diff --git a/.gitmodules b/.gitmodules index ace36122e6e..fdd48fcce01 100644 --- a/.gitmodules +++ b/.gitmodules @@ -186,3 +186,7 @@ path = contrib/cyrus-sasl url = https://github.com/cyrusimap/cyrus-sasl branch = cyrus-sasl-2.1 +[submodule "contrib/croaring"] + path = contrib/croaring + url = https://github.com/RoaringBitmap/CRoaring + branch = v0.2.66 diff --git a/CHANGELOG.md b/CHANGELOG.md index 8e4acdc293f..09ce72d20ed 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -409,7 +409,7 @@ ## ClickHouse release 20.6 -### ClickHouse release v20.6.3.28-stable +### ClickHouse release v20.6.3.28-stable #### New Feature @@ -2362,7 +2362,7 @@ No changes compared to v20.4.3.16-stable. * `Live View` table engine refactoring. [#8519](https://github.com/ClickHouse/ClickHouse/pull/8519) ([vzakaznikov](https://github.com/vzakaznikov)) * Add additional checks for external dictionaries created from DDL-queries. [#8127](https://github.com/ClickHouse/ClickHouse/pull/8127) ([alesapin](https://github.com/alesapin)) * Fix error `Column ... already exists` while using `FINAL` and `SAMPLE` together, e.g. `select count() from table final sample 1/2`. Fixes [#5186](https://github.com/ClickHouse/ClickHouse/issues/5186). [#7907](https://github.com/ClickHouse/ClickHouse/pull/7907) ([Nikolai Kochetov](https://github.com/KochetovNicolai)) -* Now table the first argument of `joinGet` function can be table indentifier. [#7707](https://github.com/ClickHouse/ClickHouse/pull/7707) ([Amos Bird](https://github.com/amosbird)) +* Now table the first argument of `joinGet` function can be table identifier. [#7707](https://github.com/ClickHouse/ClickHouse/pull/7707) ([Amos Bird](https://github.com/amosbird)) * Allow using `MaterializedView` with subqueries above `Kafka` tables. [#8197](https://github.com/ClickHouse/ClickHouse/pull/8197) ([filimonov](https://github.com/filimonov)) * Now background moves between disks run it the seprate thread pool. [#7670](https://github.com/ClickHouse/ClickHouse/pull/7670) ([Vladimir Chebotarev](https://github.com/excitoon)) * `SYSTEM RELOAD DICTIONARY` now executes synchronously. [#8240](https://github.com/ClickHouse/ClickHouse/pull/8240) ([Vitaly Baranov](https://github.com/vitlibar)) diff --git a/CMakeLists.txt b/CMakeLists.txt index 21cc74bbd2b..783a9f80b66 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -59,25 +59,6 @@ set(CMAKE_DEBUG_POSTFIX "d" CACHE STRING "Generate debug library name with a pos # For more info see https://cmake.org/cmake/help/latest/prop_gbl/USE_FOLDERS.html set_property(GLOBAL PROPERTY USE_FOLDERS ON) -# cmake 3.9+ needed. -# Usually impractical. -# See also ${ENABLE_THINLTO} -option(ENABLE_IPO "Full link time optimization") - -if(ENABLE_IPO) - cmake_policy(SET CMP0069 NEW) - include(CheckIPOSupported) - check_ipo_supported(RESULT IPO_SUPPORTED OUTPUT IPO_NOT_SUPPORTED) - if(IPO_SUPPORTED) - message(STATUS "IPO/LTO is supported, enabling") - set(CMAKE_INTERPROCEDURAL_OPTIMIZATION TRUE) - else() - message (${RECONFIGURE_MESSAGE_LEVEL} "IPO/LTO is not supported: <${IPO_NOT_SUPPORTED}>") - endif() -else() - message(STATUS "IPO/LTO not enabled.") -endif() - # Check that submodules are present only if source was downloaded with git if (EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git" AND NOT EXISTS "${ClickHouse_SOURCE_DIR}/contrib/boost/boost") message (FATAL_ERROR "Submodules are not initialized. Run\n\tgit submodule update --init --recursive") diff --git a/README.md b/README.md index 0c07021630b..03b5c988586 100644 --- a/README.md +++ b/README.md @@ -17,4 +17,6 @@ ClickHouse is an open-source column-oriented database management system that all ## Upcoming Events -* [ClickHouse virtual office hours](https://www.eventbrite.com/e/clickhouse-october-virtual-meetup-office-hours-tickets-123129500651) on October 22, 2020. +* [The Second ClickHouse Meetup East (online)](https://www.eventbrite.com/e/the-second-clickhouse-meetup-east-tickets-126787955187) on October 31, 2020. +* [ClickHouse for Enterprise Meetup (online in Russian)](https://arenadata-events.timepad.ru/event/1465249/) on November 10, 2020. + diff --git a/base/common/StringRef.h b/base/common/StringRef.h index 4376876c077..b51b95456cb 100644 --- a/base/common/StringRef.h +++ b/base/common/StringRef.h @@ -51,7 +51,7 @@ struct StringRef }; /// Here constexpr doesn't implicate inline, see https://www.viva64.com/en/w/v1043/ -/// nullptr can't be used because the StringRef values are used in SipHash's pointer arithmetics +/// nullptr can't be used because the StringRef values are used in SipHash's pointer arithmetic /// and the UBSan thinks that something like nullptr + 8 is UB. constexpr const inline char empty_string_ref_addr{}; constexpr const inline StringRef EMPTY_STRING_REF{&empty_string_ref_addr, 0}; diff --git a/base/glibc-compatibility/musl/lgammal.c b/base/glibc-compatibility/musl/lgammal.c new file mode 100644 index 00000000000..3b5d94c5051 --- /dev/null +++ b/base/glibc-compatibility/musl/lgammal.c @@ -0,0 +1,339 @@ +/* origin: OpenBSD /usr/src/lib/libm/src/ld80/e_lgammal.c */ +/* + * ==================================================== + * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved. + * + * Developed at SunPro, a Sun Microsystems, Inc. business. + * Permission to use, copy, modify, and distribute this + * software is freely granted, provided that this notice + * is preserved. + * ==================================================== + */ +/* + * Copyright (c) 2008 Stephen L. Moshier + * + * Permission to use, copy, modify, and distribute this software for any + * purpose with or without fee is hereby granted, provided that the above + * copyright notice and this permission notice appear in all copies. + * + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN + * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF + * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ +/* lgammal(x) + * Reentrant version of the logarithm of the Gamma function + * with user provide pointer for the sign of Gamma(x). + * + * Method: + * 1. Argument Reduction for 0 < x <= 8 + * Since gamma(1+s)=s*gamma(s), for x in [0,8], we may + * reduce x to a number in [1.5,2.5] by + * lgamma(1+s) = log(s) + lgamma(s) + * for example, + * lgamma(7.3) = log(6.3) + lgamma(6.3) + * = log(6.3*5.3) + lgamma(5.3) + * = log(6.3*5.3*4.3*3.3*2.3) + lgamma(2.3) + * 2. Polynomial approximation of lgamma around its + * minimun ymin=1.461632144968362245 to maintain monotonicity. + * On [ymin-0.23, ymin+0.27] (i.e., [1.23164,1.73163]), use + * Let z = x-ymin; + * lgamma(x) = -1.214862905358496078218 + z^2*poly(z) + * 2. Rational approximation in the primary interval [2,3] + * We use the following approximation: + * s = x-2.0; + * lgamma(x) = 0.5*s + s*P(s)/Q(s) + * Our algorithms are based on the following observation + * + * zeta(2)-1 2 zeta(3)-1 3 + * lgamma(2+s) = s*(1-Euler) + --------- * s - --------- * s + ... + * 2 3 + * + * where Euler = 0.5771... is the Euler constant, which is very + * close to 0.5. + * + * 3. For x>=8, we have + * lgamma(x)~(x-0.5)log(x)-x+0.5*log(2pi)+1/(12x)-1/(360x**3)+.... + * (better formula: + * lgamma(x)~(x-0.5)*(log(x)-1)-.5*(log(2pi)-1) + ...) + * Let z = 1/x, then we approximation + * f(z) = lgamma(x) - (x-0.5)(log(x)-1) + * by + * 3 5 11 + * w = w0 + w1*z + w2*z + w3*z + ... + w6*z + * + * 4. For negative x, since (G is gamma function) + * -x*G(-x)*G(x) = pi/sin(pi*x), + * we have + * G(x) = pi/(sin(pi*x)*(-x)*G(-x)) + * since G(-x) is positive, sign(G(x)) = sign(sin(pi*x)) for x<0 + * Hence, for x<0, signgam = sign(sin(pi*x)) and + * lgamma(x) = log(|Gamma(x)|) + * = log(pi/(|x*sin(pi*x)|)) - lgamma(-x); + * Note: one should avoid compute pi*(-x) directly in the + * computation of sin(pi*(-x)). + * + * 5. Special Cases + * lgamma(2+s) ~ s*(1-Euler) for tiny s + * lgamma(1)=lgamma(2)=0 + * lgamma(x) ~ -log(x) for tiny x + * lgamma(0) = lgamma(inf) = inf + * lgamma(-integer) = +-inf + * + */ + +#include +#include +#include "libm.h" + + +#if LDBL_MANT_DIG == 53 && LDBL_MAX_EXP == 1024 +double lgamma_r(double x, int *sg); + +long double lgammal_r(long double x, int *sg) +{ + return lgamma_r(x, sg); +} +#elif LDBL_MANT_DIG == 64 && LDBL_MAX_EXP == 16384 + +static const long double pi = 3.14159265358979323846264L, + +/* lgam(1+x) = 0.5 x + x a(x)/b(x) + -0.268402099609375 <= x <= 0 + peak relative error 6.6e-22 */ +a0 = -6.343246574721079391729402781192128239938E2L, +a1 = 1.856560238672465796768677717168371401378E3L, +a2 = 2.404733102163746263689288466865843408429E3L, +a3 = 8.804188795790383497379532868917517596322E2L, +a4 = 1.135361354097447729740103745999661157426E2L, +a5 = 3.766956539107615557608581581190400021285E0L, + +b0 = 8.214973713960928795704317259806842490498E3L, +b1 = 1.026343508841367384879065363925870888012E4L, +b2 = 4.553337477045763320522762343132210919277E3L, +b3 = 8.506975785032585797446253359230031874803E2L, +b4 = 6.042447899703295436820744186992189445813E1L, +/* b5 = 1.000000000000000000000000000000000000000E0 */ + + +tc = 1.4616321449683623412626595423257213284682E0L, +tf = -1.2148629053584961146050602565082954242826E-1, /* double precision */ +/* tt = (tail of tf), i.e. tf + tt has extended precision. */ +tt = 3.3649914684731379602768989080467587736363E-18L, +/* lgam ( 1.4616321449683623412626595423257213284682E0 ) = +-1.2148629053584960809551455717769158215135617312999903886372437313313530E-1 */ + +/* lgam (x + tc) = tf + tt + x g(x)/h(x) + -0.230003726999612341262659542325721328468 <= x + <= 0.2699962730003876587373404576742786715318 + peak relative error 2.1e-21 */ +g0 = 3.645529916721223331888305293534095553827E-18L, +g1 = 5.126654642791082497002594216163574795690E3L, +g2 = 8.828603575854624811911631336122070070327E3L, +g3 = 5.464186426932117031234820886525701595203E3L, +g4 = 1.455427403530884193180776558102868592293E3L, +g5 = 1.541735456969245924860307497029155838446E2L, +g6 = 4.335498275274822298341872707453445815118E0L, + +h0 = 1.059584930106085509696730443974495979641E4L, +h1 = 2.147921653490043010629481226937850618860E4L, +h2 = 1.643014770044524804175197151958100656728E4L, +h3 = 5.869021995186925517228323497501767586078E3L, +h4 = 9.764244777714344488787381271643502742293E2L, +h5 = 6.442485441570592541741092969581997002349E1L, +/* h6 = 1.000000000000000000000000000000000000000E0 */ + + +/* lgam (x+1) = -0.5 x + x u(x)/v(x) + -0.100006103515625 <= x <= 0.231639862060546875 + peak relative error 1.3e-21 */ +u0 = -8.886217500092090678492242071879342025627E1L, +u1 = 6.840109978129177639438792958320783599310E2L, +u2 = 2.042626104514127267855588786511809932433E3L, +u3 = 1.911723903442667422201651063009856064275E3L, +u4 = 7.447065275665887457628865263491667767695E2L, +u5 = 1.132256494121790736268471016493103952637E2L, +u6 = 4.484398885516614191003094714505960972894E0L, + +v0 = 1.150830924194461522996462401210374632929E3L, +v1 = 3.399692260848747447377972081399737098610E3L, +v2 = 3.786631705644460255229513563657226008015E3L, +v3 = 1.966450123004478374557778781564114347876E3L, +v4 = 4.741359068914069299837355438370682773122E2L, +v5 = 4.508989649747184050907206782117647852364E1L, +/* v6 = 1.000000000000000000000000000000000000000E0 */ + + +/* lgam (x+2) = .5 x + x s(x)/r(x) + 0 <= x <= 1 + peak relative error 7.2e-22 */ +s0 = 1.454726263410661942989109455292824853344E6L, +s1 = -3.901428390086348447890408306153378922752E6L, +s2 = -6.573568698209374121847873064292963089438E6L, +s3 = -3.319055881485044417245964508099095984643E6L, +s4 = -7.094891568758439227560184618114707107977E5L, +s5 = -6.263426646464505837422314539808112478303E4L, +s6 = -1.684926520999477529949915657519454051529E3L, + +r0 = -1.883978160734303518163008696712983134698E7L, +r1 = -2.815206082812062064902202753264922306830E7L, +r2 = -1.600245495251915899081846093343626358398E7L, +r3 = -4.310526301881305003489257052083370058799E6L, +r4 = -5.563807682263923279438235987186184968542E5L, +r5 = -3.027734654434169996032905158145259713083E4L, +r6 = -4.501995652861105629217250715790764371267E2L, +/* r6 = 1.000000000000000000000000000000000000000E0 */ + + +/* lgam(x) = ( x - 0.5 ) * log(x) - x + LS2PI + 1/x w(1/x^2) + x >= 8 + Peak relative error 1.51e-21 +w0 = LS2PI - 0.5 */ +w0 = 4.189385332046727417803e-1L, +w1 = 8.333333333333331447505E-2L, +w2 = -2.777777777750349603440E-3L, +w3 = 7.936507795855070755671E-4L, +w4 = -5.952345851765688514613E-4L, +w5 = 8.412723297322498080632E-4L, +w6 = -1.880801938119376907179E-3L, +w7 = 4.885026142432270781165E-3L; + + +long double lgammal_r(long double x, int *sg) { + long double t, y, z, nadj, p, p1, p2, q, r, w; + union ldshape u = {x}; + uint32_t ix = (u.i.se & 0x7fffU)<<16 | u.i.m>>48; + int sign = u.i.se >> 15; + int i; + + *sg = 1; + + /* purge off +-inf, NaN, +-0, tiny and negative arguments */ + if (ix >= 0x7fff0000) + return x * x; + if (ix < 0x3fc08000) { /* |x|<2**-63, return -log(|x|) */ + if (sign) { + *sg = -1; + x = -x; + } + return -logl(x); + } + if (sign) { + x = -x; + t = sin(pi * x); + if (t == 0.0) + return 1.0 / (x-x); /* -integer */ + if (t > 0.0) + *sg = -1; + else + t = -t; + nadj = logl(pi / (t * x)); + } + + /* purge off 1 and 2 (so the sign is ok with downward rounding) */ + if ((ix == 0x3fff8000 || ix == 0x40008000) && u.i.m == 0) { + r = 0; + } else if (ix < 0x40008000) { /* x < 2.0 */ + if (ix <= 0x3ffee666) { /* 8.99993896484375e-1 */ + /* lgamma(x) = lgamma(x+1) - log(x) */ + r = -logl(x); + if (ix >= 0x3ffebb4a) { /* 7.31597900390625e-1 */ + y = x - 1.0; + i = 0; + } else if (ix >= 0x3ffced33) { /* 2.31639862060546875e-1 */ + y = x - (tc - 1.0); + i = 1; + } else { /* x < 0.23 */ + y = x; + i = 2; + } + } else { + r = 0.0; + if (ix >= 0x3fffdda6) { /* 1.73162841796875 */ + /* [1.7316,2] */ + y = x - 2.0; + i = 0; + } else if (ix >= 0x3fff9da6) { /* 1.23162841796875 */ + /* [1.23,1.73] */ + y = x - tc; + i = 1; + } else { + /* [0.9, 1.23] */ + y = x - 1.0; + i = 2; + } + } + switch (i) { + case 0: + p1 = a0 + y * (a1 + y * (a2 + y * (a3 + y * (a4 + y * a5)))); + p2 = b0 + y * (b1 + y * (b2 + y * (b3 + y * (b4 + y)))); + r += 0.5 * y + y * p1/p2; + break; + case 1: + p1 = g0 + y * (g1 + y * (g2 + y * (g3 + y * (g4 + y * (g5 + y * g6))))); + p2 = h0 + y * (h1 + y * (h2 + y * (h3 + y * (h4 + y * (h5 + y))))); + p = tt + y * p1/p2; + r += (tf + p); + break; + case 2: + p1 = y * (u0 + y * (u1 + y * (u2 + y * (u3 + y * (u4 + y * (u5 + y * u6)))))); + p2 = v0 + y * (v1 + y * (v2 + y * (v3 + y * (v4 + y * (v5 + y))))); + r += (-0.5 * y + p1 / p2); + } + } else if (ix < 0x40028000) { /* 8.0 */ + /* x < 8.0 */ + i = (int)x; + y = x - (double)i; + p = y * (s0 + y * (s1 + y * (s2 + y * (s3 + y * (s4 + y * (s5 + y * s6)))))); + q = r0 + y * (r1 + y * (r2 + y * (r3 + y * (r4 + y * (r5 + y * (r6 + y)))))); + r = 0.5 * y + p / q; + z = 1.0; + /* lgamma(1+s) = log(s) + lgamma(s) */ + switch (i) { + case 7: + z *= (y + 6.0); /* FALLTHRU */ + case 6: + z *= (y + 5.0); /* FALLTHRU */ + case 5: + z *= (y + 4.0); /* FALLTHRU */ + case 4: + z *= (y + 3.0); /* FALLTHRU */ + case 3: + z *= (y + 2.0); /* FALLTHRU */ + r += logl(z); + break; + } + } else if (ix < 0x40418000) { /* 2^66 */ + /* 8.0 <= x < 2**66 */ + t = logl(x); + z = 1.0 / x; + y = z * z; + w = w0 + z * (w1 + y * (w2 + y * (w3 + y * (w4 + y * (w5 + y * (w6 + y * w7)))))); + r = (x - 0.5) * (t - 1.0) + w; + } else /* 2**66 <= x <= inf */ + r = x * (logl(x) - 1.0); + if (sign) + r = nadj - r; + return r; +} +#elif LDBL_MANT_DIG == 113 && LDBL_MAX_EXP == 16384 +// TODO: broken implementation to make things compile +double lgamma_r(double x, int *sg); + +long double lgammal_r(long double x, int *sg) +{ + return lgamma_r(x, sg); +} +#endif + + +int signgam_lgammal; + +long double lgammal(long double x) +{ + return lgammal_r(x, &signgam_lgammal); +} + diff --git a/cmake/arch.cmake b/cmake/arch.cmake index 57ed42295bb..9604ef62b31 100644 --- a/cmake/arch.cmake +++ b/cmake/arch.cmake @@ -16,8 +16,4 @@ endif () if (CMAKE_SYSTEM_PROCESSOR MATCHES "^(ppc64le.*|PPC64LE.*)") set (ARCH_PPC64LE 1) - # FIXME: move this check into tools.cmake - if (COMPILER_CLANG OR (COMPILER_GCC AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 8)) - message(FATAL_ERROR "Only gcc-8 or higher is supported for powerpc architecture") - endif () endif () diff --git a/cmake/sanitize.cmake b/cmake/sanitize.cmake index 0ccd6933dec..7a9e263778e 100644 --- a/cmake/sanitize.cmake +++ b/cmake/sanitize.cmake @@ -57,8 +57,8 @@ if (SANITIZE) endif () elseif (SANITIZE STREQUAL "undefined") - set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SAN_FLAGS} -fsanitize=undefined -fno-sanitize-recover=all -fno-sanitize=float-divide-by-zero") - set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${SAN_FLAGS} -fsanitize=undefined -fno-sanitize-recover=all -fno-sanitize=float-divide-by-zero") + set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SAN_FLAGS} -fsanitize=undefined -fno-sanitize-recover=all -fno-sanitize=float-divide-by-zero -fsanitize-blacklist=${CMAKE_SOURCE_DIR}/tests/ubsan_suppressions.txt") + set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${SAN_FLAGS} -fsanitize=undefined -fno-sanitize-recover=all -fno-sanitize=float-divide-by-zero -fsanitize-blacklist=${CMAKE_SOURCE_DIR}/tests/ubsan_suppressions.txt") if (CMAKE_CXX_COMPILER_ID STREQUAL "GNU") set (CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=undefined") endif() diff --git a/cmake/tools.cmake b/cmake/tools.cmake index 6f07cc2439c..4e9258a5cb7 100644 --- a/cmake/tools.cmake +++ b/cmake/tools.cmake @@ -15,6 +15,10 @@ if (COMPILER_GCC) elseif (COMPILER_CLANG) # Require minimum version of clang/apple-clang if (CMAKE_CXX_COMPILER_ID MATCHES "AppleClang") + # If you are developer you can figure out what exact versions of AppleClang are Ok, + # remove the following line and commit changes below. + message (FATAL_ERROR "AppleClang is not supported, you should install clang from brew.") + # AppleClang 10.0.1 (Xcode 10.2) corresponds to LLVM/Clang upstream version 7.0.0 # AppleClang 11.0.0 (Xcode 11.0) corresponds to LLVM/Clang upstream version 8.0.0 set (XCODE_MINIMUM_VERSION 10.2) @@ -80,3 +84,9 @@ if (LINKER_NAME) message(STATUS "Using custom linker by name: ${LINKER_NAME}") endif () + +if (ARCH_PPC64LE) + if (COMPILER_CLANG OR (COMPILER_GCC AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 8)) + message(FATAL_ERROR "Only gcc-8 or higher is supported for powerpc architecture") + endif () +endif () \ No newline at end of file diff --git a/cmake/yandex/ya.make.versions.inc b/cmake/yandex/ya.make.versions.inc index 3ac401cb108..6910164d0aa 100644 --- a/cmake/yandex/ya.make.versions.inc +++ b/cmake/yandex/ya.make.versions.inc @@ -11,11 +11,11 @@ CFLAGS (GLOBAL -DDBMS_VERSION_MAJOR=${VERSION_MAJOR}) CFLAGS (GLOBAL -DDBMS_VERSION_MINOR=${VERSION_MINOR}) CFLAGS (GLOBAL -DDBMS_VERSION_PATCH=${VERSION_PATCH}) CFLAGS (GLOBAL -DVERSION_FULL=\"\\\"${VERSION_FULL}\\\"\") -CFLAGS (GLOBAL -DVERSION_MAJOR=${VERSION_MAJOR}) -CFLAGS (GLOBAL -DVERSION_MINOR=${VERSION_MINOR}) +CFLAGS (GLOBAL -DVERSION_MAJOR=${VERSION_MAJOR}) +CFLAGS (GLOBAL -DVERSION_MINOR=${VERSION_MINOR}) CFLAGS (GLOBAL -DVERSION_PATCH=${VERSION_PATCH}) -# TODO: not supported yet, not sure if ya.make supports arithmetics. +# TODO: not supported yet, not sure if ya.make supports arithmetic. CFLAGS (GLOBAL -DVERSION_INTEGER=0) CFLAGS (GLOBAL -DVERSION_NAME=\"\\\"${VERSION_NAME}\\\"\") diff --git a/contrib/CMakeLists.txt b/contrib/CMakeLists.txt index 130e4b13c91..7d6b9c0e374 100644 --- a/contrib/CMakeLists.txt +++ b/contrib/CMakeLists.txt @@ -20,7 +20,6 @@ add_subdirectory (boost-cmake) add_subdirectory (cctz-cmake) add_subdirectory (consistent-hashing-sumbur) add_subdirectory (consistent-hashing) -add_subdirectory (croaring) add_subdirectory (FastMemcpy) add_subdirectory (hyperscan-cmake) add_subdirectory (jemalloc-cmake) @@ -34,6 +33,7 @@ add_subdirectory (ryu-cmake) add_subdirectory (unixodbc-cmake) add_subdirectory (poco-cmake) +add_subdirectory (croaring-cmake) # TODO: refactor the contrib libraries below this comment. diff --git a/contrib/croaring b/contrib/croaring new file mode 160000 index 00000000000..5f20740ec0d --- /dev/null +++ b/contrib/croaring @@ -0,0 +1 @@ +Subproject commit 5f20740ec0de5e153e8f4cb2ab91814e8b291a14 diff --git a/contrib/croaring-cmake/CMakeLists.txt b/contrib/croaring-cmake/CMakeLists.txt new file mode 100644 index 00000000000..3189795347b --- /dev/null +++ b/contrib/croaring-cmake/CMakeLists.txt @@ -0,0 +1,25 @@ +set(LIBRARY_DIR ${ClickHouse_SOURCE_DIR}/contrib/croaring) + +set(SRCS + ${LIBRARY_DIR}/src/array_util.c + ${LIBRARY_DIR}/src/bitset_util.c + ${LIBRARY_DIR}/src/containers/array.c + ${LIBRARY_DIR}/src/containers/bitset.c + ${LIBRARY_DIR}/src/containers/containers.c + ${LIBRARY_DIR}/src/containers/convert.c + ${LIBRARY_DIR}/src/containers/mixed_intersection.c + ${LIBRARY_DIR}/src/containers/mixed_union.c + ${LIBRARY_DIR}/src/containers/mixed_equal.c + ${LIBRARY_DIR}/src/containers/mixed_subset.c + ${LIBRARY_DIR}/src/containers/mixed_negation.c + ${LIBRARY_DIR}/src/containers/mixed_xor.c + ${LIBRARY_DIR}/src/containers/mixed_andnot.c + ${LIBRARY_DIR}/src/containers/run.c + ${LIBRARY_DIR}/src/roaring.c + ${LIBRARY_DIR}/src/roaring_priority_queue.c + ${LIBRARY_DIR}/src/roaring_array.c) + +add_library(roaring ${SRCS}) + +target_include_directories(roaring PRIVATE ${LIBRARY_DIR}/include/roaring) +target_include_directories(roaring SYSTEM BEFORE PUBLIC ${LIBRARY_DIR}/include) diff --git a/contrib/croaring/CMakeLists.txt b/contrib/croaring/CMakeLists.txt deleted file mode 100644 index da19911487f..00000000000 --- a/contrib/croaring/CMakeLists.txt +++ /dev/null @@ -1,6 +0,0 @@ -add_library(roaring - roaring.c - roaring/roaring.h - roaring/roaring.hh) - -target_include_directories (roaring SYSTEM PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}) diff --git a/contrib/croaring/LICENSE b/contrib/croaring/LICENSE deleted file mode 100644 index 3265476ea81..00000000000 --- a/contrib/croaring/LICENSE +++ /dev/null @@ -1,202 +0,0 @@ - Apache License - Version 2.0, January 2004 - http://www.apache.org/licenses/ - - TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION - - 1. Definitions. - - "License" shall mean the terms and conditions for use, reproduction, - and distribution as defined by Sections 1 through 9 of this document. - - "Licensor" shall mean the copyright owner or entity authorized by - the copyright owner that is granting the License. - - "Legal Entity" shall mean the union of the acting entity and all - other entities that control, are controlled by, or are under common - control with that entity. For the purposes of this definition, - "control" means (i) the power, direct or indirect, to cause the - direction or management of such entity, whether by contract or - otherwise, or (ii) ownership of fifty percent (50%) or more of the - outstanding shares, or (iii) beneficial ownership of such entity. - - "You" (or "Your") shall mean an individual or Legal Entity - exercising permissions granted by this License. - - "Source" form shall mean the preferred form for making modifications, - including but not limited to software source code, documentation - source, and configuration files. - - "Object" form shall mean any form resulting from mechanical - transformation or translation of a Source form, including but - not limited to compiled object code, generated documentation, - and conversions to other media types. - - "Work" shall mean the work of authorship, whether in Source or - Object form, made available under the License, as indicated by a - copyright notice that is included in or attached to the work - (an example is provided in the Appendix below). - - "Derivative Works" shall mean any work, whether in Source or Object - form, that is based on (or derived from) the Work and for which the - editorial revisions, annotations, elaborations, or other modifications - represent, as a whole, an original work of authorship. For the purposes - of this License, Derivative Works shall not include works that remain - separable from, or merely link (or bind by name) to the interfaces of, - the Work and Derivative Works thereof. - - "Contribution" shall mean any work of authorship, including - the original version of the Work and any modifications or additions - to that Work or Derivative Works thereof, that is intentionally - submitted to Licensor for inclusion in the Work by the copyright owner - or by an individual or Legal Entity authorized to submit on behalf of - the copyright owner. For the purposes of this definition, "submitted" - means any form of electronic, verbal, or written communication sent - to the Licensor or its representatives, including but not limited to - communication on electronic mailing lists, source code control systems, - and issue tracking systems that are managed by, or on behalf of, the - Licensor for the purpose of discussing and improving the Work, but - excluding communication that is conspicuously marked or otherwise - designated in writing by the copyright owner as "Not a Contribution." - - "Contributor" shall mean Licensor and any individual or Legal Entity - on behalf of whom a Contribution has been received by Licensor and - subsequently incorporated within the Work. - - 2. Grant of Copyright License. Subject to the terms and conditions of - this License, each Contributor hereby grants to You a perpetual, - worldwide, non-exclusive, no-charge, royalty-free, irrevocable - copyright license to reproduce, prepare Derivative Works of, - publicly display, publicly perform, sublicense, and distribute the - Work and such Derivative Works in Source or Object form. - - 3. Grant of Patent License. Subject to the terms and conditions of - this License, each Contributor hereby grants to You a perpetual, - worldwide, non-exclusive, no-charge, royalty-free, irrevocable - (except as stated in this section) patent license to make, have made, - use, offer to sell, sell, import, and otherwise transfer the Work, - where such license applies only to those patent claims licensable - by such Contributor that are necessarily infringed by their - Contribution(s) alone or by combination of their Contribution(s) - with the Work to which such Contribution(s) was submitted. If You - institute patent litigation against any entity (including a - cross-claim or counterclaim in a lawsuit) alleging that the Work - or a Contribution incorporated within the Work constitutes direct - or contributory patent infringement, then any patent licenses - granted to You under this License for that Work shall terminate - as of the date such litigation is filed. - - 4. Redistribution. You may reproduce and distribute copies of the - Work or Derivative Works thereof in any medium, with or without - modifications, and in Source or Object form, provided that You - meet the following conditions: - - (a) You must give any other recipients of the Work or - Derivative Works a copy of this License; and - - (b) You must cause any modified files to carry prominent notices - stating that You changed the files; and - - (c) You must retain, in the Source form of any Derivative Works - that You distribute, all copyright, patent, trademark, and - attribution notices from the Source form of the Work, - excluding those notices that do not pertain to any part of - the Derivative Works; and - - (d) If the Work includes a "NOTICE" text file as part of its - distribution, then any Derivative Works that You distribute must - include a readable copy of the attribution notices contained - within such NOTICE file, excluding those notices that do not - pertain to any part of the Derivative Works, in at least one - of the following places: within a NOTICE text file distributed - as part of the Derivative Works; within the Source form or - documentation, if provided along with the Derivative Works; or, - within a display generated by the Derivative Works, if and - wherever such third-party notices normally appear. The contents - of the NOTICE file are for informational purposes only and - do not modify the License. You may add Your own attribution - notices within Derivative Works that You distribute, alongside - or as an addendum to the NOTICE text from the Work, provided - that such additional attribution notices cannot be construed - as modifying the License. - - You may add Your own copyright statement to Your modifications and - may provide additional or different license terms and conditions - for use, reproduction, or distribution of Your modifications, or - for any such Derivative Works as a whole, provided Your use, - reproduction, and distribution of the Work otherwise complies with - the conditions stated in this License. - - 5. Submission of Contributions. Unless You explicitly state otherwise, - any Contribution intentionally submitted for inclusion in the Work - by You to the Licensor shall be under the terms and conditions of - this License, without any additional terms or conditions. - Notwithstanding the above, nothing herein shall supersede or modify - the terms of any separate license agreement you may have executed - with Licensor regarding such Contributions. - - 6. Trademarks. This License does not grant permission to use the trade - names, trademarks, service marks, or product names of the Licensor, - except as required for reasonable and customary use in describing the - origin of the Work and reproducing the content of the NOTICE file. - - 7. Disclaimer of Warranty. Unless required by applicable law or - agreed to in writing, Licensor provides the Work (and each - Contributor provides its Contributions) on an "AS IS" BASIS, - WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or - implied, including, without limitation, any warranties or conditions - of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A - PARTICULAR PURPOSE. You are solely responsible for determining the - appropriateness of using or redistributing the Work and assume any - risks associated with Your exercise of permissions under this License. - - 8. Limitation of Liability. In no event and under no legal theory, - whether in tort (including negligence), contract, or otherwise, - unless required by applicable law (such as deliberate and grossly - negligent acts) or agreed to in writing, shall any Contributor be - liable to You for damages, including any direct, indirect, special, - incidental, or consequential damages of any character arising as a - result of this License or out of the use or inability to use the - Work (including but not limited to damages for loss of goodwill, - work stoppage, computer failure or malfunction, or any and all - other commercial damages or losses), even if such Contributor - has been advised of the possibility of such damages. - - 9. Accepting Warranty or Additional Liability. While redistributing - the Work or Derivative Works thereof, You may choose to offer, - and charge a fee for, acceptance of support, warranty, indemnity, - or other liability obligations and/or rights consistent with this - License. However, in accepting such obligations, You may act only - on Your own behalf and on Your sole responsibility, not on behalf - of any other Contributor, and only if You agree to indemnify, - defend, and hold each Contributor harmless for any liability - incurred by, or claims asserted against, such Contributor by reason - of your accepting any such warranty or additional liability. - - END OF TERMS AND CONDITIONS - - APPENDIX: How to apply the Apache License to your work. - - To apply the Apache License to your work, attach the following - boilerplate notice, with the fields enclosed by brackets "{}" - replaced with your own identifying information. (Don't include - the brackets!) The text should be enclosed in the appropriate - comment syntax for the file format. We also recommend that a - file or class name and description of purpose be included on the - same "printed page" as the copyright notice for easier - identification within third-party archives. - - Copyright 2016 The CRoaring authors - - Licensed under the Apache License, Version 2.0 (the "License"); - you may not use this file except in compliance with the License. - You may obtain a copy of the License at - - http://www.apache.org/licenses/LICENSE-2.0 - - Unless required by applicable law or agreed to in writing, software - distributed under the License is distributed on an "AS IS" BASIS, - WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - See the License for the specific language governing permissions and - limitations under the License. - diff --git a/contrib/croaring/README.txt b/contrib/croaring/README.txt deleted file mode 100644 index 3daa1c43ed8..00000000000 --- a/contrib/croaring/README.txt +++ /dev/null @@ -1,2 +0,0 @@ -download from https://github.com/RoaringBitmap/CRoaring/archive/v0.2.57.tar.gz -and use ./amalgamation.sh generate diff --git a/contrib/croaring/roaring.c b/contrib/croaring/roaring.c deleted file mode 100644 index 6327db7ade3..00000000000 --- a/contrib/croaring/roaring.c +++ /dev/null @@ -1,11093 +0,0 @@ -/* auto-generated on Tue Dec 18 09:42:59 CST 2018. Do not edit! */ -#include "roaring/roaring.h" - -/* used for http://dmalloc.com/ Dmalloc - Debug Malloc Library */ -#ifdef DMALLOC -#include "dmalloc.h" -#endif - -/* begin file /opt/bitmap/CRoaring-0.2.57/src/array_util.c */ -#include -#include -#include -#include -#include -#include - -extern inline int32_t binarySearch(const uint16_t *array, int32_t lenarray, - uint16_t ikey); - -#ifdef USESSE4 -// used by intersect_vector16 -ALIGNED(0x1000) -static const uint8_t shuffle_mask16[] = { - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 8, 9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 8, 9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 6, 7, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, 8, 9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, - 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 6, 7, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, 8, 9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 6, 7, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 6, 7, 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 6, 7, - 8, 9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 10, 11, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 10, 11, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 6, 7, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 6, 7, 10, 11, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 6, 7, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 6, 7, 10, 11, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 6, 7, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, 10, 11, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 8, 9, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, 10, 11, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 8, 9, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 8, 9, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 8, 9, 10, 11, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 6, 7, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 8, 9, - 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 8, 9, 10, 11, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, - 0xFF, 0xFF, 0xFF, 0xFF, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 6, 7, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 6, 7, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 6, 7, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 6, 7, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 6, 7, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 8, 9, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 8, 9, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 8, 9, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 8, 9, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 8, 9, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, - 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 6, 7, 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 6, 7, 8, 9, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, - 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 6, 7, 8, 9, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 6, 7, 8, 9, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 6, 7, 8, 9, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 10, 11, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 10, 11, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 10, 11, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 6, 7, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 10, 11, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, - 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 8, 9, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, - 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 8, 9, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 6, 7, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, 8, 9, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, - 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 6, 7, 8, 9, 10, 11, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, 8, 9, 10, 11, - 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 6, 7, 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 6, 7, - 8, 9, 10, 11, 12, 13, 0xFF, 0xFF, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 6, 7, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 6, 7, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 6, 7, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 6, 7, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 6, 7, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 8, 9, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 8, 9, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 8, 9, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 8, 9, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 6, 7, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 8, 9, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 8, 9, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 10, 11, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 10, 11, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 6, 7, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, - 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 6, 7, 10, 11, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 6, 7, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 6, 7, 10, 11, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 6, 7, - 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 8, 9, 10, 11, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 8, 9, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 8, 9, 10, 11, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 8, 9, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 8, 9, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, - 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 6, 7, 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 6, 7, 8, 9, - 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, - 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 6, 7, 8, 9, 10, 11, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 6, 7, 8, 9, - 10, 11, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 6, 7, 8, 9, 10, 11, 14, 15, 0xFF, 0xFF, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 12, 13, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 12, 13, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 6, 7, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 8, 9, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 8, 9, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 6, 7, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, 8, 9, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, - 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 6, 7, 8, 9, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, 8, 9, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 6, 7, 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 6, 7, 8, 9, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 6, 7, - 8, 9, 12, 13, 14, 15, 0xFF, 0xFF, 10, 11, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 4, 5, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 10, 11, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 6, 7, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 6, 7, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 6, 7, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 6, 7, 10, 11, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 6, 7, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 4, 5, 6, 7, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 4, 5, 6, 7, 10, 11, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 4, 5, 6, 7, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, - 8, 9, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 8, 9, 10, 11, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 8, 9, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 2, 3, 8, 9, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 4, 5, 8, 9, 10, 11, 12, 13, - 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, - 8, 9, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, - 2, 3, 4, 5, 8, 9, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, 4, 5, 8, 9, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 6, 7, 8, 9, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0, 1, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 2, 3, 6, 7, 8, 9, 10, 11, - 12, 13, 14, 15, 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 2, 3, - 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, - 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, - 0xFF, 0xFF, 0xFF, 0xFF, 0, 1, 4, 5, 6, 7, 8, 9, - 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, 2, 3, 4, 5, - 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 0xFF, 0xFF, - 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, - 12, 13, 14, 15}; - -/** - * From Schlegel et al., Fast Sorted-Set Intersection using SIMD Instructions - * Optimized by D. Lemire on May 3rd 2013 - */ -int32_t intersect_vector16(const uint16_t *__restrict__ A, size_t s_a, - const uint16_t *__restrict__ B, size_t s_b, - uint16_t *C) { - size_t count = 0; - size_t i_a = 0, i_b = 0; - const int vectorlength = sizeof(__m128i) / sizeof(uint16_t); - const size_t st_a = (s_a / vectorlength) * vectorlength; - const size_t st_b = (s_b / vectorlength) * vectorlength; - __m128i v_a, v_b; - if ((i_a < st_a) && (i_b < st_b)) { - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - while ((A[i_a] == 0) || (B[i_b] == 0)) { - const __m128i res_v = _mm_cmpestrm( - v_b, vectorlength, v_a, vectorlength, - _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); - const int r = _mm_extract_epi32(res_v, 0); - __m128i sm16 = _mm_load_si128((const __m128i *)shuffle_mask16 + r); - __m128i p = _mm_shuffle_epi8(v_a, sm16); - _mm_storeu_si128((__m128i *)&C[count], p); // can overflow - count += _mm_popcnt_u32(r); - const uint16_t a_max = A[i_a + vectorlength - 1]; - const uint16_t b_max = B[i_b + vectorlength - 1]; - if (a_max <= b_max) { - i_a += vectorlength; - if (i_a == st_a) break; - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - } - if (b_max <= a_max) { - i_b += vectorlength; - if (i_b == st_b) break; - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - } - } - if ((i_a < st_a) && (i_b < st_b)) - while (true) { - const __m128i res_v = _mm_cmpistrm( - v_b, v_a, - _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); - const int r = _mm_extract_epi32(res_v, 0); - __m128i sm16 = - _mm_load_si128((const __m128i *)shuffle_mask16 + r); - __m128i p = _mm_shuffle_epi8(v_a, sm16); - _mm_storeu_si128((__m128i *)&C[count], p); // can overflow - count += _mm_popcnt_u32(r); - const uint16_t a_max = A[i_a + vectorlength - 1]; - const uint16_t b_max = B[i_b + vectorlength - 1]; - if (a_max <= b_max) { - i_a += vectorlength; - if (i_a == st_a) break; - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - } - if (b_max <= a_max) { - i_b += vectorlength; - if (i_b == st_b) break; - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - } - } - } - // intersect the tail using scalar intersection - while (i_a < s_a && i_b < s_b) { - uint16_t a = A[i_a]; - uint16_t b = B[i_b]; - if (a < b) { - i_a++; - } else if (b < a) { - i_b++; - } else { - C[count] = a; //==b; - count++; - i_a++; - i_b++; - } - } - return (int32_t)count; -} - -int32_t intersect_vector16_cardinality(const uint16_t *__restrict__ A, - size_t s_a, - const uint16_t *__restrict__ B, - size_t s_b) { - size_t count = 0; - size_t i_a = 0, i_b = 0; - const int vectorlength = sizeof(__m128i) / sizeof(uint16_t); - const size_t st_a = (s_a / vectorlength) * vectorlength; - const size_t st_b = (s_b / vectorlength) * vectorlength; - __m128i v_a, v_b; - if ((i_a < st_a) && (i_b < st_b)) { - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - while ((A[i_a] == 0) || (B[i_b] == 0)) { - const __m128i res_v = _mm_cmpestrm( - v_b, vectorlength, v_a, vectorlength, - _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); - const int r = _mm_extract_epi32(res_v, 0); - count += _mm_popcnt_u32(r); - const uint16_t a_max = A[i_a + vectorlength - 1]; - const uint16_t b_max = B[i_b + vectorlength - 1]; - if (a_max <= b_max) { - i_a += vectorlength; - if (i_a == st_a) break; - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - } - if (b_max <= a_max) { - i_b += vectorlength; - if (i_b == st_b) break; - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - } - } - if ((i_a < st_a) && (i_b < st_b)) - while (true) { - const __m128i res_v = _mm_cmpistrm( - v_b, v_a, - _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); - const int r = _mm_extract_epi32(res_v, 0); - count += _mm_popcnt_u32(r); - const uint16_t a_max = A[i_a + vectorlength - 1]; - const uint16_t b_max = B[i_b + vectorlength - 1]; - if (a_max <= b_max) { - i_a += vectorlength; - if (i_a == st_a) break; - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - } - if (b_max <= a_max) { - i_b += vectorlength; - if (i_b == st_b) break; - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - } - } - } - // intersect the tail using scalar intersection - while (i_a < s_a && i_b < s_b) { - uint16_t a = A[i_a]; - uint16_t b = B[i_b]; - if (a < b) { - i_a++; - } else if (b < a) { - i_b++; - } else { - count++; - i_a++; - i_b++; - } - } - return (int32_t)count; -} - -int32_t difference_vector16(const uint16_t *__restrict__ A, size_t s_a, - const uint16_t *__restrict__ B, size_t s_b, - uint16_t *C) { - // we handle the degenerate case - if (s_a == 0) return 0; - if (s_b == 0) { - if (A != C) memcpy(C, A, sizeof(uint16_t) * s_a); - return (int32_t)s_a; - } - // handle the leading zeroes, it is messy but it allows us to use the fast - // _mm_cmpistrm instrinsic safely - int32_t count = 0; - if ((A[0] == 0) || (B[0] == 0)) { - if ((A[0] == 0) && (B[0] == 0)) { - A++; - s_a--; - B++; - s_b--; - } else if (A[0] == 0) { - C[count++] = 0; - A++; - s_a--; - } else { - B++; - s_b--; - } - } - // at this point, we have two non-empty arrays, made of non-zero - // increasing values. - size_t i_a = 0, i_b = 0; - const size_t vectorlength = sizeof(__m128i) / sizeof(uint16_t); - const size_t st_a = (s_a / vectorlength) * vectorlength; - const size_t st_b = (s_b / vectorlength) * vectorlength; - if ((i_a < st_a) && (i_b < st_b)) { // this is the vectorized code path - __m128i v_a, v_b; //, v_bmax; - // we load a vector from A and a vector from B - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - // we have a runningmask which indicates which values from A have been - // spotted in B, these don't get written out. - __m128i runningmask_a_found_in_b = _mm_setzero_si128(); - /**** - * start of the main vectorized loop - *****/ - while (true) { - // afoundinb will contain a mask indicate for each entry in A - // whether it is seen - // in B - const __m128i a_found_in_b = - _mm_cmpistrm(v_b, v_a, _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | - _SIDD_BIT_MASK); - runningmask_a_found_in_b = - _mm_or_si128(runningmask_a_found_in_b, a_found_in_b); - // we always compare the last values of A and B - const uint16_t a_max = A[i_a + vectorlength - 1]; - const uint16_t b_max = B[i_b + vectorlength - 1]; - if (a_max <= b_max) { - // Ok. In this code path, we are ready to write our v_a - // because there is no need to read more from B, they will - // all be large values. - const int bitmask_belongs_to_difference = - _mm_extract_epi32(runningmask_a_found_in_b, 0) ^ 0xFF; - /*** next few lines are probably expensive *****/ - __m128i sm16 = _mm_load_si128((const __m128i *)shuffle_mask16 + - bitmask_belongs_to_difference); - __m128i p = _mm_shuffle_epi8(v_a, sm16); - _mm_storeu_si128((__m128i *)&C[count], p); // can overflow - count += _mm_popcnt_u32(bitmask_belongs_to_difference); - // we advance a - i_a += vectorlength; - if (i_a == st_a) // no more - break; - runningmask_a_found_in_b = _mm_setzero_si128(); - v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); - } - if (b_max <= a_max) { - // in this code path, the current v_b has become useless - i_b += vectorlength; - if (i_b == st_b) break; - v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); - } - } - // at this point, either we have i_a == st_a, which is the end of the - // vectorized processing, - // or we have i_b == st_b, and we are not done processing the vector... - // so we need to finish it off. - if (i_a < st_a) { // we have unfinished business... - uint16_t buffer[8]; // buffer to do a masked load - memset(buffer, 0, 8 * sizeof(uint16_t)); - memcpy(buffer, B + i_b, (s_b - i_b) * sizeof(uint16_t)); - v_b = _mm_lddqu_si128((__m128i *)buffer); - const __m128i a_found_in_b = - _mm_cmpistrm(v_b, v_a, _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | - _SIDD_BIT_MASK); - runningmask_a_found_in_b = - _mm_or_si128(runningmask_a_found_in_b, a_found_in_b); - const int bitmask_belongs_to_difference = - _mm_extract_epi32(runningmask_a_found_in_b, 0) ^ 0xFF; - __m128i sm16 = _mm_load_si128((const __m128i *)shuffle_mask16 + - bitmask_belongs_to_difference); - __m128i p = _mm_shuffle_epi8(v_a, sm16); - _mm_storeu_si128((__m128i *)&C[count], p); // can overflow - count += _mm_popcnt_u32(bitmask_belongs_to_difference); - i_a += vectorlength; - } - // at this point we should have i_a == st_a and i_b == st_b - } - // do the tail using scalar code - while (i_a < s_a && i_b < s_b) { - uint16_t a = A[i_a]; - uint16_t b = B[i_b]; - if (b < a) { - i_b++; - } else if (a < b) { - C[count] = a; - count++; - i_a++; - } else { //== - i_a++; - i_b++; - } - } - if (i_a < s_a) { - memmove(C + count, A + i_a, sizeof(uint16_t) * (s_a - i_a)); - count += (int32_t)(s_a - i_a); - } - return count; -} - -#endif // USESSE4 - - - -#ifdef USE_OLD_SKEW_INTERSECT -// TODO: given enough experience with the new skew intersect, drop the old one from the code base. - - -/* Computes the intersection between one small and one large set of uint16_t. - * Stores the result into buffer and return the number of elements. */ -int32_t intersect_skewed_uint16(const uint16_t *small, size_t size_s, - const uint16_t *large, size_t size_l, - uint16_t *buffer) { - size_t pos = 0, idx_l = 0, idx_s = 0; - - if (0 == size_s) { - return 0; - } - - uint16_t val_l = large[idx_l], val_s = small[idx_s]; - - while (true) { - if (val_l < val_s) { - idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); - if (idx_l == size_l) break; - val_l = large[idx_l]; - } else if (val_s < val_l) { - idx_s++; - if (idx_s == size_s) break; - val_s = small[idx_s]; - } else { - buffer[pos++] = val_s; - idx_s++; - if (idx_s == size_s) break; - val_s = small[idx_s]; - idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); - if (idx_l == size_l) break; - val_l = large[idx_l]; - } - } - - return (int32_t)pos; -} -#else // USE_OLD_SKEW_INTERSECT - - -/** -* Branchless binary search going after 4 values at once. -* Assumes that array is sorted. -* You have that array[*index1] >= target1, array[*index12] >= target2, ... -* except when *index1 = n, in which case you know that all values in array are -* smaller than target1, and so forth. -* It has logarithmic complexity. -*/ -static void binarySearch4(const uint16_t *array, int32_t n, uint16_t target1, - uint16_t target2, uint16_t target3, uint16_t target4, - int32_t *index1, int32_t *index2, int32_t *index3, - int32_t *index4) { - const uint16_t *base1 = array; - const uint16_t *base2 = array; - const uint16_t *base3 = array; - const uint16_t *base4 = array; - if (n == 0) - return; - while (n > 1) { - int32_t half = n >> 1; - base1 = (base1[half] < target1) ? &base1[half] : base1; - base2 = (base2[half] < target2) ? &base2[half] : base2; - base3 = (base3[half] < target3) ? &base3[half] : base3; - base4 = (base4[half] < target4) ? &base4[half] : base4; - n -= half; - } - *index1 = (int32_t)((*base1 < target1) + base1 - array); - *index2 = (int32_t)((*base2 < target2) + base2 - array); - *index3 = (int32_t)((*base3 < target3) + base3 - array); - *index4 = (int32_t)((*base4 < target4) + base4 - array); -} - -/** -* Branchless binary search going after 2 values at once. -* Assumes that array is sorted. -* You have that array[*index1] >= target1, array[*index12] >= target2. -* except when *index1 = n, in which case you know that all values in array are -* smaller than target1, and so forth. -* It has logarithmic complexity. -*/ -static void binarySearch2(const uint16_t *array, int32_t n, uint16_t target1, - uint16_t target2, int32_t *index1, int32_t *index2) { - const uint16_t *base1 = array; - const uint16_t *base2 = array; - if (n == 0) - return; - while (n > 1) { - int32_t half = n >> 1; - base1 = (base1[half] < target1) ? &base1[half] : base1; - base2 = (base2[half] < target2) ? &base2[half] : base2; - n -= half; - } - *index1 = (int32_t)((*base1 < target1) + base1 - array); - *index2 = (int32_t)((*base2 < target2) + base2 - array); -} - -/* Computes the intersection between one small and one large set of uint16_t. - * Stores the result into buffer and return the number of elements. - * Processes the small set in blocks of 4 values calling binarySearch4 - * and binarySearch2. This approach can be slightly superior to a conventional - * galloping search in some instances. - */ -int32_t intersect_skewed_uint16(const uint16_t *small, size_t size_s, - const uint16_t *large, size_t size_l, - uint16_t *buffer) { - size_t pos = 0, idx_l = 0, idx_s = 0; - - if (0 == size_s) { - return 0; - } - int32_t index1 = 0, index2 = 0, index3 = 0, index4 = 0; - while ((idx_s + 4 <= size_s) && (idx_l < size_l)) { - uint16_t target1 = small[idx_s]; - uint16_t target2 = small[idx_s + 1]; - uint16_t target3 = small[idx_s + 2]; - uint16_t target4 = small[idx_s + 3]; - binarySearch4(large + idx_l, (int32_t)(size_l - idx_l), target1, target2, target3, - target4, &index1, &index2, &index3, &index4); - if ((index1 + idx_l < size_l) && (large[idx_l + index1] == target1)) { - buffer[pos++] = target1; - } - if ((index2 + idx_l < size_l) && (large[idx_l + index2] == target2)) { - buffer[pos++] = target2; - } - if ((index3 + idx_l < size_l) && (large[idx_l + index3] == target3)) { - buffer[pos++] = target3; - } - if ((index4 + idx_l < size_l) && (large[idx_l + index4] == target4)) { - buffer[pos++] = target4; - } - idx_s += 4; - idx_l += index1; - } - if ((idx_s + 2 <= size_s) && (idx_l < size_l)) { - uint16_t target1 = small[idx_s]; - uint16_t target2 = small[idx_s + 1]; - binarySearch2(large + idx_l, (int32_t)(size_l - idx_l), target1, target2, &index1, - &index2); - if ((index1 + idx_l < size_l) && (large[idx_l + index1] == target1)) { - buffer[pos++] = target1; - } - if ((index2 + idx_l < size_l) && (large[idx_l + index2] == target2)) { - buffer[pos++] = target2; - } - idx_s += 2; - idx_l += index1; - } - if ((idx_s < size_s) && (idx_l < size_l)) { - uint16_t val_s = small[idx_s]; - int32_t index = binarySearch(large + idx_l, (int32_t)(size_l - idx_l), val_s); - if (index >= 0) - buffer[pos++] = val_s; - } - return (int32_t)pos; -} - - -#endif //USE_OLD_SKEW_INTERSECT - - -// TODO: this could be accelerated, possibly, by using binarySearch4 as above. -int32_t intersect_skewed_uint16_cardinality(const uint16_t *small, - size_t size_s, - const uint16_t *large, - size_t size_l) { - size_t pos = 0, idx_l = 0, idx_s = 0; - - if (0 == size_s) { - return 0; - } - - uint16_t val_l = large[idx_l], val_s = small[idx_s]; - - while (true) { - if (val_l < val_s) { - idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); - if (idx_l == size_l) break; - val_l = large[idx_l]; - } else if (val_s < val_l) { - idx_s++; - if (idx_s == size_s) break; - val_s = small[idx_s]; - } else { - pos++; - idx_s++; - if (idx_s == size_s) break; - val_s = small[idx_s]; - idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); - if (idx_l == size_l) break; - val_l = large[idx_l]; - } - } - - return (int32_t)pos; -} - -bool intersect_skewed_uint16_nonempty(const uint16_t *small, size_t size_s, - const uint16_t *large, size_t size_l) { - size_t idx_l = 0, idx_s = 0; - - if (0 == size_s) { - return false; - } - - uint16_t val_l = large[idx_l], val_s = small[idx_s]; - - while (true) { - if (val_l < val_s) { - idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); - if (idx_l == size_l) break; - val_l = large[idx_l]; - } else if (val_s < val_l) { - idx_s++; - if (idx_s == size_s) break; - val_s = small[idx_s]; - } else { - return true; - } - } - - return false; -} - -/** - * Generic intersection function. - */ -int32_t intersect_uint16(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB, uint16_t *out) { - const uint16_t *initout = out; - if (lenA == 0 || lenB == 0) return 0; - const uint16_t *endA = A + lenA; - const uint16_t *endB = B + lenB; - - while (1) { - while (*A < *B) { - SKIP_FIRST_COMPARE: - if (++A == endA) return (int32_t)(out - initout); - } - while (*A > *B) { - if (++B == endB) return (int32_t)(out - initout); - } - if (*A == *B) { - *out++ = *A; - if (++A == endA || ++B == endB) return (int32_t)(out - initout); - } else { - goto SKIP_FIRST_COMPARE; - } - } - return (int32_t)(out - initout); // NOTREACHED -} - -int32_t intersect_uint16_cardinality(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB) { - int32_t answer = 0; - if (lenA == 0 || lenB == 0) return 0; - const uint16_t *endA = A + lenA; - const uint16_t *endB = B + lenB; - - while (1) { - while (*A < *B) { - SKIP_FIRST_COMPARE: - if (++A == endA) return answer; - } - while (*A > *B) { - if (++B == endB) return answer; - } - if (*A == *B) { - ++answer; - if (++A == endA || ++B == endB) return answer; - } else { - goto SKIP_FIRST_COMPARE; - } - } - return answer; // NOTREACHED -} - - -bool intersect_uint16_nonempty(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB) { - if (lenA == 0 || lenB == 0) return 0; - const uint16_t *endA = A + lenA; - const uint16_t *endB = B + lenB; - - while (1) { - while (*A < *B) { - SKIP_FIRST_COMPARE: - if (++A == endA) return false; - } - while (*A > *B) { - if (++B == endB) return false; - } - if (*A == *B) { - return true; - } else { - goto SKIP_FIRST_COMPARE; - } - } - return false; // NOTREACHED -} - - - -/** - * Generic intersection function. - */ -size_t intersection_uint32(const uint32_t *A, const size_t lenA, - const uint32_t *B, const size_t lenB, - uint32_t *out) { - const uint32_t *initout = out; - if (lenA == 0 || lenB == 0) return 0; - const uint32_t *endA = A + lenA; - const uint32_t *endB = B + lenB; - - while (1) { - while (*A < *B) { - SKIP_FIRST_COMPARE: - if (++A == endA) return (out - initout); - } - while (*A > *B) { - if (++B == endB) return (out - initout); - } - if (*A == *B) { - *out++ = *A; - if (++A == endA || ++B == endB) return (out - initout); - } else { - goto SKIP_FIRST_COMPARE; - } - } - return (out - initout); // NOTREACHED -} - -size_t intersection_uint32_card(const uint32_t *A, const size_t lenA, - const uint32_t *B, const size_t lenB) { - if (lenA == 0 || lenB == 0) return 0; - size_t card = 0; - const uint32_t *endA = A + lenA; - const uint32_t *endB = B + lenB; - - while (1) { - while (*A < *B) { - SKIP_FIRST_COMPARE: - if (++A == endA) return card; - } - while (*A > *B) { - if (++B == endB) return card; - } - if (*A == *B) { - card++; - if (++A == endA || ++B == endB) return card; - } else { - goto SKIP_FIRST_COMPARE; - } - } - return card; // NOTREACHED -} - -// can one vectorize the computation of the union? (Update: Yes! See -// union_vector16). - -size_t union_uint16(const uint16_t *set_1, size_t size_1, const uint16_t *set_2, - size_t size_2, uint16_t *buffer) { - size_t pos = 0, idx_1 = 0, idx_2 = 0; - - if (0 == size_2) { - memmove(buffer, set_1, size_1 * sizeof(uint16_t)); - return size_1; - } - if (0 == size_1) { - memmove(buffer, set_2, size_2 * sizeof(uint16_t)); - return size_2; - } - - uint16_t val_1 = set_1[idx_1], val_2 = set_2[idx_2]; - - while (true) { - if (val_1 < val_2) { - buffer[pos++] = val_1; - ++idx_1; - if (idx_1 >= size_1) break; - val_1 = set_1[idx_1]; - } else if (val_2 < val_1) { - buffer[pos++] = val_2; - ++idx_2; - if (idx_2 >= size_2) break; - val_2 = set_2[idx_2]; - } else { - buffer[pos++] = val_1; - ++idx_1; - ++idx_2; - if (idx_1 >= size_1 || idx_2 >= size_2) break; - val_1 = set_1[idx_1]; - val_2 = set_2[idx_2]; - } - } - - if (idx_1 < size_1) { - const size_t n_elems = size_1 - idx_1; - memmove(buffer + pos, set_1 + idx_1, n_elems * sizeof(uint16_t)); - pos += n_elems; - } else if (idx_2 < size_2) { - const size_t n_elems = size_2 - idx_2; - memmove(buffer + pos, set_2 + idx_2, n_elems * sizeof(uint16_t)); - pos += n_elems; - } - - return pos; -} - -int difference_uint16(const uint16_t *a1, int length1, const uint16_t *a2, - int length2, uint16_t *a_out) { - int out_card = 0; - int k1 = 0, k2 = 0; - if (length1 == 0) return 0; - if (length2 == 0) { - if (a1 != a_out) memcpy(a_out, a1, sizeof(uint16_t) * length1); - return length1; - } - uint16_t s1 = a1[k1]; - uint16_t s2 = a2[k2]; - while (true) { - if (s1 < s2) { - a_out[out_card++] = s1; - ++k1; - if (k1 >= length1) { - break; - } - s1 = a1[k1]; - } else if (s1 == s2) { - ++k1; - ++k2; - if (k1 >= length1) { - break; - } - if (k2 >= length2) { - memmove(a_out + out_card, a1 + k1, - sizeof(uint16_t) * (length1 - k1)); - return out_card + length1 - k1; - } - s1 = a1[k1]; - s2 = a2[k2]; - } else { // if (val1>val2) - ++k2; - if (k2 >= length2) { - memmove(a_out + out_card, a1 + k1, - sizeof(uint16_t) * (length1 - k1)); - return out_card + length1 - k1; - } - s2 = a2[k2]; - } - } - return out_card; -} - -int32_t xor_uint16(const uint16_t *array_1, int32_t card_1, - const uint16_t *array_2, int32_t card_2, uint16_t *out) { - int32_t pos1 = 0, pos2 = 0, pos_out = 0; - while (pos1 < card_1 && pos2 < card_2) { - const uint16_t v1 = array_1[pos1]; - const uint16_t v2 = array_2[pos2]; - if (v1 == v2) { - ++pos1; - ++pos2; - continue; - } - if (v1 < v2) { - out[pos_out++] = v1; - ++pos1; - } else { - out[pos_out++] = v2; - ++pos2; - } - } - if (pos1 < card_1) { - const size_t n_elems = card_1 - pos1; - memcpy(out + pos_out, array_1 + pos1, n_elems * sizeof(uint16_t)); - pos_out += (int32_t)n_elems; - } else if (pos2 < card_2) { - const size_t n_elems = card_2 - pos2; - memcpy(out + pos_out, array_2 + pos2, n_elems * sizeof(uint16_t)); - pos_out += (int32_t)n_elems; - } - return pos_out; -} - -#ifdef USESSE4 - -/*** - * start of the SIMD 16-bit union code - * - */ - -// Assuming that vInput1 and vInput2 are sorted, produces a sorted output going -// from vecMin all the way to vecMax -// developed originally for merge sort using SIMD instructions. -// Standard merge. See, e.g., Inoue and Taura, SIMD- and Cache-Friendly -// Algorithm for Sorting an Array of Structures -static inline void sse_merge(const __m128i *vInput1, - const __m128i *vInput2, // input 1 & 2 - __m128i *vecMin, __m128i *vecMax) { // output - __m128i vecTmp; - vecTmp = _mm_min_epu16(*vInput1, *vInput2); - *vecMax = _mm_max_epu16(*vInput1, *vInput2); - vecTmp = _mm_alignr_epi8(vecTmp, vecTmp, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); - *vecMin = _mm_min_epu16(vecTmp, *vecMax); - *vecMax = _mm_max_epu16(vecTmp, *vecMax); - *vecMin = _mm_alignr_epi8(*vecMin, *vecMin, 2); -} - -// used by store_unique, generated by simdunion.py -static uint8_t uniqshuf[] = { - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, - 0xc, 0xd, 0xe, 0xf, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xa, 0xb, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0xa, 0xb, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xa, 0xb, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, - 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, - 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x8, 0x9, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0xc, 0xd, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0xc, 0xd, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xc, 0xd, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, - 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x8, 0x9, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x8, 0x9, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, 0xa, 0xb, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x6, 0x7, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0xa, 0xb, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xa, 0xb, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x8, 0x9, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0xe, 0xf, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xe, 0xf, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xe, 0xf, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, - 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x8, 0x9, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0xa, 0xb, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0xa, 0xb, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xa, 0xb, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, - 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, 0x8, 0x9, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0x8, 0x9, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x8, 0x9, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x8, 0x9, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x8, 0x9, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x6, 0x7, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0xc, 0xd, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0xc, 0xd, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xc, 0xd, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xa, 0xb, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x8, 0x9, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xa, 0xb, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0xa, 0xb, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xa, 0xb, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xa, 0xb, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0x6, 0x7, 0x8, 0x9, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x6, 0x7, - 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, 0x8, 0x9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x6, 0x7, - 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x6, 0x7, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x8, 0x9, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, - 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x4, 0x5, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x8, 0x9, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x8, 0x9, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, - 0x6, 0x7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x4, 0x5, 0x6, 0x7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, 0x6, 0x7, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0x6, 0x7, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x6, 0x7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x6, 0x7, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x2, 0x3, - 0x4, 0x5, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x2, 0x3, 0x4, 0x5, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0x4, 0x5, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x4, 0x5, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0x0, 0x1, 0x2, 0x3, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0x2, 0x3, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0, 0x1, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, - 0xFF, 0xFF, 0xFF, 0xFF}; - -// write vector new, while omitting repeated values assuming that previously -// written vector was "old" -static inline int store_unique(__m128i old, __m128i newval, uint16_t *output) { - __m128i vecTmp = _mm_alignr_epi8(newval, old, 16 - 2); - // lots of high latency instructions follow (optimize?) - int M = _mm_movemask_epi8( - _mm_packs_epi16(_mm_cmpeq_epi16(vecTmp, newval), _mm_setzero_si128())); - int numberofnewvalues = 8 - _mm_popcnt_u32(M); - __m128i key = _mm_lddqu_si128((const __m128i *)uniqshuf + M); - __m128i val = _mm_shuffle_epi8(newval, key); - _mm_storeu_si128((__m128i *)output, val); - return numberofnewvalues; -} - -// working in-place, this function overwrites the repeated values -// could be avoided? -static inline uint32_t unique(uint16_t *out, uint32_t len) { - uint32_t pos = 1; - for (uint32_t i = 1; i < len; ++i) { - if (out[i] != out[i - 1]) { - out[pos++] = out[i]; - } - } - return pos; -} - -// use with qsort, could be avoided -static int uint16_compare(const void *a, const void *b) { - return (*(uint16_t *)a - *(uint16_t *)b); -} - -// a one-pass SSE union algorithm -uint32_t union_vector16(const uint16_t *__restrict__ array1, uint32_t length1, - const uint16_t *__restrict__ array2, uint32_t length2, - uint16_t *__restrict__ output) { - if ((length1 < 8) || (length2 < 8)) { - return (uint32_t)union_uint16(array1, length1, array2, length2, output); - } - __m128i vA, vB, V, vecMin, vecMax; - __m128i laststore; - uint16_t *initoutput = output; - uint32_t len1 = length1 / 8; - uint32_t len2 = length2 / 8; - uint32_t pos1 = 0; - uint32_t pos2 = 0; - // we start the machine - vA = _mm_lddqu_si128((const __m128i *)array1 + pos1); - pos1++; - vB = _mm_lddqu_si128((const __m128i *)array2 + pos2); - pos2++; - sse_merge(&vA, &vB, &vecMin, &vecMax); - laststore = _mm_set1_epi16(-1); - output += store_unique(laststore, vecMin, output); - laststore = vecMin; - if ((pos1 < len1) && (pos2 < len2)) { - uint16_t curA, curB; - curA = array1[8 * pos1]; - curB = array2[8 * pos2]; - while (true) { - if (curA <= curB) { - V = _mm_lddqu_si128((const __m128i *)array1 + pos1); - pos1++; - if (pos1 < len1) { - curA = array1[8 * pos1]; - } else { - break; - } - } else { - V = _mm_lddqu_si128((const __m128i *)array2 + pos2); - pos2++; - if (pos2 < len2) { - curB = array2[8 * pos2]; - } else { - break; - } - } - sse_merge(&V, &vecMax, &vecMin, &vecMax); - output += store_unique(laststore, vecMin, output); - laststore = vecMin; - } - sse_merge(&V, &vecMax, &vecMin, &vecMax); - output += store_unique(laststore, vecMin, output); - laststore = vecMin; - } - // we finish the rest off using a scalar algorithm - // could be improved? - // - // copy the small end on a tmp buffer - uint32_t len = (uint32_t)(output - initoutput); - uint16_t buffer[16]; - uint32_t leftoversize = store_unique(laststore, vecMax, buffer); - if (pos1 == len1) { - memcpy(buffer + leftoversize, array1 + 8 * pos1, - (length1 - 8 * len1) * sizeof(uint16_t)); - leftoversize += length1 - 8 * len1; - qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); - - leftoversize = unique(buffer, leftoversize); - len += (uint32_t)union_uint16(buffer, leftoversize, array2 + 8 * pos2, - length2 - 8 * pos2, output); - } else { - memcpy(buffer + leftoversize, array2 + 8 * pos2, - (length2 - 8 * len2) * sizeof(uint16_t)); - leftoversize += length2 - 8 * len2; - qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); - leftoversize = unique(buffer, leftoversize); - len += (uint32_t)union_uint16(buffer, leftoversize, array1 + 8 * pos1, - length1 - 8 * pos1, output); - } - return len; -} - -/** - * End of the SIMD 16-bit union code - * - */ - -/** - * Start of SIMD 16-bit XOR code - */ - -// write vector new, while omitting repeated values assuming that previously -// written vector was "old" -static inline int store_unique_xor(__m128i old, __m128i newval, - uint16_t *output) { - __m128i vecTmp1 = _mm_alignr_epi8(newval, old, 16 - 4); - __m128i vecTmp2 = _mm_alignr_epi8(newval, old, 16 - 2); - __m128i equalleft = _mm_cmpeq_epi16(vecTmp2, vecTmp1); - __m128i equalright = _mm_cmpeq_epi16(vecTmp2, newval); - __m128i equalleftoright = _mm_or_si128(equalleft, equalright); - int M = _mm_movemask_epi8( - _mm_packs_epi16(equalleftoright, _mm_setzero_si128())); - int numberofnewvalues = 8 - _mm_popcnt_u32(M); - __m128i key = _mm_lddqu_si128((const __m128i *)uniqshuf + M); - __m128i val = _mm_shuffle_epi8(vecTmp2, key); - _mm_storeu_si128((__m128i *)output, val); - return numberofnewvalues; -} - -// working in-place, this function overwrites the repeated values -// could be avoided? Warning: assumes len > 0 -static inline uint32_t unique_xor(uint16_t *out, uint32_t len) { - uint32_t pos = 1; - for (uint32_t i = 1; i < len; ++i) { - if (out[i] != out[i - 1]) { - out[pos++] = out[i]; - } else - pos--; // if it is identical to previous, delete it - } - return pos; -} - -// a one-pass SSE xor algorithm -uint32_t xor_vector16(const uint16_t *__restrict__ array1, uint32_t length1, - const uint16_t *__restrict__ array2, uint32_t length2, - uint16_t *__restrict__ output) { - if ((length1 < 8) || (length2 < 8)) { - return xor_uint16(array1, length1, array2, length2, output); - } - __m128i vA, vB, V, vecMin, vecMax; - __m128i laststore; - uint16_t *initoutput = output; - uint32_t len1 = length1 / 8; - uint32_t len2 = length2 / 8; - uint32_t pos1 = 0; - uint32_t pos2 = 0; - // we start the machine - vA = _mm_lddqu_si128((const __m128i *)array1 + pos1); - pos1++; - vB = _mm_lddqu_si128((const __m128i *)array2 + pos2); - pos2++; - sse_merge(&vA, &vB, &vecMin, &vecMax); - laststore = _mm_set1_epi16(-1); - uint16_t buffer[17]; - output += store_unique_xor(laststore, vecMin, output); - - laststore = vecMin; - if ((pos1 < len1) && (pos2 < len2)) { - uint16_t curA, curB; - curA = array1[8 * pos1]; - curB = array2[8 * pos2]; - while (true) { - if (curA <= curB) { - V = _mm_lddqu_si128((const __m128i *)array1 + pos1); - pos1++; - if (pos1 < len1) { - curA = array1[8 * pos1]; - } else { - break; - } - } else { - V = _mm_lddqu_si128((const __m128i *)array2 + pos2); - pos2++; - if (pos2 < len2) { - curB = array2[8 * pos2]; - } else { - break; - } - } - sse_merge(&V, &vecMax, &vecMin, &vecMax); - // conditionally stores the last value of laststore as well as all - // but the - // last value of vecMin - output += store_unique_xor(laststore, vecMin, output); - laststore = vecMin; - } - sse_merge(&V, &vecMax, &vecMin, &vecMax); - // conditionally stores the last value of laststore as well as all but - // the - // last value of vecMin - output += store_unique_xor(laststore, vecMin, output); - laststore = vecMin; - } - uint32_t len = (uint32_t)(output - initoutput); - - // we finish the rest off using a scalar algorithm - // could be improved? - // conditionally stores the last value of laststore as well as all but the - // last value of vecMax, - // we store to "buffer" - int leftoversize = store_unique_xor(laststore, vecMax, buffer); - uint16_t vec7 = _mm_extract_epi16(vecMax, 7); - uint16_t vec6 = _mm_extract_epi16(vecMax, 6); - if (vec7 != vec6) buffer[leftoversize++] = vec7; - if (pos1 == len1) { - memcpy(buffer + leftoversize, array1 + 8 * pos1, - (length1 - 8 * len1) * sizeof(uint16_t)); - leftoversize += length1 - 8 * len1; - if (leftoversize == 0) { // trivial case - memcpy(output, array2 + 8 * pos2, - (length2 - 8 * pos2) * sizeof(uint16_t)); - len += (length2 - 8 * pos2); - } else { - qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); - leftoversize = unique_xor(buffer, leftoversize); - len += xor_uint16(buffer, leftoversize, array2 + 8 * pos2, - length2 - 8 * pos2, output); - } - } else { - memcpy(buffer + leftoversize, array2 + 8 * pos2, - (length2 - 8 * len2) * sizeof(uint16_t)); - leftoversize += length2 - 8 * len2; - if (leftoversize == 0) { // trivial case - memcpy(output, array1 + 8 * pos1, - (length1 - 8 * pos1) * sizeof(uint16_t)); - len += (length1 - 8 * pos1); - } else { - qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); - leftoversize = unique_xor(buffer, leftoversize); - len += xor_uint16(buffer, leftoversize, array1 + 8 * pos1, - length1 - 8 * pos1, output); - } - } - return len; -} - -/** - * End of SIMD 16-bit XOR code - */ - -#endif // USESSE4 - -size_t union_uint32(const uint32_t *set_1, size_t size_1, const uint32_t *set_2, - size_t size_2, uint32_t *buffer) { - size_t pos = 0, idx_1 = 0, idx_2 = 0; - - if (0 == size_2) { - memmove(buffer, set_1, size_1 * sizeof(uint32_t)); - return size_1; - } - if (0 == size_1) { - memmove(buffer, set_2, size_2 * sizeof(uint32_t)); - return size_2; - } - - uint32_t val_1 = set_1[idx_1], val_2 = set_2[idx_2]; - - while (true) { - if (val_1 < val_2) { - buffer[pos++] = val_1; - ++idx_1; - if (idx_1 >= size_1) break; - val_1 = set_1[idx_1]; - } else if (val_2 < val_1) { - buffer[pos++] = val_2; - ++idx_2; - if (idx_2 >= size_2) break; - val_2 = set_2[idx_2]; - } else { - buffer[pos++] = val_1; - ++idx_1; - ++idx_2; - if (idx_1 >= size_1 || idx_2 >= size_2) break; - val_1 = set_1[idx_1]; - val_2 = set_2[idx_2]; - } - } - - if (idx_1 < size_1) { - const size_t n_elems = size_1 - idx_1; - memmove(buffer + pos, set_1 + idx_1, n_elems * sizeof(uint32_t)); - pos += n_elems; - } else if (idx_2 < size_2) { - const size_t n_elems = size_2 - idx_2; - memmove(buffer + pos, set_2 + idx_2, n_elems * sizeof(uint32_t)); - pos += n_elems; - } - - return pos; -} - -size_t union_uint32_card(const uint32_t *set_1, size_t size_1, - const uint32_t *set_2, size_t size_2) { - size_t pos = 0, idx_1 = 0, idx_2 = 0; - - if (0 == size_2) { - return size_1; - } - if (0 == size_1) { - return size_2; - } - - uint32_t val_1 = set_1[idx_1], val_2 = set_2[idx_2]; - - while (true) { - if (val_1 < val_2) { - ++idx_1; - ++pos; - if (idx_1 >= size_1) break; - val_1 = set_1[idx_1]; - } else if (val_2 < val_1) { - ++idx_2; - ++pos; - if (idx_2 >= size_2) break; - val_2 = set_2[idx_2]; - } else { - ++idx_1; - ++idx_2; - ++pos; - if (idx_1 >= size_1 || idx_2 >= size_2) break; - val_1 = set_1[idx_1]; - val_2 = set_2[idx_2]; - } - } - - if (idx_1 < size_1) { - const size_t n_elems = size_1 - idx_1; - pos += n_elems; - } else if (idx_2 < size_2) { - const size_t n_elems = size_2 - idx_2; - pos += n_elems; - } - return pos; -} - - - -size_t fast_union_uint16(const uint16_t *set_1, size_t size_1, const uint16_t *set_2, - size_t size_2, uint16_t *buffer) { -#ifdef ROARING_VECTOR_OPERATIONS_ENABLED - // compute union with smallest array first - if (size_1 < size_2) { - return union_vector16(set_1, (uint32_t)size_1, - set_2, (uint32_t)size_2, buffer); - } else { - return union_vector16(set_2, (uint32_t)size_2, - set_1, (uint32_t)size_1, buffer); - } -#else - // compute union with smallest array first - if (size_1 < size_2) { - return union_uint16( - set_1, size_1, set_2, size_2, buffer); - } else { - return union_uint16( - set_2, size_2, set_1, size_1, buffer); - } -#endif -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/array_util.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/bitset_util.c */ -#include -#include -#include -#include -#include - - -#ifdef IS_X64 -static uint8_t lengthTable[256] = { - 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, 1, 2, 2, 3, 2, 3, 3, 4, - 2, 3, 3, 4, 3, 4, 4, 5, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, - 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 1, 2, 2, 3, 2, 3, 3, 4, - 2, 3, 3, 4, 3, 4, 4, 5, 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, - 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, - 4, 5, 5, 6, 5, 6, 6, 7, 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, - 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 2, 3, 3, 4, 3, 4, 4, 5, - 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, - 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, 3, 4, 4, 5, 4, 5, 5, 6, - 4, 5, 5, 6, 5, 6, 6, 7, 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, - 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8}; -#endif - -#ifdef USEAVX -ALIGNED(32) -static uint32_t vecDecodeTable[256][8] = { - {0, 0, 0, 0, 0, 0, 0, 0}, /* 0x00 (00000000) */ - {1, 0, 0, 0, 0, 0, 0, 0}, /* 0x01 (00000001) */ - {2, 0, 0, 0, 0, 0, 0, 0}, /* 0x02 (00000010) */ - {1, 2, 0, 0, 0, 0, 0, 0}, /* 0x03 (00000011) */ - {3, 0, 0, 0, 0, 0, 0, 0}, /* 0x04 (00000100) */ - {1, 3, 0, 0, 0, 0, 0, 0}, /* 0x05 (00000101) */ - {2, 3, 0, 0, 0, 0, 0, 0}, /* 0x06 (00000110) */ - {1, 2, 3, 0, 0, 0, 0, 0}, /* 0x07 (00000111) */ - {4, 0, 0, 0, 0, 0, 0, 0}, /* 0x08 (00001000) */ - {1, 4, 0, 0, 0, 0, 0, 0}, /* 0x09 (00001001) */ - {2, 4, 0, 0, 0, 0, 0, 0}, /* 0x0A (00001010) */ - {1, 2, 4, 0, 0, 0, 0, 0}, /* 0x0B (00001011) */ - {3, 4, 0, 0, 0, 0, 0, 0}, /* 0x0C (00001100) */ - {1, 3, 4, 0, 0, 0, 0, 0}, /* 0x0D (00001101) */ - {2, 3, 4, 0, 0, 0, 0, 0}, /* 0x0E (00001110) */ - {1, 2, 3, 4, 0, 0, 0, 0}, /* 0x0F (00001111) */ - {5, 0, 0, 0, 0, 0, 0, 0}, /* 0x10 (00010000) */ - {1, 5, 0, 0, 0, 0, 0, 0}, /* 0x11 (00010001) */ - {2, 5, 0, 0, 0, 0, 0, 0}, /* 0x12 (00010010) */ - {1, 2, 5, 0, 0, 0, 0, 0}, /* 0x13 (00010011) */ - {3, 5, 0, 0, 0, 0, 0, 0}, /* 0x14 (00010100) */ - {1, 3, 5, 0, 0, 0, 0, 0}, /* 0x15 (00010101) */ - {2, 3, 5, 0, 0, 0, 0, 0}, /* 0x16 (00010110) */ - {1, 2, 3, 5, 0, 0, 0, 0}, /* 0x17 (00010111) */ - {4, 5, 0, 0, 0, 0, 0, 0}, /* 0x18 (00011000) */ - {1, 4, 5, 0, 0, 0, 0, 0}, /* 0x19 (00011001) */ - {2, 4, 5, 0, 0, 0, 0, 0}, /* 0x1A (00011010) */ - {1, 2, 4, 5, 0, 0, 0, 0}, /* 0x1B (00011011) */ - {3, 4, 5, 0, 0, 0, 0, 0}, /* 0x1C (00011100) */ - {1, 3, 4, 5, 0, 0, 0, 0}, /* 0x1D (00011101) */ - {2, 3, 4, 5, 0, 0, 0, 0}, /* 0x1E (00011110) */ - {1, 2, 3, 4, 5, 0, 0, 0}, /* 0x1F (00011111) */ - {6, 0, 0, 0, 0, 0, 0, 0}, /* 0x20 (00100000) */ - {1, 6, 0, 0, 0, 0, 0, 0}, /* 0x21 (00100001) */ - {2, 6, 0, 0, 0, 0, 0, 0}, /* 0x22 (00100010) */ - {1, 2, 6, 0, 0, 0, 0, 0}, /* 0x23 (00100011) */ - {3, 6, 0, 0, 0, 0, 0, 0}, /* 0x24 (00100100) */ - {1, 3, 6, 0, 0, 0, 0, 0}, /* 0x25 (00100101) */ - {2, 3, 6, 0, 0, 0, 0, 0}, /* 0x26 (00100110) */ - {1, 2, 3, 6, 0, 0, 0, 0}, /* 0x27 (00100111) */ - {4, 6, 0, 0, 0, 0, 0, 0}, /* 0x28 (00101000) */ - {1, 4, 6, 0, 0, 0, 0, 0}, /* 0x29 (00101001) */ - {2, 4, 6, 0, 0, 0, 0, 0}, /* 0x2A (00101010) */ - {1, 2, 4, 6, 0, 0, 0, 0}, /* 0x2B (00101011) */ - {3, 4, 6, 0, 0, 0, 0, 0}, /* 0x2C (00101100) */ - {1, 3, 4, 6, 0, 0, 0, 0}, /* 0x2D (00101101) */ - {2, 3, 4, 6, 0, 0, 0, 0}, /* 0x2E (00101110) */ - {1, 2, 3, 4, 6, 0, 0, 0}, /* 0x2F (00101111) */ - {5, 6, 0, 0, 0, 0, 0, 0}, /* 0x30 (00110000) */ - {1, 5, 6, 0, 0, 0, 0, 0}, /* 0x31 (00110001) */ - {2, 5, 6, 0, 0, 0, 0, 0}, /* 0x32 (00110010) */ - {1, 2, 5, 6, 0, 0, 0, 0}, /* 0x33 (00110011) */ - {3, 5, 6, 0, 0, 0, 0, 0}, /* 0x34 (00110100) */ - {1, 3, 5, 6, 0, 0, 0, 0}, /* 0x35 (00110101) */ - {2, 3, 5, 6, 0, 0, 0, 0}, /* 0x36 (00110110) */ - {1, 2, 3, 5, 6, 0, 0, 0}, /* 0x37 (00110111) */ - {4, 5, 6, 0, 0, 0, 0, 0}, /* 0x38 (00111000) */ - {1, 4, 5, 6, 0, 0, 0, 0}, /* 0x39 (00111001) */ - {2, 4, 5, 6, 0, 0, 0, 0}, /* 0x3A (00111010) */ - {1, 2, 4, 5, 6, 0, 0, 0}, /* 0x3B (00111011) */ - {3, 4, 5, 6, 0, 0, 0, 0}, /* 0x3C (00111100) */ - {1, 3, 4, 5, 6, 0, 0, 0}, /* 0x3D (00111101) */ - {2, 3, 4, 5, 6, 0, 0, 0}, /* 0x3E (00111110) */ - {1, 2, 3, 4, 5, 6, 0, 0}, /* 0x3F (00111111) */ - {7, 0, 0, 0, 0, 0, 0, 0}, /* 0x40 (01000000) */ - {1, 7, 0, 0, 0, 0, 0, 0}, /* 0x41 (01000001) */ - {2, 7, 0, 0, 0, 0, 0, 0}, /* 0x42 (01000010) */ - {1, 2, 7, 0, 0, 0, 0, 0}, /* 0x43 (01000011) */ - {3, 7, 0, 0, 0, 0, 0, 0}, /* 0x44 (01000100) */ - {1, 3, 7, 0, 0, 0, 0, 0}, /* 0x45 (01000101) */ - {2, 3, 7, 0, 0, 0, 0, 0}, /* 0x46 (01000110) */ - {1, 2, 3, 7, 0, 0, 0, 0}, /* 0x47 (01000111) */ - {4, 7, 0, 0, 0, 0, 0, 0}, /* 0x48 (01001000) */ - {1, 4, 7, 0, 0, 0, 0, 0}, /* 0x49 (01001001) */ - {2, 4, 7, 0, 0, 0, 0, 0}, /* 0x4A (01001010) */ - {1, 2, 4, 7, 0, 0, 0, 0}, /* 0x4B (01001011) */ - {3, 4, 7, 0, 0, 0, 0, 0}, /* 0x4C (01001100) */ - {1, 3, 4, 7, 0, 0, 0, 0}, /* 0x4D (01001101) */ - {2, 3, 4, 7, 0, 0, 0, 0}, /* 0x4E (01001110) */ - {1, 2, 3, 4, 7, 0, 0, 0}, /* 0x4F (01001111) */ - {5, 7, 0, 0, 0, 0, 0, 0}, /* 0x50 (01010000) */ - {1, 5, 7, 0, 0, 0, 0, 0}, /* 0x51 (01010001) */ - {2, 5, 7, 0, 0, 0, 0, 0}, /* 0x52 (01010010) */ - {1, 2, 5, 7, 0, 0, 0, 0}, /* 0x53 (01010011) */ - {3, 5, 7, 0, 0, 0, 0, 0}, /* 0x54 (01010100) */ - {1, 3, 5, 7, 0, 0, 0, 0}, /* 0x55 (01010101) */ - {2, 3, 5, 7, 0, 0, 0, 0}, /* 0x56 (01010110) */ - {1, 2, 3, 5, 7, 0, 0, 0}, /* 0x57 (01010111) */ - {4, 5, 7, 0, 0, 0, 0, 0}, /* 0x58 (01011000) */ - {1, 4, 5, 7, 0, 0, 0, 0}, /* 0x59 (01011001) */ - {2, 4, 5, 7, 0, 0, 0, 0}, /* 0x5A (01011010) */ - {1, 2, 4, 5, 7, 0, 0, 0}, /* 0x5B (01011011) */ - {3, 4, 5, 7, 0, 0, 0, 0}, /* 0x5C (01011100) */ - {1, 3, 4, 5, 7, 0, 0, 0}, /* 0x5D (01011101) */ - {2, 3, 4, 5, 7, 0, 0, 0}, /* 0x5E (01011110) */ - {1, 2, 3, 4, 5, 7, 0, 0}, /* 0x5F (01011111) */ - {6, 7, 0, 0, 0, 0, 0, 0}, /* 0x60 (01100000) */ - {1, 6, 7, 0, 0, 0, 0, 0}, /* 0x61 (01100001) */ - {2, 6, 7, 0, 0, 0, 0, 0}, /* 0x62 (01100010) */ - {1, 2, 6, 7, 0, 0, 0, 0}, /* 0x63 (01100011) */ - {3, 6, 7, 0, 0, 0, 0, 0}, /* 0x64 (01100100) */ - {1, 3, 6, 7, 0, 0, 0, 0}, /* 0x65 (01100101) */ - {2, 3, 6, 7, 0, 0, 0, 0}, /* 0x66 (01100110) */ - {1, 2, 3, 6, 7, 0, 0, 0}, /* 0x67 (01100111) */ - {4, 6, 7, 0, 0, 0, 0, 0}, /* 0x68 (01101000) */ - {1, 4, 6, 7, 0, 0, 0, 0}, /* 0x69 (01101001) */ - {2, 4, 6, 7, 0, 0, 0, 0}, /* 0x6A (01101010) */ - {1, 2, 4, 6, 7, 0, 0, 0}, /* 0x6B (01101011) */ - {3, 4, 6, 7, 0, 0, 0, 0}, /* 0x6C (01101100) */ - {1, 3, 4, 6, 7, 0, 0, 0}, /* 0x6D (01101101) */ - {2, 3, 4, 6, 7, 0, 0, 0}, /* 0x6E (01101110) */ - {1, 2, 3, 4, 6, 7, 0, 0}, /* 0x6F (01101111) */ - {5, 6, 7, 0, 0, 0, 0, 0}, /* 0x70 (01110000) */ - {1, 5, 6, 7, 0, 0, 0, 0}, /* 0x71 (01110001) */ - {2, 5, 6, 7, 0, 0, 0, 0}, /* 0x72 (01110010) */ - {1, 2, 5, 6, 7, 0, 0, 0}, /* 0x73 (01110011) */ - {3, 5, 6, 7, 0, 0, 0, 0}, /* 0x74 (01110100) */ - {1, 3, 5, 6, 7, 0, 0, 0}, /* 0x75 (01110101) */ - {2, 3, 5, 6, 7, 0, 0, 0}, /* 0x76 (01110110) */ - {1, 2, 3, 5, 6, 7, 0, 0}, /* 0x77 (01110111) */ - {4, 5, 6, 7, 0, 0, 0, 0}, /* 0x78 (01111000) */ - {1, 4, 5, 6, 7, 0, 0, 0}, /* 0x79 (01111001) */ - {2, 4, 5, 6, 7, 0, 0, 0}, /* 0x7A (01111010) */ - {1, 2, 4, 5, 6, 7, 0, 0}, /* 0x7B (01111011) */ - {3, 4, 5, 6, 7, 0, 0, 0}, /* 0x7C (01111100) */ - {1, 3, 4, 5, 6, 7, 0, 0}, /* 0x7D (01111101) */ - {2, 3, 4, 5, 6, 7, 0, 0}, /* 0x7E (01111110) */ - {1, 2, 3, 4, 5, 6, 7, 0}, /* 0x7F (01111111) */ - {8, 0, 0, 0, 0, 0, 0, 0}, /* 0x80 (10000000) */ - {1, 8, 0, 0, 0, 0, 0, 0}, /* 0x81 (10000001) */ - {2, 8, 0, 0, 0, 0, 0, 0}, /* 0x82 (10000010) */ - {1, 2, 8, 0, 0, 0, 0, 0}, /* 0x83 (10000011) */ - {3, 8, 0, 0, 0, 0, 0, 0}, /* 0x84 (10000100) */ - {1, 3, 8, 0, 0, 0, 0, 0}, /* 0x85 (10000101) */ - {2, 3, 8, 0, 0, 0, 0, 0}, /* 0x86 (10000110) */ - {1, 2, 3, 8, 0, 0, 0, 0}, /* 0x87 (10000111) */ - {4, 8, 0, 0, 0, 0, 0, 0}, /* 0x88 (10001000) */ - {1, 4, 8, 0, 0, 0, 0, 0}, /* 0x89 (10001001) */ - {2, 4, 8, 0, 0, 0, 0, 0}, /* 0x8A (10001010) */ - {1, 2, 4, 8, 0, 0, 0, 0}, /* 0x8B (10001011) */ - {3, 4, 8, 0, 0, 0, 0, 0}, /* 0x8C (10001100) */ - {1, 3, 4, 8, 0, 0, 0, 0}, /* 0x8D (10001101) */ - {2, 3, 4, 8, 0, 0, 0, 0}, /* 0x8E (10001110) */ - {1, 2, 3, 4, 8, 0, 0, 0}, /* 0x8F (10001111) */ - {5, 8, 0, 0, 0, 0, 0, 0}, /* 0x90 (10010000) */ - {1, 5, 8, 0, 0, 0, 0, 0}, /* 0x91 (10010001) */ - {2, 5, 8, 0, 0, 0, 0, 0}, /* 0x92 (10010010) */ - {1, 2, 5, 8, 0, 0, 0, 0}, /* 0x93 (10010011) */ - {3, 5, 8, 0, 0, 0, 0, 0}, /* 0x94 (10010100) */ - {1, 3, 5, 8, 0, 0, 0, 0}, /* 0x95 (10010101) */ - {2, 3, 5, 8, 0, 0, 0, 0}, /* 0x96 (10010110) */ - {1, 2, 3, 5, 8, 0, 0, 0}, /* 0x97 (10010111) */ - {4, 5, 8, 0, 0, 0, 0, 0}, /* 0x98 (10011000) */ - {1, 4, 5, 8, 0, 0, 0, 0}, /* 0x99 (10011001) */ - {2, 4, 5, 8, 0, 0, 0, 0}, /* 0x9A (10011010) */ - {1, 2, 4, 5, 8, 0, 0, 0}, /* 0x9B (10011011) */ - {3, 4, 5, 8, 0, 0, 0, 0}, /* 0x9C (10011100) */ - {1, 3, 4, 5, 8, 0, 0, 0}, /* 0x9D (10011101) */ - {2, 3, 4, 5, 8, 0, 0, 0}, /* 0x9E (10011110) */ - {1, 2, 3, 4, 5, 8, 0, 0}, /* 0x9F (10011111) */ - {6, 8, 0, 0, 0, 0, 0, 0}, /* 0xA0 (10100000) */ - {1, 6, 8, 0, 0, 0, 0, 0}, /* 0xA1 (10100001) */ - {2, 6, 8, 0, 0, 0, 0, 0}, /* 0xA2 (10100010) */ - {1, 2, 6, 8, 0, 0, 0, 0}, /* 0xA3 (10100011) */ - {3, 6, 8, 0, 0, 0, 0, 0}, /* 0xA4 (10100100) */ - {1, 3, 6, 8, 0, 0, 0, 0}, /* 0xA5 (10100101) */ - {2, 3, 6, 8, 0, 0, 0, 0}, /* 0xA6 (10100110) */ - {1, 2, 3, 6, 8, 0, 0, 0}, /* 0xA7 (10100111) */ - {4, 6, 8, 0, 0, 0, 0, 0}, /* 0xA8 (10101000) */ - {1, 4, 6, 8, 0, 0, 0, 0}, /* 0xA9 (10101001) */ - {2, 4, 6, 8, 0, 0, 0, 0}, /* 0xAA (10101010) */ - {1, 2, 4, 6, 8, 0, 0, 0}, /* 0xAB (10101011) */ - {3, 4, 6, 8, 0, 0, 0, 0}, /* 0xAC (10101100) */ - {1, 3, 4, 6, 8, 0, 0, 0}, /* 0xAD (10101101) */ - {2, 3, 4, 6, 8, 0, 0, 0}, /* 0xAE (10101110) */ - {1, 2, 3, 4, 6, 8, 0, 0}, /* 0xAF (10101111) */ - {5, 6, 8, 0, 0, 0, 0, 0}, /* 0xB0 (10110000) */ - {1, 5, 6, 8, 0, 0, 0, 0}, /* 0xB1 (10110001) */ - {2, 5, 6, 8, 0, 0, 0, 0}, /* 0xB2 (10110010) */ - {1, 2, 5, 6, 8, 0, 0, 0}, /* 0xB3 (10110011) */ - {3, 5, 6, 8, 0, 0, 0, 0}, /* 0xB4 (10110100) */ - {1, 3, 5, 6, 8, 0, 0, 0}, /* 0xB5 (10110101) */ - {2, 3, 5, 6, 8, 0, 0, 0}, /* 0xB6 (10110110) */ - {1, 2, 3, 5, 6, 8, 0, 0}, /* 0xB7 (10110111) */ - {4, 5, 6, 8, 0, 0, 0, 0}, /* 0xB8 (10111000) */ - {1, 4, 5, 6, 8, 0, 0, 0}, /* 0xB9 (10111001) */ - {2, 4, 5, 6, 8, 0, 0, 0}, /* 0xBA (10111010) */ - {1, 2, 4, 5, 6, 8, 0, 0}, /* 0xBB (10111011) */ - {3, 4, 5, 6, 8, 0, 0, 0}, /* 0xBC (10111100) */ - {1, 3, 4, 5, 6, 8, 0, 0}, /* 0xBD (10111101) */ - {2, 3, 4, 5, 6, 8, 0, 0}, /* 0xBE (10111110) */ - {1, 2, 3, 4, 5, 6, 8, 0}, /* 0xBF (10111111) */ - {7, 8, 0, 0, 0, 0, 0, 0}, /* 0xC0 (11000000) */ - {1, 7, 8, 0, 0, 0, 0, 0}, /* 0xC1 (11000001) */ - {2, 7, 8, 0, 0, 0, 0, 0}, /* 0xC2 (11000010) */ - {1, 2, 7, 8, 0, 0, 0, 0}, /* 0xC3 (11000011) */ - {3, 7, 8, 0, 0, 0, 0, 0}, /* 0xC4 (11000100) */ - {1, 3, 7, 8, 0, 0, 0, 0}, /* 0xC5 (11000101) */ - {2, 3, 7, 8, 0, 0, 0, 0}, /* 0xC6 (11000110) */ - {1, 2, 3, 7, 8, 0, 0, 0}, /* 0xC7 (11000111) */ - {4, 7, 8, 0, 0, 0, 0, 0}, /* 0xC8 (11001000) */ - {1, 4, 7, 8, 0, 0, 0, 0}, /* 0xC9 (11001001) */ - {2, 4, 7, 8, 0, 0, 0, 0}, /* 0xCA (11001010) */ - {1, 2, 4, 7, 8, 0, 0, 0}, /* 0xCB (11001011) */ - {3, 4, 7, 8, 0, 0, 0, 0}, /* 0xCC (11001100) */ - {1, 3, 4, 7, 8, 0, 0, 0}, /* 0xCD (11001101) */ - {2, 3, 4, 7, 8, 0, 0, 0}, /* 0xCE (11001110) */ - {1, 2, 3, 4, 7, 8, 0, 0}, /* 0xCF (11001111) */ - {5, 7, 8, 0, 0, 0, 0, 0}, /* 0xD0 (11010000) */ - {1, 5, 7, 8, 0, 0, 0, 0}, /* 0xD1 (11010001) */ - {2, 5, 7, 8, 0, 0, 0, 0}, /* 0xD2 (11010010) */ - {1, 2, 5, 7, 8, 0, 0, 0}, /* 0xD3 (11010011) */ - {3, 5, 7, 8, 0, 0, 0, 0}, /* 0xD4 (11010100) */ - {1, 3, 5, 7, 8, 0, 0, 0}, /* 0xD5 (11010101) */ - {2, 3, 5, 7, 8, 0, 0, 0}, /* 0xD6 (11010110) */ - {1, 2, 3, 5, 7, 8, 0, 0}, /* 0xD7 (11010111) */ - {4, 5, 7, 8, 0, 0, 0, 0}, /* 0xD8 (11011000) */ - {1, 4, 5, 7, 8, 0, 0, 0}, /* 0xD9 (11011001) */ - {2, 4, 5, 7, 8, 0, 0, 0}, /* 0xDA (11011010) */ - {1, 2, 4, 5, 7, 8, 0, 0}, /* 0xDB (11011011) */ - {3, 4, 5, 7, 8, 0, 0, 0}, /* 0xDC (11011100) */ - {1, 3, 4, 5, 7, 8, 0, 0}, /* 0xDD (11011101) */ - {2, 3, 4, 5, 7, 8, 0, 0}, /* 0xDE (11011110) */ - {1, 2, 3, 4, 5, 7, 8, 0}, /* 0xDF (11011111) */ - {6, 7, 8, 0, 0, 0, 0, 0}, /* 0xE0 (11100000) */ - {1, 6, 7, 8, 0, 0, 0, 0}, /* 0xE1 (11100001) */ - {2, 6, 7, 8, 0, 0, 0, 0}, /* 0xE2 (11100010) */ - {1, 2, 6, 7, 8, 0, 0, 0}, /* 0xE3 (11100011) */ - {3, 6, 7, 8, 0, 0, 0, 0}, /* 0xE4 (11100100) */ - {1, 3, 6, 7, 8, 0, 0, 0}, /* 0xE5 (11100101) */ - {2, 3, 6, 7, 8, 0, 0, 0}, /* 0xE6 (11100110) */ - {1, 2, 3, 6, 7, 8, 0, 0}, /* 0xE7 (11100111) */ - {4, 6, 7, 8, 0, 0, 0, 0}, /* 0xE8 (11101000) */ - {1, 4, 6, 7, 8, 0, 0, 0}, /* 0xE9 (11101001) */ - {2, 4, 6, 7, 8, 0, 0, 0}, /* 0xEA (11101010) */ - {1, 2, 4, 6, 7, 8, 0, 0}, /* 0xEB (11101011) */ - {3, 4, 6, 7, 8, 0, 0, 0}, /* 0xEC (11101100) */ - {1, 3, 4, 6, 7, 8, 0, 0}, /* 0xED (11101101) */ - {2, 3, 4, 6, 7, 8, 0, 0}, /* 0xEE (11101110) */ - {1, 2, 3, 4, 6, 7, 8, 0}, /* 0xEF (11101111) */ - {5, 6, 7, 8, 0, 0, 0, 0}, /* 0xF0 (11110000) */ - {1, 5, 6, 7, 8, 0, 0, 0}, /* 0xF1 (11110001) */ - {2, 5, 6, 7, 8, 0, 0, 0}, /* 0xF2 (11110010) */ - {1, 2, 5, 6, 7, 8, 0, 0}, /* 0xF3 (11110011) */ - {3, 5, 6, 7, 8, 0, 0, 0}, /* 0xF4 (11110100) */ - {1, 3, 5, 6, 7, 8, 0, 0}, /* 0xF5 (11110101) */ - {2, 3, 5, 6, 7, 8, 0, 0}, /* 0xF6 (11110110) */ - {1, 2, 3, 5, 6, 7, 8, 0}, /* 0xF7 (11110111) */ - {4, 5, 6, 7, 8, 0, 0, 0}, /* 0xF8 (11111000) */ - {1, 4, 5, 6, 7, 8, 0, 0}, /* 0xF9 (11111001) */ - {2, 4, 5, 6, 7, 8, 0, 0}, /* 0xFA (11111010) */ - {1, 2, 4, 5, 6, 7, 8, 0}, /* 0xFB (11111011) */ - {3, 4, 5, 6, 7, 8, 0, 0}, /* 0xFC (11111100) */ - {1, 3, 4, 5, 6, 7, 8, 0}, /* 0xFD (11111101) */ - {2, 3, 4, 5, 6, 7, 8, 0}, /* 0xFE (11111110) */ - {1, 2, 3, 4, 5, 6, 7, 8} /* 0xFF (11111111) */ -}; - -#endif // #ifdef USEAVX - -#ifdef IS_X64 -// same as vecDecodeTable but in 16 bits -ALIGNED(32) -static uint16_t vecDecodeTable_uint16[256][8] = { - {0, 0, 0, 0, 0, 0, 0, 0}, /* 0x00 (00000000) */ - {1, 0, 0, 0, 0, 0, 0, 0}, /* 0x01 (00000001) */ - {2, 0, 0, 0, 0, 0, 0, 0}, /* 0x02 (00000010) */ - {1, 2, 0, 0, 0, 0, 0, 0}, /* 0x03 (00000011) */ - {3, 0, 0, 0, 0, 0, 0, 0}, /* 0x04 (00000100) */ - {1, 3, 0, 0, 0, 0, 0, 0}, /* 0x05 (00000101) */ - {2, 3, 0, 0, 0, 0, 0, 0}, /* 0x06 (00000110) */ - {1, 2, 3, 0, 0, 0, 0, 0}, /* 0x07 (00000111) */ - {4, 0, 0, 0, 0, 0, 0, 0}, /* 0x08 (00001000) */ - {1, 4, 0, 0, 0, 0, 0, 0}, /* 0x09 (00001001) */ - {2, 4, 0, 0, 0, 0, 0, 0}, /* 0x0A (00001010) */ - {1, 2, 4, 0, 0, 0, 0, 0}, /* 0x0B (00001011) */ - {3, 4, 0, 0, 0, 0, 0, 0}, /* 0x0C (00001100) */ - {1, 3, 4, 0, 0, 0, 0, 0}, /* 0x0D (00001101) */ - {2, 3, 4, 0, 0, 0, 0, 0}, /* 0x0E (00001110) */ - {1, 2, 3, 4, 0, 0, 0, 0}, /* 0x0F (00001111) */ - {5, 0, 0, 0, 0, 0, 0, 0}, /* 0x10 (00010000) */ - {1, 5, 0, 0, 0, 0, 0, 0}, /* 0x11 (00010001) */ - {2, 5, 0, 0, 0, 0, 0, 0}, /* 0x12 (00010010) */ - {1, 2, 5, 0, 0, 0, 0, 0}, /* 0x13 (00010011) */ - {3, 5, 0, 0, 0, 0, 0, 0}, /* 0x14 (00010100) */ - {1, 3, 5, 0, 0, 0, 0, 0}, /* 0x15 (00010101) */ - {2, 3, 5, 0, 0, 0, 0, 0}, /* 0x16 (00010110) */ - {1, 2, 3, 5, 0, 0, 0, 0}, /* 0x17 (00010111) */ - {4, 5, 0, 0, 0, 0, 0, 0}, /* 0x18 (00011000) */ - {1, 4, 5, 0, 0, 0, 0, 0}, /* 0x19 (00011001) */ - {2, 4, 5, 0, 0, 0, 0, 0}, /* 0x1A (00011010) */ - {1, 2, 4, 5, 0, 0, 0, 0}, /* 0x1B (00011011) */ - {3, 4, 5, 0, 0, 0, 0, 0}, /* 0x1C (00011100) */ - {1, 3, 4, 5, 0, 0, 0, 0}, /* 0x1D (00011101) */ - {2, 3, 4, 5, 0, 0, 0, 0}, /* 0x1E (00011110) */ - {1, 2, 3, 4, 5, 0, 0, 0}, /* 0x1F (00011111) */ - {6, 0, 0, 0, 0, 0, 0, 0}, /* 0x20 (00100000) */ - {1, 6, 0, 0, 0, 0, 0, 0}, /* 0x21 (00100001) */ - {2, 6, 0, 0, 0, 0, 0, 0}, /* 0x22 (00100010) */ - {1, 2, 6, 0, 0, 0, 0, 0}, /* 0x23 (00100011) */ - {3, 6, 0, 0, 0, 0, 0, 0}, /* 0x24 (00100100) */ - {1, 3, 6, 0, 0, 0, 0, 0}, /* 0x25 (00100101) */ - {2, 3, 6, 0, 0, 0, 0, 0}, /* 0x26 (00100110) */ - {1, 2, 3, 6, 0, 0, 0, 0}, /* 0x27 (00100111) */ - {4, 6, 0, 0, 0, 0, 0, 0}, /* 0x28 (00101000) */ - {1, 4, 6, 0, 0, 0, 0, 0}, /* 0x29 (00101001) */ - {2, 4, 6, 0, 0, 0, 0, 0}, /* 0x2A (00101010) */ - {1, 2, 4, 6, 0, 0, 0, 0}, /* 0x2B (00101011) */ - {3, 4, 6, 0, 0, 0, 0, 0}, /* 0x2C (00101100) */ - {1, 3, 4, 6, 0, 0, 0, 0}, /* 0x2D (00101101) */ - {2, 3, 4, 6, 0, 0, 0, 0}, /* 0x2E (00101110) */ - {1, 2, 3, 4, 6, 0, 0, 0}, /* 0x2F (00101111) */ - {5, 6, 0, 0, 0, 0, 0, 0}, /* 0x30 (00110000) */ - {1, 5, 6, 0, 0, 0, 0, 0}, /* 0x31 (00110001) */ - {2, 5, 6, 0, 0, 0, 0, 0}, /* 0x32 (00110010) */ - {1, 2, 5, 6, 0, 0, 0, 0}, /* 0x33 (00110011) */ - {3, 5, 6, 0, 0, 0, 0, 0}, /* 0x34 (00110100) */ - {1, 3, 5, 6, 0, 0, 0, 0}, /* 0x35 (00110101) */ - {2, 3, 5, 6, 0, 0, 0, 0}, /* 0x36 (00110110) */ - {1, 2, 3, 5, 6, 0, 0, 0}, /* 0x37 (00110111) */ - {4, 5, 6, 0, 0, 0, 0, 0}, /* 0x38 (00111000) */ - {1, 4, 5, 6, 0, 0, 0, 0}, /* 0x39 (00111001) */ - {2, 4, 5, 6, 0, 0, 0, 0}, /* 0x3A (00111010) */ - {1, 2, 4, 5, 6, 0, 0, 0}, /* 0x3B (00111011) */ - {3, 4, 5, 6, 0, 0, 0, 0}, /* 0x3C (00111100) */ - {1, 3, 4, 5, 6, 0, 0, 0}, /* 0x3D (00111101) */ - {2, 3, 4, 5, 6, 0, 0, 0}, /* 0x3E (00111110) */ - {1, 2, 3, 4, 5, 6, 0, 0}, /* 0x3F (00111111) */ - {7, 0, 0, 0, 0, 0, 0, 0}, /* 0x40 (01000000) */ - {1, 7, 0, 0, 0, 0, 0, 0}, /* 0x41 (01000001) */ - {2, 7, 0, 0, 0, 0, 0, 0}, /* 0x42 (01000010) */ - {1, 2, 7, 0, 0, 0, 0, 0}, /* 0x43 (01000011) */ - {3, 7, 0, 0, 0, 0, 0, 0}, /* 0x44 (01000100) */ - {1, 3, 7, 0, 0, 0, 0, 0}, /* 0x45 (01000101) */ - {2, 3, 7, 0, 0, 0, 0, 0}, /* 0x46 (01000110) */ - {1, 2, 3, 7, 0, 0, 0, 0}, /* 0x47 (01000111) */ - {4, 7, 0, 0, 0, 0, 0, 0}, /* 0x48 (01001000) */ - {1, 4, 7, 0, 0, 0, 0, 0}, /* 0x49 (01001001) */ - {2, 4, 7, 0, 0, 0, 0, 0}, /* 0x4A (01001010) */ - {1, 2, 4, 7, 0, 0, 0, 0}, /* 0x4B (01001011) */ - {3, 4, 7, 0, 0, 0, 0, 0}, /* 0x4C (01001100) */ - {1, 3, 4, 7, 0, 0, 0, 0}, /* 0x4D (01001101) */ - {2, 3, 4, 7, 0, 0, 0, 0}, /* 0x4E (01001110) */ - {1, 2, 3, 4, 7, 0, 0, 0}, /* 0x4F (01001111) */ - {5, 7, 0, 0, 0, 0, 0, 0}, /* 0x50 (01010000) */ - {1, 5, 7, 0, 0, 0, 0, 0}, /* 0x51 (01010001) */ - {2, 5, 7, 0, 0, 0, 0, 0}, /* 0x52 (01010010) */ - {1, 2, 5, 7, 0, 0, 0, 0}, /* 0x53 (01010011) */ - {3, 5, 7, 0, 0, 0, 0, 0}, /* 0x54 (01010100) */ - {1, 3, 5, 7, 0, 0, 0, 0}, /* 0x55 (01010101) */ - {2, 3, 5, 7, 0, 0, 0, 0}, /* 0x56 (01010110) */ - {1, 2, 3, 5, 7, 0, 0, 0}, /* 0x57 (01010111) */ - {4, 5, 7, 0, 0, 0, 0, 0}, /* 0x58 (01011000) */ - {1, 4, 5, 7, 0, 0, 0, 0}, /* 0x59 (01011001) */ - {2, 4, 5, 7, 0, 0, 0, 0}, /* 0x5A (01011010) */ - {1, 2, 4, 5, 7, 0, 0, 0}, /* 0x5B (01011011) */ - {3, 4, 5, 7, 0, 0, 0, 0}, /* 0x5C (01011100) */ - {1, 3, 4, 5, 7, 0, 0, 0}, /* 0x5D (01011101) */ - {2, 3, 4, 5, 7, 0, 0, 0}, /* 0x5E (01011110) */ - {1, 2, 3, 4, 5, 7, 0, 0}, /* 0x5F (01011111) */ - {6, 7, 0, 0, 0, 0, 0, 0}, /* 0x60 (01100000) */ - {1, 6, 7, 0, 0, 0, 0, 0}, /* 0x61 (01100001) */ - {2, 6, 7, 0, 0, 0, 0, 0}, /* 0x62 (01100010) */ - {1, 2, 6, 7, 0, 0, 0, 0}, /* 0x63 (01100011) */ - {3, 6, 7, 0, 0, 0, 0, 0}, /* 0x64 (01100100) */ - {1, 3, 6, 7, 0, 0, 0, 0}, /* 0x65 (01100101) */ - {2, 3, 6, 7, 0, 0, 0, 0}, /* 0x66 (01100110) */ - {1, 2, 3, 6, 7, 0, 0, 0}, /* 0x67 (01100111) */ - {4, 6, 7, 0, 0, 0, 0, 0}, /* 0x68 (01101000) */ - {1, 4, 6, 7, 0, 0, 0, 0}, /* 0x69 (01101001) */ - {2, 4, 6, 7, 0, 0, 0, 0}, /* 0x6A (01101010) */ - {1, 2, 4, 6, 7, 0, 0, 0}, /* 0x6B (01101011) */ - {3, 4, 6, 7, 0, 0, 0, 0}, /* 0x6C (01101100) */ - {1, 3, 4, 6, 7, 0, 0, 0}, /* 0x6D (01101101) */ - {2, 3, 4, 6, 7, 0, 0, 0}, /* 0x6E (01101110) */ - {1, 2, 3, 4, 6, 7, 0, 0}, /* 0x6F (01101111) */ - {5, 6, 7, 0, 0, 0, 0, 0}, /* 0x70 (01110000) */ - {1, 5, 6, 7, 0, 0, 0, 0}, /* 0x71 (01110001) */ - {2, 5, 6, 7, 0, 0, 0, 0}, /* 0x72 (01110010) */ - {1, 2, 5, 6, 7, 0, 0, 0}, /* 0x73 (01110011) */ - {3, 5, 6, 7, 0, 0, 0, 0}, /* 0x74 (01110100) */ - {1, 3, 5, 6, 7, 0, 0, 0}, /* 0x75 (01110101) */ - {2, 3, 5, 6, 7, 0, 0, 0}, /* 0x76 (01110110) */ - {1, 2, 3, 5, 6, 7, 0, 0}, /* 0x77 (01110111) */ - {4, 5, 6, 7, 0, 0, 0, 0}, /* 0x78 (01111000) */ - {1, 4, 5, 6, 7, 0, 0, 0}, /* 0x79 (01111001) */ - {2, 4, 5, 6, 7, 0, 0, 0}, /* 0x7A (01111010) */ - {1, 2, 4, 5, 6, 7, 0, 0}, /* 0x7B (01111011) */ - {3, 4, 5, 6, 7, 0, 0, 0}, /* 0x7C (01111100) */ - {1, 3, 4, 5, 6, 7, 0, 0}, /* 0x7D (01111101) */ - {2, 3, 4, 5, 6, 7, 0, 0}, /* 0x7E (01111110) */ - {1, 2, 3, 4, 5, 6, 7, 0}, /* 0x7F (01111111) */ - {8, 0, 0, 0, 0, 0, 0, 0}, /* 0x80 (10000000) */ - {1, 8, 0, 0, 0, 0, 0, 0}, /* 0x81 (10000001) */ - {2, 8, 0, 0, 0, 0, 0, 0}, /* 0x82 (10000010) */ - {1, 2, 8, 0, 0, 0, 0, 0}, /* 0x83 (10000011) */ - {3, 8, 0, 0, 0, 0, 0, 0}, /* 0x84 (10000100) */ - {1, 3, 8, 0, 0, 0, 0, 0}, /* 0x85 (10000101) */ - {2, 3, 8, 0, 0, 0, 0, 0}, /* 0x86 (10000110) */ - {1, 2, 3, 8, 0, 0, 0, 0}, /* 0x87 (10000111) */ - {4, 8, 0, 0, 0, 0, 0, 0}, /* 0x88 (10001000) */ - {1, 4, 8, 0, 0, 0, 0, 0}, /* 0x89 (10001001) */ - {2, 4, 8, 0, 0, 0, 0, 0}, /* 0x8A (10001010) */ - {1, 2, 4, 8, 0, 0, 0, 0}, /* 0x8B (10001011) */ - {3, 4, 8, 0, 0, 0, 0, 0}, /* 0x8C (10001100) */ - {1, 3, 4, 8, 0, 0, 0, 0}, /* 0x8D (10001101) */ - {2, 3, 4, 8, 0, 0, 0, 0}, /* 0x8E (10001110) */ - {1, 2, 3, 4, 8, 0, 0, 0}, /* 0x8F (10001111) */ - {5, 8, 0, 0, 0, 0, 0, 0}, /* 0x90 (10010000) */ - {1, 5, 8, 0, 0, 0, 0, 0}, /* 0x91 (10010001) */ - {2, 5, 8, 0, 0, 0, 0, 0}, /* 0x92 (10010010) */ - {1, 2, 5, 8, 0, 0, 0, 0}, /* 0x93 (10010011) */ - {3, 5, 8, 0, 0, 0, 0, 0}, /* 0x94 (10010100) */ - {1, 3, 5, 8, 0, 0, 0, 0}, /* 0x95 (10010101) */ - {2, 3, 5, 8, 0, 0, 0, 0}, /* 0x96 (10010110) */ - {1, 2, 3, 5, 8, 0, 0, 0}, /* 0x97 (10010111) */ - {4, 5, 8, 0, 0, 0, 0, 0}, /* 0x98 (10011000) */ - {1, 4, 5, 8, 0, 0, 0, 0}, /* 0x99 (10011001) */ - {2, 4, 5, 8, 0, 0, 0, 0}, /* 0x9A (10011010) */ - {1, 2, 4, 5, 8, 0, 0, 0}, /* 0x9B (10011011) */ - {3, 4, 5, 8, 0, 0, 0, 0}, /* 0x9C (10011100) */ - {1, 3, 4, 5, 8, 0, 0, 0}, /* 0x9D (10011101) */ - {2, 3, 4, 5, 8, 0, 0, 0}, /* 0x9E (10011110) */ - {1, 2, 3, 4, 5, 8, 0, 0}, /* 0x9F (10011111) */ - {6, 8, 0, 0, 0, 0, 0, 0}, /* 0xA0 (10100000) */ - {1, 6, 8, 0, 0, 0, 0, 0}, /* 0xA1 (10100001) */ - {2, 6, 8, 0, 0, 0, 0, 0}, /* 0xA2 (10100010) */ - {1, 2, 6, 8, 0, 0, 0, 0}, /* 0xA3 (10100011) */ - {3, 6, 8, 0, 0, 0, 0, 0}, /* 0xA4 (10100100) */ - {1, 3, 6, 8, 0, 0, 0, 0}, /* 0xA5 (10100101) */ - {2, 3, 6, 8, 0, 0, 0, 0}, /* 0xA6 (10100110) */ - {1, 2, 3, 6, 8, 0, 0, 0}, /* 0xA7 (10100111) */ - {4, 6, 8, 0, 0, 0, 0, 0}, /* 0xA8 (10101000) */ - {1, 4, 6, 8, 0, 0, 0, 0}, /* 0xA9 (10101001) */ - {2, 4, 6, 8, 0, 0, 0, 0}, /* 0xAA (10101010) */ - {1, 2, 4, 6, 8, 0, 0, 0}, /* 0xAB (10101011) */ - {3, 4, 6, 8, 0, 0, 0, 0}, /* 0xAC (10101100) */ - {1, 3, 4, 6, 8, 0, 0, 0}, /* 0xAD (10101101) */ - {2, 3, 4, 6, 8, 0, 0, 0}, /* 0xAE (10101110) */ - {1, 2, 3, 4, 6, 8, 0, 0}, /* 0xAF (10101111) */ - {5, 6, 8, 0, 0, 0, 0, 0}, /* 0xB0 (10110000) */ - {1, 5, 6, 8, 0, 0, 0, 0}, /* 0xB1 (10110001) */ - {2, 5, 6, 8, 0, 0, 0, 0}, /* 0xB2 (10110010) */ - {1, 2, 5, 6, 8, 0, 0, 0}, /* 0xB3 (10110011) */ - {3, 5, 6, 8, 0, 0, 0, 0}, /* 0xB4 (10110100) */ - {1, 3, 5, 6, 8, 0, 0, 0}, /* 0xB5 (10110101) */ - {2, 3, 5, 6, 8, 0, 0, 0}, /* 0xB6 (10110110) */ - {1, 2, 3, 5, 6, 8, 0, 0}, /* 0xB7 (10110111) */ - {4, 5, 6, 8, 0, 0, 0, 0}, /* 0xB8 (10111000) */ - {1, 4, 5, 6, 8, 0, 0, 0}, /* 0xB9 (10111001) */ - {2, 4, 5, 6, 8, 0, 0, 0}, /* 0xBA (10111010) */ - {1, 2, 4, 5, 6, 8, 0, 0}, /* 0xBB (10111011) */ - {3, 4, 5, 6, 8, 0, 0, 0}, /* 0xBC (10111100) */ - {1, 3, 4, 5, 6, 8, 0, 0}, /* 0xBD (10111101) */ - {2, 3, 4, 5, 6, 8, 0, 0}, /* 0xBE (10111110) */ - {1, 2, 3, 4, 5, 6, 8, 0}, /* 0xBF (10111111) */ - {7, 8, 0, 0, 0, 0, 0, 0}, /* 0xC0 (11000000) */ - {1, 7, 8, 0, 0, 0, 0, 0}, /* 0xC1 (11000001) */ - {2, 7, 8, 0, 0, 0, 0, 0}, /* 0xC2 (11000010) */ - {1, 2, 7, 8, 0, 0, 0, 0}, /* 0xC3 (11000011) */ - {3, 7, 8, 0, 0, 0, 0, 0}, /* 0xC4 (11000100) */ - {1, 3, 7, 8, 0, 0, 0, 0}, /* 0xC5 (11000101) */ - {2, 3, 7, 8, 0, 0, 0, 0}, /* 0xC6 (11000110) */ - {1, 2, 3, 7, 8, 0, 0, 0}, /* 0xC7 (11000111) */ - {4, 7, 8, 0, 0, 0, 0, 0}, /* 0xC8 (11001000) */ - {1, 4, 7, 8, 0, 0, 0, 0}, /* 0xC9 (11001001) */ - {2, 4, 7, 8, 0, 0, 0, 0}, /* 0xCA (11001010) */ - {1, 2, 4, 7, 8, 0, 0, 0}, /* 0xCB (11001011) */ - {3, 4, 7, 8, 0, 0, 0, 0}, /* 0xCC (11001100) */ - {1, 3, 4, 7, 8, 0, 0, 0}, /* 0xCD (11001101) */ - {2, 3, 4, 7, 8, 0, 0, 0}, /* 0xCE (11001110) */ - {1, 2, 3, 4, 7, 8, 0, 0}, /* 0xCF (11001111) */ - {5, 7, 8, 0, 0, 0, 0, 0}, /* 0xD0 (11010000) */ - {1, 5, 7, 8, 0, 0, 0, 0}, /* 0xD1 (11010001) */ - {2, 5, 7, 8, 0, 0, 0, 0}, /* 0xD2 (11010010) */ - {1, 2, 5, 7, 8, 0, 0, 0}, /* 0xD3 (11010011) */ - {3, 5, 7, 8, 0, 0, 0, 0}, /* 0xD4 (11010100) */ - {1, 3, 5, 7, 8, 0, 0, 0}, /* 0xD5 (11010101) */ - {2, 3, 5, 7, 8, 0, 0, 0}, /* 0xD6 (11010110) */ - {1, 2, 3, 5, 7, 8, 0, 0}, /* 0xD7 (11010111) */ - {4, 5, 7, 8, 0, 0, 0, 0}, /* 0xD8 (11011000) */ - {1, 4, 5, 7, 8, 0, 0, 0}, /* 0xD9 (11011001) */ - {2, 4, 5, 7, 8, 0, 0, 0}, /* 0xDA (11011010) */ - {1, 2, 4, 5, 7, 8, 0, 0}, /* 0xDB (11011011) */ - {3, 4, 5, 7, 8, 0, 0, 0}, /* 0xDC (11011100) */ - {1, 3, 4, 5, 7, 8, 0, 0}, /* 0xDD (11011101) */ - {2, 3, 4, 5, 7, 8, 0, 0}, /* 0xDE (11011110) */ - {1, 2, 3, 4, 5, 7, 8, 0}, /* 0xDF (11011111) */ - {6, 7, 8, 0, 0, 0, 0, 0}, /* 0xE0 (11100000) */ - {1, 6, 7, 8, 0, 0, 0, 0}, /* 0xE1 (11100001) */ - {2, 6, 7, 8, 0, 0, 0, 0}, /* 0xE2 (11100010) */ - {1, 2, 6, 7, 8, 0, 0, 0}, /* 0xE3 (11100011) */ - {3, 6, 7, 8, 0, 0, 0, 0}, /* 0xE4 (11100100) */ - {1, 3, 6, 7, 8, 0, 0, 0}, /* 0xE5 (11100101) */ - {2, 3, 6, 7, 8, 0, 0, 0}, /* 0xE6 (11100110) */ - {1, 2, 3, 6, 7, 8, 0, 0}, /* 0xE7 (11100111) */ - {4, 6, 7, 8, 0, 0, 0, 0}, /* 0xE8 (11101000) */ - {1, 4, 6, 7, 8, 0, 0, 0}, /* 0xE9 (11101001) */ - {2, 4, 6, 7, 8, 0, 0, 0}, /* 0xEA (11101010) */ - {1, 2, 4, 6, 7, 8, 0, 0}, /* 0xEB (11101011) */ - {3, 4, 6, 7, 8, 0, 0, 0}, /* 0xEC (11101100) */ - {1, 3, 4, 6, 7, 8, 0, 0}, /* 0xED (11101101) */ - {2, 3, 4, 6, 7, 8, 0, 0}, /* 0xEE (11101110) */ - {1, 2, 3, 4, 6, 7, 8, 0}, /* 0xEF (11101111) */ - {5, 6, 7, 8, 0, 0, 0, 0}, /* 0xF0 (11110000) */ - {1, 5, 6, 7, 8, 0, 0, 0}, /* 0xF1 (11110001) */ - {2, 5, 6, 7, 8, 0, 0, 0}, /* 0xF2 (11110010) */ - {1, 2, 5, 6, 7, 8, 0, 0}, /* 0xF3 (11110011) */ - {3, 5, 6, 7, 8, 0, 0, 0}, /* 0xF4 (11110100) */ - {1, 3, 5, 6, 7, 8, 0, 0}, /* 0xF5 (11110101) */ - {2, 3, 5, 6, 7, 8, 0, 0}, /* 0xF6 (11110110) */ - {1, 2, 3, 5, 6, 7, 8, 0}, /* 0xF7 (11110111) */ - {4, 5, 6, 7, 8, 0, 0, 0}, /* 0xF8 (11111000) */ - {1, 4, 5, 6, 7, 8, 0, 0}, /* 0xF9 (11111001) */ - {2, 4, 5, 6, 7, 8, 0, 0}, /* 0xFA (11111010) */ - {1, 2, 4, 5, 6, 7, 8, 0}, /* 0xFB (11111011) */ - {3, 4, 5, 6, 7, 8, 0, 0}, /* 0xFC (11111100) */ - {1, 3, 4, 5, 6, 7, 8, 0}, /* 0xFD (11111101) */ - {2, 3, 4, 5, 6, 7, 8, 0}, /* 0xFE (11111110) */ - {1, 2, 3, 4, 5, 6, 7, 8} /* 0xFF (11111111) */ -}; - -#endif - -#ifdef USEAVX - -size_t bitset_extract_setbits_avx2(uint64_t *array, size_t length, void *vout, - size_t outcapacity, uint32_t base) { - uint32_t *out = (uint32_t *)vout; - uint32_t *initout = out; - __m256i baseVec = _mm256_set1_epi32(base - 1); - __m256i incVec = _mm256_set1_epi32(64); - __m256i add8 = _mm256_set1_epi32(8); - uint32_t *safeout = out + outcapacity; - size_t i = 0; - for (; (i < length) && (out + 64 <= safeout); ++i) { - uint64_t w = array[i]; - if (w == 0) { - baseVec = _mm256_add_epi32(baseVec, incVec); - } else { - for (int k = 0; k < 4; ++k) { - uint8_t byteA = (uint8_t)w; - uint8_t byteB = (uint8_t)(w >> 8); - w >>= 16; - __m256i vecA = - _mm256_load_si256((const __m256i *)vecDecodeTable[byteA]); - __m256i vecB = - _mm256_load_si256((const __m256i *)vecDecodeTable[byteB]); - uint8_t advanceA = lengthTable[byteA]; - uint8_t advanceB = lengthTable[byteB]; - vecA = _mm256_add_epi32(baseVec, vecA); - baseVec = _mm256_add_epi32(baseVec, add8); - vecB = _mm256_add_epi32(baseVec, vecB); - baseVec = _mm256_add_epi32(baseVec, add8); - _mm256_storeu_si256((__m256i *)out, vecA); - out += advanceA; - _mm256_storeu_si256((__m256i *)out, vecB); - out += advanceB; - } - } - } - base += i * 64; - for (; (i < length) && (out < safeout); ++i) { - uint64_t w = array[i]; - while ((w != 0) && (out < safeout)) { - uint64_t t = w & (~w + 1); // on x64, should compile to BLSI (careful: the Intel compiler seems to fail) - int r = __builtin_ctzll(w); // on x64, should compile to TZCNT - uint32_t val = r + base; - memcpy(out, &val, - sizeof(uint32_t)); // should be compiled as a MOV on x64 - out++; - w ^= t; - } - base += 64; - } - return out - initout; -} -#endif // USEAVX - -size_t bitset_extract_setbits(uint64_t *bitset, size_t length, void *vout, - uint32_t base) { - int outpos = 0; - uint32_t *out = (uint32_t *)vout; - for (size_t i = 0; i < length; ++i) { - uint64_t w = bitset[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); // on x64, should compile to BLSI (careful: the Intel compiler seems to fail) - int r = __builtin_ctzll(w); // on x64, should compile to TZCNT - uint32_t val = r + base; - memcpy(out + outpos, &val, - sizeof(uint32_t)); // should be compiled as a MOV on x64 - outpos++; - w ^= t; - } - base += 64; - } - return outpos; -} - -size_t bitset_extract_intersection_setbits_uint16(const uint64_t * __restrict__ bitset1, - const uint64_t * __restrict__ bitset2, - size_t length, uint16_t *out, - uint16_t base) { - int outpos = 0; - for (size_t i = 0; i < length; ++i) { - uint64_t w = bitset1[i] & bitset2[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - out[outpos++] = r + base; - w ^= t; - } - base += 64; - } - return outpos; -} - -#ifdef IS_X64 -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out" as 16-bit integers, values start at "base" (can - *be set to zero). - * - * The "out" pointer should be sufficient to store the actual number of bits - *set. - * - * Returns how many values were actually decoded. - * - * This function uses SSE decoding. - */ -size_t bitset_extract_setbits_sse_uint16(const uint64_t *bitset, size_t length, - uint16_t *out, size_t outcapacity, - uint16_t base) { - uint16_t *initout = out; - __m128i baseVec = _mm_set1_epi16(base - 1); - __m128i incVec = _mm_set1_epi16(64); - __m128i add8 = _mm_set1_epi16(8); - uint16_t *safeout = out + outcapacity; - const int numberofbytes = 2; // process two bytes at a time - size_t i = 0; - for (; (i < length) && (out + numberofbytes * 8 <= safeout); ++i) { - uint64_t w = bitset[i]; - if (w == 0) { - baseVec = _mm_add_epi16(baseVec, incVec); - } else { - for (int k = 0; k < 4; ++k) { - uint8_t byteA = (uint8_t)w; - uint8_t byteB = (uint8_t)(w >> 8); - w >>= 16; - __m128i vecA = _mm_load_si128( - (const __m128i *)vecDecodeTable_uint16[byteA]); - __m128i vecB = _mm_load_si128( - (const __m128i *)vecDecodeTable_uint16[byteB]); - uint8_t advanceA = lengthTable[byteA]; - uint8_t advanceB = lengthTable[byteB]; - vecA = _mm_add_epi16(baseVec, vecA); - baseVec = _mm_add_epi16(baseVec, add8); - vecB = _mm_add_epi16(baseVec, vecB); - baseVec = _mm_add_epi16(baseVec, add8); - _mm_storeu_si128((__m128i *)out, vecA); - out += advanceA; - _mm_storeu_si128((__m128i *)out, vecB); - out += advanceB; - } - } - } - base += (uint16_t)(i * 64); - for (; (i < length) && (out < safeout); ++i) { - uint64_t w = bitset[i]; - while ((w != 0) && (out < safeout)) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - *out = r + base; - out++; - w ^= t; - } - base += 64; - } - return out - initout; -} -#endif - -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out", values start at "base" (can be set to zero). - * - * The "out" pointer should be sufficient to store the actual number of bits - *set. - * - * Returns how many values were actually decoded. - */ -size_t bitset_extract_setbits_uint16(const uint64_t *bitset, size_t length, - uint16_t *out, uint16_t base) { - int outpos = 0; - for (size_t i = 0; i < length; ++i) { - uint64_t w = bitset[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - out[outpos++] = r + base; - w ^= t; - } - base += 64; - } - return outpos; -} - -#if defined(ASMBITMANIPOPTIMIZATION) - -uint64_t bitset_set_list_withcard(void *bitset, uint64_t card, - const uint16_t *list, uint64_t length) { - uint64_t offset, load, pos; - uint64_t shift = 6; - const uint16_t *end = list + length; - if (!length) return card; - // TODO: could unroll for performance, see bitset_set_list - // bts is not available as an intrinsic in GCC - __asm volatile( - "1:\n" - "movzwq (%[list]), %[pos]\n" - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)\n" - "sbb $-1, %[card]\n" - "add $2, %[list]\n" - "cmp %[list], %[end]\n" - "jnz 1b" - : [card] "+&r"(card), [list] "+&r"(list), [load] "=&r"(load), - [pos] "=&r"(pos), [offset] "=&r"(offset) - : [end] "r"(end), [bitset] "r"(bitset), [shift] "r"(shift)); - return card; -} - -void bitset_set_list(void *bitset, const uint16_t *list, uint64_t length) { - uint64_t pos; - const uint16_t *end = list + length; - - uint64_t shift = 6; - uint64_t offset; - uint64_t load; - for (; list + 3 < end; list += 4) { - pos = list[0]; - __asm volatile( - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)" - : [load] "=&r"(load), [offset] "=&r"(offset) - : [bitset] "r"(bitset), [shift] "r"(shift), [pos] "r"(pos)); - pos = list[1]; - __asm volatile( - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)" - : [load] "=&r"(load), [offset] "=&r"(offset) - : [bitset] "r"(bitset), [shift] "r"(shift), [pos] "r"(pos)); - pos = list[2]; - __asm volatile( - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)" - : [load] "=&r"(load), [offset] "=&r"(offset) - : [bitset] "r"(bitset), [shift] "r"(shift), [pos] "r"(pos)); - pos = list[3]; - __asm volatile( - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)" - : [load] "=&r"(load), [offset] "=&r"(offset) - : [bitset] "r"(bitset), [shift] "r"(shift), [pos] "r"(pos)); - } - - while (list != end) { - pos = list[0]; - __asm volatile( - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "bts %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)" - : [load] "=&r"(load), [offset] "=&r"(offset) - : [bitset] "r"(bitset), [shift] "r"(shift), [pos] "r"(pos)); - list++; - } -} - -uint64_t bitset_clear_list(void *bitset, uint64_t card, const uint16_t *list, - uint64_t length) { - uint64_t offset, load, pos; - uint64_t shift = 6; - const uint16_t *end = list + length; - if (!length) return card; - // btr is not available as an intrinsic in GCC - __asm volatile( - "1:\n" - "movzwq (%[list]), %[pos]\n" - "shrx %[shift], %[pos], %[offset]\n" - "mov (%[bitset],%[offset],8), %[load]\n" - "btr %[pos], %[load]\n" - "mov %[load], (%[bitset],%[offset],8)\n" - "sbb $0, %[card]\n" - "add $2, %[list]\n" - "cmp %[list], %[end]\n" - "jnz 1b" - : [card] "+&r"(card), [list] "+&r"(list), [load] "=&r"(load), - [pos] "=&r"(pos), [offset] "=&r"(offset) - : [end] "r"(end), [bitset] "r"(bitset), [shift] "r"(shift) - : - /* clobbers */ "memory"); - return card; -} - -#else -uint64_t bitset_clear_list(void *bitset, uint64_t card, const uint16_t *list, - uint64_t length) { - uint64_t offset, load, newload, pos, index; - const uint16_t *end = list + length; - while (list != end) { - pos = *(const uint16_t *)list; - offset = pos >> 6; - index = pos % 64; - load = ((uint64_t *)bitset)[offset]; - newload = load & ~(UINT64_C(1) << index); - card -= (load ^ newload) >> index; - ((uint64_t *)bitset)[offset] = newload; - list++; - } - return card; -} - -uint64_t bitset_set_list_withcard(void *bitset, uint64_t card, - const uint16_t *list, uint64_t length) { - uint64_t offset, load, newload, pos, index; - const uint16_t *end = list + length; - while (list != end) { - pos = *(const uint16_t *)list; - offset = pos >> 6; - index = pos % 64; - load = ((uint64_t *)bitset)[offset]; - newload = load | (UINT64_C(1) << index); - card += (load ^ newload) >> index; - ((uint64_t *)bitset)[offset] = newload; - list++; - } - return card; -} - -void bitset_set_list(void *bitset, const uint16_t *list, uint64_t length) { - uint64_t offset, load, newload, pos, index; - const uint16_t *end = list + length; - while (list != end) { - pos = *(const uint16_t *)list; - offset = pos >> 6; - index = pos % 64; - load = ((uint64_t *)bitset)[offset]; - newload = load | (UINT64_C(1) << index); - ((uint64_t *)bitset)[offset] = newload; - list++; - } -} - -#endif - -/* flip specified bits */ -/* TODO: consider whether worthwhile to make an asm version */ - -uint64_t bitset_flip_list_withcard(void *bitset, uint64_t card, - const uint16_t *list, uint64_t length) { - uint64_t offset, load, newload, pos, index; - const uint16_t *end = list + length; - while (list != end) { - pos = *(const uint16_t *)list; - offset = pos >> 6; - index = pos % 64; - load = ((uint64_t *)bitset)[offset]; - newload = load ^ (UINT64_C(1) << index); - // todo: is a branch here all that bad? - card += - (1 - 2 * (((UINT64_C(1) << index) & load) >> index)); // +1 or -1 - ((uint64_t *)bitset)[offset] = newload; - list++; - } - return card; -} - -void bitset_flip_list(void *bitset, const uint16_t *list, uint64_t length) { - uint64_t offset, load, newload, pos, index; - const uint16_t *end = list + length; - while (list != end) { - pos = *(const uint16_t *)list; - offset = pos >> 6; - index = pos % 64; - load = ((uint64_t *)bitset)[offset]; - newload = load ^ (UINT64_C(1) << index); - ((uint64_t *)bitset)[offset] = newload; - list++; - } -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/bitset_util.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/array.c */ -/* - * array.c - * - */ - -#include -#include -#include - -extern inline uint16_t array_container_minimum(const array_container_t *arr); -extern inline uint16_t array_container_maximum(const array_container_t *arr); -extern inline int array_container_index_equalorlarger(const array_container_t *arr, uint16_t x); - -extern inline int array_container_rank(const array_container_t *arr, - uint16_t x); -extern inline bool array_container_contains(const array_container_t *arr, - uint16_t pos); -extern int array_container_cardinality(const array_container_t *array); -extern bool array_container_nonzero_cardinality(const array_container_t *array); -extern void array_container_clear(array_container_t *array); -extern int32_t array_container_serialized_size_in_bytes(int32_t card); -extern bool array_container_empty(const array_container_t *array); -extern bool array_container_full(const array_container_t *array); - -/* Create a new array with capacity size. Return NULL in case of failure. */ -array_container_t *array_container_create_given_capacity(int32_t size) { - array_container_t *container; - - if ((container = (array_container_t *)malloc(sizeof(array_container_t))) == - NULL) { - return NULL; - } - - if( size <= 0 ) { // we don't want to rely on malloc(0) - container->array = NULL; - } else if ((container->array = (uint16_t *)malloc(sizeof(uint16_t) * size)) == - NULL) { - free(container); - return NULL; - } - - container->capacity = size; - container->cardinality = 0; - - return container; -} - -/* Create a new array. Return NULL in case of failure. */ -array_container_t *array_container_create() { - return array_container_create_given_capacity(ARRAY_DEFAULT_INIT_SIZE); -} - -/* Create a new array containing all values in [min,max). */ -array_container_t * array_container_create_range(uint32_t min, uint32_t max) { - array_container_t * answer = array_container_create_given_capacity(max - min + 1); - if(answer == NULL) return answer; - answer->cardinality = 0; - for(uint32_t k = min; k < max; k++) { - answer->array[answer->cardinality++] = k; - } - return answer; -} - -/* Duplicate container */ -array_container_t *array_container_clone(const array_container_t *src) { - array_container_t *newcontainer = - array_container_create_given_capacity(src->capacity); - if (newcontainer == NULL) return NULL; - - newcontainer->cardinality = src->cardinality; - - memcpy(newcontainer->array, src->array, - src->cardinality * sizeof(uint16_t)); - - return newcontainer; -} - -int array_container_shrink_to_fit(array_container_t *src) { - if (src->cardinality == src->capacity) return 0; // nothing to do - int savings = src->capacity - src->cardinality; - src->capacity = src->cardinality; - if( src->capacity == 0) { // we do not want to rely on realloc for zero allocs - free(src->array); - src->array = NULL; - } else { - uint16_t *oldarray = src->array; - src->array = - (uint16_t *)realloc(oldarray, src->capacity * sizeof(uint16_t)); - if (src->array == NULL) free(oldarray); // should never happen? - } - return savings; -} - -/* Free memory. */ -void array_container_free(array_container_t *arr) { - if(arr->array != NULL) {// Jon Strabala reports that some tools complain otherwise - free(arr->array); - arr->array = NULL; // pedantic - } - free(arr); -} - -static inline int32_t grow_capacity(int32_t capacity) { - return (capacity <= 0) ? ARRAY_DEFAULT_INIT_SIZE - : capacity < 64 ? capacity * 2 - : capacity < 1024 ? capacity * 3 / 2 - : capacity * 5 / 4; -} - -static inline int32_t clamp(int32_t val, int32_t min, int32_t max) { - return ((val < min) ? min : (val > max) ? max : val); -} - -void array_container_grow(array_container_t *container, int32_t min, - bool preserve) { - - int32_t max = (min <= DEFAULT_MAX_SIZE ? DEFAULT_MAX_SIZE : 65536); - int32_t new_capacity = clamp(grow_capacity(container->capacity), min, max); - - container->capacity = new_capacity; - uint16_t *array = container->array; - - if (preserve) { - container->array = - (uint16_t *)realloc(array, new_capacity * sizeof(uint16_t)); - if (container->array == NULL) free(array); - } else { - // Jon Strabala reports that some tools complain otherwise - if (array != NULL) { - free(array); - } - container->array = (uint16_t *)malloc(new_capacity * sizeof(uint16_t)); - } - - // handle the case where realloc fails - if (container->array == NULL) { - fprintf(stderr, "could not allocate memory\n"); - } - assert(container->array != NULL); -} - -/* Copy one container into another. We assume that they are distinct. */ -void array_container_copy(const array_container_t *src, - array_container_t *dst) { - const int32_t cardinality = src->cardinality; - if (cardinality > dst->capacity) { - array_container_grow(dst, cardinality, false); - } - - dst->cardinality = cardinality; - memcpy(dst->array, src->array, cardinality * sizeof(uint16_t)); -} - -void array_container_add_from_range(array_container_t *arr, uint32_t min, - uint32_t max, uint16_t step) { - for (uint32_t value = min; value < max; value += step) { - array_container_append(arr, value); - } -} - -/* Computes the union of array1 and array2 and write the result to arrayout. - * It is assumed that arrayout is distinct from both array1 and array2. - */ -void array_container_union(const array_container_t *array_1, - const array_container_t *array_2, - array_container_t *out) { - const int32_t card_1 = array_1->cardinality, card_2 = array_2->cardinality; - const int32_t max_cardinality = card_1 + card_2; - - if (out->capacity < max_cardinality) { - array_container_grow(out, max_cardinality, false); - } - out->cardinality = (int32_t)fast_union_uint16(array_1->array, card_1, - array_2->array, card_2, out->array); - -} - -/* Computes the difference of array1 and array2 and write the result - * to array out. - * Array out does not need to be distinct from array_1 - */ -void array_container_andnot(const array_container_t *array_1, - const array_container_t *array_2, - array_container_t *out) { - if (out->capacity < array_1->cardinality) - array_container_grow(out, array_1->cardinality, false); -#ifdef ROARING_VECTOR_OPERATIONS_ENABLED - out->cardinality = - difference_vector16(array_1->array, array_1->cardinality, - array_2->array, array_2->cardinality, out->array); -#else - out->cardinality = - difference_uint16(array_1->array, array_1->cardinality, array_2->array, - array_2->cardinality, out->array); -#endif -} - -/* Computes the symmetric difference of array1 and array2 and write the - * result - * to arrayout. - * It is assumed that arrayout is distinct from both array1 and array2. - */ -void array_container_xor(const array_container_t *array_1, - const array_container_t *array_2, - array_container_t *out) { - const int32_t card_1 = array_1->cardinality, card_2 = array_2->cardinality; - const int32_t max_cardinality = card_1 + card_2; - if (out->capacity < max_cardinality) { - array_container_grow(out, max_cardinality, false); - } - -#ifdef ROARING_VECTOR_OPERATIONS_ENABLED - out->cardinality = - xor_vector16(array_1->array, array_1->cardinality, array_2->array, - array_2->cardinality, out->array); -#else - out->cardinality = - xor_uint16(array_1->array, array_1->cardinality, array_2->array, - array_2->cardinality, out->array); -#endif -} - -static inline int32_t minimum_int32(int32_t a, int32_t b) { - return (a < b) ? a : b; -} - -/* computes the intersection of array1 and array2 and write the result to - * arrayout. - * It is assumed that arrayout is distinct from both array1 and array2. - * */ -void array_container_intersection(const array_container_t *array1, - const array_container_t *array2, - array_container_t *out) { - int32_t card_1 = array1->cardinality, card_2 = array2->cardinality, - min_card = minimum_int32(card_1, card_2); - const int threshold = 64; // subject to tuning -#ifdef USEAVX - if (out->capacity < min_card) { - array_container_grow(out, min_card + sizeof(__m128i) / sizeof(uint16_t), - false); - } -#else - if (out->capacity < min_card) { - array_container_grow(out, min_card, false); - } -#endif - - if (card_1 * threshold < card_2) { - out->cardinality = intersect_skewed_uint16( - array1->array, card_1, array2->array, card_2, out->array); - } else if (card_2 * threshold < card_1) { - out->cardinality = intersect_skewed_uint16( - array2->array, card_2, array1->array, card_1, out->array); - } else { -#ifdef USEAVX - out->cardinality = intersect_vector16( - array1->array, card_1, array2->array, card_2, out->array); -#else - out->cardinality = intersect_uint16(array1->array, card_1, - array2->array, card_2, out->array); -#endif - } -} - -/* computes the size of the intersection of array1 and array2 - * */ -int array_container_intersection_cardinality(const array_container_t *array1, - const array_container_t *array2) { - int32_t card_1 = array1->cardinality, card_2 = array2->cardinality; - const int threshold = 64; // subject to tuning - if (card_1 * threshold < card_2) { - return intersect_skewed_uint16_cardinality(array1->array, card_1, - array2->array, card_2); - } else if (card_2 * threshold < card_1) { - return intersect_skewed_uint16_cardinality(array2->array, card_2, - array1->array, card_1); - } else { -#ifdef USEAVX - return intersect_vector16_cardinality(array1->array, card_1, - array2->array, card_2); -#else - return intersect_uint16_cardinality(array1->array, card_1, - array2->array, card_2); -#endif - } -} - -bool array_container_intersect(const array_container_t *array1, - const array_container_t *array2) { - int32_t card_1 = array1->cardinality, card_2 = array2->cardinality; - const int threshold = 64; // subject to tuning - if (card_1 * threshold < card_2) { - return intersect_skewed_uint16_nonempty( - array1->array, card_1, array2->array, card_2); - } else if (card_2 * threshold < card_1) { - return intersect_skewed_uint16_nonempty( - array2->array, card_2, array1->array, card_1); - } else { - // we do not bother vectorizing - return intersect_uint16_nonempty(array1->array, card_1, - array2->array, card_2); - } -} - -/* computes the intersection of array1 and array2 and write the result to - * array1. - * */ -void array_container_intersection_inplace(array_container_t *src_1, - const array_container_t *src_2) { - // todo: can any of this be vectorized? - int32_t card_1 = src_1->cardinality, card_2 = src_2->cardinality; - const int threshold = 64; // subject to tuning - if (card_1 * threshold < card_2) { - src_1->cardinality = intersect_skewed_uint16( - src_1->array, card_1, src_2->array, card_2, src_1->array); - } else if (card_2 * threshold < card_1) { - src_1->cardinality = intersect_skewed_uint16( - src_2->array, card_2, src_1->array, card_1, src_1->array); - } else { - src_1->cardinality = intersect_uint16( - src_1->array, card_1, src_2->array, card_2, src_1->array); - } -} - -int array_container_to_uint32_array(void *vout, const array_container_t *cont, - uint32_t base) { - int outpos = 0; - uint32_t *out = (uint32_t *)vout; - for (int i = 0; i < cont->cardinality; ++i) { - const uint32_t val = base + cont->array[i]; - memcpy(out + outpos, &val, - sizeof(uint32_t)); // should be compiled as a MOV on x64 - outpos++; - } - return outpos; -} - -void array_container_printf(const array_container_t *v) { - if (v->cardinality == 0) { - printf("{}"); - return; - } - printf("{"); - printf("%d", v->array[0]); - for (int i = 1; i < v->cardinality; ++i) { - printf(",%d", v->array[i]); - } - printf("}"); -} - -void array_container_printf_as_uint32_array(const array_container_t *v, - uint32_t base) { - if (v->cardinality == 0) { - return; - } - printf("%u", v->array[0] + base); - for (int i = 1; i < v->cardinality; ++i) { - printf(",%u", v->array[i] + base); - } -} - -/* Compute the number of runs */ -int32_t array_container_number_of_runs(const array_container_t *a) { - // Can SIMD work here? - int32_t nr_runs = 0; - int32_t prev = -2; - for (const uint16_t *p = a->array; p != a->array + a->cardinality; ++p) { - if (*p != prev + 1) nr_runs++; - prev = *p; - } - return nr_runs; -} - -int32_t array_container_serialize(const array_container_t *container, char *buf) { - int32_t l, off; - uint16_t cardinality = (uint16_t)container->cardinality; - - memcpy(buf, &cardinality, off = sizeof(cardinality)); - l = sizeof(uint16_t) * container->cardinality; - if (l) memcpy(&buf[off], container->array, l); - - return (off + l); -} - -/** - * Writes the underlying array to buf, outputs how many bytes were written. - * The number of bytes written should be - * array_container_size_in_bytes(container). - * - */ -int32_t array_container_write(const array_container_t *container, char *buf) { - memcpy(buf, container->array, container->cardinality * sizeof(uint16_t)); - return array_container_size_in_bytes(container); -} - -bool array_container_equals(const array_container_t *container1, - const array_container_t *container2) { - if (container1->cardinality != container2->cardinality) { - return false; - } - // could be vectorized: - for (int32_t i = 0; i < container1->cardinality; ++i) { - if (container1->array[i] != container2->array[i]) return false; - } - return true; -} - -bool array_container_is_subset(const array_container_t *container1, - const array_container_t *container2) { - if (container1->cardinality > container2->cardinality) { - return false; - } - int i1 = 0, i2 = 0; - while (i1 < container1->cardinality && i2 < container2->cardinality) { - if (container1->array[i1] == container2->array[i2]) { - i1++; - i2++; - } else if (container1->array[i1] > container2->array[i2]) { - i2++; - } else { // container1->array[i1] < container2->array[i2] - return false; - } - } - if (i1 == container1->cardinality) { - return true; - } else { - return false; - } -} - -int32_t array_container_read(int32_t cardinality, array_container_t *container, - const char *buf) { - if (container->capacity < cardinality) { - array_container_grow(container, cardinality, false); - } - container->cardinality = cardinality; - memcpy(container->array, buf, container->cardinality * sizeof(uint16_t)); - - return array_container_size_in_bytes(container); -} - -uint32_t array_container_serialization_len(const array_container_t *container) { - return (sizeof(uint16_t) /* container->cardinality converted to 16 bit */ + - (sizeof(uint16_t) * container->cardinality)); -} - -void *array_container_deserialize(const char *buf, size_t buf_len) { - array_container_t *ptr; - - if (buf_len < 2) /* capacity converted to 16 bit */ - return (NULL); - else - buf_len -= 2; - - if ((ptr = (array_container_t *)malloc(sizeof(array_container_t))) != - NULL) { - size_t len; - int32_t off; - uint16_t cardinality; - - memcpy(&cardinality, buf, off = sizeof(cardinality)); - - ptr->capacity = ptr->cardinality = (uint32_t)cardinality; - len = sizeof(uint16_t) * ptr->cardinality; - - if (len != buf_len) { - free(ptr); - return (NULL); - } - - if ((ptr->array = (uint16_t *)malloc(sizeof(uint16_t) * - ptr->capacity)) == NULL) { - free(ptr); - return (NULL); - } - - if (len) memcpy(ptr->array, &buf[off], len); - - /* Check if returned values are monotonically increasing */ - for (int32_t i = 0, j = 0; i < ptr->cardinality; i++) { - if (ptr->array[i] < j) { - free(ptr->array); - free(ptr); - return (NULL); - } else - j = ptr->array[i]; - } - } - - return (ptr); -} - -bool array_container_iterate(const array_container_t *cont, uint32_t base, - roaring_iterator iterator, void *ptr) { - for (int i = 0; i < cont->cardinality; i++) - if (!iterator(cont->array[i] + base, ptr)) return false; - return true; -} - -bool array_container_iterate64(const array_container_t *cont, uint32_t base, - roaring_iterator64 iterator, uint64_t high_bits, - void *ptr) { - for (int i = 0; i < cont->cardinality; i++) - if (!iterator(high_bits | (uint64_t)(cont->array[i] + base), ptr)) - return false; - return true; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/array.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/bitset.c */ -/* - * bitset.c - * - */ -#ifndef _POSIX_C_SOURCE -#define _POSIX_C_SOURCE 200809L -#endif -#include -#include -#include -#include - - -extern int bitset_container_cardinality(const bitset_container_t *bitset); -extern bool bitset_container_nonzero_cardinality(bitset_container_t *bitset); -extern void bitset_container_set(bitset_container_t *bitset, uint16_t pos); -extern void bitset_container_unset(bitset_container_t *bitset, uint16_t pos); -extern inline bool bitset_container_get(const bitset_container_t *bitset, - uint16_t pos); -extern int32_t bitset_container_serialized_size_in_bytes(); -extern bool bitset_container_add(bitset_container_t *bitset, uint16_t pos); -extern bool bitset_container_remove(bitset_container_t *bitset, uint16_t pos); -extern inline bool bitset_container_contains(const bitset_container_t *bitset, - uint16_t pos); - -void bitset_container_clear(bitset_container_t *bitset) { - memset(bitset->array, 0, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); - bitset->cardinality = 0; -} - -void bitset_container_set_all(bitset_container_t *bitset) { - memset(bitset->array, INT64_C(-1), - sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); - bitset->cardinality = (1 << 16); -} - - - -/* Create a new bitset. Return NULL in case of failure. */ -bitset_container_t *bitset_container_create(void) { - bitset_container_t *bitset = - (bitset_container_t *)malloc(sizeof(bitset_container_t)); - - if (!bitset) { - return NULL; - } - // sizeof(__m256i) == 32 - bitset->array = (uint64_t *)aligned_malloc( - 32, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); - if (!bitset->array) { - free(bitset); - return NULL; - } - bitset_container_clear(bitset); - return bitset; -} - -/* Copy one container into another. We assume that they are distinct. */ -void bitset_container_copy(const bitset_container_t *source, - bitset_container_t *dest) { - dest->cardinality = source->cardinality; - memcpy(dest->array, source->array, - sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); -} - -void bitset_container_add_from_range(bitset_container_t *bitset, uint32_t min, - uint32_t max, uint16_t step) { - if (step == 0) return; // refuse to crash - if ((64 % step) == 0) { // step divides 64 - uint64_t mask = 0; // construct the repeated mask - for (uint32_t value = (min % step); value < 64; value += step) { - mask |= ((uint64_t)1 << value); - } - uint32_t firstword = min / 64; - uint32_t endword = (max - 1) / 64; - bitset->cardinality = (max - min + step - 1) / step; - if (firstword == endword) { - bitset->array[firstword] |= - mask & (((~UINT64_C(0)) << (min % 64)) & - ((~UINT64_C(0)) >> ((~max + 1) % 64))); - return; - } - bitset->array[firstword] = mask & ((~UINT64_C(0)) << (min % 64)); - for (uint32_t i = firstword + 1; i < endword; i++) - bitset->array[i] = mask; - bitset->array[endword] = mask & ((~UINT64_C(0)) >> ((~max + 1) % 64)); - } else { - for (uint32_t value = min; value < max; value += step) { - bitset_container_add(bitset, value); - } - } -} - -/* Free memory. */ -void bitset_container_free(bitset_container_t *bitset) { - if(bitset->array != NULL) {// Jon Strabala reports that some tools complain otherwise - aligned_free(bitset->array); - bitset->array = NULL; // pedantic - } - free(bitset); -} - -/* duplicate container. */ -bitset_container_t *bitset_container_clone(const bitset_container_t *src) { - bitset_container_t *bitset = - (bitset_container_t *)malloc(sizeof(bitset_container_t)); - - if (!bitset) { - return NULL; - } - // sizeof(__m256i) == 32 - bitset->array = (uint64_t *)aligned_malloc( - 32, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); - if (!bitset->array) { - free(bitset); - return NULL; - } - bitset->cardinality = src->cardinality; - memcpy(bitset->array, src->array, - sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); - return bitset; -} - -void bitset_container_set_range(bitset_container_t *bitset, uint32_t begin, - uint32_t end) { - bitset_set_range(bitset->array, begin, end); - bitset->cardinality = - bitset_container_compute_cardinality(bitset); // could be smarter -} - - -bool bitset_container_intersect(const bitset_container_t *src_1, - const bitset_container_t *src_2) { - // could vectorize, but this is probably already quite fast in practice - const uint64_t * __restrict__ array_1 = src_1->array; - const uint64_t * __restrict__ array_2 = src_2->array; - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i ++) { - if((array_1[i] & array_2[i]) != 0) return true; - } - return false; -} - - -#ifdef USEAVX -#ifndef WORDS_IN_AVX2_REG -#define WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) -#endif -/* Get the number of bits set (force computation) */ -int bitset_container_compute_cardinality(const bitset_container_t *bitset) { - return (int) avx2_harley_seal_popcount256( - (const __m256i *)bitset->array, - BITSET_CONTAINER_SIZE_IN_WORDS / (WORDS_IN_AVX2_REG)); -} -#else - -/* Get the number of bits set (force computation) */ -int bitset_container_compute_cardinality(const bitset_container_t *bitset) { - const uint64_t *array = bitset->array; - int32_t sum = 0; - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i += 4) { - sum += hamming(array[i]); - sum += hamming(array[i + 1]); - sum += hamming(array[i + 2]); - sum += hamming(array[i + 3]); - } - return sum; -} - -#endif - -#ifdef USEAVX - -#define BITSET_CONTAINER_FN_REPEAT 8 -#ifndef WORDS_IN_AVX2_REG -#define WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) -#endif -#define LOOP_SIZE \ - BITSET_CONTAINER_SIZE_IN_WORDS / \ - ((WORDS_IN_AVX2_REG)*BITSET_CONTAINER_FN_REPEAT) - -/* Computes a binary operation (eg union) on bitset1 and bitset2 and write the - result to bitsetout */ -// clang-format off -#define BITSET_CONTAINER_FN(opname, opsymbol, avx_intrinsic) \ -int bitset_container_##opname##_nocard(const bitset_container_t *src_1, \ - const bitset_container_t *src_2, \ - bitset_container_t *dst) { \ - const uint8_t * __restrict__ array_1 = (const uint8_t *)src_1->array; \ - const uint8_t * __restrict__ array_2 = (const uint8_t *)src_2->array; \ - /* not using the blocking optimization for some reason*/ \ - uint8_t *out = (uint8_t*)dst->array; \ - const int innerloop = 8; \ - for (size_t i = 0; \ - i < BITSET_CONTAINER_SIZE_IN_WORDS / (WORDS_IN_AVX2_REG); \ - i+=innerloop) {\ - __m256i A1, A2, AO; \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)out, AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 32)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 32)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+32), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 64)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 64)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+64), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 96)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 96)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+96), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 128)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 128)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+128), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 160)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 160)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+160), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 192)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 192)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+192), AO); \ - A1 = _mm256_lddqu_si256((const __m256i *)(array_1 + 224)); \ - A2 = _mm256_lddqu_si256((const __m256i *)(array_2 + 224)); \ - AO = avx_intrinsic(A2, A1); \ - _mm256_storeu_si256((__m256i *)(out+224), AO); \ - out+=256; \ - array_1 += 256; \ - array_2 += 256; \ - } \ - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; \ - return dst->cardinality; \ -} \ -/* next, a version that updates cardinality*/ \ -int bitset_container_##opname(const bitset_container_t *src_1, \ - const bitset_container_t *src_2, \ - bitset_container_t *dst) { \ - const __m256i * __restrict__ array_1 = (const __m256i *) src_1->array; \ - const __m256i * __restrict__ array_2 = (const __m256i *) src_2->array; \ - __m256i *out = (__m256i *) dst->array; \ - dst->cardinality = (int32_t)avx2_harley_seal_popcount256andstore_##opname(array_2,\ - array_1, out,BITSET_CONTAINER_SIZE_IN_WORDS / (WORDS_IN_AVX2_REG));\ - return dst->cardinality; \ -} \ -/* next, a version that just computes the cardinality*/ \ -int bitset_container_##opname##_justcard(const bitset_container_t *src_1, \ - const bitset_container_t *src_2) { \ - const __m256i * __restrict__ data1 = (const __m256i *) src_1->array; \ - const __m256i * __restrict__ data2 = (const __m256i *) src_2->array; \ - return (int)avx2_harley_seal_popcount256_##opname(data2, \ - data1, BITSET_CONTAINER_SIZE_IN_WORDS / (WORDS_IN_AVX2_REG));\ -} - - - -#else /* not USEAVX */ - -#define BITSET_CONTAINER_FN(opname, opsymbol, avxintrinsic) \ -int bitset_container_##opname(const bitset_container_t *src_1, \ - const bitset_container_t *src_2, \ - bitset_container_t *dst) { \ - const uint64_t * __restrict__ array_1 = src_1->array; \ - const uint64_t * __restrict__ array_2 = src_2->array; \ - uint64_t *out = dst->array; \ - int32_t sum = 0; \ - for (size_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i += 2) { \ - const uint64_t word_1 = (array_1[i])opsymbol(array_2[i]), \ - word_2 = (array_1[i + 1])opsymbol(array_2[i + 1]); \ - out[i] = word_1; \ - out[i + 1] = word_2; \ - sum += hamming(word_1); \ - sum += hamming(word_2); \ - } \ - dst->cardinality = sum; \ - return dst->cardinality; \ -} \ -int bitset_container_##opname##_nocard(const bitset_container_t *src_1, \ - const bitset_container_t *src_2, \ - bitset_container_t *dst) { \ - const uint64_t * __restrict__ array_1 = src_1->array; \ - const uint64_t * __restrict__ array_2 = src_2->array; \ - uint64_t *out = dst->array; \ - for (size_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i++) { \ - out[i] = (array_1[i])opsymbol(array_2[i]); \ - } \ - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; \ - return dst->cardinality; \ -} \ -int bitset_container_##opname##_justcard(const bitset_container_t *src_1, \ - const bitset_container_t *src_2) { \ - const uint64_t * __restrict__ array_1 = src_1->array; \ - const uint64_t * __restrict__ array_2 = src_2->array; \ - int32_t sum = 0; \ - for (size_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i += 2) { \ - const uint64_t word_1 = (array_1[i])opsymbol(array_2[i]), \ - word_2 = (array_1[i + 1])opsymbol(array_2[i + 1]); \ - sum += hamming(word_1); \ - sum += hamming(word_2); \ - } \ - return sum; \ -} - -#endif - -// we duplicate the function because other containers use the "or" term, makes API more consistent -BITSET_CONTAINER_FN(or, |, _mm256_or_si256) -BITSET_CONTAINER_FN(union, |, _mm256_or_si256) - -// we duplicate the function because other containers use the "intersection" term, makes API more consistent -BITSET_CONTAINER_FN(and, &, _mm256_and_si256) -BITSET_CONTAINER_FN(intersection, &, _mm256_and_si256) - -BITSET_CONTAINER_FN(xor, ^, _mm256_xor_si256) -BITSET_CONTAINER_FN(andnot, &~, _mm256_andnot_si256) -// clang-format On - - - -int bitset_container_to_uint32_array( void *vout, const bitset_container_t *cont, uint32_t base) { -#ifdef USEAVX2FORDECODING - if(cont->cardinality >= 8192)// heuristic - return (int) bitset_extract_setbits_avx2(cont->array, BITSET_CONTAINER_SIZE_IN_WORDS, vout,cont->cardinality,base); - else - return (int) bitset_extract_setbits(cont->array, BITSET_CONTAINER_SIZE_IN_WORDS, vout,base); -#else - return (int) bitset_extract_setbits(cont->array, BITSET_CONTAINER_SIZE_IN_WORDS, vout,base); -#endif -} - -/* - * Print this container using printf (useful for debugging). - */ -void bitset_container_printf(const bitset_container_t * v) { - printf("{"); - uint32_t base = 0; - bool iamfirst = true;// TODO: rework so that this is not necessary yet still readable - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { - uint64_t w = v->array[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - if(iamfirst) {// predicted to be false - printf("%u",base + r); - iamfirst = false; - } else { - printf(",%u",base + r); - } - w ^= t; - } - base += 64; - } - printf("}"); -} - - -/* - * Print this container using printf as a comma-separated list of 32-bit integers starting at base. - */ -void bitset_container_printf_as_uint32_array(const bitset_container_t * v, uint32_t base) { - bool iamfirst = true;// TODO: rework so that this is not necessary yet still readable - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { - uint64_t w = v->array[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - if(iamfirst) {// predicted to be false - printf("%u", r + base); - iamfirst = false; - } else { - printf(",%u",r + base); - } - w ^= t; - } - base += 64; - } -} - - -// TODO: use the fast lower bound, also -int bitset_container_number_of_runs(bitset_container_t *b) { - int num_runs = 0; - uint64_t next_word = b->array[0]; - - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS-1; ++i) { - uint64_t word = next_word; - next_word = b->array[i+1]; - num_runs += hamming((~word) & (word << 1)) + ( (word >> 63) & ~next_word); - } - - uint64_t word = next_word; - num_runs += hamming((~word) & (word << 1)); - if((word & 0x8000000000000000ULL) != 0) - num_runs++; - return num_runs; -} - -int32_t bitset_container_serialize(const bitset_container_t *container, char *buf) { - int32_t l = sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS; - memcpy(buf, container->array, l); - return(l); -} - - - -int32_t bitset_container_write(const bitset_container_t *container, - char *buf) { - memcpy(buf, container->array, BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t)); - return bitset_container_size_in_bytes(container); -} - - -int32_t bitset_container_read(int32_t cardinality, bitset_container_t *container, - const char *buf) { - container->cardinality = cardinality; - memcpy(container->array, buf, BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t)); - return bitset_container_size_in_bytes(container); -} - -uint32_t bitset_container_serialization_len() { - return(sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); -} - -void* bitset_container_deserialize(const char *buf, size_t buf_len) { - bitset_container_t *ptr; - size_t l = sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS; - - if(l != buf_len) - return(NULL); - - if((ptr = (bitset_container_t *)malloc(sizeof(bitset_container_t))) != NULL) { - memcpy(ptr, buf, sizeof(bitset_container_t)); - // sizeof(__m256i) == 32 - ptr->array = (uint64_t *) aligned_malloc(32, l); - if (! ptr->array) { - free(ptr); - return NULL; - } - memcpy(ptr->array, buf, l); - ptr->cardinality = bitset_container_compute_cardinality(ptr); - } - - return((void*)ptr); -} - -bool bitset_container_iterate(const bitset_container_t *cont, uint32_t base, roaring_iterator iterator, void *ptr) { - for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { - uint64_t w = cont->array[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - if(!iterator(r + base, ptr)) return false; - w ^= t; - } - base += 64; - } - return true; -} - -bool bitset_container_iterate64(const bitset_container_t *cont, uint32_t base, roaring_iterator64 iterator, uint64_t high_bits, void *ptr) { - for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { - uint64_t w = cont->array[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - if(!iterator(high_bits | (uint64_t)(r + base), ptr)) return false; - w ^= t; - } - base += 64; - } - return true; -} - - -bool bitset_container_equals(const bitset_container_t *container1, const bitset_container_t *container2) { - if((container1->cardinality != BITSET_UNKNOWN_CARDINALITY) && (container2->cardinality != BITSET_UNKNOWN_CARDINALITY)) { - if(container1->cardinality != container2->cardinality) { - return false; - } - } - for(int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { - if(container1->array[i] != container2->array[i]) { - return false; - } - } - return true; -} - -bool bitset_container_is_subset(const bitset_container_t *container1, - const bitset_container_t *container2) { - if((container1->cardinality != BITSET_UNKNOWN_CARDINALITY) && (container2->cardinality != BITSET_UNKNOWN_CARDINALITY)) { - if(container1->cardinality > container2->cardinality) { - return false; - } - } - for(int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { - if((container1->array[i] & container2->array[i]) != container1->array[i]) { - return false; - } - } - return true; -} - -bool bitset_container_select(const bitset_container_t *container, uint32_t *start_rank, uint32_t rank, uint32_t *element) { - int card = bitset_container_cardinality(container); - if(rank >= *start_rank + card) { - *start_rank += card; - return false; - } - const uint64_t *array = container->array; - int32_t size; - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i += 1) { - size = hamming(array[i]); - if(rank <= *start_rank + size) { - uint64_t w = container->array[i]; - uint16_t base = i*64; - while (w != 0) { - uint64_t t = w & (~w + 1); - int r = __builtin_ctzll(w); - if(*start_rank == rank) { - *element = r+base; - return true; - } - w ^= t; - *start_rank += 1; - } - } - else - *start_rank += size; - } - assert(false); - __builtin_unreachable(); -} - - -/* Returns the smallest value (assumes not empty) */ -uint16_t bitset_container_minimum(const bitset_container_t *container) { - for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { - uint64_t w = container->array[i]; - if (w != 0) { - int r = __builtin_ctzll(w); - return r + i * 64; - } - } - return UINT16_MAX; -} - -/* Returns the largest value (assumes not empty) */ -uint16_t bitset_container_maximum(const bitset_container_t *container) { - for (int32_t i = BITSET_CONTAINER_SIZE_IN_WORDS - 1; i > 0; --i ) { - uint64_t w = container->array[i]; - if (w != 0) { - int r = __builtin_clzll(w); - return i * 64 + 63 - r; - } - } - return 0; -} - -/* Returns the number of values equal or smaller than x */ -int bitset_container_rank(const bitset_container_t *container, uint16_t x) { - uint32_t x32 = x; - int sum = 0; - uint32_t k = 0; - for (; k + 63 <= x32; k += 64) { - sum += hamming(container->array[k / 64]); - } - // at this point, we have covered everything up to k, k not included. - // we have that k < x, but not so large that k+63<=x - // k is a power of 64 - int bitsleft = x32 - k + 1;// will be in [0,64) - uint64_t leftoverword = container->array[k / 64];// k / 64 should be within scope - leftoverword = leftoverword & ((UINT64_C(1) << bitsleft) - 1); - sum += hamming(leftoverword); - return sum; -} - -/* Returns the index of the first value equal or larger than x, or -1 */ -int bitset_container_index_equalorlarger(const bitset_container_t *container, uint16_t x) { - uint32_t x32 = x; - uint32_t k = x32 / 64; - uint64_t word = container->array[k]; - const int diff = x32 - k * 64; // in [0,64) - word = (word >> diff) << diff; // a mask is faster, but we don't care - while(word == 0) { - k++; - if(k == BITSET_CONTAINER_SIZE_IN_WORDS) return -1; - word = container->array[k]; - } - return k * 64 + __builtin_ctzll(word); -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/bitset.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/containers.c */ - - -extern inline const void *container_unwrap_shared( - const void *candidate_shared_container, uint8_t *type); -extern inline void *container_mutable_unwrap_shared( - void *candidate_shared_container, uint8_t *type); - -extern const char *get_container_name(uint8_t typecode); - -extern int container_get_cardinality(const void *container, uint8_t typecode); - -extern void *container_iand(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_ior(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_ixor(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_iandnot(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -void container_free(void *container, uint8_t typecode) { - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - bitset_container_free((bitset_container_t *)container); - break; - case ARRAY_CONTAINER_TYPE_CODE: - array_container_free((array_container_t *)container); - break; - case RUN_CONTAINER_TYPE_CODE: - run_container_free((run_container_t *)container); - break; - case SHARED_CONTAINER_TYPE_CODE: - shared_container_free((shared_container_t *)container); - break; - default: - assert(false); - __builtin_unreachable(); - } -} - -void container_printf(const void *container, uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - bitset_container_printf((const bitset_container_t *)container); - return; - case ARRAY_CONTAINER_TYPE_CODE: - array_container_printf((const array_container_t *)container); - return; - case RUN_CONTAINER_TYPE_CODE: - run_container_printf((const run_container_t *)container); - return; - default: - __builtin_unreachable(); - } -} - -void container_printf_as_uint32_array(const void *container, uint8_t typecode, - uint32_t base) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - bitset_container_printf_as_uint32_array( - (const bitset_container_t *)container, base); - return; - case ARRAY_CONTAINER_TYPE_CODE: - array_container_printf_as_uint32_array( - (const array_container_t *)container, base); - return; - case RUN_CONTAINER_TYPE_CODE: - run_container_printf_as_uint32_array( - (const run_container_t *)container, base); - return; - return; - default: - __builtin_unreachable(); - } -} - -int32_t container_serialize(const void *container, uint8_t typecode, - char *buf) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return (bitset_container_serialize((const bitset_container_t *)container, - buf)); - case ARRAY_CONTAINER_TYPE_CODE: - return ( - array_container_serialize((const array_container_t *)container, buf)); - case RUN_CONTAINER_TYPE_CODE: - return (run_container_serialize((const run_container_t *)container, buf)); - default: - assert(0); - __builtin_unreachable(); - return (-1); - } -} - -uint32_t container_serialization_len(const void *container, uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_serialization_len(); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_serialization_len( - (const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_serialization_len( - (const run_container_t *)container); - default: - assert(0); - __builtin_unreachable(); - return (0); - } -} - -void *container_deserialize(uint8_t typecode, const char *buf, size_t buf_len) { - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return (bitset_container_deserialize(buf, buf_len)); - case ARRAY_CONTAINER_TYPE_CODE: - return (array_container_deserialize(buf, buf_len)); - case RUN_CONTAINER_TYPE_CODE: - return (run_container_deserialize(buf, buf_len)); - case SHARED_CONTAINER_TYPE_CODE: - printf("this should never happen.\n"); - assert(0); - __builtin_unreachable(); - return (NULL); - default: - assert(0); - __builtin_unreachable(); - return (NULL); - } -} - -extern bool container_nonzero_cardinality(const void *container, - uint8_t typecode); - -extern void container_free(void *container, uint8_t typecode); - -extern int container_to_uint32_array(uint32_t *output, const void *container, - uint8_t typecode, uint32_t base); - -extern void *container_add(void *container, uint16_t val, uint8_t typecode, - uint8_t *new_typecode); - -extern inline bool container_contains(const void *container, uint16_t val, - uint8_t typecode); - -extern void *container_clone(const void *container, uint8_t typecode); - -extern void *container_and(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_or(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_xor(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -void *get_copy_of_container(void *container, uint8_t *typecode, - bool copy_on_write) { - if (copy_on_write) { - shared_container_t *shared_container; - if (*typecode == SHARED_CONTAINER_TYPE_CODE) { - shared_container = (shared_container_t *)container; - shared_container->counter += 1; - return shared_container; - } - assert(*typecode != SHARED_CONTAINER_TYPE_CODE); - - if ((shared_container = (shared_container_t *)malloc( - sizeof(shared_container_t))) == NULL) { - return NULL; - } - - shared_container->container = container; - shared_container->typecode = *typecode; - - shared_container->counter = 2; - *typecode = SHARED_CONTAINER_TYPE_CODE; - - return shared_container; - } // copy_on_write - // otherwise, no copy on write... - const void *actualcontainer = - container_unwrap_shared((const void *)container, typecode); - assert(*typecode != SHARED_CONTAINER_TYPE_CODE); - return container_clone(actualcontainer, *typecode); -} -/** - * Copies a container, requires a typecode. This allocates new memory, caller - * is responsible for deallocation. - */ -void *container_clone(const void *container, uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_clone((const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_clone((const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_clone((const run_container_t *)container); - case SHARED_CONTAINER_TYPE_CODE: - printf("shared containers are not cloneable\n"); - assert(false); - return NULL; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -void *shared_container_extract_copy(shared_container_t *container, - uint8_t *typecode) { - assert(container->counter > 0); - assert(container->typecode != SHARED_CONTAINER_TYPE_CODE); - container->counter--; - *typecode = container->typecode; - void *answer; - if (container->counter == 0) { - answer = container->container; - container->container = NULL; // paranoid - free(container); - } else { - answer = container_clone(container->container, *typecode); - } - assert(*typecode != SHARED_CONTAINER_TYPE_CODE); - return answer; -} - -void shared_container_free(shared_container_t *container) { - assert(container->counter > 0); - container->counter--; - if (container->counter == 0) { - assert(container->typecode != SHARED_CONTAINER_TYPE_CODE); - container_free(container->container, container->typecode); - container->container = NULL; // paranoid - free(container); - } -} - -extern void *container_not(const void *c1, uint8_t type1, uint8_t *result_type); - -extern void *container_not_range(const void *c1, uint8_t type1, - uint32_t range_start, uint32_t range_end, - uint8_t *result_type); - -extern void *container_inot(void *c1, uint8_t type1, uint8_t *result_type); - -extern void *container_inot_range(void *c1, uint8_t type1, uint32_t range_start, - uint32_t range_end, uint8_t *result_type); - -extern void *container_range_of_ones(uint32_t range_start, uint32_t range_end, - uint8_t *result_type); - -// where are the correponding things for union and intersection?? -extern void *container_lazy_xor(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_lazy_ixor(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); - -extern void *container_andnot(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type); -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/containers.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/convert.c */ -#include - - -// file contains grubby stuff that must know impl. details of all container -// types. -bitset_container_t *bitset_container_from_array(const array_container_t *a) { - bitset_container_t *ans = bitset_container_create(); - int limit = array_container_cardinality(a); - for (int i = 0; i < limit; ++i) bitset_container_set(ans, a->array[i]); - return ans; -} - -bitset_container_t *bitset_container_from_run(const run_container_t *arr) { - int card = run_container_cardinality(arr); - bitset_container_t *answer = bitset_container_create(); - for (int rlepos = 0; rlepos < arr->n_runs; ++rlepos) { - rle16_t vl = arr->runs[rlepos]; - bitset_set_lenrange(answer->array, vl.value, vl.length); - } - answer->cardinality = card; - return answer; -} - -array_container_t *array_container_from_run(const run_container_t *arr) { - array_container_t *answer = - array_container_create_given_capacity(run_container_cardinality(arr)); - answer->cardinality = 0; - for (int rlepos = 0; rlepos < arr->n_runs; ++rlepos) { - int run_start = arr->runs[rlepos].value; - int run_end = run_start + arr->runs[rlepos].length; - - for (int run_value = run_start; run_value <= run_end; ++run_value) { - answer->array[answer->cardinality++] = (uint16_t)run_value; - } - } - return answer; -} - -array_container_t *array_container_from_bitset(const bitset_container_t *bits) { - array_container_t *result = - array_container_create_given_capacity(bits->cardinality); - result->cardinality = bits->cardinality; - // sse version ends up being slower here - // (bitset_extract_setbits_sse_uint16) - // because of the sparsity of the data - bitset_extract_setbits_uint16(bits->array, BITSET_CONTAINER_SIZE_IN_WORDS, - result->array, 0); - return result; -} - -/* assumes that container has adequate space. Run from [s,e] (inclusive) */ -static void add_run(run_container_t *r, int s, int e) { - r->runs[r->n_runs].value = s; - r->runs[r->n_runs].length = e - s; - r->n_runs++; -} - -run_container_t *run_container_from_array(const array_container_t *c) { - int32_t n_runs = array_container_number_of_runs(c); - run_container_t *answer = run_container_create_given_capacity(n_runs); - int prev = -2; - int run_start = -1; - int32_t card = c->cardinality; - if (card == 0) return answer; - for (int i = 0; i < card; ++i) { - const uint16_t cur_val = c->array[i]; - if (cur_val != prev + 1) { - // new run starts; flush old one, if any - if (run_start != -1) add_run(answer, run_start, prev); - run_start = cur_val; - } - prev = c->array[i]; - } - // now prev is the last seen value - add_run(answer, run_start, prev); - // assert(run_container_cardinality(answer) == c->cardinality); - return answer; -} - -/** - * Convert the runcontainer to either a Bitmap or an Array Container, depending - * on the cardinality. Frees the container. - * Allocates and returns new container, which caller is responsible for freeing - */ - -void *convert_to_bitset_or_array_container(run_container_t *r, int32_t card, - uint8_t *resulttype) { - if (card <= DEFAULT_MAX_SIZE) { - array_container_t *answer = array_container_create_given_capacity(card); - answer->cardinality = 0; - for (int rlepos = 0; rlepos < r->n_runs; ++rlepos) { - uint16_t run_start = r->runs[rlepos].value; - uint16_t run_end = run_start + r->runs[rlepos].length; - for (uint16_t run_value = run_start; run_value <= run_end; - ++run_value) { - answer->array[answer->cardinality++] = run_value; - } - } - assert(card == answer->cardinality); - *resulttype = ARRAY_CONTAINER_TYPE_CODE; - run_container_free(r); - return answer; - } - bitset_container_t *answer = bitset_container_create(); - for (int rlepos = 0; rlepos < r->n_runs; ++rlepos) { - uint16_t run_start = r->runs[rlepos].value; - bitset_set_lenrange(answer->array, run_start, r->runs[rlepos].length); - } - answer->cardinality = card; - *resulttype = BITSET_CONTAINER_TYPE_CODE; - run_container_free(r); - return answer; -} - -/* Converts a run container to either an array or a bitset, IF it saves space. - */ -/* If a conversion occurs, the caller is responsible to free the original - * container and - * he becomes responsible to free the new one. */ -void *convert_run_to_efficient_container(run_container_t *c, - uint8_t *typecode_after) { - int32_t size_as_run_container = - run_container_serialized_size_in_bytes(c->n_runs); - - int32_t size_as_bitset_container = - bitset_container_serialized_size_in_bytes(); - int32_t card = run_container_cardinality(c); - int32_t size_as_array_container = - array_container_serialized_size_in_bytes(card); - - int32_t min_size_non_run = - size_as_bitset_container < size_as_array_container - ? size_as_bitset_container - : size_as_array_container; - if (size_as_run_container <= min_size_non_run) { // no conversion - *typecode_after = RUN_CONTAINER_TYPE_CODE; - return c; - } - if (card <= DEFAULT_MAX_SIZE) { - // to array - array_container_t *answer = array_container_create_given_capacity(card); - answer->cardinality = 0; - for (int rlepos = 0; rlepos < c->n_runs; ++rlepos) { - int run_start = c->runs[rlepos].value; - int run_end = run_start + c->runs[rlepos].length; - - for (int run_value = run_start; run_value <= run_end; ++run_value) { - answer->array[answer->cardinality++] = (uint16_t)run_value; - } - } - *typecode_after = ARRAY_CONTAINER_TYPE_CODE; - return answer; - } - - // else to bitset - bitset_container_t *answer = bitset_container_create(); - - for (int rlepos = 0; rlepos < c->n_runs; ++rlepos) { - int start = c->runs[rlepos].value; - int end = start + c->runs[rlepos].length; - bitset_set_range(answer->array, start, end + 1); - } - answer->cardinality = card; - *typecode_after = BITSET_CONTAINER_TYPE_CODE; - return answer; -} - -// like convert_run_to_efficient_container but frees the old result if needed -void *convert_run_to_efficient_container_and_free(run_container_t *c, - uint8_t *typecode_after) { - void *answer = convert_run_to_efficient_container(c, typecode_after); - if (answer != c) run_container_free(c); - return answer; -} - -/* once converted, the original container is disposed here, rather than - in roaring_array -*/ - -// TODO: split into run- array- and bitset- subfunctions for sanity; -// a few function calls won't really matter. - -void *convert_run_optimize(void *c, uint8_t typecode_original, - uint8_t *typecode_after) { - if (typecode_original == RUN_CONTAINER_TYPE_CODE) { - void *newc = convert_run_to_efficient_container((run_container_t *)c, - typecode_after); - if (newc != c) { - container_free(c, typecode_original); - } - return newc; - } else if (typecode_original == ARRAY_CONTAINER_TYPE_CODE) { - // it might need to be converted to a run container. - array_container_t *c_qua_array = (array_container_t *)c; - int32_t n_runs = array_container_number_of_runs(c_qua_array); - int32_t size_as_run_container = - run_container_serialized_size_in_bytes(n_runs); - int32_t card = array_container_cardinality(c_qua_array); - int32_t size_as_array_container = - array_container_serialized_size_in_bytes(card); - - if (size_as_run_container >= size_as_array_container) { - *typecode_after = ARRAY_CONTAINER_TYPE_CODE; - return c; - } - // else convert array to run container - run_container_t *answer = run_container_create_given_capacity(n_runs); - int prev = -2; - int run_start = -1; - - assert(card > 0); - for (int i = 0; i < card; ++i) { - uint16_t cur_val = c_qua_array->array[i]; - if (cur_val != prev + 1) { - // new run starts; flush old one, if any - if (run_start != -1) add_run(answer, run_start, prev); - run_start = cur_val; - } - prev = c_qua_array->array[i]; - } - assert(run_start >= 0); - // now prev is the last seen value - add_run(answer, run_start, prev); - *typecode_after = RUN_CONTAINER_TYPE_CODE; - array_container_free(c_qua_array); - return answer; - } else if (typecode_original == - BITSET_CONTAINER_TYPE_CODE) { // run conversions on bitset - // does bitset need conversion to run? - bitset_container_t *c_qua_bitset = (bitset_container_t *)c; - int32_t n_runs = bitset_container_number_of_runs(c_qua_bitset); - int32_t size_as_run_container = - run_container_serialized_size_in_bytes(n_runs); - int32_t size_as_bitset_container = - bitset_container_serialized_size_in_bytes(); - - if (size_as_bitset_container <= size_as_run_container) { - // no conversion needed. - *typecode_after = BITSET_CONTAINER_TYPE_CODE; - return c; - } - // bitset to runcontainer (ported from Java RunContainer( - // BitmapContainer bc, int nbrRuns)) - assert(n_runs > 0); // no empty bitmaps - run_container_t *answer = run_container_create_given_capacity(n_runs); - - int long_ctr = 0; - uint64_t cur_word = c_qua_bitset->array[0]; - int run_count = 0; - while (true) { - while (cur_word == UINT64_C(0) && - long_ctr < BITSET_CONTAINER_SIZE_IN_WORDS - 1) - cur_word = c_qua_bitset->array[++long_ctr]; - - if (cur_word == UINT64_C(0)) { - bitset_container_free(c_qua_bitset); - *typecode_after = RUN_CONTAINER_TYPE_CODE; - return answer; - } - - int local_run_start = __builtin_ctzll(cur_word); - int run_start = local_run_start + 64 * long_ctr; - uint64_t cur_word_with_1s = cur_word | (cur_word - 1); - - int run_end = 0; - while (cur_word_with_1s == UINT64_C(0xFFFFFFFFFFFFFFFF) && - long_ctr < BITSET_CONTAINER_SIZE_IN_WORDS - 1) - cur_word_with_1s = c_qua_bitset->array[++long_ctr]; - - if (cur_word_with_1s == UINT64_C(0xFFFFFFFFFFFFFFFF)) { - run_end = 64 + long_ctr * 64; // exclusive, I guess - add_run(answer, run_start, run_end - 1); - bitset_container_free(c_qua_bitset); - *typecode_after = RUN_CONTAINER_TYPE_CODE; - return answer; - } - int local_run_end = __builtin_ctzll(~cur_word_with_1s); - run_end = local_run_end + long_ctr * 64; - add_run(answer, run_start, run_end - 1); - run_count++; - cur_word = cur_word_with_1s & (cur_word_with_1s + 1); - } - return answer; - } else { - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -bitset_container_t *bitset_container_from_run_range(const run_container_t *run, - uint32_t min, uint32_t max) { - bitset_container_t *bitset = bitset_container_create(); - int32_t union_cardinality = 0; - for (int32_t i = 0; i < run->n_runs; ++i) { - uint32_t rle_min = run->runs[i].value; - uint32_t rle_max = rle_min + run->runs[i].length; - bitset_set_lenrange(bitset->array, rle_min, rle_max - rle_min); - union_cardinality += run->runs[i].length + 1; - } - union_cardinality += max - min + 1; - union_cardinality -= bitset_lenrange_cardinality(bitset->array, min, max-min); - bitset_set_lenrange(bitset->array, min, max - min); - bitset->cardinality = union_cardinality; - return bitset; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/convert.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_andnot.c */ -/* - * mixed_andnot.c. More methods since operation is not symmetric, - * except no "wide" andnot , so no lazy options motivated. - */ - -#include -#include - - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst, a valid array container that could be the same as dst.*/ -void array_bitset_container_andnot(const array_container_t *src_1, - const bitset_container_t *src_2, - array_container_t *dst) { - // follows Java implementation as of June 2016 - if (dst->capacity < src_1->cardinality) { - array_container_grow(dst, src_1->cardinality, false); - } - int32_t newcard = 0; - const int32_t origcard = src_1->cardinality; - for (int i = 0; i < origcard; ++i) { - uint16_t key = src_1->array[i]; - dst->array[newcard] = key; - newcard += 1 - bitset_container_contains(src_2, key); - } - dst->cardinality = newcard; -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * src_1 */ - -void array_bitset_container_iandnot(array_container_t *src_1, - const bitset_container_t *src_2) { - array_bitset_container_andnot(src_1, src_2, src_1); -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst, which does not initially have a valid container. - * Return true for a bitset result; false for array - */ - -bool bitset_array_container_andnot(const bitset_container_t *src_1, - const array_container_t *src_2, void **dst) { - // Java did this directly, but we have option of asm or avx - bitset_container_t *result = bitset_container_create(); - bitset_container_copy(src_1, result); - result->cardinality = - (int32_t)bitset_clear_list(result->array, (uint64_t)result->cardinality, - src_2->array, (uint64_t)src_2->cardinality); - - // do required type conversions. - if (result->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(result); - bitset_container_free(result); - return false; - } - *dst = result; - return true; -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_array_container_iandnot(bitset_container_t *src_1, - const array_container_t *src_2, - void **dst) { - *dst = src_1; - src_1->cardinality = - (int32_t)bitset_clear_list(src_1->array, (uint64_t)src_1->cardinality, - src_2->array, (uint64_t)src_2->cardinality); - - if (src_1->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(src_1); - bitset_container_free(src_1); - return false; // not bitset - } else - return true; -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_andnot(const run_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - // follows the Java implementation as of June 2016 - int card = run_container_cardinality(src_1); - if (card <= DEFAULT_MAX_SIZE) { - // must be an array - array_container_t *answer = array_container_create_given_capacity(card); - answer->cardinality = 0; - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - for (int run_value = rle.value; run_value <= rle.value + rle.length; - ++run_value) { - if (!bitset_container_get(src_2, (uint16_t)run_value)) { - answer->array[answer->cardinality++] = (uint16_t)run_value; - } - } - } - *dst = answer; - return false; - } else { // we guess it will be a bitset, though have to check guess when - // done - bitset_container_t *answer = bitset_container_clone(src_2); - - uint32_t last_pos = 0; - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - - uint32_t start = rle.value; - uint32_t end = start + rle.length + 1; - bitset_reset_range(answer->array, last_pos, start); - bitset_flip_range(answer->array, start, end); - last_pos = end; - } - bitset_reset_range(answer->array, last_pos, (uint32_t)(1 << 16)); - - answer->cardinality = bitset_container_compute_cardinality(answer); - - if (answer->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(answer); - bitset_container_free(answer); - return false; // not bitset - } - *dst = answer; - return true; // bitset - } -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_iandnot(run_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - // dummy implementation - bool ans = run_bitset_container_andnot(src_1, src_2, dst); - run_container_free(src_1); - return ans; -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool bitset_run_container_andnot(const bitset_container_t *src_1, - const run_container_t *src_2, void **dst) { - // follows Java implementation - bitset_container_t *result = bitset_container_create(); - - bitset_container_copy(src_1, result); - for (int32_t rlepos = 0; rlepos < src_2->n_runs; ++rlepos) { - rle16_t rle = src_2->runs[rlepos]; - bitset_reset_range(result->array, rle.value, - rle.value + rle.length + UINT32_C(1)); - } - result->cardinality = bitset_container_compute_cardinality(result); - - if (result->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(result); - bitset_container_free(result); - return false; // not bitset - } - *dst = result; - return true; // bitset -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_run_container_iandnot(bitset_container_t *src_1, - const run_container_t *src_2, void **dst) { - *dst = src_1; - - for (int32_t rlepos = 0; rlepos < src_2->n_runs; ++rlepos) { - rle16_t rle = src_2->runs[rlepos]; - bitset_reset_range(src_1->array, rle.value, - rle.value + rle.length + UINT32_C(1)); - } - src_1->cardinality = bitset_container_compute_cardinality(src_1); - - if (src_1->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(src_1); - bitset_container_free(src_1); - return false; // not bitset - } else - return true; -} - -/* helper. a_out must be a valid array container with adequate capacity. - * Returns the cardinality of the output container. Partly Based on Java - * implementation Util.unsignedDifference. - * - * TODO: Util.unsignedDifference does not use advanceUntil. Is it cheaper - * to avoid advanceUntil? - */ - -static int run_array_array_subtract(const run_container_t *r, - const array_container_t *a_in, - array_container_t *a_out) { - int out_card = 0; - int32_t in_array_pos = - -1; // since advanceUntil always assumes we start the search AFTER this - - for (int rlepos = 0; rlepos < r->n_runs; rlepos++) { - int32_t start = r->runs[rlepos].value; - int32_t end = start + r->runs[rlepos].length + 1; - - in_array_pos = advanceUntil(a_in->array, in_array_pos, - a_in->cardinality, (uint16_t)start); - - if (in_array_pos >= a_in->cardinality) { // run has no items subtracted - for (int32_t i = start; i < end; ++i) - a_out->array[out_card++] = (uint16_t)i; - } else { - uint16_t next_nonincluded = a_in->array[in_array_pos]; - if (next_nonincluded >= end) { - // another case when run goes unaltered - for (int32_t i = start; i < end; ++i) - a_out->array[out_card++] = (uint16_t)i; - in_array_pos--; // ensure we see this item again if necessary - } else { - for (int32_t i = start; i < end; ++i) - if (i != next_nonincluded) - a_out->array[out_card++] = (uint16_t)i; - else // 0 should ensure we don't match - next_nonincluded = - (in_array_pos + 1 >= a_in->cardinality) - ? 0 - : a_in->array[++in_array_pos]; - in_array_pos--; // see again - } - } - } - return out_card; -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any type of container. - */ - -int run_array_container_andnot(const run_container_t *src_1, - const array_container_t *src_2, void **dst) { - // follows the Java impl as of June 2016 - - int card = run_container_cardinality(src_1); - const int arbitrary_threshold = 32; - - if (card <= arbitrary_threshold) { - if (src_2->cardinality == 0) { - *dst = run_container_clone(src_1); - return RUN_CONTAINER_TYPE_CODE; - } - // Java's "lazyandNot.toEfficientContainer" thing - run_container_t *answer = run_container_create_given_capacity( - card + array_container_cardinality(src_2)); - - int rlepos = 0; - int xrlepos = 0; // "x" is src_2 - rle16_t rle = src_1->runs[rlepos]; - int32_t start = rle.value; - int32_t end = start + rle.length + 1; - int32_t xstart = src_2->array[xrlepos]; - - while ((rlepos < src_1->n_runs) && (xrlepos < src_2->cardinality)) { - if (end <= xstart) { - // output the first run - answer->runs[answer->n_runs++] = - (rle16_t){.value = (uint16_t)start, - .length = (uint16_t)(end - start - 1)}; - rlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - } else if (xstart + 1 <= start) { - // exit the second run - xrlepos++; - if (xrlepos < src_2->cardinality) { - xstart = src_2->array[xrlepos]; - } - } else { - if (start < xstart) { - answer->runs[answer->n_runs++] = - (rle16_t){.value = (uint16_t)start, - .length = (uint16_t)(xstart - start - 1)}; - } - if (xstart + 1 < end) { - start = xstart + 1; - } else { - rlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - } - } - } - if (rlepos < src_1->n_runs) { - answer->runs[answer->n_runs++] = - (rle16_t){.value = (uint16_t)start, - .length = (uint16_t)(end - start - 1)}; - rlepos++; - if (rlepos < src_1->n_runs) { - memcpy(answer->runs + answer->n_runs, src_1->runs + rlepos, - (src_1->n_runs - rlepos) * sizeof(rle16_t)); - answer->n_runs += (src_1->n_runs - rlepos); - } - } - uint8_t return_type; - *dst = convert_run_to_efficient_container(answer, &return_type); - if (answer != *dst) run_container_free(answer); - return return_type; - } - // else it's a bitmap or array - - if (card <= DEFAULT_MAX_SIZE) { - array_container_t *ac = array_container_create_given_capacity(card); - // nb Java code used a generic iterator-based merge to compute - // difference - ac->cardinality = run_array_array_subtract(src_1, src_2, ac); - *dst = ac; - return ARRAY_CONTAINER_TYPE_CODE; - } - bitset_container_t *ans = bitset_container_from_run(src_1); - bool result_is_bitset = bitset_array_container_iandnot(ans, src_2, dst); - return (result_is_bitset ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE); -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -int run_array_container_iandnot(run_container_t *src_1, - const array_container_t *src_2, void **dst) { - // dummy implementation same as June 2016 Java - int ans = run_array_container_andnot(src_1, src_2, dst); - run_container_free(src_1); - return ans; -} - -/* dst must be a valid array container, allowed to be src_1 */ - -void array_run_container_andnot(const array_container_t *src_1, - const run_container_t *src_2, - array_container_t *dst) { - // basically following Java impl as of June 2016 - if (src_1->cardinality > dst->capacity) { - array_container_grow(dst, src_1->cardinality, false); - } - - if (src_2->n_runs == 0) { - memmove(dst->array, src_1->array, - sizeof(uint16_t) * src_1->cardinality); - dst->cardinality = src_1->cardinality; - return; - } - int32_t run_start = src_2->runs[0].value; - int32_t run_end = run_start + src_2->runs[0].length; - int which_run = 0; - - uint16_t val = 0; - int dest_card = 0; - for (int i = 0; i < src_1->cardinality; ++i) { - val = src_1->array[i]; - if (val < run_start) - dst->array[dest_card++] = val; - else if (val <= run_end) { - ; // omitted item - } else { - do { - if (which_run + 1 < src_2->n_runs) { - ++which_run; - run_start = src_2->runs[which_run].value; - run_end = run_start + src_2->runs[which_run].length; - - } else - run_start = run_end = (1 << 16) + 1; - } while (val > run_end); - --i; - } - } - dst->cardinality = dest_card; -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -void array_run_container_iandnot(array_container_t *src_1, - const run_container_t *src_2) { - array_run_container_andnot(src_1, src_2, src_1); -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int run_run_container_andnot(const run_container_t *src_1, - const run_container_t *src_2, void **dst) { - run_container_t *ans = run_container_create(); - run_container_andnot(src_1, src_2, ans); - uint8_t typecode_after; - *dst = convert_run_to_efficient_container_and_free(ans, &typecode_after); - return typecode_after; -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -int run_run_container_iandnot(run_container_t *src_1, - const run_container_t *src_2, void **dst) { - // following Java impl as of June 2016 (dummy) - int ans = run_run_container_andnot(src_1, src_2, dst); - run_container_free(src_1); - return ans; -} - -/* - * dst is a valid array container and may be the same as src_1 - */ - -void array_array_container_andnot(const array_container_t *src_1, - const array_container_t *src_2, - array_container_t *dst) { - array_container_andnot(src_1, src_2, dst); -} - -/* inplace array-array andnot will always be able to reuse the space of - * src_1 */ -void array_array_container_iandnot(array_container_t *src_1, - const array_container_t *src_2) { - array_container_andnot(src_1, src_2, src_1); -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). Return value is - * "dst is a bitset" - */ - -bool bitset_bitset_container_andnot(const bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst) { - bitset_container_t *ans = bitset_container_create(); - int card = bitset_container_andnot(src_1, src_2, ans); - if (card <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(ans); - bitset_container_free(ans); - return false; // not bitset - } else { - *dst = ans; - return true; - } -} - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_bitset_container_iandnot(bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst) { - int card = bitset_container_andnot(src_1, src_2, src_1); - if (card <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(src_1); - bitset_container_free(src_1); - return false; // not bitset - } else { - *dst = src_1; - return true; - } -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_andnot.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_equal.c */ - -bool array_container_equal_bitset(const array_container_t* container1, - const bitset_container_t* container2) { - if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { - if (container2->cardinality != container1->cardinality) { - return false; - } - } - int32_t pos = 0; - for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { - uint64_t w = container2->array[i]; - while (w != 0) { - uint64_t t = w & (~w + 1); - uint16_t r = i * 64 + __builtin_ctzll(w); - if (pos >= container1->cardinality) { - return false; - } - if (container1->array[pos] != r) { - return false; - } - ++pos; - w ^= t; - } - } - return (pos == container1->cardinality); -} - -bool run_container_equals_array(const run_container_t* container1, - const array_container_t* container2) { - if (run_container_cardinality(container1) != container2->cardinality) - return false; - int32_t pos = 0; - for (int i = 0; i < container1->n_runs; ++i) { - const uint32_t run_start = container1->runs[i].value; - const uint32_t le = container1->runs[i].length; - - if (container2->array[pos] != run_start) { - return false; - } - - if (container2->array[pos + le] != run_start + le) { - return false; - } - - pos += le + 1; - } - return true; -} - -bool run_container_equals_bitset(const run_container_t* container1, - const bitset_container_t* container2) { - if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { - if (container2->cardinality != run_container_cardinality(container1)) { - return false; - } - } else { - int32_t card = bitset_container_compute_cardinality( - container2); // modify container2? - if (card != run_container_cardinality(container1)) { - return false; - } - } - for (int i = 0; i < container1->n_runs; ++i) { - uint32_t run_start = container1->runs[i].value; - uint32_t le = container1->runs[i].length; - for (uint32_t j = run_start; j <= run_start + le; ++j) { - // todo: this code could be much faster - if (!bitset_container_contains(container2, j)) { - return false; - } - } - } - return true; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_equal.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_intersection.c */ -/* - * mixed_intersection.c - * - */ - - -/* Compute the intersection of src_1 and src_2 and write the result to - * dst. */ -void array_bitset_container_intersection(const array_container_t *src_1, - const bitset_container_t *src_2, - array_container_t *dst) { - if (dst->capacity < src_1->cardinality) { - array_container_grow(dst, src_1->cardinality, false); - } - int32_t newcard = 0; // dst could be src_1 - const int32_t origcard = src_1->cardinality; - for (int i = 0; i < origcard; ++i) { - uint16_t key = src_1->array[i]; - // this branchless approach is much faster... - dst->array[newcard] = key; - newcard += bitset_container_contains(src_2, key); - /** - * we could do it this way instead... - * if (bitset_container_contains(src_2, key)) { - * dst->array[newcard++] = key; - * } - * but if the result is unpredictible, the processor generates - * many mispredicted branches. - * Difference can be huge (from 3 cycles when predictible all the way - * to 16 cycles when unpredictible. - * See - * https://github.com/lemire/Code-used-on-Daniel-Lemire-s-blog/blob/master/extra/bitset/c/arraybitsetintersection.c - */ - } - dst->cardinality = newcard; -} - -/* Compute the size of the intersection of src_1 and src_2. */ -int array_bitset_container_intersection_cardinality( - const array_container_t *src_1, const bitset_container_t *src_2) { - int32_t newcard = 0; - const int32_t origcard = src_1->cardinality; - for (int i = 0; i < origcard; ++i) { - uint16_t key = src_1->array[i]; - newcard += bitset_container_contains(src_2, key); - } - return newcard; -} - - -bool array_bitset_container_intersect(const array_container_t *src_1, - const bitset_container_t *src_2) { - const int32_t origcard = src_1->cardinality; - for (int i = 0; i < origcard; ++i) { - uint16_t key = src_1->array[i]; - if(bitset_container_contains(src_2, key)) return true; - } - return false; -} - -/* Compute the intersection of src_1 and src_2 and write the result to - * dst. It is allowed for dst to be equal to src_1. We assume that dst is a - * valid container. */ -void array_run_container_intersection(const array_container_t *src_1, - const run_container_t *src_2, - array_container_t *dst) { - if (run_container_is_full(src_2)) { - if (dst != src_1) array_container_copy(src_1, dst); - return; - } - if (dst->capacity < src_1->cardinality) { - array_container_grow(dst, src_1->cardinality, false); - } - if (src_2->n_runs == 0) { - return; - } - int32_t rlepos = 0; - int32_t arraypos = 0; - rle16_t rle = src_2->runs[rlepos]; - int32_t newcard = 0; - while (arraypos < src_1->cardinality) { - const uint16_t arrayval = src_1->array[arraypos]; - while (rle.value + rle.length < - arrayval) { // this will frequently be false - ++rlepos; - if (rlepos == src_2->n_runs) { - dst->cardinality = newcard; - return; // we are done - } - rle = src_2->runs[rlepos]; - } - if (rle.value > arrayval) { - arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, - rle.value); - } else { - dst->array[newcard] = arrayval; - newcard++; - arraypos++; - } - } - dst->cardinality = newcard; -} - -/* Compute the intersection of src_1 and src_2 and write the result to - * *dst. If the result is true then the result is a bitset_container_t - * otherwise is a array_container_t. If *dst == src_2, an in-place processing - * is attempted.*/ -bool run_bitset_container_intersection(const run_container_t *src_1, - const bitset_container_t *src_2, - void **dst) { - if (run_container_is_full(src_1)) { - if (*dst != src_2) *dst = bitset_container_clone(src_2); - return true; - } - int32_t card = run_container_cardinality(src_1); - if (card <= DEFAULT_MAX_SIZE) { - // result can only be an array (assuming that we never make a - // RunContainer) - if (card > src_2->cardinality) { - card = src_2->cardinality; - } - array_container_t *answer = array_container_create_given_capacity(card); - *dst = answer; - if (*dst == NULL) { - return false; - } - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - uint32_t endofrun = (uint32_t)rle.value + rle.length; - for (uint32_t runValue = rle.value; runValue <= endofrun; - ++runValue) { - answer->array[answer->cardinality] = (uint16_t)runValue; - answer->cardinality += - bitset_container_contains(src_2, runValue); - } - } - return false; - } - if (*dst == src_2) { // we attempt in-place - bitset_container_t *answer = (bitset_container_t *)*dst; - uint32_t start = 0; - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - const rle16_t rle = src_1->runs[rlepos]; - uint32_t end = rle.value; - bitset_reset_range(src_2->array, start, end); - - start = end + rle.length + 1; - } - bitset_reset_range(src_2->array, start, UINT32_C(1) << 16); - answer->cardinality = bitset_container_compute_cardinality(answer); - if (src_2->cardinality > DEFAULT_MAX_SIZE) { - return true; - } else { - array_container_t *newanswer = array_container_from_bitset(src_2); - if (newanswer == NULL) { - *dst = NULL; - return false; - } - *dst = newanswer; - return false; - } - } else { // no inplace - // we expect the answer to be a bitmap (if we are lucky) - bitset_container_t *answer = bitset_container_clone(src_2); - - *dst = answer; - if (answer == NULL) { - return true; - } - uint32_t start = 0; - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - const rle16_t rle = src_1->runs[rlepos]; - uint32_t end = rle.value; - bitset_reset_range(answer->array, start, end); - start = end + rle.length + 1; - } - bitset_reset_range(answer->array, start, UINT32_C(1) << 16); - answer->cardinality = bitset_container_compute_cardinality(answer); - - if (answer->cardinality > DEFAULT_MAX_SIZE) { - return true; - } else { - array_container_t *newanswer = array_container_from_bitset(answer); - bitset_container_free((bitset_container_t *)*dst); - if (newanswer == NULL) { - *dst = NULL; - return false; - } - *dst = newanswer; - return false; - } - } -} - -/* Compute the size of the intersection between src_1 and src_2 . */ -int array_run_container_intersection_cardinality(const array_container_t *src_1, - const run_container_t *src_2) { - if (run_container_is_full(src_2)) { - return src_1->cardinality; - } - if (src_2->n_runs == 0) { - return 0; - } - int32_t rlepos = 0; - int32_t arraypos = 0; - rle16_t rle = src_2->runs[rlepos]; - int32_t newcard = 0; - while (arraypos < src_1->cardinality) { - const uint16_t arrayval = src_1->array[arraypos]; - while (rle.value + rle.length < - arrayval) { // this will frequently be false - ++rlepos; - if (rlepos == src_2->n_runs) { - return newcard; // we are done - } - rle = src_2->runs[rlepos]; - } - if (rle.value > arrayval) { - arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, - rle.value); - } else { - newcard++; - arraypos++; - } - } - return newcard; -} - -/* Compute the intersection between src_1 and src_2 - **/ -int run_bitset_container_intersection_cardinality( - const run_container_t *src_1, const bitset_container_t *src_2) { - if (run_container_is_full(src_1)) { - return bitset_container_cardinality(src_2); - } - int answer = 0; - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - answer += - bitset_lenrange_cardinality(src_2->array, rle.value, rle.length); - } - return answer; -} - - -bool array_run_container_intersect(const array_container_t *src_1, - const run_container_t *src_2) { - if( run_container_is_full(src_2) ) { - return !array_container_empty(src_1); - } - if (src_2->n_runs == 0) { - return false; - } - int32_t rlepos = 0; - int32_t arraypos = 0; - rle16_t rle = src_2->runs[rlepos]; - while (arraypos < src_1->cardinality) { - const uint16_t arrayval = src_1->array[arraypos]; - while (rle.value + rle.length < - arrayval) { // this will frequently be false - ++rlepos; - if (rlepos == src_2->n_runs) { - return false; // we are done - } - rle = src_2->runs[rlepos]; - } - if (rle.value > arrayval) { - arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, - rle.value); - } else { - return true; - } - } - return false; -} - -/* Compute the intersection between src_1 and src_2 - **/ -bool run_bitset_container_intersect(const run_container_t *src_1, - const bitset_container_t *src_2) { - if( run_container_is_full(src_1) ) { - return !bitset_container_empty(src_2); - } - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - if(!bitset_lenrange_empty(src_2->array, rle.value,rle.length)) return true; - } - return false; -} - -/* - * Compute the intersection between src_1 and src_2 and write the result - * to *dst. If the return function is true, the result is a bitset_container_t - * otherwise is a array_container_t. - */ -bool bitset_bitset_container_intersection(const bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst) { - const int newCardinality = bitset_container_and_justcard(src_1, src_2); - if (newCardinality > DEFAULT_MAX_SIZE) { - *dst = bitset_container_create(); - if (*dst != NULL) { - bitset_container_and_nocard(src_1, src_2, - (bitset_container_t *)*dst); - ((bitset_container_t *)*dst)->cardinality = newCardinality; - } - return true; // it is a bitset - } - *dst = array_container_create_given_capacity(newCardinality); - if (*dst != NULL) { - ((array_container_t *)*dst)->cardinality = newCardinality; - bitset_extract_intersection_setbits_uint16( - ((const bitset_container_t *)src_1)->array, - ((const bitset_container_t *)src_2)->array, - BITSET_CONTAINER_SIZE_IN_WORDS, ((array_container_t *)*dst)->array, - 0); - } - return false; // not a bitset -} - -bool bitset_bitset_container_intersection_inplace( - bitset_container_t *src_1, const bitset_container_t *src_2, void **dst) { - const int newCardinality = bitset_container_and_justcard(src_1, src_2); - if (newCardinality > DEFAULT_MAX_SIZE) { - *dst = src_1; - bitset_container_and_nocard(src_1, src_2, src_1); - ((bitset_container_t *)*dst)->cardinality = newCardinality; - return true; // it is a bitset - } - *dst = array_container_create_given_capacity(newCardinality); - if (*dst != NULL) { - ((array_container_t *)*dst)->cardinality = newCardinality; - bitset_extract_intersection_setbits_uint16( - ((const bitset_container_t *)src_1)->array, - ((const bitset_container_t *)src_2)->array, - BITSET_CONTAINER_SIZE_IN_WORDS, ((array_container_t *)*dst)->array, - 0); - } - return false; // not a bitset -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_intersection.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_negation.c */ -/* - * mixed_negation.c - * - */ - -#include -#include - - -// TODO: make simplified and optimized negation code across -// the full range. - -/* Negation across the entire range of the container. - * Compute the negation of src and write the result - * to *dst. The complement of a - * sufficiently sparse set will always be dense and a hence a bitmap -' * We assume that dst is pre-allocated and a valid bitset container - * There can be no in-place version. - */ -void array_container_negation(const array_container_t *src, - bitset_container_t *dst) { - uint64_t card = UINT64_C(1 << 16); - bitset_container_set_all(dst); - - dst->cardinality = (int32_t)bitset_clear_list(dst->array, card, src->array, - (uint64_t)src->cardinality); -} - -/* Negation across the entire range of the container - * Compute the negation of src and write the result - * to *dst. A true return value indicates a bitset result, - * otherwise the result is an array container. - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool bitset_container_negation(const bitset_container_t *src, void **dst) { - return bitset_container_negation_range(src, 0, (1 << 16), dst); -} - -/* inplace version */ -/* - * Same as bitset_container_negation except that if the output is to - * be a - * bitset_container_t, then src is modified and no allocation is made. - * If the output is to be an array_container_t, then caller is responsible - * to free the container. - * In all cases, the result is in *dst. - */ -bool bitset_container_negation_inplace(bitset_container_t *src, void **dst) { - return bitset_container_negation_range_inplace(src, 0, (1 << 16), dst); -} - -/* Negation across the entire range of container - * Compute the negation of src and write the result - * to *dst. Return values are the *_TYPECODES as defined * in containers.h - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -int run_container_negation(const run_container_t *src, void **dst) { - return run_container_negation_range(src, 0, (1 << 16), dst); -} - -/* - * Same as run_container_negation except that if the output is to - * be a - * run_container_t, and has the capacity to hold the result, - * then src is modified and no allocation is made. - * In all cases, the result is in *dst. - */ -int run_container_negation_inplace(run_container_t *src, void **dst) { - return run_container_negation_range_inplace(src, 0, (1 << 16), dst); -} - -/* Negation across a range of the container. - * Compute the negation of src and write the result - * to *dst. Returns true if the result is a bitset container - * and false for an array container. *dst is not preallocated. - */ -bool array_container_negation_range(const array_container_t *src, - const int range_start, const int range_end, - void **dst) { - /* close port of the Java implementation */ - if (range_start >= range_end) { - *dst = array_container_clone(src); - return false; - } - - int32_t start_index = - binarySearch(src->array, src->cardinality, (uint16_t)range_start); - if (start_index < 0) start_index = -start_index - 1; - - int32_t last_index = - binarySearch(src->array, src->cardinality, (uint16_t)(range_end - 1)); - if (last_index < 0) last_index = -last_index - 2; - - const int32_t current_values_in_range = last_index - start_index + 1; - const int32_t span_to_be_flipped = range_end - range_start; - const int32_t new_values_in_range = - span_to_be_flipped - current_values_in_range; - const int32_t cardinality_change = - new_values_in_range - current_values_in_range; - const int32_t new_cardinality = src->cardinality + cardinality_change; - - if (new_cardinality > DEFAULT_MAX_SIZE) { - bitset_container_t *temp = bitset_container_from_array(src); - bitset_flip_range(temp->array, (uint32_t)range_start, - (uint32_t)range_end); - temp->cardinality = new_cardinality; - *dst = temp; - return true; - } - - array_container_t *arr = - array_container_create_given_capacity(new_cardinality); - *dst = (void *)arr; - if(new_cardinality == 0) { - arr->cardinality = new_cardinality; - return false; // we are done. - } - // copy stuff before the active area - memcpy(arr->array, src->array, start_index * sizeof(uint16_t)); - - // work on the range - int32_t out_pos = start_index, in_pos = start_index; - int32_t val_in_range = range_start; - for (; val_in_range < range_end && in_pos <= last_index; ++val_in_range) { - if ((uint16_t)val_in_range != src->array[in_pos]) { - arr->array[out_pos++] = (uint16_t)val_in_range; - } else { - ++in_pos; - } - } - for (; val_in_range < range_end; ++val_in_range) - arr->array[out_pos++] = (uint16_t)val_in_range; - - // content after the active range - memcpy(arr->array + out_pos, src->array + (last_index + 1), - (src->cardinality - (last_index + 1)) * sizeof(uint16_t)); - arr->cardinality = new_cardinality; - return false; -} - -/* Even when the result would fit, it is unclear how to make an - * inplace version without inefficient copying. - */ - -bool array_container_negation_range_inplace(array_container_t *src, - const int range_start, - const int range_end, void **dst) { - bool ans = array_container_negation_range(src, range_start, range_end, dst); - // TODO : try a real inplace version - array_container_free(src); - return ans; -} - -/* Negation across a range of the container - * Compute the negation of src and write the result - * to *dst. A true return value indicates a bitset result, - * otherwise the result is an array container. - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool bitset_container_negation_range(const bitset_container_t *src, - const int range_start, const int range_end, - void **dst) { - // TODO maybe consider density-based estimate - // and sometimes build result directly as array, with - // conversion back to bitset if wrong. Or determine - // actual result cardinality, then go directly for the known final cont. - - // keep computation using bitsets as long as possible. - bitset_container_t *t = bitset_container_clone(src); - bitset_flip_range(t->array, (uint32_t)range_start, (uint32_t)range_end); - t->cardinality = bitset_container_compute_cardinality(t); - - if (t->cardinality > DEFAULT_MAX_SIZE) { - *dst = t; - return true; - } else { - *dst = array_container_from_bitset(t); - bitset_container_free(t); - return false; - } -} - -/* inplace version */ -/* - * Same as bitset_container_negation except that if the output is to - * be a - * bitset_container_t, then src is modified and no allocation is made. - * If the output is to be an array_container_t, then caller is responsible - * to free the container. - * In all cases, the result is in *dst. - */ -bool bitset_container_negation_range_inplace(bitset_container_t *src, - const int range_start, - const int range_end, void **dst) { - bitset_flip_range(src->array, (uint32_t)range_start, (uint32_t)range_end); - src->cardinality = bitset_container_compute_cardinality(src); - if (src->cardinality > DEFAULT_MAX_SIZE) { - *dst = src; - return true; - } - *dst = array_container_from_bitset(src); - bitset_container_free(src); - return false; -} - -/* Negation across a range of container - * Compute the negation of src and write the result - * to *dst. Return values are the *_TYPECODES as defined * in containers.h - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -int run_container_negation_range(const run_container_t *src, - const int range_start, const int range_end, - void **dst) { - uint8_t return_typecode; - - // follows the Java implementation - if (range_end <= range_start) { - *dst = run_container_clone(src); - return RUN_CONTAINER_TYPE_CODE; - } - - run_container_t *ans = run_container_create_given_capacity( - src->n_runs + 1); // src->n_runs + 1); - int k = 0; - for (; k < src->n_runs && src->runs[k].value < range_start; ++k) { - ans->runs[k] = src->runs[k]; - ans->n_runs++; - } - - run_container_smart_append_exclusive( - ans, (uint16_t)range_start, (uint16_t)(range_end - range_start - 1)); - - for (; k < src->n_runs; ++k) { - run_container_smart_append_exclusive(ans, src->runs[k].value, - src->runs[k].length); - } - - *dst = convert_run_to_efficient_container(ans, &return_typecode); - if (return_typecode != RUN_CONTAINER_TYPE_CODE) run_container_free(ans); - - return return_typecode; -} - -/* - * Same as run_container_negation except that if the output is to - * be a - * run_container_t, and has the capacity to hold the result, - * then src is modified and no allocation is made. - * In all cases, the result is in *dst. - */ -int run_container_negation_range_inplace(run_container_t *src, - const int range_start, - const int range_end, void **dst) { - uint8_t return_typecode; - - if (range_end <= range_start) { - *dst = src; - return RUN_CONTAINER_TYPE_CODE; - } - - // TODO: efficient special case when range is 0 to 65535 inclusive - - if (src->capacity == src->n_runs) { - // no excess room. More checking to see if result can fit - bool last_val_before_range = false; - bool first_val_in_range = false; - bool last_val_in_range = false; - bool first_val_past_range = false; - - if (range_start > 0) - last_val_before_range = - run_container_contains(src, (uint16_t)(range_start - 1)); - first_val_in_range = run_container_contains(src, (uint16_t)range_start); - - if (last_val_before_range == first_val_in_range) { - last_val_in_range = - run_container_contains(src, (uint16_t)(range_end - 1)); - if (range_end != 0x10000) - first_val_past_range = - run_container_contains(src, (uint16_t)range_end); - - if (last_val_in_range == - first_val_past_range) { // no space for inplace - int ans = run_container_negation_range(src, range_start, - range_end, dst); - run_container_free(src); - return ans; - } - } - } - // all other cases: result will fit - - run_container_t *ans = src; - int my_nbr_runs = src->n_runs; - - ans->n_runs = 0; - int k = 0; - for (; (k < my_nbr_runs) && (src->runs[k].value < range_start); ++k) { - // ans->runs[k] = src->runs[k]; (would be self-copy) - ans->n_runs++; - } - - // as with Java implementation, use locals to give self a buffer of depth 1 - rle16_t buffered = (rle16_t){.value = (uint16_t)0, .length = (uint16_t)0}; - rle16_t next = buffered; - if (k < my_nbr_runs) buffered = src->runs[k]; - - run_container_smart_append_exclusive( - ans, (uint16_t)range_start, (uint16_t)(range_end - range_start - 1)); - - for (; k < my_nbr_runs; ++k) { - if (k + 1 < my_nbr_runs) next = src->runs[k + 1]; - - run_container_smart_append_exclusive(ans, buffered.value, - buffered.length); - buffered = next; - } - - *dst = convert_run_to_efficient_container(ans, &return_typecode); - if (return_typecode != RUN_CONTAINER_TYPE_CODE) run_container_free(ans); - - return return_typecode; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_negation.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_subset.c */ - -bool array_container_is_subset_bitset(const array_container_t* container1, - const bitset_container_t* container2) { - if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { - if (container2->cardinality < container1->cardinality) { - return false; - } - } - for (int i = 0; i < container1->cardinality; ++i) { - if (!bitset_container_contains(container2, container1->array[i])) { - return false; - } - } - return true; -} - -bool run_container_is_subset_array(const run_container_t* container1, - const array_container_t* container2) { - if (run_container_cardinality(container1) > container2->cardinality) - return false; - int32_t start_pos = -1, stop_pos = -1; - for (int i = 0; i < container1->n_runs; ++i) { - int32_t start = container1->runs[i].value; - int32_t stop = start + container1->runs[i].length; - start_pos = advanceUntil(container2->array, stop_pos, - container2->cardinality, start); - stop_pos = advanceUntil(container2->array, stop_pos, - container2->cardinality, stop); - if (start_pos == container2->cardinality) { - return false; - } else if (stop_pos - start_pos != stop - start || - container2->array[start_pos] != start || - container2->array[stop_pos] != stop) { - return false; - } - } - return true; -} - -bool array_container_is_subset_run(const array_container_t* container1, - const run_container_t* container2) { - if (container1->cardinality > run_container_cardinality(container2)) - return false; - int i_array = 0, i_run = 0; - while (i_array < container1->cardinality && i_run < container2->n_runs) { - uint32_t start = container2->runs[i_run].value; - uint32_t stop = start + container2->runs[i_run].length; - if (container1->array[i_array] < start) { - return false; - } else if (container1->array[i_array] > stop) { - i_run++; - } else { // the value of the array is in the run - i_array++; - } - } - if (i_array == container1->cardinality) { - return true; - } else { - return false; - } -} - -bool run_container_is_subset_bitset(const run_container_t* container1, - const bitset_container_t* container2) { - // todo: this code could be much faster - if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { - if (container2->cardinality < run_container_cardinality(container1)) { - return false; - } - } else { - int32_t card = bitset_container_compute_cardinality( - container2); // modify container2? - if (card < run_container_cardinality(container1)) { - return false; - } - } - for (int i = 0; i < container1->n_runs; ++i) { - uint32_t run_start = container1->runs[i].value; - uint32_t le = container1->runs[i].length; - for (uint32_t j = run_start; j <= run_start + le; ++j) { - if (!bitset_container_contains(container2, j)) { - return false; - } - } - } - return true; -} - -bool bitset_container_is_subset_run(const bitset_container_t* container1, - const run_container_t* container2) { - // todo: this code could be much faster - if (container1->cardinality != BITSET_UNKNOWN_CARDINALITY) { - if (container1->cardinality > run_container_cardinality(container2)) { - return false; - } - } - int32_t i_bitset = 0, i_run = 0; - while (i_bitset < BITSET_CONTAINER_SIZE_IN_WORDS && - i_run < container2->n_runs) { - uint64_t w = container1->array[i_bitset]; - while (w != 0 && i_run < container2->n_runs) { - uint32_t start = container2->runs[i_run].value; - uint32_t stop = start + container2->runs[i_run].length; - uint64_t t = w & (~w + 1); - uint16_t r = i_bitset * 64 + __builtin_ctzll(w); - if (r < start) { - return false; - } else if (r > stop) { - i_run++; - continue; - } else { - w ^= t; - } - } - if (w == 0) { - i_bitset++; - } else { - return false; - } - } - if (i_bitset < BITSET_CONTAINER_SIZE_IN_WORDS) { - // terminated iterating on the run containers, check that rest of bitset - // is empty - for (; i_bitset < BITSET_CONTAINER_SIZE_IN_WORDS; i_bitset++) { - if (container1->array[i_bitset] != 0) { - return false; - } - } - } - return true; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_subset.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_union.c */ -/* - * mixed_union.c - * - */ - -#include -#include - - -/* Compute the union of src_1 and src_2 and write the result to - * dst. */ -void array_bitset_container_union(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - if (src_2 != dst) bitset_container_copy(src_2, dst); - dst->cardinality = (int32_t)bitset_set_list_withcard( - dst->array, dst->cardinality, src_1->array, src_1->cardinality); -} - -/* Compute the union of src_1 and src_2 and write the result to - * dst. It is allowed for src_2 to be dst. This version does not - * update the cardinality of dst (it is set to BITSET_UNKNOWN_CARDINALITY). */ -void array_bitset_container_lazy_union(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - if (src_2 != dst) bitset_container_copy(src_2, dst); - bitset_set_list(dst->array, src_1->array, src_1->cardinality); - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; -} - -void run_bitset_container_union(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - assert(!run_container_is_full(src_1)); // catch this case upstream - if (src_2 != dst) bitset_container_copy(src_2, dst); - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - bitset_set_lenrange(dst->array, rle.value, rle.length); - } - dst->cardinality = bitset_container_compute_cardinality(dst); -} - -void run_bitset_container_lazy_union(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - assert(!run_container_is_full(src_1)); // catch this case upstream - if (src_2 != dst) bitset_container_copy(src_2, dst); - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - bitset_set_lenrange(dst->array, rle.value, rle.length); - } - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; -} - -// why do we leave the result as a run container?? -void array_run_container_union(const array_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst) { - if (run_container_is_full(src_2)) { - run_container_copy(src_2, dst); - return; - } - // TODO: see whether the "2*" is spurious - run_container_grow(dst, 2 * (src_1->cardinality + src_2->n_runs), false); - int32_t rlepos = 0; - int32_t arraypos = 0; - rle16_t previousrle; - if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { - previousrle = run_container_append_first(dst, src_2->runs[rlepos]); - rlepos++; - } else { - previousrle = - run_container_append_value_first(dst, src_1->array[arraypos]); - arraypos++; - } - while ((rlepos < src_2->n_runs) && (arraypos < src_1->cardinality)) { - if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { - run_container_append(dst, src_2->runs[rlepos], &previousrle); - rlepos++; - } else { - run_container_append_value(dst, src_1->array[arraypos], - &previousrle); - arraypos++; - } - } - if (arraypos < src_1->cardinality) { - while (arraypos < src_1->cardinality) { - run_container_append_value(dst, src_1->array[arraypos], - &previousrle); - arraypos++; - } - } else { - while (rlepos < src_2->n_runs) { - run_container_append(dst, src_2->runs[rlepos], &previousrle); - rlepos++; - } - } -} - -void array_run_container_inplace_union(const array_container_t *src_1, - run_container_t *src_2) { - if (run_container_is_full(src_2)) { - return; - } - const int32_t maxoutput = src_1->cardinality + src_2->n_runs; - const int32_t neededcapacity = maxoutput + src_2->n_runs; - if (src_2->capacity < neededcapacity) - run_container_grow(src_2, neededcapacity, true); - memmove(src_2->runs + maxoutput, src_2->runs, - src_2->n_runs * sizeof(rle16_t)); - rle16_t *inputsrc2 = src_2->runs + maxoutput; - int32_t rlepos = 0; - int32_t arraypos = 0; - int src2nruns = src_2->n_runs; - src_2->n_runs = 0; - - rle16_t previousrle; - - if (inputsrc2[rlepos].value <= src_1->array[arraypos]) { - previousrle = run_container_append_first(src_2, inputsrc2[rlepos]); - rlepos++; - } else { - previousrle = - run_container_append_value_first(src_2, src_1->array[arraypos]); - arraypos++; - } - - while ((rlepos < src2nruns) && (arraypos < src_1->cardinality)) { - if (inputsrc2[rlepos].value <= src_1->array[arraypos]) { - run_container_append(src_2, inputsrc2[rlepos], &previousrle); - rlepos++; - } else { - run_container_append_value(src_2, src_1->array[arraypos], - &previousrle); - arraypos++; - } - } - if (arraypos < src_1->cardinality) { - while (arraypos < src_1->cardinality) { - run_container_append_value(src_2, src_1->array[arraypos], - &previousrle); - arraypos++; - } - } else { - while (rlepos < src2nruns) { - run_container_append(src_2, inputsrc2[rlepos], &previousrle); - rlepos++; - } - } -} - -bool array_array_container_union(const array_container_t *src_1, - const array_container_t *src_2, void **dst) { - int totalCardinality = src_1->cardinality + src_2->cardinality; - if (totalCardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_create_given_capacity(totalCardinality); - if (*dst != NULL) { - array_container_union(src_1, src_2, (array_container_t *)*dst); - } else { - return true; // otherwise failure won't be caught - } - return false; // not a bitset - } - *dst = bitset_container_create(); - bool returnval = true; // expect a bitset - if (*dst != NULL) { - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - bitset_set_list(ourbitset->array, src_1->array, src_1->cardinality); - ourbitset->cardinality = (int32_t)bitset_set_list_withcard( - ourbitset->array, src_1->cardinality, src_2->array, - src_2->cardinality); - if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { - // need to convert! - *dst = array_container_from_bitset(ourbitset); - bitset_container_free(ourbitset); - returnval = false; // not going to be a bitset - } - } - return returnval; -} - -bool array_array_container_inplace_union(array_container_t *src_1, - const array_container_t *src_2, void **dst) { - int totalCardinality = src_1->cardinality + src_2->cardinality; - *dst = NULL; - if (totalCardinality <= DEFAULT_MAX_SIZE) { - if(src_1->capacity < totalCardinality) { - *dst = array_container_create_given_capacity(2 * totalCardinality); // be purposefully generous - if (*dst != NULL) { - array_container_union(src_1, src_2, (array_container_t *)*dst); - } else { - return true; // otherwise failure won't be caught - } - return false; // not a bitset - } else { - memmove(src_1->array + src_2->cardinality, src_1->array, src_1->cardinality * sizeof(uint16_t)); - src_1->cardinality = (int32_t)fast_union_uint16(src_1->array + src_2->cardinality, src_1->cardinality, - src_2->array, src_2->cardinality, src_1->array); - return false; // not a bitset - } - } - *dst = bitset_container_create(); - bool returnval = true; // expect a bitset - if (*dst != NULL) { - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - bitset_set_list(ourbitset->array, src_1->array, src_1->cardinality); - ourbitset->cardinality = (int32_t)bitset_set_list_withcard( - ourbitset->array, src_1->cardinality, src_2->array, - src_2->cardinality); - if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { - // need to convert! - if(src_1->capacity < ourbitset->cardinality) { - array_container_grow(src_1, ourbitset->cardinality, false); - } - - bitset_extract_setbits_uint16(ourbitset->array, BITSET_CONTAINER_SIZE_IN_WORDS, - src_1->array, 0); - src_1->cardinality = ourbitset->cardinality; - *dst = src_1; - bitset_container_free(ourbitset); - returnval = false; // not going to be a bitset - } - } - return returnval; -} - - -bool array_array_container_lazy_union(const array_container_t *src_1, - const array_container_t *src_2, - void **dst) { - int totalCardinality = src_1->cardinality + src_2->cardinality; - if (totalCardinality <= ARRAY_LAZY_LOWERBOUND) { - *dst = array_container_create_given_capacity(totalCardinality); - if (*dst != NULL) { - array_container_union(src_1, src_2, (array_container_t *)*dst); - } else { - return true; // otherwise failure won't be caught - } - return false; // not a bitset - } - *dst = bitset_container_create(); - bool returnval = true; // expect a bitset - if (*dst != NULL) { - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - bitset_set_list(ourbitset->array, src_1->array, src_1->cardinality); - bitset_set_list(ourbitset->array, src_2->array, src_2->cardinality); - ourbitset->cardinality = BITSET_UNKNOWN_CARDINALITY; - } - return returnval; -} - - -bool array_array_container_lazy_inplace_union(array_container_t *src_1, - const array_container_t *src_2, - void **dst) { - int totalCardinality = src_1->cardinality + src_2->cardinality; - *dst = NULL; - if (totalCardinality <= ARRAY_LAZY_LOWERBOUND) { - if(src_1->capacity < totalCardinality) { - *dst = array_container_create_given_capacity(2 * totalCardinality); // be purposefully generous - if (*dst != NULL) { - array_container_union(src_1, src_2, (array_container_t *)*dst); - } else { - return true; // otherwise failure won't be caught - } - return false; // not a bitset - } else { - memmove(src_1->array + src_2->cardinality, src_1->array, src_1->cardinality * sizeof(uint16_t)); - src_1->cardinality = (int32_t)fast_union_uint16(src_1->array + src_2->cardinality, src_1->cardinality, - src_2->array, src_2->cardinality, src_1->array); - return false; // not a bitset - } - } - *dst = bitset_container_create(); - bool returnval = true; // expect a bitset - if (*dst != NULL) { - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - bitset_set_list(ourbitset->array, src_1->array, src_1->cardinality); - bitset_set_list(ourbitset->array, src_2->array, src_2->cardinality); - ourbitset->cardinality = BITSET_UNKNOWN_CARDINALITY; - } - return returnval; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_union.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_xor.c */ -/* - * mixed_xor.c - */ - -#include -#include - - -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). - * Result is true iff dst is a bitset */ -bool array_bitset_container_xor(const array_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bitset_container_t *result = bitset_container_create(); - bitset_container_copy(src_2, result); - result->cardinality = (int32_t)bitset_flip_list_withcard( - result->array, result->cardinality, src_1->array, src_1->cardinality); - - // do required type conversions. - if (result->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(result); - bitset_container_free(result); - return false; // not bitset - } - *dst = result; - return true; // bitset -} - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. It is allowed for src_2 to be dst. This version does not - * update the cardinality of dst (it is set to BITSET_UNKNOWN_CARDINALITY). - */ - -void array_bitset_container_lazy_xor(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - if (src_2 != dst) bitset_container_copy(src_2, dst); - bitset_flip_list(dst->array, src_1->array, src_1->cardinality); - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; -} - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_xor(const run_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bitset_container_t *result = bitset_container_create(); - - bitset_container_copy(src_2, result); - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - bitset_flip_range(result->array, rle.value, - rle.value + rle.length + UINT32_C(1)); - } - result->cardinality = bitset_container_compute_cardinality(result); - - if (result->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(result); - bitset_container_free(result); - return false; // not bitset - } - *dst = result; - return true; // bitset -} - -/* lazy xor. Dst is initialized and may be equal to src_2. - * Result is left as a bitset container, even if actual - * cardinality would dictate an array container. - */ - -void run_bitset_container_lazy_xor(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst) { - if (src_2 != dst) bitset_container_copy(src_2, dst); - for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { - rle16_t rle = src_1->runs[rlepos]; - bitset_flip_range(dst->array, rle.value, - rle.value + rle.length + UINT32_C(1)); - } - dst->cardinality = BITSET_UNKNOWN_CARDINALITY; -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int array_run_container_xor(const array_container_t *src_1, - const run_container_t *src_2, void **dst) { - // semi following Java XOR implementation as of May 2016 - // the C OR implementation works quite differently and can return a run - // container - // TODO could optimize for full run containers. - - // use of lazy following Java impl. - const int arbitrary_threshold = 32; - if (src_1->cardinality < arbitrary_threshold) { - run_container_t *ans = run_container_create(); - array_run_container_lazy_xor(src_1, src_2, ans); // keeps runs. - uint8_t typecode_after; - *dst = - convert_run_to_efficient_container_and_free(ans, &typecode_after); - return typecode_after; - } - - int card = run_container_cardinality(src_2); - if (card <= DEFAULT_MAX_SIZE) { - // Java implementation works with the array, xoring the run elements via - // iterator - array_container_t *temp = array_container_from_run(src_2); - bool ret_is_bitset = array_array_container_xor(temp, src_1, dst); - array_container_free(temp); - return ret_is_bitset ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - } else { // guess that it will end up as a bitset - bitset_container_t *result = bitset_container_from_run(src_2); - bool is_bitset = bitset_array_container_ixor(result, src_1, dst); - // any necessary type conversion has been done by the ixor - int retval = (is_bitset ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE); - return retval; - } -} - -/* Dst is a valid run container. (Can it be src_2? Let's say not.) - * Leaves result as run container, even if other options are - * smaller. - */ - -void array_run_container_lazy_xor(const array_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst) { - run_container_grow(dst, src_1->cardinality + src_2->n_runs, false); - int32_t rlepos = 0; - int32_t arraypos = 0; - dst->n_runs = 0; - - while ((rlepos < src_2->n_runs) && (arraypos < src_1->cardinality)) { - if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { - run_container_smart_append_exclusive(dst, src_2->runs[rlepos].value, - src_2->runs[rlepos].length); - rlepos++; - } else { - run_container_smart_append_exclusive(dst, src_1->array[arraypos], - 0); - arraypos++; - } - } - while (arraypos < src_1->cardinality) { - run_container_smart_append_exclusive(dst, src_1->array[arraypos], 0); - arraypos++; - } - while (rlepos < src_2->n_runs) { - run_container_smart_append_exclusive(dst, src_2->runs[rlepos].value, - src_2->runs[rlepos].length); - rlepos++; - } -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int run_run_container_xor(const run_container_t *src_1, - const run_container_t *src_2, void **dst) { - run_container_t *ans = run_container_create(); - run_container_xor(src_1, src_2, ans); - uint8_t typecode_after; - *dst = convert_run_to_efficient_container_and_free(ans, &typecode_after); - return typecode_after; -} - -/* - * Java implementation (as of May 2016) for array_run, run_run - * and bitset_run don't do anything different for inplace. - * Could adopt the mixed_union.c approach instead (ie, using - * smart_append_exclusive) - * - */ - -bool array_array_container_xor(const array_container_t *src_1, - const array_container_t *src_2, void **dst) { - int totalCardinality = - src_1->cardinality + src_2->cardinality; // upper bound - if (totalCardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_create_given_capacity(totalCardinality); - array_container_xor(src_1, src_2, (array_container_t *)*dst); - return false; // not a bitset - } - *dst = bitset_container_from_array(src_1); - bool returnval = true; // expect a bitset - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - ourbitset->cardinality = (uint32_t)bitset_flip_list_withcard( - ourbitset->array, src_1->cardinality, src_2->array, src_2->cardinality); - if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { - // need to convert! - *dst = array_container_from_bitset(ourbitset); - bitset_container_free(ourbitset); - returnval = false; // not going to be a bitset - } - - return returnval; -} - -bool array_array_container_lazy_xor(const array_container_t *src_1, - const array_container_t *src_2, - void **dst) { - int totalCardinality = src_1->cardinality + src_2->cardinality; - // upper bound, but probably poor estimate for xor - if (totalCardinality <= ARRAY_LAZY_LOWERBOUND) { - *dst = array_container_create_given_capacity(totalCardinality); - if (*dst != NULL) - array_container_xor(src_1, src_2, (array_container_t *)*dst); - return false; // not a bitset - } - *dst = bitset_container_from_array(src_1); - bool returnval = true; // expect a bitset (maybe, for XOR??) - if (*dst != NULL) { - bitset_container_t *ourbitset = (bitset_container_t *)*dst; - bitset_flip_list(ourbitset->array, src_2->array, src_2->cardinality); - ourbitset->cardinality = BITSET_UNKNOWN_CARDINALITY; - } - return returnval; -} - -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). Return value is - * "dst is a bitset" - */ - -bool bitset_bitset_container_xor(const bitset_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bitset_container_t *ans = bitset_container_create(); - int card = bitset_container_xor(src_1, src_2, ans); - if (card <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(ans); - bitset_container_free(ans); - return false; // not bitset - } else { - *dst = ans; - return true; - } -} - -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_array_container_ixor(bitset_container_t *src_1, - const array_container_t *src_2, void **dst) { - *dst = src_1; - src_1->cardinality = (uint32_t)bitset_flip_list_withcard( - src_1->array, src_1->cardinality, src_2->array, src_2->cardinality); - - if (src_1->cardinality <= DEFAULT_MAX_SIZE) { - *dst = array_container_from_bitset(src_1); - bitset_container_free(src_1); - return false; // not bitset - } else - return true; -} - -/* a bunch of in-place, some of which may not *really* be inplace. - * TODO: write actual inplace routine if efficiency warrants it - * Anything inplace with a bitset is a good candidate - */ - -bool bitset_bitset_container_ixor(bitset_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bool ans = bitset_bitset_container_xor(src_1, src_2, dst); - bitset_container_free(src_1); - return ans; -} - -bool array_bitset_container_ixor(array_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bool ans = array_bitset_container_xor(src_1, src_2, dst); - array_container_free(src_1); - return ans; -} - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_ixor(run_container_t *src_1, - const bitset_container_t *src_2, void **dst) { - bool ans = run_bitset_container_xor(src_1, src_2, dst); - run_container_free(src_1); - return ans; -} - -bool bitset_run_container_ixor(bitset_container_t *src_1, - const run_container_t *src_2, void **dst) { - bool ans = run_bitset_container_xor(src_2, src_1, dst); - bitset_container_free(src_1); - return ans; -} - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int array_run_container_ixor(array_container_t *src_1, - const run_container_t *src_2, void **dst) { - int ans = array_run_container_xor(src_1, src_2, dst); - array_container_free(src_1); - return ans; -} - -int run_array_container_ixor(run_container_t *src_1, - const array_container_t *src_2, void **dst) { - int ans = array_run_container_xor(src_2, src_1, dst); - run_container_free(src_1); - return ans; -} - -bool array_array_container_ixor(array_container_t *src_1, - const array_container_t *src_2, void **dst) { - bool ans = array_array_container_xor(src_1, src_2, dst); - array_container_free(src_1); - return ans; -} - -int run_run_container_ixor(run_container_t *src_1, const run_container_t *src_2, - void **dst) { - int ans = run_run_container_xor(src_1, src_2, dst); - run_container_free(src_1); - return ans; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/mixed_xor.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/containers/run.c */ -#include -#include - - -extern inline uint16_t run_container_minimum(const run_container_t *run); -extern inline uint16_t run_container_maximum(const run_container_t *run); -extern inline int32_t interleavedBinarySearch(const rle16_t *array, - int32_t lenarray, uint16_t ikey); -extern inline bool run_container_contains(const run_container_t *run, - uint16_t pos); -extern inline int run_container_index_equalorlarger(const run_container_t *arr, uint16_t x); -extern bool run_container_is_full(const run_container_t *run); -extern bool run_container_nonzero_cardinality(const run_container_t *r); -extern void run_container_clear(run_container_t *run); -extern int32_t run_container_serialized_size_in_bytes(int32_t num_runs); -extern run_container_t *run_container_create_range(uint32_t start, - uint32_t stop); - -bool run_container_add(run_container_t *run, uint16_t pos) { - int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); - if (index >= 0) return false; // already there - index = -index - 2; // points to preceding value, possibly -1 - if (index >= 0) { // possible match - int32_t offset = pos - run->runs[index].value; - int32_t le = run->runs[index].length; - if (offset <= le) return false; // already there - if (offset == le + 1) { - // we may need to fuse - if (index + 1 < run->n_runs) { - if (run->runs[index + 1].value == pos + 1) { - // indeed fusion is needed - run->runs[index].length = run->runs[index + 1].value + - run->runs[index + 1].length - - run->runs[index].value; - recoverRoomAtIndex(run, (uint16_t)(index + 1)); - return true; - } - } - run->runs[index].length++; - return true; - } - if (index + 1 < run->n_runs) { - // we may need to fuse - if (run->runs[index + 1].value == pos + 1) { - // indeed fusion is needed - run->runs[index + 1].value = pos; - run->runs[index + 1].length = run->runs[index + 1].length + 1; - return true; - } - } - } - if (index == -1) { - // we may need to extend the first run - if (0 < run->n_runs) { - if (run->runs[0].value == pos + 1) { - run->runs[0].length++; - run->runs[0].value--; - return true; - } - } - } - makeRoomAtIndex(run, (uint16_t)(index + 1)); - run->runs[index + 1].value = pos; - run->runs[index + 1].length = 0; - return true; -} - -/* Create a new run container. Return NULL in case of failure. */ -run_container_t *run_container_create_given_capacity(int32_t size) { - run_container_t *run; - /* Allocate the run container itself. */ - if ((run = (run_container_t *)malloc(sizeof(run_container_t))) == NULL) { - return NULL; - } - if (size <= 0 ) { // we don't want to rely on malloc(0) - run->runs = NULL; - } else if ((run->runs = (rle16_t *)malloc(sizeof(rle16_t) * size)) == NULL) { - free(run); - return NULL; - } - run->capacity = size; - run->n_runs = 0; - return run; -} - -int run_container_shrink_to_fit(run_container_t *src) { - if (src->n_runs == src->capacity) return 0; // nothing to do - int savings = src->capacity - src->n_runs; - src->capacity = src->n_runs; - rle16_t *oldruns = src->runs; - src->runs = (rle16_t *)realloc(oldruns, src->capacity * sizeof(rle16_t)); - if (src->runs == NULL) free(oldruns); // should never happen? - return savings; -} -/* Create a new run container. Return NULL in case of failure. */ -run_container_t *run_container_create(void) { - return run_container_create_given_capacity(RUN_DEFAULT_INIT_SIZE); -} - -run_container_t *run_container_clone(const run_container_t *src) { - run_container_t *run = run_container_create_given_capacity(src->capacity); - if (run == NULL) return NULL; - run->capacity = src->capacity; - run->n_runs = src->n_runs; - memcpy(run->runs, src->runs, src->n_runs * sizeof(rle16_t)); - return run; -} - -/* Free memory. */ -void run_container_free(run_container_t *run) { - if(run->runs != NULL) {// Jon Strabala reports that some tools complain otherwise - free(run->runs); - run->runs = NULL; // pedantic - } - free(run); -} - -void run_container_grow(run_container_t *run, int32_t min, bool copy) { - int32_t newCapacity = - (run->capacity == 0) - ? RUN_DEFAULT_INIT_SIZE - : run->capacity < 64 ? run->capacity * 2 - : run->capacity < 1024 ? run->capacity * 3 / 2 - : run->capacity * 5 / 4; - if (newCapacity < min) newCapacity = min; - run->capacity = newCapacity; - assert(run->capacity >= min); - if (copy) { - rle16_t *oldruns = run->runs; - run->runs = - (rle16_t *)realloc(oldruns, run->capacity * sizeof(rle16_t)); - if (run->runs == NULL) free(oldruns); - } else { - // Jon Strabala reports that some tools complain otherwise - if (run->runs != NULL) { - free(run->runs); - } - run->runs = (rle16_t *)malloc(run->capacity * sizeof(rle16_t)); - } - // handle the case where realloc fails - if (run->runs == NULL) { - fprintf(stderr, "could not allocate memory\n"); - } - assert(run->runs != NULL); -} - -/* copy one container into another */ -void run_container_copy(const run_container_t *src, run_container_t *dst) { - const int32_t n_runs = src->n_runs; - if (src->n_runs > dst->capacity) { - run_container_grow(dst, n_runs, false); - } - dst->n_runs = n_runs; - memcpy(dst->runs, src->runs, sizeof(rle16_t) * n_runs); -} - -/* Compute the union of `src_1' and `src_2' and write the result to `dst' - * It is assumed that `dst' is distinct from both `src_1' and `src_2'. */ -void run_container_union(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst) { - // TODO: this could be a lot more efficient - - // we start out with inexpensive checks - const bool if1 = run_container_is_full(src_1); - const bool if2 = run_container_is_full(src_2); - if (if1 || if2) { - if (if1) { - run_container_copy(src_1, dst); - return; - } - if (if2) { - run_container_copy(src_2, dst); - return; - } - } - const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; - if (dst->capacity < neededcapacity) - run_container_grow(dst, neededcapacity, false); - dst->n_runs = 0; - int32_t rlepos = 0; - int32_t xrlepos = 0; - - rle16_t previousrle; - if (src_1->runs[rlepos].value <= src_2->runs[xrlepos].value) { - previousrle = run_container_append_first(dst, src_1->runs[rlepos]); - rlepos++; - } else { - previousrle = run_container_append_first(dst, src_2->runs[xrlepos]); - xrlepos++; - } - - while ((xrlepos < src_2->n_runs) && (rlepos < src_1->n_runs)) { - rle16_t newrl; - if (src_1->runs[rlepos].value <= src_2->runs[xrlepos].value) { - newrl = src_1->runs[rlepos]; - rlepos++; - } else { - newrl = src_2->runs[xrlepos]; - xrlepos++; - } - run_container_append(dst, newrl, &previousrle); - } - while (xrlepos < src_2->n_runs) { - run_container_append(dst, src_2->runs[xrlepos], &previousrle); - xrlepos++; - } - while (rlepos < src_1->n_runs) { - run_container_append(dst, src_1->runs[rlepos], &previousrle); - rlepos++; - } -} - -/* Compute the union of `src_1' and `src_2' and write the result to `src_1' - */ -void run_container_union_inplace(run_container_t *src_1, - const run_container_t *src_2) { - // TODO: this could be a lot more efficient - - // we start out with inexpensive checks - const bool if1 = run_container_is_full(src_1); - const bool if2 = run_container_is_full(src_2); - if (if1 || if2) { - if (if1) { - return; - } - if (if2) { - run_container_copy(src_2, src_1); - return; - } - } - // we move the data to the end of the current array - const int32_t maxoutput = src_1->n_runs + src_2->n_runs; - const int32_t neededcapacity = maxoutput + src_1->n_runs; - if (src_1->capacity < neededcapacity) - run_container_grow(src_1, neededcapacity, true); - memmove(src_1->runs + maxoutput, src_1->runs, - src_1->n_runs * sizeof(rle16_t)); - rle16_t *inputsrc1 = src_1->runs + maxoutput; - const int32_t input1nruns = src_1->n_runs; - src_1->n_runs = 0; - int32_t rlepos = 0; - int32_t xrlepos = 0; - - rle16_t previousrle; - if (inputsrc1[rlepos].value <= src_2->runs[xrlepos].value) { - previousrle = run_container_append_first(src_1, inputsrc1[rlepos]); - rlepos++; - } else { - previousrle = run_container_append_first(src_1, src_2->runs[xrlepos]); - xrlepos++; - } - while ((xrlepos < src_2->n_runs) && (rlepos < input1nruns)) { - rle16_t newrl; - if (inputsrc1[rlepos].value <= src_2->runs[xrlepos].value) { - newrl = inputsrc1[rlepos]; - rlepos++; - } else { - newrl = src_2->runs[xrlepos]; - xrlepos++; - } - run_container_append(src_1, newrl, &previousrle); - } - while (xrlepos < src_2->n_runs) { - run_container_append(src_1, src_2->runs[xrlepos], &previousrle); - xrlepos++; - } - while (rlepos < input1nruns) { - run_container_append(src_1, inputsrc1[rlepos], &previousrle); - rlepos++; - } -} - -/* Compute the symmetric difference of `src_1' and `src_2' and write the result - * to `dst' - * It is assumed that `dst' is distinct from both `src_1' and `src_2'. */ -void run_container_xor(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst) { - // don't bother to convert xor with full range into negation - // since negation is implemented similarly - - const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; - if (dst->capacity < neededcapacity) - run_container_grow(dst, neededcapacity, false); - - int32_t pos1 = 0; - int32_t pos2 = 0; - dst->n_runs = 0; - - while ((pos1 < src_1->n_runs) && (pos2 < src_2->n_runs)) { - if (src_1->runs[pos1].value <= src_2->runs[pos2].value) { - run_container_smart_append_exclusive(dst, src_1->runs[pos1].value, - src_1->runs[pos1].length); - pos1++; - } else { - run_container_smart_append_exclusive(dst, src_2->runs[pos2].value, - src_2->runs[pos2].length); - pos2++; - } - } - while (pos1 < src_1->n_runs) { - run_container_smart_append_exclusive(dst, src_1->runs[pos1].value, - src_1->runs[pos1].length); - pos1++; - } - - while (pos2 < src_2->n_runs) { - run_container_smart_append_exclusive(dst, src_2->runs[pos2].value, - src_2->runs[pos2].length); - pos2++; - } -} - -/* Compute the intersection of src_1 and src_2 and write the result to - * dst. It is assumed that dst is distinct from both src_1 and src_2. */ -void run_container_intersection(const run_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst) { - const bool if1 = run_container_is_full(src_1); - const bool if2 = run_container_is_full(src_2); - if (if1 || if2) { - if (if1) { - run_container_copy(src_2, dst); - return; - } - if (if2) { - run_container_copy(src_1, dst); - return; - } - } - // TODO: this could be a lot more efficient, could use SIMD optimizations - const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; - if (dst->capacity < neededcapacity) - run_container_grow(dst, neededcapacity, false); - dst->n_runs = 0; - int32_t rlepos = 0; - int32_t xrlepos = 0; - int32_t start = src_1->runs[rlepos].value; - int32_t end = start + src_1->runs[rlepos].length + 1; - int32_t xstart = src_2->runs[xrlepos].value; - int32_t xend = xstart + src_2->runs[xrlepos].length + 1; - while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { - if (end <= xstart) { - ++rlepos; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - } else if (xend <= start) { - ++xrlepos; - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } else { // they overlap - const int32_t lateststart = start > xstart ? start : xstart; - int32_t earliestend; - if (end == xend) { // improbable - earliestend = end; - rlepos++; - xrlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } else if (end < xend) { - earliestend = end; - rlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - - } else { // end > xend - earliestend = xend; - xrlepos++; - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } - dst->runs[dst->n_runs].value = (uint16_t)lateststart; - dst->runs[dst->n_runs].length = - (uint16_t)(earliestend - lateststart - 1); - dst->n_runs++; - } - } -} - -/* Compute the size of the intersection of src_1 and src_2 . */ -int run_container_intersection_cardinality(const run_container_t *src_1, - const run_container_t *src_2) { - const bool if1 = run_container_is_full(src_1); - const bool if2 = run_container_is_full(src_2); - if (if1 || if2) { - if (if1) { - return run_container_cardinality(src_2); - } - if (if2) { - return run_container_cardinality(src_1); - } - } - int answer = 0; - int32_t rlepos = 0; - int32_t xrlepos = 0; - int32_t start = src_1->runs[rlepos].value; - int32_t end = start + src_1->runs[rlepos].length + 1; - int32_t xstart = src_2->runs[xrlepos].value; - int32_t xend = xstart + src_2->runs[xrlepos].length + 1; - while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { - if (end <= xstart) { - ++rlepos; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - } else if (xend <= start) { - ++xrlepos; - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } else { // they overlap - const int32_t lateststart = start > xstart ? start : xstart; - int32_t earliestend; - if (end == xend) { // improbable - earliestend = end; - rlepos++; - xrlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } else if (end < xend) { - earliestend = end; - rlepos++; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - - } else { // end > xend - earliestend = xend; - xrlepos++; - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } - answer += earliestend - lateststart; - } - } - return answer; -} - -bool run_container_intersect(const run_container_t *src_1, - const run_container_t *src_2) { - const bool if1 = run_container_is_full(src_1); - const bool if2 = run_container_is_full(src_2); - if (if1 || if2) { - if (if1) { - return !run_container_empty(src_2); - } - if (if2) { - return !run_container_empty(src_1); - } - } - int32_t rlepos = 0; - int32_t xrlepos = 0; - int32_t start = src_1->runs[rlepos].value; - int32_t end = start + src_1->runs[rlepos].length + 1; - int32_t xstart = src_2->runs[xrlepos].value; - int32_t xend = xstart + src_2->runs[xrlepos].length + 1; - while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { - if (end <= xstart) { - ++rlepos; - if (rlepos < src_1->n_runs) { - start = src_1->runs[rlepos].value; - end = start + src_1->runs[rlepos].length + 1; - } - } else if (xend <= start) { - ++xrlepos; - if (xrlepos < src_2->n_runs) { - xstart = src_2->runs[xrlepos].value; - xend = xstart + src_2->runs[xrlepos].length + 1; - } - } else { // they overlap - return true; - } - } - return false; -} - - -/* Compute the difference of src_1 and src_2 and write the result to - * dst. It is assumed that dst is distinct from both src_1 and src_2. */ -void run_container_andnot(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst) { - // following Java implementation as of June 2016 - - if (dst->capacity < src_1->n_runs + src_2->n_runs) - run_container_grow(dst, src_1->n_runs + src_2->n_runs, false); - - dst->n_runs = 0; - - int rlepos1 = 0; - int rlepos2 = 0; - int32_t start = src_1->runs[rlepos1].value; - int32_t end = start + src_1->runs[rlepos1].length + 1; - int32_t start2 = src_2->runs[rlepos2].value; - int32_t end2 = start2 + src_2->runs[rlepos2].length + 1; - - while ((rlepos1 < src_1->n_runs) && (rlepos2 < src_2->n_runs)) { - if (end <= start2) { - // output the first run - dst->runs[dst->n_runs++] = - (rle16_t){.value = (uint16_t)start, - .length = (uint16_t)(end - start - 1)}; - rlepos1++; - if (rlepos1 < src_1->n_runs) { - start = src_1->runs[rlepos1].value; - end = start + src_1->runs[rlepos1].length + 1; - } - } else if (end2 <= start) { - // exit the second run - rlepos2++; - if (rlepos2 < src_2->n_runs) { - start2 = src_2->runs[rlepos2].value; - end2 = start2 + src_2->runs[rlepos2].length + 1; - } - } else { - if (start < start2) { - dst->runs[dst->n_runs++] = - (rle16_t){.value = (uint16_t)start, - .length = (uint16_t)(start2 - start - 1)}; - } - if (end2 < end) { - start = end2; - } else { - rlepos1++; - if (rlepos1 < src_1->n_runs) { - start = src_1->runs[rlepos1].value; - end = start + src_1->runs[rlepos1].length + 1; - } - } - } - } - if (rlepos1 < src_1->n_runs) { - dst->runs[dst->n_runs++] = (rle16_t){ - .value = (uint16_t)start, .length = (uint16_t)(end - start - 1)}; - rlepos1++; - if (rlepos1 < src_1->n_runs) { - memcpy(dst->runs + dst->n_runs, src_1->runs + rlepos1, - sizeof(rle16_t) * (src_1->n_runs - rlepos1)); - dst->n_runs += src_1->n_runs - rlepos1; - } - } -} - -int run_container_to_uint32_array(void *vout, const run_container_t *cont, - uint32_t base) { - int outpos = 0; - uint32_t *out = (uint32_t *)vout; - for (int i = 0; i < cont->n_runs; ++i) { - uint32_t run_start = base + cont->runs[i].value; - uint16_t le = cont->runs[i].length; - for (int j = 0; j <= le; ++j) { - uint32_t val = run_start + j; - memcpy(out + outpos, &val, - sizeof(uint32_t)); // should be compiled as a MOV on x64 - outpos++; - } - } - return outpos; -} - -/* - * Print this container using printf (useful for debugging). - */ -void run_container_printf(const run_container_t *cont) { - for (int i = 0; i < cont->n_runs; ++i) { - uint16_t run_start = cont->runs[i].value; - uint16_t le = cont->runs[i].length; - printf("[%d,%d]", run_start, run_start + le); - } -} - -/* - * Print this container using printf as a comma-separated list of 32-bit - * integers starting at base. - */ -void run_container_printf_as_uint32_array(const run_container_t *cont, - uint32_t base) { - if (cont->n_runs == 0) return; - { - uint32_t run_start = base + cont->runs[0].value; - uint16_t le = cont->runs[0].length; - printf("%u", run_start); - for (uint32_t j = 1; j <= le; ++j) printf(",%u", run_start + j); - } - for (int32_t i = 1; i < cont->n_runs; ++i) { - uint32_t run_start = base + cont->runs[i].value; - uint16_t le = cont->runs[i].length; - for (uint32_t j = 0; j <= le; ++j) printf(",%u", run_start + j); - } -} - -int32_t run_container_serialize(const run_container_t *container, char *buf) { - int32_t l, off; - - memcpy(buf, &container->n_runs, off = sizeof(container->n_runs)); - memcpy(&buf[off], &container->capacity, sizeof(container->capacity)); - off += sizeof(container->capacity); - - l = sizeof(rle16_t) * container->n_runs; - memcpy(&buf[off], container->runs, l); - return (off + l); -} - -int32_t run_container_write(const run_container_t *container, char *buf) { - memcpy(buf, &container->n_runs, sizeof(uint16_t)); - memcpy(buf + sizeof(uint16_t), container->runs, - container->n_runs * sizeof(rle16_t)); - return run_container_size_in_bytes(container); -} - -int32_t run_container_read(int32_t cardinality, run_container_t *container, - const char *buf) { - (void)cardinality; - memcpy(&container->n_runs, buf, sizeof(uint16_t)); - if (container->n_runs > container->capacity) - run_container_grow(container, container->n_runs, false); - if(container->n_runs > 0) { - memcpy(container->runs, buf + sizeof(uint16_t), - container->n_runs * sizeof(rle16_t)); - } - return run_container_size_in_bytes(container); -} - -uint32_t run_container_serialization_len(const run_container_t *container) { - return (sizeof(container->n_runs) + sizeof(container->capacity) + - sizeof(rle16_t) * container->n_runs); -} - -void *run_container_deserialize(const char *buf, size_t buf_len) { - run_container_t *ptr; - - if (buf_len < 8 /* n_runs + capacity */) - return (NULL); - else - buf_len -= 8; - - if ((ptr = (run_container_t *)malloc(sizeof(run_container_t))) != NULL) { - size_t len; - int32_t off; - - memcpy(&ptr->n_runs, buf, off = 4); - memcpy(&ptr->capacity, &buf[off], 4); - off += 4; - - len = sizeof(rle16_t) * ptr->n_runs; - - if (len != buf_len) { - free(ptr); - return (NULL); - } - - if ((ptr->runs = (rle16_t *)malloc(len)) == NULL) { - free(ptr); - return (NULL); - } - - memcpy(ptr->runs, &buf[off], len); - - /* Check if returned values are monotonically increasing */ - for (int32_t i = 0, j = 0; i < ptr->n_runs; i++) { - if (ptr->runs[i].value < j) { - free(ptr->runs); - free(ptr); - return (NULL); - } else - j = ptr->runs[i].value; - } - } - - return (ptr); -} - -bool run_container_iterate(const run_container_t *cont, uint32_t base, - roaring_iterator iterator, void *ptr) { - for (int i = 0; i < cont->n_runs; ++i) { - uint32_t run_start = base + cont->runs[i].value; - uint16_t le = cont->runs[i].length; - - for (int j = 0; j <= le; ++j) - if (!iterator(run_start + j, ptr)) return false; - } - return true; -} - -bool run_container_iterate64(const run_container_t *cont, uint32_t base, - roaring_iterator64 iterator, uint64_t high_bits, - void *ptr) { - for (int i = 0; i < cont->n_runs; ++i) { - uint32_t run_start = base + cont->runs[i].value; - uint16_t le = cont->runs[i].length; - - for (int j = 0; j <= le; ++j) - if (!iterator(high_bits | (uint64_t)(run_start + j), ptr)) - return false; - } - return true; -} - -bool run_container_equals(const run_container_t *container1, - const run_container_t *container2) { - if (container1->n_runs != container2->n_runs) { - return false; - } - for (int32_t i = 0; i < container1->n_runs; ++i) { - if ((container1->runs[i].value != container2->runs[i].value) || - (container1->runs[i].length != container2->runs[i].length)) - return false; - } - return true; -} - -bool run_container_is_subset(const run_container_t *container1, - const run_container_t *container2) { - int i1 = 0, i2 = 0; - while (i1 < container1->n_runs && i2 < container2->n_runs) { - int start1 = container1->runs[i1].value; - int stop1 = start1 + container1->runs[i1].length; - int start2 = container2->runs[i2].value; - int stop2 = start2 + container2->runs[i2].length; - if (start1 < start2) { - return false; - } else { // start1 >= start2 - if (stop1 < stop2) { - i1++; - } else if (stop1 == stop2) { - i1++; - i2++; - } else { // stop1 > stop2 - i2++; - } - } - } - if (i1 == container1->n_runs) { - return true; - } else { - return false; - } -} - -// TODO: write smart_append_exclusive version to match the overloaded 1 param -// Java version (or is it even used?) - -// follows the Java implementation closely -// length is the rle-value. Ie, run [10,12) uses a length value 1. -void run_container_smart_append_exclusive(run_container_t *src, - const uint16_t start, - const uint16_t length) { - int old_end; - rle16_t *last_run = src->n_runs ? src->runs + (src->n_runs - 1) : NULL; - rle16_t *appended_last_run = src->runs + src->n_runs; - - if (!src->n_runs || - (start > (old_end = last_run->value + last_run->length + 1))) { - *appended_last_run = (rle16_t){.value = start, .length = length}; - src->n_runs++; - return; - } - if (old_end == start) { - // we merge - last_run->length += (length + 1); - return; - } - int new_end = start + length + 1; - - if (start == last_run->value) { - // wipe out previous - if (new_end < old_end) { - *last_run = (rle16_t){.value = (uint16_t)new_end, - .length = (uint16_t)(old_end - new_end - 1)}; - return; - } else if (new_end > old_end) { - *last_run = (rle16_t){.value = (uint16_t)old_end, - .length = (uint16_t)(new_end - old_end - 1)}; - return; - } else { - src->n_runs--; - return; - } - } - last_run->length = start - last_run->value - 1; - if (new_end < old_end) { - *appended_last_run = - (rle16_t){.value = (uint16_t)new_end, - .length = (uint16_t)(old_end - new_end - 1)}; - src->n_runs++; - } else if (new_end > old_end) { - *appended_last_run = - (rle16_t){.value = (uint16_t)old_end, - .length = (uint16_t)(new_end - old_end - 1)}; - src->n_runs++; - } -} - -bool run_container_select(const run_container_t *container, - uint32_t *start_rank, uint32_t rank, - uint32_t *element) { - for (int i = 0; i < container->n_runs; i++) { - uint16_t length = container->runs[i].length; - if (rank <= *start_rank + length) { - uint16_t value = container->runs[i].value; - *element = value + rank - (*start_rank); - return true; - } else - *start_rank += length + 1; - } - return false; -} - -int run_container_rank(const run_container_t *container, uint16_t x) { - int sum = 0; - uint32_t x32 = x; - for (int i = 0; i < container->n_runs; i++) { - uint32_t startpoint = container->runs[i].value; - uint32_t length = container->runs[i].length; - uint32_t endpoint = length + startpoint; - if (x <= endpoint) { - if (x < startpoint) break; - return sum + (x32 - startpoint) + 1; - } else { - sum += length + 1; - } - } - return sum; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/containers/run.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/roaring.c */ -#include -#include -#include -#include -#include -#include - -extern inline bool roaring_bitmap_contains(const roaring_bitmap_t *r, - uint32_t val); - -// this is like roaring_bitmap_add, but it populates pointer arguments in such a -// way -// that we can recover the container touched, which, in turn can be used to -// accelerate some functions (when you repeatedly need to add to the same -// container) -void *containerptr_roaring_bitmap_add(roaring_bitmap_t *r, - uint32_t val, - uint8_t *typecode, - int *index) { - uint16_t hb = val >> 16; - const int i = ra_get_index(&r->high_low_container, hb); - if (i >= 0) { - ra_unshare_container_at_index(&r->high_low_container, i); - void *container = - ra_get_container_at_index(&r->high_low_container, i, typecode); - uint8_t newtypecode = *typecode; - void *container2 = - container_add(container, val & 0xFFFF, *typecode, &newtypecode); - *index = i; - if (container2 != container) { - container_free(container, *typecode); - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - *typecode = newtypecode; - return container2; - } else { - return container; - } - } else { - array_container_t *newac = array_container_create(); - void *container = container_add(newac, val & 0xFFFF, - ARRAY_CONTAINER_TYPE_CODE, typecode); - // we could just assume that it stays an array container - ra_insert_new_key_value_at(&r->high_low_container, -i - 1, hb, - container, *typecode); - *index = -i - 1; - return container; - } -} - -roaring_bitmap_t *roaring_bitmap_create() { - roaring_bitmap_t *ans = - (roaring_bitmap_t *)malloc(sizeof(roaring_bitmap_t)); - if (!ans) { - return NULL; - } - bool is_ok = ra_init(&ans->high_low_container); - if (!is_ok) { - free(ans); - return NULL; - } - ans->copy_on_write = false; - return ans; -} - -roaring_bitmap_t *roaring_bitmap_create_with_capacity(uint32_t cap) { - roaring_bitmap_t *ans = - (roaring_bitmap_t *)malloc(sizeof(roaring_bitmap_t)); - if (!ans) { - return NULL; - } - bool is_ok = ra_init_with_capacity(&ans->high_low_container, cap); - if (!is_ok) { - free(ans); - return NULL; - } - ans->copy_on_write = false; - return ans; -} - -void roaring_bitmap_add_many(roaring_bitmap_t *r, size_t n_args, - const uint32_t *vals) { - void *container = NULL; // hold value of last container touched - uint8_t typecode = 0; // typecode of last container touched - uint32_t prev = 0; // previous valued inserted - size_t i = 0; // index of value - int containerindex = 0; - if (n_args == 0) return; - uint32_t val; - memcpy(&val, vals + i, sizeof(val)); - container = - containerptr_roaring_bitmap_add(r, val, &typecode, &containerindex); - prev = val; - i++; - for (; i < n_args; i++) { - memcpy(&val, vals + i, sizeof(val)); - if (((prev ^ val) >> 16) == - 0) { // no need to seek the container, it is at hand - // because we already have the container at hand, we can do the - // insertion - // automatically, bypassing the roaring_bitmap_add call - uint8_t newtypecode = typecode; - void *container2 = - container_add(container, val & 0xFFFF, typecode, &newtypecode); - if (container2 != container) { // rare instance when we need to - // change the container type - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, - containerindex, container2, - newtypecode); - typecode = newtypecode; - container = container2; - } - } else { - container = containerptr_roaring_bitmap_add(r, val, &typecode, - &containerindex); - } - prev = val; - } -} - -roaring_bitmap_t *roaring_bitmap_of_ptr(size_t n_args, const uint32_t *vals) { - roaring_bitmap_t *answer = roaring_bitmap_create(); - roaring_bitmap_add_many(answer, n_args, vals); - return answer; -} - -roaring_bitmap_t *roaring_bitmap_of(size_t n_args, ...) { - // todo: could be greatly optimized but we do not expect this call to ever - // include long lists - roaring_bitmap_t *answer = roaring_bitmap_create(); - va_list ap; - va_start(ap, n_args); - for (size_t i = 1; i <= n_args; i++) { - uint32_t val = va_arg(ap, uint32_t); - roaring_bitmap_add(answer, val); - } - va_end(ap); - return answer; -} - -static inline uint32_t minimum_uint32(uint32_t a, uint32_t b) { - return (a < b) ? a : b; -} - -static inline uint64_t minimum_uint64(uint64_t a, uint64_t b) { - return (a < b) ? a : b; -} - -roaring_bitmap_t *roaring_bitmap_from_range(uint64_t min, uint64_t max, - uint32_t step) { - if(max >= UINT64_C(0x100000000)) { - max = UINT64_C(0x100000000); - } - if (step == 0) return NULL; - if (max <= min) return NULL; - roaring_bitmap_t *answer = roaring_bitmap_create(); - if (step >= (1 << 16)) { - for (uint32_t value = (uint32_t)min; value < max; value += step) { - roaring_bitmap_add(answer, value); - } - return answer; - } - uint64_t min_tmp = min; - do { - uint32_t key = (uint32_t)min_tmp >> 16; - uint32_t container_min = min_tmp & 0xFFFF; - uint32_t container_max = (uint32_t)minimum_uint64(max - (key << 16), 1 << 16); - uint8_t type; - void *container = container_from_range(&type, container_min, - container_max, (uint16_t)step); - ra_append(&answer->high_low_container, key, container, type); - uint32_t gap = container_max - container_min + step - 1; - min_tmp += gap - (gap % step); - } while (min_tmp < max); - // cardinality of bitmap will be ((uint64_t) max - min + step - 1 ) / step - return answer; -} - -void roaring_bitmap_add_range_closed(roaring_bitmap_t *ra, uint32_t min, uint32_t max) { - if (min > max) { - return; - } - - uint32_t min_key = min >> 16; - uint32_t max_key = max >> 16; - - int32_t num_required_containers = max_key - min_key + 1; - int32_t suffix_length = count_greater(ra->high_low_container.keys, - ra->high_low_container.size, - max_key); - int32_t prefix_length = count_less(ra->high_low_container.keys, - ra->high_low_container.size - suffix_length, - min_key); - int32_t common_length = ra->high_low_container.size - prefix_length - suffix_length; - - if (num_required_containers > common_length) { - ra_shift_tail(&ra->high_low_container, suffix_length, - num_required_containers - common_length); - } - - int32_t src = prefix_length + common_length - 1; - int32_t dst = ra->high_low_container.size - suffix_length - 1; - for (uint32_t key = max_key; key != min_key-1; key--) { // beware of min_key==0 - uint32_t container_min = (min_key == key) ? (min & 0xffff) : 0; - uint32_t container_max = (max_key == key) ? (max & 0xffff) : 0xffff; - void* new_container; - uint8_t new_type; - - if (src >= 0 && ra->high_low_container.keys[src] == key) { - ra_unshare_container_at_index(&ra->high_low_container, src); - new_container = container_add_range(ra->high_low_container.containers[src], - ra->high_low_container.typecodes[src], - container_min, container_max, &new_type); - if (new_container != ra->high_low_container.containers[src]) { - container_free(ra->high_low_container.containers[src], - ra->high_low_container.typecodes[src]); - } - src--; - } else { - new_container = container_from_range(&new_type, container_min, - container_max+1, 1); - } - ra_replace_key_and_container_at_index(&ra->high_low_container, dst, - key, new_container, new_type); - dst--; - } -} - -void roaring_bitmap_remove_range_closed(roaring_bitmap_t *ra, uint32_t min, uint32_t max) { - if (min > max) { - return; - } - - uint32_t min_key = min >> 16; - uint32_t max_key = max >> 16; - - int32_t src = count_less(ra->high_low_container.keys, ra->high_low_container.size, min_key); - int32_t dst = src; - while (src < ra->high_low_container.size && ra->high_low_container.keys[src] <= max_key) { - uint32_t container_min = (min_key == ra->high_low_container.keys[src]) ? (min & 0xffff) : 0; - uint32_t container_max = (max_key == ra->high_low_container.keys[src]) ? (max & 0xffff) : 0xffff; - ra_unshare_container_at_index(&ra->high_low_container, src); - void *new_container; - uint8_t new_type; - new_container = container_remove_range(ra->high_low_container.containers[src], - ra->high_low_container.typecodes[src], - container_min, container_max, - &new_type); - if (new_container != ra->high_low_container.containers[src]) { - container_free(ra->high_low_container.containers[src], - ra->high_low_container.typecodes[src]); - } - if (new_container) { - ra_replace_key_and_container_at_index(&ra->high_low_container, dst, - ra->high_low_container.keys[src], - new_container, new_type); - dst++; - } - src++; - } - if (src > dst) { - ra_shift_tail(&ra->high_low_container, ra->high_low_container.size - src, dst - src); - } -} - -void roaring_bitmap_add_range(roaring_bitmap_t *ra, uint64_t min, uint64_t max); -void roaring_bitmap_remove_range(roaring_bitmap_t *ra, uint64_t min, uint64_t max); - -void roaring_bitmap_printf(const roaring_bitmap_t *ra) { - printf("{"); - for (int i = 0; i < ra->high_low_container.size; ++i) { - container_printf_as_uint32_array( - ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - ((uint32_t)ra->high_low_container.keys[i]) << 16); - if (i + 1 < ra->high_low_container.size) printf(","); - } - printf("}"); -} - -void roaring_bitmap_printf_describe(const roaring_bitmap_t *ra) { - printf("{"); - for (int i = 0; i < ra->high_low_container.size; ++i) { - printf("%d: %s (%d)", ra->high_low_container.keys[i], - get_full_container_name(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]), - container_get_cardinality(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i])); - if (ra->high_low_container.typecodes[i] == SHARED_CONTAINER_TYPE_CODE) { - printf( - "(shared count = %" PRIu32 " )", - ((shared_container_t *)(ra->high_low_container.containers[i])) - ->counter); - } - - if (i + 1 < ra->high_low_container.size) printf(", "); - } - printf("}"); -} - -typedef struct min_max_sum_s { - uint32_t min; - uint32_t max; - uint64_t sum; -} min_max_sum_t; - -static bool min_max_sum_fnc(uint32_t value, void *param) { - min_max_sum_t *mms = (min_max_sum_t *)param; - if (value > mms->max) mms->max = value; - if (value < mms->min) mms->min = value; - mms->sum += value; - return true; // we always process all data points -} - -/** -* (For advanced users.) -* Collect statistics about the bitmap -*/ -void roaring_bitmap_statistics(const roaring_bitmap_t *ra, - roaring_statistics_t *stat) { - memset(stat, 0, sizeof(*stat)); - stat->n_containers = ra->high_low_container.size; - stat->cardinality = roaring_bitmap_get_cardinality(ra); - min_max_sum_t mms; - mms.min = UINT32_C(0xFFFFFFFF); - mms.max = UINT32_C(0); - mms.sum = 0; - roaring_iterate(ra, &min_max_sum_fnc, &mms); - stat->min_value = mms.min; - stat->max_value = mms.max; - stat->sum_value = mms.sum; - - for (int i = 0; i < ra->high_low_container.size; ++i) { - uint8_t truetype = - get_container_type(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - uint32_t card = - container_get_cardinality(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - uint32_t sbytes = - container_size_in_bytes(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - switch (truetype) { - case BITSET_CONTAINER_TYPE_CODE: - stat->n_bitset_containers++; - stat->n_values_bitset_containers += card; - stat->n_bytes_bitset_containers += sbytes; - break; - case ARRAY_CONTAINER_TYPE_CODE: - stat->n_array_containers++; - stat->n_values_array_containers += card; - stat->n_bytes_array_containers += sbytes; - break; - case RUN_CONTAINER_TYPE_CODE: - stat->n_run_containers++; - stat->n_values_run_containers += card; - stat->n_bytes_run_containers += sbytes; - break; - default: - assert(false); - __builtin_unreachable(); - } - } -} - -roaring_bitmap_t *roaring_bitmap_copy(const roaring_bitmap_t *r) { - roaring_bitmap_t *ans = - (roaring_bitmap_t *)malloc(sizeof(roaring_bitmap_t)); - if (!ans) { - return NULL; - } - bool is_ok = ra_copy(&r->high_low_container, &ans->high_low_container, - r->copy_on_write); - if (!is_ok) { - free(ans); - return NULL; - } - ans->copy_on_write = r->copy_on_write; - return ans; -} - -bool roaring_bitmap_overwrite(roaring_bitmap_t *dest, - const roaring_bitmap_t *src) { - return ra_overwrite(&src->high_low_container, &dest->high_low_container, - src->copy_on_write); -} - -void roaring_bitmap_free(roaring_bitmap_t *r) { - ra_clear(&r->high_low_container); - free(r); -} - -void roaring_bitmap_clear(roaring_bitmap_t *r) { - ra_reset(&r->high_low_container); -} - -void roaring_bitmap_add(roaring_bitmap_t *r, uint32_t val) { - const uint16_t hb = val >> 16; - const int i = ra_get_index(&r->high_low_container, hb); - uint8_t typecode; - if (i >= 0) { - ra_unshare_container_at_index(&r->high_low_container, i); - void *container = - ra_get_container_at_index(&r->high_low_container, i, &typecode); - uint8_t newtypecode = typecode; - void *container2 = - container_add(container, val & 0xFFFF, typecode, &newtypecode); - if (container2 != container) { - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - } - } else { - array_container_t *newac = array_container_create(); - void *container = container_add(newac, val & 0xFFFF, - ARRAY_CONTAINER_TYPE_CODE, &typecode); - // we could just assume that it stays an array container - ra_insert_new_key_value_at(&r->high_low_container, -i - 1, hb, - container, typecode); - } -} - -bool roaring_bitmap_add_checked(roaring_bitmap_t *r, uint32_t val) { - const uint16_t hb = val >> 16; - const int i = ra_get_index(&r->high_low_container, hb); - uint8_t typecode; - bool result = false; - if (i >= 0) { - ra_unshare_container_at_index(&r->high_low_container, i); - void *container = - ra_get_container_at_index(&r->high_low_container, i, &typecode); - - const int oldCardinality = - container_get_cardinality(container, typecode); - - uint8_t newtypecode = typecode; - void *container2 = - container_add(container, val & 0xFFFF, typecode, &newtypecode); - if (container2 != container) { - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - result = true; - } else { - const int newCardinality = - container_get_cardinality(container, newtypecode); - - result = oldCardinality != newCardinality; - } - } else { - array_container_t *newac = array_container_create(); - void *container = container_add(newac, val & 0xFFFF, - ARRAY_CONTAINER_TYPE_CODE, &typecode); - // we could just assume that it stays an array container - ra_insert_new_key_value_at(&r->high_low_container, -i - 1, hb, - container, typecode); - result = true; - } - - return result; -} - -void roaring_bitmap_remove(roaring_bitmap_t *r, uint32_t val) { - const uint16_t hb = val >> 16; - const int i = ra_get_index(&r->high_low_container, hb); - uint8_t typecode; - if (i >= 0) { - ra_unshare_container_at_index(&r->high_low_container, i); - void *container = - ra_get_container_at_index(&r->high_low_container, i, &typecode); - uint8_t newtypecode = typecode; - void *container2 = - container_remove(container, val & 0xFFFF, typecode, &newtypecode); - if (container2 != container) { - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - } - if (container_get_cardinality(container2, newtypecode) != 0) { - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - } else { - ra_remove_at_index_and_free(&r->high_low_container, i); - } - } -} - -bool roaring_bitmap_remove_checked(roaring_bitmap_t *r, uint32_t val) { - const uint16_t hb = val >> 16; - const int i = ra_get_index(&r->high_low_container, hb); - uint8_t typecode; - bool result = false; - if (i >= 0) { - ra_unshare_container_at_index(&r->high_low_container, i); - void *container = - ra_get_container_at_index(&r->high_low_container, i, &typecode); - - const int oldCardinality = - container_get_cardinality(container, typecode); - - uint8_t newtypecode = typecode; - void *container2 = - container_remove(container, val & 0xFFFF, typecode, &newtypecode); - if (container2 != container) { - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - } - - const int newCardinality = - container_get_cardinality(container2, newtypecode); - - if (newCardinality != 0) { - ra_set_container_at_index(&r->high_low_container, i, container2, - newtypecode); - } else { - ra_remove_at_index_and_free(&r->high_low_container, i); - } - - result = oldCardinality != newCardinality; - } - return result; -} - -void roaring_bitmap_remove_many(roaring_bitmap_t *r, size_t n_args, - const uint32_t *vals) { - if (n_args == 0 || r->high_low_container.size == 0) { - return; - } - int32_t pos = -1; // position of the container used in the previous iteration - for (size_t i = 0; i < n_args; i++) { - uint16_t key = (uint16_t)(vals[i] >> 16); - if (pos < 0 || key != r->high_low_container.keys[pos]) { - pos = ra_get_index(&r->high_low_container, key); - } - if (pos >= 0) { - uint8_t new_typecode; - void *new_container; - new_container = container_remove(r->high_low_container.containers[pos], - vals[i] & 0xffff, - r->high_low_container.typecodes[pos], - &new_typecode); - if (new_container != r->high_low_container.containers[pos]) { - container_free(r->high_low_container.containers[pos], - r->high_low_container.typecodes[pos]); - ra_replace_key_and_container_at_index(&r->high_low_container, - pos, key, new_container, - new_typecode); - } - if (!container_nonzero_cardinality(new_container, new_typecode)) { - container_free(new_container, new_typecode); - ra_remove_at_index(&r->high_low_container, pos); - pos = -1; - } - } - } -} - -// there should be some SIMD optimizations possible here -roaring_bitmap_t *roaring_bitmap_and(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - uint32_t neededcap = length1 > length2 ? length2 : length1; - roaring_bitmap_t *answer = roaring_bitmap_create_with_capacity(neededcap); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - - int pos1 = 0, pos2 = 0; - - while (pos1 < length1 && pos2 < length2) { - const uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - const uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - if (s1 == s2) { - uint8_t container_type_1, container_type_2; - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = container_and(c1, container_type_1, c2, container_type_2, - &container_result_type); - if (container_nonzero_cardinality(c, container_result_type)) { - ra_append(&answer->high_low_container, s1, c, - container_result_type); - } else { - container_free( - c, container_result_type); // otherwise:memory leak! - } - ++pos1; - ++pos2; - } else if (s1 < s2) { // s1 < s2 - pos1 = ra_advance_until(&x1->high_low_container, s2, pos1); - } else { // s1 > s2 - pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); - } - } - return answer; -} - -/** - * Compute the union of 'number' bitmaps. - */ -roaring_bitmap_t *roaring_bitmap_or_many(size_t number, - const roaring_bitmap_t **x) { - if (number == 0) { - return roaring_bitmap_create(); - } - if (number == 1) { - return roaring_bitmap_copy(x[0]); - } - roaring_bitmap_t *answer = - roaring_bitmap_lazy_or(x[0], x[1], LAZY_OR_BITSET_CONVERSION); - for (size_t i = 2; i < number; i++) { - roaring_bitmap_lazy_or_inplace(answer, x[i], LAZY_OR_BITSET_CONVERSION); - } - roaring_bitmap_repair_after_lazy(answer); - return answer; -} - -/** - * Compute the xor of 'number' bitmaps. - */ -roaring_bitmap_t *roaring_bitmap_xor_many(size_t number, - const roaring_bitmap_t **x) { - if (number == 0) { - return roaring_bitmap_create(); - } - if (number == 1) { - return roaring_bitmap_copy(x[0]); - } - roaring_bitmap_t *answer = roaring_bitmap_lazy_xor(x[0], x[1]); - for (size_t i = 2; i < number; i++) { - roaring_bitmap_lazy_xor_inplace(answer, x[i]); - } - roaring_bitmap_repair_after_lazy(answer); - return answer; -} - -// inplace and (modifies its first argument). -void roaring_bitmap_and_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - if (x1 == x2) return; - int pos1 = 0, pos2 = 0, intersection_size = 0; - const int length1 = ra_get_size(&x1->high_low_container); - const int length2 = ra_get_size(&x2->high_low_container); - - // any skipped-over or newly emptied containers in x1 - // have to be freed. - while (pos1 < length1 && pos2 < length2) { - const uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - const uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - if (s1 == s2) { - uint8_t typecode1, typecode2, typecode_result; - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &typecode1); - c1 = get_writable_copy_if_shared(c1, &typecode1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &typecode2); - void *c = - container_iand(c1, typecode1, c2, typecode2, &typecode_result); - if (c != c1) { // in this instance a new container was created, and - // we need to free the old one - container_free(c1, typecode1); - } - if (container_nonzero_cardinality(c, typecode_result)) { - ra_replace_key_and_container_at_index(&x1->high_low_container, - intersection_size, s1, c, - typecode_result); - intersection_size++; - } else { - container_free(c, typecode_result); - } - ++pos1; - ++pos2; - } else if (s1 < s2) { - pos1 = ra_advance_until_freeing(&x1->high_low_container, s2, pos1); - } else { // s1 > s2 - pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); - } - } - - // if we ended early because x2 ran out, then all remaining in x1 should be - // freed - while (pos1 < length1) { - container_free(x1->high_low_container.containers[pos1], - x1->high_low_container.typecodes[pos1]); - ++pos1; - } - - // all containers after this have either been copied or freed - ra_downsize(&x1->high_low_container, intersection_size); -} - -roaring_bitmap_t *roaring_bitmap_or(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - if (0 == length1) { - return roaring_bitmap_copy(x2); - } - if (0 == length2) { - return roaring_bitmap_copy(x1); - } - roaring_bitmap_t *answer = - roaring_bitmap_create_with_capacity(length1 + length2); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = container_or(c1, container_type_1, c2, container_type_2, - &container_result_type); - // since we assume that the initial containers are non-empty, the - // result here - // can only be non-empty - ra_append(&answer->high_low_container, s1, c, - container_result_type); - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - // c1 = container_clone(c1, container_type_1); - c1 = - get_copy_of_container(c1, &container_type_1, x1->copy_on_write); - if (x1->copy_on_write) { - ra_set_container_at_index(&x1->high_low_container, pos1, c1, - container_type_1); - } - ra_append(&answer->high_low_container, s1, c1, container_type_1); - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - // c2 = container_clone(c2, container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_append(&answer->high_low_container, s2, c2, container_type_2); - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&answer->high_low_container, - &x2->high_low_container, pos2, length2, - x2->copy_on_write); - } else if (pos2 == length2) { - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1, - x1->copy_on_write); - } - return answer; -} - -// inplace or (modifies its first argument). -void roaring_bitmap_or_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - int length1 = x1->high_low_container.size; - const int length2 = x2->high_low_container.size; - - if (0 == length2) return; - - if (0 == length1) { - roaring_bitmap_overwrite(x1, x2); - return; - } - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - if (!container_is_full(c1, container_type_1)) { - c1 = get_writable_copy_if_shared(c1, &container_type_1); - - void *c2 = ra_get_container_at_index(&x2->high_low_container, - pos2, &container_type_2); - void *c = - container_ior(c1, container_type_1, c2, container_type_2, - &container_result_type); - if (c != - c1) { // in this instance a new container was created, and - // we need to free the old one - container_free(c1, container_type_1); - } - - ra_set_container_at_index(&x1->high_low_container, pos1, c, - container_result_type); - } - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - - // void *c2_clone = container_clone(c2, container_type_2); - ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, - container_type_2); - pos1++; - length1++; - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, - pos2, length2, x2->copy_on_write); - } -} - -roaring_bitmap_t *roaring_bitmap_xor(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - if (0 == length1) { - return roaring_bitmap_copy(x2); - } - if (0 == length2) { - return roaring_bitmap_copy(x1); - } - roaring_bitmap_t *answer = - roaring_bitmap_create_with_capacity(length1 + length2); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = container_xor(c1, container_type_1, c2, container_type_2, - &container_result_type); - - if (container_nonzero_cardinality(c, container_result_type)) { - ra_append(&answer->high_low_container, s1, c, - container_result_type); - } else { - container_free(c, container_result_type); - } - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = - get_copy_of_container(c1, &container_type_1, x1->copy_on_write); - if (x1->copy_on_write) { - ra_set_container_at_index(&x1->high_low_container, pos1, c1, - container_type_1); - } - ra_append(&answer->high_low_container, s1, c1, container_type_1); - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_append(&answer->high_low_container, s2, c2, container_type_2); - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&answer->high_low_container, - &x2->high_low_container, pos2, length2, - x2->copy_on_write); - } else if (pos2 == length2) { - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1, - x1->copy_on_write); - } - return answer; -} - -// inplace xor (modifies its first argument). - -void roaring_bitmap_xor_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - assert(x1 != x2); - uint8_t container_result_type = 0; - int length1 = x1->high_low_container.size; - const int length2 = x2->high_low_container.size; - - if (0 == length2) return; - - if (0 == length1) { - roaring_bitmap_overwrite(x1, x2); - return; - } - - // XOR can have new containers inserted from x2, but can also - // lose containers when x1 and x2 are nonempty and identical. - - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = get_writable_copy_if_shared(c1, &container_type_1); - - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = container_ixor(c1, container_type_1, c2, container_type_2, - &container_result_type); - - if (container_nonzero_cardinality(c, container_result_type)) { - ra_set_container_at_index(&x1->high_low_container, pos1, c, - container_result_type); - ++pos1; - } else { - container_free(c, container_result_type); - ra_remove_at_index(&x1->high_low_container, pos1); - --length1; - } - - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - - ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, - container_type_2); - pos1++; - length1++; - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, - pos2, length2, x2->copy_on_write); - } -} - -roaring_bitmap_t *roaring_bitmap_andnot(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - if (0 == length1) { - roaring_bitmap_t *empty_bitmap = roaring_bitmap_create(); - empty_bitmap->copy_on_write = x1->copy_on_write && x2->copy_on_write; - return empty_bitmap; - } - if (0 == length2) { - return roaring_bitmap_copy(x1); - } - roaring_bitmap_t *answer = roaring_bitmap_create_with_capacity(length1); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = 0; - uint16_t s2 = 0; - while (true) { - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = - container_andnot(c1, container_type_1, c2, container_type_2, - &container_result_type); - - if (container_nonzero_cardinality(c, container_result_type)) { - ra_append(&answer->high_low_container, s1, c, - container_result_type); - } else { - container_free(c, container_result_type); - } - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - } else if (s1 < s2) { // s1 < s2 - const int next_pos1 = - ra_advance_until(&x1->high_low_container, s2, pos1); - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, next_pos1, - x1->copy_on_write); - // TODO : perhaps some of the copy_on_write should be based on - // answer rather than x1 (more stringent?). Many similar cases - pos1 = next_pos1; - if (pos1 == length1) break; - } else { // s1 > s2 - pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); - if (pos2 == length2) break; - } - } - if (pos2 == length2) { - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1, - x1->copy_on_write); - } - return answer; -} - -// inplace andnot (modifies its first argument). - -void roaring_bitmap_andnot_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - assert(x1 != x2); - - uint8_t container_result_type = 0; - int length1 = x1->high_low_container.size; - const int length2 = x2->high_low_container.size; - int intersection_size = 0; - - if (0 == length2) return; - - if (0 == length1) { - roaring_bitmap_clear(x1); - return; - } - - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = get_writable_copy_if_shared(c1, &container_type_1); - - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = - container_iandnot(c1, container_type_1, c2, container_type_2, - &container_result_type); - - if (container_nonzero_cardinality(c, container_result_type)) { - ra_replace_key_and_container_at_index(&x1->high_low_container, - intersection_size++, s1, - c, container_result_type); - } else { - container_free(c, container_result_type); - } - - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - if (pos1 != intersection_size) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, - pos1, &container_type_1); - - ra_replace_key_and_container_at_index(&x1->high_low_container, - intersection_size, s1, c1, - container_type_1); - } - intersection_size++; - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - - if (pos1 < length1) { - // all containers between intersection_size and - // pos1 are junk. However, they have either been moved - // (thus still referenced) or involved in an iandnot - // that will clean up all containers that could not be reused. - // Thus we should not free the junk containers between - // intersection_size and pos1. - if (pos1 > intersection_size) { - // left slide of remaining items - ra_copy_range(&x1->high_low_container, pos1, length1, - intersection_size); - } - // else current placement is fine - intersection_size += (length1 - pos1); - } - ra_downsize(&x1->high_low_container, intersection_size); -} - -uint64_t roaring_bitmap_get_cardinality(const roaring_bitmap_t *ra) { - uint64_t card = 0; - for (int i = 0; i < ra->high_low_container.size; ++i) - card += container_get_cardinality(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - return card; -} - -uint64_t roaring_bitmap_range_cardinality(const roaring_bitmap_t *ra, - uint64_t range_start, - uint64_t range_end) { - if (range_end > UINT32_MAX) { - range_end = UINT32_MAX + UINT64_C(1); - } - if (range_start >= range_end) { - return 0; - } - range_end--; // make range_end inclusive - // now we have: 0 <= range_start <= range_end <= UINT32_MAX - - int minhb = range_start >> 16; - int maxhb = range_end >> 16; - - uint64_t card = 0; - - int i = ra_get_index(&ra->high_low_container, minhb); - if (i >= 0) { - if (minhb == maxhb) { - card += container_rank(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - range_end & 0xffff); - } else { - card += container_get_cardinality(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - } - if ((range_start & 0xffff) != 0) { - card -= container_rank(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - (range_start & 0xffff) - 1); - } - i++; - } else { - i = -i - 1; - } - - for (; i < ra->high_low_container.size; i++) { - uint16_t key = ra->high_low_container.keys[i]; - if (key < maxhb) { - card += container_get_cardinality(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i]); - } else if (key == maxhb) { - card += container_rank(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - range_end & 0xffff); - break; - } else { - break; - } - } - - return card; -} - - -bool roaring_bitmap_is_empty(const roaring_bitmap_t *ra) { - return ra->high_low_container.size == 0; -} - -void roaring_bitmap_to_uint32_array(const roaring_bitmap_t *ra, uint32_t *ans) { - ra_to_uint32_array(&ra->high_low_container, ans); -} - -bool roaring_bitmap_range_uint32_array(const roaring_bitmap_t *ra, size_t offset, size_t limit, uint32_t *ans) { - return ra_range_uint32_array(&ra->high_low_container, offset, limit, ans); -} - -/** convert array and bitmap containers to run containers when it is more - * efficient; - * also convert from run containers when more space efficient. Returns - * true if the result has at least one run container. -*/ -bool roaring_bitmap_run_optimize(roaring_bitmap_t *r) { - bool answer = false; - for (int i = 0; i < r->high_low_container.size; i++) { - uint8_t typecode_original, typecode_after; - ra_unshare_container_at_index( - &r->high_low_container, i); // TODO: this introduces extra cloning! - void *c = ra_get_container_at_index(&r->high_low_container, i, - &typecode_original); - void *c1 = convert_run_optimize(c, typecode_original, &typecode_after); - if (typecode_after == RUN_CONTAINER_TYPE_CODE) answer = true; - ra_set_container_at_index(&r->high_low_container, i, c1, - typecode_after); - } - return answer; -} - -size_t roaring_bitmap_shrink_to_fit(roaring_bitmap_t *r) { - size_t answer = 0; - for (int i = 0; i < r->high_low_container.size; i++) { - uint8_t typecode_original; - void *c = ra_get_container_at_index(&r->high_low_container, i, - &typecode_original); - answer += container_shrink_to_fit(c, typecode_original); - } - answer += ra_shrink_to_fit(&r->high_low_container); - return answer; -} - -/** - * Remove run-length encoding even when it is more space efficient - * return whether a change was applied - */ -bool roaring_bitmap_remove_run_compression(roaring_bitmap_t *r) { - bool answer = false; - for (int i = 0; i < r->high_low_container.size; i++) { - uint8_t typecode_original, typecode_after; - void *c = ra_get_container_at_index(&r->high_low_container, i, - &typecode_original); - if (get_container_type(c, typecode_original) == - RUN_CONTAINER_TYPE_CODE) { - answer = true; - if (typecode_original == SHARED_CONTAINER_TYPE_CODE) { - run_container_t *truec = - (run_container_t *)((shared_container_t *)c)->container; - int32_t card = run_container_cardinality(truec); - void *c1 = convert_to_bitset_or_array_container( - truec, card, &typecode_after); - shared_container_free((shared_container_t *)c); - ra_set_container_at_index(&r->high_low_container, i, c1, - typecode_after); - - } else { - int32_t card = run_container_cardinality((run_container_t *)c); - void *c1 = convert_to_bitset_or_array_container( - (run_container_t *)c, card, &typecode_after); - ra_set_container_at_index(&r->high_low_container, i, c1, - typecode_after); - } - } - } - return answer; -} - -size_t roaring_bitmap_serialize(const roaring_bitmap_t *ra, char *buf) { - size_t portablesize = roaring_bitmap_portable_size_in_bytes(ra); - uint64_t cardinality = roaring_bitmap_get_cardinality(ra); - uint64_t sizeasarray = cardinality * sizeof(uint32_t) + sizeof(uint32_t); - if (portablesize < sizeasarray) { - buf[0] = SERIALIZATION_CONTAINER; - return roaring_bitmap_portable_serialize(ra, buf + 1) + 1; - } else { - buf[0] = SERIALIZATION_ARRAY_UINT32; - memcpy(buf + 1, &cardinality, sizeof(uint32_t)); - roaring_bitmap_to_uint32_array( - ra, (uint32_t *)(buf + 1 + sizeof(uint32_t))); - return 1 + (size_t)sizeasarray; - } -} - -size_t roaring_bitmap_size_in_bytes(const roaring_bitmap_t *ra) { - size_t portablesize = roaring_bitmap_portable_size_in_bytes(ra); - uint64_t sizeasarray = roaring_bitmap_get_cardinality(ra) * sizeof(uint32_t) + - sizeof(uint32_t); - return portablesize < sizeasarray ? portablesize + 1 : (size_t)sizeasarray + 1; -} - -size_t roaring_bitmap_portable_size_in_bytes(const roaring_bitmap_t *ra) { - return ra_portable_size_in_bytes(&ra->high_low_container); -} - - -roaring_bitmap_t *roaring_bitmap_portable_deserialize_safe(const char *buf, size_t maxbytes) { - roaring_bitmap_t *ans = - (roaring_bitmap_t *)malloc(sizeof(roaring_bitmap_t)); - if (ans == NULL) { - return NULL; - } - size_t bytesread; - bool is_ok = ra_portable_deserialize(&ans->high_low_container, buf, maxbytes, &bytesread); - if(is_ok) assert(bytesread <= maxbytes); - ans->copy_on_write = false; - if (!is_ok) { - free(ans); - return NULL; - } - return ans; -} - -roaring_bitmap_t *roaring_bitmap_portable_deserialize(const char *buf) { - return roaring_bitmap_portable_deserialize_safe(buf, SIZE_MAX); -} - - -size_t roaring_bitmap_portable_deserialize_size(const char *buf, size_t maxbytes) { - return ra_portable_deserialize_size(buf, maxbytes); -} - - -size_t roaring_bitmap_portable_serialize(const roaring_bitmap_t *ra, - char *buf) { - return ra_portable_serialize(&ra->high_low_container, buf); -} - -roaring_bitmap_t *roaring_bitmap_deserialize(const void *buf) { - const char *bufaschar = (const char *)buf; - if (*(const unsigned char *)buf == SERIALIZATION_ARRAY_UINT32) { - /* This looks like a compressed set of uint32_t elements */ - uint32_t card; - memcpy(&card, bufaschar + 1, sizeof(uint32_t)); - const uint32_t *elems = - (const uint32_t *)(bufaschar + 1 + sizeof(uint32_t)); - - return roaring_bitmap_of_ptr(card, elems); - } else if (bufaschar[0] == SERIALIZATION_CONTAINER) { - return roaring_bitmap_portable_deserialize(bufaschar + 1); - } else - return (NULL); -} - -bool roaring_iterate(const roaring_bitmap_t *ra, roaring_iterator iterator, - void *ptr) { - for (int i = 0; i < ra->high_low_container.size; ++i) - if (!container_iterate(ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - ((uint32_t)ra->high_low_container.keys[i]) << 16, - iterator, ptr)) { - return false; - } - return true; -} - -bool roaring_iterate64(const roaring_bitmap_t *ra, roaring_iterator64 iterator, - uint64_t high_bits, void *ptr) { - for (int i = 0; i < ra->high_low_container.size; ++i) - if (!container_iterate64( - ra->high_low_container.containers[i], - ra->high_low_container.typecodes[i], - ((uint32_t)ra->high_low_container.keys[i]) << 16, iterator, - high_bits, ptr)) { - return false; - } - return true; -} - -/**** -* begin roaring_uint32_iterator_t -*****/ - -static bool loadfirstvalue(roaring_uint32_iterator_t *newit) { - newit->in_container_index = 0; - newit->run_index = 0; - newit->current_value = 0; - if (newit->container_index >= - newit->parent->high_low_container.size) { // otherwise nothing - newit->current_value = UINT32_MAX; - return (newit->has_value = false); - } - // assume not empty - newit->has_value = true; - // we precompute container, typecode and highbits so that successive - // iterators do not have to grab them from odd memory locations - // and have to worry about the (easily predicted) container_unwrap_shared - // call. - newit->container = - newit->parent->high_low_container.containers[newit->container_index]; - newit->typecode = - newit->parent->high_low_container.typecodes[newit->container_index]; - newit->highbits = - ((uint32_t) - newit->parent->high_low_container.keys[newit->container_index]) - << 16; - newit->container = - container_unwrap_shared(newit->container, &(newit->typecode)); - uint32_t wordindex; - uint64_t word; // used for bitsets - switch (newit->typecode) { - case BITSET_CONTAINER_TYPE_CODE: - wordindex = 0; - while ((word = ((const bitset_container_t *)(newit->container)) - ->array[wordindex]) == 0) - wordindex++; // advance - // here "word" is non-zero - newit->in_container_index = wordindex * 64 + __builtin_ctzll(word); - newit->current_value = newit->highbits | newit->in_container_index; - break; - case ARRAY_CONTAINER_TYPE_CODE: - newit->current_value = - newit->highbits | - ((const array_container_t *)(newit->container))->array[0]; - break; - case RUN_CONTAINER_TYPE_CODE: - newit->current_value = - newit->highbits | - (((const run_container_t *)(newit->container))->runs[0].value); - newit->in_run_index = - newit->current_value + - (((const run_container_t *)(newit->container))->runs[0].length); - break; - default: - // if this ever happens, bug! - assert(false); - } // switch (typecode) - return true; -} - -// prerequesite: the value should be in range of the container -static bool loadfirstvalue_largeorequal(roaring_uint32_iterator_t *newit, uint32_t val) { - uint16_t lb = val & 0xFFFF; - newit->in_container_index = 0; - newit->run_index = 0; - newit->current_value = 0; - // assume it is found - newit->has_value = true; - newit->container = - newit->parent->high_low_container.containers[newit->container_index]; - newit->typecode = - newit->parent->high_low_container.typecodes[newit->container_index]; - newit->highbits = - ((uint32_t) - newit->parent->high_low_container.keys[newit->container_index]) - << 16; - newit->container = - container_unwrap_shared(newit->container, &(newit->typecode)); - switch (newit->typecode) { - case BITSET_CONTAINER_TYPE_CODE: - newit->in_container_index = bitset_container_index_equalorlarger((const bitset_container_t *)(newit->container), lb); - newit->current_value = newit->highbits | newit->in_container_index; - break; - case ARRAY_CONTAINER_TYPE_CODE: - newit->in_container_index = array_container_index_equalorlarger((const array_container_t *)(newit->container), lb); - newit->current_value = - newit->highbits | - ((const array_container_t *)(newit->container))->array[newit->in_container_index]; - break; - case RUN_CONTAINER_TYPE_CODE: - newit->run_index = run_container_index_equalorlarger((const run_container_t *)(newit->container), lb); - if(((const run_container_t *)(newit->container))->runs[newit->run_index].value <= lb) { - newit->current_value = val; - } else { - newit->current_value = - newit->highbits | - (((const run_container_t *)(newit->container))->runs[newit->run_index].value); - } - newit->in_run_index = - (newit->highbits | (((const run_container_t *)(newit->container))->runs[newit->run_index].value)) + - (((const run_container_t *)(newit->container))->runs[newit->run_index].length); - - break; - default: - // if this ever happens, bug! - assert(false); - } // switch (typecode) - return true; -} - -void roaring_init_iterator(const roaring_bitmap_t *ra, - roaring_uint32_iterator_t *newit) { - newit->parent = ra; - newit->container_index = 0; - newit->has_value = loadfirstvalue(newit); -} - -roaring_uint32_iterator_t *roaring_create_iterator(const roaring_bitmap_t *ra) { - roaring_uint32_iterator_t *newit = - (roaring_uint32_iterator_t *)malloc(sizeof(roaring_uint32_iterator_t)); - if (newit == NULL) return NULL; - roaring_init_iterator(ra, newit); - return newit; -} - -roaring_uint32_iterator_t *roaring_copy_uint32_iterator( - const roaring_uint32_iterator_t *it) { - roaring_uint32_iterator_t *newit = - (roaring_uint32_iterator_t *)malloc(sizeof(roaring_uint32_iterator_t)); - memcpy(newit, it, sizeof(roaring_uint32_iterator_t)); - return newit; -} - -bool roaring_move_uint32_iterator_equalorlarger(roaring_uint32_iterator_t *it, uint32_t val) { - uint16_t hb = val >> 16; - const int i = ra_get_index(& it->parent->high_low_container, hb); - if (i >= 0) { - uint32_t lowvalue = container_maximum(it->parent->high_low_container.containers[i], it->parent->high_low_container.typecodes[i]); - uint16_t lb = val & 0xFFFF; - if(lowvalue < lb ) { - it->container_index = i+1; // will have to load first value of next container - } else {// the value is necessarily within the range of the container - it->container_index = i; - it->has_value = loadfirstvalue_largeorequal(it, val); - return it->has_value; - } - } else { - // there is no matching, so we are going for the next container - it->container_index = -i-1; - } - it->has_value = loadfirstvalue(it); - return it->has_value; -} - - -bool roaring_advance_uint32_iterator(roaring_uint32_iterator_t *it) { - if (it->container_index >= it->parent->high_low_container.size) { - return (it->has_value = false); - } - uint32_t wordindex; // used for bitsets - uint64_t word; // used for bitsets - switch (it->typecode) { - case BITSET_CONTAINER_TYPE_CODE: - it->in_container_index++; - wordindex = it->in_container_index / 64; - if (wordindex >= BITSET_CONTAINER_SIZE_IN_WORDS) break; - word = ((const bitset_container_t *)(it->container)) - ->array[wordindex] & - (UINT64_MAX << (it->in_container_index % 64)); - // next part could be optimized/simplified - while ((word == 0) && - (wordindex + 1 < BITSET_CONTAINER_SIZE_IN_WORDS)) { - wordindex++; - word = ((const bitset_container_t *)(it->container)) - ->array[wordindex]; - } - if (word != 0) { - it->in_container_index = wordindex * 64 + __builtin_ctzll(word); - it->current_value = it->highbits | it->in_container_index; - return (it->has_value = true); - } - break; - case ARRAY_CONTAINER_TYPE_CODE: - it->in_container_index++; - if (it->in_container_index < - ((const array_container_t *)(it->container))->cardinality) { - it->current_value = it->highbits | - ((const array_container_t *)(it->container)) - ->array[it->in_container_index]; - return true; - } - break; - case RUN_CONTAINER_TYPE_CODE: - if(it->current_value == UINT32_MAX) { - return (it->has_value = false); // without this, we risk an overflow to zero - } - it->current_value++; - if (it->current_value <= it->in_run_index) { - return (it->has_value = true); - } - it->run_index++; - if (it->run_index < - ((const run_container_t *)(it->container))->n_runs) { - it->current_value = - it->highbits | (((const run_container_t *)(it->container)) - ->runs[it->run_index] - .value); - it->in_run_index = it->current_value + - ((const run_container_t *)(it->container)) - ->runs[it->run_index] - .length; - return (it->has_value = true); - } - break; - default: - // if this ever happens, bug! - assert(false); - } // switch (typecode) - // moving to next container - it->container_index++; - return (it->has_value = loadfirstvalue(it)); -} - -uint32_t roaring_read_uint32_iterator(roaring_uint32_iterator_t *it, uint32_t* buf, uint32_t count) { - uint32_t ret = 0; - uint32_t num_values; - uint32_t wordindex; // used for bitsets - uint64_t word; // used for bitsets - const array_container_t* acont; //TODO remove - const run_container_t* rcont; //TODO remove - const bitset_container_t* bcont; //TODO remove - - while (it->has_value && ret < count) { - switch (it->typecode) { - case BITSET_CONTAINER_TYPE_CODE: - bcont = (const bitset_container_t*)(it->container); - wordindex = it->in_container_index / 64; - word = bcont->array[wordindex] & (UINT64_MAX << (it->in_container_index % 64)); - do { - while (word != 0 && ret < count) { - buf[0] = it->highbits | (wordindex * 64 + __builtin_ctzll(word)); - word = word & (word - 1); - buf++; - ret++; - } - while (word == 0 && wordindex+1 < BITSET_CONTAINER_SIZE_IN_WORDS) { - wordindex++; - word = bcont->array[wordindex]; - } - } while (word != 0 && ret < count); - it->has_value = (word != 0); - if (it->has_value) { - it->in_container_index = wordindex * 64 + __builtin_ctzll(word); - it->current_value = it->highbits | it->in_container_index; - } - break; - case ARRAY_CONTAINER_TYPE_CODE: - acont = (const array_container_t *)(it->container); - num_values = minimum_uint32(acont->cardinality - it->in_container_index, count - ret); - for (uint32_t i = 0; i < num_values; i++) { - buf[i] = it->highbits | acont->array[it->in_container_index + i]; - } - buf += num_values; - ret += num_values; - it->in_container_index += num_values; - it->has_value = (it->in_container_index < acont->cardinality); - if (it->has_value) { - it->current_value = it->highbits | acont->array[it->in_container_index]; - } - break; - case RUN_CONTAINER_TYPE_CODE: - rcont = (const run_container_t*)(it->container); - //"in_run_index" name is misleading, read it as "max_value_in_current_run" - do { - num_values = minimum_uint32(it->in_run_index - it->current_value + 1, count - ret); - for (uint32_t i = 0; i < num_values; i++) { - buf[i] = it->current_value + i; - } - it->current_value += num_values; // this can overflow to zero: UINT32_MAX+1=0 - buf += num_values; - ret += num_values; - - if (it->current_value > it->in_run_index || it->current_value == 0) { - it->run_index++; - if (it->run_index < rcont->n_runs) { - it->current_value = it->highbits | rcont->runs[it->run_index].value; - it->in_run_index = it->current_value + rcont->runs[it->run_index].length; - } else { - it->has_value = false; - } - } - } while ((ret < count) && it->has_value); - break; - default: - assert(false); - } - if (it->has_value) { - assert(ret == count); - return ret; - } - it->container_index++; - it->has_value = loadfirstvalue(it); - } - return ret; -} - - - -void roaring_free_uint32_iterator(roaring_uint32_iterator_t *it) { free(it); } - -/**** -* end of roaring_uint32_iterator_t -*****/ - -bool roaring_bitmap_equals(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2) { - if (ra1->high_low_container.size != ra2->high_low_container.size) { - return false; - } - for (int i = 0; i < ra1->high_low_container.size; ++i) { - if (ra1->high_low_container.keys[i] != - ra2->high_low_container.keys[i]) { - return false; - } - } - for (int i = 0; i < ra1->high_low_container.size; ++i) { - bool areequal = container_equals(ra1->high_low_container.containers[i], - ra1->high_low_container.typecodes[i], - ra2->high_low_container.containers[i], - ra2->high_low_container.typecodes[i]); - if (!areequal) { - return false; - } - } - return true; -} - -bool roaring_bitmap_is_subset(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2) { - const int length1 = ra1->high_low_container.size, - length2 = ra2->high_low_container.size; - - int pos1 = 0, pos2 = 0; - - while (pos1 < length1 && pos2 < length2) { - const uint16_t s1 = ra_get_key_at_index(&ra1->high_low_container, pos1); - const uint16_t s2 = ra_get_key_at_index(&ra2->high_low_container, pos2); - - if (s1 == s2) { - uint8_t container_type_1, container_type_2; - void *c1 = ra_get_container_at_index(&ra1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&ra2->high_low_container, pos2, - &container_type_2); - bool subset = - container_is_subset(c1, container_type_1, c2, container_type_2); - if (!subset) return false; - ++pos1; - ++pos2; - } else if (s1 < s2) { // s1 < s2 - return false; - } else { // s1 > s2 - pos2 = ra_advance_until(&ra2->high_low_container, s1, pos2); - } - } - if (pos1 == length1) - return true; - else - return false; -} - -static void insert_flipped_container(roaring_array_t *ans_arr, - const roaring_array_t *x1_arr, uint16_t hb, - uint16_t lb_start, uint16_t lb_end) { - const int i = ra_get_index(x1_arr, hb); - const int j = ra_get_index(ans_arr, hb); - uint8_t ctype_in, ctype_out; - void *flipped_container = NULL; - if (i >= 0) { - void *container_to_flip = - ra_get_container_at_index(x1_arr, i, &ctype_in); - flipped_container = - container_not_range(container_to_flip, ctype_in, (uint32_t)lb_start, - (uint32_t)(lb_end + 1), &ctype_out); - - if (container_get_cardinality(flipped_container, ctype_out)) - ra_insert_new_key_value_at(ans_arr, -j - 1, hb, flipped_container, - ctype_out); - else { - container_free(flipped_container, ctype_out); - } - } else { - flipped_container = container_range_of_ones( - (uint32_t)lb_start, (uint32_t)(lb_end + 1), &ctype_out); - ra_insert_new_key_value_at(ans_arr, -j - 1, hb, flipped_container, - ctype_out); - } -} - -static void inplace_flip_container(roaring_array_t *x1_arr, uint16_t hb, - uint16_t lb_start, uint16_t lb_end) { - const int i = ra_get_index(x1_arr, hb); - uint8_t ctype_in, ctype_out; - void *flipped_container = NULL; - if (i >= 0) { - void *container_to_flip = - ra_get_container_at_index(x1_arr, i, &ctype_in); - flipped_container = container_inot_range( - container_to_flip, ctype_in, (uint32_t)lb_start, - (uint32_t)(lb_end + 1), &ctype_out); - // if a new container was created, the old one was already freed - if (container_get_cardinality(flipped_container, ctype_out)) { - ra_set_container_at_index(x1_arr, i, flipped_container, ctype_out); - } else { - container_free(flipped_container, ctype_out); - ra_remove_at_index(x1_arr, i); - } - - } else { - flipped_container = container_range_of_ones( - (uint32_t)lb_start, (uint32_t)(lb_end + 1), &ctype_out); - ra_insert_new_key_value_at(x1_arr, -i - 1, hb, flipped_container, - ctype_out); - } -} - -static void insert_fully_flipped_container(roaring_array_t *ans_arr, - const roaring_array_t *x1_arr, - uint16_t hb) { - const int i = ra_get_index(x1_arr, hb); - const int j = ra_get_index(ans_arr, hb); - uint8_t ctype_in, ctype_out; - void *flipped_container = NULL; - if (i >= 0) { - void *container_to_flip = - ra_get_container_at_index(x1_arr, i, &ctype_in); - flipped_container = - container_not(container_to_flip, ctype_in, &ctype_out); - if (container_get_cardinality(flipped_container, ctype_out)) - ra_insert_new_key_value_at(ans_arr, -j - 1, hb, flipped_container, - ctype_out); - else { - container_free(flipped_container, ctype_out); - } - } else { - flipped_container = container_range_of_ones(0U, 0x10000U, &ctype_out); - ra_insert_new_key_value_at(ans_arr, -j - 1, hb, flipped_container, - ctype_out); - } -} - -static void inplace_fully_flip_container(roaring_array_t *x1_arr, uint16_t hb) { - const int i = ra_get_index(x1_arr, hb); - uint8_t ctype_in, ctype_out; - void *flipped_container = NULL; - if (i >= 0) { - void *container_to_flip = - ra_get_container_at_index(x1_arr, i, &ctype_in); - flipped_container = - container_inot(container_to_flip, ctype_in, &ctype_out); - - if (container_get_cardinality(flipped_container, ctype_out)) { - ra_set_container_at_index(x1_arr, i, flipped_container, ctype_out); - } else { - container_free(flipped_container, ctype_out); - ra_remove_at_index(x1_arr, i); - } - - } else { - flipped_container = container_range_of_ones(0U, 0x10000U, &ctype_out); - ra_insert_new_key_value_at(x1_arr, -i - 1, hb, flipped_container, - ctype_out); - } -} - -roaring_bitmap_t *roaring_bitmap_flip(const roaring_bitmap_t *x1, - uint64_t range_start, - uint64_t range_end) { - if (range_start >= range_end) { - return roaring_bitmap_copy(x1); - } - if(range_end >= UINT64_C(0x100000000)) { - range_end = UINT64_C(0x100000000); - } - - roaring_bitmap_t *ans = roaring_bitmap_create(); - ans->copy_on_write = x1->copy_on_write; - - uint16_t hb_start = (uint16_t)(range_start >> 16); - const uint16_t lb_start = (uint16_t)range_start; // & 0xFFFF; - uint16_t hb_end = (uint16_t)((range_end - 1) >> 16); - const uint16_t lb_end = (uint16_t)(range_end - 1); // & 0xFFFF; - - ra_append_copies_until(&ans->high_low_container, &x1->high_low_container, - hb_start, x1->copy_on_write); - if (hb_start == hb_end) { - insert_flipped_container(&ans->high_low_container, - &x1->high_low_container, hb_start, lb_start, - lb_end); - } else { - // start and end containers are distinct - if (lb_start > 0) { - // handle first (partial) container - insert_flipped_container(&ans->high_low_container, - &x1->high_low_container, hb_start, - lb_start, 0xFFFF); - ++hb_start; // for the full containers. Can't wrap. - } - - if (lb_end != 0xFFFF) --hb_end; // later we'll handle the partial block - - for (uint32_t hb = hb_start; hb <= hb_end; ++hb) { - insert_fully_flipped_container(&ans->high_low_container, - &x1->high_low_container, hb); - } - - // handle a partial final container - if (lb_end != 0xFFFF) { - insert_flipped_container(&ans->high_low_container, - &x1->high_low_container, hb_end + 1, 0, - lb_end); - ++hb_end; - } - } - ra_append_copies_after(&ans->high_low_container, &x1->high_low_container, - hb_end, x1->copy_on_write); - return ans; -} - -void roaring_bitmap_flip_inplace(roaring_bitmap_t *x1, uint64_t range_start, - uint64_t range_end) { - if (range_start >= range_end) { - return; // empty range - } - if(range_end >= UINT64_C(0x100000000)) { - range_end = UINT64_C(0x100000000); - } - - uint16_t hb_start = (uint16_t)(range_start >> 16); - const uint16_t lb_start = (uint16_t)range_start; - uint16_t hb_end = (uint16_t)((range_end - 1) >> 16); - const uint16_t lb_end = (uint16_t)(range_end - 1); - - if (hb_start == hb_end) { - inplace_flip_container(&x1->high_low_container, hb_start, lb_start, - lb_end); - } else { - // start and end containers are distinct - if (lb_start > 0) { - // handle first (partial) container - inplace_flip_container(&x1->high_low_container, hb_start, lb_start, - 0xFFFF); - ++hb_start; // for the full containers. Can't wrap. - } - - if (lb_end != 0xFFFF) --hb_end; - - for (uint32_t hb = hb_start; hb <= hb_end; ++hb) { - inplace_fully_flip_container(&x1->high_low_container, hb); - } - // handle a partial final container - if (lb_end != 0xFFFF) { - inplace_flip_container(&x1->high_low_container, hb_end + 1, 0, - lb_end); - ++hb_end; - } - } -} - -roaring_bitmap_t *roaring_bitmap_lazy_or(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2, - const bool bitsetconversion) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - if (0 == length1) { - return roaring_bitmap_copy(x2); - } - if (0 == length2) { - return roaring_bitmap_copy(x1); - } - roaring_bitmap_t *answer = - roaring_bitmap_create_with_capacity(length1 + length2); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c; - if (bitsetconversion && (get_container_type(c1, container_type_1) != - BITSET_CONTAINER_TYPE_CODE) && - (get_container_type(c2, container_type_2) != - BITSET_CONTAINER_TYPE_CODE)) { - void *newc1 = - container_mutable_unwrap_shared(c1, &container_type_1); - newc1 = container_to_bitset(newc1, container_type_1); - container_type_1 = BITSET_CONTAINER_TYPE_CODE; - c = container_lazy_ior(newc1, container_type_1, c2, - container_type_2, - &container_result_type); - if (c != newc1) { // should not happen - container_free(newc1, container_type_1); - } - } else { - c = container_lazy_or(c1, container_type_1, c2, - container_type_2, &container_result_type); - } - // since we assume that the initial containers are non-empty, - // the - // result here - // can only be non-empty - ra_append(&answer->high_low_container, s1, c, - container_result_type); - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = - get_copy_of_container(c1, &container_type_1, x1->copy_on_write); - if (x1->copy_on_write) { - ra_set_container_at_index(&x1->high_low_container, pos1, c1, - container_type_1); - } - ra_append(&answer->high_low_container, s1, c1, container_type_1); - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_append(&answer->high_low_container, s2, c2, container_type_2); - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&answer->high_low_container, - &x2->high_low_container, pos2, length2, - x2->copy_on_write); - } else if (pos2 == length2) { - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1, - x1->copy_on_write); - } - return answer; -} - -void roaring_bitmap_lazy_or_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2, - const bool bitsetconversion) { - uint8_t container_result_type = 0; - int length1 = x1->high_low_container.size; - const int length2 = x2->high_low_container.size; - - if (0 == length2) return; - - if (0 == length1) { - roaring_bitmap_overwrite(x1, x2); - return; - } - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - if (!container_is_full(c1, container_type_1)) { - if ((bitsetconversion == false) || - (get_container_type(c1, container_type_1) == - BITSET_CONTAINER_TYPE_CODE)) { - c1 = get_writable_copy_if_shared(c1, &container_type_1); - } else { - // convert to bitset - void *oldc1 = c1; - uint8_t oldt1 = container_type_1; - c1 = container_mutable_unwrap_shared(c1, &container_type_1); - c1 = container_to_bitset(c1, container_type_1); - container_free(oldc1, oldt1); - container_type_1 = BITSET_CONTAINER_TYPE_CODE; - } - - void *c2 = ra_get_container_at_index(&x2->high_low_container, - pos2, &container_type_2); - void *c = container_lazy_ior(c1, container_type_1, c2, - container_type_2, - &container_result_type); - if (c != - c1) { // in this instance a new container was created, and - // we need to free the old one - container_free(c1, container_type_1); - } - - ra_set_container_at_index(&x1->high_low_container, pos1, c, - container_result_type); - } - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - // void *c2_clone = container_clone(c2, container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, - container_type_2); - pos1++; - length1++; - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, - pos2, length2, x2->copy_on_write); - } -} - -roaring_bitmap_t *roaring_bitmap_lazy_xor(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - if (0 == length1) { - return roaring_bitmap_copy(x2); - } - if (0 == length2) { - return roaring_bitmap_copy(x1); - } - roaring_bitmap_t *answer = - roaring_bitmap_create_with_capacity(length1 + length2); - answer->copy_on_write = x1->copy_on_write && x2->copy_on_write; - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = - container_lazy_xor(c1, container_type_1, c2, container_type_2, - &container_result_type); - - if (container_nonzero_cardinality(c, container_result_type)) { - ra_append(&answer->high_low_container, s1, c, - container_result_type); - } else { - container_free(c, container_result_type); - } - - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = - get_copy_of_container(c1, &container_type_1, x1->copy_on_write); - if (x1->copy_on_write) { - ra_set_container_at_index(&x1->high_low_container, pos1, c1, - container_type_1); - } - ra_append(&answer->high_low_container, s1, c1, container_type_1); - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_append(&answer->high_low_container, s2, c2, container_type_2); - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&answer->high_low_container, - &x2->high_low_container, pos2, length2, - x2->copy_on_write); - } else if (pos2 == length2) { - ra_append_copy_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1, - x1->copy_on_write); - } - return answer; -} - -void roaring_bitmap_lazy_xor_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - assert(x1 != x2); - uint8_t container_result_type = 0; - int length1 = x1->high_low_container.size; - const int length2 = x2->high_low_container.size; - - if (0 == length2) return; - - if (0 == length1) { - roaring_bitmap_overwrite(x1, x2); - return; - } - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - c1 = get_writable_copy_if_shared(c1, &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - void *c = - container_lazy_ixor(c1, container_type_1, c2, container_type_2, - &container_result_type); - if (container_nonzero_cardinality(c, container_result_type)) { - ra_set_container_at_index(&x1->high_low_container, pos1, c, - container_result_type); - ++pos1; - } else { - container_free(c, container_result_type); - ra_remove_at_index(&x1->high_low_container, pos1); - --length1; - } - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - // void *c2_clone = container_clone(c2, container_type_2); - c2 = - get_copy_of_container(c2, &container_type_2, x2->copy_on_write); - if (x2->copy_on_write) { - ra_set_container_at_index(&x2->high_low_container, pos2, c2, - container_type_2); - } - ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, - container_type_2); - pos1++; - length1++; - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, - pos2, length2, x2->copy_on_write); - } -} - -void roaring_bitmap_repair_after_lazy(roaring_bitmap_t *ra) { - for (int i = 0; i < ra->high_low_container.size; ++i) { - const uint8_t original_typecode = ra->high_low_container.typecodes[i]; - void *container = ra->high_low_container.containers[i]; - uint8_t new_typecode = original_typecode; - void *newcontainer = - container_repair_after_lazy(container, &new_typecode); - ra->high_low_container.containers[i] = newcontainer; - ra->high_low_container.typecodes[i] = new_typecode; - } -} - - - -/** -* roaring_bitmap_rank returns the number of integers that are smaller or equal -* to x. -*/ -uint64_t roaring_bitmap_rank(const roaring_bitmap_t *bm, uint32_t x) { - uint64_t size = 0; - uint32_t xhigh = x >> 16; - for (int i = 0; i < bm->high_low_container.size; i++) { - uint32_t key = bm->high_low_container.keys[i]; - if (xhigh > key) { - size += - container_get_cardinality(bm->high_low_container.containers[i], - bm->high_low_container.typecodes[i]); - } else if (xhigh == key) { - return size + container_rank(bm->high_low_container.containers[i], - bm->high_low_container.typecodes[i], - x & 0xFFFF); - } else { - return size; - } - } - return size; -} - -/** -* roaring_bitmap_smallest returns the smallest value in the set. -* Returns UINT32_MAX if the set is empty. -*/ -uint32_t roaring_bitmap_minimum(const roaring_bitmap_t *bm) { - if (bm->high_low_container.size > 0) { - void *container = bm->high_low_container.containers[0]; - uint8_t typecode = bm->high_low_container.typecodes[0]; - uint32_t key = bm->high_low_container.keys[0]; - uint32_t lowvalue = container_minimum(container, typecode); - return lowvalue | (key << 16); - } - return UINT32_MAX; -} - -/** -* roaring_bitmap_smallest returns the greatest value in the set. -* Returns 0 if the set is empty. -*/ -uint32_t roaring_bitmap_maximum(const roaring_bitmap_t *bm) { - if (bm->high_low_container.size > 0) { - void *container = - bm->high_low_container.containers[bm->high_low_container.size - 1]; - uint8_t typecode = - bm->high_low_container.typecodes[bm->high_low_container.size - 1]; - uint32_t key = - bm->high_low_container.keys[bm->high_low_container.size - 1]; - uint32_t lowvalue = container_maximum(container, typecode); - return lowvalue | (key << 16); - } - return 0; -} - -bool roaring_bitmap_select(const roaring_bitmap_t *bm, uint32_t rank, - uint32_t *element) { - void *container; - uint8_t typecode; - uint16_t key; - uint32_t start_rank = 0; - int i = 0; - bool valid = false; - while (!valid && i < bm->high_low_container.size) { - container = bm->high_low_container.containers[i]; - typecode = bm->high_low_container.typecodes[i]; - valid = - container_select(container, typecode, &start_rank, rank, element); - i++; - } - - if (valid) { - key = bm->high_low_container.keys[i - 1]; - *element |= (key << 16); - return true; - } else - return false; -} - -bool roaring_bitmap_intersect(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - uint64_t answer = 0; - int pos1 = 0, pos2 = 0; - - while (pos1 < length1 && pos2 < length2) { - const uint16_t s1 = ra_get_key_at_index(& x1->high_low_container, pos1); - const uint16_t s2 = ra_get_key_at_index(& x2->high_low_container, pos2); - - if (s1 == s2) { - uint8_t container_type_1, container_type_2; - void *c1 = ra_get_container_at_index(& x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(& x2->high_low_container, pos2, - &container_type_2); - if( container_intersect(c1, container_type_1, c2, container_type_2) ) return true; - ++pos1; - ++pos2; - } else if (s1 < s2) { // s1 < s2 - pos1 = ra_advance_until(& x1->high_low_container, s2, pos1); - } else { // s1 > s2 - pos2 = ra_advance_until(& x2->high_low_container, s1, pos2); - } - } - return answer; -} - - -uint64_t roaring_bitmap_and_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const int length1 = x1->high_low_container.size, - length2 = x2->high_low_container.size; - uint64_t answer = 0; - int pos1 = 0, pos2 = 0; - - while (pos1 < length1 && pos2 < length2) { - const uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - const uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - if (s1 == s2) { - uint8_t container_type_1, container_type_2; - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - answer += container_and_cardinality(c1, container_type_1, c2, - container_type_2); - ++pos1; - ++pos2; - } else if (s1 < s2) { // s1 < s2 - pos1 = ra_advance_until(&x1->high_low_container, s2, pos1); - } else { // s1 > s2 - pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); - } - } - return answer; -} - -double roaring_bitmap_jaccard_index(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const uint64_t c1 = roaring_bitmap_get_cardinality(x1); - const uint64_t c2 = roaring_bitmap_get_cardinality(x2); - const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); - return (double)inter / (double)(c1 + c2 - inter); -} - -uint64_t roaring_bitmap_or_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const uint64_t c1 = roaring_bitmap_get_cardinality(x1); - const uint64_t c2 = roaring_bitmap_get_cardinality(x2); - const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); - return c1 + c2 - inter; -} - -uint64_t roaring_bitmap_andnot_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const uint64_t c1 = roaring_bitmap_get_cardinality(x1); - const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); - return c1 - inter; -} - -uint64_t roaring_bitmap_xor_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2) { - const uint64_t c1 = roaring_bitmap_get_cardinality(x1); - const uint64_t c2 = roaring_bitmap_get_cardinality(x2); - const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); - return c1 + c2 - 2 * inter; -} - - -/** - * Check whether a range of values from range_start (included) to range_end (excluded) is present - */ -bool roaring_bitmap_contains_range(const roaring_bitmap_t *r, uint64_t range_start, uint64_t range_end) { - if(range_end >= UINT64_C(0x100000000)) { - range_end = UINT64_C(0x100000000); - } - if (range_start >= range_end) return true; // empty range are always contained! - if (range_end - range_start == 1) return roaring_bitmap_contains(r, (uint32_t)range_start); - uint16_t hb_rs = (uint16_t)(range_start >> 16); - uint16_t hb_re = (uint16_t)((range_end - 1) >> 16); - const int32_t span = hb_re - hb_rs; - const int32_t hlc_sz = ra_get_size(&r->high_low_container); - if (hlc_sz < span + 1) { - return false; - } - int32_t is = ra_get_index(&r->high_low_container, hb_rs); - int32_t ie = ra_get_index(&r->high_low_container, hb_re); - ie = (ie < 0 ? -ie - 1 : ie); - if ((is < 0) || ((ie - is) != span)) { - return false; - } - const uint32_t lb_rs = range_start & 0xFFFF; - const uint32_t lb_re = ((range_end - 1) & 0xFFFF) + 1; - uint8_t typecode; - void *container = ra_get_container_at_index(&r->high_low_container, is, &typecode); - if (hb_rs == hb_re) { - return container_contains_range(container, lb_rs, lb_re, typecode); - } - if (!container_contains_range(container, lb_rs, 1 << 16, typecode)) { - return false; - } - assert(ie < hlc_sz); // would indicate an algorithmic bug - container = ra_get_container_at_index(&r->high_low_container, ie, &typecode); - if (!container_contains_range(container, 0, lb_re, typecode)) { - return false; - } - for (int32_t i = is + 1; i < ie; ++i) { - container = ra_get_container_at_index(&r->high_low_container, i, &typecode); - if (!container_is_full(container, typecode) ) { - return false; - } - } - return true; -} - - -bool roaring_bitmap_is_strict_subset(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2) { - return (roaring_bitmap_get_cardinality(ra2) > - roaring_bitmap_get_cardinality(ra1) && - roaring_bitmap_is_subset(ra1, ra2)); -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/roaring.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/roaring_array.c */ -#include -#include -#include -#include -#include -#include - - -// Convention: [0,ra->size) all elements are initialized -// [ra->size, ra->allocation_size) is junk and contains nothing needing freeing - -extern inline int32_t ra_get_size(const roaring_array_t *ra); -extern inline int32_t ra_get_index(const roaring_array_t *ra, uint16_t x); -extern inline void *ra_get_container_at_index(const roaring_array_t *ra, - uint16_t i, uint8_t *typecode); -extern inline void ra_unshare_container_at_index(roaring_array_t *ra, - uint16_t i); -extern inline void ra_replace_key_and_container_at_index(roaring_array_t *ra, - int32_t i, - uint16_t key, void *c, - uint8_t typecode); -extern inline void ra_set_container_at_index(const roaring_array_t *ra, - int32_t i, void *c, - uint8_t typecode); - -#define INITIAL_CAPACITY 4 - -static bool realloc_array(roaring_array_t *ra, int32_t new_capacity) { - // because we combine the allocations, it is not possible to use realloc - /*ra->keys = - (uint16_t *)realloc(ra->keys, sizeof(uint16_t) * new_capacity); -ra->containers = - (void **)realloc(ra->containers, sizeof(void *) * new_capacity); -ra->typecodes = - (uint8_t *)realloc(ra->typecodes, sizeof(uint8_t) * new_capacity); -if (!ra->keys || !ra->containers || !ra->typecodes) { - free(ra->keys); - free(ra->containers); - free(ra->typecodes); - return false; -}*/ - - if ( new_capacity == 0 ) { - free(ra->containers); - ra->containers = NULL; - ra->keys = NULL; - ra->typecodes = NULL; - ra->allocation_size = 0; - return true; - } - const size_t memoryneeded = - new_capacity * (sizeof(uint16_t) + sizeof(void *) + sizeof(uint8_t)); - void *bigalloc = malloc(memoryneeded); - if (!bigalloc) return false; - void *oldbigalloc = ra->containers; - void **newcontainers = (void **)bigalloc; - uint16_t *newkeys = (uint16_t *)(newcontainers + new_capacity); - uint8_t *newtypecodes = (uint8_t *)(newkeys + new_capacity); - assert((char *)(newtypecodes + new_capacity) == - (char *)bigalloc + memoryneeded); - if(ra->size > 0) { - memcpy(newcontainers, ra->containers, sizeof(void *) * ra->size); - memcpy(newkeys, ra->keys, sizeof(uint16_t) * ra->size); - memcpy(newtypecodes, ra->typecodes, sizeof(uint8_t) * ra->size); - } - ra->containers = newcontainers; - ra->keys = newkeys; - ra->typecodes = newtypecodes; - ra->allocation_size = new_capacity; - free(oldbigalloc); - return true; -} - -bool ra_init_with_capacity(roaring_array_t *new_ra, uint32_t cap) { - if (!new_ra) return false; - new_ra->keys = NULL; - new_ra->containers = NULL; - new_ra->typecodes = NULL; - - new_ra->allocation_size = cap; - new_ra->size = 0; - if(cap > 0) { - void *bigalloc = - malloc(cap * (sizeof(uint16_t) + sizeof(void *) + sizeof(uint8_t))); - if( bigalloc == NULL ) return false; - new_ra->containers = (void **)bigalloc; - new_ra->keys = (uint16_t *)(new_ra->containers + cap); - new_ra->typecodes = (uint8_t *)(new_ra->keys + cap); - } - return true; -} - -int ra_shrink_to_fit(roaring_array_t *ra) { - int savings = (ra->allocation_size - ra->size) * - (sizeof(uint16_t) + sizeof(void *) + sizeof(uint8_t)); - if (!realloc_array(ra, ra->size)) { - return 0; - } - ra->allocation_size = ra->size; - return savings; -} - -bool ra_init(roaring_array_t *t) { - return ra_init_with_capacity(t, INITIAL_CAPACITY); -} - -bool ra_copy(const roaring_array_t *source, roaring_array_t *dest, - bool copy_on_write) { - if (!ra_init_with_capacity(dest, source->size)) return false; - dest->size = source->size; - dest->allocation_size = source->size; - if(dest->size > 0) { - memcpy(dest->keys, source->keys, dest->size * sizeof(uint16_t)); - } - // we go through the containers, turning them into shared containers... - if (copy_on_write) { - for (int32_t i = 0; i < dest->size; ++i) { - source->containers[i] = get_copy_of_container( - source->containers[i], &source->typecodes[i], copy_on_write); - } - // we do a shallow copy to the other bitmap - if(dest->size > 0) { - memcpy(dest->containers, source->containers, - dest->size * sizeof(void *)); - memcpy(dest->typecodes, source->typecodes, - dest->size * sizeof(uint8_t)); - } - } else { - if(dest->size > 0) { - memcpy(dest->typecodes, source->typecodes, - dest->size * sizeof(uint8_t)); - } - for (int32_t i = 0; i < dest->size; i++) { - dest->containers[i] = - container_clone(source->containers[i], source->typecodes[i]); - if (dest->containers[i] == NULL) { - for (int32_t j = 0; j < i; j++) { - container_free(dest->containers[j], dest->typecodes[j]); - } - ra_clear_without_containers(dest); - return false; - } - } - } - return true; -} - -bool ra_overwrite(const roaring_array_t *source, roaring_array_t *dest, - bool copy_on_write) { - ra_clear_containers(dest); // we are going to overwrite them - if (dest->allocation_size < source->size) { - if (!realloc_array(dest, source->size)) { - return false; - } - } - dest->size = source->size; - memcpy(dest->keys, source->keys, dest->size * sizeof(uint16_t)); - // we go through the containers, turning them into shared containers... - if (copy_on_write) { - for (int32_t i = 0; i < dest->size; ++i) { - source->containers[i] = get_copy_of_container( - source->containers[i], &source->typecodes[i], copy_on_write); - } - // we do a shallow copy to the other bitmap - memcpy(dest->containers, source->containers, - dest->size * sizeof(void *)); - memcpy(dest->typecodes, source->typecodes, - dest->size * sizeof(uint8_t)); - } else { - memcpy(dest->typecodes, source->typecodes, - dest->size * sizeof(uint8_t)); - for (int32_t i = 0; i < dest->size; i++) { - dest->containers[i] = - container_clone(source->containers[i], source->typecodes[i]); - if (dest->containers[i] == NULL) { - for (int32_t j = 0; j < i; j++) { - container_free(dest->containers[j], dest->typecodes[j]); - } - ra_clear_without_containers(dest); - return false; - } - } - } - return true; -} - -void ra_clear_containers(roaring_array_t *ra) { - for (int32_t i = 0; i < ra->size; ++i) { - container_free(ra->containers[i], ra->typecodes[i]); - } -} - -void ra_reset(roaring_array_t *ra) { - ra_clear_containers(ra); - ra->size = 0; - ra_shrink_to_fit(ra); -} - -void ra_clear_without_containers(roaring_array_t *ra) { - free(ra->containers); // keys and typecodes are allocated with containers - ra->size = 0; - ra->allocation_size = 0; - ra->containers = NULL; - ra->keys = NULL; - ra->typecodes = NULL; -} - -void ra_clear(roaring_array_t *ra) { - ra_clear_containers(ra); - ra_clear_without_containers(ra); -} - -bool extend_array(roaring_array_t *ra, int32_t k) { - int32_t desired_size = ra->size + k; - assert(desired_size <= MAX_CONTAINERS); - if (desired_size > ra->allocation_size) { - int32_t new_capacity = - (ra->size < 1024) ? 2 * desired_size : 5 * desired_size / 4; - if (new_capacity > MAX_CONTAINERS) { - new_capacity = MAX_CONTAINERS; - } - - return realloc_array(ra, new_capacity); - } - return true; -} - -void ra_append(roaring_array_t *ra, uint16_t key, void *container, - uint8_t typecode) { - extend_array(ra, 1); - const int32_t pos = ra->size; - - ra->keys[pos] = key; - ra->containers[pos] = container; - ra->typecodes[pos] = typecode; - ra->size++; -} - -void ra_append_copy(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t index, bool copy_on_write) { - extend_array(ra, 1); - const int32_t pos = ra->size; - - // old contents is junk not needing freeing - ra->keys[pos] = sa->keys[index]; - // the shared container will be in two bitmaps - if (copy_on_write) { - sa->containers[index] = get_copy_of_container( - sa->containers[index], &sa->typecodes[index], copy_on_write); - ra->containers[pos] = sa->containers[index]; - ra->typecodes[pos] = sa->typecodes[index]; - } else { - ra->containers[pos] = - container_clone(sa->containers[index], sa->typecodes[index]); - ra->typecodes[pos] = sa->typecodes[index]; - } - ra->size++; -} - -void ra_append_copies_until(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t stopping_key, bool copy_on_write) { - for (int32_t i = 0; i < sa->size; ++i) { - if (sa->keys[i] >= stopping_key) break; - ra_append_copy(ra, sa, i, copy_on_write); - } -} - -void ra_append_copy_range(roaring_array_t *ra, const roaring_array_t *sa, - int32_t start_index, int32_t end_index, - bool copy_on_write) { - extend_array(ra, end_index - start_index); - for (int32_t i = start_index; i < end_index; ++i) { - const int32_t pos = ra->size; - ra->keys[pos] = sa->keys[i]; - if (copy_on_write) { - sa->containers[i] = get_copy_of_container( - sa->containers[i], &sa->typecodes[i], copy_on_write); - ra->containers[pos] = sa->containers[i]; - ra->typecodes[pos] = sa->typecodes[i]; - } else { - ra->containers[pos] = - container_clone(sa->containers[i], sa->typecodes[i]); - ra->typecodes[pos] = sa->typecodes[i]; - } - ra->size++; - } -} - -void ra_append_copies_after(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t before_start, bool copy_on_write) { - int start_location = ra_get_index(sa, before_start); - if (start_location >= 0) - ++start_location; - else - start_location = -start_location - 1; - ra_append_copy_range(ra, sa, start_location, sa->size, copy_on_write); -} - -void ra_append_move_range(roaring_array_t *ra, roaring_array_t *sa, - int32_t start_index, int32_t end_index) { - extend_array(ra, end_index - start_index); - - for (int32_t i = start_index; i < end_index; ++i) { - const int32_t pos = ra->size; - - ra->keys[pos] = sa->keys[i]; - ra->containers[pos] = sa->containers[i]; - ra->typecodes[pos] = sa->typecodes[i]; - ra->size++; - } -} - -void ra_append_range(roaring_array_t *ra, roaring_array_t *sa, - int32_t start_index, int32_t end_index, - bool copy_on_write) { - extend_array(ra, end_index - start_index); - - for (int32_t i = start_index; i < end_index; ++i) { - const int32_t pos = ra->size; - ra->keys[pos] = sa->keys[i]; - if (copy_on_write) { - sa->containers[i] = get_copy_of_container( - sa->containers[i], &sa->typecodes[i], copy_on_write); - ra->containers[pos] = sa->containers[i]; - ra->typecodes[pos] = sa->typecodes[i]; - } else { - ra->containers[pos] = - container_clone(sa->containers[i], sa->typecodes[i]); - ra->typecodes[pos] = sa->typecodes[i]; - } - ra->size++; - } -} - -void *ra_get_container(roaring_array_t *ra, uint16_t x, uint8_t *typecode) { - int i = binarySearch(ra->keys, (int32_t)ra->size, x); - if (i < 0) return NULL; - *typecode = ra->typecodes[i]; - return ra->containers[i]; -} - -extern void *ra_get_container_at_index(const roaring_array_t *ra, uint16_t i, - uint8_t *typecode); - -void *ra_get_writable_container(roaring_array_t *ra, uint16_t x, - uint8_t *typecode) { - int i = binarySearch(ra->keys, (int32_t)ra->size, x); - if (i < 0) return NULL; - *typecode = ra->typecodes[i]; - return get_writable_copy_if_shared(ra->containers[i], typecode); -} - -void *ra_get_writable_container_at_index(roaring_array_t *ra, uint16_t i, - uint8_t *typecode) { - assert(i < ra->size); - *typecode = ra->typecodes[i]; - return get_writable_copy_if_shared(ra->containers[i], typecode); -} - -uint16_t ra_get_key_at_index(const roaring_array_t *ra, uint16_t i) { - return ra->keys[i]; -} - -extern int32_t ra_get_index(const roaring_array_t *ra, uint16_t x); - -extern int32_t ra_advance_until(const roaring_array_t *ra, uint16_t x, - int32_t pos); - -// everything skipped over is freed -int32_t ra_advance_until_freeing(roaring_array_t *ra, uint16_t x, int32_t pos) { - while (pos < ra->size && ra->keys[pos] < x) { - container_free(ra->containers[pos], ra->typecodes[pos]); - ++pos; - } - return pos; -} - -void ra_insert_new_key_value_at(roaring_array_t *ra, int32_t i, uint16_t key, - void *container, uint8_t typecode) { - extend_array(ra, 1); - // May be an optimization opportunity with DIY memmove - memmove(&(ra->keys[i + 1]), &(ra->keys[i]), - sizeof(uint16_t) * (ra->size - i)); - memmove(&(ra->containers[i + 1]), &(ra->containers[i]), - sizeof(void *) * (ra->size - i)); - memmove(&(ra->typecodes[i + 1]), &(ra->typecodes[i]), - sizeof(uint8_t) * (ra->size - i)); - ra->keys[i] = key; - ra->containers[i] = container; - ra->typecodes[i] = typecode; - ra->size++; -} - -// note: Java routine set things to 0, enabling GC. -// Java called it "resize" but it was always used to downsize. -// Allowing upsize would break the conventions about -// valid containers below ra->size. - -void ra_downsize(roaring_array_t *ra, int32_t new_length) { - assert(new_length <= ra->size); - ra->size = new_length; -} - -void ra_remove_at_index(roaring_array_t *ra, int32_t i) { - memmove(&(ra->containers[i]), &(ra->containers[i + 1]), - sizeof(void *) * (ra->size - i - 1)); - memmove(&(ra->keys[i]), &(ra->keys[i + 1]), - sizeof(uint16_t) * (ra->size - i - 1)); - memmove(&(ra->typecodes[i]), &(ra->typecodes[i + 1]), - sizeof(uint8_t) * (ra->size - i - 1)); - ra->size--; -} - -void ra_remove_at_index_and_free(roaring_array_t *ra, int32_t i) { - container_free(ra->containers[i], ra->typecodes[i]); - ra_remove_at_index(ra, i); -} - -// used in inplace andNot only, to slide left the containers from -// the mutated RoaringBitmap that are after the largest container of -// the argument RoaringBitmap. In use it should be followed by a call to -// downsize. -// -void ra_copy_range(roaring_array_t *ra, uint32_t begin, uint32_t end, - uint32_t new_begin) { - assert(begin <= end); - assert(new_begin < begin); - - const int range = end - begin; - - // We ensure to previously have freed overwritten containers - // that are not copied elsewhere - - memmove(&(ra->containers[new_begin]), &(ra->containers[begin]), - sizeof(void *) * range); - memmove(&(ra->keys[new_begin]), &(ra->keys[begin]), - sizeof(uint16_t) * range); - memmove(&(ra->typecodes[new_begin]), &(ra->typecodes[begin]), - sizeof(uint8_t) * range); -} - -void ra_shift_tail(roaring_array_t *ra, int32_t count, int32_t distance) { - if (distance > 0) { - extend_array(ra, distance); - } - int32_t srcpos = ra->size - count; - int32_t dstpos = srcpos + distance; - memmove(&(ra->keys[dstpos]), &(ra->keys[srcpos]), - sizeof(uint16_t) * count); - memmove(&(ra->containers[dstpos]), &(ra->containers[srcpos]), - sizeof(void *) * count); - memmove(&(ra->typecodes[dstpos]), &(ra->typecodes[srcpos]), - sizeof(uint8_t) * count); - ra->size += distance; -} - - -size_t ra_size_in_bytes(roaring_array_t *ra) { - size_t cardinality = 0; - size_t tot_len = - 1 /* initial byte type */ + 4 /* tot_len */ + sizeof(roaring_array_t) + - ra->size * (sizeof(uint16_t) + sizeof(void *) + sizeof(uint8_t)); - for (int32_t i = 0; i < ra->size; i++) { - tot_len += - (container_serialization_len(ra->containers[i], ra->typecodes[i]) + - sizeof(uint16_t)); - cardinality += - container_get_cardinality(ra->containers[i], ra->typecodes[i]); - } - - if ((cardinality * sizeof(uint32_t) + sizeof(uint32_t)) < tot_len) { - return cardinality * sizeof(uint32_t) + 1 + sizeof(uint32_t); - } - return tot_len; -} - -void ra_to_uint32_array(const roaring_array_t *ra, uint32_t *ans) { - size_t ctr = 0; - for (int32_t i = 0; i < ra->size; ++i) { - int num_added = container_to_uint32_array( - ans + ctr, ra->containers[i], ra->typecodes[i], - ((uint32_t)ra->keys[i]) << 16); - ctr += num_added; - } -} - -bool ra_range_uint32_array(const roaring_array_t *ra, size_t offset, size_t limit, uint32_t *ans) { - size_t ctr = 0; - size_t dtr = 0; - - size_t t_limit = 0; - - bool first = false; - size_t first_skip = 0; - - uint32_t *t_ans = NULL; - size_t cur_len = 0; - - for (int i = 0; i < ra->size; ++i) { - - const void *container = container_unwrap_shared(ra->containers[i], &ra->typecodes[i]); - switch (ra->typecodes[i]) { - case BITSET_CONTAINER_TYPE_CODE: - t_limit = ((const bitset_container_t *)container)->cardinality; - break; - case ARRAY_CONTAINER_TYPE_CODE: - t_limit = ((const array_container_t *)container)->cardinality; - break; - case RUN_CONTAINER_TYPE_CODE: - t_limit = run_container_cardinality((const run_container_t *)container); - break; - } - if (ctr + t_limit - 1 >= offset && ctr < offset + limit){ - if (!first){ - //first_skip = t_limit - (ctr + t_limit - offset); - first_skip = offset - ctr; - first = true; - t_ans = (uint32_t *)malloc(sizeof(*t_ans) * (first_skip + limit)); - if(t_ans == NULL) { - return false; - } - memset(t_ans, 0, sizeof(*t_ans) * (first_skip + limit)) ; - cur_len = first_skip + limit; - } - if (dtr + t_limit > cur_len){ - uint32_t * append_ans = (uint32_t *)malloc(sizeof(*append_ans) * (cur_len + t_limit)); - if(append_ans == NULL) { - if(t_ans != NULL) free(t_ans); - return false; - } - memset(append_ans, 0, sizeof(*append_ans) * (cur_len + t_limit)); - cur_len = cur_len + t_limit; - memcpy(append_ans, t_ans, dtr * sizeof(uint32_t)); - free(t_ans); - t_ans = append_ans; - } - switch (ra->typecodes[i]) { - case BITSET_CONTAINER_TYPE_CODE: - container_to_uint32_array( - t_ans + dtr, (const bitset_container_t *)container, ra->typecodes[i], - ((uint32_t)ra->keys[i]) << 16); - break; - case ARRAY_CONTAINER_TYPE_CODE: - container_to_uint32_array( - t_ans + dtr, (const array_container_t *)container, ra->typecodes[i], - ((uint32_t)ra->keys[i]) << 16); - break; - case RUN_CONTAINER_TYPE_CODE: - container_to_uint32_array( - t_ans + dtr, (const run_container_t *)container, ra->typecodes[i], - ((uint32_t)ra->keys[i]) << 16); - break; - } - dtr += t_limit; - } - ctr += t_limit; - if (dtr-first_skip >= limit) break; - } - if(t_ans != NULL) { - memcpy(ans, t_ans+first_skip, limit * sizeof(uint32_t)); - free(t_ans); - } - return true; -} - -bool ra_has_run_container(const roaring_array_t *ra) { - for (int32_t k = 0; k < ra->size; ++k) { - if (get_container_type(ra->containers[k], ra->typecodes[k]) == - RUN_CONTAINER_TYPE_CODE) - return true; - } - return false; -} - -uint32_t ra_portable_header_size(const roaring_array_t *ra) { - if (ra_has_run_container(ra)) { - if (ra->size < - NO_OFFSET_THRESHOLD) { // for small bitmaps, we omit the offsets - return 4 + (ra->size + 7) / 8 + 4 * ra->size; - } - return 4 + (ra->size + 7) / 8 + - 8 * ra->size; // - 4 because we pack the size with the cookie - } else { - return 4 + 4 + 8 * ra->size; - } -} - -size_t ra_portable_size_in_bytes(const roaring_array_t *ra) { - size_t count = ra_portable_header_size(ra); - - for (int32_t k = 0; k < ra->size; ++k) { - count += container_size_in_bytes(ra->containers[k], ra->typecodes[k]); - } - return count; -} - -size_t ra_portable_serialize(const roaring_array_t *ra, char *buf) { - char *initbuf = buf; - uint32_t startOffset = 0; - bool hasrun = ra_has_run_container(ra); - if (hasrun) { - uint32_t cookie = SERIAL_COOKIE | ((ra->size - 1) << 16); - memcpy(buf, &cookie, sizeof(cookie)); - buf += sizeof(cookie); - uint32_t s = (ra->size + 7) / 8; - uint8_t *bitmapOfRunContainers = (uint8_t *)calloc(s, 1); - assert(bitmapOfRunContainers != NULL); // todo: handle - for (int32_t i = 0; i < ra->size; ++i) { - if (get_container_type(ra->containers[i], ra->typecodes[i]) == - RUN_CONTAINER_TYPE_CODE) { - bitmapOfRunContainers[i / 8] |= (1 << (i % 8)); - } - } - memcpy(buf, bitmapOfRunContainers, s); - buf += s; - free(bitmapOfRunContainers); - if (ra->size < NO_OFFSET_THRESHOLD) { - startOffset = 4 + 4 * ra->size + s; - } else { - startOffset = 4 + 8 * ra->size + s; - } - } else { // backwards compatibility - uint32_t cookie = SERIAL_COOKIE_NO_RUNCONTAINER; - - memcpy(buf, &cookie, sizeof(cookie)); - buf += sizeof(cookie); - memcpy(buf, &ra->size, sizeof(ra->size)); - buf += sizeof(ra->size); - - startOffset = 4 + 4 + 4 * ra->size + 4 * ra->size; - } - for (int32_t k = 0; k < ra->size; ++k) { - memcpy(buf, &ra->keys[k], sizeof(ra->keys[k])); - buf += sizeof(ra->keys[k]); - // get_cardinality returns a value in [1,1<<16], subtracting one - // we get [0,1<<16 - 1] which fits in 16 bits - uint16_t card = (uint16_t)( - container_get_cardinality(ra->containers[k], ra->typecodes[k]) - 1); - memcpy(buf, &card, sizeof(card)); - buf += sizeof(card); - } - if ((!hasrun) || (ra->size >= NO_OFFSET_THRESHOLD)) { - // writing the containers offsets - for (int32_t k = 0; k < ra->size; k++) { - memcpy(buf, &startOffset, sizeof(startOffset)); - buf += sizeof(startOffset); - startOffset = - startOffset + - container_size_in_bytes(ra->containers[k], ra->typecodes[k]); - } - } - for (int32_t k = 0; k < ra->size; ++k) { - buf += container_write(ra->containers[k], ra->typecodes[k], buf); - } - return buf - initbuf; -} - -// Quickly checks whether there is a serialized bitmap at the pointer, -// not exceeding size "maxbytes" in bytes. This function does not allocate -// memory dynamically. -// -// This function returns 0 if and only if no valid bitmap is found. -// Otherwise, it returns how many bytes are occupied. -// -size_t ra_portable_deserialize_size(const char *buf, const size_t maxbytes) { - size_t bytestotal = sizeof(int32_t);// for cookie - if(bytestotal > maxbytes) return 0; - uint32_t cookie; - memcpy(&cookie, buf, sizeof(int32_t)); - buf += sizeof(uint32_t); - if ((cookie & 0xFFFF) != SERIAL_COOKIE && - cookie != SERIAL_COOKIE_NO_RUNCONTAINER) { - return 0; - } - int32_t size; - - if ((cookie & 0xFFFF) == SERIAL_COOKIE) - size = (cookie >> 16) + 1; - else { - bytestotal += sizeof(int32_t); - if(bytestotal > maxbytes) return 0; - memcpy(&size, buf, sizeof(int32_t)); - buf += sizeof(uint32_t); - } - if (size > (1<<16)) { - return 0; // logically impossible - } - char *bitmapOfRunContainers = NULL; - bool hasrun = (cookie & 0xFFFF) == SERIAL_COOKIE; - if (hasrun) { - int32_t s = (size + 7) / 8; - bytestotal += s; - if(bytestotal > maxbytes) return 0; - bitmapOfRunContainers = (char *)buf; - buf += s; - } - bytestotal += size * 2 * sizeof(uint16_t); - if(bytestotal > maxbytes) return 0; - uint16_t *keyscards = (uint16_t *)buf; - buf += size * 2 * sizeof(uint16_t); - if ((!hasrun) || (size >= NO_OFFSET_THRESHOLD)) { - // skipping the offsets - bytestotal += size * 4; - if(bytestotal > maxbytes) return 0; - buf += size * 4; - } - // Reading the containers - for (int32_t k = 0; k < size; ++k) { - uint16_t tmp; - memcpy(&tmp, keyscards + 2*k+1, sizeof(tmp)); - uint32_t thiscard = tmp + 1; - bool isbitmap = (thiscard > DEFAULT_MAX_SIZE); - bool isrun = false; - if(hasrun) { - if((bitmapOfRunContainers[k / 8] & (1 << (k % 8))) != 0) { - isbitmap = false; - isrun = true; - } - } - if (isbitmap) { - size_t containersize = BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); - bytestotal += containersize; - if(bytestotal > maxbytes) return 0; - buf += containersize; - } else if (isrun) { - bytestotal += sizeof(uint16_t); - if(bytestotal > maxbytes) return 0; - uint16_t n_runs; - memcpy(&n_runs, buf, sizeof(uint16_t)); - buf += sizeof(uint16_t); - size_t containersize = n_runs * sizeof(rle16_t); - bytestotal += containersize; - if(bytestotal > maxbytes) return 0; - buf += containersize; - } else { - size_t containersize = thiscard * sizeof(uint16_t); - bytestotal += containersize; - if(bytestotal > maxbytes) return 0; - buf += containersize; - } - } - return bytestotal; -} - - -// this function populates answer from the content of buf (reading up to maxbytes bytes). -// The function returns false if a properly serialized bitmap cannot be found. -// if it returns true, readbytes is populated by how many bytes were read, we have that *readbytes <= maxbytes. -bool ra_portable_deserialize(roaring_array_t *answer, const char *buf, const size_t maxbytes, size_t * readbytes) { - *readbytes = sizeof(int32_t);// for cookie - if(*readbytes > maxbytes) { - fprintf(stderr, "Ran out of bytes while reading first 4 bytes.\n"); - return false; - } - uint32_t cookie; - memcpy(&cookie, buf, sizeof(int32_t)); - buf += sizeof(uint32_t); - if ((cookie & 0xFFFF) != SERIAL_COOKIE && - cookie != SERIAL_COOKIE_NO_RUNCONTAINER) { - fprintf(stderr, "I failed to find one of the right cookies. Found %" PRIu32 "\n", - cookie); - return false; - } - int32_t size; - - if ((cookie & 0xFFFF) == SERIAL_COOKIE) - size = (cookie >> 16) + 1; - else { - *readbytes += sizeof(int32_t); - if(*readbytes > maxbytes) { - fprintf(stderr, "Ran out of bytes while reading second part of the cookie.\n"); - return false; - } - memcpy(&size, buf, sizeof(int32_t)); - buf += sizeof(uint32_t); - } - if (size > (1<<16)) { - fprintf(stderr, "You cannot have so many containers, the data must be corrupted: %" PRId32 "\n", - size); - return false; // logically impossible - } - const char *bitmapOfRunContainers = NULL; - bool hasrun = (cookie & 0xFFFF) == SERIAL_COOKIE; - if (hasrun) { - int32_t s = (size + 7) / 8; - *readbytes += s; - if(*readbytes > maxbytes) {// data is corrupted? - fprintf(stderr, "Ran out of bytes while reading run bitmap.\n"); - return false; - } - bitmapOfRunContainers = buf; - buf += s; - } - uint16_t *keyscards = (uint16_t *)buf; - - *readbytes += size * 2 * sizeof(uint16_t); - if(*readbytes > maxbytes) { - fprintf(stderr, "Ran out of bytes while reading key-cardinality array.\n"); - return false; - } - buf += size * 2 * sizeof(uint16_t); - - bool is_ok = ra_init_with_capacity(answer, size); - if (!is_ok) { - fprintf(stderr, "Failed to allocate memory for roaring array. Bailing out.\n"); - return false; - } - - for (int32_t k = 0; k < size; ++k) { - uint16_t tmp; - memcpy(&tmp, keyscards + 2*k, sizeof(tmp)); - answer->keys[k] = tmp; - } - if ((!hasrun) || (size >= NO_OFFSET_THRESHOLD)) { - *readbytes += size * 4; - if(*readbytes > maxbytes) {// data is corrupted? - fprintf(stderr, "Ran out of bytes while reading offsets.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - - // skipping the offsets - buf += size * 4; - } - // Reading the containers - for (int32_t k = 0; k < size; ++k) { - uint16_t tmp; - memcpy(&tmp, keyscards + 2*k+1, sizeof(tmp)); - uint32_t thiscard = tmp + 1; - bool isbitmap = (thiscard > DEFAULT_MAX_SIZE); - bool isrun = false; - if(hasrun) { - if((bitmapOfRunContainers[k / 8] & (1 << (k % 8))) != 0) { - isbitmap = false; - isrun = true; - } - } - if (isbitmap) { - // we check that the read is allowed - size_t containersize = BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); - *readbytes += containersize; - if(*readbytes > maxbytes) { - fprintf(stderr, "Running out of bytes while reading a bitset container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - // it is now safe to read - bitset_container_t *c = bitset_container_create(); - if(c == NULL) {// memory allocation failure - fprintf(stderr, "Failed to allocate memory for a bitset container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - answer->size++; - buf += bitset_container_read(thiscard, c, buf); - answer->containers[k] = c; - answer->typecodes[k] = BITSET_CONTAINER_TYPE_CODE; - } else if (isrun) { - // we check that the read is allowed - *readbytes += sizeof(uint16_t); - if(*readbytes > maxbytes) { - fprintf(stderr, "Running out of bytes while reading a run container (header).\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - uint16_t n_runs; - memcpy(&n_runs, buf, sizeof(uint16_t)); - size_t containersize = n_runs * sizeof(rle16_t); - *readbytes += containersize; - if(*readbytes > maxbytes) {// data is corrupted? - fprintf(stderr, "Running out of bytes while reading a run container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - // it is now safe to read - - run_container_t *c = run_container_create(); - if(c == NULL) {// memory allocation failure - fprintf(stderr, "Failed to allocate memory for a run container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - answer->size++; - buf += run_container_read(thiscard, c, buf); - answer->containers[k] = c; - answer->typecodes[k] = RUN_CONTAINER_TYPE_CODE; - } else { - // we check that the read is allowed - size_t containersize = thiscard * sizeof(uint16_t); - *readbytes += containersize; - if(*readbytes > maxbytes) {// data is corrupted? - fprintf(stderr, "Running out of bytes while reading an array container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - // it is now safe to read - array_container_t *c = - array_container_create_given_capacity(thiscard); - if(c == NULL) {// memory allocation failure - fprintf(stderr, "Failed to allocate memory for an array container.\n"); - ra_clear(answer);// we need to clear the containers already allocated, and the roaring array - return false; - } - answer->size++; - buf += array_container_read(thiscard, c, buf); - answer->containers[k] = c; - answer->typecodes[k] = ARRAY_CONTAINER_TYPE_CODE; - } - } - return true; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/roaring_array.c */ -/* begin file /opt/bitmap/CRoaring-0.2.57/src/roaring_priority_queue.c */ - -struct roaring_pq_element_s { - uint64_t size; - bool is_temporary; - roaring_bitmap_t *bitmap; -}; - -typedef struct roaring_pq_element_s roaring_pq_element_t; - -struct roaring_pq_s { - roaring_pq_element_t *elements; - uint64_t size; -}; - -typedef struct roaring_pq_s roaring_pq_t; - -static inline bool compare(roaring_pq_element_t *t1, roaring_pq_element_t *t2) { - return t1->size < t2->size; -} - -static void pq_add(roaring_pq_t *pq, roaring_pq_element_t *t) { - uint64_t i = pq->size; - pq->elements[pq->size++] = *t; - while (i > 0) { - uint64_t p = (i - 1) >> 1; - roaring_pq_element_t ap = pq->elements[p]; - if (!compare(t, &ap)) break; - pq->elements[i] = ap; - i = p; - } - pq->elements[i] = *t; -} - -static void pq_free(roaring_pq_t *pq) { - free(pq->elements); - pq->elements = NULL; // paranoid - free(pq); -} - -static void percolate_down(roaring_pq_t *pq, uint32_t i) { - uint32_t size = (uint32_t)pq->size; - uint32_t hsize = size >> 1; - roaring_pq_element_t ai = pq->elements[i]; - while (i < hsize) { - uint32_t l = (i << 1) + 1; - uint32_t r = l + 1; - roaring_pq_element_t bestc = pq->elements[l]; - if (r < size) { - if (compare(pq->elements + r, &bestc)) { - l = r; - bestc = pq->elements[r]; - } - } - if (!compare(&bestc, &ai)) { - break; - } - pq->elements[i] = bestc; - i = l; - } - pq->elements[i] = ai; -} - -static roaring_pq_t *create_pq(const roaring_bitmap_t **arr, uint32_t length) { - roaring_pq_t *answer = (roaring_pq_t *)malloc(sizeof(roaring_pq_t)); - answer->elements = - (roaring_pq_element_t *)malloc(sizeof(roaring_pq_element_t) * length); - answer->size = length; - for (uint32_t i = 0; i < length; i++) { - answer->elements[i].bitmap = (roaring_bitmap_t *)arr[i]; - answer->elements[i].is_temporary = false; - answer->elements[i].size = - roaring_bitmap_portable_size_in_bytes(arr[i]); - } - for (int32_t i = (length >> 1); i >= 0; i--) { - percolate_down(answer, i); - } - return answer; -} - -static roaring_pq_element_t pq_poll(roaring_pq_t *pq) { - roaring_pq_element_t ans = *pq->elements; - if (pq->size > 1) { - pq->elements[0] = pq->elements[--pq->size]; - percolate_down(pq, 0); - } else - --pq->size; - // memmove(pq->elements,pq->elements+1,(pq->size-1)*sizeof(roaring_pq_element_t));--pq->size; - return ans; -} - -// this function consumes and frees the inputs -static roaring_bitmap_t *lazy_or_from_lazy_inputs(roaring_bitmap_t *x1, - roaring_bitmap_t *x2) { - uint8_t container_result_type = 0; - const int length1 = ra_get_size(&x1->high_low_container), - length2 = ra_get_size(&x2->high_low_container); - if (0 == length1) { - roaring_bitmap_free(x1); - return x2; - } - if (0 == length2) { - roaring_bitmap_free(x2); - return x1; - } - uint32_t neededcap = length1 > length2 ? length2 : length1; - roaring_bitmap_t *answer = roaring_bitmap_create_with_capacity(neededcap); - int pos1 = 0, pos2 = 0; - uint8_t container_type_1, container_type_2; - uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - while (true) { - if (s1 == s2) { - // todo: unsharing can be inefficient as it may create a clone where - // none - // is needed, but it has the benefit of being easy to reason about. - ra_unshare_container_at_index(&x1->high_low_container, pos1); - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - assert(container_type_1 != SHARED_CONTAINER_TYPE_CODE); - ra_unshare_container_at_index(&x2->high_low_container, pos2); - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - assert(container_type_2 != SHARED_CONTAINER_TYPE_CODE); - void *c; - - if ((container_type_2 == BITSET_CONTAINER_TYPE_CODE) && - (container_type_1 != BITSET_CONTAINER_TYPE_CODE)) { - c = container_lazy_ior(c2, container_type_2, c1, - container_type_1, - &container_result_type); - container_free(c1, container_type_1); - if (c != c2) { - container_free(c2, container_type_2); - } - } else { - c = container_lazy_ior(c1, container_type_1, c2, - container_type_2, - &container_result_type); - container_free(c2, container_type_2); - if (c != c1) { - container_free(c1, container_type_1); - } - } - // since we assume that the initial containers are non-empty, the - // result here - // can only be non-empty - ra_append(&answer->high_low_container, s1, c, - container_result_type); - ++pos1; - ++pos2; - if (pos1 == length1) break; - if (pos2 == length2) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - - } else if (s1 < s2) { // s1 < s2 - void *c1 = ra_get_container_at_index(&x1->high_low_container, pos1, - &container_type_1); - ra_append(&answer->high_low_container, s1, c1, container_type_1); - pos1++; - if (pos1 == length1) break; - s1 = ra_get_key_at_index(&x1->high_low_container, pos1); - - } else { // s1 > s2 - void *c2 = ra_get_container_at_index(&x2->high_low_container, pos2, - &container_type_2); - ra_append(&answer->high_low_container, s2, c2, container_type_2); - pos2++; - if (pos2 == length2) break; - s2 = ra_get_key_at_index(&x2->high_low_container, pos2); - } - } - if (pos1 == length1) { - ra_append_move_range(&answer->high_low_container, - &x2->high_low_container, pos2, length2); - } else if (pos2 == length2) { - ra_append_move_range(&answer->high_low_container, - &x1->high_low_container, pos1, length1); - } - ra_clear_without_containers(&x1->high_low_container); - ra_clear_without_containers(&x2->high_low_container); - free(x1); - free(x2); - return answer; -} - -/** - * Compute the union of 'number' bitmaps using a heap. This can - * sometimes be faster than roaring_bitmap_or_many which uses - * a naive algorithm. Caller is responsible for freeing the - * result. - */ -roaring_bitmap_t *roaring_bitmap_or_many_heap(uint32_t number, - const roaring_bitmap_t **x) { - if (number == 0) { - return roaring_bitmap_create(); - } - if (number == 1) { - return roaring_bitmap_copy(x[0]); - } - roaring_pq_t *pq = create_pq(x, number); - while (pq->size > 1) { - roaring_pq_element_t x1 = pq_poll(pq); - roaring_pq_element_t x2 = pq_poll(pq); - - if (x1.is_temporary && x2.is_temporary) { - roaring_bitmap_t *newb = - lazy_or_from_lazy_inputs(x1.bitmap, x2.bitmap); - // should normally return a fresh new bitmap *except* that - // it can return x1.bitmap or x2.bitmap in degenerate cases - bool temporary = !((newb == x1.bitmap) && (newb == x2.bitmap)); - uint64_t bsize = roaring_bitmap_portable_size_in_bytes(newb); - roaring_pq_element_t newelement = { - .size = bsize, .is_temporary = temporary, .bitmap = newb}; - pq_add(pq, &newelement); - } else if (x2.is_temporary) { - roaring_bitmap_lazy_or_inplace(x2.bitmap, x1.bitmap, false); - x2.size = roaring_bitmap_portable_size_in_bytes(x2.bitmap); - pq_add(pq, &x2); - } else if (x1.is_temporary) { - roaring_bitmap_lazy_or_inplace(x1.bitmap, x2.bitmap, false); - x1.size = roaring_bitmap_portable_size_in_bytes(x1.bitmap); - - pq_add(pq, &x1); - } else { - roaring_bitmap_t *newb = - roaring_bitmap_lazy_or(x1.bitmap, x2.bitmap, false); - uint64_t bsize = roaring_bitmap_portable_size_in_bytes(newb); - roaring_pq_element_t newelement = { - .size = bsize, .is_temporary = true, .bitmap = newb}; - - pq_add(pq, &newelement); - } - } - roaring_pq_element_t X = pq_poll(pq); - roaring_bitmap_t *answer = X.bitmap; - roaring_bitmap_repair_after_lazy(answer); - pq_free(pq); - return answer; -} -/* end file /opt/bitmap/CRoaring-0.2.57/src/roaring_priority_queue.c */ diff --git a/contrib/croaring/roaring/roaring.h b/contrib/croaring/roaring/roaring.h deleted file mode 100644 index 53413b2a06d..00000000000 --- a/contrib/croaring/roaring/roaring.h +++ /dev/null @@ -1,7187 +0,0 @@ -/* auto-generated on Tue Dec 18 09:42:59 CST 2018. Do not edit! */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_version.h */ -// /include/roaring/roaring_version.h automatically generated by release.py, do not change by hand -#ifndef ROARING_INCLUDE_ROARING_VERSION -#define ROARING_INCLUDE_ROARING_VERSION -#define ROARING_VERSION = 0.2.57, -enum { - ROARING_VERSION_MAJOR = 0, - ROARING_VERSION_MINOR = 2, - ROARING_VERSION_REVISION = 57 -}; -#endif // ROARING_INCLUDE_ROARING_VERSION -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_version.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/portability.h */ -/* - * portability.h - * - */ - - -#if defined(__clang__) -#pragma clang diagnostic ignored "-Wold-style-cast" -#pragma clang diagnostic ignored "-Wzero-as-null-pointer-constant" -#pragma clang diagnostic ignored "-Wold-style-cast" -#pragma clang diagnostic ignored "-Wcast-align" -#pragma clang diagnostic ignored "-Wcast-qual" -#pragma clang diagnostic ignored "-Wundef" -#endif - -#ifndef INCLUDE_PORTABILITY_H_ -#define INCLUDE_PORTABILITY_H_ - -#ifdef __cplusplus -extern "C" { -#endif - - -#ifndef _GNU_SOURCE -#define _GNU_SOURCE -#endif -//#ifndef __STDC_FORMAT_MACROS -//#define __STDC_FORMAT_MACROS 1 -//#endif - -#if !(defined(_POSIX_C_SOURCE)) || (_POSIX_C_SOURCE < 200809L) -#define _POSIX_C_SOURCE 200809L -#endif -#if !(defined(_XOPEN_SOURCE)) || (_XOPEN_SOURCE < 700) -#define _XOPEN_SOURCE 700 -#endif - -#include -#include -#include // will provide posix_memalign with _POSIX_C_SOURCE as defined above -#if !(defined(__APPLE__)) && !(defined(__FreeBSD__)) -#include // this should never be needed but there are some reports that it is needed. -#endif - - -#if defined(_MSC_VER) && !defined(__clang__) && !defined(_WIN64) -#pragma message( \ - "You appear to be attempting a 32-bit build under Visual Studio. We recommend a 64-bit build instead.") -#endif - -#if defined(__SIZEOF_LONG_LONG__) && __SIZEOF_LONG_LONG__ != 8 -#error This code assumes 64-bit long longs (by use of the GCC intrinsics). Your system is not currently supported. -#endif - -#if defined(_MSC_VER) -#define __restrict__ __restrict -#endif - -#ifndef DISABLE_X64 // some users may want to compile as if they did not have - // an x64 processor - -/////////////////////// -/// We support X64 hardware in the following manner: -/// -/// if IS_X64 is defined then we have at least SSE and SSE2 -/// (All Intel processors sold in the recent past have at least SSE and SSE2 support, -/// going back to the Pentium 4.) -/// -/// if USESSE4 is defined then we assume at least SSE4.2, SSE4.1, -/// SSSE3, SSE3... + IS_X64 -/// if USEAVX is defined, then we assume AVX2, AVX + USESSE4 -/// -/// So if you have hardware that supports AVX but not AVX2, then "USEAVX" -/// won't be enabled. -/// If you have hardware that supports SSE4.1, but not SSE4.2, then USESSE4 -/// won't be defined. -////////////////////// - -// unless DISABLEAVX was defined, if we have __AVX2__, we enable AVX -#if (!defined(USEAVX)) && (!defined(DISABLEAVX)) && (defined(__AVX2__)) -#define USEAVX -#endif - -// if we have __SSE4_2__, we enable SSE4 -#if (defined(__POPCNT__)) && (defined(__SSE4_2__)) -#define USESSE4 -#endif - -#if defined(USEAVX) || defined(__x86_64__) || defined(_M_X64) -// we have an x64 processor -#define IS_X64 -// we include the intrinsic header -#ifndef _MSC_VER -/* Non-Microsoft C/C++-compatible compiler */ -#include // on some recent GCC, this will declare posix_memalign -#endif -#endif - -#ifndef _MSC_VER -/* Non-Microsoft C/C++-compatible compiler, assumes that it supports inline - * assembly */ -#define ROARING_INLINE_ASM -#endif - -#ifdef USEAVX -#define USESSE4 // if we have AVX, then we have SSE4 -#define USE_BMI // we assume that AVX2 and BMI go hand and hand -#define USEAVX2FORDECODING // optimization -// vector operations should work on not just AVX -#define ROARING_VECTOR_OPERATIONS_ENABLED // vector unions (optimization) -#endif - -#endif // DISABLE_X64 - -#ifdef _MSC_VER -/* Microsoft C/C++-compatible compiler */ -#include - -#ifndef __clang__ // if one compiles with MSVC *with* clang, then these - // intrinsics are defined!!! -// sadly there is no way to check whether we are missing these intrinsics -// specifically. - -/* wrappers for Visual Studio built-ins that look like gcc built-ins */ -/* result might be undefined when input_num is zero */ -static inline int __builtin_ctzll(unsigned long long input_num) { - unsigned long index; -#ifdef _WIN64 // highly recommended!!! - _BitScanForward64(&index, input_num); -#else // if we must support 32-bit Windows - if ((uint32_t)input_num != 0) { - _BitScanForward(&index, (uint32_t)input_num); - } else { - _BitScanForward(&index, (uint32_t)(input_num >> 32)); - index += 32; - } -#endif - return index; -} - -/* result might be undefined when input_num is zero */ -static inline int __builtin_clzll(unsigned long long input_num) { - unsigned long index; -#ifdef _WIN64 // highly recommended!!! - _BitScanReverse64(&index, input_num); -#else // if we must support 32-bit Windows - if (input_num > 0xFFFFFFFF) { - _BitScanReverse(&index, (uint32_t)(input_num >> 32)); - index += 32; - } else { - _BitScanReverse(&index, (uint32_t)(input_num)); - } -#endif - return 63 - index; -} - -/* result might be undefined when input_num is zero */ -#ifdef USESSE4 -/* POPCNT support was added to processors around the release of SSE4.2 */ -/* USESSE4 flag guarantees POPCNT support */ -static inline int __builtin_popcountll(unsigned long long input_num) { -#ifdef _WIN64 // highly recommended!!! - return (int)__popcnt64(input_num); -#else // if we must support 32-bit Windows - return (int)(__popcnt((uint32_t)input_num) + - __popcnt((uint32_t)(input_num >> 32))); -#endif -} -#else -/* software implementation avoids POPCNT */ -static inline int __builtin_popcountll(unsigned long long input_num) { - const uint64_t m1 = 0x5555555555555555; //binary: 0101... - const uint64_t m2 = 0x3333333333333333; //binary: 00110011.. - const uint64_t m4 = 0x0f0f0f0f0f0f0f0f; //binary: 4 zeros, 4 ones ... - const uint64_t h01 = 0x0101010101010101; //the sum of 256 to the power of 0,1,2,3... - - input_num -= (input_num >> 1) & m1; - input_num = (input_num & m2) + ((input_num >> 2) & m2); - input_num = (input_num + (input_num >> 4)) & m4; - return (input_num * h01) >> 56; -} -#endif - -/* Use #define so this is effective even under /Ob0 (no inline) */ -#define __builtin_unreachable() __assume(0) -#endif - -#endif - -// without the following, we get lots of warnings about posix_memalign -#ifndef __cplusplus -extern int posix_memalign(void **__memptr, size_t __alignment, size_t __size); -#endif //__cplusplus // C++ does not have a well defined signature - -// portable version of posix_memalign -static inline void *aligned_malloc(size_t alignment, size_t size) { - void *p; -#ifdef _MSC_VER - p = _aligned_malloc(size, alignment); -#elif defined(__MINGW32__) || defined(__MINGW64__) - p = __mingw_aligned_malloc(size, alignment); -#else - // somehow, if this is used before including "x86intrin.h", it creates an - // implicit defined warning. - if (posix_memalign(&p, alignment, size) != 0) return NULL; -#endif - return p; -} - -static inline void aligned_free(void *memblock) { -#ifdef _MSC_VER - _aligned_free(memblock); -#elif defined(__MINGW32__) || defined(__MINGW64__) - __mingw_aligned_free(memblock); -#else - free(memblock); -#endif -} - -#if defined(_MSC_VER) -#define ALIGNED(x) __declspec(align(x)) -#else -#if defined(__GNUC__) -#define ALIGNED(x) __attribute__((aligned(x))) -#endif -#endif - -#ifdef __GNUC__ -#define WARN_UNUSED __attribute__((warn_unused_result)) -#else -#define WARN_UNUSED -#endif - -#define IS_BIG_ENDIAN (*(uint16_t *)"\0\xff" < 0x100) - -static inline int hamming(uint64_t x) { -#ifdef USESSE4 - return (int) _mm_popcnt_u64(x); -#else - // won't work under visual studio, but hopeful we have _mm_popcnt_u64 in - // many cases - return __builtin_popcountll(x); -#endif -} - -#ifndef UINT64_C -#define UINT64_C(c) (c##ULL) -#endif - -#ifndef UINT32_C -#define UINT32_C(c) (c##UL) -#endif - -#ifdef __cplusplus -} -#endif - -#endif /* INCLUDE_PORTABILITY_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/portability.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/perfparameters.h */ -#ifndef PERFPARAMETERS_H_ -#define PERFPARAMETERS_H_ - -#include - -/** -During lazy computations, we can transform array containers into bitset -containers as -long as we can expect them to have ARRAY_LAZY_LOWERBOUND values. -*/ -enum { ARRAY_LAZY_LOWERBOUND = 1024 }; - -/* default initial size of a run container - setting it to zero delays the malloc.*/ -enum { RUN_DEFAULT_INIT_SIZE = 0 }; - -/* default initial size of an array container - setting it to zero delays the malloc */ -enum { ARRAY_DEFAULT_INIT_SIZE = 0 }; - -/* automatic bitset conversion during lazy or */ -#ifndef LAZY_OR_BITSET_CONVERSION -#define LAZY_OR_BITSET_CONVERSION true -#endif - -/* automatically attempt to convert a bitset to a full run during lazy - * evaluation */ -#ifndef LAZY_OR_BITSET_CONVERSION_TO_FULL -#define LAZY_OR_BITSET_CONVERSION_TO_FULL true -#endif - -/* automatically attempt to convert a bitset to a full run */ -#ifndef OR_BITSET_CONVERSION_TO_FULL -#define OR_BITSET_CONVERSION_TO_FULL true -#endif - -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/perfparameters.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/array_util.h */ -#ifndef ARRAY_UTIL_H -#define ARRAY_UTIL_H - -#include // for size_t -#include - - -/* - * Good old binary search. - * Assumes that array is sorted, has logarithmic complexity. - * if the result is x, then: - * if ( x>0 ) you have array[x] = ikey - * if ( x<0 ) then inserting ikey at position -x-1 in array (insuring that array[-x-1]=ikey) - * keys the array sorted. - */ -inline int32_t binarySearch(const uint16_t *array, int32_t lenarray, - uint16_t ikey) { - int32_t low = 0; - int32_t high = lenarray - 1; - while (low <= high) { - int32_t middleIndex = (low + high) >> 1; - uint16_t middleValue = array[middleIndex]; - if (middleValue < ikey) { - low = middleIndex + 1; - } else if (middleValue > ikey) { - high = middleIndex - 1; - } else { - return middleIndex; - } - } - return -(low + 1); -} - -/** - * Galloping search - * Assumes that array is sorted, has logarithmic complexity. - * if the result is x, then if x = length, you have that all values in array between pos and length - * are smaller than min. - * otherwise returns the first index x such that array[x] >= min. - */ -static inline int32_t advanceUntil(const uint16_t *array, int32_t pos, - int32_t length, uint16_t min) { - int32_t lower = pos + 1; - - if ((lower >= length) || (array[lower] >= min)) { - return lower; - } - - int32_t spansize = 1; - - while ((lower + spansize < length) && (array[lower + spansize] < min)) { - spansize <<= 1; - } - int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; - - if (array[upper] == min) { - return upper; - } - if (array[upper] < min) { - // means - // array - // has no - // item - // >= min - // pos = array.length; - return length; - } - - // we know that the next-smallest span was too small - lower += (spansize >> 1); - - int32_t mid = 0; - while (lower + 1 != upper) { - mid = (lower + upper) >> 1; - if (array[mid] == min) { - return mid; - } else if (array[mid] < min) { - lower = mid; - } else { - upper = mid; - } - } - return upper; -} - -/** - * Returns number of elements which are less then $ikey. - * Array elements must be unique and sorted. - */ -static inline int32_t count_less(const uint16_t *array, int32_t lenarray, - uint16_t ikey) { - if (lenarray == 0) return 0; - int32_t pos = binarySearch(array, lenarray, ikey); - return pos >= 0 ? pos : -(pos+1); -} - -/** - * Returns number of elements which are greater then $ikey. - * Array elements must be unique and sorted. - */ -static inline int32_t count_greater(const uint16_t *array, int32_t lenarray, - uint16_t ikey) { - if (lenarray == 0) return 0; - int32_t pos = binarySearch(array, lenarray, ikey); - if (pos >= 0) { - return lenarray - (pos+1); - } else { - return lenarray - (-pos-1); - } -} - -/** - * From Schlegel et al., Fast Sorted-Set Intersection using SIMD Instructions - * Optimized by D. Lemire on May 3rd 2013 - * - * C should have capacity greater than the minimum of s_1 and s_b + 8 - * where 8 is sizeof(__m128i)/sizeof(uint16_t). - */ -int32_t intersect_vector16(const uint16_t *__restrict__ A, size_t s_a, - const uint16_t *__restrict__ B, size_t s_b, - uint16_t *C); - -/** - * Compute the cardinality of the intersection using SSE4 instructions - */ -int32_t intersect_vector16_cardinality(const uint16_t *__restrict__ A, - size_t s_a, - const uint16_t *__restrict__ B, - size_t s_b); - -/* Computes the intersection between one small and one large set of uint16_t. - * Stores the result into buffer and return the number of elements. */ -int32_t intersect_skewed_uint16(const uint16_t *smallarray, size_t size_s, - const uint16_t *largearray, size_t size_l, - uint16_t *buffer); - -/* Computes the size of the intersection between one small and one large set of - * uint16_t. */ -int32_t intersect_skewed_uint16_cardinality(const uint16_t *smallarray, - size_t size_s, - const uint16_t *largearray, - size_t size_l); - - -/* Check whether the size of the intersection between one small and one large set of uint16_t is non-zero. */ -bool intersect_skewed_uint16_nonempty(const uint16_t *smallarray, size_t size_s, - const uint16_t *largearray, size_t size_l); -/** - * Generic intersection function. - */ -int32_t intersect_uint16(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB, uint16_t *out); -/** - * Compute the size of the intersection (generic). - */ -int32_t intersect_uint16_cardinality(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB); - -/** - * Checking whether the size of the intersection is non-zero. - */ -bool intersect_uint16_nonempty(const uint16_t *A, const size_t lenA, - const uint16_t *B, const size_t lenB); -/** - * Generic union function. - */ -size_t union_uint16(const uint16_t *set_1, size_t size_1, const uint16_t *set_2, - size_t size_2, uint16_t *buffer); - -/** - * Generic XOR function. - */ -int32_t xor_uint16(const uint16_t *array_1, int32_t card_1, - const uint16_t *array_2, int32_t card_2, uint16_t *out); - -/** - * Generic difference function (ANDNOT). - */ -int difference_uint16(const uint16_t *a1, int length1, const uint16_t *a2, - int length2, uint16_t *a_out); - -/** - * Generic intersection function. - */ -size_t intersection_uint32(const uint32_t *A, const size_t lenA, - const uint32_t *B, const size_t lenB, uint32_t *out); - -/** - * Generic intersection function, returns just the cardinality. - */ -size_t intersection_uint32_card(const uint32_t *A, const size_t lenA, - const uint32_t *B, const size_t lenB); - -/** - * Generic union function. - */ -size_t union_uint32(const uint32_t *set_1, size_t size_1, const uint32_t *set_2, - size_t size_2, uint32_t *buffer); - -/** - * A fast SSE-based union function. - */ -uint32_t union_vector16(const uint16_t *__restrict__ set_1, uint32_t size_1, - const uint16_t *__restrict__ set_2, uint32_t size_2, - uint16_t *__restrict__ buffer); -/** - * A fast SSE-based XOR function. - */ -uint32_t xor_vector16(const uint16_t *__restrict__ array1, uint32_t length1, - const uint16_t *__restrict__ array2, uint32_t length2, - uint16_t *__restrict__ output); - -/** - * A fast SSE-based difference function. - */ -int32_t difference_vector16(const uint16_t *__restrict__ A, size_t s_a, - const uint16_t *__restrict__ B, size_t s_b, - uint16_t *C); - -/** - * Generic union function, returns just the cardinality. - */ -size_t union_uint32_card(const uint32_t *set_1, size_t size_1, - const uint32_t *set_2, size_t size_2); - -/** -* combines union_uint16 and union_vector16 optimally -*/ -size_t fast_union_uint16(const uint16_t *set_1, size_t size_1, const uint16_t *set_2, - size_t size_2, uint16_t *buffer); - - -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/array_util.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_types.h */ -/* - Typedefs used by various components -*/ - -#ifndef ROARING_TYPES_H -#define ROARING_TYPES_H - -typedef bool (*roaring_iterator)(uint32_t value, void *param); -typedef bool (*roaring_iterator64)(uint64_t value, void *param); - -/** -* (For advanced users.) -* The roaring_statistics_t can be used to collect detailed statistics about -* the composition of a roaring bitmap. -*/ -typedef struct roaring_statistics_s { - uint32_t n_containers; /* number of containers */ - - uint32_t n_array_containers; /* number of array containers */ - uint32_t n_run_containers; /* number of run containers */ - uint32_t n_bitset_containers; /* number of bitmap containers */ - - uint32_t - n_values_array_containers; /* number of values in array containers */ - uint32_t n_values_run_containers; /* number of values in run containers */ - uint32_t - n_values_bitset_containers; /* number of values in bitmap containers */ - - uint32_t n_bytes_array_containers; /* number of allocated bytes in array - containers */ - uint32_t n_bytes_run_containers; /* number of allocated bytes in run - containers */ - uint32_t n_bytes_bitset_containers; /* number of allocated bytes in bitmap - containers */ - - uint32_t - max_value; /* the maximal value, undefined if cardinality is zero */ - uint32_t - min_value; /* the minimal value, undefined if cardinality is zero */ - uint64_t sum_value; /* the sum of all values (could be used to compute - average) */ - - uint64_t cardinality; /* total number of values stored in the bitmap */ - - // and n_values_arrays, n_values_rle, n_values_bitmap -} roaring_statistics_t; - -#endif /* ROARING_TYPES_H */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_types.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/utilasm.h */ -/* - * utilasm.h - * - */ - -#ifndef INCLUDE_UTILASM_H_ -#define INCLUDE_UTILASM_H_ - - -#if defined(USE_BMI) & defined(ROARING_INLINE_ASM) -#define ASMBITMANIPOPTIMIZATION // optimization flag - -#define ASM_SHIFT_RIGHT(srcReg, bitsReg, destReg) \ - __asm volatile("shrx %1, %2, %0" \ - : "=r"(destReg) \ - : /* write */ \ - "r"(bitsReg), /* read only */ \ - "r"(srcReg) /* read only */ \ - ) - -#define ASM_INPLACESHIFT_RIGHT(srcReg, bitsReg) \ - __asm volatile("shrx %1, %0, %0" \ - : "+r"(srcReg) \ - : /* read/write */ \ - "r"(bitsReg) /* read only */ \ - ) - -#define ASM_SHIFT_LEFT(srcReg, bitsReg, destReg) \ - __asm volatile("shlx %1, %2, %0" \ - : "=r"(destReg) \ - : /* write */ \ - "r"(bitsReg), /* read only */ \ - "r"(srcReg) /* read only */ \ - ) -// set bit at position testBit within testByte to 1 and -// copy cmovDst to cmovSrc if that bit was previously clear -#define ASM_SET_BIT_INC_WAS_CLEAR(testByte, testBit, count) \ - __asm volatile( \ - "bts %2, %0\n" \ - "sbb $-1, %1\n" \ - : "+r"(testByte), /* read/write */ \ - "+r"(count) \ - : /* read/write */ \ - "r"(testBit) /* read only */ \ - ) - -#define ASM_CLEAR_BIT_DEC_WAS_SET(testByte, testBit, count) \ - __asm volatile( \ - "btr %2, %0\n" \ - "sbb $0, %1\n" \ - : "+r"(testByte), /* read/write */ \ - "+r"(count) \ - : /* read/write */ \ - "r"(testBit) /* read only */ \ - ) - -#define ASM_BT64(testByte, testBit, count) \ - __asm volatile( \ - "bt %2,%1\n" \ - "sbb %0,%0" /*could use setb */ \ - : "=r"(count) \ - : /* write */ \ - "r"(testByte), /* read only */ \ - "r"(testBit) /* read only */ \ - ) - -#endif // USE_BMI -#endif /* INCLUDE_UTILASM_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/utilasm.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/bitset_util.h */ -#ifndef BITSET_UTIL_H -#define BITSET_UTIL_H - -#include - - -/* - * Set all bits in indexes [begin,end) to true. - */ -static inline void bitset_set_range(uint64_t *bitmap, uint32_t start, - uint32_t end) { - if (start == end) return; - uint32_t firstword = start / 64; - uint32_t endword = (end - 1) / 64; - if (firstword == endword) { - bitmap[firstword] |= ((~UINT64_C(0)) << (start % 64)) & - ((~UINT64_C(0)) >> ((~end + 1) % 64)); - return; - } - bitmap[firstword] |= (~UINT64_C(0)) << (start % 64); - for (uint32_t i = firstword + 1; i < endword; i++) bitmap[i] = ~UINT64_C(0); - bitmap[endword] |= (~UINT64_C(0)) >> ((~end + 1) % 64); -} - - -/* - * Find the cardinality of the bitset in [begin,begin+lenminusone] - */ -static inline int bitset_lenrange_cardinality(uint64_t *bitmap, uint32_t start, - uint32_t lenminusone) { - uint32_t firstword = start / 64; - uint32_t endword = (start + lenminusone) / 64; - if (firstword == endword) { - return hamming(bitmap[firstword] & - ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) - << (start % 64)); - } - int answer = hamming(bitmap[firstword] & ((~UINT64_C(0)) << (start % 64))); - for (uint32_t i = firstword + 1; i < endword; i++) { - answer += hamming(bitmap[i]); - } - answer += - hamming(bitmap[endword] & - (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64)); - return answer; -} - -/* - * Check whether the cardinality of the bitset in [begin,begin+lenminusone] is 0 - */ -static inline bool bitset_lenrange_empty(uint64_t *bitmap, uint32_t start, - uint32_t lenminusone) { - uint32_t firstword = start / 64; - uint32_t endword = (start + lenminusone) / 64; - if (firstword == endword) { - return (bitmap[firstword] & ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) - << (start % 64)) == 0; - } - if(((bitmap[firstword] & ((~UINT64_C(0)) << (start%64)))) != 0) return false; - for (uint32_t i = firstword + 1; i < endword; i++) { - if(bitmap[i] != 0) return false; - } - if((bitmap[endword] & (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64)) != 0) return false; - return true; -} - - -/* - * Set all bits in indexes [begin,begin+lenminusone] to true. - */ -static inline void bitset_set_lenrange(uint64_t *bitmap, uint32_t start, - uint32_t lenminusone) { - uint32_t firstword = start / 64; - uint32_t endword = (start + lenminusone) / 64; - if (firstword == endword) { - bitmap[firstword] |= ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) - << (start % 64); - return; - } - uint64_t temp = bitmap[endword]; - bitmap[firstword] |= (~UINT64_C(0)) << (start % 64); - for (uint32_t i = firstword + 1; i < endword; i += 2) - bitmap[i] = bitmap[i + 1] = ~UINT64_C(0); - bitmap[endword] = - temp | (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64); -} - -/* - * Flip all the bits in indexes [begin,end). - */ -static inline void bitset_flip_range(uint64_t *bitmap, uint32_t start, - uint32_t end) { - if (start == end) return; - uint32_t firstword = start / 64; - uint32_t endword = (end - 1) / 64; - bitmap[firstword] ^= ~((~UINT64_C(0)) << (start % 64)); - for (uint32_t i = firstword; i < endword; i++) bitmap[i] = ~bitmap[i]; - bitmap[endword] ^= ((~UINT64_C(0)) >> ((~end + 1) % 64)); -} - -/* - * Set all bits in indexes [begin,end) to false. - */ -static inline void bitset_reset_range(uint64_t *bitmap, uint32_t start, - uint32_t end) { - if (start == end) return; - uint32_t firstword = start / 64; - uint32_t endword = (end - 1) / 64; - if (firstword == endword) { - bitmap[firstword] &= ~(((~UINT64_C(0)) << (start % 64)) & - ((~UINT64_C(0)) >> ((~end + 1) % 64))); - return; - } - bitmap[firstword] &= ~((~UINT64_C(0)) << (start % 64)); - for (uint32_t i = firstword + 1; i < endword; i++) bitmap[i] = UINT64_C(0); - bitmap[endword] &= ~((~UINT64_C(0)) >> ((~end + 1) % 64)); -} - -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out", values start at "base". - * - * The "out" pointer should be sufficient to store the actual number of bits - * set. - * - * Returns how many values were actually decoded. - * - * This function should only be expected to be faster than - * bitset_extract_setbits - * when the density of the bitset is high. - * - * This function uses AVX2 decoding. - */ -size_t bitset_extract_setbits_avx2(uint64_t *bitset, size_t length, void *vout, - size_t outcapacity, uint32_t base); - -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out", values start at "base". - * - * The "out" pointer should be sufficient to store the actual number of bits - *set. - * - * Returns how many values were actually decoded. - */ -size_t bitset_extract_setbits(uint64_t *bitset, size_t length, void *vout, - uint32_t base); - -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out" as 16-bit integers, values start at "base" (can - *be set to zero) - * - * The "out" pointer should be sufficient to store the actual number of bits - *set. - * - * Returns how many values were actually decoded. - * - * This function should only be expected to be faster than - *bitset_extract_setbits_uint16 - * when the density of the bitset is high. - * - * This function uses SSE decoding. - */ -size_t bitset_extract_setbits_sse_uint16(const uint64_t *bitset, size_t length, - uint16_t *out, size_t outcapacity, - uint16_t base); - -/* - * Given a bitset containing "length" 64-bit words, write out the position - * of all the set bits to "out", values start at "base" - * (can be set to zero) - * - * The "out" pointer should be sufficient to store the actual number of bits - *set. - * - * Returns how many values were actually decoded. - */ -size_t bitset_extract_setbits_uint16(const uint64_t *bitset, size_t length, - uint16_t *out, uint16_t base); - -/* - * Given two bitsets containing "length" 64-bit words, write out the position - * of all the common set bits to "out", values start at "base" - * (can be set to zero) - * - * The "out" pointer should be sufficient to store the actual number of bits - * set. - * - * Returns how many values were actually decoded. - */ -size_t bitset_extract_intersection_setbits_uint16(const uint64_t * __restrict__ bitset1, - const uint64_t * __restrict__ bitset2, - size_t length, uint16_t *out, - uint16_t base); - -/* - * Given a bitset having cardinality card, set all bit values in the list (there - * are length of them) - * and return the updated cardinality. This evidently assumes that the bitset - * already contained data. - */ -uint64_t bitset_set_list_withcard(void *bitset, uint64_t card, - const uint16_t *list, uint64_t length); -/* - * Given a bitset, set all bit values in the list (there - * are length of them). - */ -void bitset_set_list(void *bitset, const uint16_t *list, uint64_t length); - -/* - * Given a bitset having cardinality card, unset all bit values in the list - * (there are length of them) - * and return the updated cardinality. This evidently assumes that the bitset - * already contained data. - */ -uint64_t bitset_clear_list(void *bitset, uint64_t card, const uint16_t *list, - uint64_t length); - -/* - * Given a bitset having cardinality card, toggle all bit values in the list - * (there are length of them) - * and return the updated cardinality. This evidently assumes that the bitset - * already contained data. - */ - -uint64_t bitset_flip_list_withcard(void *bitset, uint64_t card, - const uint16_t *list, uint64_t length); - -void bitset_flip_list(void *bitset, const uint16_t *list, uint64_t length); - -#ifdef USEAVX -/*** - * BEGIN Harley-Seal popcount functions. - */ - -/** - * Compute the population count of a 256-bit word - * This is not especially fast, but it is convenient as part of other functions. - */ -static inline __m256i popcount256(__m256i v) { - const __m256i lookuppos = _mm256_setr_epi8( - /* 0 */ 4 + 0, /* 1 */ 4 + 1, /* 2 */ 4 + 1, /* 3 */ 4 + 2, - /* 4 */ 4 + 1, /* 5 */ 4 + 2, /* 6 */ 4 + 2, /* 7 */ 4 + 3, - /* 8 */ 4 + 1, /* 9 */ 4 + 2, /* a */ 4 + 2, /* b */ 4 + 3, - /* c */ 4 + 2, /* d */ 4 + 3, /* e */ 4 + 3, /* f */ 4 + 4, - - /* 0 */ 4 + 0, /* 1 */ 4 + 1, /* 2 */ 4 + 1, /* 3 */ 4 + 2, - /* 4 */ 4 + 1, /* 5 */ 4 + 2, /* 6 */ 4 + 2, /* 7 */ 4 + 3, - /* 8 */ 4 + 1, /* 9 */ 4 + 2, /* a */ 4 + 2, /* b */ 4 + 3, - /* c */ 4 + 2, /* d */ 4 + 3, /* e */ 4 + 3, /* f */ 4 + 4); - const __m256i lookupneg = _mm256_setr_epi8( - /* 0 */ 4 - 0, /* 1 */ 4 - 1, /* 2 */ 4 - 1, /* 3 */ 4 - 2, - /* 4 */ 4 - 1, /* 5 */ 4 - 2, /* 6 */ 4 - 2, /* 7 */ 4 - 3, - /* 8 */ 4 - 1, /* 9 */ 4 - 2, /* a */ 4 - 2, /* b */ 4 - 3, - /* c */ 4 - 2, /* d */ 4 - 3, /* e */ 4 - 3, /* f */ 4 - 4, - - /* 0 */ 4 - 0, /* 1 */ 4 - 1, /* 2 */ 4 - 1, /* 3 */ 4 - 2, - /* 4 */ 4 - 1, /* 5 */ 4 - 2, /* 6 */ 4 - 2, /* 7 */ 4 - 3, - /* 8 */ 4 - 1, /* 9 */ 4 - 2, /* a */ 4 - 2, /* b */ 4 - 3, - /* c */ 4 - 2, /* d */ 4 - 3, /* e */ 4 - 3, /* f */ 4 - 4); - const __m256i low_mask = _mm256_set1_epi8(0x0f); - - const __m256i lo = _mm256_and_si256(v, low_mask); - const __m256i hi = _mm256_and_si256(_mm256_srli_epi16(v, 4), low_mask); - const __m256i popcnt1 = _mm256_shuffle_epi8(lookuppos, lo); - const __m256i popcnt2 = _mm256_shuffle_epi8(lookupneg, hi); - return _mm256_sad_epu8(popcnt1, popcnt2); -} - -/** - * Simple CSA over 256 bits - */ -static inline void CSA(__m256i *h, __m256i *l, __m256i a, __m256i b, - __m256i c) { - const __m256i u = _mm256_xor_si256(a, b); - *h = _mm256_or_si256(_mm256_and_si256(a, b), _mm256_and_si256(u, c)); - *l = _mm256_xor_si256(u, c); -} - -/** - * Fast Harley-Seal AVX population count function - */ -inline static uint64_t avx2_harley_seal_popcount256(const __m256i *data, - const uint64_t size) { - __m256i total = _mm256_setzero_si256(); - __m256i ones = _mm256_setzero_si256(); - __m256i twos = _mm256_setzero_si256(); - __m256i fours = _mm256_setzero_si256(); - __m256i eights = _mm256_setzero_si256(); - __m256i sixteens = _mm256_setzero_si256(); - __m256i twosA, twosB, foursA, foursB, eightsA, eightsB; - - const uint64_t limit = size - size % 16; - uint64_t i = 0; - - for (; i < limit; i += 16) { - CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i), - _mm256_lddqu_si256(data + i + 1)); - CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 2), - _mm256_lddqu_si256(data + i + 3)); - CSA(&foursA, &twos, twos, twosA, twosB); - CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 4), - _mm256_lddqu_si256(data + i + 5)); - CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 6), - _mm256_lddqu_si256(data + i + 7)); - CSA(&foursB, &twos, twos, twosA, twosB); - CSA(&eightsA, &fours, fours, foursA, foursB); - CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 8), - _mm256_lddqu_si256(data + i + 9)); - CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 10), - _mm256_lddqu_si256(data + i + 11)); - CSA(&foursA, &twos, twos, twosA, twosB); - CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 12), - _mm256_lddqu_si256(data + i + 13)); - CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 14), - _mm256_lddqu_si256(data + i + 15)); - CSA(&foursB, &twos, twos, twosA, twosB); - CSA(&eightsB, &fours, fours, foursA, foursB); - CSA(&sixteens, &eights, eights, eightsA, eightsB); - - total = _mm256_add_epi64(total, popcount256(sixteens)); - } - - total = _mm256_slli_epi64(total, 4); // * 16 - total = _mm256_add_epi64( - total, _mm256_slli_epi64(popcount256(eights), 3)); // += 8 * ... - total = _mm256_add_epi64( - total, _mm256_slli_epi64(popcount256(fours), 2)); // += 4 * ... - total = _mm256_add_epi64( - total, _mm256_slli_epi64(popcount256(twos), 1)); // += 2 * ... - total = _mm256_add_epi64(total, popcount256(ones)); - for (; i < size; i++) - total = - _mm256_add_epi64(total, popcount256(_mm256_lddqu_si256(data + i))); - - return (uint64_t)(_mm256_extract_epi64(total, 0)) + - (uint64_t)(_mm256_extract_epi64(total, 1)) + - (uint64_t)(_mm256_extract_epi64(total, 2)) + - (uint64_t)(_mm256_extract_epi64(total, 3)); -} - -#define AVXPOPCNTFNC(opname, avx_intrinsic) \ - static inline uint64_t avx2_harley_seal_popcount256_##opname( \ - const __m256i *data1, const __m256i *data2, const uint64_t size) { \ - __m256i total = _mm256_setzero_si256(); \ - __m256i ones = _mm256_setzero_si256(); \ - __m256i twos = _mm256_setzero_si256(); \ - __m256i fours = _mm256_setzero_si256(); \ - __m256i eights = _mm256_setzero_si256(); \ - __m256i sixteens = _mm256_setzero_si256(); \ - __m256i twosA, twosB, foursA, foursB, eightsA, eightsB; \ - __m256i A1, A2; \ - const uint64_t limit = size - size % 16; \ - uint64_t i = 0; \ - for (; i < limit; i += 16) { \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i), \ - _mm256_lddqu_si256(data2 + i)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 1), \ - _mm256_lddqu_si256(data2 + i + 1)); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 2), \ - _mm256_lddqu_si256(data2 + i + 2)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 3), \ - _mm256_lddqu_si256(data2 + i + 3)); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursA, &twos, twos, twosA, twosB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 4), \ - _mm256_lddqu_si256(data2 + i + 4)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 5), \ - _mm256_lddqu_si256(data2 + i + 5)); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 6), \ - _mm256_lddqu_si256(data2 + i + 6)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 7), \ - _mm256_lddqu_si256(data2 + i + 7)); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursB, &twos, twos, twosA, twosB); \ - CSA(&eightsA, &fours, fours, foursA, foursB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 8), \ - _mm256_lddqu_si256(data2 + i + 8)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 9), \ - _mm256_lddqu_si256(data2 + i + 9)); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 10), \ - _mm256_lddqu_si256(data2 + i + 10)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 11), \ - _mm256_lddqu_si256(data2 + i + 11)); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursA, &twos, twos, twosA, twosB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 12), \ - _mm256_lddqu_si256(data2 + i + 12)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 13), \ - _mm256_lddqu_si256(data2 + i + 13)); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 14), \ - _mm256_lddqu_si256(data2 + i + 14)); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 15), \ - _mm256_lddqu_si256(data2 + i + 15)); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursB, &twos, twos, twosA, twosB); \ - CSA(&eightsB, &fours, fours, foursA, foursB); \ - CSA(&sixteens, &eights, eights, eightsA, eightsB); \ - total = _mm256_add_epi64(total, popcount256(sixteens)); \ - } \ - total = _mm256_slli_epi64(total, 4); \ - total = _mm256_add_epi64(total, \ - _mm256_slli_epi64(popcount256(eights), 3)); \ - total = \ - _mm256_add_epi64(total, _mm256_slli_epi64(popcount256(fours), 2)); \ - total = \ - _mm256_add_epi64(total, _mm256_slli_epi64(popcount256(twos), 1)); \ - total = _mm256_add_epi64(total, popcount256(ones)); \ - for (; i < size; i++) { \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i), \ - _mm256_lddqu_si256(data2 + i)); \ - total = _mm256_add_epi64(total, popcount256(A1)); \ - } \ - return (uint64_t)(_mm256_extract_epi64(total, 0)) + \ - (uint64_t)(_mm256_extract_epi64(total, 1)) + \ - (uint64_t)(_mm256_extract_epi64(total, 2)) + \ - (uint64_t)(_mm256_extract_epi64(total, 3)); \ - } \ - static inline uint64_t avx2_harley_seal_popcount256andstore_##opname( \ - const __m256i *__restrict__ data1, const __m256i *__restrict__ data2, \ - __m256i *__restrict__ out, const uint64_t size) { \ - __m256i total = _mm256_setzero_si256(); \ - __m256i ones = _mm256_setzero_si256(); \ - __m256i twos = _mm256_setzero_si256(); \ - __m256i fours = _mm256_setzero_si256(); \ - __m256i eights = _mm256_setzero_si256(); \ - __m256i sixteens = _mm256_setzero_si256(); \ - __m256i twosA, twosB, foursA, foursB, eightsA, eightsB; \ - __m256i A1, A2; \ - const uint64_t limit = size - size % 16; \ - uint64_t i = 0; \ - for (; i < limit; i += 16) { \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i), \ - _mm256_lddqu_si256(data2 + i)); \ - _mm256_storeu_si256(out + i, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 1), \ - _mm256_lddqu_si256(data2 + i + 1)); \ - _mm256_storeu_si256(out + i + 1, A2); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 2), \ - _mm256_lddqu_si256(data2 + i + 2)); \ - _mm256_storeu_si256(out + i + 2, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 3), \ - _mm256_lddqu_si256(data2 + i + 3)); \ - _mm256_storeu_si256(out + i + 3, A2); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursA, &twos, twos, twosA, twosB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 4), \ - _mm256_lddqu_si256(data2 + i + 4)); \ - _mm256_storeu_si256(out + i + 4, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 5), \ - _mm256_lddqu_si256(data2 + i + 5)); \ - _mm256_storeu_si256(out + i + 5, A2); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 6), \ - _mm256_lddqu_si256(data2 + i + 6)); \ - _mm256_storeu_si256(out + i + 6, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 7), \ - _mm256_lddqu_si256(data2 + i + 7)); \ - _mm256_storeu_si256(out + i + 7, A2); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursB, &twos, twos, twosA, twosB); \ - CSA(&eightsA, &fours, fours, foursA, foursB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 8), \ - _mm256_lddqu_si256(data2 + i + 8)); \ - _mm256_storeu_si256(out + i + 8, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 9), \ - _mm256_lddqu_si256(data2 + i + 9)); \ - _mm256_storeu_si256(out + i + 9, A2); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 10), \ - _mm256_lddqu_si256(data2 + i + 10)); \ - _mm256_storeu_si256(out + i + 10, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 11), \ - _mm256_lddqu_si256(data2 + i + 11)); \ - _mm256_storeu_si256(out + i + 11, A2); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursA, &twos, twos, twosA, twosB); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 12), \ - _mm256_lddqu_si256(data2 + i + 12)); \ - _mm256_storeu_si256(out + i + 12, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 13), \ - _mm256_lddqu_si256(data2 + i + 13)); \ - _mm256_storeu_si256(out + i + 13, A2); \ - CSA(&twosA, &ones, ones, A1, A2); \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 14), \ - _mm256_lddqu_si256(data2 + i + 14)); \ - _mm256_storeu_si256(out + i + 14, A1); \ - A2 = avx_intrinsic(_mm256_lddqu_si256(data1 + i + 15), \ - _mm256_lddqu_si256(data2 + i + 15)); \ - _mm256_storeu_si256(out + i + 15, A2); \ - CSA(&twosB, &ones, ones, A1, A2); \ - CSA(&foursB, &twos, twos, twosA, twosB); \ - CSA(&eightsB, &fours, fours, foursA, foursB); \ - CSA(&sixteens, &eights, eights, eightsA, eightsB); \ - total = _mm256_add_epi64(total, popcount256(sixteens)); \ - } \ - total = _mm256_slli_epi64(total, 4); \ - total = _mm256_add_epi64(total, \ - _mm256_slli_epi64(popcount256(eights), 3)); \ - total = \ - _mm256_add_epi64(total, _mm256_slli_epi64(popcount256(fours), 2)); \ - total = \ - _mm256_add_epi64(total, _mm256_slli_epi64(popcount256(twos), 1)); \ - total = _mm256_add_epi64(total, popcount256(ones)); \ - for (; i < size; i++) { \ - A1 = avx_intrinsic(_mm256_lddqu_si256(data1 + i), \ - _mm256_lddqu_si256(data2 + i)); \ - _mm256_storeu_si256(out + i, A1); \ - total = _mm256_add_epi64(total, popcount256(A1)); \ - } \ - return (uint64_t)(_mm256_extract_epi64(total, 0)) + \ - (uint64_t)(_mm256_extract_epi64(total, 1)) + \ - (uint64_t)(_mm256_extract_epi64(total, 2)) + \ - (uint64_t)(_mm256_extract_epi64(total, 3)); \ - } - -AVXPOPCNTFNC(or, _mm256_or_si256) -AVXPOPCNTFNC(union, _mm256_or_si256) -AVXPOPCNTFNC(and, _mm256_and_si256) -AVXPOPCNTFNC(intersection, _mm256_and_si256) -AVXPOPCNTFNC (xor, _mm256_xor_si256) -AVXPOPCNTFNC(andnot, _mm256_andnot_si256) - -/*** - * END Harley-Seal popcount functions. - */ - -#endif // USEAVX - -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/bitset_util.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/array.h */ -/* - * array.h - * - */ - -#ifndef INCLUDE_CONTAINERS_ARRAY_H_ -#define INCLUDE_CONTAINERS_ARRAY_H_ - -#ifdef __cplusplus -extern "C" { -#endif - -#include - - -/* Containers with DEFAULT_MAX_SIZE or less integers should be arrays */ -enum { DEFAULT_MAX_SIZE = 4096 }; - -/* struct array_container - sparse representation of a bitmap - * - * @cardinality: number of indices in `array` (and the bitmap) - * @capacity: allocated size of `array` - * @array: sorted list of integers - */ -struct array_container_s { - int32_t cardinality; - int32_t capacity; - uint16_t *array; -}; - -typedef struct array_container_s array_container_t; - -/* Create a new array with default. Return NULL in case of failure. See also - * array_container_create_given_capacity. */ -array_container_t *array_container_create(void); - -/* Create a new array with a specified capacity size. Return NULL in case of - * failure. */ -array_container_t *array_container_create_given_capacity(int32_t size); - -/* Create a new array containing all values in [min,max). */ -array_container_t * array_container_create_range(uint32_t min, uint32_t max); - -/* - * Shrink the capacity to the actual size, return the number of bytes saved. - */ -int array_container_shrink_to_fit(array_container_t *src); - -/* Free memory owned by `array'. */ -void array_container_free(array_container_t *array); - -/* Duplicate container */ -array_container_t *array_container_clone(const array_container_t *src); - -int32_t array_container_serialize(const array_container_t *container, - char *buf) WARN_UNUSED; - -uint32_t array_container_serialization_len(const array_container_t *container); - -void *array_container_deserialize(const char *buf, size_t buf_len); - -/* Get the cardinality of `array'. */ -static inline int array_container_cardinality(const array_container_t *array) { - return array->cardinality; -} - -static inline bool array_container_nonzero_cardinality( - const array_container_t *array) { - return array->cardinality > 0; -} - -/* Copy one container into another. We assume that they are distinct. */ -void array_container_copy(const array_container_t *src, array_container_t *dst); - -/* Add all the values in [min,max) (included) at a distance k*step from min. - The container must have a size less or equal to DEFAULT_MAX_SIZE after this - addition. */ -void array_container_add_from_range(array_container_t *arr, uint32_t min, - uint32_t max, uint16_t step); - -/* Set the cardinality to zero (does not release memory). */ -static inline void array_container_clear(array_container_t *array) { - array->cardinality = 0; -} - -static inline bool array_container_empty(const array_container_t *array) { - return array->cardinality == 0; -} - -/* check whether the cardinality is equal to the capacity (this does not mean -* that it contains 1<<16 elements) */ -static inline bool array_container_full(const array_container_t *array) { - return array->cardinality == array->capacity; -} - - -/* Compute the union of `src_1' and `src_2' and write the result to `dst' - * It is assumed that `dst' is distinct from both `src_1' and `src_2'. */ -void array_container_union(const array_container_t *src_1, - const array_container_t *src_2, - array_container_t *dst); - -/* symmetric difference, see array_container_union */ -void array_container_xor(const array_container_t *array_1, - const array_container_t *array_2, - array_container_t *out); - -/* Computes the intersection of src_1 and src_2 and write the result to - * dst. It is assumed that dst is distinct from both src_1 and src_2. */ -void array_container_intersection(const array_container_t *src_1, - const array_container_t *src_2, - array_container_t *dst); - -/* Check whether src_1 and src_2 intersect. */ -bool array_container_intersect(const array_container_t *src_1, - const array_container_t *src_2); - - -/* computers the size of the intersection between two arrays. - */ -int array_container_intersection_cardinality(const array_container_t *src_1, - const array_container_t *src_2); - -/* computes the intersection of array1 and array2 and write the result to - * array1. - * */ -void array_container_intersection_inplace(array_container_t *src_1, - const array_container_t *src_2); - -/* - * Write out the 16-bit integers contained in this container as a list of 32-bit - * integers using base - * as the starting value (it might be expected that base has zeros in its 16 - * least significant bits). - * The function returns the number of values written. - * The caller is responsible for allocating enough memory in out. - */ -int array_container_to_uint32_array(void *vout, const array_container_t *cont, - uint32_t base); - -/* Compute the number of runs */ -int32_t array_container_number_of_runs(const array_container_t *a); - -/* - * Print this container using printf (useful for debugging). - */ -void array_container_printf(const array_container_t *v); - -/* - * Print this container using printf as a comma-separated list of 32-bit - * integers starting at base. - */ -void array_container_printf_as_uint32_array(const array_container_t *v, - uint32_t base); - -/** - * Return the serialized size in bytes of a container having cardinality "card". - */ -static inline int32_t array_container_serialized_size_in_bytes(int32_t card) { - return card * 2 + 2; -} - -/** - * Increase capacity to at least min. - * Whether the existing data needs to be copied over depends on the "preserve" - * parameter. If preserve is false, then the new content will be uninitialized, - * otherwise the old content is copied. - */ -void array_container_grow(array_container_t *container, int32_t min, - bool preserve); - -bool array_container_iterate(const array_container_t *cont, uint32_t base, - roaring_iterator iterator, void *ptr); -bool array_container_iterate64(const array_container_t *cont, uint32_t base, - roaring_iterator64 iterator, uint64_t high_bits, - void *ptr); - -/** - * Writes the underlying array to buf, outputs how many bytes were written. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes written should be - * array_container_size_in_bytes(container). - * - */ -int32_t array_container_write(const array_container_t *container, char *buf); -/** - * Reads the instance from buf, outputs how many bytes were read. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes read should be array_container_size_in_bytes(container). - * You need to provide the (known) cardinality. - */ -int32_t array_container_read(int32_t cardinality, array_container_t *container, - const char *buf); - -/** - * Return the serialized size in bytes of a container (see - * bitset_container_write) - * This is meant to be compatible with the Java and Go versions of Roaring and - * assumes - * that the cardinality of the container is already known. - * - */ -static inline int32_t array_container_size_in_bytes( - const array_container_t *container) { - return container->cardinality * sizeof(uint16_t); -} - -/** - * Return true if the two arrays have the same content. - */ -bool array_container_equals(const array_container_t *container1, - const array_container_t *container2); - -/** - * Return true if container1 is a subset of container2. - */ -bool array_container_is_subset(const array_container_t *container1, - const array_container_t *container2); - -/** - * If the element of given rank is in this container, supposing that the first - * element has rank start_rank, then the function returns true and sets element - * accordingly. - * Otherwise, it returns false and update start_rank. - */ -static inline bool array_container_select(const array_container_t *container, - uint32_t *start_rank, uint32_t rank, - uint32_t *element) { - int card = array_container_cardinality(container); - if (*start_rank + card <= rank) { - *start_rank += card; - return false; - } else { - *element = container->array[rank - *start_rank]; - return true; - } -} - -/* Computes the difference of array1 and array2 and write the result - * to array out. - * Array out does not need to be distinct from array_1 - */ -void array_container_andnot(const array_container_t *array_1, - const array_container_t *array_2, - array_container_t *out); - -/* Append x to the set. Assumes that the value is larger than any preceding - * values. */ -static inline void array_container_append(array_container_t *arr, - uint16_t pos) { - const int32_t capacity = arr->capacity; - - if (array_container_full(arr)) { - array_container_grow(arr, capacity + 1, true); - } - - arr->array[arr->cardinality++] = pos; -} - -/** - * Add value to the set if final cardinality doesn't exceed max_cardinality. - * Return code: - * 1 -- value was added - * 0 -- value was already present - * -1 -- value was not added because cardinality would exceed max_cardinality - */ -static inline int array_container_try_add(array_container_t *arr, uint16_t value, - int32_t max_cardinality) { - const int32_t cardinality = arr->cardinality; - - // best case, we can append. - if ((array_container_empty(arr) || arr->array[cardinality - 1] < value) && - cardinality < max_cardinality) { - array_container_append(arr, value); - return 1; - } - - const int32_t loc = binarySearch(arr->array, cardinality, value); - - if (loc >= 0) { - return 0; - } else if (cardinality < max_cardinality) { - if (array_container_full(arr)) { - array_container_grow(arr, arr->capacity + 1, true); - } - const int32_t insert_idx = -loc - 1; - memmove(arr->array + insert_idx + 1, arr->array + insert_idx, - (cardinality - insert_idx) * sizeof(uint16_t)); - arr->array[insert_idx] = value; - arr->cardinality++; - return 1; - } else { - return -1; - } -} - -/* Add value to the set. Returns true if x was not already present. */ -static inline bool array_container_add(array_container_t *arr, uint16_t value) { - return array_container_try_add(arr, value, INT32_MAX) == 1; -} - -/* Remove x from the set. Returns true if x was present. */ -static inline bool array_container_remove(array_container_t *arr, - uint16_t pos) { - const int32_t idx = binarySearch(arr->array, arr->cardinality, pos); - const bool is_present = idx >= 0; - if (is_present) { - memmove(arr->array + idx, arr->array + idx + 1, - (arr->cardinality - idx - 1) * sizeof(uint16_t)); - arr->cardinality--; - } - - return is_present; -} - -/* Check whether x is present. */ -inline bool array_container_contains(const array_container_t *arr, - uint16_t pos) { - // return binarySearch(arr->array, arr->cardinality, pos) >= 0; - // binary search with fallback to linear search for short ranges - int32_t low = 0; - const uint16_t * carr = (const uint16_t *) arr->array; - int32_t high = arr->cardinality - 1; - // while (high - low >= 0) { - while(high >= low + 16) { - int32_t middleIndex = (low + high)>>1; - uint16_t middleValue = carr[middleIndex]; - if (middleValue < pos) { - low = middleIndex + 1; - } else if (middleValue > pos) { - high = middleIndex - 1; - } else { - return true; - } - } - - for (int i=low; i <= high; i++) { - uint16_t v = carr[i]; - if (v == pos) { - return true; - } - if ( v > pos ) return false; - } - return false; - -} - - -//* Check whether a range of values from range_start (included) to range_end (excluded) is present. */ -static inline bool array_container_contains_range(const array_container_t *arr, - uint32_t range_start, uint32_t range_end) { - - const uint16_t rs_included = range_start; - const uint16_t re_included = range_end - 1; - - const uint16_t *carr = (const uint16_t *) arr->array; - - const int32_t start = advanceUntil(carr, -1, arr->cardinality, rs_included); - const int32_t end = advanceUntil(carr, start - 1, arr->cardinality, re_included); - - return (start < arr->cardinality) && (end < arr->cardinality) - && (((uint16_t)(end - start)) == re_included - rs_included) - && (carr[start] == rs_included) && (carr[end] == re_included); -} - -/* Returns the smallest value (assumes not empty) */ -inline uint16_t array_container_minimum(const array_container_t *arr) { - if (arr->cardinality == 0) return 0; - return arr->array[0]; -} - -/* Returns the largest value (assumes not empty) */ -inline uint16_t array_container_maximum(const array_container_t *arr) { - if (arr->cardinality == 0) return 0; - return arr->array[arr->cardinality - 1]; -} - -/* Returns the number of values equal or smaller than x */ -inline int array_container_rank(const array_container_t *arr, uint16_t x) { - const int32_t idx = binarySearch(arr->array, arr->cardinality, x); - const bool is_present = idx >= 0; - if (is_present) { - return idx + 1; - } else { - return -idx - 1; - } -} - -/* Returns the index of the first value equal or smaller than x, or -1 */ -inline int array_container_index_equalorlarger(const array_container_t *arr, uint16_t x) { - const int32_t idx = binarySearch(arr->array, arr->cardinality, x); - const bool is_present = idx >= 0; - if (is_present) { - return idx; - } else { - int32_t candidate = - idx - 1; - if(candidate < arr->cardinality) return candidate; - return -1; - } -} - -/* - * Adds all values in range [min,max] using hint: - * nvals_less is the number of array values less than $min - * nvals_greater is the number of array values greater than $max - */ -static inline void array_container_add_range_nvals(array_container_t *array, - uint32_t min, uint32_t max, - int32_t nvals_less, - int32_t nvals_greater) { - int32_t union_cardinality = nvals_less + (max - min + 1) + nvals_greater; - if (union_cardinality > array->capacity) { - array_container_grow(array, union_cardinality, true); - } - memmove(&(array->array[union_cardinality - nvals_greater]), - &(array->array[array->cardinality - nvals_greater]), - nvals_greater * sizeof(uint16_t)); - for (uint32_t i = 0; i <= max - min; i++) { - array->array[nvals_less + i] = min + i; - } - array->cardinality = union_cardinality; -} - -/** - * Adds all values in range [min,max]. - */ -static inline void array_container_add_range(array_container_t *array, - uint32_t min, uint32_t max) { - int32_t nvals_greater = count_greater(array->array, array->cardinality, max); - int32_t nvals_less = count_less(array->array, array->cardinality - nvals_greater, min); - array_container_add_range_nvals(array, min, max, nvals_less, nvals_greater); -} - -/* - * Removes all elements array[pos] .. array[pos+count-1] - */ -static inline void array_container_remove_range(array_container_t *array, - uint32_t pos, uint32_t count) { - if (count != 0) { - memmove(&(array->array[pos]), &(array->array[pos+count]), - (array->cardinality - pos - count) * sizeof(uint16_t)); - array->cardinality -= count; - } -} - -#ifdef __cplusplus -} -#endif - -#endif /* INCLUDE_CONTAINERS_ARRAY_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/array.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/bitset.h */ -/* - * bitset.h - * - */ - -#ifndef INCLUDE_CONTAINERS_BITSET_H_ -#define INCLUDE_CONTAINERS_BITSET_H_ - -#include -#include - -#ifdef USEAVX -#define ALIGN_AVX __attribute__((aligned(sizeof(__m256i)))) -#else -#define ALIGN_AVX -#endif - -enum { - BITSET_CONTAINER_SIZE_IN_WORDS = (1 << 16) / 64, - BITSET_UNKNOWN_CARDINALITY = -1 -}; - -struct bitset_container_s { - int32_t cardinality; - uint64_t *array; -}; - -typedef struct bitset_container_s bitset_container_t; - -/* Create a new bitset. Return NULL in case of failure. */ -bitset_container_t *bitset_container_create(void); - -/* Free memory. */ -void bitset_container_free(bitset_container_t *bitset); - -/* Clear bitset (sets bits to 0). */ -void bitset_container_clear(bitset_container_t *bitset); - -/* Set all bits to 1. */ -void bitset_container_set_all(bitset_container_t *bitset); - -/* Duplicate bitset */ -bitset_container_t *bitset_container_clone(const bitset_container_t *src); - -int32_t bitset_container_serialize(const bitset_container_t *container, - char *buf) WARN_UNUSED; - -uint32_t bitset_container_serialization_len(void); - -void *bitset_container_deserialize(const char *buf, size_t buf_len); - -/* Set the bit in [begin,end). WARNING: as of April 2016, this method is slow - * and - * should not be used in performance-sensitive code. Ever. */ -void bitset_container_set_range(bitset_container_t *bitset, uint32_t begin, - uint32_t end); - -#ifdef ASMBITMANIPOPTIMIZATION -/* Set the ith bit. */ -static inline void bitset_container_set(bitset_container_t *bitset, - uint16_t pos) { - uint64_t shift = 6; - uint64_t offset; - uint64_t p = pos; - ASM_SHIFT_RIGHT(p, shift, offset); - uint64_t load = bitset->array[offset]; - ASM_SET_BIT_INC_WAS_CLEAR(load, p, bitset->cardinality); - bitset->array[offset] = load; -} - -/* Unset the ith bit. */ -static inline void bitset_container_unset(bitset_container_t *bitset, - uint16_t pos) { - uint64_t shift = 6; - uint64_t offset; - uint64_t p = pos; - ASM_SHIFT_RIGHT(p, shift, offset); - uint64_t load = bitset->array[offset]; - ASM_CLEAR_BIT_DEC_WAS_SET(load, p, bitset->cardinality); - bitset->array[offset] = load; -} - -/* Add `pos' to `bitset'. Returns true if `pos' was not present. Might be slower - * than bitset_container_set. */ -static inline bool bitset_container_add(bitset_container_t *bitset, - uint16_t pos) { - uint64_t shift = 6; - uint64_t offset; - uint64_t p = pos; - ASM_SHIFT_RIGHT(p, shift, offset); - uint64_t load = bitset->array[offset]; - // could be possibly slightly further optimized - const int32_t oldcard = bitset->cardinality; - ASM_SET_BIT_INC_WAS_CLEAR(load, p, bitset->cardinality); - bitset->array[offset] = load; - return bitset->cardinality - oldcard; -} - -/* Remove `pos' from `bitset'. Returns true if `pos' was present. Might be - * slower than bitset_container_unset. */ -static inline bool bitset_container_remove(bitset_container_t *bitset, - uint16_t pos) { - uint64_t shift = 6; - uint64_t offset; - uint64_t p = pos; - ASM_SHIFT_RIGHT(p, shift, offset); - uint64_t load = bitset->array[offset]; - // could be possibly slightly further optimized - const int32_t oldcard = bitset->cardinality; - ASM_CLEAR_BIT_DEC_WAS_SET(load, p, bitset->cardinality); - bitset->array[offset] = load; - return oldcard - bitset->cardinality; -} - -/* Get the value of the ith bit. */ -inline bool bitset_container_get(const bitset_container_t *bitset, - uint16_t pos) { - uint64_t word = bitset->array[pos >> 6]; - const uint64_t p = pos; - ASM_INPLACESHIFT_RIGHT(word, p); - return word & 1; -} - -#else - -/* Set the ith bit. */ -static inline void bitset_container_set(bitset_container_t *bitset, - uint16_t pos) { - const uint64_t old_word = bitset->array[pos >> 6]; - const int index = pos & 63; - const uint64_t new_word = old_word | (UINT64_C(1) << index); - bitset->cardinality += (uint32_t)((old_word ^ new_word) >> index); - bitset->array[pos >> 6] = new_word; -} - -/* Unset the ith bit. */ -static inline void bitset_container_unset(bitset_container_t *bitset, - uint16_t pos) { - const uint64_t old_word = bitset->array[pos >> 6]; - const int index = pos & 63; - const uint64_t new_word = old_word & (~(UINT64_C(1) << index)); - bitset->cardinality -= (uint32_t)((old_word ^ new_word) >> index); - bitset->array[pos >> 6] = new_word; -} - -/* Add `pos' to `bitset'. Returns true if `pos' was not present. Might be slower - * than bitset_container_set. */ -static inline bool bitset_container_add(bitset_container_t *bitset, - uint16_t pos) { - const uint64_t old_word = bitset->array[pos >> 6]; - const int index = pos & 63; - const uint64_t new_word = old_word | (UINT64_C(1) << index); - const uint64_t increment = (old_word ^ new_word) >> index; - bitset->cardinality += (uint32_t)increment; - bitset->array[pos >> 6] = new_word; - return increment > 0; -} - -/* Remove `pos' from `bitset'. Returns true if `pos' was present. Might be - * slower than bitset_container_unset. */ -static inline bool bitset_container_remove(bitset_container_t *bitset, - uint16_t pos) { - const uint64_t old_word = bitset->array[pos >> 6]; - const int index = pos & 63; - const uint64_t new_word = old_word & (~(UINT64_C(1) << index)); - const uint64_t increment = (old_word ^ new_word) >> index; - bitset->cardinality -= (uint32_t)increment; - bitset->array[pos >> 6] = new_word; - return increment > 0; -} - -/* Get the value of the ith bit. */ -inline bool bitset_container_get(const bitset_container_t *bitset, - uint16_t pos) { - const uint64_t word = bitset->array[pos >> 6]; - return (word >> (pos & 63)) & 1; -} - -#endif - -/* -* Check if all bits are set in a range of positions from pos_start (included) to -* pos_end (excluded). -*/ -static inline bool bitset_container_get_range(const bitset_container_t *bitset, - uint32_t pos_start, uint32_t pos_end) { - - const uint32_t start = pos_start >> 6; - const uint32_t end = pos_end >> 6; - - const uint64_t first = ~((1ULL << (pos_start & 0x3F)) - 1); - const uint64_t last = (1ULL << (pos_end & 0x3F)) - 1; - - if (start == end) return ((bitset->array[end] & first & last) == (first & last)); - if ((bitset->array[start] & first) != first) return false; - - if ((end < BITSET_CONTAINER_SIZE_IN_WORDS) && ((bitset->array[end] & last) != last)){ - - return false; - } - - for (uint16_t i = start + 1; (i < BITSET_CONTAINER_SIZE_IN_WORDS) && (i < end); ++i){ - - if (bitset->array[i] != UINT64_C(0xFFFFFFFFFFFFFFFF)) return false; - } - - return true; -} - -/* Check whether `bitset' is present in `array'. Calls bitset_container_get. */ -inline bool bitset_container_contains(const bitset_container_t *bitset, - uint16_t pos) { - return bitset_container_get(bitset, pos); -} - -/* -* Check whether a range of bits from position `pos_start' (included) to `pos_end' (excluded) -* is present in `bitset'. Calls bitset_container_get_all. -*/ -static inline bool bitset_container_contains_range(const bitset_container_t *bitset, - uint32_t pos_start, uint32_t pos_end) { - return bitset_container_get_range(bitset, pos_start, pos_end); -} - -/* Get the number of bits set */ -static inline int bitset_container_cardinality( - const bitset_container_t *bitset) { - return bitset->cardinality; -} - - - - -/* Copy one container into another. We assume that they are distinct. */ -void bitset_container_copy(const bitset_container_t *source, - bitset_container_t *dest); - -/* Add all the values [min,max) at a distance k*step from min: min, - * min+step,.... */ -void bitset_container_add_from_range(bitset_container_t *bitset, uint32_t min, - uint32_t max, uint16_t step); - -/* Get the number of bits set (force computation). This does not modify bitset. - * To update the cardinality, you should do - * bitset->cardinality = bitset_container_compute_cardinality(bitset).*/ -int bitset_container_compute_cardinality(const bitset_container_t *bitset); - -/* Get whether there is at least one bit set (see bitset_container_empty for the reverse), - when the cardinality is unknown, it is computed and stored in the struct */ -static inline bool bitset_container_nonzero_cardinality( - bitset_container_t *bitset) { - // account for laziness - if (bitset->cardinality == BITSET_UNKNOWN_CARDINALITY) { - // could bail early instead with a nonzero result - bitset->cardinality = bitset_container_compute_cardinality(bitset); - } - return bitset->cardinality > 0; -} - -/* Check whether this bitset is empty (see bitset_container_nonzero_cardinality for the reverse), - * it never modifies the bitset struct. */ -static inline bool bitset_container_empty( - const bitset_container_t *bitset) { - if (bitset->cardinality == BITSET_UNKNOWN_CARDINALITY) { - for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i ++) { - if((bitset->array[i]) != 0) return false; - } - return true; - } - return bitset->cardinality == 0; -} - - -/* Get whether there is at least one bit set (see bitset_container_empty for the reverse), - the bitset is never modified */ -static inline bool bitset_container_const_nonzero_cardinality( - const bitset_container_t *bitset) { - return !bitset_container_empty(bitset); -} - -/* - * Check whether the two bitsets intersect - */ -bool bitset_container_intersect(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the union of bitsets `src_1' and `src_2' into `dst' and return the - * cardinality. */ -int bitset_container_or(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the union of bitsets `src_1' and `src_2' and return the cardinality. - */ -int bitset_container_or_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the union of bitsets `src_1' and `src_2' into `dst' and return the - * cardinality. Same as bitset_container_or. */ -int bitset_container_union(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the union of bitsets `src_1' and `src_2' and return the - * cardinality. Same as bitset_container_or_justcard. */ -int bitset_container_union_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the union of bitsets `src_1' and `src_2' into `dst', but does not - * update the cardinality. Provided to optimize chained operations. */ -int bitset_container_or_nocard(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the intersection of bitsets `src_1' and `src_2' into `dst' and - * return the cardinality. */ -int bitset_container_and(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the intersection of bitsets `src_1' and `src_2' and return the - * cardinality. */ -int bitset_container_and_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the intersection of bitsets `src_1' and `src_2' into `dst' and - * return the cardinality. Same as bitset_container_and. */ -int bitset_container_intersection(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the intersection of bitsets `src_1' and `src_2' and return the - * cardinality. Same as bitset_container_and_justcard. */ -int bitset_container_intersection_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the intersection of bitsets `src_1' and `src_2' into `dst', but does - * not update the cardinality. Provided to optimize chained operations. */ -int bitset_container_and_nocard(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the exclusive or of bitsets `src_1' and `src_2' into `dst' and - * return the cardinality. */ -int bitset_container_xor(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the exclusive or of bitsets `src_1' and `src_2' and return the - * cardinality. */ -int bitset_container_xor_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the exclusive or of bitsets `src_1' and `src_2' into `dst', but does - * not update the cardinality. Provided to optimize chained operations. */ -int bitset_container_xor_nocard(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the and not of bitsets `src_1' and `src_2' into `dst' and return the - * cardinality. */ -int bitset_container_andnot(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Computes the and not of bitsets `src_1' and `src_2' and return the - * cardinality. */ -int bitset_container_andnot_justcard(const bitset_container_t *src_1, - const bitset_container_t *src_2); - -/* Computes the and not or of bitsets `src_1' and `src_2' into `dst', but does - * not update the cardinality. Provided to optimize chained operations. */ -int bitset_container_andnot_nocard(const bitset_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* - * Write out the 16-bit integers contained in this container as a list of 32-bit - * integers using base - * as the starting value (it might be expected that base has zeros in its 16 - * least significant bits). - * The function returns the number of values written. - * The caller is responsible for allocating enough memory in out. - * The out pointer should point to enough memory (the cardinality times 32 - * bits). - */ -int bitset_container_to_uint32_array(void *out, const bitset_container_t *cont, - uint32_t base); - -/* - * Print this container using printf (useful for debugging). - */ -void bitset_container_printf(const bitset_container_t *v); - -/* - * Print this container using printf as a comma-separated list of 32-bit - * integers starting at base. - */ -void bitset_container_printf_as_uint32_array(const bitset_container_t *v, - uint32_t base); - -/** - * Return the serialized size in bytes of a container. - */ -static inline int32_t bitset_container_serialized_size_in_bytes(void) { - return BITSET_CONTAINER_SIZE_IN_WORDS * 8; -} - -/** - * Return the the number of runs. - */ -int bitset_container_number_of_runs(bitset_container_t *b); - -bool bitset_container_iterate(const bitset_container_t *cont, uint32_t base, - roaring_iterator iterator, void *ptr); -bool bitset_container_iterate64(const bitset_container_t *cont, uint32_t base, - roaring_iterator64 iterator, uint64_t high_bits, - void *ptr); - -/** - * Writes the underlying array to buf, outputs how many bytes were written. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes written should be - * bitset_container_size_in_bytes(container). - */ -int32_t bitset_container_write(const bitset_container_t *container, char *buf); - -/** - * Reads the instance from buf, outputs how many bytes were read. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes read should be bitset_container_size_in_bytes(container). - * You need to provide the (known) cardinality. - */ -int32_t bitset_container_read(int32_t cardinality, - bitset_container_t *container, const char *buf); -/** - * Return the serialized size in bytes of a container (see - * bitset_container_write). - * This is meant to be compatible with the Java and Go versions of Roaring and - * assumes - * that the cardinality of the container is already known or can be computed. - */ -static inline int32_t bitset_container_size_in_bytes( - const bitset_container_t *container) { - (void)container; - return BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); -} - -/** - * Return true if the two containers have the same content. - */ -bool bitset_container_equals(const bitset_container_t *container1, - const bitset_container_t *container2); - -/** -* Return true if container1 is a subset of container2. -*/ -bool bitset_container_is_subset(const bitset_container_t *container1, - const bitset_container_t *container2); - -/** - * If the element of given rank is in this container, supposing that the first - * element has rank start_rank, then the function returns true and sets element - * accordingly. - * Otherwise, it returns false and update start_rank. - */ -bool bitset_container_select(const bitset_container_t *container, - uint32_t *start_rank, uint32_t rank, - uint32_t *element); - -/* Returns the smallest value (assumes not empty) */ -uint16_t bitset_container_minimum(const bitset_container_t *container); - -/* Returns the largest value (assumes not empty) */ -uint16_t bitset_container_maximum(const bitset_container_t *container); - -/* Returns the number of values equal or smaller than x */ -int bitset_container_rank(const bitset_container_t *container, uint16_t x); - -/* Returns the index of the first value equal or larger than x, or -1 */ -int bitset_container_index_equalorlarger(const bitset_container_t *container, uint16_t x); -#endif /* INCLUDE_CONTAINERS_BITSET_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/bitset.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/run.h */ -/* - * run.h - * - */ - -#ifndef INCLUDE_CONTAINERS_RUN_H_ -#define INCLUDE_CONTAINERS_RUN_H_ - -#ifdef __cplusplus -extern "C" { -#endif - -#include -#include -#include -#include - - -/* struct rle16_s - run length pair - * - * @value: start position of the run - * @length: length of the run is `length + 1` - * - * An RLE pair {v, l} would represent the integers between the interval - * [v, v+l+1], e.g. {3, 2} = [3, 4, 5]. - */ -struct rle16_s { - uint16_t value; - uint16_t length; -}; - -typedef struct rle16_s rle16_t; - -/* struct run_container_s - run container bitmap - * - * @n_runs: number of rle_t pairs in `runs`. - * @capacity: capacity in rle_t pairs `runs` can hold. - * @runs: pairs of rle_t. - * - */ -struct run_container_s { - int32_t n_runs; - int32_t capacity; - rle16_t *runs; -}; - -typedef struct run_container_s run_container_t; - -/* Create a new run container. Return NULL in case of failure. */ -run_container_t *run_container_create(void); - -/* Create a new run container with given capacity. Return NULL in case of - * failure. */ -run_container_t *run_container_create_given_capacity(int32_t size); - -/* - * Shrink the capacity to the actual size, return the number of bytes saved. - */ -int run_container_shrink_to_fit(run_container_t *src); - -/* Free memory owned by `run'. */ -void run_container_free(run_container_t *run); - -/* Duplicate container */ -run_container_t *run_container_clone(const run_container_t *src); - -int32_t run_container_serialize(const run_container_t *container, - char *buf) WARN_UNUSED; - -uint32_t run_container_serialization_len(const run_container_t *container); - -void *run_container_deserialize(const char *buf, size_t buf_len); - -/* - * Effectively deletes the value at index index, repacking data. - */ -static inline void recoverRoomAtIndex(run_container_t *run, uint16_t index) { - memmove(run->runs + index, run->runs + (1 + index), - (run->n_runs - index - 1) * sizeof(rle16_t)); - run->n_runs--; -} - -/** - * Good old binary search through rle data - */ -inline int32_t interleavedBinarySearch(const rle16_t *array, int32_t lenarray, - uint16_t ikey) { - int32_t low = 0; - int32_t high = lenarray - 1; - while (low <= high) { - int32_t middleIndex = (low + high) >> 1; - uint16_t middleValue = array[middleIndex].value; - if (middleValue < ikey) { - low = middleIndex + 1; - } else if (middleValue > ikey) { - high = middleIndex - 1; - } else { - return middleIndex; - } - } - return -(low + 1); -} - -/* - * Returns index of the run which contains $ikey - */ -static inline int32_t rle16_find_run(const rle16_t *array, int32_t lenarray, - uint16_t ikey) { - int32_t low = 0; - int32_t high = lenarray - 1; - while (low <= high) { - int32_t middleIndex = (low + high) >> 1; - uint16_t min = array[middleIndex].value; - uint16_t max = array[middleIndex].value + array[middleIndex].length; - if (ikey > max) { - low = middleIndex + 1; - } else if (ikey < min) { - high = middleIndex - 1; - } else { - return middleIndex; - } - } - return -(low + 1); -} - - -/** - * Returns number of runs which can'be be merged with the key because they - * are less than the key. - * Note that [5,6,7,8] can be merged with the key 9 and won't be counted. - */ -static inline int32_t rle16_count_less(const rle16_t* array, int32_t lenarray, - uint16_t key) { - if (lenarray == 0) return 0; - int32_t low = 0; - int32_t high = lenarray - 1; - while (low <= high) { - int32_t middleIndex = (low + high) >> 1; - uint16_t min_value = array[middleIndex].value; - uint16_t max_value = array[middleIndex].value + array[middleIndex].length; - if (max_value + UINT32_C(1) < key) { // uint32 arithmetic - low = middleIndex + 1; - } else if (key < min_value) { - high = middleIndex - 1; - } else { - return middleIndex; - } - } - return low; -} - -static inline int32_t rle16_count_greater(const rle16_t* array, int32_t lenarray, - uint16_t key) { - if (lenarray == 0) return 0; - int32_t low = 0; - int32_t high = lenarray - 1; - while (low <= high) { - int32_t middleIndex = (low + high) >> 1; - uint16_t min_value = array[middleIndex].value; - uint16_t max_value = array[middleIndex].value + array[middleIndex].length; - if (max_value < key) { - low = middleIndex + 1; - } else if (key + UINT32_C(1) < min_value) { // uint32 arithmetic - high = middleIndex - 1; - } else { - return lenarray - (middleIndex + 1); - } - } - return lenarray - low; -} - -/** - * increase capacity to at least min. Whether the - * existing data needs to be copied over depends on copy. If "copy" is false, - * then the new content will be uninitialized, otherwise a copy is made. - */ -void run_container_grow(run_container_t *run, int32_t min, bool copy); - -/** - * Moves the data so that we can write data at index - */ -static inline void makeRoomAtIndex(run_container_t *run, uint16_t index) { - /* This function calls realloc + memmove sequentially to move by one index. - * Potentially copying twice the array. - */ - if (run->n_runs + 1 > run->capacity) - run_container_grow(run, run->n_runs + 1, true); - memmove(run->runs + 1 + index, run->runs + index, - (run->n_runs - index) * sizeof(rle16_t)); - run->n_runs++; -} - -/* Add `pos' to `run'. Returns true if `pos' was not present. */ -bool run_container_add(run_container_t *run, uint16_t pos); - -/* Remove `pos' from `run'. Returns true if `pos' was present. */ -static inline bool run_container_remove(run_container_t *run, uint16_t pos) { - int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); - if (index >= 0) { - int32_t le = run->runs[index].length; - if (le == 0) { - recoverRoomAtIndex(run, (uint16_t)index); - } else { - run->runs[index].value++; - run->runs[index].length--; - } - return true; - } - index = -index - 2; // points to preceding value, possibly -1 - if (index >= 0) { // possible match - int32_t offset = pos - run->runs[index].value; - int32_t le = run->runs[index].length; - if (offset < le) { - // need to break in two - run->runs[index].length = (uint16_t)(offset - 1); - // need to insert - uint16_t newvalue = pos + 1; - int32_t newlength = le - offset - 1; - makeRoomAtIndex(run, (uint16_t)(index + 1)); - run->runs[index + 1].value = newvalue; - run->runs[index + 1].length = (uint16_t)newlength; - return true; - - } else if (offset == le) { - run->runs[index].length--; - return true; - } - } - // no match - return false; -} - -/* Check whether `pos' is present in `run'. */ -inline bool run_container_contains(const run_container_t *run, uint16_t pos) { - int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); - if (index >= 0) return true; - index = -index - 2; // points to preceding value, possibly -1 - if (index != -1) { // possible match - int32_t offset = pos - run->runs[index].value; - int32_t le = run->runs[index].length; - if (offset <= le) return true; - } - return false; -} - -/* -* Check whether all positions in a range of positions from pos_start (included) -* to pos_end (excluded) is present in `run'. -*/ -static inline bool run_container_contains_range(const run_container_t *run, - uint32_t pos_start, uint32_t pos_end) { - uint32_t count = 0; - int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos_start); - if (index < 0) { - index = -index - 2; - if ((index == -1) || ((pos_start - run->runs[index].value) > run->runs[index].length)){ - return false; - } - } - for (int32_t i = index; i < run->n_runs; ++i) { - const uint32_t stop = run->runs[i].value + run->runs[i].length; - if (run->runs[i].value >= pos_end) break; - if (stop >= pos_end) { - count += (((pos_end - run->runs[i].value) > 0) ? (pos_end - run->runs[i].value) : 0); - break; - } - const uint32_t min = (stop - pos_start) > 0 ? (stop - pos_start) : 0; - count += (min < run->runs[i].length) ? min : run->runs[i].length; - } - return count >= (pos_end - pos_start - 1); -} - -#ifdef USEAVX - -/* Get the cardinality of `run'. Requires an actual computation. */ -static inline int run_container_cardinality(const run_container_t *run) { - const int32_t n_runs = run->n_runs; - const rle16_t *runs = run->runs; - - /* by initializing with n_runs, we omit counting the +1 for each pair. */ - int sum = n_runs; - int32_t k = 0; - const int32_t step = sizeof(__m256i) / sizeof(rle16_t); - if (n_runs > step) { - __m256i total = _mm256_setzero_si256(); - for (; k + step <= n_runs; k += step) { - __m256i ymm1 = _mm256_lddqu_si256((const __m256i *)(runs + k)); - __m256i justlengths = _mm256_srli_epi32(ymm1, 16); - total = _mm256_add_epi32(total, justlengths); - } - // a store might be faster than extract? - uint32_t buffer[sizeof(__m256i) / sizeof(rle16_t)]; - _mm256_storeu_si256((__m256i *)buffer, total); - sum += (buffer[0] + buffer[1]) + (buffer[2] + buffer[3]) + - (buffer[4] + buffer[5]) + (buffer[6] + buffer[7]); - } - for (; k < n_runs; ++k) { - sum += runs[k].length; - } - - return sum; -} - -#else - -/* Get the cardinality of `run'. Requires an actual computation. */ -static inline int run_container_cardinality(const run_container_t *run) { - const int32_t n_runs = run->n_runs; - const rle16_t *runs = run->runs; - - /* by initializing with n_runs, we omit counting the +1 for each pair. */ - int sum = n_runs; - for (int k = 0; k < n_runs; ++k) { - sum += runs[k].length; - } - - return sum; -} -#endif - -/* Card > 0?, see run_container_empty for the reverse */ -static inline bool run_container_nonzero_cardinality( - const run_container_t *run) { - return run->n_runs > 0; // runs never empty -} - -/* Card == 0?, see run_container_nonzero_cardinality for the reverse */ -static inline bool run_container_empty( - const run_container_t *run) { - return run->n_runs == 0; // runs never empty -} - - - -/* Copy one container into another. We assume that they are distinct. */ -void run_container_copy(const run_container_t *src, run_container_t *dst); - -/* Set the cardinality to zero (does not release memory). */ -static inline void run_container_clear(run_container_t *run) { - run->n_runs = 0; -} - -/** - * Append run described by vl to the run container, possibly merging. - * It is assumed that the run would be inserted at the end of the container, no - * check is made. - * It is assumed that the run container has the necessary capacity: caller is - * responsible for checking memory capacity. - * - * - * This is not a safe function, it is meant for performance: use with care. - */ -static inline void run_container_append(run_container_t *run, rle16_t vl, - rle16_t *previousrl) { - const uint32_t previousend = previousrl->value + previousrl->length; - if (vl.value > previousend + 1) { // we add a new one - run->runs[run->n_runs] = vl; - run->n_runs++; - *previousrl = vl; - } else { - uint32_t newend = vl.value + vl.length + UINT32_C(1); - if (newend > previousend) { // we merge - previousrl->length = (uint16_t)(newend - 1 - previousrl->value); - run->runs[run->n_runs - 1] = *previousrl; - } - } -} - -/** - * Like run_container_append but it is assumed that the content of run is empty. - */ -static inline rle16_t run_container_append_first(run_container_t *run, - rle16_t vl) { - run->runs[run->n_runs] = vl; - run->n_runs++; - return vl; -} - -/** - * append a single value given by val to the run container, possibly merging. - * It is assumed that the value would be inserted at the end of the container, - * no check is made. - * It is assumed that the run container has the necessary capacity: caller is - * responsible for checking memory capacity. - * - * This is not a safe function, it is meant for performance: use with care. - */ -static inline void run_container_append_value(run_container_t *run, - uint16_t val, - rle16_t *previousrl) { - const uint32_t previousend = previousrl->value + previousrl->length; - if (val > previousend + 1) { // we add a new one - //*previousrl = (rle16_t){.value = val, .length = 0};// requires C99 - previousrl->value = val; - previousrl->length = 0; - - run->runs[run->n_runs] = *previousrl; - run->n_runs++; - } else if (val == previousend + 1) { // we merge - previousrl->length++; - run->runs[run->n_runs - 1] = *previousrl; - } -} - -/** - * Like run_container_append_value but it is assumed that the content of run is - * empty. - */ -static inline rle16_t run_container_append_value_first(run_container_t *run, - uint16_t val) { - // rle16_t newrle = (rle16_t){.value = val, .length = 0};// requires C99 - rle16_t newrle; - newrle.value = val; - newrle.length = 0; - - run->runs[run->n_runs] = newrle; - run->n_runs++; - return newrle; -} - -/* Check whether the container spans the whole chunk (cardinality = 1<<16). - * This check can be done in constant time (inexpensive). */ -static inline bool run_container_is_full(const run_container_t *run) { - rle16_t vl = run->runs[0]; - return (run->n_runs == 1) && (vl.value == 0) && (vl.length == 0xFFFF); -} - -/* Compute the union of `src_1' and `src_2' and write the result to `dst' - * It is assumed that `dst' is distinct from both `src_1' and `src_2'. */ -void run_container_union(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst); - -/* Compute the union of `src_1' and `src_2' and write the result to `src_1' */ -void run_container_union_inplace(run_container_t *src_1, - const run_container_t *src_2); - -/* Compute the intersection of src_1 and src_2 and write the result to - * dst. It is assumed that dst is distinct from both src_1 and src_2. */ -void run_container_intersection(const run_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst); - -/* Compute the size of the intersection of src_1 and src_2 . */ -int run_container_intersection_cardinality(const run_container_t *src_1, - const run_container_t *src_2); - -/* Check whether src_1 and src_2 intersect. */ -bool run_container_intersect(const run_container_t *src_1, - const run_container_t *src_2); - -/* Compute the symmetric difference of `src_1' and `src_2' and write the result - * to `dst' - * It is assumed that `dst' is distinct from both `src_1' and `src_2'. */ -void run_container_xor(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst); - -/* - * Write out the 16-bit integers contained in this container as a list of 32-bit - * integers using base - * as the starting value (it might be expected that base has zeros in its 16 - * least significant bits). - * The function returns the number of values written. - * The caller is responsible for allocating enough memory in out. - */ -int run_container_to_uint32_array(void *vout, const run_container_t *cont, - uint32_t base); - -/* - * Print this container using printf (useful for debugging). - */ -void run_container_printf(const run_container_t *v); - -/* - * Print this container using printf as a comma-separated list of 32-bit - * integers starting at base. - */ -void run_container_printf_as_uint32_array(const run_container_t *v, - uint32_t base); - -/** - * Return the serialized size in bytes of a container having "num_runs" runs. - */ -static inline int32_t run_container_serialized_size_in_bytes(int32_t num_runs) { - return sizeof(uint16_t) + - sizeof(rle16_t) * num_runs; // each run requires 2 2-byte entries. -} - -bool run_container_iterate(const run_container_t *cont, uint32_t base, - roaring_iterator iterator, void *ptr); -bool run_container_iterate64(const run_container_t *cont, uint32_t base, - roaring_iterator64 iterator, uint64_t high_bits, - void *ptr); - -/** - * Writes the underlying array to buf, outputs how many bytes were written. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes written should be run_container_size_in_bytes(container). - */ -int32_t run_container_write(const run_container_t *container, char *buf); - -/** - * Reads the instance from buf, outputs how many bytes were read. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes read should be bitset_container_size_in_bytes(container). - * The cardinality parameter is provided for consistency with other containers, - * but - * it might be effectively ignored.. - */ -int32_t run_container_read(int32_t cardinality, run_container_t *container, - const char *buf); - -/** - * Return the serialized size in bytes of a container (see run_container_write). - * This is meant to be compatible with the Java and Go versions of Roaring. - */ -static inline int32_t run_container_size_in_bytes( - const run_container_t *container) { - return run_container_serialized_size_in_bytes(container->n_runs); -} - -/** - * Return true if the two containers have the same content. - */ -bool run_container_equals(const run_container_t *container1, - const run_container_t *container2); - -/** -* Return true if container1 is a subset of container2. -*/ -bool run_container_is_subset(const run_container_t *container1, - const run_container_t *container2); - -/** - * Used in a start-finish scan that appends segments, for XOR and NOT - */ - -void run_container_smart_append_exclusive(run_container_t *src, - const uint16_t start, - const uint16_t length); - -/** -* The new container consists of a single run [start,stop). -* It is required that stop>start, the caller is responsability for this check. -* It is required that stop <= (1<<16), the caller is responsability for this check. -* The cardinality of the created container is stop - start. -* Returns NULL on failure -*/ -static inline run_container_t *run_container_create_range(uint32_t start, - uint32_t stop) { - run_container_t *rc = run_container_create_given_capacity(1); - if (rc) { - rle16_t r; - r.value = (uint16_t)start; - r.length = (uint16_t)(stop - start - 1); - run_container_append_first(rc, r); - } - return rc; -} - -/** - * If the element of given rank is in this container, supposing that the first - * element has rank start_rank, then the function returns true and sets element - * accordingly. - * Otherwise, it returns false and update start_rank. - */ -bool run_container_select(const run_container_t *container, - uint32_t *start_rank, uint32_t rank, - uint32_t *element); - -/* Compute the difference of src_1 and src_2 and write the result to - * dst. It is assumed that dst is distinct from both src_1 and src_2. */ - -void run_container_andnot(const run_container_t *src_1, - const run_container_t *src_2, run_container_t *dst); - -/* Returns the smallest value (assumes not empty) */ -inline uint16_t run_container_minimum(const run_container_t *run) { - if (run->n_runs == 0) return 0; - return run->runs[0].value; -} - -/* Returns the largest value (assumes not empty) */ -inline uint16_t run_container_maximum(const run_container_t *run) { - if (run->n_runs == 0) return 0; - return run->runs[run->n_runs - 1].value + run->runs[run->n_runs - 1].length; -} - -/* Returns the number of values equal or smaller than x */ -int run_container_rank(const run_container_t *arr, uint16_t x); - -/* Returns the index of the first run containing a value at least as large as x, or -1 */ -inline int run_container_index_equalorlarger(const run_container_t *arr, uint16_t x) { - int32_t index = interleavedBinarySearch(arr->runs, arr->n_runs, x); - if (index >= 0) return index; - index = -index - 2; // points to preceding run, possibly -1 - if (index != -1) { // possible match - int32_t offset = x - arr->runs[index].value; - int32_t le = arr->runs[index].length; - if (offset <= le) return index; - } - index += 1; - if(index < arr->n_runs) { - return index; - } - return -1; -} - -/* - * Add all values in range [min, max] using hint. - */ -static inline void run_container_add_range_nruns(run_container_t* run, - uint32_t min, uint32_t max, - int32_t nruns_less, - int32_t nruns_greater) { - int32_t nruns_common = run->n_runs - nruns_less - nruns_greater; - if (nruns_common == 0) { - makeRoomAtIndex(run, nruns_less); - run->runs[nruns_less].value = min; - run->runs[nruns_less].length = max - min; - } else { - uint32_t common_min = run->runs[nruns_less].value; - uint32_t common_max = run->runs[nruns_less + nruns_common - 1].value + - run->runs[nruns_less + nruns_common - 1].length; - uint32_t result_min = (common_min < min) ? common_min : min; - uint32_t result_max = (common_max > max) ? common_max : max; - - run->runs[nruns_less].value = result_min; - run->runs[nruns_less].length = result_max - result_min; - - memmove(&(run->runs[nruns_less + 1]), - &(run->runs[run->n_runs - nruns_greater]), - nruns_greater*sizeof(rle16_t)); - run->n_runs = nruns_less + 1 + nruns_greater; - } -} - -/** - * Add all values in range [min, max] - */ -static inline void run_container_add_range(run_container_t* run, - uint32_t min, uint32_t max) { - int32_t nruns_greater = rle16_count_greater(run->runs, run->n_runs, max); - int32_t nruns_less = rle16_count_less(run->runs, run->n_runs - nruns_greater, min); - run_container_add_range_nruns(run, min, max, nruns_less, nruns_greater); -} - -/** - * Shifts last $count elements either left (distance < 0) or right (distance > 0) - */ -static inline void run_container_shift_tail(run_container_t* run, - int32_t count, int32_t distance) { - if (distance > 0) { - if (run->capacity < count+distance) { - run_container_grow(run, count+distance, true); - } - } - int32_t srcpos = run->n_runs - count; - int32_t dstpos = srcpos + distance; - memmove(&(run->runs[dstpos]), &(run->runs[srcpos]), sizeof(rle16_t) * count); - run->n_runs += distance; -} - -/** - * Remove all elements in range [min, max] - */ -static inline void run_container_remove_range(run_container_t *run, uint32_t min, uint32_t max) { - int32_t first = rle16_find_run(run->runs, run->n_runs, min); - int32_t last = rle16_find_run(run->runs, run->n_runs, max); - - if (first >= 0 && min > run->runs[first].value && - max < run->runs[first].value + run->runs[first].length) { - // split this run into two adjacent runs - - // right subinterval - makeRoomAtIndex(run, first+1); - run->runs[first+1].value = max + 1; - run->runs[first+1].length = (run->runs[first].value + run->runs[first].length) - (max + 1); - - // left subinterval - run->runs[first].length = (min - 1) - run->runs[first].value; - - return; - } - - // update left-most partial run - if (first >= 0) { - if (min > run->runs[first].value) { - run->runs[first].length = (min - 1) - run->runs[first].value; - first++; - } - } else { - first = -first-1; - } - - // update right-most run - if (last >= 0) { - uint16_t run_max = run->runs[last].value + run->runs[last].length; - if (run_max > max) { - run->runs[last].value = max + 1; - run->runs[last].length = run_max - (max + 1); - last--; - } - } else { - last = (-last-1) - 1; - } - - // remove intermediate runs - if (first <= last) { - run_container_shift_tail(run, run->n_runs - (last+1), -(last-first+1)); - } -} - -#ifdef __cplusplus -} -#endif - -#endif /* INCLUDE_CONTAINERS_RUN_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/run.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/convert.h */ -/* - * convert.h - * - */ - -#ifndef INCLUDE_CONTAINERS_CONVERT_H_ -#define INCLUDE_CONTAINERS_CONVERT_H_ - -#ifdef __cplusplus -extern "C" { -#endif - -/* Convert an array into a bitset. The input container is not freed or modified. - */ -bitset_container_t *bitset_container_from_array(const array_container_t *arr); - -/* Convert a run into a bitset. The input container is not freed or modified. */ -bitset_container_t *bitset_container_from_run(const run_container_t *arr); - -/* Convert a run into an array. The input container is not freed or modified. */ -array_container_t *array_container_from_run(const run_container_t *arr); - -/* Convert a bitset into an array. The input container is not freed or modified. - */ -array_container_t *array_container_from_bitset(const bitset_container_t *bits); - -/* Convert an array into a run. The input container is not freed or modified. - */ -run_container_t *run_container_from_array(const array_container_t *c); - -/* convert a run into either an array or a bitset - * might free the container */ -void *convert_to_bitset_or_array_container(run_container_t *r, int32_t card, - uint8_t *resulttype); - -/* convert containers to and from runcontainers, as is most space efficient. - * The container might be freed. */ -void *convert_run_optimize(void *c, uint8_t typecode_original, - uint8_t *typecode_after); - -/* converts a run container to either an array or a bitset, IF it saves space. - */ -/* If a conversion occurs, the caller is responsible to free the original - * container and - * he becomes reponsible to free the new one. */ -void *convert_run_to_efficient_container(run_container_t *c, - uint8_t *typecode_after); -// like convert_run_to_efficient_container but frees the old result if needed -void *convert_run_to_efficient_container_and_free(run_container_t *c, - uint8_t *typecode_after); - -/** - * Create new bitset container which is a union of run container and - * range [min, max]. Caller is responsible for freeing run container. - */ -bitset_container_t *bitset_container_from_run_range(const run_container_t *run, - uint32_t min, uint32_t max); - - -#ifdef __cplusplus -} -#endif - -#endif /* INCLUDE_CONTAINERS_CONVERT_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/convert.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_equal.h */ -/* - * mixed_equal.h - * - */ - -#ifndef CONTAINERS_MIXED_EQUAL_H_ -#define CONTAINERS_MIXED_EQUAL_H_ - - -/** - * Return true if the two containers have the same content. - */ -bool array_container_equal_bitset(const array_container_t* container1, - const bitset_container_t* container2); - -/** - * Return true if the two containers have the same content. - */ -bool run_container_equals_array(const run_container_t* container1, - const array_container_t* container2); -/** - * Return true if the two containers have the same content. - */ -bool run_container_equals_bitset(const run_container_t* container1, - const bitset_container_t* container2); - -#endif /* CONTAINERS_MIXED_EQUAL_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_equal.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_subset.h */ -/* - * mixed_subset.h - * - */ - -#ifndef CONTAINERS_MIXED_SUBSET_H_ -#define CONTAINERS_MIXED_SUBSET_H_ - - -/** - * Return true if container1 is a subset of container2. - */ -bool array_container_is_subset_bitset(const array_container_t* container1, - const bitset_container_t* container2); - -/** -* Return true if container1 is a subset of container2. - */ -bool run_container_is_subset_array(const run_container_t* container1, - const array_container_t* container2); - -/** -* Return true if container1 is a subset of container2. - */ -bool array_container_is_subset_run(const array_container_t* container1, - const run_container_t* container2); - -/** -* Return true if container1 is a subset of container2. - */ -bool run_container_is_subset_bitset(const run_container_t* container1, - const bitset_container_t* container2); - -/** -* Return true if container1 is a subset of container2. -*/ -bool bitset_container_is_subset_run(const bitset_container_t* container1, - const run_container_t* container2); - -#endif /* CONTAINERS_MIXED_SUBSET_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_subset.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_andnot.h */ -/* - * mixed_andnot.h - */ -#ifndef INCLUDE_CONTAINERS_MIXED_ANDNOT_H_ -#define INCLUDE_CONTAINERS_MIXED_ANDNOT_H_ - - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst, a valid array container that could be the same as dst.*/ -void array_bitset_container_andnot(const array_container_t *src_1, - const bitset_container_t *src_2, - array_container_t *dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * src_1 */ - -void array_bitset_container_iandnot(array_container_t *src_1, - const bitset_container_t *src_2); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst, which does not initially have a valid container. - * Return true for a bitset result; false for array - */ - -bool bitset_array_container_andnot(const bitset_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_array_container_iandnot(bitset_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_andnot(const run_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_iandnot(run_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool bitset_run_container_andnot(const bitset_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_run_container_iandnot(bitset_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* dst does not indicate a valid container initially. Eventually it - * can become any type of container. - */ - -int run_array_container_andnot(const run_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -int run_array_container_iandnot(run_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* dst must be a valid array container, allowed to be src_1 */ - -void array_run_container_andnot(const array_container_t *src_1, - const run_container_t *src_2, - array_container_t *dst); - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -void array_run_container_iandnot(array_container_t *src_1, - const run_container_t *src_2); - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int run_run_container_andnot(const run_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -int run_run_container_iandnot(run_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* - * dst is a valid array container and may be the same as src_1 - */ - -void array_array_container_andnot(const array_container_t *src_1, - const array_container_t *src_2, - array_container_t *dst); - -/* inplace array-array andnot will always be able to reuse the space of - * src_1 */ -void array_array_container_iandnot(array_container_t *src_1, - const array_container_t *src_2); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). Return value is - * "dst is a bitset" - */ - -bool bitset_bitset_container_andnot(const bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst); - -/* Compute the andnot of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_bitset_container_iandnot(bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst); -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_andnot.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_intersection.h */ -/* - * mixed_intersection.h - * - */ - -#ifndef INCLUDE_CONTAINERS_MIXED_INTERSECTION_H_ -#define INCLUDE_CONTAINERS_MIXED_INTERSECTION_H_ - -/* These functions appear to exclude cases where the - * inputs have the same type and the output is guaranteed - * to have the same type as the inputs. Eg, array intersection - */ - - -/* Compute the intersection of src_1 and src_2 and write the result to - * dst. It is allowed for dst to be equal to src_1. We assume that dst is a - * valid container. */ -void array_bitset_container_intersection(const array_container_t *src_1, - const bitset_container_t *src_2, - array_container_t *dst); - -/* Compute the size of the intersection of src_1 and src_2. */ -int array_bitset_container_intersection_cardinality( - const array_container_t *src_1, const bitset_container_t *src_2); - - - -/* Checking whether src_1 and src_2 intersect. */ -bool array_bitset_container_intersect(const array_container_t *src_1, - const bitset_container_t *src_2); - -/* - * Compute the intersection between src_1 and src_2 and write the result - * to *dst. If the return function is true, the result is a bitset_container_t - * otherwise is a array_container_t. We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool bitset_bitset_container_intersection(const bitset_container_t *src_1, - const bitset_container_t *src_2, - void **dst); - -/* Compute the intersection between src_1 and src_2 and write the result to - * dst. It is allowed for dst to be equal to src_1. We assume that dst is a - * valid container. */ -void array_run_container_intersection(const array_container_t *src_1, - const run_container_t *src_2, - array_container_t *dst); - -/* Compute the intersection between src_1 and src_2 and write the result to - * *dst. If the result is true then the result is a bitset_container_t - * otherwise is a array_container_t. - * If *dst == src_2, then an in-place intersection is attempted - **/ -bool run_bitset_container_intersection(const run_container_t *src_1, - const bitset_container_t *src_2, - void **dst); - -/* Compute the size of the intersection between src_1 and src_2 . */ -int array_run_container_intersection_cardinality(const array_container_t *src_1, - const run_container_t *src_2); - -/* Compute the size of the intersection between src_1 and src_2 - **/ -int run_bitset_container_intersection_cardinality(const run_container_t *src_1, - const bitset_container_t *src_2); - - -/* Check that src_1 and src_2 intersect. */ -bool array_run_container_intersect(const array_container_t *src_1, - const run_container_t *src_2); - -/* Check that src_1 and src_2 intersect. - **/ -bool run_bitset_container_intersect(const run_container_t *src_1, - const bitset_container_t *src_2); - -/* - * Same as bitset_bitset_container_intersection except that if the output is to - * be a - * bitset_container_t, then src_1 is modified and no allocation is made. - * If the output is to be an array_container_t, then caller is responsible - * to free the container. - * In all cases, the result is in *dst. - */ -bool bitset_bitset_container_intersection_inplace( - bitset_container_t *src_1, const bitset_container_t *src_2, void **dst); - -#endif /* INCLUDE_CONTAINERS_MIXED_INTERSECTION_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_intersection.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_negation.h */ -/* - * mixed_negation.h - * - */ - -#ifndef INCLUDE_CONTAINERS_MIXED_NEGATION_H_ -#define INCLUDE_CONTAINERS_MIXED_NEGATION_H_ - - -/* Negation across the entire range of the container. - * Compute the negation of src and write the result - * to *dst. The complement of a - * sufficiently sparse set will always be dense and a hence a bitmap - * We assume that dst is pre-allocated and a valid bitset container - * There can be no in-place version. - */ -void array_container_negation(const array_container_t *src, - bitset_container_t *dst); - -/* Negation across the entire range of the container - * Compute the negation of src and write the result - * to *dst. A true return value indicates a bitset result, - * otherwise the result is an array container. - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool bitset_container_negation(const bitset_container_t *src, void **dst); - -/* inplace version */ -/* - * Same as bitset_container_negation except that if the output is to - * be a - * bitset_container_t, then src is modified and no allocation is made. - * If the output is to be an array_container_t, then caller is responsible - * to free the container. - * In all cases, the result is in *dst. - */ -bool bitset_container_negation_inplace(bitset_container_t *src, void **dst); - -/* Negation across the entire range of container - * Compute the negation of src and write the result - * to *dst. - * Return values are the *_TYPECODES as defined * in containers.h - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -int run_container_negation(const run_container_t *src, void **dst); - -/* - * Same as run_container_negation except that if the output is to - * be a - * run_container_t, and has the capacity to hold the result, - * then src is modified and no allocation is made. - * In all cases, the result is in *dst. - */ -int run_container_negation_inplace(run_container_t *src, void **dst); - -/* Negation across a range of the container. - * Compute the negation of src and write the result - * to *dst. Returns true if the result is a bitset container - * and false for an array container. *dst is not preallocated. - */ -bool array_container_negation_range(const array_container_t *src, - const int range_start, const int range_end, - void **dst); - -/* Even when the result would fit, it is unclear how to make an - * inplace version without inefficient copying. Thus this routine - * may be a wrapper for the non-in-place version - */ -bool array_container_negation_range_inplace(array_container_t *src, - const int range_start, - const int range_end, void **dst); - -/* Negation across a range of the container - * Compute the negation of src and write the result - * to *dst. A true return value indicates a bitset result, - * otherwise the result is an array container. - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool bitset_container_negation_range(const bitset_container_t *src, - const int range_start, const int range_end, - void **dst); - -/* inplace version */ -/* - * Same as bitset_container_negation except that if the output is to - * be a - * bitset_container_t, then src is modified and no allocation is made. - * If the output is to be an array_container_t, then caller is responsible - * to free the container. - * In all cases, the result is in *dst. - */ -bool bitset_container_negation_range_inplace(bitset_container_t *src, - const int range_start, - const int range_end, void **dst); - -/* Negation across a range of container - * Compute the negation of src and write the result - * to *dst. Return values are the *_TYPECODES as defined * in containers.h - * We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -int run_container_negation_range(const run_container_t *src, - const int range_start, const int range_end, - void **dst); - -/* - * Same as run_container_negation except that if the output is to - * be a - * run_container_t, and has the capacity to hold the result, - * then src is modified and no allocation is made. - * In all cases, the result is in *dst. - */ -int run_container_negation_range_inplace(run_container_t *src, - const int range_start, - const int range_end, void **dst); - -#endif /* INCLUDE_CONTAINERS_MIXED_NEGATION_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_negation.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_union.h */ -/* - * mixed_intersection.h - * - */ - -#ifndef INCLUDE_CONTAINERS_MIXED_UNION_H_ -#define INCLUDE_CONTAINERS_MIXED_UNION_H_ - -/* These functions appear to exclude cases where the - * inputs have the same type and the output is guaranteed - * to have the same type as the inputs. Eg, bitset unions - */ - - -/* Compute the union of src_1 and src_2 and write the result to - * dst. It is allowed for src_2 to be dst. */ -void array_bitset_container_union(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Compute the union of src_1 and src_2 and write the result to - * dst. It is allowed for src_2 to be dst. This version does not - * update the cardinality of dst (it is set to BITSET_UNKNOWN_CARDINALITY). */ -void array_bitset_container_lazy_union(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* - * Compute the union between src_1 and src_2 and write the result - * to *dst. If the return function is true, the result is a bitset_container_t - * otherwise is a array_container_t. We assume that dst is not pre-allocated. In - * case of failure, *dst will be NULL. - */ -bool array_array_container_union(const array_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* - * Compute the union between src_1 and src_2 and write the result - * to *dst if it cannot be written to src_1. If the return function is true, - * the result is a bitset_container_t - * otherwise is a array_container_t. When the result is an array_container_t, it - * it either written to src_1 (if *dst is null) or to *dst. - * If the result is a bitset_container_t and *dst is null, then there was a failure. - */ -bool array_array_container_inplace_union(array_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* - * Same as array_array_container_union except that it will more eagerly produce - * a bitset. - */ -bool array_array_container_lazy_union(const array_container_t *src_1, - const array_container_t *src_2, - void **dst); - -/* - * Same as array_array_container_inplace_union except that it will more eagerly produce - * a bitset. - */ -bool array_array_container_lazy_inplace_union(array_container_t *src_1, - const array_container_t *src_2, - void **dst); - -/* Compute the union of src_1 and src_2 and write the result to - * dst. We assume that dst is a - * valid container. The result might need to be further converted to array or - * bitset container, - * the caller is responsible for the eventual conversion. */ -void array_run_container_union(const array_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst); - -/* Compute the union of src_1 and src_2 and write the result to - * src2. The result might need to be further converted to array or - * bitset container, - * the caller is responsible for the eventual conversion. */ -void array_run_container_inplace_union(const array_container_t *src_1, - run_container_t *src_2); - -/* Compute the union of src_1 and src_2 and write the result to - * dst. It is allowed for dst to be src_2. - * If run_container_is_full(src_1) is true, you must not be calling this - *function. - **/ -void run_bitset_container_union(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* Compute the union of src_1 and src_2 and write the result to - * dst. It is allowed for dst to be src_2. This version does not - * update the cardinality of dst (it is set to BITSET_UNKNOWN_CARDINALITY). - * If run_container_is_full(src_1) is true, you must not be calling this - * function. - * */ -void run_bitset_container_lazy_union(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -#endif /* INCLUDE_CONTAINERS_MIXED_UNION_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_union.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_xor.h */ -/* - * mixed_xor.h - * - */ - -#ifndef INCLUDE_CONTAINERS_MIXED_XOR_H_ -#define INCLUDE_CONTAINERS_MIXED_XOR_H_ - -/* These functions appear to exclude cases where the - * inputs have the same type and the output is guaranteed - * to have the same type as the inputs. Eg, bitset unions - */ - -/* - * Java implementation (as of May 2016) for array_run, run_run - * and bitset_run don't do anything different for inplace. - * (They are not truly in place.) - */ - - - -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). - * Result is true iff dst is a bitset */ -bool array_bitset_container_xor(const array_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. It is allowed for src_2 to be dst. This version does not - * update the cardinality of dst (it is set to BITSET_UNKNOWN_CARDINALITY). - */ - -void array_bitset_container_lazy_xor(const array_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). Return value is - * "dst is a bitset" - */ - -bool bitset_bitset_container_xor(const bitset_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_xor(const run_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* lazy xor. Dst is initialized and may be equal to src_2. - * Result is left as a bitset container, even if actual - * cardinality would dictate an array container. - */ - -void run_bitset_container_lazy_xor(const run_container_t *src_1, - const bitset_container_t *src_2, - bitset_container_t *dst); - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int array_run_container_xor(const array_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* dst does not initially have a valid container. Creates either - * an array or a bitset container, indicated by return code - */ - -bool array_array_container_xor(const array_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* dst does not initially have a valid container. Creates either - * an array or a bitset container, indicated by return code. - * A bitset container will not have a valid cardinality and the - * container type might not be correct for the actual cardinality - */ - -bool array_array_container_lazy_xor(const array_container_t *src_1, - const array_container_t *src_2, void **dst); - -/* Dst is a valid run container. (Can it be src_2? Let's say not.) - * Leaves result as run container, even if other options are - * smaller. - */ - -void array_run_container_lazy_xor(const array_container_t *src_1, - const run_container_t *src_2, - run_container_t *dst); - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int run_run_container_xor(const run_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* INPLACE versions (initial implementation may not exploit all inplace - * opportunities (if any...) - */ - -/* Compute the xor of src_1 and src_2 and write the result to - * dst (which has no container initially). It will modify src_1 - * to be dst if the result is a bitset. Otherwise, it will - * free src_1 and dst will be a new array container. In both - * cases, the caller is responsible for deallocating dst. - * Returns true iff dst is a bitset */ - -bool bitset_array_container_ixor(bitset_container_t *src_1, - const array_container_t *src_2, void **dst); - -bool bitset_bitset_container_ixor(bitset_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -bool array_bitset_container_ixor(array_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -/* Compute the xor of src_1 and src_2 and write the result to - * dst. Result may be either a bitset or an array container - * (returns "result is bitset"). dst does not initially have - * any container, but becomes either a bitset container (return - * result true) or an array container. - */ - -bool run_bitset_container_ixor(run_container_t *src_1, - const bitset_container_t *src_2, void **dst); - -bool bitset_run_container_ixor(bitset_container_t *src_1, - const run_container_t *src_2, void **dst); - -/* dst does not indicate a valid container initially. Eventually it - * can become any kind of container. - */ - -int array_run_container_ixor(array_container_t *src_1, - const run_container_t *src_2, void **dst); - -int run_array_container_ixor(run_container_t *src_1, - const array_container_t *src_2, void **dst); - -bool array_array_container_ixor(array_container_t *src_1, - const array_container_t *src_2, void **dst); - -int run_run_container_ixor(run_container_t *src_1, const run_container_t *src_2, - void **dst); -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/mixed_xor.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/containers.h */ -#ifndef CONTAINERS_CONTAINERS_H -#define CONTAINERS_CONTAINERS_H - -#ifdef __cplusplus -extern "C" { -#endif - -#include -#include -#include - - -// would enum be possible or better? - -/** - * The switch case statements follow - * BITSET_CONTAINER_TYPE_CODE -- ARRAY_CONTAINER_TYPE_CODE -- - * RUN_CONTAINER_TYPE_CODE - * so it makes more sense to number them 1, 2, 3 (in the vague hope that the - * compiler might exploit this ordering). - */ - -#define BITSET_CONTAINER_TYPE_CODE 1 -#define ARRAY_CONTAINER_TYPE_CODE 2 -#define RUN_CONTAINER_TYPE_CODE 3 -#define SHARED_CONTAINER_TYPE_CODE 4 - -// macro for pairing container type codes -#define CONTAINER_PAIR(c1, c2) (4 * (c1) + (c2)) - -/** - * A shared container is a wrapper around a container - * with reference counting. - */ - -struct shared_container_s { - void *container; - uint8_t typecode; - uint32_t counter; // to be managed atomically -}; - -typedef struct shared_container_s shared_container_t; - -/* - * With copy_on_write = true - * Create a new shared container if the typecode is not SHARED_CONTAINER_TYPE, - * otherwise, increase the count - * If copy_on_write = false, then clone. - * Return NULL in case of failure. - **/ -void *get_copy_of_container(void *container, uint8_t *typecode, - bool copy_on_write); - -/* Frees a shared container (actually decrement its counter and only frees when - * the counter falls to zero). */ -void shared_container_free(shared_container_t *container); - -/* extract a copy from the shared container, freeing the shared container if -there is just one instance left, -clone instances when the counter is higher than one -*/ -void *shared_container_extract_copy(shared_container_t *container, - uint8_t *typecode); - -/* access to container underneath */ -inline const void *container_unwrap_shared( - const void *candidate_shared_container, uint8_t *type) { - if (*type == SHARED_CONTAINER_TYPE_CODE) { - *type = - ((const shared_container_t *)candidate_shared_container)->typecode; - assert(*type != SHARED_CONTAINER_TYPE_CODE); - return ((const shared_container_t *)candidate_shared_container)->container; - } else { - return candidate_shared_container; - } -} - - -/* access to container underneath */ -inline void *container_mutable_unwrap_shared( - void *candidate_shared_container, uint8_t *type) { - if (*type == SHARED_CONTAINER_TYPE_CODE) { - *type = - ((shared_container_t *)candidate_shared_container)->typecode; - assert(*type != SHARED_CONTAINER_TYPE_CODE); - return ((shared_container_t *)candidate_shared_container)->container; - } else { - return candidate_shared_container; - } -} - -/* access to container underneath and queries its type */ -static inline uint8_t get_container_type(const void *container, uint8_t type) { - if (type == SHARED_CONTAINER_TYPE_CODE) { - return ((const shared_container_t *)container)->typecode; - } else { - return type; - } -} - -/** - * Copies a container, requires a typecode. This allocates new memory, caller - * is responsible for deallocation. If the container is not shared, then it is - * physically cloned. Sharable containers are not cloneable. - */ -void *container_clone(const void *container, uint8_t typecode); - -/* access to container underneath, cloning it if needed */ -static inline void *get_writable_copy_if_shared( - void *candidate_shared_container, uint8_t *type) { - if (*type == SHARED_CONTAINER_TYPE_CODE) { - return shared_container_extract_copy( - (shared_container_t *)candidate_shared_container, type); - } else { - return candidate_shared_container; - } -} - -/** - * End of shared container code - */ - -static const char *container_names[] = {"bitset", "array", "run", "shared"}; -static const char *shared_container_names[] = { - "bitset (shared)", "array (shared)", "run (shared)"}; - -// no matter what the initial container was, convert it to a bitset -// if a new container is produced, caller responsible for freeing the previous -// one -// container should not be a shared container -static inline void *container_to_bitset(void *container, uint8_t typecode) { - bitset_container_t *result = NULL; - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return container; // nothing to do - case ARRAY_CONTAINER_TYPE_CODE: - result = - bitset_container_from_array((array_container_t *)container); - return result; - case RUN_CONTAINER_TYPE_CODE: - result = bitset_container_from_run((run_container_t *)container); - return result; - case SHARED_CONTAINER_TYPE_CODE: - assert(false); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * Get the container name from the typecode - */ -static inline const char *get_container_name(uint8_t typecode) { - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return container_names[0]; - case ARRAY_CONTAINER_TYPE_CODE: - return container_names[1]; - case RUN_CONTAINER_TYPE_CODE: - return container_names[2]; - case SHARED_CONTAINER_TYPE_CODE: - return container_names[3]; - default: - assert(false); - __builtin_unreachable(); - return "unknown"; - } -} - -static inline const char *get_full_container_name(const void *container, - uint8_t typecode) { - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return container_names[0]; - case ARRAY_CONTAINER_TYPE_CODE: - return container_names[1]; - case RUN_CONTAINER_TYPE_CODE: - return container_names[2]; - case SHARED_CONTAINER_TYPE_CODE: - switch (((const shared_container_t *)container)->typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return shared_container_names[0]; - case ARRAY_CONTAINER_TYPE_CODE: - return shared_container_names[1]; - case RUN_CONTAINER_TYPE_CODE: - return shared_container_names[2]; - default: - assert(false); - __builtin_unreachable(); - return "unknown"; - } - break; - default: - assert(false); - __builtin_unreachable(); - return "unknown"; - } - __builtin_unreachable(); - return NULL; -} - -/** - * Get the container cardinality (number of elements), requires a typecode - */ -static inline int container_get_cardinality(const void *container, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_cardinality( - (const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_cardinality( - (const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_cardinality( - (const run_container_t *)container); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - - - -// returns true if a container is known to be full. Note that a lazy bitset -// container -// might be full without us knowing -static inline bool container_is_full(const void *container, uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_cardinality( - (const bitset_container_t *)container) == (1 << 16); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_cardinality( - (const array_container_t *)container) == (1 << 16); - case RUN_CONTAINER_TYPE_CODE: - return run_container_is_full((const run_container_t *)container); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -static inline int container_shrink_to_fit(void *container, uint8_t typecode) { - container = container_mutable_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return 0; // no shrinking possible - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_shrink_to_fit( - (array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_shrink_to_fit((run_container_t *)container); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - - -/** - * make a container with a run of ones - */ -/* initially always use a run container, even if an array might be - * marginally - * smaller */ -static inline void *container_range_of_ones(uint32_t range_start, - uint32_t range_end, - uint8_t *result_type) { - assert(range_end >= range_start); - uint64_t cardinality = range_end - range_start + 1; - if(cardinality <= 2) { - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return array_container_create_range(range_start, range_end); - } else { - *result_type = RUN_CONTAINER_TYPE_CODE; - return run_container_create_range(range_start, range_end); - } -} - - -/* Create a container with all the values between in [min,max) at a - distance k*step from min. */ -static inline void *container_from_range(uint8_t *type, uint32_t min, - uint32_t max, uint16_t step) { - if (step == 0) return NULL; // being paranoid - if (step == 1) { - return container_range_of_ones(min,max,type); - // Note: the result is not always a run (need to check the cardinality) - //*type = RUN_CONTAINER_TYPE_CODE; - //return run_container_create_range(min, max); - } - int size = (max - min + step - 1) / step; - if (size <= DEFAULT_MAX_SIZE) { // array container - *type = ARRAY_CONTAINER_TYPE_CODE; - array_container_t *array = array_container_create_given_capacity(size); - array_container_add_from_range(array, min, max, step); - assert(array->cardinality == size); - return array; - } else { // bitset container - *type = BITSET_CONTAINER_TYPE_CODE; - bitset_container_t *bitset = bitset_container_create(); - bitset_container_add_from_range(bitset, min, max, step); - assert(bitset->cardinality == size); - return bitset; - } -} - -/** - * "repair" the container after lazy operations. - */ -static inline void *container_repair_after_lazy(void *container, - uint8_t *typecode) { - container = get_writable_copy_if_shared( - container, typecode); // TODO: this introduces unnecessary cloning - void *result = NULL; - switch (*typecode) { - case BITSET_CONTAINER_TYPE_CODE: - ((bitset_container_t *)container)->cardinality = - bitset_container_compute_cardinality( - (bitset_container_t *)container); - if (((bitset_container_t *)container)->cardinality <= - DEFAULT_MAX_SIZE) { - result = array_container_from_bitset( - (const bitset_container_t *)container); - bitset_container_free((bitset_container_t *)container); - *typecode = ARRAY_CONTAINER_TYPE_CODE; - return result; - } - return container; - case ARRAY_CONTAINER_TYPE_CODE: - return container; // nothing to do - case RUN_CONTAINER_TYPE_CODE: - return convert_run_to_efficient_container_and_free( - (run_container_t *)container, typecode); - case SHARED_CONTAINER_TYPE_CODE: - assert(false); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * Writes the underlying array to buf, outputs how many bytes were written. - * This is meant to be byte-by-byte compatible with the Java and Go versions of - * Roaring. - * The number of bytes written should be - * container_write(container, buf). - * - */ -static inline int32_t container_write(const void *container, uint8_t typecode, - char *buf) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_write((const bitset_container_t *)container, buf); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_write((const array_container_t *)container, buf); - case RUN_CONTAINER_TYPE_CODE: - return run_container_write((const run_container_t *)container, buf); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * Get the container size in bytes under portable serialization (see - * container_write), requires a - * typecode - */ -static inline int32_t container_size_in_bytes(const void *container, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_size_in_bytes( - (const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_size_in_bytes( - (const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_size_in_bytes((const run_container_t *)container); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * print the container (useful for debugging), requires a typecode - */ -void container_printf(const void *container, uint8_t typecode); - -/** - * print the content of the container as a comma-separated list of 32-bit values - * starting at base, requires a typecode - */ -void container_printf_as_uint32_array(const void *container, uint8_t typecode, - uint32_t base); - -/** - * Checks whether a container is not empty, requires a typecode - */ -static inline bool container_nonzero_cardinality(const void *container, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_const_nonzero_cardinality( - (const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_nonzero_cardinality( - (const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_nonzero_cardinality( - (const run_container_t *)container); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * Recover memory from a container, requires a typecode - */ -void container_free(void *container, uint8_t typecode); - -/** - * Convert a container to an array of values, requires a typecode as well as a - * "base" (most significant values) - * Returns number of ints added. - */ -static inline int container_to_uint32_array(uint32_t *output, - const void *container, - uint8_t typecode, uint32_t base) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_to_uint32_array( - output, (const bitset_container_t *)container, base); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_to_uint32_array( - output, (const array_container_t *)container, base); - case RUN_CONTAINER_TYPE_CODE: - return run_container_to_uint32_array( - output, (const run_container_t *)container, base); - } - assert(false); - __builtin_unreachable(); - return 0; // unreached -} - -/** - * Add a value to a container, requires a typecode, fills in new_typecode and - * return (possibly different) container. - * This function may allocate a new container, and caller is responsible for - * memory deallocation - */ -static inline void *container_add(void *container, uint16_t val, - uint8_t typecode, uint8_t *new_typecode) { - container = get_writable_copy_if_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - bitset_container_set((bitset_container_t *)container, val); - *new_typecode = BITSET_CONTAINER_TYPE_CODE; - return container; - case ARRAY_CONTAINER_TYPE_CODE: { - array_container_t *ac = (array_container_t *)container; - if (array_container_try_add(ac, val, DEFAULT_MAX_SIZE) != -1) { - *new_typecode = ARRAY_CONTAINER_TYPE_CODE; - return ac; - } else { - bitset_container_t* bitset = bitset_container_from_array(ac); - bitset_container_add(bitset, val); - *new_typecode = BITSET_CONTAINER_TYPE_CODE; - return bitset; - } - } break; - case RUN_CONTAINER_TYPE_CODE: - // per Java, no container type adjustments are done (revisit?) - run_container_add((run_container_t *)container, val); - *new_typecode = RUN_CONTAINER_TYPE_CODE; - return container; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Remove a value from a container, requires a typecode, fills in new_typecode - * and - * return (possibly different) container. - * This function may allocate a new container, and caller is responsible for - * memory deallocation - */ -static inline void *container_remove(void *container, uint16_t val, - uint8_t typecode, uint8_t *new_typecode) { - container = get_writable_copy_if_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - if (bitset_container_remove((bitset_container_t *)container, val)) { - if (bitset_container_cardinality( - (bitset_container_t *)container) <= DEFAULT_MAX_SIZE) { - *new_typecode = ARRAY_CONTAINER_TYPE_CODE; - return array_container_from_bitset( - (bitset_container_t *)container); - } - } - *new_typecode = typecode; - return container; - case ARRAY_CONTAINER_TYPE_CODE: - *new_typecode = typecode; - array_container_remove((array_container_t *)container, val); - return container; - case RUN_CONTAINER_TYPE_CODE: - // per Java, no container type adjustments are done (revisit?) - run_container_remove((run_container_t *)container, val); - *new_typecode = RUN_CONTAINER_TYPE_CODE; - return container; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Check whether a value is in a container, requires a typecode - */ -inline bool container_contains(const void *container, uint16_t val, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_get((const bitset_container_t *)container, - val); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_contains( - (const array_container_t *)container, val); - case RUN_CONTAINER_TYPE_CODE: - return run_container_contains((const run_container_t *)container, - val); - default: - assert(false); - __builtin_unreachable(); - return false; - } -} - -/** - * Check whether a range of values from range_start (included) to range_end (excluded) - * is in a container, requires a typecode - */ -static inline bool container_contains_range(const void *container, uint32_t range_start, - uint32_t range_end, uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_get_range((const bitset_container_t *)container, - range_start, range_end); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_contains_range((const array_container_t *)container, - range_start, range_end); - case RUN_CONTAINER_TYPE_CODE: - return run_container_contains_range((const run_container_t *)container, - range_start, range_end); - default: - assert(false); - __builtin_unreachable(); - return false; - } -} - -int32_t container_serialize(const void *container, uint8_t typecode, - char *buf) WARN_UNUSED; - -uint32_t container_serialization_len(const void *container, uint8_t typecode); - -void *container_deserialize(uint8_t typecode, const char *buf, size_t buf_len); - -/** - * Returns true if the two containers have the same content. Note that - * two containers having different types can be "equal" in this sense. - */ -static inline bool container_equals(const void *c1, uint8_t type1, - const void *c2, uint8_t type2) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return bitset_container_equals((const bitset_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - return run_container_equals_bitset((const run_container_t *)c2, - (const bitset_container_t *)c1); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return run_container_equals_bitset((const run_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - // java would always return false? - return array_container_equal_bitset((const array_container_t *)c2, - (const bitset_container_t *)c1); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - // java would always return false? - return array_container_equal_bitset((const array_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return run_container_equals_array((const run_container_t *)c2, - (const array_container_t *)c1); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - return run_container_equals_array((const run_container_t *)c1, - (const array_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_container_equals((const array_container_t *)c1, - (const array_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return run_container_equals((const run_container_t *)c1, - (const run_container_t *)c2); - default: - assert(false); - __builtin_unreachable(); - return false; - } -} - -/** - * Returns true if the container c1 is a subset of the container c2. Note that - * c1 can be a subset of c2 even if they have a different type. - */ -static inline bool container_is_subset(const void *c1, uint8_t type1, - const void *c2, uint8_t type2) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return bitset_container_is_subset((const bitset_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - return bitset_container_is_subset_run((const bitset_container_t *)c1, - (const run_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return run_container_is_subset_bitset((const run_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return false; // by construction, size(c1) > size(c2) - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return array_container_is_subset_bitset((const array_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return array_container_is_subset_run((const array_container_t *)c1, - (const run_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - return run_container_is_subset_array((const run_container_t *)c1, - (const array_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_container_is_subset((const array_container_t *)c1, - (const array_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return run_container_is_subset((const run_container_t *)c1, - (const run_container_t *)c2); - default: - assert(false); - __builtin_unreachable(); - return false; - } -} - -// macro-izations possibilities for generic non-inplace binary-op dispatch - -/** - * Compute intersection between two containers, generate a new container (having - * type result_type), requires a typecode. This allocates new memory, caller - * is responsible for deallocation. - */ -static inline void *container_and(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = bitset_bitset_container_intersection( - (const bitset_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = array_container_create(); - array_container_intersection((const array_container_t *)c1, - (const array_container_t *)c2, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - run_container_intersection((const run_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - return convert_run_to_efficient_container_and_free( - (run_container_t *)result, result_type); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = array_container_create(); - array_bitset_container_intersection((const array_container_t *)c2, - (const bitset_container_t *)c1, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_bitset_container_intersection((const array_container_t *)c1, - (const bitset_container_t *)c2, - (array_container_t *)result); - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_intersection( - (const run_container_t *)c2, - (const bitset_container_t *)c1, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_intersection( - (const run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_run_container_intersection((const array_container_t *)c1, - (const run_container_t *)c2, - (array_container_t *)result); - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_run_container_intersection((const array_container_t *)c2, - (const run_container_t *)c1, - (array_container_t *)result); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Compute the size of the intersection between two containers. - */ -static inline int container_and_cardinality(const void *c1, uint8_t type1, - const void *c2, uint8_t type2) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return bitset_container_and_justcard( - (const bitset_container_t *)c1, (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_container_intersection_cardinality( - (const array_container_t *)c1, (const array_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return run_container_intersection_cardinality( - (const run_container_t *)c1, (const run_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_bitset_container_intersection_cardinality( - (const array_container_t *)c2, (const bitset_container_t *)c1); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return array_bitset_container_intersection_cardinality( - (const array_container_t *)c1, (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - return run_bitset_container_intersection_cardinality( - (const run_container_t *)c2, (const bitset_container_t *)c1); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return run_bitset_container_intersection_cardinality( - (const run_container_t *)c1, (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return array_run_container_intersection_cardinality( - (const array_container_t *)c1, (const run_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - return array_run_container_intersection_cardinality( - (const array_container_t *)c2, (const run_container_t *)c1); - default: - assert(false); - __builtin_unreachable(); - return 0; - } -} - -/** - * Check whether two containers intersect. - */ -static inline bool container_intersect(const void *c1, uint8_t type1, const void *c2, - uint8_t type2) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return bitset_container_intersect( - (const bitset_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_container_intersect((const array_container_t *)c1, - (const array_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return run_container_intersect((const run_container_t *)c1, - (const run_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - return array_bitset_container_intersect((const array_container_t *)c2, - (const bitset_container_t *)c1); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return array_bitset_container_intersect((const array_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - return run_bitset_container_intersect( - (const run_container_t *)c2, - (const bitset_container_t *)c1); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - return run_bitset_container_intersect( - (const run_container_t *)c1, - (const bitset_container_t *)c2); - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - return array_run_container_intersect((const array_container_t *)c1, - (const run_container_t *)c2); - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - return array_run_container_intersect((const array_container_t *)c2, - (const run_container_t *)c1); - default: - assert(false); - __builtin_unreachable(); - return 0; - } -} - -/** - * Compute intersection between two containers, with result in the first - container if possible. If the returned pointer is identical to c1, - then the container has been modified. If the returned pointer is different - from c1, then a new container has been created and the caller is responsible - for freeing it. - The type of the first container may change. Returns the modified - (and possibly new) container. -*/ -static inline void *container_iand(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = get_writable_copy_if_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = - bitset_bitset_container_intersection_inplace( - (bitset_container_t *)c1, (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - array_container_intersection_inplace((array_container_t *)c1, - (const array_container_t *)c2); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - run_container_intersection((const run_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - // as of January 2016, Java code used non-in-place intersection for - // two runcontainers - return convert_run_to_efficient_container_and_free( - (run_container_t *)result, result_type); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - // c1 is a bitmap so no inplace possible - result = array_container_create(); - array_bitset_container_intersection((const array_container_t *)c2, - (const bitset_container_t *)c1, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_bitset_container_intersection( - (const array_container_t *)c1, (const bitset_container_t *)c2, - (array_container_t *)c1); // allowed - return c1; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - // will attempt in-place computation - *result_type = run_bitset_container_intersection( - (const run_container_t *)c2, - (const bitset_container_t *)c1, &c1) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_intersection( - (const run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_run_container_intersection((const array_container_t *)c1, - (const run_container_t *)c2, - (array_container_t *)result); - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; // never bitset - array_run_container_intersection((const array_container_t *)c2, - (const run_container_t *)c1, - (array_container_t *)result); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Compute union between two containers, generate a new container (having type - * result_type), requires a typecode. This allocates new memory, caller - * is responsible for deallocation. - */ -static inline void *container_or(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - bitset_container_or((const bitset_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_union( - (const array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - run_container_union((const run_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // todo: could be optimized since will never convert to array - result = convert_run_to_efficient_container_and_free( - (run_container_t *)result, (uint8_t *)result_type); - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - array_bitset_container_union((const array_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - array_bitset_container_union((const array_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c2, - (run_container_t *)result); - return result; - } - result = bitset_container_create(); - run_bitset_container_union((const run_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c1)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c1, - (run_container_t *)result); - return result; - } - result = bitset_container_create(); - run_bitset_container_union((const run_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union((const array_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - result = convert_run_to_efficient_container_and_free( - (run_container_t *)result, (uint8_t *)result_type); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union((const array_container_t *)c2, - (const run_container_t *)c1, - (run_container_t *)result); - result = convert_run_to_efficient_container_and_free( - (run_container_t *)result, (uint8_t *)result_type); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; // unreached - } -} - -/** - * Compute union between two containers, generate a new container (having type - * result_type), requires a typecode. This allocates new memory, caller - * is responsible for deallocation. - * - * This lazy version delays some operations such as the maintenance of the - * cardinality. It requires repair later on the generated containers. - */ -static inline void *container_lazy_or(const void *c1, uint8_t type1, - const void *c2, uint8_t type2, - uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - bitset_container_or_nocard( - (const bitset_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_lazy_union( - (const array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - run_container_union((const run_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // we are being lazy - result = convert_run_to_efficient_container( - (run_container_t *)result, result_type); - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - array_bitset_container_lazy_union( - (const array_container_t *)c2, (const bitset_container_t *)c1, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - array_bitset_container_lazy_union( - (const array_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c2, - (run_container_t *)result); - return result; - } - result = bitset_container_create(); - run_bitset_container_lazy_union( - (const run_container_t *)c2, (const bitset_container_t *)c1, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c1)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c1, - (run_container_t *)result); - return result; - } - result = bitset_container_create(); - run_bitset_container_lazy_union( - (const run_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union((const array_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container(result, result_type); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union( - (const array_container_t *)c2, (const run_container_t *)c1, - (run_container_t *)result); // TODO make lazy - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container(result, result_type); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; // unreached - } -} - -/** - * Compute the union between two containers, with result in the first container. - * If the returned pointer is identical to c1, then the container has been - * modified. - * If the returned pointer is different from c1, then a new container has been - * created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container -*/ -static inline void *container_ior(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = get_writable_copy_if_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - bitset_container_or((const bitset_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)c1); -#ifdef OR_BITSET_CONVERSION_TO_FULL - if (((bitset_container_t *)c1)->cardinality == - (1 << 16)) { // we convert - result = run_container_create_range(0, (1 << 16)); - *result_type = RUN_CONTAINER_TYPE_CODE; - return result; - } -#endif - *result_type = BITSET_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_inplace_union( - (array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - if((result == NULL) - && (*result_type == ARRAY_CONTAINER_TYPE_CODE)) { - return c1; // the computation was done in-place! - } - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - run_container_union_inplace((run_container_t *)c1, - (const run_container_t *)c2); - return convert_run_to_efficient_container((run_container_t *)c1, - result_type); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - array_bitset_container_union((const array_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)c1); - *result_type = BITSET_CONTAINER_TYPE_CODE; // never array - return c1; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - // c1 is an array, so no in-place possible - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_bitset_container_union((const array_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c2, - (run_container_t *)result); - return result; - } - run_bitset_container_union((const run_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)c1); // allowed - *result_type = BITSET_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c1)) { - *result_type = RUN_CONTAINER_TYPE_CODE; - - return c1; - } - result = bitset_container_create(); - run_bitset_container_union((const run_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union((const array_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - result = convert_run_to_efficient_container_and_free( - (run_container_t *)result, result_type); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - array_run_container_inplace_union((const array_container_t *)c2, - (run_container_t *)c1); - c1 = convert_run_to_efficient_container((run_container_t *)c1, - result_type); - return c1; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Compute the union between two containers, with result in the first container. - * If the returned pointer is identical to c1, then the container has been - * modified. - * If the returned pointer is different from c1, then a new container has been - * created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container - * - * This lazy version delays some operations such as the maintenance of the - * cardinality. It requires repair later on the generated containers. -*/ -static inline void *container_lazy_ior(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - assert(type1 != SHARED_CONTAINER_TYPE_CODE); - // c1 = get_writable_copy_if_shared(c1,&type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): -#ifdef LAZY_OR_BITSET_CONVERSION_TO_FULL - // if we have two bitsets, we might as well compute the cardinality - bitset_container_or((const bitset_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)c1); - // it is possible that two bitsets can lead to a full container - if (((bitset_container_t *)c1)->cardinality == - (1 << 16)) { // we convert - result = run_container_create_range(0, (1 << 16)); - *result_type = RUN_CONTAINER_TYPE_CODE; - return result; - } -#else - bitset_container_or_nocard((const bitset_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)c1); - -#endif - *result_type = BITSET_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_lazy_inplace_union( - (array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - if((result == NULL) - && (*result_type == ARRAY_CONTAINER_TYPE_CODE)) { - return c1; // the computation was done in-place! - } - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - run_container_union_inplace((run_container_t *)c1, - (const run_container_t *)c2); - *result_type = RUN_CONTAINER_TYPE_CODE; - return convert_run_to_efficient_container((run_container_t *)c1, - result_type); - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - array_bitset_container_lazy_union( - (const array_container_t *)c2, (const bitset_container_t *)c1, - (bitset_container_t *)c1); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; // never array - return c1; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - // c1 is an array, so no in-place possible - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_bitset_container_lazy_union( - (const array_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // is lazy - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = run_container_create(); - *result_type = RUN_CONTAINER_TYPE_CODE; - run_container_copy((const run_container_t *)c2, - (run_container_t *)result); - return result; - } - run_bitset_container_lazy_union( - (const run_container_t *)c2, (const bitset_container_t *)c1, - (bitset_container_t *)c1); // allowed // lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return c1; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c1)) { - *result_type = RUN_CONTAINER_TYPE_CODE; - return c1; - } - result = bitset_container_create(); - run_bitset_container_lazy_union( - (const run_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_union((const array_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container_and_free(result, - // result_type); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - array_run_container_inplace_union((const array_container_t *)c2, - (run_container_t *)c1); - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container_and_free(result, - // result_type); - return c1; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Compute symmetric difference (xor) between two containers, generate a new - * container (having type result_type), requires a typecode. This allocates new - * memory, caller is responsible for deallocation. - */ -static inline void *container_xor(const void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = bitset_bitset_container_xor( - (const bitset_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_xor( - (const array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = - run_run_container_xor((const run_container_t *)c1, - (const run_container_t *)c2, &result); - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_bitset_container_xor( - (const array_container_t *)c2, - (const bitset_container_t *)c1, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = array_bitset_container_xor( - (const array_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_xor( - (const run_container_t *)c2, - (const bitset_container_t *)c1, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - - *result_type = run_bitset_container_xor( - (const run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = - array_run_container_xor((const array_container_t *)c1, - (const run_container_t *)c2, &result); - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - *result_type = - array_run_container_xor((const array_container_t *)c2, - (const run_container_t *)c1, &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - return NULL; // unreached - } -} - -/** - * Compute xor between two containers, generate a new container (having type - * result_type), requires a typecode. This allocates new memory, caller - * is responsible for deallocation. - * - * This lazy version delays some operations such as the maintenance of the - * cardinality. It requires repair later on the generated containers. - */ -static inline void *container_lazy_xor(const void *c1, uint8_t type1, - const void *c2, uint8_t type2, - uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - bitset_container_xor_nocard( - (const bitset_container_t *)c1, (const bitset_container_t *)c2, - (bitset_container_t *)result); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_lazy_xor( - (const array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - // nothing special done yet. - *result_type = - run_run_container_xor((const run_container_t *)c1, - (const run_container_t *)c2, &result); - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_bitset_container_lazy_xor((const array_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)result); - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_bitset_container_lazy_xor((const array_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - run_bitset_container_lazy_xor((const run_container_t *)c2, - (const bitset_container_t *)c1, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = bitset_container_create(); - run_bitset_container_lazy_xor((const run_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)result); - *result_type = BITSET_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_lazy_xor((const array_container_t *)c1, - (const run_container_t *)c2, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container(result, result_type); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - result = run_container_create(); - array_run_container_lazy_xor((const array_container_t *)c2, - (const run_container_t *)c1, - (run_container_t *)result); - *result_type = RUN_CONTAINER_TYPE_CODE; - // next line skipped since we are lazy - // result = convert_run_to_efficient_container(result, result_type); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; // unreached - } -} - -/** - * Compute the xor between two containers, with result in the first container. - * If the returned pointer is identical to c1, then the container has been - * modified. - * If the returned pointer is different from c1, then a new container has been - * created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container -*/ -static inline void *container_ixor(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = get_writable_copy_if_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = bitset_bitset_container_ixor( - (bitset_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = array_array_container_ixor( - (array_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = run_run_container_ixor( - (run_container_t *)c1, (const run_container_t *)c2, &result); - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = bitset_array_container_ixor( - (bitset_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = array_bitset_container_ixor( - (array_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - *result_type = - bitset_run_container_ixor((bitset_container_t *)c1, - (const run_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_ixor( - (run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - return result; - - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = array_run_container_ixor( - (array_container_t *)c1, (const run_container_t *)c2, &result); - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - *result_type = run_array_container_ixor( - (run_container_t *)c1, (const array_container_t *)c2, &result); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Compute the xor between two containers, with result in the first container. - * If the returned pointer is identical to c1, then the container has been - * modified. - * If the returned pointer is different from c1, then a new container has been - * created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container - * - * This lazy version delays some operations such as the maintenance of the - * cardinality. It requires repair later on the generated containers. -*/ -static inline void *container_lazy_ixor(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - assert(type1 != SHARED_CONTAINER_TYPE_CODE); - // c1 = get_writable_copy_if_shared(c1,&type1); - c2 = container_unwrap_shared(c2, &type2); - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - bitset_container_xor_nocard((bitset_container_t *)c1, - (const bitset_container_t *)c2, - (bitset_container_t *)c1); // is lazy - *result_type = BITSET_CONTAINER_TYPE_CODE; - return c1; - // TODO: other cases being lazy, esp. when we know inplace not likely - // could see the corresponding code for union - default: - // we may have a dirty bitset (without a precomputed cardinality) and - // calling container_ixor on it might be unsafe. - if( (type1 == BITSET_CONTAINER_TYPE_CODE) - && (((const bitset_container_t *)c1)->cardinality == BITSET_UNKNOWN_CARDINALITY)) { - ((bitset_container_t *)c1)->cardinality = bitset_container_compute_cardinality((bitset_container_t *)c1); - } - return container_ixor(c1, type1, c2, type2, result_type); - } -} - -/** - * Compute difference (andnot) between two containers, generate a new - * container (having type result_type), requires a typecode. This allocates new - * memory, caller is responsible for deallocation. - */ -static inline void *container_andnot(const void *c1, uint8_t type1, - const void *c2, uint8_t type2, - uint8_t *result_type) { - c1 = container_unwrap_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = bitset_bitset_container_andnot( - (const bitset_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - result = array_container_create(); - array_array_container_andnot((const array_container_t *)c1, - (const array_container_t *)c2, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - } - *result_type = - run_run_container_andnot((const run_container_t *)c1, - (const run_container_t *)c2, &result); - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = bitset_array_container_andnot( - (const bitset_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - result = array_container_create(); - array_bitset_container_andnot((const array_container_t *)c1, - (const bitset_container_t *)c2, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - } - *result_type = bitset_run_container_andnot( - (const bitset_container_t *)c1, - (const run_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - - *result_type = run_bitset_container_andnot( - (const run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - if (run_container_is_full((const run_container_t *)c2)) { - result = array_container_create(); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - } - result = array_container_create(); - array_run_container_andnot((const array_container_t *)c1, - (const run_container_t *)c2, - (array_container_t *)result); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - *result_type = run_array_container_andnot( - (const run_container_t *)c1, (const array_container_t *)c2, - &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - return NULL; // unreached - } -} - -/** - * Compute the andnot between two containers, with result in the first - * container. - * If the returned pointer is identical to c1, then the container has been - * modified. - * If the returned pointer is different from c1, then a new container has been - * created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container -*/ -static inline void *container_iandnot(void *c1, uint8_t type1, const void *c2, - uint8_t type2, uint8_t *result_type) { - c1 = get_writable_copy_if_shared(c1, &type1); - c2 = container_unwrap_shared(c2, &type2); - void *result = NULL; - switch (CONTAINER_PAIR(type1, type2)) { - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = bitset_bitset_container_iandnot( - (bitset_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - array_array_container_iandnot((array_container_t *)c1, - (const array_container_t *)c2); - *result_type = ARRAY_CONTAINER_TYPE_CODE; - return c1; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = run_run_container_iandnot( - (run_container_t *)c1, (const run_container_t *)c2, &result); - return result; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - ARRAY_CONTAINER_TYPE_CODE): - *result_type = bitset_array_container_iandnot( - (bitset_container_t *)c1, - (const array_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = ARRAY_CONTAINER_TYPE_CODE; - - array_bitset_container_iandnot((array_container_t *)c1, - (const bitset_container_t *)c2); - return c1; - - case CONTAINER_PAIR(BITSET_CONTAINER_TYPE_CODE, - RUN_CONTAINER_TYPE_CODE): - *result_type = bitset_run_container_iandnot( - (bitset_container_t *)c1, - (const run_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - return result; - - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, - BITSET_CONTAINER_TYPE_CODE): - *result_type = run_bitset_container_iandnot( - (run_container_t *)c1, - (const bitset_container_t *)c2, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - - return result; - - case CONTAINER_PAIR(ARRAY_CONTAINER_TYPE_CODE, RUN_CONTAINER_TYPE_CODE): - *result_type = ARRAY_CONTAINER_TYPE_CODE; - array_run_container_iandnot((array_container_t *)c1, - (const run_container_t *)c2); - return c1; - case CONTAINER_PAIR(RUN_CONTAINER_TYPE_CODE, ARRAY_CONTAINER_TYPE_CODE): - *result_type = run_array_container_iandnot( - (run_container_t *)c1, (const array_container_t *)c2, &result); - return result; - default: - assert(false); - __builtin_unreachable(); - return NULL; - } -} - -/** - * Visit all values x of the container once, passing (base+x,ptr) - * to iterator. You need to specify a container and its type. - * Returns true if the iteration should continue. - */ -static inline bool container_iterate(const void *container, uint8_t typecode, - uint32_t base, roaring_iterator iterator, - void *ptr) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_iterate( - (const bitset_container_t *)container, base, iterator, ptr); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_iterate((const array_container_t *)container, - base, iterator, ptr); - case RUN_CONTAINER_TYPE_CODE: - return run_container_iterate((const run_container_t *)container, - base, iterator, ptr); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -static inline bool container_iterate64(const void *container, uint8_t typecode, - uint32_t base, - roaring_iterator64 iterator, - uint64_t high_bits, void *ptr) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_iterate64( - (const bitset_container_t *)container, base, iterator, - high_bits, ptr); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_iterate64( - (const array_container_t *)container, base, iterator, high_bits, - ptr); - case RUN_CONTAINER_TYPE_CODE: - return run_container_iterate64((const run_container_t *)container, - base, iterator, high_bits, ptr); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -static inline void *container_not(const void *c, uint8_t typ, - uint8_t *result_type) { - c = container_unwrap_shared(c, &typ); - void *result = NULL; - switch (typ) { - case BITSET_CONTAINER_TYPE_CODE: - *result_type = bitset_container_negation( - (const bitset_container_t *)c, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case ARRAY_CONTAINER_TYPE_CODE: - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_container_negation((const array_container_t *)c, - (bitset_container_t *)result); - return result; - case RUN_CONTAINER_TYPE_CODE: - *result_type = - run_container_negation((const run_container_t *)c, &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return NULL; -} - -static inline void *container_not_range(const void *c, uint8_t typ, - uint32_t range_start, - uint32_t range_end, - uint8_t *result_type) { - c = container_unwrap_shared(c, &typ); - void *result = NULL; - switch (typ) { - case BITSET_CONTAINER_TYPE_CODE: - *result_type = - bitset_container_negation_range((const bitset_container_t *)c, - range_start, range_end, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case ARRAY_CONTAINER_TYPE_CODE: - *result_type = - array_container_negation_range((const array_container_t *)c, - range_start, range_end, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case RUN_CONTAINER_TYPE_CODE: - *result_type = run_container_negation_range( - (const run_container_t *)c, range_start, range_end, &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return NULL; -} - -static inline void *container_inot(void *c, uint8_t typ, uint8_t *result_type) { - c = get_writable_copy_if_shared(c, &typ); - void *result = NULL; - switch (typ) { - case BITSET_CONTAINER_TYPE_CODE: - *result_type = bitset_container_negation_inplace( - (bitset_container_t *)c, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case ARRAY_CONTAINER_TYPE_CODE: - // will never be inplace - result = bitset_container_create(); - *result_type = BITSET_CONTAINER_TYPE_CODE; - array_container_negation((array_container_t *)c, - (bitset_container_t *)result); - array_container_free((array_container_t *)c); - return result; - case RUN_CONTAINER_TYPE_CODE: - *result_type = - run_container_negation_inplace((run_container_t *)c, &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return NULL; -} - -static inline void *container_inot_range(void *c, uint8_t typ, - uint32_t range_start, - uint32_t range_end, - uint8_t *result_type) { - c = get_writable_copy_if_shared(c, &typ); - void *result = NULL; - switch (typ) { - case BITSET_CONTAINER_TYPE_CODE: - *result_type = - bitset_container_negation_range_inplace( - (bitset_container_t *)c, range_start, range_end, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case ARRAY_CONTAINER_TYPE_CODE: - *result_type = - array_container_negation_range_inplace( - (array_container_t *)c, range_start, range_end, &result) - ? BITSET_CONTAINER_TYPE_CODE - : ARRAY_CONTAINER_TYPE_CODE; - return result; - case RUN_CONTAINER_TYPE_CODE: - *result_type = run_container_negation_range_inplace( - (run_container_t *)c, range_start, range_end, &result); - return result; - - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return NULL; -} - -/** - * If the element of given rank is in this container, supposing that - * the first - * element has rank start_rank, then the function returns true and - * sets element - * accordingly. - * Otherwise, it returns false and update start_rank. - */ -static inline bool container_select(const void *container, uint8_t typecode, - uint32_t *start_rank, uint32_t rank, - uint32_t *element) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_select((const bitset_container_t *)container, - start_rank, rank, element); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_select((const array_container_t *)container, - start_rank, rank, element); - case RUN_CONTAINER_TYPE_CODE: - return run_container_select((const run_container_t *)container, - start_rank, rank, element); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -static inline uint16_t container_maximum(const void *container, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_maximum((const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_maximum((const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_maximum((const run_container_t *)container); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -static inline uint16_t container_minimum(const void *container, - uint8_t typecode) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_minimum((const bitset_container_t *)container); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_minimum((const array_container_t *)container); - case RUN_CONTAINER_TYPE_CODE: - return run_container_minimum((const run_container_t *)container); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -// number of values smaller or equal to x -static inline int container_rank(const void *container, uint8_t typecode, - uint16_t x) { - container = container_unwrap_shared(container, &typecode); - switch (typecode) { - case BITSET_CONTAINER_TYPE_CODE: - return bitset_container_rank((const bitset_container_t *)container, x); - case ARRAY_CONTAINER_TYPE_CODE: - return array_container_rank((const array_container_t *)container, x); - case RUN_CONTAINER_TYPE_CODE: - return run_container_rank((const run_container_t *)container, x); - default: - assert(false); - __builtin_unreachable(); - } - assert(false); - __builtin_unreachable(); - return false; -} - -/** - * Add all values in range [min, max] to a given container. - * - * If the returned pointer is different from $container, then a new container - * has been created and the caller is responsible for freeing it. - * The type of the first container may change. Returns the modified - * (and possibly new) container. - */ -static inline void *container_add_range(void *container, uint8_t type, - uint32_t min, uint32_t max, - uint8_t *result_type) { - // NB: when selecting new container type, we perform only inexpensive checks - switch (type) { - case BITSET_CONTAINER_TYPE_CODE: { - bitset_container_t *bitset = (bitset_container_t *) container; - - int32_t union_cardinality = 0; - union_cardinality += bitset->cardinality; - union_cardinality += max - min + 1; - union_cardinality -= bitset_lenrange_cardinality(bitset->array, min, max-min); - - if (union_cardinality == INT32_C(0x10000)) { - *result_type = RUN_CONTAINER_TYPE_CODE; - return run_container_create_range(0, INT32_C(0x10000)); - } else { - *result_type = BITSET_CONTAINER_TYPE_CODE; - bitset_set_lenrange(bitset->array, min, max - min); - bitset->cardinality = union_cardinality; - return bitset; - } - } - case ARRAY_CONTAINER_TYPE_CODE: { - array_container_t *array = (array_container_t *) container; - - int32_t nvals_greater = count_greater(array->array, array->cardinality, max); - int32_t nvals_less = count_less(array->array, array->cardinality - nvals_greater, min); - int32_t union_cardinality = nvals_less + (max - min + 1) + nvals_greater; - - if (union_cardinality == INT32_C(0x10000)) { - *result_type = RUN_CONTAINER_TYPE_CODE; - return run_container_create_range(0, INT32_C(0x10000)); - } else if (union_cardinality <= DEFAULT_MAX_SIZE) { - *result_type = ARRAY_CONTAINER_TYPE_CODE; - array_container_add_range_nvals(array, min, max, nvals_less, nvals_greater); - return array; - } else { - *result_type = BITSET_CONTAINER_TYPE_CODE; - bitset_container_t *bitset = bitset_container_from_array(array); - bitset_set_lenrange(bitset->array, min, max - min); - bitset->cardinality = union_cardinality; - return bitset; - } - } - case RUN_CONTAINER_TYPE_CODE: { - run_container_t *run = (run_container_t *) container; - - int32_t nruns_greater = rle16_count_greater(run->runs, run->n_runs, max); - int32_t nruns_less = rle16_count_less(run->runs, run->n_runs - nruns_greater, min); - - int32_t run_size_bytes = (nruns_less + 1 + nruns_greater) * sizeof(rle16_t); - int32_t bitset_size_bytes = BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); - - if (run_size_bytes <= bitset_size_bytes) { - run_container_add_range_nruns(run, min, max, nruns_less, nruns_greater); - *result_type = RUN_CONTAINER_TYPE_CODE; - return run; - } else { - *result_type = BITSET_CONTAINER_TYPE_CODE; - return bitset_container_from_run_range(run, min, max); - } - } - default: - __builtin_unreachable(); - } -} - -/* - * Removes all elements in range [min, max]. - * Returns one of: - * - NULL if no elements left - * - pointer to the original container - * - pointer to a newly-allocated container (if it is more efficient) - * - * If the returned pointer is different from $container, then a new container - * has been created and the caller is responsible for freeing the original container. - */ -static inline void *container_remove_range(void *container, uint8_t type, - uint32_t min, uint32_t max, - uint8_t *result_type) { - switch (type) { - case BITSET_CONTAINER_TYPE_CODE: { - bitset_container_t *bitset = (bitset_container_t *) container; - - int32_t result_cardinality = bitset->cardinality - - bitset_lenrange_cardinality(bitset->array, min, max-min); - - if (result_cardinality == 0) { - return NULL; - } else if (result_cardinality < DEFAULT_MAX_SIZE) { - *result_type = ARRAY_CONTAINER_TYPE_CODE; - bitset_reset_range(bitset->array, min, max+1); - bitset->cardinality = result_cardinality; - return array_container_from_bitset(bitset); - } else { - *result_type = BITSET_CONTAINER_TYPE_CODE; - bitset_reset_range(bitset->array, min, max+1); - bitset->cardinality = result_cardinality; - return bitset; - } - } - case ARRAY_CONTAINER_TYPE_CODE: { - array_container_t *array = (array_container_t *) container; - - int32_t nvals_greater = count_greater(array->array, array->cardinality, max); - int32_t nvals_less = count_less(array->array, array->cardinality - nvals_greater, min); - int32_t result_cardinality = nvals_less + nvals_greater; - - if (result_cardinality == 0) { - return NULL; - } else { - *result_type = ARRAY_CONTAINER_TYPE_CODE; - array_container_remove_range(array, nvals_less, - array->cardinality - result_cardinality); - return array; - } - } - case RUN_CONTAINER_TYPE_CODE: { - run_container_t *run = (run_container_t *) container; - - if (run->n_runs == 0) { - return NULL; - } - if (min <= run_container_minimum(run) && max >= run_container_maximum(run)) { - return NULL; - } - - run_container_remove_range(run, min, max); - - if (run_container_serialized_size_in_bytes(run->n_runs) <= - bitset_container_serialized_size_in_bytes()) { - *result_type = RUN_CONTAINER_TYPE_CODE; - return run; - } else { - *result_type = BITSET_CONTAINER_TYPE_CODE; - return bitset_container_from_run(run); - } - } - default: - __builtin_unreachable(); - } -} - -#ifdef __cplusplus -} -#endif - -#endif /* CONTAINERS_CONTAINERS_H */ - -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/containers/containers.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_array.h */ -#ifndef INCLUDE_ROARING_ARRAY_H -#define INCLUDE_ROARING_ARRAY_H -#ifdef __cplusplus -extern "C" { -#endif - -#include -#include -#include - -#define MAX_CONTAINERS 65536 - -#define SERIALIZATION_ARRAY_UINT32 1 -#define SERIALIZATION_CONTAINER 2 - -enum { - SERIAL_COOKIE_NO_RUNCONTAINER = 12346, - SERIAL_COOKIE = 12347, - NO_OFFSET_THRESHOLD = 4 -}; - -/** - * Roaring arrays are array-based key-value pairs having containers as values - * and 16-bit integer keys. A roaring bitmap might be implemented as such. - */ - -// parallel arrays. Element sizes quite different. -// Alternative is array -// of structs. Which would have better -// cache performance through binary searches? - -typedef struct roaring_array_s { - int32_t size; - int32_t allocation_size; - void **containers; - uint16_t *keys; - uint8_t *typecodes; -} roaring_array_t; - -/** - * Create a new roaring array - */ -roaring_array_t *ra_create(void); - -/** - * Initialize an existing roaring array with the specified capacity (in number - * of containers) - */ -bool ra_init_with_capacity(roaring_array_t *new_ra, uint32_t cap); - -/** - * Initialize with default capacity - */ -bool ra_init(roaring_array_t *t); - -/** - * Copies this roaring array, we assume that dest is not initialized - */ -bool ra_copy(const roaring_array_t *source, roaring_array_t *dest, - bool copy_on_write); - -/* - * Shrinks the capacity, returns the number of bytes saved. - */ -int ra_shrink_to_fit(roaring_array_t *ra); - -/** - * Copies this roaring array, we assume that dest is initialized - */ -bool ra_overwrite(const roaring_array_t *source, roaring_array_t *dest, - bool copy_on_write); - -/** - * Frees the memory used by a roaring array - */ -void ra_clear(roaring_array_t *r); - -/** - * Frees the memory used by a roaring array, but does not free the containers - */ -void ra_clear_without_containers(roaring_array_t *r); - -/** - * Frees just the containers - */ -void ra_clear_containers(roaring_array_t *ra); - -/** - * Get the index corresponding to a 16-bit key - */ -inline int32_t ra_get_index(const roaring_array_t *ra, uint16_t x) { - if ((ra->size == 0) || ra->keys[ra->size - 1] == x) return ra->size - 1; - return binarySearch(ra->keys, (int32_t)ra->size, x); -} - -/** - * Retrieves the container at index i, filling in the typecode - */ -inline void *ra_get_container_at_index(const roaring_array_t *ra, uint16_t i, - uint8_t *typecode) { - *typecode = ra->typecodes[i]; - return ra->containers[i]; -} - -/** - * Retrieves the key at index i - */ -uint16_t ra_get_key_at_index(const roaring_array_t *ra, uint16_t i); - -/** - * Add a new key-value pair at index i - */ -void ra_insert_new_key_value_at(roaring_array_t *ra, int32_t i, uint16_t key, - void *container, uint8_t typecode); - -/** - * Append a new key-value pair - */ -void ra_append(roaring_array_t *ra, uint16_t s, void *c, uint8_t typecode); - -/** - * Append a new key-value pair to ra, cloning (in COW sense) a value from sa - * at index index - */ -void ra_append_copy(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t index, bool copy_on_write); - -/** - * Append new key-value pairs to ra, cloning (in COW sense) values from sa - * at indexes - * [start_index, end_index) - */ -void ra_append_copy_range(roaring_array_t *ra, const roaring_array_t *sa, - int32_t start_index, int32_t end_index, - bool copy_on_write); - -/** appends from sa to ra, ending with the greatest key that is - * is less or equal stopping_key - */ -void ra_append_copies_until(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t stopping_key, bool copy_on_write); - -/** appends from sa to ra, starting with the smallest key that is - * is strictly greater than before_start - */ - -void ra_append_copies_after(roaring_array_t *ra, const roaring_array_t *sa, - uint16_t before_start, bool copy_on_write); - -/** - * Move the key-value pairs to ra from sa at indexes - * [start_index, end_index), old array should not be freed - * (use ra_clear_without_containers) - **/ -void ra_append_move_range(roaring_array_t *ra, roaring_array_t *sa, - int32_t start_index, int32_t end_index); -/** - * Append new key-value pairs to ra, from sa at indexes - * [start_index, end_index) - */ -void ra_append_range(roaring_array_t *ra, roaring_array_t *sa, - int32_t start_index, int32_t end_index, - bool copy_on_write); - -/** - * Set the container at the corresponding index using the specified - * typecode. - */ -inline void ra_set_container_at_index(const roaring_array_t *ra, int32_t i, - void *c, uint8_t typecode) { - assert(i < ra->size); - ra->containers[i] = c; - ra->typecodes[i] = typecode; -} - -/** - * If needed, increase the capacity of the array so that it can fit k values - * (at - * least); - */ -bool extend_array(roaring_array_t *ra, int32_t k); - -inline int32_t ra_get_size(const roaring_array_t *ra) { return ra->size; } - -static inline int32_t ra_advance_until(const roaring_array_t *ra, uint16_t x, - int32_t pos) { - return advanceUntil(ra->keys, pos, ra->size, x); -} - -int32_t ra_advance_until_freeing(roaring_array_t *ra, uint16_t x, int32_t pos); - -void ra_downsize(roaring_array_t *ra, int32_t new_length); - -inline void ra_replace_key_and_container_at_index(roaring_array_t *ra, - int32_t i, uint16_t key, - void *c, uint8_t typecode) { - assert(i < ra->size); - - ra->keys[i] = key; - ra->containers[i] = c; - ra->typecodes[i] = typecode; -} - -// write set bits to an array -void ra_to_uint32_array(const roaring_array_t *ra, uint32_t *ans); - -bool ra_range_uint32_array(const roaring_array_t *ra, size_t offset, size_t limit, uint32_t *ans); - -/** - * write a bitmap to a buffer. This is meant to be compatible with - * the - * Java and Go versions. Return the size in bytes of the serialized - * output (which should be ra_portable_size_in_bytes(ra)). - */ -size_t ra_portable_serialize(const roaring_array_t *ra, char *buf); - -/** - * read a bitmap from a serialized version. This is meant to be compatible - * with the Java and Go versions. - * maxbytes indicates how many bytes available from buf. - * When the function returns true, roaring_array_t is populated with the data - * and *readbytes indicates how many bytes were read. In all cases, if the function - * returns true, then maxbytes >= *readbytes. - */ -bool ra_portable_deserialize(roaring_array_t *ra, const char *buf, const size_t maxbytes, size_t * readbytes); - -/** - * Quickly checks whether there is a serialized bitmap at the pointer, - * not exceeding size "maxbytes" in bytes. This function does not allocate - * memory dynamically. - * - * This function returns 0 if and only if no valid bitmap is found. - * Otherwise, it returns how many bytes are occupied by the bitmap data. - */ -size_t ra_portable_deserialize_size(const char *buf, const size_t maxbytes); - -/** - * How many bytes are required to serialize this bitmap (meant to be - * compatible - * with Java and Go versions) - */ -size_t ra_portable_size_in_bytes(const roaring_array_t *ra); - -/** - * return true if it contains at least one run container. - */ -bool ra_has_run_container(const roaring_array_t *ra); - -/** - * Size of the header when serializing (meant to be compatible - * with Java and Go versions) - */ -uint32_t ra_portable_header_size(const roaring_array_t *ra); - -/** - * If the container at the index i is share, unshare it (creating a local - * copy if needed). - */ -static inline void ra_unshare_container_at_index(roaring_array_t *ra, - uint16_t i) { - assert(i < ra->size); - ra->containers[i] = - get_writable_copy_if_shared(ra->containers[i], &ra->typecodes[i]); -} - -/** - * remove at index i, sliding over all entries after i - */ -void ra_remove_at_index(roaring_array_t *ra, int32_t i); - - -/** -* clears all containers, sets the size at 0 and shrinks the memory usage. -*/ -void ra_reset(roaring_array_t *ra); - -/** - * remove at index i, sliding over all entries after i. Free removed container. - */ -void ra_remove_at_index_and_free(roaring_array_t *ra, int32_t i); - -/** - * remove a chunk of indices, sliding over entries after it - */ -// void ra_remove_index_range(roaring_array_t *ra, int32_t begin, int32_t end); - -// used in inplace andNot only, to slide left the containers from -// the mutated RoaringBitmap that are after the largest container of -// the argument RoaringBitmap. It is followed by a call to resize. -// -void ra_copy_range(roaring_array_t *ra, uint32_t begin, uint32_t end, - uint32_t new_begin); - -/** - * Shifts rightmost $count containers to the left (distance < 0) or - * to the right (distance > 0). - * Allocates memory if necessary. - * This function doesn't free or create new containers. - * Caller is responsible for that. - */ -void ra_shift_tail(roaring_array_t *ra, int32_t count, int32_t distance); - -#ifdef __cplusplus -} -#endif - -#endif -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring_array.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/misc/configreport.h */ -/* - * configreport.h - * - */ - -#ifndef INCLUDE_MISC_CONFIGREPORT_H_ -#define INCLUDE_MISC_CONFIGREPORT_H_ - -#include // for size_t -#include -#include - - -#ifdef IS_X64 -// useful for basic info (0) -static inline void native_cpuid(unsigned int *eax, unsigned int *ebx, - unsigned int *ecx, unsigned int *edx) { -#ifdef ROARING_INLINE_ASM - __asm volatile("cpuid" - : "=a"(*eax), "=b"(*ebx), "=c"(*ecx), "=d"(*edx) - : "0"(*eax), "2"(*ecx)); -#endif /* not sure what to do when inline assembly is unavailable*/ -} - -// CPUID instruction takes no parameters as CPUID implicitly uses the EAX -// register. -// The EAX register should be loaded with a value specifying what information to -// return -static inline void cpuinfo(int code, int *eax, int *ebx, int *ecx, int *edx) { -#ifdef ROARING_INLINE_ASM - __asm__ volatile("cpuid;" // call cpuid instruction - : "=a"(*eax), "=b"(*ebx), "=c"(*ecx), - "=d"(*edx) // output equal to "movl %%eax %1" - : "a"(code) // input equal to "movl %1, %%eax" - //:"%eax","%ebx","%ecx","%edx"// clobbered register - ); -#endif /* not sure what to do when inline assembly is unavailable*/ -} - -static inline int computecacheline() { - int eax = 0, ebx = 0, ecx = 0, edx = 0; - cpuinfo((int)0x80000006, &eax, &ebx, &ecx, &edx); - return ecx & 0xFF; -} - -// this is quite imperfect, but can be handy -static inline const char *guessprocessor() { - unsigned eax = 1, ebx = 0, ecx = 0, edx = 0; - native_cpuid(&eax, &ebx, &ecx, &edx); - const char *codename; - switch (eax >> 4) { - case 0x506E: - codename = "Skylake"; - break; - case 0x406C: - codename = "CherryTrail"; - break; - case 0x306D: - codename = "Broadwell"; - break; - case 0x306C: - codename = "Haswell"; - break; - case 0x306A: - codename = "IvyBridge"; - break; - case 0x206A: - case 0x206D: - codename = "SandyBridge"; - break; - case 0x2065: - case 0x206C: - case 0x206F: - codename = "Westmere"; - break; - case 0x106E: - case 0x106A: - case 0x206E: - codename = "Nehalem"; - break; - case 0x1067: - case 0x106D: - codename = "Penryn"; - break; - case 0x006F: - case 0x1066: - codename = "Merom"; - break; - case 0x0066: - codename = "Presler"; - break; - case 0x0063: - case 0x0064: - codename = "Prescott"; - break; - case 0x006D: - codename = "Dothan"; - break; - case 0x0366: - codename = "Cedarview"; - break; - case 0x0266: - codename = "Lincroft"; - break; - case 0x016C: - codename = "Pineview"; - break; - default: - codename = "UNKNOWN"; - break; - } - return codename; -} - -static inline void tellmeall() { - printf("Intel processor: %s\t", guessprocessor()); - -#ifdef __VERSION__ - printf(" compiler version: %s\t", __VERSION__); -#endif - printf("\tBuild option USEAVX "); -#ifdef USEAVX - printf("enabled\n"); -#else - printf("disabled\n"); -#endif -#ifndef __AVX2__ - printf("AVX2 is NOT available.\n"); -#endif - - if ((sizeof(int) != 4) || (sizeof(long) != 8)) { - printf("number of bytes: int = %lu long = %lu \n", - (long unsigned int)sizeof(size_t), - (long unsigned int)sizeof(int)); - } -#if __LITTLE_ENDIAN__ -// This is what we expect! -// printf("you have little endian machine"); -#endif -#if __BIG_ENDIAN__ - printf("you have a big endian machine"); -#endif -#if __CHAR_BIT__ - if (__CHAR_BIT__ != 8) printf("on your machine, chars don't have 8bits???"); -#endif - if (computecacheline() != 64) - printf("cache line: %d bytes\n", computecacheline()); -} -#else - -static inline void tellmeall() { - printf("Non-X64 processor\n"); -#ifdef __arm__ - printf("ARM processor detected\n"); -#endif -#ifdef __VERSION__ - printf(" compiler version: %s\t", __VERSION__); -#endif - if ((sizeof(int) != 4) || (sizeof(long) != 8)) { - printf("number of bytes: int = %lu long = %lu \n", - (long unsigned int)sizeof(size_t), - (long unsigned int)sizeof(int)); - } -#if __LITTLE_ENDIAN__ -// This is what we expect! -// printf("you have little endian machine"); -#endif -#if __BIG_ENDIAN__ - printf("you have a big endian machine"); -#endif -#if __CHAR_BIT__ - if (__CHAR_BIT__ != 8) printf("on your machine, chars don't have 8bits???"); -#endif -} - -#endif - -#endif /* INCLUDE_MISC_CONFIGREPORT_H_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/misc/configreport.h */ -/* begin file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring.h */ -/* -An implementation of Roaring Bitmaps in C. -*/ - -#ifndef ROARING_H -#define ROARING_H -#ifdef __cplusplus -extern "C" { -#endif - -#include - -typedef struct roaring_bitmap_s { - roaring_array_t high_low_container; - bool copy_on_write; /* copy_on_write: whether you want to use copy-on-write - (saves memory and avoids - copies but needs more care in a threaded context). - Most users should ignore this flag. - Note: if you do turn this flag to 'true', enabling - COW, then ensure that you do so for all of your bitmaps since - interactions between bitmaps with and without COW is unsafe. */ -} roaring_bitmap_t; - - -void *containerptr_roaring_bitmap_add(roaring_bitmap_t *r, - uint32_t val, - uint8_t *typecode, - int *index); -/** - * Creates a new bitmap (initially empty) - */ -roaring_bitmap_t *roaring_bitmap_create(void); - -/** - * Add all the values between min (included) and max (excluded) that are at a - * distance k*step from min. -*/ -roaring_bitmap_t *roaring_bitmap_from_range(uint64_t min, uint64_t max, - uint32_t step); - -/** - * Creates a new bitmap (initially empty) with a provided - * container-storage capacity (it is a performance hint). - */ -roaring_bitmap_t *roaring_bitmap_create_with_capacity(uint32_t cap); - -/** - * Creates a new bitmap from a pointer of uint32_t integers - */ -roaring_bitmap_t *roaring_bitmap_of_ptr(size_t n_args, const uint32_t *vals); - -/** - * Describe the inner structure of the bitmap. - */ -void roaring_bitmap_printf_describe(const roaring_bitmap_t *ra); - -/** - * Creates a new bitmap from a list of uint32_t integers - */ -roaring_bitmap_t *roaring_bitmap_of(size_t n, ...); - -/** - * Copies a bitmap. This does memory allocation. The caller is responsible for - * memory management. - * - */ -roaring_bitmap_t *roaring_bitmap_copy(const roaring_bitmap_t *r); - - -/** - * Copies a bitmap from src to dest. It is assumed that the pointer dest - * is to an already allocated bitmap. The content of the dest bitmap is - * freed/deleted. - * - * It might be preferable and simpler to call roaring_bitmap_copy except - * that roaring_bitmap_overwrite can save on memory allocations. - * - */ -bool roaring_bitmap_overwrite(roaring_bitmap_t *dest, - const roaring_bitmap_t *src); - -/** - * Print the content of the bitmap. - */ -void roaring_bitmap_printf(const roaring_bitmap_t *ra); - -/** - * Computes the intersection between two bitmaps and returns new bitmap. The - * caller is - * responsible for memory management. - * - */ -roaring_bitmap_t *roaring_bitmap_and(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the size of the intersection between two bitmaps. - * - */ -uint64_t roaring_bitmap_and_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - - -/** - * Check whether two bitmaps intersect. - * - */ -bool roaring_bitmap_intersect(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the Jaccard index between two bitmaps. (Also known as the Tanimoto - * distance, - * or the Jaccard similarity coefficient) - * - * The Jaccard index is undefined if both bitmaps are empty. - * - */ -double roaring_bitmap_jaccard_index(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the size of the union between two bitmaps. - * - */ -uint64_t roaring_bitmap_or_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the size of the difference (andnot) between two bitmaps. - * - */ -uint64_t roaring_bitmap_andnot_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the size of the symmetric difference (andnot) between two bitmaps. - * - */ -uint64_t roaring_bitmap_xor_cardinality(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Inplace version modifies x1, x1 == x2 is allowed - */ -void roaring_bitmap_and_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Computes the union between two bitmaps and returns new bitmap. The caller is - * responsible for memory management. - */ -roaring_bitmap_t *roaring_bitmap_or(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Inplace version of roaring_bitmap_or, modifies x1. TDOO: decide whether x1 == - *x2 ok - * - */ -void roaring_bitmap_or_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Compute the union of 'number' bitmaps. See also roaring_bitmap_or_many_heap. - * Caller is responsible for freeing the - * result. - * - */ -roaring_bitmap_t *roaring_bitmap_or_many(size_t number, - const roaring_bitmap_t **x); - -/** - * Compute the union of 'number' bitmaps using a heap. This can - * sometimes be faster than roaring_bitmap_or_many which uses - * a naive algorithm. Caller is responsible for freeing the - * result. - * - */ -roaring_bitmap_t *roaring_bitmap_or_many_heap(uint32_t number, - const roaring_bitmap_t **x); - -/** - * Computes the symmetric difference (xor) between two bitmaps - * and returns new bitmap. The caller is responsible for memory management. - */ -roaring_bitmap_t *roaring_bitmap_xor(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Inplace version of roaring_bitmap_xor, modifies x1. x1 != x2. - * - */ -void roaring_bitmap_xor_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Compute the xor of 'number' bitmaps. - * Caller is responsible for freeing the - * result. - * - */ -roaring_bitmap_t *roaring_bitmap_xor_many(size_t number, - const roaring_bitmap_t **x); - -/** - * Computes the difference (andnot) between two bitmaps - * and returns new bitmap. The caller is responsible for memory management. - */ -roaring_bitmap_t *roaring_bitmap_andnot(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * Inplace version of roaring_bitmap_andnot, modifies x1. x1 != x2. - * - */ -void roaring_bitmap_andnot_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * TODO: consider implementing: - * Compute the xor of 'number' bitmaps using a heap. This can - * sometimes be faster than roaring_bitmap_xor_many which uses - * a naive algorithm. Caller is responsible for freeing the - * result. - * - * roaring_bitmap_t *roaring_bitmap_xor_many_heap(uint32_t number, - * const roaring_bitmap_t **x); - */ - -/** - * Frees the memory. - */ -void roaring_bitmap_free(roaring_bitmap_t *r); - -/** - * Add value n_args from pointer vals, faster than repeatedly calling - * roaring_bitmap_add - * - */ -void roaring_bitmap_add_many(roaring_bitmap_t *r, size_t n_args, - const uint32_t *vals); - -/** - * Add value x - * - */ -void roaring_bitmap_add(roaring_bitmap_t *r, uint32_t x); - -/** - * Add value x - * Returns true if a new value was added, false if the value was already existing. - */ -bool roaring_bitmap_add_checked(roaring_bitmap_t *r, uint32_t x); - -/** - * Add all values in range [min, max] - */ -void roaring_bitmap_add_range_closed(roaring_bitmap_t *ra, uint32_t min, uint32_t max); - -/** - * Add all values in range [min, max) - */ -inline void roaring_bitmap_add_range(roaring_bitmap_t *ra, uint64_t min, uint64_t max) { - if(max == min) return; - roaring_bitmap_add_range_closed(ra, (uint32_t)min, (uint32_t)(max - 1)); -} - -/** - * Remove value x - * - */ -void roaring_bitmap_remove(roaring_bitmap_t *r, uint32_t x); - -/** Remove all values in range [min, max] */ -void roaring_bitmap_remove_range_closed(roaring_bitmap_t *ra, uint32_t min, uint32_t max); - -/** Remove all values in range [min, max) */ -inline void roaring_bitmap_remove_range(roaring_bitmap_t *ra, uint64_t min, uint64_t max) { - if(max == min) return; - roaring_bitmap_remove_range_closed(ra, (uint32_t)min, (uint32_t)(max - 1)); -} - -/** Remove multiple values */ -void roaring_bitmap_remove_many(roaring_bitmap_t *r, size_t n_args, - const uint32_t *vals); - -/** - * Remove value x - * Returns true if a new value was removed, false if the value was not existing. - */ -bool roaring_bitmap_remove_checked(roaring_bitmap_t *r, uint32_t x); - -/** - * Check if value x is present - */ -inline bool roaring_bitmap_contains(const roaring_bitmap_t *r, uint32_t val) { - const uint16_t hb = val >> 16; - /* - * the next function call involves a binary search and lots of branching. - */ - int32_t i = ra_get_index(&r->high_low_container, hb); - if (i < 0) return false; - - uint8_t typecode; - // next call ought to be cheap - void *container = - ra_get_container_at_index(&r->high_low_container, i, &typecode); - // rest might be a tad expensive, possibly involving another round of binary search - return container_contains(container, val & 0xFFFF, typecode); -} - -/** - * Check whether a range of values from range_start (included) to range_end (excluded) is present - */ -bool roaring_bitmap_contains_range(const roaring_bitmap_t *r, uint64_t range_start, uint64_t range_end); - -/** - * Get the cardinality of the bitmap (number of elements). - */ -uint64_t roaring_bitmap_get_cardinality(const roaring_bitmap_t *ra); - -/** - * Returns number of elements in range [range_start, range_end). - */ -uint64_t roaring_bitmap_range_cardinality(const roaring_bitmap_t *ra, - uint64_t range_start, uint64_t range_end); - -/** -* Returns true if the bitmap is empty (cardinality is zero). -*/ -bool roaring_bitmap_is_empty(const roaring_bitmap_t *ra); - - -/** -* Empties the bitmap -*/ -void roaring_bitmap_clear(roaring_bitmap_t *ra); - -/** - * Convert the bitmap to an array. Write the output to "ans", - * caller is responsible to ensure that there is enough memory - * allocated - * (e.g., ans = malloc(roaring_bitmap_get_cardinality(mybitmap) - * * sizeof(uint32_t)) - */ -void roaring_bitmap_to_uint32_array(const roaring_bitmap_t *ra, uint32_t *ans); - - -/** - * Convert the bitmap to an array from "offset" by "limit". Write the output to "ans". - * so, you can get data in paging. - * caller is responsible to ensure that there is enough memory - * allocated - * (e.g., ans = malloc(roaring_bitmap_get_cardinality(limit) - * * sizeof(uint32_t)) - * Return false in case of failure (e.g., insufficient memory) - */ -bool roaring_bitmap_range_uint32_array(const roaring_bitmap_t *ra, size_t offset, size_t limit, uint32_t *ans); - -/** - * Remove run-length encoding even when it is more space efficient - * return whether a change was applied - */ -bool roaring_bitmap_remove_run_compression(roaring_bitmap_t *r); - -/** convert array and bitmap containers to run containers when it is more - * efficient; - * also convert from run containers when more space efficient. Returns - * true if the result has at least one run container. - * Additional savings might be possible by calling shrinkToFit(). - */ -bool roaring_bitmap_run_optimize(roaring_bitmap_t *r); - -/** - * If needed, reallocate memory to shrink the memory usage. Returns - * the number of bytes saved. -*/ -size_t roaring_bitmap_shrink_to_fit(roaring_bitmap_t *r); - -/** -* write the bitmap to an output pointer, this output buffer should refer to -* at least roaring_bitmap_size_in_bytes(ra) allocated bytes. -* -* see roaring_bitmap_portable_serialize if you want a format that's compatible -* with Java and Go implementations -* -* this format has the benefit of being sometimes more space efficient than -* roaring_bitmap_portable_serialize -* e.g., when the data is sparse. -* -* Returns how many bytes were written which should be -* roaring_bitmap_size_in_bytes(ra). -*/ -size_t roaring_bitmap_serialize(const roaring_bitmap_t *ra, char *buf); - -/** use with roaring_bitmap_serialize -* see roaring_bitmap_portable_deserialize if you want a format that's -* compatible with Java and Go implementations -*/ -roaring_bitmap_t *roaring_bitmap_deserialize(const void *buf); - -/** - * How many bytes are required to serialize this bitmap (NOT compatible - * with Java and Go versions) - */ -size_t roaring_bitmap_size_in_bytes(const roaring_bitmap_t *ra); - -/** - * read a bitmap from a serialized version. This is meant to be compatible with - * the Java and Go versions. See format specification at - * https://github.com/RoaringBitmap/RoaringFormatSpec - * In case of failure, a null pointer is returned. - * This function is unsafe in the sense that if there is no valid serialized - * bitmap at the pointer, then many bytes could be read, possibly causing a buffer - * overflow. For a safer approach, - * call roaring_bitmap_portable_deserialize_safe. - */ -roaring_bitmap_t *roaring_bitmap_portable_deserialize(const char *buf); - -/** - * read a bitmap from a serialized version in a safe manner (reading up to maxbytes). - * This is meant to be compatible with - * the Java and Go versions. See format specification at - * https://github.com/RoaringBitmap/RoaringFormatSpec - * In case of failure, a null pointer is returned. - */ -roaring_bitmap_t *roaring_bitmap_portable_deserialize_safe(const char *buf, size_t maxbytes); - -/** - * Check how many bytes would be read (up to maxbytes) at this pointer if there - * is a bitmap, returns zero if there is no valid bitmap. - * This is meant to be compatible with - * the Java and Go versions. See format specification at - * https://github.com/RoaringBitmap/RoaringFormatSpec - */ -size_t roaring_bitmap_portable_deserialize_size(const char *buf, size_t maxbytes); - - -/** - * How many bytes are required to serialize this bitmap (meant to be compatible - * with Java and Go versions). See format specification at - * https://github.com/RoaringBitmap/RoaringFormatSpec - */ -size_t roaring_bitmap_portable_size_in_bytes(const roaring_bitmap_t *ra); - -/** - * write a bitmap to a char buffer. The output buffer should refer to at least - * roaring_bitmap_portable_size_in_bytes(ra) bytes of allocated memory. - * This is meant to be compatible with - * the - * Java and Go versions. Returns how many bytes were written which should be - * roaring_bitmap_portable_size_in_bytes(ra). See format specification at - * https://github.com/RoaringBitmap/RoaringFormatSpec - */ -size_t roaring_bitmap_portable_serialize(const roaring_bitmap_t *ra, char *buf); - -/** - * Iterate over the bitmap elements. The function iterator is called once for - * all the values with ptr (can be NULL) as the second parameter of each call. - * - * roaring_iterator is simply a pointer to a function that returns bool - * (true means that the iteration should continue while false means that it - * should stop), - * and takes (uint32_t,void*) as inputs. - * - * Returns true if the roaring_iterator returned true throughout (so that - * all data points were necessarily visited). - */ -bool roaring_iterate(const roaring_bitmap_t *ra, roaring_iterator iterator, - void *ptr); - -bool roaring_iterate64(const roaring_bitmap_t *ra, roaring_iterator64 iterator, - uint64_t high_bits, void *ptr); - -/** - * Return true if the two bitmaps contain the same elements. - */ -bool roaring_bitmap_equals(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2); - -/** - * Return true if all the elements of ra1 are also in ra2. - */ -bool roaring_bitmap_is_subset(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2); - -/** - * Return true if all the elements of ra1 are also in ra2 and ra2 is strictly - * greater - * than ra1. - */ -bool roaring_bitmap_is_strict_subset(const roaring_bitmap_t *ra1, - const roaring_bitmap_t *ra2); - -/** - * (For expert users who seek high performance.) - * - * Computes the union between two bitmaps and returns new bitmap. The caller is - * responsible for memory management. - * - * The lazy version defers some computations such as the maintenance of the - * cardinality counts. Thus you need - * to call roaring_bitmap_repair_after_lazy after executing "lazy" computations. - * It is safe to repeatedly call roaring_bitmap_lazy_or_inplace on the result. - * The bitsetconversion conversion is a flag which determines - * whether container-container operations force a bitset conversion. - **/ -roaring_bitmap_t *roaring_bitmap_lazy_or(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2, - const bool bitsetconversion); - -/** - * (For expert users who seek high performance.) - * Inplace version of roaring_bitmap_lazy_or, modifies x1 - * The bitsetconversion conversion is a flag which determines - * whether container-container operations force a bitset conversion. - */ -void roaring_bitmap_lazy_or_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2, - const bool bitsetconversion); - -/** - * (For expert users who seek high performance.) - * - * Execute maintenance operations on a bitmap created from - * roaring_bitmap_lazy_or - * or modified with roaring_bitmap_lazy_or_inplace. - */ -void roaring_bitmap_repair_after_lazy(roaring_bitmap_t *x1); - -/** - * Computes the symmetric difference between two bitmaps and returns new bitmap. - *The caller is - * responsible for memory management. - * - * The lazy version defers some computations such as the maintenance of the - * cardinality counts. Thus you need - * to call roaring_bitmap_repair_after_lazy after executing "lazy" computations. - * It is safe to repeatedly call roaring_bitmap_lazy_xor_inplace on the result. - * - */ -roaring_bitmap_t *roaring_bitmap_lazy_xor(const roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * (For expert users who seek high performance.) - * Inplace version of roaring_bitmap_lazy_xor, modifies x1. x1 != x2 - * - */ -void roaring_bitmap_lazy_xor_inplace(roaring_bitmap_t *x1, - const roaring_bitmap_t *x2); - -/** - * compute the negation of the roaring bitmap within a specified - * interval: [range_start, range_end). The number of negated values is - * range_end - range_start. - * Areas outside the range are passed through unchanged. - */ - -roaring_bitmap_t *roaring_bitmap_flip(const roaring_bitmap_t *x1, - uint64_t range_start, uint64_t range_end); - -/** - * compute (in place) the negation of the roaring bitmap within a specified - * interval: [range_start, range_end). The number of negated values is - * range_end - range_start. - * Areas outside the range are passed through unchanged. - */ - -void roaring_bitmap_flip_inplace(roaring_bitmap_t *x1, uint64_t range_start, - uint64_t range_end); - -/** - * If the size of the roaring bitmap is strictly greater than rank, then this - function returns true and set element to the element of given rank. - Otherwise, it returns false. - */ -bool roaring_bitmap_select(const roaring_bitmap_t *ra, uint32_t rank, - uint32_t *element); -/** -* roaring_bitmap_rank returns the number of integers that are smaller or equal -* to x. -*/ -uint64_t roaring_bitmap_rank(const roaring_bitmap_t *bm, uint32_t x); - -/** -* roaring_bitmap_smallest returns the smallest value in the set. -* Returns UINT32_MAX if the set is empty. -*/ -uint32_t roaring_bitmap_minimum(const roaring_bitmap_t *bm); - -/** -* roaring_bitmap_smallest returns the greatest value in the set. -* Returns 0 if the set is empty. -*/ -uint32_t roaring_bitmap_maximum(const roaring_bitmap_t *bm); - -/** -* (For advanced users.) -* Collect statistics about the bitmap, see roaring_types.h for -* a description of roaring_statistics_t -*/ -void roaring_bitmap_statistics(const roaring_bitmap_t *ra, - roaring_statistics_t *stat); - -/********************* -* What follows is code use to iterate through values in a roaring bitmap - -roaring_bitmap_t *ra =... -roaring_uint32_iterator_t i; -roaring_create_iterator(ra, &i); -while(i.has_value) { - printf("value = %d\n", i.current_value); - roaring_advance_uint32_iterator(&i); -} - -Obviously, if you modify the underlying bitmap, the iterator -becomes invalid. So don't. -*/ - -typedef struct roaring_uint32_iterator_s { - const roaring_bitmap_t *parent; // owner - int32_t container_index; // point to the current container index - int32_t in_container_index; // for bitset and array container, this is out - // index - int32_t run_index; // for run container, this points at the run - uint32_t in_run_index; // within a run, this is our index (points at the - // end of the current run) - - uint32_t current_value; - bool has_value; - - const void - *container; // should be: - // parent->high_low_container.containers[container_index]; - uint8_t typecode; // should be: - // parent->high_low_container.typecodes[container_index]; - uint32_t highbits; // should be: - // parent->high_low_container.keys[container_index]) << - // 16; - -} roaring_uint32_iterator_t; - -/** -* Initialize an iterator object that can be used to iterate through the -* values. If there is a value, then it->has_value is true. -* The first value is in it->current_value. The iterator traverses the values -* in increasing order. -*/ -void roaring_init_iterator(const roaring_bitmap_t *ra, - roaring_uint32_iterator_t *newit); - -/** -* Create an iterator object that can be used to iterate through the -* values. Caller is responsible for calling roaring_free_iterator. -* The iterator is initialized. If there is a value, then it->has_value is true. -* The first value is in it->current_value. The iterator traverses the values -* in increasing order. -* -* This function calls roaring_init_iterator. -*/ -roaring_uint32_iterator_t *roaring_create_iterator(const roaring_bitmap_t *ra); - -/** -* Advance the iterator. If there is a new value, then it->has_value is true. -* The new value is in it->current_value. Values are traversed in increasing -* orders. For convenience, returns it->has_value. -*/ -bool roaring_advance_uint32_iterator(roaring_uint32_iterator_t *it); - -/** -* Move the iterator to the first value >= val. If there is a such a value, then it->has_value is true. -* The new value is in it->current_value. For convenience, returns it->has_value. -*/ -bool roaring_move_uint32_iterator_equalorlarger(roaring_uint32_iterator_t *it, uint32_t val) ; -/** -* Creates a copy of an iterator. -* Caller must free it. -*/ -roaring_uint32_iterator_t *roaring_copy_uint32_iterator( - const roaring_uint32_iterator_t *it); - -/** -* Free memory following roaring_create_iterator -*/ -void roaring_free_uint32_iterator(roaring_uint32_iterator_t *it); - -/* - * Reads next ${count} values from iterator into user-supplied ${buf}. - * Returns the number of read elements. - * This number can be smaller than ${count}, which means that iterator is drained. - * - * This function satisfies semantics of iteration and can be used together with - * other iterator functions. - * - first value is copied from ${it}->current_value - * - after function returns, iterator is positioned at the next element - */ -uint32_t roaring_read_uint32_iterator(roaring_uint32_iterator_t *it, uint32_t* buf, uint32_t count); - -#ifdef __cplusplus -} -#endif - -#endif - -/* end file /opt/bitmap/CRoaring-0.2.57/include/roaring/roaring.h */ diff --git a/contrib/croaring/roaring/roaring.hh b/contrib/croaring/roaring/roaring.hh deleted file mode 100644 index 6266fff2758..00000000000 --- a/contrib/croaring/roaring/roaring.hh +++ /dev/null @@ -1,1732 +0,0 @@ -/* auto-generated on Tue Dec 18 09:42:59 CST 2018. Do not edit! */ -#include "roaring.h" -/* begin file /opt/bitmap/CRoaring-0.2.57/cpp/roaring.hh */ -/* -A C++ header for Roaring Bitmaps. -*/ -#ifndef INCLUDE_ROARING_HH_ -#define INCLUDE_ROARING_HH_ - -#include - -#include -#include -#include -#include - -class RoaringSetBitForwardIterator; - -class Roaring { - public: - /** - * Create an empty bitmap - */ - Roaring() { - bool is_ok = ra_init(&roaring.high_low_container); - if (!is_ok) { - throw std::runtime_error("failed memory alloc in constructor"); - } - roaring.copy_on_write = false; - } - - /** - * Construct a bitmap from a list of integer values. - */ - Roaring(size_t n, const uint32_t *data) : Roaring() { - roaring_bitmap_add_many(&roaring, n, data); - } - - /** - * Copy constructor - */ - Roaring(const Roaring &r) { - bool is_ok = - ra_copy(&r.roaring.high_low_container, &roaring.high_low_container, - r.roaring.copy_on_write); - if (!is_ok) { - throw std::runtime_error("failed memory alloc in constructor"); - } - roaring.copy_on_write = r.roaring.copy_on_write; - } - - /** - * Move constructor. The moved object remains valid, i.e. - * all methods can still be called on it. - */ - Roaring(Roaring &&r) { - roaring = std::move(r.roaring); - - // left the moved object in a valid state - bool is_ok = ra_init_with_capacity(&r.roaring.high_low_container, 1); - if (!is_ok) { - throw std::runtime_error("failed memory alloc in constructor"); - } - } - - /** - * Construct a roaring object from the C struct. - * - * Passing a NULL point is unsafe. - * the pointer to the C struct will be invalid after the call. - */ - Roaring(roaring_bitmap_t *s) { - // steal the interior struct - roaring.high_low_container = s->high_low_container; - roaring.copy_on_write = s->copy_on_write; - // deallocate the old container - free(s); - } - - /** - * Construct a bitmap from a list of integer values. - */ - static Roaring bitmapOf(size_t n, ...) { - Roaring ans; - va_list vl; - va_start(vl, n); - for (size_t i = 0; i < n; i++) { - ans.add(va_arg(vl, uint32_t)); - } - va_end(vl); - return ans; - } - - /** - * Add value x - * - */ - void add(uint32_t x) { roaring_bitmap_add(&roaring, x); } - - /** - * Add value x - * Returns true if a new value was added, false if the value was already existing. - */ - bool addChecked(uint32_t x) { - return roaring_bitmap_add_checked(&roaring, x); - } - - /** - * add if all values from x (included) to y (excluded) - */ - void addRange(const uint64_t x, const uint64_t y) { - return roaring_bitmap_add_range(&roaring, x, y); - } - - /** - * Add value n_args from pointer vals - * - */ - void addMany(size_t n_args, const uint32_t *vals) { - roaring_bitmap_add_many(&roaring, n_args, vals); - } - - /** - * Remove value x - * - */ - void remove(uint32_t x) { roaring_bitmap_remove(&roaring, x); } - - /** - * Remove value x - * Returns true if a new value was removed, false if the value was not existing. - */ - bool removeChecked(uint32_t x) { - return roaring_bitmap_remove_checked(&roaring, x); - } - - /** - * Return the largest value (if not empty) - * - */ - uint32_t maximum() const { return roaring_bitmap_maximum(&roaring); } - - /** - * Return the smallest value (if not empty) - * - */ - uint32_t minimum() const { return roaring_bitmap_minimum(&roaring); } - - /** - * Check if value x is present - */ - bool contains(uint32_t x) const { - return roaring_bitmap_contains(&roaring, x); - } - - /** - * Check if all values from x (included) to y (excluded) are present - */ - bool containsRange(const uint64_t x, const uint64_t y) const { - return roaring_bitmap_contains_range(&roaring, x, y); - } - - /** - * Destructor - */ - ~Roaring() { ra_clear(&roaring.high_low_container); } - - /** - * Copies the content of the provided bitmap, and - * discard the current content. - */ - Roaring &operator=(const Roaring &r) { - ra_clear(&roaring.high_low_container); - bool is_ok = - ra_copy(&r.roaring.high_low_container, &roaring.high_low_container, - r.roaring.copy_on_write); - if (!is_ok) { - throw std::runtime_error("failed memory alloc in assignment"); - } - roaring.copy_on_write = r.roaring.copy_on_write; - return *this; - } - - /** - * Moves the content of the provided bitmap, and - * discard the current content. - */ - Roaring &operator=(Roaring &&r) { - ra_clear(&roaring.high_low_container); - - roaring = std::move(r.roaring); - bool is_ok = ra_init_with_capacity(&r.roaring.high_low_container, 1); - if (!is_ok) { - throw std::runtime_error("failed memory alloc in assignment"); - } - - return *this; - } - - /** - * Compute the intersection between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring &operator&=(const Roaring &r) { - roaring_bitmap_and_inplace(&roaring, &r.roaring); - return *this; - } - - /** - * Compute the difference between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring &operator-=(const Roaring &r) { - roaring_bitmap_andnot_inplace(&roaring, &r.roaring); - return *this; - } - - /** - * Compute the union between the current bitmap and the provided bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - * - * See also the fastunion function to aggregate many bitmaps more quickly. - */ - Roaring &operator|=(const Roaring &r) { - roaring_bitmap_or_inplace(&roaring, &r.roaring); - return *this; - } - - /** - * Compute the symmetric union between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring &operator^=(const Roaring &r) { - roaring_bitmap_xor_inplace(&roaring, &r.roaring); - return *this; - } - - /** - * Exchange the content of this bitmap with another. - */ - void swap(Roaring &r) { std::swap(r.roaring, roaring); } - - /** - * Get the cardinality of the bitmap (number of elements). - */ - uint64_t cardinality() const { - return roaring_bitmap_get_cardinality(&roaring); - } - - /** - * Returns true if the bitmap is empty (cardinality is zero). - */ - bool isEmpty() const { return roaring_bitmap_is_empty(&roaring); } - - /** - * Returns true if the bitmap is subset of the other. - */ - bool isSubset(const Roaring &r) const { - return roaring_bitmap_is_subset(&roaring, &r.roaring); - } - - /** - * Returns true if the bitmap is strict subset of the other. - */ - bool isStrictSubset(const Roaring &r) const { - return roaring_bitmap_is_strict_subset(&roaring, &r.roaring); - } - - /** - * Convert the bitmap to an array. Write the output to "ans", - * caller is responsible to ensure that there is enough memory - * allocated - * (e.g., ans = new uint32[mybitmap.cardinality()];) - */ - void toUint32Array(uint32_t *ans) const { - roaring_bitmap_to_uint32_array(&roaring, ans); - } - /** - * to int array with pagination - * - */ - void rangeUint32Array(uint32_t *ans, size_t offset, size_t limit) const { - roaring_bitmap_range_uint32_array(&roaring, offset, limit, ans); - } - - /** - * Return true if the two bitmaps contain the same elements. - */ - bool operator==(const Roaring &r) const { - return roaring_bitmap_equals(&roaring, &r.roaring); - } - - /** - * compute the negation of the roaring bitmap within a specified interval. - * areas outside the range are passed through unchanged. - */ - void flip(uint64_t range_start, uint64_t range_end) { - roaring_bitmap_flip_inplace(&roaring, range_start, range_end); - } - - /** - * Remove run-length encoding even when it is more space efficient - * return whether a change was applied - */ - bool removeRunCompression() { - return roaring_bitmap_remove_run_compression(&roaring); - } - - /** convert array and bitmap containers to run containers when it is more - * efficient; - * also convert from run containers when more space efficient. Returns - * true if the result has at least one run container. - * Additional savings might be possible by calling shrinkToFit(). - */ - bool runOptimize() { return roaring_bitmap_run_optimize(&roaring); } - - /** - * If needed, reallocate memory to shrink the memory usage. Returns - * the number of bytes saved. - */ - size_t shrinkToFit() { return roaring_bitmap_shrink_to_fit(&roaring); } - - /** - * Iterate over the bitmap elements. The function iterator is called once for - * all the values with ptr (can be NULL) as the second parameter of each call. - * - * roaring_iterator is simply a pointer to a function that returns bool - * (true means that the iteration should continue while false means that it - * should stop), and takes (uint32_t,void*) as inputs. - */ - void iterate(roaring_iterator iterator, void *ptr) const { - roaring_iterate(&roaring, iterator, ptr); - } - - /** - * If the size of the roaring bitmap is strictly greater than rank, then - * this function returns true and set element to the element of given rank. - * Otherwise, it returns false. - */ - bool select(uint32_t rnk, uint32_t *element) const { - return roaring_bitmap_select(&roaring, rnk, element); - } - - /** - * Computes the size of the intersection between two bitmaps. - * - */ - uint64_t and_cardinality(const Roaring &r) const { - return roaring_bitmap_and_cardinality(&roaring, &r.roaring); - } - - /** - * Check whether the two bitmaps intersect. - * - */ - bool intersect(const Roaring &r) const { - return roaring_bitmap_intersect(&roaring, &r.roaring); - } - - /** - * Computes the Jaccard index between two bitmaps. (Also known as the - * Tanimoto distance, - * or the Jaccard similarity coefficient) - * - * The Jaccard index is undefined if both bitmaps are empty. - * - */ - double jaccard_index(const Roaring &r) const { - return roaring_bitmap_jaccard_index(&roaring, &r.roaring); - } - - /** - * Computes the size of the union between two bitmaps. - * - */ - uint64_t or_cardinality(const Roaring &r) const { - return roaring_bitmap_or_cardinality(&roaring, &r.roaring); - } - - /** - * Computes the size of the difference (andnot) between two bitmaps. - * - */ - uint64_t andnot_cardinality(const Roaring &r) const { - return roaring_bitmap_andnot_cardinality(&roaring, &r.roaring); - } - - /** - * Computes the size of the symmetric difference (andnot) between two - * bitmaps. - * - */ - uint64_t xor_cardinality(const Roaring &r) const { - return roaring_bitmap_xor_cardinality(&roaring, &r.roaring); - } - - /** - * Returns the number of integers that are smaller or equal to x. - */ - uint64_t rank(uint32_t x) const { return roaring_bitmap_rank(&roaring, x); } - - /** - * write a bitmap to a char buffer. This is meant to be compatible with - * the - * Java and Go versions. Returns how many bytes were written which should be - * getSizeInBytes(). - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - * - * Boost users can serialize bitmaps in this manner: - * - * BOOST_SERIALIZATION_SPLIT_FREE(Roaring) - * namespace boost { - * namespace serialization { - * - * template - * void save(Archive& ar, const Roaring& bitmask, - * const unsigned int version) { - * std::size_t expected_size_in_bytes = bitmask.getSizeInBytes(); - * std::vector buffer(expected_size_in_bytes); - * std::size_t size_in_bytes = bitmask.write(buffer.data()); - * - * ar& size_in_bytes; - * ar& boost::serialization::make_binary_object(buffer.data(), - * size_in_bytes); - * } - * template - * void load(Archive& ar, Roaring& bitmask, - * const unsigned int version) { - * std::size_t size_in_bytes = 0; - * ar& size_in_bytes; - * std::vector buffer(size_in_bytes); - * ar& boost::serialization::make_binary_object(buffer.data(), - * size_in_bytes); - * bitmask = Roaring::readSafe(buffer.data(), size_in_bytes); - *} - *} // namespace serialization - *} // namespace boost - */ - size_t write(char *buf, bool portable = true) const { - if (portable) - return roaring_bitmap_portable_serialize(&roaring, buf); - else - return roaring_bitmap_serialize(&roaring, buf); - } - - /** - * read a bitmap from a serialized version. This is meant to be compatible - * with the Java and Go versions. - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - * - * This function is unsafe in the sense that if you provide bad data, - * many, many bytes could be read. See also readSafe. - */ - static Roaring read(const char *buf, bool portable = true) { - roaring_bitmap_t * r = portable ? roaring_bitmap_portable_deserialize(buf) : roaring_bitmap_deserialize(buf); - if (r == NULL) { - throw std::runtime_error("failed alloc while reading"); - } - return Roaring(r); - } - /** - * read a bitmap from a serialized version, reading no more than maxbytes bytes. - * This is meant to be compatible with the Java and Go versions. - * - */ - static Roaring readSafe(const char *buf, size_t maxbytes) { - roaring_bitmap_t * r = roaring_bitmap_portable_deserialize_safe(buf,maxbytes); - if (r == NULL) { - throw std::runtime_error("failed alloc while reading"); - } - return Roaring(r); - } - /** - * How many bytes are required to serialize this bitmap (meant to be - * compatible - * with Java and Go versions) - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - */ - size_t getSizeInBytes(bool portable = true) const { - if (portable) - return roaring_bitmap_portable_size_in_bytes(&roaring); - else - return roaring_bitmap_size_in_bytes(&roaring); - } - - /** - * Computes the intersection between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring operator&(const Roaring &o) const { - roaring_bitmap_t *r = roaring_bitmap_and(&roaring, &o.roaring); - if (r == NULL) { - throw std::runtime_error("failed materalization in and"); - } - return Roaring(r); - } - - /** - * Computes the difference between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring operator-(const Roaring &o) const { - roaring_bitmap_t *r = roaring_bitmap_andnot(&roaring, &o.roaring); - if (r == NULL) { - throw std::runtime_error("failed materalization in andnot"); - } - return Roaring(r); - } - - /** - * Computes the union between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring operator|(const Roaring &o) const { - roaring_bitmap_t *r = roaring_bitmap_or(&roaring, &o.roaring); - if (r == NULL) { - throw std::runtime_error("failed materalization in or"); - } - return Roaring(r); - } - - /** - * Computes the symmetric union between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring operator^(const Roaring &o) const { - roaring_bitmap_t *r = roaring_bitmap_xor(&roaring, &o.roaring); - if (r == NULL) { - throw std::runtime_error("failed materalization in xor"); - } - return Roaring(r); - } - - /** - * Whether or not we apply copy and write. - */ - void setCopyOnWrite(bool val) { roaring.copy_on_write = val; } - - /** - * Print the content of the bitmap - */ - void printf() const { roaring_bitmap_printf(&roaring); } - - /** - * Print the content of the bitmap into a string - */ - std::string toString() const { - struct iter_data { - std::string str; - char first_char = '{'; - } outer_iter_data; - if (!isEmpty()) { - iterate( - [](uint32_t value, void *inner_iter_data) -> bool { - ((iter_data *)inner_iter_data)->str += - ((iter_data *)inner_iter_data)->first_char; - ((iter_data *)inner_iter_data)->str += - std::to_string(value); - ((iter_data *)inner_iter_data)->first_char = ','; - return true; - }, - (void *)&outer_iter_data); - } else - outer_iter_data.str = '{'; - outer_iter_data.str += '}'; - return outer_iter_data.str; - } - - /** - * Whether or not copy and write is active. - */ - bool getCopyOnWrite() const { return roaring.copy_on_write; } - - /** - * computes the logical or (union) between "n" bitmaps (referenced by a - * pointer). - */ - static Roaring fastunion(size_t n, const Roaring **inputs) { - const roaring_bitmap_t **x = - (const roaring_bitmap_t **)malloc(n * sizeof(roaring_bitmap_t *)); - if (x == NULL) { - throw std::runtime_error("failed memory alloc in fastunion"); - } - for (size_t k = 0; k < n; ++k) x[k] = &inputs[k]->roaring; - - roaring_bitmap_t *c_ans = roaring_bitmap_or_many(n, x); - if (c_ans == NULL) { - free(x); - throw std::runtime_error("failed memory alloc in fastunion"); - } - Roaring ans(c_ans); - free(x); - return ans; - } - - typedef RoaringSetBitForwardIterator const_iterator; - - /** - * Returns an iterator that can be used to access the position of the - * set bits. The running time complexity of a full scan is proportional to - * the - * number - * of set bits: be aware that if you have long strings of 1s, this can be - * very inefficient. - * - * It can be much faster to use the toArray method if you want to - * retrieve the set bits. - */ - const_iterator begin() const; - - /** - * A bogus iterator that can be used together with begin() - * for constructions such as for(auto i = b.begin(); - * i!=b.end(); ++i) {} - */ - const_iterator &end() const; - - roaring_bitmap_t roaring; -}; - -/** - * Used to go through the set bits. Not optimally fast, but convenient. - */ -class RoaringSetBitForwardIterator final { - public: - typedef std::forward_iterator_tag iterator_category; - typedef uint32_t *pointer; - typedef uint32_t &reference_type; - typedef uint32_t value_type; - typedef int32_t difference_type; - typedef RoaringSetBitForwardIterator type_of_iterator; - - /** - * Provides the location of the set bit. - */ - value_type operator*() const { return i.current_value; } - - bool operator<(const type_of_iterator &o) { - if (!i.has_value) return false; - if (!o.i.has_value) return true; - return i.current_value < *o; - } - - bool operator<=(const type_of_iterator &o) { - if (!o.i.has_value) return true; - if (!i.has_value) return false; - return i.current_value <= *o; - } - - bool operator>(const type_of_iterator &o) { - if (!o.i.has_value) return false; - if (!i.has_value) return true; - return i.current_value > *o; - } - - bool operator>=(const type_of_iterator &o) { - if (!i.has_value) return true; - if (!o.i.has_value) return false; - return i.current_value >= *o; - } - - /** - * Move the iterator to the first value >= val. - */ - void equalorlarger(uint32_t val) { - roaring_move_uint32_iterator_equalorlarger(&i,val); - } - - type_of_iterator &operator++() { // ++i, must returned inc. value - roaring_advance_uint32_iterator(&i); - return *this; - } - - type_of_iterator operator++(int) { // i++, must return orig. value - RoaringSetBitForwardIterator orig(*this); - roaring_advance_uint32_iterator(&i); - return orig; - } - - bool operator==(const RoaringSetBitForwardIterator &o) const { - return i.current_value == *o && i.has_value == o.i.has_value; - } - - bool operator!=(const RoaringSetBitForwardIterator &o) const { - return i.current_value != *o || i.has_value != o.i.has_value; - } - - RoaringSetBitForwardIterator(const Roaring &parent, - bool exhausted = false) { - if (exhausted) { - i.parent = &parent.roaring; - i.container_index = INT32_MAX; - i.has_value = false; - i.current_value = UINT32_MAX; - } else { - roaring_init_iterator(&parent.roaring, &i); - } - } - - RoaringSetBitForwardIterator &operator=( - const RoaringSetBitForwardIterator &o) = default; - RoaringSetBitForwardIterator &operator=(RoaringSetBitForwardIterator &&o) = - default; - - ~RoaringSetBitForwardIterator() = default; - - RoaringSetBitForwardIterator(const RoaringSetBitForwardIterator &o) - : i(o.i) {} - - roaring_uint32_iterator_t i; -}; - -inline RoaringSetBitForwardIterator Roaring::begin() const { - return RoaringSetBitForwardIterator(*this); -} - -inline RoaringSetBitForwardIterator &Roaring::end() const { - static RoaringSetBitForwardIterator e(*this, true); - return e; -} - -#endif /* INCLUDE_ROARING_HH_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/cpp/roaring.hh */ -/* begin file /opt/bitmap/CRoaring-0.2.57/cpp/roaring64map.hh */ -/* -A C++ header for 64-bit Roaring Bitmaps, implemented by way of a map of many -32-bit Roaring Bitmaps. -*/ -#ifndef INCLUDE_ROARING_64_MAP_HH_ -#define INCLUDE_ROARING_64_MAP_HH_ - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - - -class Roaring64MapSetBitForwardIterator; - -class Roaring64Map { - public: - /** - * Create an empty bitmap - */ - Roaring64Map() = default; - - /** - * Construct a bitmap from a list of 32-bit integer values. - */ - Roaring64Map(size_t n, const uint32_t *data) { addMany(n, data); } - - /** - * Construct a bitmap from a list of 64-bit integer values. - */ - Roaring64Map(size_t n, const uint64_t *data) { addMany(n, data); } - - /** - * Copy constructor - */ - Roaring64Map(const Roaring64Map &r) = default; - - /** - * Move constructor - */ - Roaring64Map(Roaring64Map &&r) = default; - - /** - * Construct a 64-bit map from a 32-bit one - */ - Roaring64Map(const Roaring &r) { emplaceOrInsert(0, r); } - - /** - * Construct a roaring object from the C struct. - * - * Passing a NULL point is unsafe. - */ - Roaring64Map(roaring_bitmap_t *s) { emplaceOrInsert(0, s); } - - /** - * Construct a bitmap from a list of integer values. - */ - static Roaring64Map bitmapOf(size_t n...) { - Roaring64Map ans; - va_list vl; - va_start(vl, n); - for (size_t i = 0; i < n; i++) { - ans.add(va_arg(vl, uint64_t)); - } - va_end(vl); - return ans; - } - - /** - * Add value x - * - */ - void add(uint32_t x) { - roarings[0].add(x); - roarings[0].setCopyOnWrite(copyOnWrite); - } - void add(uint64_t x) { - roarings[highBytes(x)].add(lowBytes(x)); - roarings[highBytes(x)].setCopyOnWrite(copyOnWrite); - } - - /** - * Add value x - * Returns true if a new value was added, false if the value was already existing. - */ - bool addChecked(uint32_t x) { - bool result = roarings[0].addChecked(x); - roarings[0].setCopyOnWrite(copyOnWrite); - return result; - } - bool addChecked(uint64_t x) { - bool result = roarings[highBytes(x)].addChecked(lowBytes(x)); - roarings[highBytes(x)].setCopyOnWrite(copyOnWrite); - return result; - } - - /** - * Add value n_args from pointer vals - * - */ - void addMany(size_t n_args, const uint32_t *vals) { - for (size_t lcv = 0; lcv < n_args; lcv++) { - roarings[0].add(vals[lcv]); - roarings[0].setCopyOnWrite(copyOnWrite); - } - } - void addMany(size_t n_args, const uint64_t *vals) { - for (size_t lcv = 0; lcv < n_args; lcv++) { - roarings[highBytes(vals[lcv])].add(lowBytes(vals[lcv])); - roarings[highBytes(vals[lcv])].setCopyOnWrite(copyOnWrite); - } - } - - /** - * Remove value x - * - */ - void remove(uint32_t x) { roarings[0].remove(x); } - void remove(uint64_t x) { - auto roaring_iter = roarings.find(highBytes(x)); - if (roaring_iter != roarings.cend()) - roaring_iter->second.remove(lowBytes(x)); - } - - /** - * Remove value x - * Returns true if a new value was removed, false if the value was not existing. - */ - bool removeChecked(uint32_t x) { - return roarings[0].removeChecked(x); - } - bool removeChecked(uint64_t x) { - auto roaring_iter = roarings.find(highBytes(x)); - if (roaring_iter != roarings.cend()) - return roaring_iter->second.removeChecked(lowBytes(x)); - return false; - } - - /** - * Return the largest value (if not empty) - * - */ - uint64_t maximum() const { - for (auto roaring_iter = roarings.crbegin(); - roaring_iter != roarings.crend(); ++roaring_iter) { - if (!roaring_iter->second.isEmpty()) { - return uniteBytes(roaring_iter->first, - roaring_iter->second.maximum()); - } - } - // we put std::numeric_limits<>::max/min in parenthesis - // to avoid a clash with the Windows.h header under Windows - return (std::numeric_limits::min)(); - } - - /** - * Return the smallest value (if not empty) - * - */ - uint64_t minimum() const { - for (auto roaring_iter = roarings.cbegin(); - roaring_iter != roarings.cend(); ++roaring_iter) { - if (!roaring_iter->second.isEmpty()) { - return uniteBytes(roaring_iter->first, - roaring_iter->second.minimum()); - } - } - // we put std::numeric_limits<>::max/min in parenthesis - // to avoid a clash with the Windows.h header under Windows - return (std::numeric_limits::max)(); - } - - /** - * Check if value x is present - */ - bool contains(uint32_t x) const { - return roarings.count(0) == 0 ? false : roarings.at(0).contains(x); - } - bool contains(uint64_t x) const { - return roarings.count(highBytes(x)) == 0 - ? false - : roarings.at(highBytes(x)).contains(lowBytes(x)); - } - - /** - * Destructor - */ - ~Roaring64Map() = default; - - /** - * Copies the content of the provided bitmap, and - * discards the current content. - */ - Roaring64Map &operator=(const Roaring64Map &r) { - roarings = r.roarings; - copyOnWrite = r.copyOnWrite; - return *this; - } - - /** - * Moves the content of the provided bitmap, and - * discards the current content. - */ - Roaring64Map &operator=(Roaring64Map &&r) { - roarings = std::move(r.roarings); - copyOnWrite = r.copyOnWrite; - return *this; - } - - /** - * Compute the intersection between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring64Map &operator&=(const Roaring64Map &r) { - for (auto &map_entry : roarings) { - if (r.roarings.count(map_entry.first) == 1) - map_entry.second &= r.roarings.at(map_entry.first); - else - map_entry.second = Roaring(); - } - return *this; - } - - /** - * Compute the difference between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring64Map &operator-=(const Roaring64Map &r) { - for (auto &map_entry : roarings) { - if (r.roarings.count(map_entry.first) == 1) - map_entry.second -= r.roarings.at(map_entry.first); - } - return *this; - } - - /** - * Compute the union between the current bitmap and the provided bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - * - * See also the fastunion function to aggregate many bitmaps more quickly. - */ - Roaring64Map &operator|=(const Roaring64Map &r) { - for (const auto &map_entry : r.roarings) { - if (roarings.count(map_entry.first) == 0) { - roarings[map_entry.first] = map_entry.second; - roarings[map_entry.first].setCopyOnWrite(copyOnWrite); - } else - roarings[map_entry.first] |= map_entry.second; - } - return *this; - } - - /** - * Compute the symmetric union between the current bitmap and the provided - * bitmap, - * writing the result in the current bitmap. The provided bitmap is not - * modified. - */ - Roaring64Map &operator^=(const Roaring64Map &r) { - for (const auto &map_entry : r.roarings) { - if (roarings.count(map_entry.first) == 0) { - roarings[map_entry.first] = map_entry.second; - roarings[map_entry.first].setCopyOnWrite(copyOnWrite); - } else - roarings[map_entry.first] ^= map_entry.second; - } - return *this; - } - - /** - * Exchange the content of this bitmap with another. - */ - void swap(Roaring64Map &r) { roarings.swap(r.roarings); } - - /** - * Get the cardinality of the bitmap (number of elements). - * Throws std::length_error in the special case where the bitmap is full - * (cardinality() == 2^64). Check isFull() before calling to avoid - * exception. - */ - uint64_t cardinality() const { - if (isFull()) { - throw std::length_error( - "bitmap is full, cardinality is 2^64, " - "unable to represent in a 64-bit integer"); - } - return std::accumulate( - roarings.cbegin(), roarings.cend(), (uint64_t)0, - [](uint64_t previous, - const std::pair &map_entry) { - return previous + map_entry.second.cardinality(); - }); - } - - /** - * Returns true if the bitmap is empty (cardinality is zero). - */ - bool isEmpty() const { - return std::all_of(roarings.cbegin(), roarings.cend(), - [](const std::pair &map_entry) { - return map_entry.second.isEmpty(); - }); - } - - /** - * Returns true if the bitmap is full (cardinality is max uint64_t + 1). - */ - bool isFull() const { - // only bother to check if map is fully saturated - // - // we put std::numeric_limits<>::max/min in parenthesis - // to avoid a clash with the Windows.h header under Windows - return roarings.size() == - ((size_t)(std::numeric_limits::max)()) + 1 - ? std::all_of( - roarings.cbegin(), roarings.cend(), - [](const std::pair &roaring_map_entry) { - // roarings within map are saturated if cardinality - // is uint32_t max + 1 - return roaring_map_entry.second.cardinality() == - ((uint64_t) - (std::numeric_limits::max)()) + - 1; - }) - : false; - } - - /** - * Returns true if the bitmap is subset of the other. - */ - bool isSubset(const Roaring64Map &r) const { - for (const auto &map_entry : roarings) { - auto roaring_iter = r.roarings.find(map_entry.first); - if (roaring_iter == roarings.cend()) - return false; - else if (!map_entry.second.isSubset(roaring_iter->second)) - return false; - } - return true; - } - - /** - * Returns true if the bitmap is strict subset of the other. - * Throws std::length_error in the special case where the bitmap is full - * (cardinality() == 2^64). Check isFull() before calling to avoid exception. - */ - bool isStrictSubset(const Roaring64Map &r) const { - return isSubset(r) && cardinality() != r.cardinality(); - } - - /** - * Convert the bitmap to an array. Write the output to "ans", - * caller is responsible to ensure that there is enough memory - * allocated - * (e.g., ans = new uint32[mybitmap.cardinality()];) - */ - void toUint64Array(uint64_t *ans) const { - // Annoyingly, VS 2017 marks std::accumulate() as [[nodiscard]] - (void)std::accumulate(roarings.cbegin(), roarings.cend(), ans, - [](uint64_t *previous, - const std::pair &map_entry) { - for (uint32_t low_bits : map_entry.second) - *previous++ = - uniteBytes(map_entry.first, low_bits); - return previous; - }); - } - - /** - * Return true if the two bitmaps contain the same elements. - */ - bool operator==(const Roaring64Map &r) const { - // we cannot use operator == on the map because either side may contain - // empty Roaring Bitmaps - auto lhs_iter = roarings.cbegin(); - auto rhs_iter = r.roarings.cbegin(); - do { - // if the left map has reached its end, ensure that the right map - // contains only empty Bitmaps - if (lhs_iter == roarings.cend()) { - while (rhs_iter != r.roarings.cend()) { - if (rhs_iter->second.isEmpty()) { - ++rhs_iter; - continue; - } - return false; - } - return true; - } - // if the left map has an empty bitmap, skip it - if (lhs_iter->second.isEmpty()) { - ++lhs_iter; - continue; - } - - do { - // if the right map has reached its end, ensure that the right - // map contains only empty Bitmaps - if (rhs_iter == r.roarings.cend()) { - while (lhs_iter != roarings.cend()) { - if (lhs_iter->second.isEmpty()) { - ++lhs_iter; - continue; - } - return false; - } - return true; - } - // if the right map has an empty bitmap, skip it - if (rhs_iter->second.isEmpty()) { - ++rhs_iter; - continue; - } - } while (false); - // if neither map has reached its end ensure elements are equal and - // move to the next element in both - } while (lhs_iter++->second == rhs_iter++->second); - return false; - } - - /** - * compute the negation of the roaring bitmap within a specified interval. - * areas outside the range are passed through unchanged. - */ - void flip(uint64_t range_start, uint64_t range_end) { - uint32_t start_high = highBytes(range_start); - uint32_t start_low = lowBytes(range_start); - uint32_t end_high = highBytes(range_end); - uint32_t end_low = lowBytes(range_end); - - if (start_high == end_high) { - roarings[start_high].flip(start_low, end_low); - return; - } - // we put std::numeric_limits<>::max/min in parenthesis - // to avoid a clash with the Windows.h header under Windows - roarings[start_high].flip(start_low, - (std::numeric_limits::max)()); - roarings[start_high++].setCopyOnWrite(copyOnWrite); - - for (; start_high <= highBytes(range_end) - 1; ++start_high) { - roarings[start_high].flip((std::numeric_limits::min)(), - (std::numeric_limits::max)()); - roarings[start_high].setCopyOnWrite(copyOnWrite); - } - - roarings[start_high].flip((std::numeric_limits::min)(), - end_low); - roarings[start_high].setCopyOnWrite(copyOnWrite); - } - - /** - * Remove run-length encoding even when it is more space efficient - * return whether a change was applied - */ - bool removeRunCompression() { - return std::accumulate( - roarings.begin(), roarings.end(), false, - [](bool previous, std::pair &map_entry) { - return map_entry.second.removeRunCompression() && previous; - }); - } - - /** convert array and bitmap containers to run containers when it is more - * efficient; - * also convert from run containers when more space efficient. Returns - * true if the result has at least one run container. - * Additional savings might be possible by calling shrinkToFit(). - */ - bool runOptimize() { - return std::accumulate( - roarings.begin(), roarings.end(), false, - [](bool previous, std::pair &map_entry) { - return map_entry.second.runOptimize() && previous; - }); - } - - /** - * If needed, reallocate memory to shrink the memory usage. Returns - * the number of bytes saved. - */ - size_t shrinkToFit() { - size_t savedBytes = 0; - auto iter = roarings.begin(); - while (iter != roarings.cend()) { - if (iter->second.isEmpty()) { - // empty Roarings are 84 bytes - savedBytes += 88; - roarings.erase(iter++); - } else { - savedBytes += iter->second.shrinkToFit(); - iter++; - } - } - return savedBytes; - } - - /** - * Iterate over the bitmap elements. The function iterator is called once - * for all the values with ptr (can be NULL) as the second parameter of each - * call. - * - * roaring_iterator is simply a pointer to a function that returns bool - * (true means that the iteration should continue while false means that it - * should stop), and takes (uint32_t,void*) as inputs. - */ - void iterate(roaring_iterator64 iterator, void *ptr) const { - std::for_each(roarings.begin(), roarings.cend(), - [=](const std::pair &map_entry) { - roaring_iterate64(&map_entry.second.roaring, iterator, - uint64_t(map_entry.first) << 32, - ptr); - }); - } - - /** - * If the size of the roaring bitmap is strictly greater than rank, then - this - function returns true and set element to the element of given rank. - Otherwise, it returns false. - */ - bool select(uint64_t rnk, uint64_t *element) const { - for (const auto &map_entry : roarings) { - uint64_t sub_cardinality = (uint64_t)map_entry.second.cardinality(); - if (rnk < sub_cardinality) { - *element = ((uint64_t)map_entry.first) << 32; - // assuming little endian - return map_entry.second.select((uint32_t)rnk, - ((uint32_t *)element)); - } - rnk -= sub_cardinality; - } - return false; - } - - /** - * Returns the number of integers that are smaller or equal to x. - */ - uint64_t rank(uint64_t x) const { - uint64_t result = 0; - auto roaring_destination = roarings.find(highBytes(x)); - if (roaring_destination != roarings.cend()) { - for (auto roaring_iter = roarings.cbegin(); - roaring_iter != roaring_destination; ++roaring_iter) { - result += roaring_iter->second.cardinality(); - } - result += roaring_destination->second.rank(lowBytes(x)); - return result; - } - roaring_destination = roarings.lower_bound(highBytes(x)); - for (auto roaring_iter = roarings.cbegin(); - roaring_iter != roaring_destination; ++roaring_iter) { - result += roaring_iter->second.cardinality(); - } - return result; - } - - /** - * write a bitmap to a char buffer. This is meant to be compatible with - * the - * Java and Go versions. Returns how many bytes were written which should be - * getSizeInBytes(). - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - */ - size_t write(char *buf, bool portable = true) const { - const char *orig = buf; - // push map size - *((uint64_t *)buf) = roarings.size(); - buf += sizeof(uint64_t); - std::for_each( - roarings.cbegin(), roarings.cend(), - [&buf, portable](const std::pair &map_entry) { - // push map key - memcpy(buf, &map_entry.first, - sizeof(uint32_t)); // this is undefined: - // *((uint32_t*)buf) = - // map_entry.first; - buf += sizeof(uint32_t); - // push map value Roaring - buf += map_entry.second.write(buf, portable); - }); - return buf - orig; - } - - /** - * read a bitmap from a serialized version. This is meant to be compatible - * with - * the - * Java and Go versions. - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - * - * This function is unsafe in the sense that if you provide bad data, - * many bytes could be read, possibly causing a buffer overflow. See also readSafe. - */ - static Roaring64Map read(const char *buf, bool portable = true) { - Roaring64Map result; - // get map size - uint64_t map_size = *((uint64_t *)buf); - buf += sizeof(uint64_t); - for (uint64_t lcv = 0; lcv < map_size; lcv++) { - // get map key - uint32_t key; - memcpy(&key, buf, sizeof(uint32_t)); // this is undefined: uint32_t - // key = *((uint32_t*)buf); - buf += sizeof(uint32_t); - // read map value Roaring - Roaring read = Roaring::read(buf, portable); - result.emplaceOrInsert(key, read); - // forward buffer past the last Roaring Bitmap - buf += read.getSizeInBytes(portable); - } - return result; - } - - /** - * read a bitmap from a serialized version, reading no more than maxbytes bytes. - * This is meant to be compatible with the Java and Go versions. - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - */ - static Roaring64Map readSafe(const char *buf, size_t maxbytes) { - Roaring64Map result; - // get map size - uint64_t map_size = *((uint64_t *)buf); - buf += sizeof(uint64_t); - for (uint64_t lcv = 0; lcv < map_size; lcv++) { - // get map key - if(maxbytes < sizeof(uint32_t)) { - throw std::runtime_error("ran out of bytes"); - } - uint32_t key; - memcpy(&key, buf, sizeof(uint32_t)); // this is undefined: uint32_t - // key = *((uint32_t*)buf); - buf += sizeof(uint32_t); - maxbytes -= sizeof(uint32_t); - // read map value Roaring - Roaring read = Roaring::readSafe(buf, maxbytes); - result.emplaceOrInsert(key, read); - // forward buffer past the last Roaring Bitmap - size_t tz = read.getSizeInBytes(true); - buf += tz; - maxbytes -= tz; - } - return result; - } - - /** - * How many bytes are required to serialize this bitmap (meant to be - * compatible - * with Java and Go versions) - * - * Setting the portable flag to false enable a custom format that - * can save space compared to the portable format (e.g., for very - * sparse bitmaps). - */ - size_t getSizeInBytes(bool portable = true) const { - // start with, respectively, map size and size of keys for each map - // entry - return std::accumulate( - roarings.cbegin(), roarings.cend(), - sizeof(uint64_t) + roarings.size() * sizeof(uint32_t), - [=](size_t previous, - const std::pair &map_entry) { - // add in bytes used by each Roaring - return previous + map_entry.second.getSizeInBytes(portable); - }); - } - - /** - * Computes the intersection between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring64Map operator&(const Roaring64Map &o) const { - return Roaring64Map(*this) &= o; - } - - /** - * Computes the difference between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring64Map operator-(const Roaring64Map &o) const { - return Roaring64Map(*this) -= o; - } - - /** - * Computes the union between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring64Map operator|(const Roaring64Map &o) const { - return Roaring64Map(*this) |= o; - } - - /** - * Computes the symmetric union between two bitmaps and returns new bitmap. - * The current bitmap and the provided bitmap are unchanged. - */ - Roaring64Map operator^(const Roaring64Map &o) const { - return Roaring64Map(*this) ^= o; - } - - /** - * Whether or not we apply copy and write. - */ - void setCopyOnWrite(bool val) { - if (copyOnWrite == val) return; - copyOnWrite = val; - std::for_each(roarings.begin(), roarings.end(), - [=](std::pair &map_entry) { - map_entry.second.setCopyOnWrite(val); - }); - } - - /** - * Print the content of the bitmap - */ - void printf() const { - if (!isEmpty()) { - auto map_iter = roarings.cbegin(); - while (map_iter->second.isEmpty()) ++map_iter; - struct iter_data { - uint32_t high_bits; - char first_char = '{'; - } outer_iter_data; - outer_iter_data.high_bits = roarings.begin()->first; - map_iter->second.iterate( - [](uint32_t low_bits, void *inner_iter_data) -> bool { - std::printf("%c%llu", - ((iter_data *)inner_iter_data)->first_char, - (long long unsigned)uniteBytes( - ((iter_data *)inner_iter_data)->high_bits, - low_bits)); - ((iter_data *)inner_iter_data)->first_char = ','; - return true; - }, - (void *)&outer_iter_data); - std::for_each( - ++map_iter, roarings.cend(), - [](const std::pair &map_entry) { - map_entry.second.iterate( - [](uint32_t low_bits, void *high_bits) -> bool { - std::printf(",%llu", - (long long unsigned)uniteBytes( - *(uint32_t *)high_bits, low_bits)); - return true; - }, - (void *)&map_entry.first); - }); - } else - std::printf("{"); - std::printf("}\n"); - } - - /** - * Print the content of the bitmap into a string - */ - std::string toString() const { - struct iter_data { - std::string str; - uint32_t high_bits; - char first_char = '{'; - } outer_iter_data; - if (!isEmpty()) { - auto map_iter = roarings.cbegin(); - while (map_iter->second.isEmpty()) ++map_iter; - outer_iter_data.high_bits = roarings.begin()->first; - map_iter->second.iterate( - [](uint32_t low_bits, void *inner_iter_data) -> bool { - ((iter_data *)inner_iter_data)->str += - ((iter_data *)inner_iter_data)->first_char; - ((iter_data *)inner_iter_data)->str += std::to_string( - uniteBytes(((iter_data *)inner_iter_data)->high_bits, - low_bits)); - ((iter_data *)inner_iter_data)->first_char = ','; - return true; - }, - (void *)&outer_iter_data); - std::for_each( - ++map_iter, roarings.cend(), - [&outer_iter_data]( - const std::pair &map_entry) { - outer_iter_data.high_bits = map_entry.first; - map_entry.second.iterate( - [](uint32_t low_bits, void *inner_iter_data) -> bool { - ((iter_data *)inner_iter_data)->str += - ((iter_data *)inner_iter_data)->first_char; - ((iter_data *)inner_iter_data)->str += - std::to_string(uniteBytes( - ((iter_data *)inner_iter_data)->high_bits, - low_bits)); - return true; - }, - (void *)&outer_iter_data); - }); - } else - outer_iter_data.str = '{'; - outer_iter_data.str += '}'; - return outer_iter_data.str; - } - - /** - * Whether or not copy and write is active. - */ - bool getCopyOnWrite() const { return copyOnWrite; } - - /** - * computes the logical or (union) between "n" bitmaps (referenced by a - * pointer). - */ - static Roaring64Map fastunion(size_t n, const Roaring64Map **inputs) { - Roaring64Map ans; - // not particularly fast - for (size_t lcv = 0; lcv < n; ++lcv) { - ans |= *(inputs[lcv]); - } - return ans; - } - - friend class Roaring64MapSetBitForwardIterator; - typedef Roaring64MapSetBitForwardIterator const_iterator; - - /** - * Returns an iterator that can be used to access the position of the - * set bits. The running time complexity of a full scan is proportional to - * the - * number - * of set bits: be aware that if you have long strings of 1s, this can be - * very inefficient. - * - * It can be much faster to use the toArray method if you want to - * retrieve the set bits. - */ - const_iterator begin() const; - - /** - * A bogus iterator that can be used together with begin() - * for constructions such as for(auto i = b.begin(); - * i!=b.end(); ++i) {} - */ - const_iterator end() const; - - private: - std::map roarings; - bool copyOnWrite = false; - static uint32_t highBytes(const uint64_t in) { return uint32_t(in >> 32); } - static uint32_t lowBytes(const uint64_t in) { return uint32_t(in); } - static uint64_t uniteBytes(const uint32_t highBytes, - const uint32_t lowBytes) { - return (uint64_t(highBytes) << 32) | uint64_t(lowBytes); - } - // this is needed to tolerate gcc's C++11 libstdc++ lacking emplace - // prior to version 4.8 - void emplaceOrInsert(const uint32_t key, const Roaring &value) { -#if defined(__GLIBCXX__) && __GLIBCXX__ < 20130322 - roarings.insert(std::make_pair(key, value)); -#else - roarings.emplace(std::make_pair(key, value)); -#endif - } -}; - -/** - * Used to go through the set bits. Not optimally fast, but convenient. - */ -class Roaring64MapSetBitForwardIterator final { - public: - typedef std::forward_iterator_tag iterator_category; - typedef uint64_t *pointer; - typedef uint64_t &reference_type; - typedef uint64_t value_type; - typedef int64_t difference_type; - typedef Roaring64MapSetBitForwardIterator type_of_iterator; - - /** - * Provides the location of the set bit. - */ - value_type operator*() const { - return Roaring64Map::uniteBytes(map_iter->first, i.current_value); - } - - bool operator<(const type_of_iterator &o) { - if (map_iter == map_end) return false; - if (o.map_iter == o.map_end) return true; - return **this < *o; - } - - bool operator<=(const type_of_iterator &o) { - if (o.map_iter == o.map_end) return true; - if (map_iter == map_end) return false; - return **this <= *o; - } - - bool operator>(const type_of_iterator &o) { - if (o.map_iter == o.map_end) return false; - if (map_iter == map_end) return true; - return **this > *o; - } - - bool operator>=(const type_of_iterator &o) { - if (map_iter == map_end) return true; - if (o.map_iter == o.map_end) return false; - return **this >= *o; - } - - type_of_iterator &operator++() { // ++i, must returned inc. value - if (i.has_value == true) roaring_advance_uint32_iterator(&i); - while (!i.has_value) { - map_iter++; - if (map_iter == map_end) return *this; - roaring_init_iterator(&map_iter->second.roaring, &i); - } - return *this; - } - - type_of_iterator operator++(int) { // i++, must return orig. value - Roaring64MapSetBitForwardIterator orig(*this); - roaring_advance_uint32_iterator(&i); - while (!i.has_value) { - map_iter++; - if (map_iter == map_end) return orig; - roaring_init_iterator(&map_iter->second.roaring, &i); - } - return orig; - } - - bool operator==(const Roaring64MapSetBitForwardIterator &o) { - if (map_iter == map_end && o.map_iter == o.map_end) return true; - if (o.map_iter == o.map_end) return false; - return **this == *o; - } - - bool operator!=(const Roaring64MapSetBitForwardIterator &o) { - if (map_iter == map_end && o.map_iter == o.map_end) return false; - if (o.map_iter == o.map_end) return true; - return **this != *o; - } - - Roaring64MapSetBitForwardIterator(const Roaring64Map &parent, - bool exhausted = false) - : map_end(parent.roarings.cend()) { - if (exhausted || parent.roarings.empty()) { - map_iter = parent.roarings.cend(); - } else { - map_iter = parent.roarings.cbegin(); - roaring_init_iterator(&map_iter->second.roaring, &i); - while (!i.has_value) { - map_iter++; - if (map_iter == map_end) return; - roaring_init_iterator(&map_iter->second.roaring, &i); - } - } - } - - ~Roaring64MapSetBitForwardIterator() = default; - - Roaring64MapSetBitForwardIterator( - const Roaring64MapSetBitForwardIterator &o) = default; - - private: - std::map::const_iterator map_iter; - std::map::const_iterator map_end; - roaring_uint32_iterator_t i; -}; - -inline Roaring64MapSetBitForwardIterator Roaring64Map::begin() const { - return Roaring64MapSetBitForwardIterator(*this); -} - -inline Roaring64MapSetBitForwardIterator Roaring64Map::end() const { - return Roaring64MapSetBitForwardIterator(*this, true); -} - -#endif /* INCLUDE_ROARING_64_MAP_HH_ */ -/* end file /opt/bitmap/CRoaring-0.2.57/cpp/roaring64map.hh */ diff --git a/contrib/libhdfs3-cmake/CMakeLists.txt b/contrib/libhdfs3-cmake/CMakeLists.txt index ab4857f0387..3aa5cb970db 100644 --- a/contrib/libhdfs3-cmake/CMakeLists.txt +++ b/contrib/libhdfs3-cmake/CMakeLists.txt @@ -192,7 +192,7 @@ set(SRCS ${HDFS3_SOURCE_DIR}/common/FileWrapper.h ) -# old kernels (< 3.17) doens't have SYS_getrandom. Always use POSIX implementation to have better compatibility +# old kernels (< 3.17) doesn't have SYS_getrandom. Always use POSIX implementation to have better compatibility set_source_files_properties(${HDFS3_SOURCE_DIR}/rpc/RpcClient.cpp PROPERTIES COMPILE_FLAGS "-DBOOST_UUID_RANDOM_PROVIDER_FORCE_POSIX=1") # target diff --git a/contrib/mariadb-connector-c b/contrib/mariadb-connector-c index f5638e954a7..1485b0de3ea 160000 --- a/contrib/mariadb-connector-c +++ b/contrib/mariadb-connector-c @@ -1 +1 @@ -Subproject commit f5638e954a79f50bac7c7a5deaa5a241e0ce8b5f +Subproject commit 1485b0de3eaa1508dfe49a5ba1e4aa2a71fd8335 diff --git a/debian/clickhouse-server.init b/debian/clickhouse-server.init index f56164759bf..8f10153a682 100755 --- a/debian/clickhouse-server.init +++ b/debian/clickhouse-server.init @@ -153,82 +153,19 @@ initdb() start() { - [ -x $CLICKHOUSE_BINDIR/$PROGRAM ] || exit 0 - local EXIT_STATUS - EXIT_STATUS=0 - - echo -n "Start $PROGRAM service: " - - if is_running; then - echo -n "already running " - EXIT_STATUS=1 - else - ulimit -n 262144 - mkdir -p $CLICKHOUSE_PIDDIR - chown -R $CLICKHOUSE_USER:$CLICKHOUSE_GROUP $CLICKHOUSE_PIDDIR - initdb - if ! is_running; then - # Lock should not be held while running child process, so we release the lock. Note: obviously, there is race condition. - # But clickhouse-server has protection from simultaneous runs with same data directory. - su -s $SHELL ${CLICKHOUSE_USER} -c "$FLOCK -u 9; $CLICKHOUSE_PROGRAM_ENV exec -a \"$PROGRAM\" \"$CLICKHOUSE_BINDIR/$PROGRAM\" --daemon --pid-file=\"$CLICKHOUSE_PIDFILE\" --config-file=\"$CLICKHOUSE_CONFIG\"" - EXIT_STATUS=$? - if [ $EXIT_STATUS -ne 0 ]; then - return $EXIT_STATUS - fi - fi - fi - - if [ $EXIT_STATUS -eq 0 ]; then - attempts=0 - while ! is_running && [ $attempts -le ${CLICKHOUSE_START_TIMEOUT:=10} ]; do - attempts=$(($attempts + 1)) - sleep 1 - done - if is_running; then - echo "DONE" - else - echo "UNKNOWN" - fi - else - echo "FAILED" - fi - - return $EXIT_STATUS + ${CLICKHOUSE_GENERIC_PROGRAM} start --user "${CLICKHOUSE_USER}" --pid-path "${CLICKHOUSE_PIDDIR}" --config-path "${CLICKHOUSE_CONFDIR}" --binary-path "${CLICKHOUSE_BINDIR}" } stop() { - #local EXIT_STATUS - EXIT_STATUS=0 - - if [ -f $CLICKHOUSE_PIDFILE ]; then - - echo -n "Stop $PROGRAM service: " - - kill -TERM $(cat "$CLICKHOUSE_PIDFILE") - - if ! wait_for_done ${CLICKHOUSE_STOP_TIMEOUT}; then - EXIT_STATUS=2 - echo "TIMEOUT" - else - echo "DONE" - fi - - fi - return $EXIT_STATUS + ${CLICKHOUSE_GENERIC_PROGRAM} stop --pid-path "${CLICKHOUSE_PIDDIR}" } restart() { - check_config - if stop; then - if start; then - return 0 - fi - fi - return 1 + ${CLICKHOUSE_GENERIC_PROGRAM} restart --user "${CLICKHOUSE_USER}" --pid-path "${CLICKHOUSE_PIDDIR}" --config-path "${CLICKHOUSE_CONFDIR}" --binary-path "${CLICKHOUSE_BINDIR}" } diff --git a/debian/clickhouse-server.postinst b/debian/clickhouse-server.postinst index 6e031ae8f44..dc876f45954 100644 --- a/debian/clickhouse-server.postinst +++ b/debian/clickhouse-server.postinst @@ -2,6 +2,7 @@ set -e # set -x +PROGRAM=clickhouse-server CLICKHOUSE_USER=${CLICKHOUSE_USER:=clickhouse} CLICKHOUSE_GROUP=${CLICKHOUSE_GROUP:=${CLICKHOUSE_USER}} # Please note that we don't support paths with whitespaces. This is rather ignorant. @@ -12,6 +13,7 @@ CLICKHOUSE_BINDIR=${CLICKHOUSE_BINDIR:=/usr/bin} CLICKHOUSE_GENERIC_PROGRAM=${CLICKHOUSE_GENERIC_PROGRAM:=clickhouse} EXTRACT_FROM_CONFIG=${CLICKHOUSE_GENERIC_PROGRAM}-extract-from-config CLICKHOUSE_CONFIG=$CLICKHOUSE_CONFDIR/config.xml +CLICKHOUSE_PIDDIR=/var/run/$PROGRAM [ -f /usr/share/debconf/confmodule ] && . /usr/share/debconf/confmodule [ -f /etc/default/clickhouse ] && . /etc/default/clickhouse @@ -41,105 +43,5 @@ if [ "$1" = configure ] || [ -n "$not_deb_os" ]; then fi fi - # Make sure the administrative user exists - if ! getent passwd ${CLICKHOUSE_USER} > /dev/null; then - if [ -n "$not_deb_os" ]; then - useradd -r -s /bin/false --home-dir /nonexistent ${CLICKHOUSE_USER} > /dev/null - else - adduser --system --disabled-login --no-create-home --home /nonexistent \ - --shell /bin/false --group --gecos "ClickHouse server" ${CLICKHOUSE_USER} > /dev/null - fi - fi - - # if the user was created manually, make sure the group is there as well - if ! getent group ${CLICKHOUSE_GROUP} > /dev/null; then - groupadd -r ${CLICKHOUSE_GROUP} > /dev/null - fi - - # make sure user is in the correct group - if ! id -Gn ${CLICKHOUSE_USER} | grep -qw ${CLICKHOUSE_USER}; then - usermod -a -G ${CLICKHOUSE_GROUP} ${CLICKHOUSE_USER} > /dev/null - fi - - # check validity of user and group - if [ "$(id -u ${CLICKHOUSE_USER})" -eq 0 ]; then - echo "The ${CLICKHOUSE_USER} system user must not have uid 0 (root). -Please fix this and reinstall this package." >&2 - exit 1 - fi - - if [ "$(id -g ${CLICKHOUSE_GROUP})" -eq 0 ]; then - echo "The ${CLICKHOUSE_USER} system user must not have root as primary group. -Please fix this and reinstall this package." >&2 - exit 1 - fi - - if [ -x "$CLICKHOUSE_BINDIR/$EXTRACT_FROM_CONFIG" ] && [ -f "$CLICKHOUSE_CONFIG" ]; then - if [ -z "$SHELL" ]; then - SHELL="/bin/sh" - fi - CLICKHOUSE_DATADIR_FROM_CONFIG=$(su -s $SHELL ${CLICKHOUSE_USER} -c "$CLICKHOUSE_BINDIR/$EXTRACT_FROM_CONFIG --config-file=\"$CLICKHOUSE_CONFIG\" --key=path") ||: - echo "Path to data directory in ${CLICKHOUSE_CONFIG}: ${CLICKHOUSE_DATADIR_FROM_CONFIG}" - fi - CLICKHOUSE_DATADIR_FROM_CONFIG=${CLICKHOUSE_DATADIR_FROM_CONFIG:=$CLICKHOUSE_DATADIR} - - if [ ! -d ${CLICKHOUSE_DATADIR_FROM_CONFIG} ]; then - mkdir -p ${CLICKHOUSE_DATADIR_FROM_CONFIG} - chown ${CLICKHOUSE_USER}:${CLICKHOUSE_GROUP} ${CLICKHOUSE_DATADIR_FROM_CONFIG} - chmod 700 ${CLICKHOUSE_DATADIR_FROM_CONFIG} - fi - - if [ -d ${CLICKHOUSE_CONFDIR} ]; then - mkdir -p ${CLICKHOUSE_CONFDIR}/users.d - mkdir -p ${CLICKHOUSE_CONFDIR}/config.d - rm -fv ${CLICKHOUSE_CONFDIR}/*-preprocessed.xml ||: - fi - - [ -e ${CLICKHOUSE_CONFDIR}/preprocessed ] || ln -s ${CLICKHOUSE_DATADIR_FROM_CONFIG}/preprocessed_configs ${CLICKHOUSE_CONFDIR}/preprocessed ||: - - if [ ! -d ${CLICKHOUSE_LOGDIR} ]; then - mkdir -p ${CLICKHOUSE_LOGDIR} - chown root:${CLICKHOUSE_GROUP} ${CLICKHOUSE_LOGDIR} - # Allow everyone to read logs, root and clickhouse to read-write - chmod 775 ${CLICKHOUSE_LOGDIR} - fi - - # Set net_admin capabilities to support introspection of "taskstats" performance metrics from the kernel - # and ipc_lock capabilities to allow mlock of clickhouse binary. - - # 1. Check that "setcap" tool exists. - # 2. Check that an arbitrary program with installed capabilities can run. - # 3. Set the capabilities. - - # The second is important for Docker and systemd-nspawn. - # When the container has no capabilities, - # but the executable file inside the container has capabilities, - # then attempt to run this file will end up with a cryptic "Operation not permitted" message. - - TMPFILE=/tmp/test_setcap.sh - - command -v setcap >/dev/null \ - && echo > $TMPFILE && chmod a+x $TMPFILE && $TMPFILE && setcap "cap_net_admin,cap_ipc_lock,cap_sys_nice+ep" $TMPFILE && $TMPFILE && rm $TMPFILE \ - && setcap "cap_net_admin,cap_ipc_lock,cap_sys_nice+ep" "${CLICKHOUSE_BINDIR}/${CLICKHOUSE_GENERIC_PROGRAM}" \ - || echo "Cannot set 'net_admin' or 'ipc_lock' or 'sys_nice' capability for clickhouse binary. This is optional. Taskstats accounting will be disabled. To enable taskstats accounting you may add the required capability later manually." - - # Clean old dynamic compilation results - if [ -d "${CLICKHOUSE_DATADIR_FROM_CONFIG}/build" ]; then - rm -f ${CLICKHOUSE_DATADIR_FROM_CONFIG}/build/*.cpp ${CLICKHOUSE_DATADIR_FROM_CONFIG}/build/*.so ||: - fi - - if [ -f /usr/share/debconf/confmodule ]; then - db_get clickhouse-server/default-password - defaultpassword="$RET" - if [ -n "$defaultpassword" ]; then - echo "$defaultpassword" > ${CLICKHOUSE_CONFDIR}/users.d/default-password.xml - chown ${CLICKHOUSE_USER}:${CLICKHOUSE_GROUP} ${CLICKHOUSE_CONFDIR}/users.d/default-password.xml - chmod 600 ${CLICKHOUSE_CONFDIR}/users.d/default-password.xml - fi - - # everything went well, so now let's reset the password - db_set clickhouse-server/default-password "" - # ... done with debconf here - db_stop - fi + ${CLICKHOUSE_GENERIC_PROGRAM} install --user "${CLICKHOUSE_USER}" --group "${CLICKHOUSE_GROUP}" --pid-path "${CLICKHOUSE_PIDDIR}" --config-path "${CLICKHOUSE_CONFDIR}" --binary-path "${CLICKHOUSE_BINDIR}" --log-path "${CLICKHOUSE_LOGDIR}" --data-path "${CLICKHOUSE_DATADIR}" fi diff --git a/docker/packager/deb/Dockerfile b/docker/packager/deb/Dockerfile index 5bd8a74cb99..8fd89d60f85 100644 --- a/docker/packager/deb/Dockerfile +++ b/docker/packager/deb/Dockerfile @@ -31,14 +31,6 @@ RUN curl -O https://clickhouse-builds.s3.yandex.net/utils/1/dpkg-deb \ && chmod +x dpkg-deb \ && cp dpkg-deb /usr/bin -ENV APACHE_PUBKEY_HASH="bba6987b63c63f710fd4ed476121c588bc3812e99659d27a855f8c4d312783ee66ad6adfce238765691b04d62fa3688f" - -RUN export CODENAME="$(lsb_release --codename --short | tr 'A-Z' 'a-z')" \ - && wget -nv -O /tmp/arrow-keyring.deb "https://apache.bintray.com/arrow/ubuntu/apache-arrow-archive-keyring-latest-${CODENAME}.deb" \ - && echo "${APACHE_PUBKEY_HASH} /tmp/arrow-keyring.deb" | sha384sum -c \ - && dpkg -i /tmp/arrow-keyring.deb - - # Libraries from OS are only needed to test the "unbundled" build (this is not used in production). RUN apt-get update \ && apt-get install \ diff --git a/docker/packager/unbundled/Dockerfile b/docker/packager/unbundled/Dockerfile index 604f187cc58..50671011a23 100644 --- a/docker/packager/unbundled/Dockerfile +++ b/docker/packager/unbundled/Dockerfile @@ -1,6 +1,10 @@ # docker build -t yandex/clickhouse-unbundled-builder . FROM yandex/clickhouse-deb-builder +RUN export CODENAME="$(lsb_release --codename --short | tr 'A-Z' 'a-z')" \ + && wget -nv -O /tmp/arrow-keyring.deb "https://apache.bintray.com/arrow/ubuntu/apache-arrow-archive-keyring-latest-${CODENAME}.deb" \ + && dpkg -i /tmp/arrow-keyring.deb + # Libraries from OS are only needed to test the "unbundled" build (that is not used in production). RUN apt-get update \ && apt-get install \ diff --git a/docker/server/.dockerignore b/docker/server/.dockerignore new file mode 100644 index 00000000000..468a8cafb00 --- /dev/null +++ b/docker/server/.dockerignore @@ -0,0 +1,8 @@ +# post / preinstall scripts (not needed, we do it in Dockerfile) +alpine-root/install/* + +# docs (looks useless) +alpine-root/usr/share/doc/* + +# packages, etc. (used by prepare.sh) +alpine-root/tgz-packages/* \ No newline at end of file diff --git a/docker/server/.gitignore b/docker/server/.gitignore new file mode 100644 index 00000000000..4081b5f124c --- /dev/null +++ b/docker/server/.gitignore @@ -0,0 +1 @@ +alpine-root/* \ No newline at end of file diff --git a/docker/server/Dockerfile.alpine b/docker/server/Dockerfile.alpine new file mode 100644 index 00000000000..fc2756eac8c --- /dev/null +++ b/docker/server/Dockerfile.alpine @@ -0,0 +1,26 @@ +FROM alpine + +ENV LANG=en_US.UTF-8 \ + LANGUAGE=en_US:en \ + LC_ALL=en_US.UTF-8 \ + TZ=UTC \ + CLICKHOUSE_CONFIG=/etc/clickhouse-server/config.xml + +COPY alpine-root/ / + +# from https://github.com/ClickHouse/ClickHouse/blob/master/debian/clickhouse-server.postinst +RUN addgroup clickhouse \ + && adduser -S -H -h /nonexistent -s /bin/false -G clickhouse -g "ClickHouse server" clickhouse \ + && chown clickhouse:clickhouse /var/lib/clickhouse \ + && chmod 700 /var/lib/clickhouse \ + && chown root:clickhouse /var/log/clickhouse-server \ + && chmod 775 /var/log/clickhouse-server \ + && chmod +x /entrypoint.sh \ + && apk add --no-cache su-exec + +EXPOSE 9000 8123 9009 + +VOLUME /var/lib/clickhouse \ + /var/log/clickhouse-server + +ENTRYPOINT ["/entrypoint.sh"] diff --git a/docker/server/alpine-build.sh b/docker/server/alpine-build.sh new file mode 100755 index 00000000000..30101225b3e --- /dev/null +++ b/docker/server/alpine-build.sh @@ -0,0 +1,59 @@ +#!/bin/bash +set -x + +REPO_CHANNEL="${REPO_CHANNEL:-stable}" # lts / testing / prestable / etc +REPO_URL="${REPO_URL:-"https://repo.yandex.ru/clickhouse/tgz/${REPO_CHANNEL}"}" +VERSION="${VERSION:-20.9.3.45}" + +# where original files live +DOCKER_BUILD_FOLDER="${BASH_SOURCE%/*}" + +# we will create root for our image here +CONTAINER_ROOT_FOLDER="${DOCKER_BUILD_FOLDER}/alpine-root" + +# where to put downloaded tgz +TGZ_PACKAGES_FOLDER="${CONTAINER_ROOT_FOLDER}/tgz-packages" + +# clean up the root from old runs +rm -rf "$CONTAINER_ROOT_FOLDER" + +mkdir -p "$TGZ_PACKAGES_FOLDER" + +PACKAGES=( "clickhouse-client" "clickhouse-server" "clickhouse-common-static" ) + +# download tars from the repo +for package in "${PACKAGES[@]}" +do + wget -q --show-progress "${REPO_URL}/${package}-${VERSION}.tgz" -O "${TGZ_PACKAGES_FOLDER}/${package}-${VERSION}.tgz" +done + +# unpack tars +for package in "${PACKAGES[@]}" +do + tar xvzf "${TGZ_PACKAGES_FOLDER}/${package}-${VERSION}.tgz" --strip-components=2 -C "$CONTAINER_ROOT_FOLDER" +done + +# prepare few more folders +mkdir -p "${CONTAINER_ROOT_FOLDER}/etc/clickhouse-server/users.d" \ + "${CONTAINER_ROOT_FOLDER}/etc/clickhouse-server/config.d" \ + "${CONTAINER_ROOT_FOLDER}/var/log/clickhouse-server" \ + "${CONTAINER_ROOT_FOLDER}/var/lib/clickhouse" \ + "${CONTAINER_ROOT_FOLDER}/docker-entrypoint-initdb.d" \ + "${CONTAINER_ROOT_FOLDER}/lib64" + +cp "${DOCKER_BUILD_FOLDER}/docker_related_config.xml" "${CONTAINER_ROOT_FOLDER}/etc/clickhouse-server/config.d/" +cp "${DOCKER_BUILD_FOLDER}/entrypoint.alpine.sh" "${CONTAINER_ROOT_FOLDER}/entrypoint.sh" + +## get glibc components from ubuntu 20.04 and put them to expected place +docker pull ubuntu:20.04 +ubuntu20image=$(docker create --rm ubuntu:20.04) +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libc.so.6 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libdl.so.2 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libm.so.6 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libpthread.so.0 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/librt.so.1 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libnss_dns.so.2 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib/x86_64-linux-gnu/libresolv.so.2 "${CONTAINER_ROOT_FOLDER}/lib" +docker cp -L ${ubuntu20image}:/lib64/ld-linux-x86-64.so.2 "${CONTAINER_ROOT_FOLDER}/lib64" + +docker build "$DOCKER_BUILD_FOLDER" -f Dockerfile.alpine -t "yandex/clickhouse-server:${VERSION}-alpine" --pull \ No newline at end of file diff --git a/docker/server/entrypoint.alpine.sh b/docker/server/entrypoint.alpine.sh new file mode 100755 index 00000000000..e2edda9ca26 --- /dev/null +++ b/docker/server/entrypoint.alpine.sh @@ -0,0 +1,152 @@ +#!/bin/sh +#set -x + +DO_CHOWN=1 +if [ "$CLICKHOUSE_DO_NOT_CHOWN" = 1 ]; then + DO_CHOWN=0 +fi + +CLICKHOUSE_UID="${CLICKHOUSE_UID:-"$(id -u clickhouse)"}" +CLICKHOUSE_GID="${CLICKHOUSE_GID:-"$(id -g clickhouse)"}" + +# support --user +if [ "$(id -u)" = "0" ]; then + USER=$CLICKHOUSE_UID + GROUP=$CLICKHOUSE_GID + # busybox has setuidgid & chpst buildin + gosu="su-exec $USER:$GROUP" +else + USER="$(id -u)" + GROUP="$(id -g)" + gosu="" + DO_CHOWN=0 +fi + +# set some vars +CLICKHOUSE_CONFIG="${CLICKHOUSE_CONFIG:-/etc/clickhouse-server/config.xml}" + +# port is needed to check if clickhouse-server is ready for connections +HTTP_PORT="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=http_port)" + +# get CH directories locations +DATA_DIR="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=path || true)" +TMP_DIR="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=tmp_path || true)" +USER_PATH="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=user_files_path || true)" +LOG_PATH="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=logger.log || true)" +LOG_DIR="$(dirname $LOG_PATH || true)" +ERROR_LOG_PATH="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=logger.errorlog || true)" +ERROR_LOG_DIR="$(dirname $ERROR_LOG_PATH || true)" +FORMAT_SCHEMA_PATH="$(clickhouse extract-from-config --config-file $CLICKHOUSE_CONFIG --key=format_schema_path || true)" + +CLICKHOUSE_USER="${CLICKHOUSE_USER:-default}" +CLICKHOUSE_PASSWORD="${CLICKHOUSE_PASSWORD:-}" +CLICKHOUSE_DB="${CLICKHOUSE_DB:-}" + +for dir in "$DATA_DIR" \ + "$ERROR_LOG_DIR" \ + "$LOG_DIR" \ + "$TMP_DIR" \ + "$USER_PATH" \ + "$FORMAT_SCHEMA_PATH" +do + # check if variable not empty + [ -z "$dir" ] && continue + # ensure directories exist + if ! mkdir -p "$dir"; then + echo "Couldn't create necessary directory: $dir" + exit 1 + fi + + if [ "$DO_CHOWN" = "1" ]; then + # ensure proper directories permissions + chown -R "$USER:$GROUP" "$dir" + elif [ "$(stat -c %u "$dir")" != "$USER" ]; then + echo "Necessary directory '$dir' isn't owned by user with id '$USER'" + exit 1 + fi +done + +# if clickhouse user is defined - create it (user "default" already exists out of box) +if [ -n "$CLICKHOUSE_USER" ] && [ "$CLICKHOUSE_USER" != "default" ] || [ -n "$CLICKHOUSE_PASSWORD" ]; then + echo "$0: create new user '$CLICKHOUSE_USER' instead 'default'" + cat < /etc/clickhouse-server/users.d/default-user.xml + + + + + + + + <${CLICKHOUSE_USER}> + default + + ::/0 + + ${CLICKHOUSE_PASSWORD} + default + + + +EOT +fi + +if [ -n "$(ls /docker-entrypoint-initdb.d/)" ] || [ -n "$CLICKHOUSE_DB" ]; then + # Listen only on localhost until the initialization is done + $gosu /usr/bin/clickhouse-server --config-file=$CLICKHOUSE_CONFIG -- --listen_host=127.0.0.1 & + pid="$!" + + # check if clickhouse is ready to accept connections + # will try to send ping clickhouse via http_port (max 6 retries, with 1 sec timeout and 1 sec delay between retries) + tries=6 + while ! wget --spider -T 1 -q "http://localhost:$HTTP_PORT/ping" 2>/dev/null; do + if [ "$tries" -le "0" ]; then + echo >&2 'ClickHouse init process failed.' + exit 1 + fi + tries=$(( tries-1 )) + sleep 1 + done + + if [ ! -z "$CLICKHOUSE_PASSWORD" ]; then + printf -v WITH_PASSWORD '%s %q' "--password" "$CLICKHOUSE_PASSWORD" + fi + + clickhouseclient="clickhouse-client --multiquery -u $CLICKHOUSE_USER $WITH_PASSWORD " + + # create default database, if defined + if [ -n "$CLICKHOUSE_DB" ]; then + echo "$0: create database '$CLICKHOUSE_DB'" + "$clickhouseclient" -q "CREATE DATABASE IF NOT EXISTS $CLICKHOUSE_DB"; + fi + + for f in /docker-entrypoint-initdb.d/*; do + case "$f" in + *.sh) + if [ -x "$f" ]; then + echo "$0: running $f" + "$f" + else + echo "$0: sourcing $f" + . "$f" + fi + ;; + *.sql) echo "$0: running $f"; cat "$f" | "$clickhouseclient" ; echo ;; + *.sql.gz) echo "$0: running $f"; gunzip -c "$f" | "$clickhouseclient"; echo ;; + *) echo "$0: ignoring $f" ;; + esac + echo + done + + if ! kill -s TERM "$pid" || ! wait "$pid"; then + echo >&2 'Finishing of ClickHouse init process failed.' + exit 1 + fi +fi + +# if no args passed to `docker run` or first argument start with `--`, then the user is passing clickhouse-server arguments +if [[ $# -lt 1 ]] || [[ "$1" == "--"* ]]; then + exec $gosu /usr/bin/clickhouse-server --config-file=$CLICKHOUSE_CONFIG "$@" +fi + +# Otherwise, we assume the user want to run his own process, for example a `bash` shell to explore this image +exec "$@" diff --git a/docker/test/fasttest/Dockerfile b/docker/test/fasttest/Dockerfile index bed438a6579..f6c665ff3fd 100644 --- a/docker/test/fasttest/Dockerfile +++ b/docker/test/fasttest/Dockerfile @@ -82,6 +82,7 @@ RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone ENV COMMIT_SHA='' ENV PULL_REQUEST_NUMBER='' +ENV COPY_CLICKHOUSE_BINARY_TO_OUTPUT=0 COPY run.sh / CMD ["/bin/bash", "/run.sh"] diff --git a/docker/test/fasttest/run.sh b/docker/test/fasttest/run.sh index f12ecbb2c9c..bcba50e9d81 100755 --- a/docker/test/fasttest/run.sh +++ b/docker/test/fasttest/run.sh @@ -20,6 +20,7 @@ FASTTEST_SOURCE=$(readlink -f "${FASTTEST_SOURCE:-$FASTTEST_WORKSPACE/ch}") FASTTEST_BUILD=$(readlink -f "${FASTTEST_BUILD:-${BUILD:-$FASTTEST_WORKSPACE/build}}") FASTTEST_DATA=$(readlink -f "${FASTTEST_DATA:-$FASTTEST_WORKSPACE/db-fasttest}") FASTTEST_OUTPUT=$(readlink -f "${FASTTEST_OUTPUT:-$FASTTEST_WORKSPACE}") +PATH="$FASTTEST_BUILD/programs:$FASTTEST_SOURCE/tests:$PATH" # Export these variables, so that all subsequent invocations of the script # use them, and not try to guess them anew, which leads to weird effects. @@ -28,6 +29,7 @@ export FASTTEST_SOURCE export FASTTEST_BUILD export FASTTEST_DATA export FASTTEST_OUT +export PATH server_pid=none @@ -125,7 +127,7 @@ function clone_submodules ( cd "$FASTTEST_SOURCE" -SUBMODULES_TO_UPDATE=(contrib/boost contrib/zlib-ng contrib/libxml2 contrib/poco contrib/libunwind contrib/ryu contrib/fmtlib contrib/base64 contrib/cctz contrib/libcpuid contrib/double-conversion contrib/libcxx contrib/libcxxabi contrib/libc-headers contrib/lz4 contrib/zstd contrib/fastops contrib/rapidjson contrib/re2 contrib/sparsehash-c11) +SUBMODULES_TO_UPDATE=(contrib/boost contrib/zlib-ng contrib/libxml2 contrib/poco contrib/libunwind contrib/ryu contrib/fmtlib contrib/base64 contrib/cctz contrib/libcpuid contrib/double-conversion contrib/libcxx contrib/libcxxabi contrib/libc-headers contrib/lz4 contrib/zstd contrib/fastops contrib/rapidjson contrib/re2 contrib/sparsehash-c11 contrib/croaring) git submodule sync git submodule update --init --recursive "${SUBMODULES_TO_UPDATE[@]}" @@ -137,7 +139,14 @@ git submodule foreach git clean -xfd function run_cmake { -CMAKE_LIBS_CONFIG=("-DENABLE_LIBRARIES=0" "-DENABLE_TESTS=0" "-DENABLE_UTILS=0" "-DENABLE_EMBEDDED_COMPILER=0" "-DENABLE_THINLTO=0" "-DUSE_UNWIND=1") +CMAKE_LIBS_CONFIG=( + "-DENABLE_LIBRARIES=0" + "-DENABLE_TESTS=0" + "-DENABLE_UTILS=0" + "-DENABLE_EMBEDDED_COMPILER=0" + "-DENABLE_THINLTO=0" + "-DUSE_UNWIND=1" +) # TODO remove this? we don't use ccache anyway. An option would be to download it # from S3 simultaneously with cloning. @@ -163,6 +172,9 @@ function build ( cd "$FASTTEST_BUILD" time ninja clickhouse-bundle | ts '%Y-%m-%d %H:%M:%S' | tee "$FASTTEST_OUTPUT/build_log.txt" +if [ "$COPY_CLICKHOUSE_BINARY_TO_OUTPUT" -eq "1" ]; then + cp programs/clickhouse "$FASTTEST_OUTPUT/clickhouse" +fi ccache --show-stats ||: ) } @@ -219,6 +231,8 @@ TESTS_TO_SKIP=( 01268_dictionary_direct_layout 01280_ssd_complex_key_dictionary 01281_group_by_limit_memory_tracking # max_memory_usage_for_user can interfere another queries running concurrently + 01318_encrypt # Depends on OpenSSL + 01318_decrypt # Depends on OpenSSL 01281_unsucceeded_insert_select_queries_counter 01292_create_user 01294_lazy_database_concurrent @@ -258,6 +272,13 @@ TESTS_TO_SKIP=( # Look at DistributedFilesToInsert, so cannot run in parallel. 01460_DistributedFilesToInsert + + 01541_max_memory_usage_for_user + + # Require python libraries like scipy, pandas and numpy + 01322_ttest_scipy + + 01545_system_errors ) time clickhouse-test -j 8 --order=random --no-long --testname --shard --zookeeper --skip "${TESTS_TO_SKIP[@]}" 2>&1 | ts '%Y-%m-%d %H:%M:%S' | tee "$FASTTEST_OUTPUT/test_log.txt" @@ -327,8 +348,6 @@ case "$stage" in ;& "build") build - PATH="$FASTTEST_BUILD/programs:$FASTTEST_SOURCE/tests:$PATH" - export PATH ;& "configure") # The `install_log.txt` is also needed for compatibility with old CI task -- diff --git a/docker/test/fuzzer/run-fuzzer.sh b/docker/test/fuzzer/run-fuzzer.sh index 6429c1bc19c..41559a9b1f5 100755 --- a/docker/test/fuzzer/run-fuzzer.sh +++ b/docker/test/fuzzer/run-fuzzer.sh @@ -164,7 +164,7 @@ case "$stage" in # Lost connection to the server. This probably means that the server died # with abort. echo "failure" > status.txt - if ! grep -ao "Received signal.*\|Logical error.*\|Assertion.*failed" server.log > description.txt + if ! grep -ao "Received signal.*\|Logical error.*\|Assertion.*failed\|Failed assertion.*" server.log > description.txt then echo "Lost connection to server. See the logs" > description.txt fi diff --git a/docker/test/integration/base/Dockerfile b/docker/test/integration/base/Dockerfile index b6a46f6d934..99095de60fb 100644 --- a/docker/test/integration/base/Dockerfile +++ b/docker/test/integration/base/Dockerfile @@ -17,7 +17,8 @@ RUN apt-get update \ sqlite3 \ curl \ tar \ - krb5-user + krb5-user \ + iproute2 RUN rm -rf \ /var/lib/apt/lists/* \ /var/cache/debconf \ diff --git a/docker/test/integration/runner/Dockerfile b/docker/test/integration/runner/Dockerfile index 795d0d371f6..70b57b245d3 100644 --- a/docker/test/integration/runner/Dockerfile +++ b/docker/test/integration/runner/Dockerfile @@ -37,7 +37,28 @@ RUN apt-get update \ ENV TZ=Europe/Moscow RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone -RUN python3 -m pip install urllib3==1.23 pytest docker-compose==1.22.0 docker dicttoxml kazoo PyMySQL psycopg2==2.7.5 pymongo tzlocal kafka-python protobuf redis aerospike pytest-timeout minio grpcio grpcio-tools cassandra-driver confluent-kafka avro +RUN python3 -m pip install \ + PyMySQL \ + aerospike \ + avro \ + cassandra-driver \ + confluent-kafka \ + dicttoxml \ + docker \ + docker-compose==1.22.0 \ + grpcio \ + grpcio-tools \ + kafka-python \ + kazoo \ + minio \ + protobuf \ + psycopg2-binary==2.7.5 \ + pymongo \ + pytest \ + pytest-timeout \ + redis \ + tzlocal \ + urllib3 ENV DOCKER_CHANNEL stable ENV DOCKER_VERSION 17.09.1-ce diff --git a/docker/test/performance-comparison/Dockerfile b/docker/test/performance-comparison/Dockerfile index a4f8af2f388..76cadc3ce11 100644 --- a/docker/test/performance-comparison/Dockerfile +++ b/docker/test/performance-comparison/Dockerfile @@ -9,6 +9,7 @@ RUN apt-get update \ && DEBIAN_FRONTEND=noninteractive apt-get install --yes --no-install-recommends \ bash \ curl \ + dmidecode \ g++ \ gdb \ git \ @@ -37,7 +38,18 @@ RUN apt-get update \ COPY * / -CMD /entrypoint.sh +# Bind everything to one NUMA node, if there's more than one. Theoretically the +# node #0 should be less stable because of system interruptions. We bind +# randomly to node 1 or 0 to gather some statistics on that. We have to bind +# both servers and the tmpfs on which the database is stored. How to do it +# through Yandex Sandbox API is unclear, but by default tmpfs uses +# 'process allocation policy', not sure which process but hopefully the one that +# writes to it, so just bind the downloader script as well. We could also try to +# remount it with proper options in Sandbox task. +# https://www.kernel.org/doc/Documentation/filesystems/tmpfs.txt +# Double-escaped backslashes are a tribute to the engineering wonder of docker -- +# it gives '/bin/sh: 1: [bash,: not found' otherwise. +CMD ["bash", "-c", "node=$((RANDOM % $(numactl --hardware | sed -n 's/^.*available:\\(.*\\)nodes.*$/\\1/p'))); echo Will bind to NUMA node $node; numactl --cpunodebind=$node --membind=$node /entrypoint.sh"] # docker run --network=host --volume :/workspace --volume=:/output -e PR_TO_TEST=<> -e SHA_TO_TEST=<> yandex/clickhouse-performance-comparison diff --git a/docker/test/performance-comparison/README.md b/docker/test/performance-comparison/README.md index 3953e99fc0f..782644a81dd 100644 --- a/docker/test/performance-comparison/README.md +++ b/docker/test/performance-comparison/README.md @@ -48,12 +48,13 @@ This table shows queries that take significantly longer to process on the client #### Unexpected Query Duration Action required for every item -- these are errors that must be fixed. -Queries that have "short" duration (on the order of 0.1 s) can't be reliably tested in a normal way, where we perform a small (about ten) measurements for each server, because the signal-to-noise ratio is much smaller. There is a special mode for such queries that instead runs them for a fixed amount of time, normally with much higher number of measurements (up to thousands). This mode must be explicitly enabled by the test author to avoid accidental errors. It must be used only for queries that are meant to complete "immediately", such as `select count(*)`. If your query is not supposed to be "immediate", try to make it run longer, by e.g. processing more data. +A query is supposed to run longer than 0.1 second. If your query runs faster, increase the amount of processed data to bring the run time above this threshold. You can use a bigger table (e.g. `hits_100m` instead of `hits_10m`), increase a `LIMIT`, make a query single-threaded, and so on. Queries that are too fast suffer from poor stability and precision. -This table shows queries for which the "short" marking is not consistent with the actual query run time -- i.e., a query runs for a long time but is marked as short, or it runs very fast but is not marked as short. +Sometimes you want to test a query that is supposed to complete "instantaneously", i.e. in sublinear time. This might be `count(*)`, or parsing a complicated tuple. It might not be practical or even possible to increase the run time of such queries by adding more data. For such queries there is a specal comparison mode which runs them for a fixed amount of time, instead of a fixed number of iterations like we do normally. This mode is inferior to the normal mode, because the influence of noise and overhead is higher, which leads to less precise and stable results. -If your query is really supposed to complete "immediately" and can't be made to run longer, you have to mark it as "short". To do so, write `...` in the test file. The value of "short" attribute is evaluated as a python expression, and substitutions are performed, so you can write something like `select count(*) from table where {column1} > {column2}`, to mark only a particular combination of variables as short. +If it is impossible to increase the run time of a query and it is supposed to complete "immediately", you have to explicitly mark this in the test. To do so, add a `short` attribute to the query tag in the test file: `...`. The value of the `short` attribute is evaluated as a python expression, and substitutions are performed, so you can write something like `select count(*) from table where {column1} > {column2}`, to mark only a particular combination of variables as short. +This table shows queries for which the `short` marking is not consistent with the actual query run time -- i.e., a query runs for a normal time but is marked as `short`, or it runs faster than normal but is not marked as `short`. #### Partial Queries Action required for the cells marked in red. diff --git a/docker/test/performance-comparison/compare.sh b/docker/test/performance-comparison/compare.sh index 74f1e81e8f3..d0cf12baa9e 100755 --- a/docker/test/performance-comparison/compare.sh +++ b/docker/test/performance-comparison/compare.sh @@ -63,7 +63,7 @@ function configure # Make copies of the original db for both servers. Use hardlinks instead # of copying to save space. Before that, remove preprocessed configs and # system tables, because sharing them between servers with hardlinks may - # lead to weird effects. + # lead to weird effects. rm -r left/db ||: rm -r right/db ||: rm -r db0/preprocessed_configs ||: @@ -77,20 +77,30 @@ function restart while killall clickhouse-server; do echo . ; sleep 1 ; done echo all killed + # Change the jemalloc settings here. + # https://github.com/jemalloc/jemalloc/wiki/Getting-Started + export MALLOC_CONF="confirm_conf:true" + set -m # Spawn servers in their own process groups - left/clickhouse-server --config-file=left/config/config.xml -- --path left/db --user_files_path left/db/user_files &>> left-server-log.log & + left/clickhouse-server --config-file=left/config/config.xml \ + -- --path left/db --user_files_path left/db/user_files \ + &>> left-server-log.log & left_pid=$! kill -0 $left_pid disown $left_pid - right/clickhouse-server --config-file=right/config/config.xml -- --path right/db --user_files_path right/db/user_files &>> right-server-log.log & + right/clickhouse-server --config-file=right/config/config.xml \ + -- --path right/db --user_files_path right/db/user_files \ + &>> right-server-log.log & right_pid=$! kill -0 $right_pid disown $right_pid set +m + unset MALLOC_CONF + wait_for_server 9001 $left_pid echo left ok @@ -198,7 +208,7 @@ function run_tests echo test "$test_name" # Don't profile if we're past the time limit. - # Use awk because bash doesn't support floating point arithmetics. + # Use awk because bash doesn't support floating point arithmetic. profile_seconds=$(awk "BEGIN { print ($profile_seconds_left > 0 ? 10 : 0) }") TIMEFORMAT=$(printf "$test_name\t%%3R\t%%3U\t%%3S\n") @@ -449,7 +459,12 @@ wait unset IFS ) -parallel --joblog analyze/parallel-log.txt --null < analyze/commands.txt 2>> analyze/errors.log +# The comparison script might be bound to one NUMA node for better test +# stability, and the calculation runs out of memory because of this. Use +# all nodes. +numactl --show +numactl --cpunodebind=all --membind=all numactl --show +numactl --cpunodebind=all --membind=all parallel --joblog analyze/parallel-log.txt --null < analyze/commands.txt 2>> analyze/errors.log clickhouse-local --query " -- Join the metric names back to the metric statistics we've calculated, and make @@ -526,10 +541,10 @@ create table queries engine File(TSVWithNamesAndTypes, 'report/queries.tsv') as select abs(diff) > report_threshold and abs(diff) > stat_threshold as changed_fail, abs(diff) > report_threshold - 0.05 and abs(diff) > stat_threshold as changed_show, - + not changed_fail and stat_threshold > report_threshold + 0.10 as unstable_fail, not changed_show and stat_threshold > report_threshold - 0.05 as unstable_show, - + left, right, diff, stat_threshold, if(report_threshold > 0, report_threshold, 0.10) as report_threshold, query_metric_stats.test test, query_metric_stats.query_index query_index, @@ -752,7 +767,7 @@ create table all_tests_report engine File(TSV, 'report/all-queries.tsv') as -- The threshold for 2) is significantly larger than the threshold for 1), to -- avoid jitter. create view shortness - as select + as select (test, query_index) in (select * from file('analyze/marked-short-queries.tsv', TSV, 'test text, query_index int')) @@ -1070,8 +1085,10 @@ case "$stage" in time configure ;& "restart") + numactl --show ||: numactl --hardware ||: lscpu ||: + dmidecode -t 4 ||: time restart ;& "run_tests") diff --git a/docker/test/performance-comparison/config/users.d/perf-comparison-tweaks-users.xml b/docker/test/performance-comparison/config/users.d/perf-comparison-tweaks-users.xml index cee7dc3ff16..ce7a6ae094a 100644 --- a/docker/test/performance-comparison/config/users.d/perf-comparison-tweaks-users.xml +++ b/docker/test/performance-comparison/config/users.d/perf-comparison-tweaks-users.xml @@ -14,6 +14,9 @@ we might also add time check to perf.py script. --> 300 + + + 20 diff --git a/docker/test/performance-comparison/report.py b/docker/test/performance-comparison/report.py index 1f55300661b..9d3ccabb788 100755 --- a/docker/test/performance-comparison/report.py +++ b/docker/test/performance-comparison/report.py @@ -468,14 +468,14 @@ if args.report == 'main': return columns = [ - 'Test', #0 - 'Wall clock time, s', #1 - 'Total client time, s', #2 - 'Total queries', #3 - 'Longest query
(sum for all runs), s', #4 - 'Avg wall clock time
(sum for all runs), s', #5 - 'Shortest query
(sum for all runs), s', #6 - '', # Runs #7 + 'Test', #0 + 'Wall clock time, entire test, s', #1 + 'Total client time for measured query runs, s', #2 + 'Queries', #3 + 'Longest query, total for measured runs, s', #4 + 'Wall clock time per query, s', #5 + 'Shortest query, total for measured runs, s', #6 + '', # Runs #7 ] attrs = ['' for c in columns] attrs[7] = None diff --git a/docker/test/stateless/Dockerfile b/docker/test/stateless/Dockerfile index 33eb1c29103..47c45e57508 100644 --- a/docker/test/stateless/Dockerfile +++ b/docker/test/stateless/Dockerfile @@ -16,6 +16,7 @@ RUN apt-get update -y \ python3-lxml \ python3-requests \ python3-termcolor \ + python3-pip \ qemu-user-static \ sudo \ telnet \ @@ -23,6 +24,8 @@ RUN apt-get update -y \ unixodbc \ wget +RUN pip3 install numpy scipy pandas + RUN mkdir -p /tmp/clickhouse-odbc-tmp \ && wget -nv -O - ${odbc_driver_url} | tar --strip-components=1 -xz -C /tmp/clickhouse-odbc-tmp \ && cp /tmp/clickhouse-odbc-tmp/lib64/*.so /usr/local/lib/ \ @@ -33,5 +36,8 @@ RUN mkdir -p /tmp/clickhouse-odbc-tmp \ ENV TZ=Europe/Moscow RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone +ENV NUM_TRIES=1 +ENV MAX_RUN_TIME=0 + COPY run.sh / CMD ["/bin/bash", "/run.sh"] diff --git a/docker/test/stateless/run.sh b/docker/test/stateless/run.sh index 3d5a2dffe32..8793f3d22b4 100755 --- a/docker/test/stateless/run.sh +++ b/docker/test/stateless/run.sh @@ -1,6 +1,7 @@ #!/bin/bash -set -e -x +# fail on errors, verbose and export all env variables +set -e -x -a dpkg -i package_folder/clickhouse-common-static_*.deb dpkg -i package_folder/clickhouse-common-static-dbg_*.deb @@ -17,8 +18,26 @@ if grep -q -- "--use-skip-list" /usr/bin/clickhouse-test; then SKIP_LIST_OPT="--use-skip-list" fi -# We can have several additional options so we path them as array because it's -# more idiologically correct. -read -ra ADDITIONAL_OPTIONS <<< "${ADDITIONAL_OPTIONS:-}" +function run_tests() +{ + # We can have several additional options so we path them as array because it's + # more idiologically correct. + read -ra ADDITIONAL_OPTIONS <<< "${ADDITIONAL_OPTIONS:-}" -clickhouse-test --testname --shard --zookeeper --hung-check --print-time "$SKIP_LIST_OPT" "${ADDITIONAL_OPTIONS[@]}" "$SKIP_TESTS_OPTION" 2>&1 | ts '%Y-%m-%d %H:%M:%S' | tee test_output/test_result.txt + # Skip these tests, because they fail when we rerun them multiple times + if [ "$NUM_TRIES" -gt "1" ]; then + ADDITIONAL_OPTIONS+=('--skip') + ADDITIONAL_OPTIONS+=('00000_no_tests_to_skip') + fi + + for i in $(seq 1 $NUM_TRIES); do + clickhouse-test --testname --shard --zookeeper --hung-check --print-time "$SKIP_LIST_OPT" "${ADDITIONAL_OPTIONS[@]}" 2>&1 | ts '%Y-%m-%d %H:%M:%S' | tee -a test_output/test_result.txt + if [ ${PIPESTATUS[0]} -ne "0" ]; then + break; + fi + done +} + +export -f run_tests + +timeout $MAX_RUN_TIME bash -c run_tests ||: diff --git a/docker/test/stateless_unbundled/Dockerfile b/docker/test/stateless_unbundled/Dockerfile index f2fd28e4078..345ba905412 100644 --- a/docker/test/stateless_unbundled/Dockerfile +++ b/docker/test/stateless_unbundled/Dockerfile @@ -58,6 +58,7 @@ RUN apt-get --allow-unauthenticated update -y \ python3-lxml \ python3-requests \ python3-termcolor \ + python3-pip \ qemu-user-static \ sudo \ telnet \ @@ -68,6 +69,8 @@ RUN apt-get --allow-unauthenticated update -y \ wget \ zlib1g-dev +RUN pip3 install numpy scipy pandas + RUN mkdir -p /tmp/clickhouse-odbc-tmp \ && wget -nv -O - ${odbc_driver_url} | tar --strip-components=1 -xz -C /tmp/clickhouse-odbc-tmp \ && cp /tmp/clickhouse-odbc-tmp/lib64/*.so /usr/local/lib/ \ diff --git a/docker/test/stress/run.sh b/docker/test/stress/run.sh index ba529c401ea..9da2f3d3ada 100755 --- a/docker/test/stress/run.sh +++ b/docker/test/stress/run.sh @@ -45,7 +45,7 @@ function start() # for clickhouse-server (via service) echo "ASAN_OPTIONS='malloc_context_size=10 verbosity=1 allocator_release_to_os_interval_ms=10000'" >> /etc/environment # for clickhouse-client -export ASAN_OPTIONS='malloc_context_size=10 verbosity=1 allocator_release_to_os_interval_ms=10000' +export ASAN_OPTIONS='malloc_context_size=10 allocator_release_to_os_interval_ms=10000' start diff --git a/docker/test/stress/stress b/docker/test/stress/stress index a81391d56a7..f29ebc55141 100755 --- a/docker/test/stress/stress +++ b/docker/test/stress/stress @@ -28,8 +28,18 @@ def get_options(i): options = "" if 0 < i: options += " --order=random" + if i % 2 == 1: options += " --db-engine=Ordinary" + + # If database name is not specified, new database is created for each functional test. + # Run some threads with one database for all tests. + if i % 3 == 1: + options += " --database=test_{}".format(i) + + if i == 13: + options += " --client-option='memory_tracker_fault_probability=0.00001'" + return options diff --git a/docker/test/testflows/runner/Dockerfile b/docker/test/testflows/runner/Dockerfile index ed49743319c..06241d6d497 100644 --- a/docker/test/testflows/runner/Dockerfile +++ b/docker/test/testflows/runner/Dockerfile @@ -35,7 +35,7 @@ RUN apt-get update \ ENV TZ=Europe/Moscow RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone -RUN pip3 install urllib3 testflows==1.6.48 docker-compose docker dicttoxml kazoo tzlocal +RUN pip3 install urllib3 testflows==1.6.59 docker-compose docker dicttoxml kazoo tzlocal ENV DOCKER_CHANNEL stable ENV DOCKER_VERSION 17.09.1-ce @@ -72,5 +72,5 @@ RUN set -x \ VOLUME /var/lib/docker EXPOSE 2375 ENTRYPOINT ["dockerd-entrypoint.sh"] -CMD ["sh", "-c", "python3 regression.py --no-color --local --clickhouse-binary-path ${CLICKHOUSE_TESTS_SERVER_BIN_PATH} --log test.log ${TESTFLOWS_OPTS}; cat test.log | tfs report results --format json > results.json"] +CMD ["sh", "-c", "python3 regression.py --no-color -o classic --local --clickhouse-binary-path ${CLICKHOUSE_TESTS_SERVER_BIN_PATH} --log test.log ${TESTFLOWS_OPTS}; cat test.log | tfs report results --format json > results.json"] diff --git a/docs/README.md b/docs/README.md index c7fa0221726..8b3066501bf 100644 --- a/docs/README.md +++ b/docs/README.md @@ -195,7 +195,7 @@ Templates: - [Function](_description_templates/template-function.md) - [Setting](_description_templates/template-setting.md) -- [Table engine](_description_templates/template-table-engine.md) +- [Database or Table engine](_description_templates/template-engine.md) - [System table](_description_templates/template-system-table.md) diff --git a/docs/_description_templates/template-table-engine.md b/docs/_description_templates/template-engine.md similarity index 59% rename from docs/_description_templates/template-table-engine.md rename to docs/_description_templates/template-engine.md index c1bfcb3ec86..35181881134 100644 --- a/docs/_description_templates/template-table-engine.md +++ b/docs/_description_templates/template-engine.md @@ -1,8 +1,14 @@ # EngineName {#enginename} -- What the engine does. +- What the Database/Table engine does. - Relations with other engines if they exist. +## Creating a Database {#creating-a-database} +``` sql + CREATE DATABASE ... +``` +or + ## Creating a Table {#creating-a-table} ``` sql CREATE TABLE ... @@ -10,12 +16,19 @@ **Engine Parameters** -**Query Clauses** +**Query Clauses** (for Table engines only) -## Virtual columns {#virtual-columns} +## Virtual columns {#virtual-columns} (for Table engines only) List and virtual columns with description, if they exist. +## Data Types Support {#data_types-support} (for Database engines only) + +| EngineName | ClickHouse | +|-----------------------|------------------------------------| +| NativeDataTypeName | [ClickHouseDataTypeName](link#) | + + ## Specifics and recommendations {#specifics-and-recommendations} Algorithms diff --git a/docs/en/commercial/cloud.md b/docs/en/commercial/cloud.md index 07d843f724a..06216517db8 100644 --- a/docs/en/commercial/cloud.md +++ b/docs/en/commercial/cloud.md @@ -18,4 +18,14 @@ toc_title: Cloud - Encryption and isolation - Automated maintenance +## Altinity.Cloud {#altinity.cloud} + +[Altinity.Cloud](https://altinity.com/cloud-database/) is a fully managed ClickHouse-as-a-Service for the Amazon public cloud. +- Fast deployment of ClickHouse clusters on Amazon resources +- Easy scale-out/scale-in as well as vertical scaling of nodes +- Isolated per-tenant VPCs with public endpoint or VPC peering +- Configurable storage types and volume configurations +- Cross-AZ scaling for performance and high availability +- Built-in monitoring and SQL query editor + {## [Original article](https://clickhouse.tech/docs/en/commercial/cloud/) ##} diff --git a/docs/en/development/architecture.md b/docs/en/development/architecture.md index bea75140606..19caa5241b0 100644 --- a/docs/en/development/architecture.md +++ b/docs/en/development/architecture.md @@ -189,7 +189,7 @@ Replication is implemented in the `ReplicatedMergeTree` storage engine. The path Replication uses an asynchronous multi-master scheme. You can insert data into any replica that has a session with `ZooKeeper`, and data is replicated to all other replicas asynchronously. Because ClickHouse doesn’t support UPDATEs, replication is conflict-free. As there is no quorum acknowledgment of inserts, just-inserted data might be lost if one node fails. -Metadata for replication is stored in ZooKeeper. There is a replication log that lists what actions to do. Actions are: get part; merge parts; drop a partition, and so on. Each replica copies the replication log to its queue and then executes the actions from the queue. For example, on insertion, the “get the part” action is created in the log, and every replica downloads that part. Merges are coordinated between replicas to get byte-identical results. All parts are merged in the same way on all replicas. It is achieved by electing one replica as the leader, and that replica initiates merges and writes “merge parts” actions to the log. +Metadata for replication is stored in ZooKeeper. There is a replication log that lists what actions to do. Actions are: get part; merge parts; drop a partition, and so on. Each replica copies the replication log to its queue and then executes the actions from the queue. For example, on insertion, the “get the part” action is created in the log, and every replica downloads that part. Merges are coordinated between replicas to get byte-identical results. All parts are merged in the same way on all replicas. One of the leaders initiates a new merge first and writes “merge parts” actions to the log. Multiple replicas (or all) can be leaders at the same time. A replica can be prevented from becoming a leader using the `merge_tree` setting `replicated_can_become_leader`. The leaders are responsible for scheduling background merges. Replication is physical: only compressed parts are transferred between nodes, not queries. Merges are processed on each replica independently in most cases to lower the network costs by avoiding network amplification. Large merged parts are sent over the network only in cases of significant replication lag. diff --git a/docs/en/development/tests.md b/docs/en/development/tests.md index 8c06c868e66..e6141f7ab02 100644 --- a/docs/en/development/tests.md +++ b/docs/en/development/tests.md @@ -47,6 +47,8 @@ select x; -- { serverError 49 } ``` This test ensures that the server returns an error with code 49 about unknown column `x`. If there is no error, or the error is different, the test will fail. If you want to ensure that an error occurs on the client side, use `clientError` annotation instead. +Do not check for a particular wording of error message, it may change in the future, and the test will needlessly break. Check only the error code. If the existing error code is not precise enough for your needs, consider adding a new one. + ### Testing a Distributed Query If you want to use distributed queries in functional tests, you can leverage `remote` table function with `127.0.0.{1..2}` addresses for the server to query itself; or you can use predefined test clusters in server configuration file like `test_shard_localhost`. Remember to add the words `shard` or `distributed` to the test name, so that it is ran in CI in correct configurations, where the server is configured to support distributed queries. diff --git a/docs/en/engines/table-engines/integrations/rabbitmq.md b/docs/en/engines/table-engines/integrations/rabbitmq.md index dd14ee3b4b1..b0901ee6f6e 100644 --- a/docs/en/engines/table-engines/integrations/rabbitmq.md +++ b/docs/en/engines/table-engines/integrations/rabbitmq.md @@ -51,7 +51,7 @@ Optional parameters: - `rabbitmq_row_delimiter` – Delimiter character, which ends the message. - `rabbitmq_schema` – Parameter that must be used if the format requires a schema definition. For example, [Cap’n Proto](https://capnproto.org/) requires the path to the schema file and the name of the root `schema.capnp:Message` object. - `rabbitmq_num_consumers` – The number of consumers per table. Default: `1`. Specify more consumers if the throughput of one consumer is insufficient. -- `rabbitmq_num_queues` – The number of queues per consumer. Default: `1`. Specify more queues if the capacity of one queue per consumer is insufficient. +- `rabbitmq_num_queues` – Total number of queues. Default: `1`. Increasing this number can significantly improve performance. - `rabbitmq_queue_base` - Specify a hint for queue names. Use cases of this setting are described below. - `rabbitmq_deadletter_exchange` - Specify name for a [dead letter exchange](https://www.rabbitmq.com/dlx.html). You can create another table with this exchange name and collect messages in cases when they are republished to dead letter exchange. By default dead letter exchange is not specified. - `rabbitmq_persistent` - If set to 1 (true), in insert query delivery mode will be set to 2 (marks messages as 'persistent'). Default: `0`. @@ -148,4 +148,5 @@ Example: - `_channel_id` - ChannelID, on which consumer, who received the message, was declared. - `_delivery_tag` - DeliveryTag of the received message. Scoped per channel. - `_redelivered` - `redelivered` flag of the message. -- `_message_id` - MessageID of the received message; non-empty if was set, when message was published. +- `_message_id` - messageID of the received message; non-empty if was set, when message was published. +- `_timestamp` - timestamp of the received message; non-empty if was set, when message was published. diff --git a/docs/en/engines/table-engines/mergetree-family/mergetree.md b/docs/en/engines/table-engines/mergetree-family/mergetree.md index c96e31a6eed..d48c05326c1 100644 --- a/docs/en/engines/table-engines/mergetree-family/mergetree.md +++ b/docs/en/engines/table-engines/mergetree-family/mergetree.md @@ -88,7 +88,7 @@ For a description of parameters, see the [CREATE query description](../../../sql - `index_granularity` — Maximum number of data rows between the marks of an index. Default value: 8192. See [Data Storage](#mergetree-data-storage). - `index_granularity_bytes` — Maximum size of data granules in bytes. Default value: 10Mb. To restrict the granule size only by number of rows, set to 0 (not recommended). See [Data Storage](#mergetree-data-storage). - - `min_index_granularity_bytes` — Min allowed size of data granules in bytes. Default value: 1024b. To provide safeguard against accidentally creating tables with very low index_granularity_bytes. See [Data Storage](#mergetree-data-storage). + - `min_index_granularity_bytes` — Min allowed size of data granules in bytes. Default value: 1024b. To provide a safeguard against accidentally creating tables with very low index_granularity_bytes. See [Data Storage](#mergetree-data-storage). - `enable_mixed_granularity_parts` — Enables or disables transitioning to control the granule size with the `index_granularity_bytes` setting. Before version 19.11, there was only the `index_granularity` setting for restricting granule size. The `index_granularity_bytes` setting improves ClickHouse performance when selecting data from tables with big rows (tens and hundreds of megabytes). If you have tables with big rows, you can enable this setting for the tables to improve the efficiency of `SELECT` queries. - `use_minimalistic_part_header_in_zookeeper` — Storage method of the data parts headers in ZooKeeper. If `use_minimalistic_part_header_in_zookeeper=1`, then ZooKeeper stores less data. For more information, see the [setting description](../../../operations/server-configuration-parameters/settings.md#server-settings-use_minimalistic_part_header_in_zookeeper) in “Server configuration parameters”. - `min_merge_bytes_to_use_direct_io` — The minimum data volume for merge operation that is required for using direct I/O access to the storage disk. When merging data parts, ClickHouse calculates the total storage volume of all the data to be merged. If the volume exceeds `min_merge_bytes_to_use_direct_io` bytes, ClickHouse reads and writes the data to the storage disk using the direct I/O interface (`O_DIRECT` option). If `min_merge_bytes_to_use_direct_io = 0`, then direct I/O is disabled. Default value: `10 * 1024 * 1024 * 1024` bytes. diff --git a/docs/en/engines/table-engines/mergetree-family/replication.md b/docs/en/engines/table-engines/mergetree-family/replication.md index cdb2d718415..932facc9ddc 100644 --- a/docs/en/engines/table-engines/mergetree-family/replication.md +++ b/docs/en/engines/table-engines/mergetree-family/replication.md @@ -117,7 +117,9 @@ CREATE TABLE table_name -As the example shows, these parameters can contain substitutions in curly brackets. The substituted values are taken from the ‘macros’ section of the configuration file. Example: +As the example shows, these parameters can contain substitutions in curly brackets. The substituted values are taken from the «[macros](../../../operations/server-configuration-parameters/settings/#macros) section of the configuration file. + +Example: ``` xml @@ -137,12 +139,40 @@ In this case, the path consists of the following parts: `table_name` is the name of the node for the table in ZooKeeper. It is a good idea to make it the same as the table name. It is defined explicitly, because in contrast to the table name, it doesn’t change after a RENAME query. *HINT*: you could add a database name in front of `table_name` as well. E.g. `db_name.table_name` +The two built-in substitutions `{database}` and `{table}` can be used, they expand into the table name and the database name respectively (unless these macros are defined in the `macros` section). So the zookeeper path can be specified as `'/clickhouse/tables/{layer}-{shard}/{database}/{table}'`. +Be careful with table renames when using these built-in substitutions. The path in Zookeeper cannot be changed, and when the table is renamed, the macros will expand into a different path, the table will refer to a path that does not exist in Zookeeper, and will go into read-only mode. + The replica name identifies different replicas of the same table. You can use the server name for this, as in the example. The name only needs to be unique within each shard. You can define the parameters explicitly instead of using substitutions. This might be convenient for testing and for configuring small clusters. However, you can’t use distributed DDL queries (`ON CLUSTER`) in this case. When working with large clusters, we recommend using substitutions because they reduce the probability of error. +You can specify default arguments for `Replicated` table engine in the server configuration file. For instance: + +```xml +/clickhouse/tables/{shard}/{database}/{table} +{replica} +``` + +In this case, you can omit arguments when creating tables: + +``` sql +CREATE TABLE table_name ( + x UInt32 +) ENGINE = ReplicatedMergeTree +ORDER BY x; +``` + +It is equivalent to: + +``` sql +CREATE TABLE table_name ( + x UInt32 +) ENGINE = ReplicatedMergeTree('/clickhouse/tables/{shard}/{database}/table_name', '{replica}') +ORDER BY x; +``` + Run the `CREATE TABLE` query on each replica. This query creates a new replicated table, or adds a new replica to an existing one. If you add a new replica after the table already contains some data on other replicas, the data will be copied from the other replicas to the new one after running the query. In other words, the new replica syncs itself with the others. diff --git a/docs/en/faq/integration/json-import.md b/docs/en/faq/integration/json-import.md index 067b407a079..fb94f226f2b 100644 --- a/docs/en/faq/integration/json-import.md +++ b/docs/en/faq/integration/json-import.md @@ -30,4 +30,4 @@ Instead of inserting data manually, you might consider to use one of [client lib - `input_format_import_nested_json` allows to insert nested JSON objects into columns of [Nested](../../sql-reference/data-types/nested-data-structures/nested.md) type. !!! note "Note" - Settings are specified as `GET` parameters for the HTTP interface or as additional command-line arguments prefixed with `--` for the CLI interface. + Settings are specified as `GET` parameters for the HTTP interface or as additional command-line arguments prefixed with `--` for the `CLI` interface. \ No newline at end of file diff --git a/docs/en/getting-started/example-datasets/amplab-benchmark.md b/docs/en/getting-started/example-datasets/amplab-benchmark.md index 1b740f6194c..27ddfd27f78 100644 --- a/docs/en/getting-started/example-datasets/amplab-benchmark.md +++ b/docs/en/getting-started/example-datasets/amplab-benchmark.md @@ -1,5 +1,5 @@ --- -toc_priority: 17 +toc_priority: 19 toc_title: AMPLab Big Data Benchmark --- diff --git a/docs/en/getting-started/example-datasets/criteo.md b/docs/en/getting-started/example-datasets/criteo.md index a38a1b14210..261d1606fa4 100644 --- a/docs/en/getting-started/example-datasets/criteo.md +++ b/docs/en/getting-started/example-datasets/criteo.md @@ -1,5 +1,5 @@ --- -toc_priority: 19 +toc_priority: 18 toc_title: Terabyte Click Logs from Criteo --- diff --git a/docs/en/getting-started/example-datasets/index.md b/docs/en/getting-started/example-datasets/index.md index eedf4196b23..35ac90f9beb 100644 --- a/docs/en/getting-started/example-datasets/index.md +++ b/docs/en/getting-started/example-datasets/index.md @@ -1,6 +1,6 @@ --- toc_folder_title: Example Datasets -toc_priority: 15 +toc_priority: 14 toc_title: Introduction --- @@ -18,4 +18,4 @@ The list of documented datasets: - [New York Taxi Data](../../getting-started/example-datasets/nyc-taxi.md) - [OnTime](../../getting-started/example-datasets/ontime.md) -[Original article](https://clickhouse.tech/docs/en/getting_started/example_datasets) +[Original article](https://clickhouse.tech/docs/en/getting_started/example_datasets) \ No newline at end of file diff --git a/docs/en/getting-started/example-datasets/metrica.md b/docs/en/getting-started/example-datasets/metrica.md index 4131dca78fe..b036973b255 100644 --- a/docs/en/getting-started/example-datasets/metrica.md +++ b/docs/en/getting-started/example-datasets/metrica.md @@ -1,5 +1,5 @@ --- -toc_priority: 14 +toc_priority: 15 toc_title: Yandex.Metrica Data --- diff --git a/docs/en/getting-started/example-datasets/nyc-taxi.md b/docs/en/getting-started/example-datasets/nyc-taxi.md index a180f19be8a..9b9a12ba724 100644 --- a/docs/en/getting-started/example-datasets/nyc-taxi.md +++ b/docs/en/getting-started/example-datasets/nyc-taxi.md @@ -1,5 +1,5 @@ --- -toc_priority: 16 +toc_priority: 20 toc_title: New York Taxi Data --- diff --git a/docs/en/getting-started/example-datasets/ontime.md b/docs/en/getting-started/example-datasets/ontime.md index 91b6913e223..c2c8d5d930e 100644 --- a/docs/en/getting-started/example-datasets/ontime.md +++ b/docs/en/getting-started/example-datasets/ontime.md @@ -1,5 +1,5 @@ --- -toc_priority: 15 +toc_priority: 21 toc_title: OnTime --- diff --git a/docs/en/getting-started/example-datasets/star-schema.md b/docs/en/getting-started/example-datasets/star-schema.md index 7940807ee49..8a019e7ded3 100644 --- a/docs/en/getting-started/example-datasets/star-schema.md +++ b/docs/en/getting-started/example-datasets/star-schema.md @@ -1,5 +1,5 @@ --- -toc_priority: 20 +toc_priority: 16 toc_title: Star Schema Benchmark --- diff --git a/docs/en/getting-started/example-datasets/wikistat.md b/docs/en/getting-started/example-datasets/wikistat.md index 874af8c9423..619711582f4 100644 --- a/docs/en/getting-started/example-datasets/wikistat.md +++ b/docs/en/getting-started/example-datasets/wikistat.md @@ -1,5 +1,5 @@ --- -toc_priority: 18 +toc_priority: 17 toc_title: WikiStat --- diff --git a/docs/en/interfaces/formats.md b/docs/en/interfaces/formats.md index d96d48bdca3..d310705d1c1 100644 --- a/docs/en/interfaces/formats.md +++ b/docs/en/interfaces/formats.md @@ -460,7 +460,7 @@ See also the [JSONEachRow](#jsoneachrow) format. ## JSONString {#jsonstring} -Differs from JSON only in that data fields are output in strings, not in typed json values. +Differs from JSON only in that data fields are output in strings, not in typed JSON values. Example: @@ -596,7 +596,7 @@ When inserting the data, you should provide a separate JSON value for each row. ## JSONEachRowWithProgress {#jsoneachrowwithprogress} ## JSONStringEachRowWithProgress {#jsonstringeachrowwithprogress} -Differs from JSONEachRow/JSONStringEachRow in that ClickHouse will also yield progress information as JSON objects. +Differs from `JSONEachRow`/`JSONStringEachRow` in that ClickHouse will also yield progress information as JSON values. ```json {"row":{"'hello'":"hello","multiply(42, number)":"0","range(5)":[0,1,2,3,4]}} @@ -608,7 +608,7 @@ Differs from JSONEachRow/JSONStringEachRow in that ClickHouse will also yield pr ## JSONCompactEachRowWithNamesAndTypes {#jsoncompacteachrowwithnamesandtypes} ## JSONCompactStringEachRowWithNamesAndTypes {#jsoncompactstringeachrowwithnamesandtypes} -Differs from JSONCompactEachRow/JSONCompactStringEachRow in that the column names and types are written as the first two rows. +Differs from `JSONCompactEachRow`/`JSONCompactStringEachRow` in that the column names and types are written as the first two rows. ```json ["'hello'", "multiply(42, number)", "range(5)"] diff --git a/docs/en/interfaces/http.md b/docs/en/interfaces/http.md index 4840737fd48..310286e3d44 100644 --- a/docs/en/interfaces/http.md +++ b/docs/en/interfaces/http.md @@ -79,7 +79,7 @@ By default, data is returned in TabSeparated format (for more information, see t You use the FORMAT clause of the query to request any other format. -Also, you can use the ‘default_format’ URL parameter or ‘X-ClickHouse-Format’ header to specify a default format other than TabSeparated. +Also, you can use the ‘default_format’ URL parameter or the ‘X-ClickHouse-Format’ header to specify a default format other than TabSeparated. ``` bash $ echo 'SELECT 1 FORMAT Pretty' | curl 'http://localhost:8123/?' --data-binary @- @@ -170,7 +170,7 @@ $ echo "SELECT 1" | gzip -c | curl -sS --data-binary @- -H 'Content-Encoding: gz !!! note "Note" Some HTTP clients might decompress data from the server by default (with `gzip` and `deflate`) and you might get decompressed data even if you use the compression settings correctly. -You can use the ‘database’ URL parameter or ‘X-ClickHouse-Database’ header to specify the default database. +You can use the ‘database’ URL parameter or the ‘X-ClickHouse-Database’ header to specify the default database. ``` bash $ echo 'SELECT number FROM numbers LIMIT 10' | curl 'http://localhost:8123/?database=system' --data-binary @- diff --git a/docs/en/interfaces/third-party/client-libraries.md b/docs/en/interfaces/third-party/client-libraries.md index bb98c5f936b..4e8347c9a6e 100644 --- a/docs/en/interfaces/third-party/client-libraries.md +++ b/docs/en/interfaces/third-party/client-libraries.md @@ -6,7 +6,7 @@ toc_title: Client Libraries # Client Libraries from Third-party Developers {#client-libraries-from-third-party-developers} !!! warning "Disclaimer" - Yandex does **not** maintain the libraries listed below and haven’t done any extensive testing to ensure their quality. + Yandex does **not** maintain the libraries listed below and hasn’t done any extensive testing to ensure their quality. - Python - [infi.clickhouse_orm](https://github.com/Infinidat/infi.clickhouse_orm) diff --git a/docs/en/introduction/adopters.md b/docs/en/introduction/adopters.md index b85aa9b261f..d08e7833b33 100644 --- a/docs/en/introduction/adopters.md +++ b/docs/en/introduction/adopters.md @@ -77,6 +77,7 @@ toc_title: Adopters | Rambler | Internet services | Analytics | — | — | [Talk in Russian, April 2018](https://medium.com/@ramblertop/разработка-api-clickhouse-для-рамблер-топ-100-f4c7e56f3141) | | Retell | Speech synthesis | Analytics | — | — | [Blog Article, August 2020](https://vc.ru/services/153732-kak-sozdat-audiostati-na-vashem-sayte-i-zachem-eto-nuzhno) | | Rspamd | Antispam | Analytics | — | — | [Official Website](https://rspamd.com/doc/modules/clickhouse.html) | +| RuSIEM | SIEM | Main Product | — | — | [Official Website](https://rusiem.com/en/products/architecture) | | S7 Airlines | Airlines | Metrics, Logging | — | — | [Talk in Russian, March 2019](https://www.youtube.com/watch?v=nwG68klRpPg&t=15s) | | scireum GmbH | e-Commerce | Main product | — | — | [Talk in German, February 2020](https://www.youtube.com/watch?v=7QWAn5RbyR4) | | Segment | Data processing | Main product | 9 * i3en.3xlarge nodes 7.5TB NVME SSDs, 96GB Memory, 12 vCPUs | — | [Slides, 2019](https://slides.com/abraithwaite/segment-clickhouse) | @@ -88,6 +89,8 @@ toc_title: Adopters | SMI2 | News | Analytics | — | — | [Blog Post in Russian, November 2017](https://habr.com/ru/company/smi2/blog/314558/) | | Splunk | Business Analytics | Main product | — | — | [Slides in English, January 2018](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup12/splunk.pdf) | | Spotify | Music | Experimentation | — | — | [Slides, July 2018](https://www.slideshare.net/glebus/using-clickhouse-for-experimentation-104247173) | +| Staffcop | Information Security | Main Product | — | — | [Official website, Documentation](https://www.staffcop.ru/sce43) | +| Teralytics | Mobility | Analytics | — | — | [Tech blog](https://www.teralytics.net/knowledge-hub/visualizing-mobility-data-the-scalability-challenge) | | Tencent | Big Data | Data processing | — | — | [Slides in Chinese, October 2018](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup19/5.%20ClickHouse大数据集群应用_李俊飞腾讯网媒事业部.pdf) | | Tencent | Messaging | Logging | — | — | [Talk in Chinese, November 2019](https://youtu.be/T-iVQRuw-QY?t=5050) | | Traffic Stars | AD network | — | — | — | [Slides in Russian, May 2018](https://github.com/ClickHouse/clickhouse-presentations/blob/master/meetup15/lightning/ninja.pdf) | diff --git a/docs/en/operations/opentelemetry.md b/docs/en/operations/opentelemetry.md new file mode 100644 index 00000000000..45533d3733f --- /dev/null +++ b/docs/en/operations/opentelemetry.md @@ -0,0 +1,69 @@ +--- +toc_priority: 62 +toc_title: OpenTelemetry Support +--- + +# [experimental] OpenTelemetry Support + +[OpenTelemetry](https://opentelemetry.io/) is an open standard for collecting +traces and metrics from distributed application. ClickHouse has some support +for OpenTelemetry. + +!!! warning "Warning" +This is an experimental feature that will change in backwards-incompatible ways in the future releases. + + +## Supplying Trace Context to ClickHouse + +ClickHouse accepts trace context HTTP headers, as described by +the [W3C recommendation](https://www.w3.org/TR/trace-context/). +It also accepts trace context over native protocol that is used for +communication between ClickHouse servers or between the client and server. +For manual testing, trace context headers conforming to the Trace Context +recommendation can be supplied to `clickhouse-client` using +`--opentelemetry-traceparent` and `--opentelemetry-tracestate` flags. + +If no parent trace context is supplied, ClickHouse can start a new trace, with +probability controlled by the `opentelemetry_start_trace_probability` setting. + + +## Propagating the Trace Context + +The trace context is propagated to downstream services in the following cases: + +* Queries to remote ClickHouse servers, such as when using `Distributed` table + engine. + +* `URL` table function. Trace context information is sent in HTTP headers. + + +## Tracing the ClickHouse Itself + +ClickHouse creates _trace spans_ for each query and some of the query execution +stages, such as query planning or distributed queries. + +To be useful, the tracing information has to be exported to a monitoring system +that supports OpenTelemetry, such as Jaeger or Prometheus. ClickHouse avoids +a dependency on a particular monitoring system, instead only +providing the tracing data conforming to the standard. A natural way to do so +in an SQL RDBMS is a system table. OpenTelemetry trace span information +[required by the standard](https://github.com/open-telemetry/opentelemetry-specification/blob/master/specification/overview.md#span) +is stored in the system table called `system.opentelemetry_span_log`. + +The table must be enabled in the server configuration, see the `opentelemetry_span_log` +element in the default config file `config.xml`. It is enabled by default. + +The table has the following columns: + +- `trace_id` +- `span_id` +- `parent_span_id` +- `operation_name` +- `start_time` +- `finish_time` +- `finish_date` +- `attribute.name` +- `attribute.values` + +The tags or attributes are saved as two parallel arrays, containing the keys +and values. Use `ARRAY JOIN` to work with them. diff --git a/docs/en/operations/server-configuration-parameters/settings.md b/docs/en/operations/server-configuration-parameters/settings.md index 368ef8e6f7c..31a8e896438 100644 --- a/docs/en/operations/server-configuration-parameters/settings.md +++ b/docs/en/operations/server-configuration-parameters/settings.md @@ -479,6 +479,26 @@ The maximum number of simultaneously processed requests. 100 ``` +## max_concurrent_queries_for_all_users {#max-concurrent-queries-for-all-users} + +Throw exception if the value of this setting is less or equal than the current number of simultaneously processed queries. + +Example: `max_concurrent_queries_for_all_users` can be set to 99 for all users and database administrator can set it to 100 for itself to run queries for investigation even when the server is overloaded. + +Modifying the setting for one query or user does not affect other queries. + +Default value: `0` that means no limit. + +**Example** + +``` xml +99 +``` + +**See Also** + +- [max_concurrent_queries](#max-concurrent-queries) + ## max_connections {#max-connections} The maximum number of inbound connections. diff --git a/docs/en/operations/settings/settings.md b/docs/en/operations/settings/settings.md index f93a8b680b5..359a3579e46 100644 --- a/docs/en/operations/settings/settings.md +++ b/docs/en/operations/settings/settings.md @@ -305,6 +305,10 @@ When enabled, replace empty input fields in TSV with default values. For complex Disabled by default. +## input_format_tsv_enum_as_number {#settings-input_format_tsv_enum_as_number} + +For TSV input format switches to parsing enum values as enum ids. + ## input_format_null_as_default {#settings-input-format-null-as-default} Enables or disables using default values if input data contain `NULL`, but the data type of the corresponding column in not `Nullable(T)` (for text input formats). @@ -676,6 +680,21 @@ Example: log_queries=1 ``` +## log_queries_min_query_duration_ms {#settings-log-queries-min-query-duration-ms} + +Minimal time for the query to run to get to the following tables: + +- `system.query_log` +- `system.query_thread_log` + +Only the queries with the following type will get to the log: + +- `QUERY_FINISH` +- `EXCEPTION_WHILE_PROCESSING` + +- Type: milliseconds +- Default value: 0 (any query) + ## log_queries_min_type {#settings-log-queries-min-type} `query_log` minimal type to log. @@ -1161,6 +1180,10 @@ The character is interpreted as a delimiter in the CSV data. By default, the del For CSV input format enables or disables parsing of unquoted `NULL` as literal (synonym for `\N`). +## input_format_csv_enum_as_number {#settings-input_format_csv_enum_as_number} + +For CSV input format switches to parsing enum values as enum ids. + ## output_format_csv_crlf_end_of_line {#settings-output-format-csv-crlf-end-of-line} Use DOS/Windows-style line separator (CRLF) in CSV instead of Unix style (LF). @@ -1398,6 +1421,17 @@ Possible values: Default value: 0 +## allow_nondeterministic_optimize_skip_unused_shards {#allow-nondeterministic-optimize-skip-unused-shards} + +Allow nondeterministic (like `rand` or `dictGet`, since later has some caveats with updates) functions in sharding key. + +Possible values: + +- 0 — Disallowed. +- 1 — Allowed. + +Default value: 0 + ## optimize_skip_unused_shards_nesting {#optimize-skip-unused-shards-nesting} Controls [`optimize_skip_unused_shards`](#optimize-skip-unused-shards) (hence still requires [`optimize_skip_unused_shards`](#optimize-skip-unused-shards)) depends on the nesting level of the distributed query (case when you have `Distributed` table that look into another `Distributed` table). @@ -2129,7 +2163,34 @@ Result: └───────────────┘ ``` -[Original article](https://clickhouse.tech/docs/en/operations/settings/settings/) +## output_format_pretty_row_numbers {#output_format_pretty_row_numbers} + +Adds row numbers to output in the [Pretty](../../interfaces/formats.md#pretty) format. + +Possible values: + +- 0 — Output without row numbers. +- 1 — Output with row numbers. + +Default value: `0`. + +**Example** + +Query: + +```sql +SET output_format_pretty_row_numbers = 1; +SELECT TOP 3 name, value FROM system.settings; +``` + +Result: +```text + ┌─name────────────────────┬─value───┐ +1. │ min_compress_block_size │ 65536 │ +2. │ max_compress_block_size │ 1048576 │ +3. │ max_block_size │ 65505 │ + └─────────────────────────┴─────────┘ +``` ## allow_experimental_bigint_types {#allow_experimental_bigint_types} @@ -2141,3 +2202,5 @@ Possible values: - 0 — The bigint data type is disabled. Default value: `0`. + +[Original article](https://clickhouse.tech/docs/en/operations/settings/settings/) diff --git a/docs/en/operations/system-tables/asynchronous_metric_log.md b/docs/en/operations/system-tables/asynchronous_metric_log.md index 75607cc30b0..5dcfca5fbda 100644 --- a/docs/en/operations/system-tables/asynchronous_metric_log.md +++ b/docs/en/operations/system-tables/asynchronous_metric_log.md @@ -1,6 +1,6 @@ ## system.asynchronous_metric_log {#system-tables-async-log} -Contains the historical values for `system.asynchronous_metrics`, which are saved once per minute. This feature is enabled by default. +Contains the historical values for `system.asynchronous_metrics`, which are saved once per minute. Enabled by default. Columns: @@ -33,7 +33,7 @@ SELECT * FROM system.asynchronous_metric_log LIMIT 10 **See Also** -- [system.asynchronous_metrics](../system-tables/asynchronous_metrics.md) — Contains metrics that are calculated periodically in the background. +- [system.asynchronous_metrics](../system-tables/asynchronous_metrics.md) — Contains metrics, calculated periodically in the background. - [system.metric_log](../system-tables/metric_log.md) — Contains history of metrics values from tables `system.metrics` and `system.events`, periodically flushed to disk. [Original article](https://clickhouse.tech/docs/en/operations/system_tables/asynchronous_metric_log) diff --git a/docs/en/operations/system-tables/columns.md b/docs/en/operations/system-tables/columns.md index 4d8077ddeac..92cbdd19ca8 100644 --- a/docs/en/operations/system-tables/columns.md +++ b/docs/en/operations/system-tables/columns.md @@ -6,19 +6,21 @@ You can use this table to get information similar to the [DESCRIBE TABLE](../../ The `system.columns` table contains the following columns (the column type is shown in brackets): -- `database` (String) — Database name. -- `table` (String) — Table name. -- `name` (String) — Column name. -- `type` (String) — Column type. -- `default_kind` (String) — Expression type (`DEFAULT`, `MATERIALIZED`, `ALIAS`) for the default value, or an empty string if it is not defined. -- `default_expression` (String) — Expression for the default value, or an empty string if it is not defined. -- `data_compressed_bytes` (UInt64) — The size of compressed data, in bytes. -- `data_uncompressed_bytes` (UInt64) — The size of decompressed data, in bytes. -- `marks_bytes` (UInt64) — The size of marks, in bytes. -- `comment` (String) — Comment on the column, or an empty string if it is not defined. -- `is_in_partition_key` (UInt8) — Flag that indicates whether the column is in the partition expression. -- `is_in_sorting_key` (UInt8) — Flag that indicates whether the column is in the sorting key expression. -- `is_in_primary_key` (UInt8) — Flag that indicates whether the column is in the primary key expression. -- `is_in_sampling_key` (UInt8) — Flag that indicates whether the column is in the sampling key expression. +- `database` ([String](../../sql-reference/data-types/string.md)) — Database name. +- `table` ([String](../../sql-reference/data-types/string.md)) — Table name. +- `name` ([String](../../sql-reference/data-types/string.md)) — Column name. +- `type` ([String](../../sql-reference/data-types/string.md)) — Column type. +- `position` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Ordinal position of a column in a table starting with 1. +- `default_kind` ([String](../../sql-reference/data-types/string.md)) — Expression type (`DEFAULT`, `MATERIALIZED`, `ALIAS`) for the default value, or an empty string if it is not defined. +- `default_expression` ([String](../../sql-reference/data-types/string.md)) — Expression for the default value, or an empty string if it is not defined. +- `data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The size of compressed data, in bytes. +- `data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The size of decompressed data, in bytes. +- `marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The size of marks, in bytes. +- `comment` ([String](../../sql-reference/data-types/string.md)) — Comment on the column, or an empty string if it is not defined. +- `is_in_partition_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — Flag that indicates whether the column is in the partition expression. +- `is_in_sorting_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — Flag that indicates whether the column is in the sorting key expression. +- `is_in_primary_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — Flag that indicates whether the column is in the primary key expression. +- `is_in_sampling_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — Flag that indicates whether the column is in the sampling key expression. +- `compression_codec` ([String](../../sql-reference/data-types/string.md)) — Compression codec name. [Original article](https://clickhouse.tech/docs/en/operations/system_tables/columns) diff --git a/docs/en/operations/system-tables/crash-log.md b/docs/en/operations/system-tables/crash-log.md new file mode 100644 index 00000000000..5e9fec53429 --- /dev/null +++ b/docs/en/operations/system-tables/crash-log.md @@ -0,0 +1,48 @@ +# system.crash_log {#system-tables_crash_log} + +Contains information about stack traces for fatal errors. The table does not exist in the database by default, it is created only when fatal errors occur. + +Columns: + +- `event_date` ([Datetime](../../sql-reference/data-types/datetime.md)) — Date of the event. +- `event_time` ([Datetime](../../sql-reference/data-types/datetime.md)) — Time of the event. +- `timestamp_ns` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Timestamp of the event with nanoseconds. +- `signal` ([Int32](../../sql-reference/data-types/int-uint.md)) — Signal number. +- `thread_id` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Thread ID. +- `query_id` ([String](../../sql-reference/data-types/string.md)) — Query ID. +- `trace` ([Array](../../sql-reference/data-types/array.md)([UInt64](../../sql-reference/data-types/int-uint.md))) — Stack trace at the moment of crash. Each element is a virtual memory address inside ClickHouse server process. +- `trace_full` ([Array](../../sql-reference/data-types/array.md)([String](../../sql-reference/data-types/string.md))) — Stack trace at the moment of crash. Each element contains a called method inside ClickHouse server process. +- `version` ([String](../../sql-reference/data-types/string.md)) — ClickHouse server version. +- `revision` ([UInt32](../../sql-reference/data-types/int-uint.md)) — ClickHouse server revision. +- `build_id` ([String](../../sql-reference/data-types/string.md)) — BuildID that is generated by compiler. + +**Example** + +Query: + +``` sql +SELECT * FROM system.crash_log ORDER BY event_time DESC LIMIT 1; +``` + +Result (not full): + +``` text +Row 1: +────── +event_date: 2020-10-14 +event_time: 2020-10-14 15:47:40 +timestamp_ns: 1602679660271312710 +signal: 11 +thread_id: 23624 +query_id: 428aab7c-8f5c-44e9-9607-d16b44467e69 +trace: [188531193,...] +trace_full: ['3. DB::(anonymous namespace)::FunctionFormatReadableTimeDelta::executeImpl(std::__1::vector >&, std::__1::vector > const&, unsigned long, unsigned long) const @ 0xb3cc1f9 in /home/username/work/ClickHouse/build/programs/clickhouse',...] +version: ClickHouse 20.11.1.1 +revision: 54442 +build_id: +``` + +**See also** +- [trace_log](../../operations/system-tables/trace_log.md) system table + +[Original article](https://clickhouse.tech/docs/en/operations/system-tables/crash-log) diff --git a/docs/en/operations/system-tables/errors.md b/docs/en/operations/system-tables/errors.md new file mode 100644 index 00000000000..53e8a397217 --- /dev/null +++ b/docs/en/operations/system-tables/errors.md @@ -0,0 +1,23 @@ +# system.errors {#system_tables-errors} + +Contains error codes with number of times they have been triggered. + +Columns: + +- `name` ([String](../../sql-reference/data-types/string.md)) — name of the error (`errorCodeToName`). +- `code` ([Int32](../../sql-reference/data-types/int-uint.md)) — code number of the error. +- `value` ([UInt64](../../sql-reference/data-types/int-uint.md)) - number of times this error has been happened. + +**Example** + +``` sql +SELECT * +FROM system.errors +WHERE value > 0 +ORDER BY code ASC +LIMIT 1 + +┌─name─────────────┬─code─┬─value─┐ +│ CANNOT_OPEN_FILE │ 76 │ 1 │ +└──────────────────┴──────┴───────┘ +``` diff --git a/docs/en/operations/system-tables/metric_log.md b/docs/en/operations/system-tables/metric_log.md index f1910407949..c6aefc5034f 100644 --- a/docs/en/operations/system-tables/metric_log.md +++ b/docs/en/operations/system-tables/metric_log.md @@ -1,6 +1,7 @@ # system.metric_log {#system_tables-metric_log} Contains history of metrics values from tables `system.metrics` and `system.events`, periodically flushed to disk. + To turn on metrics history collection on `system.metric_log`, create `/etc/clickhouse-server/config.d/metric_log.xml` with following content: ``` xml @@ -14,6 +15,11 @@ To turn on metrics history collection on `system.metric_log`, create `/etc/click ``` +Columns: +- `event_date` ([Date](../../sql-reference/data-types/date.md)) — Event date. +- `event_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — Event time. +- `event_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — Event time with microseconds resolution. + **Example** ``` sql diff --git a/docs/en/operations/system-tables/parts_columns.md b/docs/en/operations/system-tables/parts_columns.md new file mode 100644 index 00000000000..5c3dd7155f7 --- /dev/null +++ b/docs/en/operations/system-tables/parts_columns.md @@ -0,0 +1,148 @@ +# system.parts_columns {#system_tables-parts_columns} + +Contains information about parts and columns of [MergeTree](../../engines/table-engines/mergetree-family/mergetree.md) tables. + +Each row describes one data part. + +Columns: + +- `partition` ([String](../../sql-reference/data-types/string.md)) — The partition name. To learn what a partition is, see the description of the [ALTER](../../sql-reference/statements/alter/index.md#query_language_queries_alter) query. + + Formats: + + - `YYYYMM` for automatic partitioning by month. + - `any_string` when partitioning manually. + +- `name` ([String](../../sql-reference/data-types/string.md)) — Name of the data part. + +- `part_type` ([String](../../sql-reference/data-types/string.md)) — The data part storing format. + + Possible values: + + - `Wide` — Each column is stored in a separate file in a filesystem. + - `Compact` — All columns are stored in one file in a filesystem. + + Data storing format is controlled by the `min_bytes_for_wide_part` and `min_rows_for_wide_part` settings of the [MergeTree](../../engines/table-engines/mergetree-family/mergetree.md) table. + +- `active` ([UInt8](../../sql-reference/data-types/int-uint.md)) — Flag that indicates whether the data part is active. If a data part is active, it’s used in a table. Otherwise, it’s deleted. Inactive data parts remain after merging. + +- `marks` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The number of marks. To get the approximate number of rows in a data part, multiply `marks` by the index granularity (usually 8192) (this hint doesn’t work for adaptive granularity). + +- `rows` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The number of rows. + +- `bytes_on_disk` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of all the data part files in bytes. + +- `data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of compressed data in the data part. All the auxiliary files (for example, files with marks) are not included. + +- `data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of uncompressed data in the data part. All the auxiliary files (for example, files with marks) are not included. + +- `marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The size of the file with marks. + +- `modification_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — The time the directory with the data part was modified. This usually corresponds to the time of data part creation. + +- `remove_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — The time when the data part became inactive. + +- `refcount` ([UInt32](../../sql-reference/data-types/int-uint.md)) — The number of places where the data part is used. A value greater than 2 indicates that the data part is used in queries or merges. + +- `min_date` ([Date](../../sql-reference/data-types/date.md)) — The minimum value of the date key in the data part. + +- `max_date` ([Date](../../sql-reference/data-types/date.md)) — The maximum value of the date key in the data part. + +- `partition_id` ([String](../../sql-reference/data-types/string.md)) — ID of the partition. + +- `min_block_number` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The minimum number of data parts that make up the current part after merging. + +- `max_block_number` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The maximum number of data parts that make up the current part after merging. + +- `level` ([UInt32](../../sql-reference/data-types/int-uint.md)) — Depth of the merge tree. Zero means that the current part was created by insert rather than by merging other parts. + +- `data_version` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Number that is used to determine which mutations should be applied to the data part (mutations with a version higher than `data_version`). + +- `primary_key_bytes_in_memory` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The amount of memory (in bytes) used by primary key values. + +- `primary_key_bytes_in_memory_allocated` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The amount of memory (in bytes) reserved for primary key values. + +- `database` ([String](../../sql-reference/data-types/string.md)) — Name of the database. + +- `table` ([String](../../sql-reference/data-types/string.md)) — Name of the table. + +- `engine` ([String](../../sql-reference/data-types/string.md)) — Name of the table engine without parameters. + +- `disk_name` ([String](../../sql-reference/data-types/string.md)) — Name of a disk that stores the data part. + +- `path` ([String](../../sql-reference/data-types/string.md)) — Absolute path to the folder with data part files. + +- `column` ([String](../../sql-reference/data-types/string.md)) — Name of the column. + +- `type` ([String](../../sql-reference/data-types/string.md)) — Column type. + +- `column_position` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Ordinal position of a column in a table starting with 1. + +- `default_kind` ([String](../../sql-reference/data-types/string.md)) — Expression type (`DEFAULT`, `MATERIALIZED`, `ALIAS`) for the default value, or an empty string if it is not defined. + +- `default_expression` ([String](../../sql-reference/data-types/string.md)) — Expression for the default value, or an empty string if it is not defined. + +- `column_bytes_on_disk` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of the column in bytes. + +- `column_data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of compressed data in the column, in bytes. + +- `column_data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Total size of the decompressed data in the column, in bytes. + +- `column_marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — The size of the column with marks, in bytes. + +- `bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Alias for `bytes_on_disk`. + +- `marks_size` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Alias for `marks_bytes`. + +**Example** + +``` sql +SELECT * FROM system.parts_columns LIMIT 1 FORMAT Vertical; +``` + +``` text +Row 1: +────── +partition: tuple() +name: all_1_2_1 +part_type: Wide +active: 1 +marks: 2 +rows: 2 +bytes_on_disk: 155 +data_compressed_bytes: 56 +data_uncompressed_bytes: 4 +marks_bytes: 96 +modification_time: 2020-09-23 10:13:36 +remove_time: 2106-02-07 06:28:15 +refcount: 1 +min_date: 1970-01-01 +max_date: 1970-01-01 +partition_id: all +min_block_number: 1 +max_block_number: 2 +level: 1 +data_version: 1 +primary_key_bytes_in_memory: 2 +primary_key_bytes_in_memory_allocated: 64 +database: default +table: 53r93yleapyears +engine: MergeTree +disk_name: default +path: /var/lib/clickhouse/data/default/53r93yleapyears/all_1_2_1/ +column: id +type: Int8 +column_position: 1 +default_kind: +default_expression: +column_bytes_on_disk: 76 +column_data_compressed_bytes: 28 +column_data_uncompressed_bytes: 2 +column_marks_bytes: 48 +``` + +**See Also** + +- [MergeTree family](../../engines/table-engines/mergetree-family/mergetree.md) + +[Original article](https://clickhouse.tech/docs/en/operations/system_tables/parts_columns) diff --git a/docs/en/operations/system-tables/query_log.md b/docs/en/operations/system-tables/query_log.md index ddd5d327d59..32b2bdf2133 100644 --- a/docs/en/operations/system-tables/query_log.md +++ b/docs/en/operations/system-tables/query_log.md @@ -20,8 +20,8 @@ The `system.query_log` table registers two kinds of queries: Each query creates one or two rows in the `query_log` table, depending on the status (see the `type` column) of the query: -1. If the query execution was successful, two rows with the `QueryStart` and `QueryFinish` types are created . -2. If an error occurred during query processing, two events with the `QueryStart` and `ExceptionWhileProcessing` types are created . +1. If the query execution was successful, two rows with the `QueryStart` and `QueryFinish` types are created. +2. If an error occurred during query processing, two events with the `QueryStart` and `ExceptionWhileProcessing` types are created. 3. If an error occurred before launching the query, a single event with the `ExceptionBeforeStart` type is created. Columns: @@ -37,8 +37,8 @@ Columns: - `query_start_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — Start time of query execution. - `query_start_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — Start time of query execution with microsecond precision. - `query_duration_ms` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Duration of query execution in milliseconds. -- `read_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number or rows read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_rows` includes the total number of rows read at all replicas. Each replica sends it’s `read_rows` value, and the server-initiator of the query summarize all received and local values. The cache volumes doesn’t affect this value. -- `read_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number or bytes read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_bytes` includes the total number of rows read at all replicas. Each replica sends it’s `read_bytes` value, and the server-initiator of the query summarize all received and local values. The cache volumes doesn’t affect this value. +- `read_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number of rows read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_rows` includes the total number of rows read at all replicas. Each replica sends it’s `read_rows` value, and the server-initiator of the query summarizes all received and local values. The cache volumes don’t affect this value. +- `read_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Total number of bytes read from all tables and table functions participated in query. It includes usual subqueries, subqueries for `IN` and `JOIN`. For distributed queries `read_bytes` includes the total number of rows read at all replicas. Each replica sends it’s `read_bytes` value, and the server-initiator of the query summarizes all received and local values. The cache volumes don’t affect this value. - `written_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — For `INSERT` queries, the number of written rows. For other queries, the column value is 0. - `written_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — For `INSERT` queries, the number of written bytes. For other queries, the column value is 0. - `result_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Number of rows in a result of the `SELECT` query, or a number of rows in the `INSERT` query. diff --git a/docs/en/operations/system-tables/query_thread_log.md b/docs/en/operations/system-tables/query_thread_log.md index a6f090e99f3..0ae2e7d5d3b 100644 --- a/docs/en/operations/system-tables/query_thread_log.md +++ b/docs/en/operations/system-tables/query_thread_log.md @@ -1,6 +1,6 @@ # system.query_thread_log {#system_tables-query_thread_log} -Contains information about threads which execute queries, for example, thread name, thread start time, duration of query processing. +Contains information about threads that execute queries, for example, thread name, thread start time, duration of query processing. To start logging: diff --git a/docs/en/operations/system-tables/replicas.md b/docs/en/operations/system-tables/replicas.md index f31f7fd27ea..8da68d2d2ab 100644 --- a/docs/en/operations/system-tables/replicas.md +++ b/docs/en/operations/system-tables/replicas.md @@ -53,9 +53,9 @@ Columns: - `table` (`String`) - Table name - `engine` (`String`) - Table engine name - `is_leader` (`UInt8`) - Whether the replica is the leader. - Only one replica at a time can be the leader. The leader is responsible for selecting background merges to perform. + Multiple replicas can be leaders at the same time. A replica can be prevented from becoming a leader using the `merge_tree` setting `replicated_can_become_leader`. The leaders are responsible for scheduling background merges. Note that writes can be performed to any replica that is available and has a session in ZK, regardless of whether it is a leader. -- `can_become_leader` (`UInt8`) - Whether the replica can be elected as a leader. +- `can_become_leader` (`UInt8`) - Whether the replica can be a leader. - `is_readonly` (`UInt8`) - Whether the replica is in read-only mode. This mode is turned on if the config doesn’t have sections with ZooKeeper, if an unknown error occurred when reinitializing sessions in ZooKeeper, and during session reinitialization in ZooKeeper. - `is_session_expired` (`UInt8`) - the session with ZooKeeper has expired. Basically the same as `is_readonly`. diff --git a/docs/en/operations/system-tables/text_log.md b/docs/en/operations/system-tables/text_log.md index 34f37c2cae7..f5f53c95653 100644 --- a/docs/en/operations/system-tables/text_log.md +++ b/docs/en/operations/system-tables/text_log.md @@ -1,6 +1,6 @@ # system.text_log {#system_tables-text_log} -Contains logging entries. Logging level which goes to this table can be limited with `text_log.level` server setting. +Contains logging entries. The logging level which goes to this table can be limited to the `text_log.level` server setting. Columns: diff --git a/docs/en/operations/system-tables/trace_log.md b/docs/en/operations/system-tables/trace_log.md index bc955ebe4de..8107f60b808 100644 --- a/docs/en/operations/system-tables/trace_log.md +++ b/docs/en/operations/system-tables/trace_log.md @@ -18,7 +18,7 @@ Columns: - `revision` ([UInt32](../../sql-reference/data-types/int-uint.md)) — ClickHouse server build revision. - When connecting to server by `clickhouse-client`, you see the string similar to `Connected to ClickHouse server version 19.18.1 revision 54429.`. This field contains the `revision`, but not the `version` of a server. + When connecting to the server by `clickhouse-client`, you see the string similar to `Connected to ClickHouse server version 19.18.1 revision 54429.`. This field contains the `revision`, but not the `version` of a server. - `timer_type` ([Enum8](../../sql-reference/data-types/enum.md)) — Timer type: diff --git a/docs/en/operations/utilities/clickhouse-local.md b/docs/en/operations/utilities/clickhouse-local.md index 132f50fa24e..af3d06898fd 100644 --- a/docs/en/operations/utilities/clickhouse-local.md +++ b/docs/en/operations/utilities/clickhouse-local.md @@ -16,7 +16,7 @@ By default `clickhouse-local` does not have access to data on the same host, but !!! warning "Warning" It is not recommended to load production server configuration into `clickhouse-local` because data can be damaged in case of human error. -For temporary data an unique temporary data directory is created by default. If you want to override this behavior the data directory can be explicitly specified with the `-- --path` option. +For temporary data, a unique temporary data directory is created by default. If you want to override this behavior, the data directory can be explicitly specified with the `-- --path` option. ## Usage {#usage} diff --git a/docs/en/sql-reference/data-types/special-data-types/interval.md b/docs/en/sql-reference/data-types/special-data-types/interval.md index 8a4b9ae7886..7c0c5b00c0d 100644 --- a/docs/en/sql-reference/data-types/special-data-types/interval.md +++ b/docs/en/sql-reference/data-types/special-data-types/interval.md @@ -80,4 +80,4 @@ Code: 43. DB::Exception: Received from localhost:9000. DB::Exception: Wrong argu ## See Also {#see-also} - [INTERVAL](../../../sql-reference/operators/index.md#operator-interval) operator -- [toInterval](../../../sql-reference/functions/type-conversion-functions.md#function-tointerval) type convertion functions +- [toInterval](../../../sql-reference/functions/type-conversion-functions.md#function-tointerval) type conversion functions diff --git a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md index ae2de0ba6f8..a49ddf62eb9 100644 --- a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md +++ b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md @@ -59,7 +59,8 @@ LAYOUT(LAYOUT_TYPE(param value)) -- layout settings - [range_hashed](#range-hashed) - [complex_key_hashed](#complex-key-hashed) - [complex_key_cache](#complex-key-cache) -- [ssd_complex_key_cache](#ssd-cache) +- [ssd_cache](#ssd-cache) +- [ssd_complex_key_cache](#complex-key-ssd-cache) - [complex_key_direct](#complex-key-direct) - [ip_trie](#ip-trie) diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 15214cc4c00..a5a347e553a 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -23,8 +23,6 @@ SELECT └─────────────────────┴────────────┴────────────┴─────────────────────┘ ``` -Only time zones that differ from UTC by a whole number of hours are supported. - ## toTimeZone {#totimezone} Convert time or date and time to the specified time zone. diff --git a/docs/en/sql-reference/functions/encoding-functions.md b/docs/en/sql-reference/functions/encoding-functions.md index 44008fbbcbe..bc3f5ca4345 100644 --- a/docs/en/sql-reference/functions/encoding-functions.md +++ b/docs/en/sql-reference/functions/encoding-functions.md @@ -6,7 +6,7 @@ toc_title: Encoding # Encoding Functions {#encoding-functions} ## char {#char} - + Returns the string with the length as the number of passed arguments and each byte has the value of corresponding argument. Accepts multiple arguments of numeric types. If the value of argument is out of range of UInt8 data type, it is converted to UInt8 with possible rounding and overflow. **Syntax** diff --git a/docs/en/sql-reference/functions/other-functions.md b/docs/en/sql-reference/functions/other-functions.md index 1145efbbc5f..31ed47c3195 100644 --- a/docs/en/sql-reference/functions/other-functions.md +++ b/docs/en/sql-reference/functions/other-functions.md @@ -551,7 +551,7 @@ formatReadableTimeDelta(column[, maximum_unit]) **Parameters** - `column` — A column with numeric time delta. -- `maximum_unit` — Optional. Maximum unit to show. Acceptable values seconds, minutes, hours, days, months, years. +- `maximum_unit` — Optional. Maximum unit to show. Acceptable values seconds, minutes, hours, days, months, years. Example: @@ -626,7 +626,12 @@ neighbor(column, offset[, default_value]) ``` The result of the function depends on the affected data blocks and the order of data in the block. -If you make a subquery with ORDER BY and call the function from outside the subquery, you can get the expected result. + +!!! warning "Warning" + It can reach the neighbor rows only inside the currently processed data block. + +The rows order used during the calculation of `neighbor` can differ from the order of rows returned to the user. +To prevent that you can make a subquery with ORDER BY and call the function from outside the subquery. **Parameters** @@ -731,8 +736,13 @@ Result: Calculates the difference between successive row values ​​in the data block. Returns 0 for the first row and the difference from the previous row for each subsequent row. +!!! warning "Warning" + It can reach the previos row only inside the currently processed data block. + The result of the function depends on the affected data blocks and the order of data in the block. -If you make a subquery with ORDER BY and call the function from outside the subquery, you can get the expected result. + +The rows order used during the calculation of `runningDifference` can differ from the order of rows returned to the user. +To prevent that you can make a subquery with ORDER BY and call the function from outside the subquery. Example: @@ -1584,7 +1594,7 @@ isDecimalOverflow(d, [p]) **Parameters** - `d` — value. [Decimal](../../sql-reference/data-types/decimal.md). -- `p` — precision. Optional. If omitted, the initial presicion of the first argument is used. Using of this paratemer could be helpful for data extraction to another DBMS or file. [UInt8](../../sql-reference/data-types/int-uint.md#uint-ranges). +- `p` — precision. Optional. If omitted, the initial precision of the first argument is used. Using of this paratemer could be helpful for data extraction to another DBMS or file. [UInt8](../../sql-reference/data-types/int-uint.md#uint-ranges). **Returned values** @@ -1647,4 +1657,24 @@ Result: 10 10 19 19 39 39 ``` +## errorCodeToName {#error-code-to-name} + +**Returned value** + +- Variable name for the error code. + +Type: [LowCardinality(String)](../../sql-reference/data-types/lowcardinality.md). + +**Syntax** + +``` sql +errorCodeToName(1) +``` + +Result: + +``` text +UNSUPPORTED_METHOD +``` + [Original article](https://clickhouse.tech/docs/en/query_language/functions/other_functions/) diff --git a/docs/en/sql-reference/functions/string-search-functions.md b/docs/en/sql-reference/functions/string-search-functions.md index 5f08417f349..881139f103c 100644 --- a/docs/en/sql-reference/functions/string-search-functions.md +++ b/docs/en/sql-reference/functions/string-search-functions.md @@ -461,6 +461,66 @@ For other regular expressions, the code is the same as for the ‘match’ funct The same thing as ‘like’, but negative. +## ilike {#ilike} + +Case insensitive variant of [like](https://clickhouse.tech/docs/en/sql-reference/functions/string-search-functions/#function-like) function. You can use `ILIKE` operator instead of the `ilike` function. + +**Syntax** + +``` sql +ilike(haystack, pattern) +``` + +**Parameters** + +- `haystack` — Input string. [String](../../sql-reference/syntax.md#syntax-string-literal). +- `pattern` — If `pattern` doesn't contain percent signs or underscores, then the `pattern` only represents the string itself. An underscore (`_`) in `pattern` stands for (matches) any single character. A percent sign (`%`) matches any sequence of zero or more characters. + +Some `pattern` examples: + +``` text +'abc' ILIKE 'abc' true +'abc' ILIKE 'a%' true +'abc' ILIKE '_b_' true +'abc' ILIKE 'c' false +``` + +**Returned values** + +- True, if the string matches `pattern`. +- False, if the string doesn't match `pattern`. + +**Example** + +Input table: + +``` text +┌─id─┬─name─────┬─days─┐ +│ 1 │ January │ 31 │ +│ 2 │ February │ 29 │ +│ 3 │ March │ 31 │ +│ 4 │ April │ 30 │ +└────┴──────────┴──────┘ +``` + +Query: + +``` sql +SELECT * FROM Months WHERE ilike(name, '%j%') +``` + +Result: + +``` text +┌─id─┬─name────┬─days─┐ +│ 1 │ January │ 31 │ +└────┴─────────┴──────┘ +``` + +**See Also** + +- [like](https://clickhouse.tech/docs/en/sql-reference/functions/string-search-functions/#function-like) + ## ngramDistance(haystack, needle) {#ngramdistancehaystack-needle} Calculates the 4-gram distance between `haystack` and `needle`: counts the symmetric difference between two multisets of 4-grams and normalizes it by the sum of their cardinalities. Returns float number from 0 to 1 – the closer to zero, the more strings are similar to each other. If the constant `needle` or `haystack` is more than 32Kb, throws an exception. If some of the non-constant `haystack` or `needle` strings are more than 32Kb, the distance is always one. diff --git a/docs/en/sql-reference/functions/type-conversion-functions.md b/docs/en/sql-reference/functions/type-conversion-functions.md index ba8d379ccec..fc066a34b0b 100644 --- a/docs/en/sql-reference/functions/type-conversion-functions.md +++ b/docs/en/sql-reference/functions/type-conversion-functions.md @@ -323,6 +323,10 @@ This function accepts a number or date or date with time, and returns a string c This function accepts a number or date or date with time, and returns a FixedString containing bytes representing the corresponding value in host order (little endian). Null bytes are dropped from the end. For example, a UInt32 type value of 255 is a FixedString that is one byte long. +## reinterpretAsUUID {#reinterpretasuuid} + +This function accepts FixedString, and returns UUID. Takes 16 bytes string. If the string isn't long enough, the functions work as if the string is padded with the necessary number of null bytes to the end. If the string longer than 16 bytes, the extra bytes at the end are ignored. + ## CAST(x, T) {#type_conversion_function-cast} Converts ‘x’ to the ‘t’ data type. The syntax CAST(x AS t) is also supported. @@ -780,4 +784,42 @@ Result: └──────────────────────────────────┘ ``` +## formatRowNoNewline {#formatrownonewline} + +Converts arbitrary expressions into a string via given format. The function trims the last `\n` if any. + +**Syntax** + +``` sql +formatRowNoNewline(format, x, y, ...) +``` + +**Parameters** + +- `format` — Text format. For example, [CSV](../../interfaces/formats.md#csv), [TSV](../../interfaces/formats.md#tabseparated). +- `x`,`y`, ... — Expressions. + +**Returned value** + +- A formatted string. + +**Example** + +Query: + +``` sql +SELECT formatRowNoNewline('CSV', number, 'good') +FROM numbers(3) +``` + +Result: + +``` text +┌─formatRowNoNewline('CSV', number, 'good')─┐ +│ 0,"good" │ +│ 1,"good" │ +│ 2,"good" │ +└───────────────────────────────────────────┘ +``` + [Original article](https://clickhouse.tech/docs/en/query_language/functions/type_conversion_functions/) diff --git a/docs/en/sql-reference/functions/uuid-functions.md b/docs/en/sql-reference/functions/uuid-functions.md index f608c643ee8..01a61c65b67 100644 --- a/docs/en/sql-reference/functions/uuid-functions.md +++ b/docs/en/sql-reference/functions/uuid-functions.md @@ -61,6 +61,54 @@ SELECT toUUID('61f0c404-5cb3-11e7-907b-a6006ad3dba0') AS uuid └──────────────────────────────────────┘ ``` +## toUUIDOrNull (x) {#touuidornull-x} + +It takes an argument of type String and tries to parse it into UUID. If failed, returns NULL. + +``` sql +toUUIDOrNull(String) +``` + +**Returned value** + +The Nullable(UUID) type value. + +**Usage example** + +``` sql +SELECT toUUIDOrNull('61f0c404-5cb3-11e7-907b-a6006ad3dba0T') AS uuid +``` + +``` text +┌─uuid─┐ +│ ᴺᵁᴸᴸ │ +└──────┘ +``` + +## toUUIDOrZero (x) {#touuidorzero-x} + +It takes an argument of type String and tries to parse it into UUID. If failed, returns zero UUID. + +``` sql +toUUIDOrZero(String) +``` + +**Returned value** + +The UUID type value. + +**Usage example** + +``` sql +SELECT toUUIDOrZero('61f0c404-5cb3-11e7-907b-a6006ad3dba0T') AS uuid +``` + +``` text +┌─────────────────────────────────uuid─┐ +│ 00000000-0000-0000-0000-000000000000 │ +└──────────────────────────────────────┘ +``` + ## UUIDStringToNum {#uuidstringtonum} Accepts a string containing 36 characters in the format `xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx`, and returns it as a set of bytes in a [FixedString(16)](../../sql-reference/data-types/fixedstring.md). diff --git a/docs/en/sql-reference/operators/index.md b/docs/en/sql-reference/operators/index.md index e07febf9ec9..274f7269bc8 100644 --- a/docs/en/sql-reference/operators/index.md +++ b/docs/en/sql-reference/operators/index.md @@ -1,5 +1,5 @@ --- -toc_priority: 37 +toc_priority: 38 toc_title: Operators --- @@ -53,6 +53,8 @@ ClickHouse transforms operators to their corresponding functions at the query pa `a NOT LIKE s` – The `notLike(a, b)` function. +`a ILIKE s` – The `ilike(a, b)` function. + `a BETWEEN b AND c` – The same as `a >= b AND a <= c`. `a NOT BETWEEN b AND c` – The same as `a < b OR a > c`. @@ -149,25 +151,47 @@ Types of intervals: - `QUARTER` - `YEAR` +You can also use a string literal when setting the `INTERVAL` value. For example, `INTERVAL 1 HOUR` is identical to the `INTERVAL '1 hour'` or `INTERVAL '1' hour`. + !!! warning "Warning" Intervals with different types can’t be combined. You can’t use expressions like `INTERVAL 4 DAY 1 HOUR`. Specify intervals in units that are smaller or equal to the smallest unit of the interval, for example, `INTERVAL 25 HOUR`. You can use consecutive operations, like in the example below. -Example: +Examples: ``` sql -SELECT now() AS current_date_time, current_date_time + INTERVAL 4 DAY + INTERVAL 3 HOUR +SELECT now() AS current_date_time, current_date_time + INTERVAL 4 DAY + INTERVAL 3 HOUR; ``` ``` text ┌───current_date_time─┬─plus(plus(now(), toIntervalDay(4)), toIntervalHour(3))─┐ -│ 2019-10-23 11:16:28 │ 2019-10-27 14:16:28 │ +│ 2020-11-03 22:09:50 │ 2020-11-08 01:09:50 │ └─────────────────────┴────────────────────────────────────────────────────────┘ ``` +``` sql +SELECT now() AS current_date_time, current_date_time + INTERVAL '4 day' + INTERVAL '3 hour'; +``` + +``` text +┌───current_date_time─┬─plus(plus(now(), toIntervalDay(4)), toIntervalHour(3))─┐ +│ 2020-11-03 22:12:10 │ 2020-11-08 01:12:10 │ +└─────────────────────┴────────────────────────────────────────────────────────┘ +``` + +``` sql +SELECT now() AS current_date_time, current_date_time + INTERVAL '4' day + INTERVAL '3' hour; +``` + +``` text +┌───current_date_time─┬─plus(plus(now(), toIntervalDay('4')), toIntervalHour('3'))─┐ +│ 2020-11-03 22:33:19 │ 2020-11-08 01:33:19 │ +└─────────────────────┴────────────────────────────────────────────────────────────┘ +``` + **See Also** - [Interval](../../sql-reference/data-types/special-data-types/interval.md) data type -- [toInterval](../../sql-reference/functions/type-conversion-functions.md#function-tointerval) type convertion functions +- [toInterval](../../sql-reference/functions/type-conversion-functions.md#function-tointerval) type conversion functions ## Logical Negation Operator {#logical-negation-operator} diff --git a/docs/en/sql-reference/statements/alter/index.md b/docs/en/sql-reference/statements/alter/index.md index 60b7375f76d..30603122096 100644 --- a/docs/en/sql-reference/statements/alter/index.md +++ b/docs/en/sql-reference/statements/alter/index.md @@ -1,5 +1,5 @@ --- -toc_priority: 36 +toc_priority: 35 toc_title: ALTER --- diff --git a/docs/en/sql-reference/statements/alter/sample-by.md b/docs/en/sql-reference/statements/alter/sample-by.md index bea6364e4ea..df8ff90f196 100644 --- a/docs/en/sql-reference/statements/alter/sample-by.md +++ b/docs/en/sql-reference/statements/alter/sample-by.md @@ -5,16 +5,16 @@ toc_title: SAMPLE BY # Manipulating Sampling-Key Expressions {#manipulations-with-sampling-key-expressions} +Syntax: + ``` sql ALTER TABLE [db].name [ON CLUSTER cluster] MODIFY SAMPLE BY new_expression ``` The command changes the [sampling key](../../../engines/table-engines/mergetree-family/mergetree.md) of the table to `new_expression` (an expression or a tuple of expressions). -The command is lightweight in a sense that it only changes metadata. The primary key must contain the new sample key. +The command is lightweight in the sense that it only changes metadata. The primary key must contain the new sample key. !!! note "Note" - It only works for tables in the [`MergeTree`](../../../engines/table-engines/mergetree-family/mergetree.md) family (including -[replicated](../../../engines/table-engines/mergetree-family/replication.md) tables). - - + It only works for tables in the [MergeTree](../../../engines/table-engines/mergetree-family/mergetree.md) family (including +[replicated](../../../engines/table-engines/mergetree-family/replication.md) tables). \ No newline at end of file diff --git a/docs/en/sql-reference/statements/attach.md b/docs/en/sql-reference/statements/attach.md index 6260e15cfa9..035441ef5f1 100644 --- a/docs/en/sql-reference/statements/attach.md +++ b/docs/en/sql-reference/statements/attach.md @@ -1,5 +1,5 @@ --- -toc_priority: 42 +toc_priority: 40 toc_title: ATTACH --- diff --git a/docs/en/sql-reference/statements/check-table.md b/docs/en/sql-reference/statements/check-table.md index b55955eeb40..450447acaf8 100644 --- a/docs/en/sql-reference/statements/check-table.md +++ b/docs/en/sql-reference/statements/check-table.md @@ -1,5 +1,5 @@ --- -toc_priority: 43 +toc_priority: 41 toc_title: CHECK --- diff --git a/docs/en/sql-reference/statements/create/database.md b/docs/en/sql-reference/statements/create/database.md index e874672471d..bdb31d44b0b 100644 --- a/docs/en/sql-reference/statements/create/database.md +++ b/docs/en/sql-reference/statements/create/database.md @@ -1,5 +1,5 @@ --- -toc_priority: 1 +toc_priority: 35 toc_title: DATABASE --- diff --git a/docs/en/sql-reference/statements/create/dictionary.md b/docs/en/sql-reference/statements/create/dictionary.md index 5313c59396e..b1098c54703 100644 --- a/docs/en/sql-reference/statements/create/dictionary.md +++ b/docs/en/sql-reference/statements/create/dictionary.md @@ -1,5 +1,5 @@ --- -toc_priority: 4 +toc_priority: 38 toc_title: DICTIONARY --- diff --git a/docs/en/sql-reference/statements/create/index.md b/docs/en/sql-reference/statements/create/index.md index 71ace2b664c..902a4348bac 100644 --- a/docs/en/sql-reference/statements/create/index.md +++ b/docs/en/sql-reference/statements/create/index.md @@ -1,6 +1,6 @@ --- toc_folder_title: CREATE -toc_priority: 35 +toc_priority: 34 toc_title: Overview --- diff --git a/docs/en/sql-reference/statements/create/quota.md b/docs/en/sql-reference/statements/create/quota.md index 7919dc0f6db..29752050b69 100644 --- a/docs/en/sql-reference/statements/create/quota.md +++ b/docs/en/sql-reference/statements/create/quota.md @@ -1,5 +1,5 @@ --- -toc_priority: 8 +toc_priority: 42 toc_title: QUOTA --- diff --git a/docs/en/sql-reference/statements/create/role.md b/docs/en/sql-reference/statements/create/role.md index 8a682875d63..19db2b79b61 100644 --- a/docs/en/sql-reference/statements/create/role.md +++ b/docs/en/sql-reference/statements/create/role.md @@ -1,5 +1,5 @@ --- -toc_priority: 6 +toc_priority: 40 toc_title: ROLE --- diff --git a/docs/en/sql-reference/statements/create/row-policy.md b/docs/en/sql-reference/statements/create/row-policy.md index b1e7c8e9006..9f8b6b87d3b 100644 --- a/docs/en/sql-reference/statements/create/row-policy.md +++ b/docs/en/sql-reference/statements/create/row-policy.md @@ -1,5 +1,5 @@ --- -toc_priority: 7 +toc_priority: 41 toc_title: ROW POLICY --- diff --git a/docs/en/sql-reference/statements/create/settings-profile.md b/docs/en/sql-reference/statements/create/settings-profile.md index 6fcd1d4e840..196c336bb4d 100644 --- a/docs/en/sql-reference/statements/create/settings-profile.md +++ b/docs/en/sql-reference/statements/create/settings-profile.md @@ -1,5 +1,5 @@ --- -toc_priority: 9 +toc_priority: 43 toc_title: SETTINGS PROFILE --- diff --git a/docs/en/sql-reference/statements/create/table.md b/docs/en/sql-reference/statements/create/table.md index 016641b958d..82326bf51cf 100644 --- a/docs/en/sql-reference/statements/create/table.md +++ b/docs/en/sql-reference/statements/create/table.md @@ -1,5 +1,5 @@ --- -toc_priority: 2 +toc_priority: 36 toc_title: TABLE --- @@ -121,7 +121,7 @@ Defines storage time for values. Can be specified only for MergeTree-family tabl ## Column Compression Codecs {#codecs} -By default, ClickHouse applies the `lz4` compression method. For `MergeTree`-engine family you can change the default compression method in the [compression](../../../operations/server-configuration-parameters/settings.md#server-settings-compression) section of a server configuration. +By default, ClickHouse applies the `lz4` compression method. For `MergeTree`-engine family you can change the default compression method in the [compression](../../../operations/server-configuration-parameters/settings.md#server-settings-compression) section of a server configuration. You can also define the compression method for each individual column in the `CREATE TABLE` query. @@ -138,8 +138,8 @@ ENGINE = ... ``` -The `Default` codec can be specified to reference default compression which may dependend on different settings (and properties of data) in runtime. -Example: `value UInt64 CODEC(Default)` - the same as lack of codec specification. +The `Default` codec can be specified to reference default compression which may depend on different settings (and properties of data) in runtime. +Example: `value UInt64 CODEC(Default)` — the same as lack of codec specification. Also you can remove current CODEC from the column and use default compression from config.xml: @@ -149,7 +149,7 @@ ALTER TABLE codec_example MODIFY COLUMN float_value CODEC(Default); Codecs can be combined in a pipeline, for example, `CODEC(Delta, Default)`. -To select the best codec combination for you project, pass benchmarks similar to described in the Altinity [New Encodings to Improve ClickHouse Efficiency](https://www.altinity.com/blog/2019/7/new-encodings-to-improve-clickhouse) article. One thing to note is that codec can't be applied for ALIAS column type. +To select the best codec combination for you project, pass benchmarks similar to described in the Altinity [New Encodings to Improve ClickHouse Efficiency](https://www.altinity.com/blog/2019/7/new-encodings-to-improve-clickhouse) article. One thing to note is that codec can't be applied for ALIAS column type. !!! warning "Warning" You can’t decompress ClickHouse database files with external utilities like `lz4`. Instead, use the special [clickhouse-compressor](https://github.com/ClickHouse/ClickHouse/tree/master/programs/compressor) utility. diff --git a/docs/en/sql-reference/statements/create/user.md b/docs/en/sql-reference/statements/create/user.md index 98c29363102..273882fd639 100644 --- a/docs/en/sql-reference/statements/create/user.md +++ b/docs/en/sql-reference/statements/create/user.md @@ -1,5 +1,5 @@ --- -toc_priority: 5 +toc_priority: 39 toc_title: USER --- diff --git a/docs/en/sql-reference/statements/create/view.md b/docs/en/sql-reference/statements/create/view.md index dc1200e5979..4370735b8d9 100644 --- a/docs/en/sql-reference/statements/create/view.md +++ b/docs/en/sql-reference/statements/create/view.md @@ -1,5 +1,5 @@ --- -toc_priority: 3 +toc_priority: 37 toc_title: VIEW --- @@ -15,7 +15,7 @@ Syntax: CREATE [OR REPLACE] VIEW [IF NOT EXISTS] [db.]table_name [ON CLUSTER] AS SELECT ... ``` -Normal views don’t store any data, they just perform a read from another table on each access. In other words, a normal view is nothing more than a saved query. When reading from a view, this saved query is used as a subquery in the [FROM](../../../sql-reference/statements/select/from.md) clause. +Normal views don’t store any data. They just perform a read from another table on each access. In other words, a normal view is nothing more than a saved query. When reading from a view, this saved query is used as a subquery in the [FROM](../../../sql-reference/statements/select/from.md) clause. As an example, assume you’ve created a view: diff --git a/docs/en/sql-reference/statements/describe-table.md b/docs/en/sql-reference/statements/describe-table.md index b7ffec1db07..bc197bf0f72 100644 --- a/docs/en/sql-reference/statements/describe-table.md +++ b/docs/en/sql-reference/statements/describe-table.md @@ -1,5 +1,5 @@ --- -toc_priority: 44 +toc_priority: 42 toc_title: DESCRIBE --- diff --git a/docs/en/sql-reference/statements/detach.md b/docs/en/sql-reference/statements/detach.md index 0d236c132df..62a7c0cc1e0 100644 --- a/docs/en/sql-reference/statements/detach.md +++ b/docs/en/sql-reference/statements/detach.md @@ -1,5 +1,5 @@ --- -toc_priority: 45 +toc_priority: 43 toc_title: DETACH --- diff --git a/docs/en/sql-reference/statements/drop.md b/docs/en/sql-reference/statements/drop.md index de4eda64ef6..4317a20419e 100644 --- a/docs/en/sql-reference/statements/drop.md +++ b/docs/en/sql-reference/statements/drop.md @@ -1,88 +1,100 @@ --- -toc_priority: 46 +toc_priority: 44 toc_title: DROP --- # DROP Statements {#drop} -Deletes existing entity. If `IF EXISTS` clause is specified, these queries doesn’t return an error if the entity doesn’t exist. +Deletes existing entity. If the `IF EXISTS` clause is specified, these queries don’t return an error if the entity doesn’t exist. ## DROP DATABASE {#drop-database} +Deletes all tables inside the `db` database, then deletes the `db` database itself. + +Syntax: + ``` sql DROP DATABASE [IF EXISTS] db [ON CLUSTER cluster] ``` -Deletes all tables inside the `db` database, then deletes the ‘db’ database itself. - ## DROP TABLE {#drop-table} +Deletes the table. + +Syntax: + ``` sql DROP [TEMPORARY] TABLE [IF EXISTS] [db.]name [ON CLUSTER cluster] ``` -Deletes the table. - ## DROP DICTIONARY {#drop-dictionary} +Deletes the dictionary. + +Syntax: + ``` sql DROP DICTIONARY [IF EXISTS] [db.]name ``` -Deletes the dictionary. - ## DROP USER {#drop-user-statement} +Deletes a user. + +Syntax: + ``` sql DROP USER [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` -Deletes a user. - ## DROP ROLE {#drop-role-statement} +Deletes a role. The deleted role is revoked from all the entities where it was assigned. + +Syntax: + ``` sql DROP ROLE [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` -Deletes a role. - -Deleted role is revoked from all the entities where it was assigned. - ## DROP ROW POLICY {#drop-row-policy-statement} +Deletes a row policy. Deleted row policy is revoked from all the entities where it was assigned. + +Syntax: + ``` sql DROP [ROW] POLICY [IF EXISTS] name [,...] ON [database.]table [,...] [ON CLUSTER cluster_name] ``` -Deletes a row policy. - -Deleted row policy is revoked from all the entities where it was assigned. - ## DROP QUOTA {#drop-quota-statement} +Deletes a quota. The deleted quota is revoked from all the entities where it was assigned. + +Syntax: + ``` sql DROP QUOTA [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` -Deletes a quota. - -Deleted quota is revoked from all the entities where it was assigned. - ## DROP SETTINGS PROFILE {#drop-settings-profile-statement} +Deletes a settings profile. The deleted settings profile is revoked from all the entities where it was assigned. + +Syntax: + ``` sql DROP [SETTINGS] PROFILE [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` -Deletes a settings profile. - -Deleted settings profile is revoked from all the entities where it was assigned. - ## DROP VIEW {#drop-view} +Deletes a view. Views can be deleted by a `DROP TABLE` command as well but `DROP VIEW` checks that `[db.]name` is a view. + +Syntax: + ``` sql DROP VIEW [IF EXISTS] [db.]name [ON CLUSTER cluster] ``` -Deletes a view. Views can be deleted by a `DROP TABLE` command as well but `DROP VIEW` checks that `[db.]name` is a view. +[Оriginal article](https://clickhouse.tech/docs/en/sql-reference/statements/drop/) \ No newline at end of file diff --git a/docs/en/sql-reference/statements/exists.md b/docs/en/sql-reference/statements/exists.md index 2e8072125c7..3b0f4b66343 100644 --- a/docs/en/sql-reference/statements/exists.md +++ b/docs/en/sql-reference/statements/exists.md @@ -1,5 +1,5 @@ --- -toc_priority: 47 +toc_priority: 45 toc_title: EXISTS --- diff --git a/docs/en/sql-reference/statements/grant.md b/docs/en/sql-reference/statements/grant.md index c3e18ca3c75..f3829de2fbb 100644 --- a/docs/en/sql-reference/statements/grant.md +++ b/docs/en/sql-reference/statements/grant.md @@ -1,5 +1,5 @@ --- -toc_priority: 39 +toc_priority: 38 toc_title: GRANT --- diff --git a/docs/en/sql-reference/statements/insert-into.md b/docs/en/sql-reference/statements/insert-into.md index a2afbb2e10a..ae5e074fd15 100644 --- a/docs/en/sql-reference/statements/insert-into.md +++ b/docs/en/sql-reference/statements/insert-into.md @@ -1,5 +1,5 @@ --- -toc_priority: 34 +toc_priority: 33 toc_title: INSERT INTO --- @@ -13,12 +13,61 @@ Basic query format: INSERT INTO [db.]table [(c1, c2, c3)] VALUES (v11, v12, v13), (v21, v22, v23), ... ``` -The query can specify a list of columns to insert `[(c1, c2, c3)]`. In this case, the rest of the columns are filled with: +You can specify a list of columns to insert using the `(c1, c2, c3)` or `COLUMNS(c1,c2,c3)` syntax. + +Instead of listing all the required columns you can use the `(* EXCEPT(column_list))` syntax. + +For example, consider the table: + +``` sql +SHOW CREATE insert_select_testtable; +``` + +``` +┌─statement────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐ +│ CREATE TABLE insert_select_testtable +( + `a` Int8, + `b` String, + `c` Int8 +) +ENGINE = MergeTree() +ORDER BY a +SETTINGS index_granularity = 8192 │ +└──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘ +``` + +``` sql +INSERT INTO insert_select_testtable (*) VALUES (1, 'a', 1) ; +``` + +If you want to insert data in all the columns, except 'b', you need to pass so many values how many columns you chose in parenthesis then: + +``` sql +INSERT INTO insert_select_testtable (* EXCEPT(b)) Values (2, 2); +``` + +``` sql +SELECT * FROM insert_select_testtable; +``` + +``` +┌─a─┬─b─┬─c─┐ +│ 2 │ │ 2 │ +└───┴───┴───┘ +┌─a─┬─b─┬─c─┐ +│ 1 │ a │ 1 │ +└───┴───┴───┘ +``` + +In this example, we see that the second inserted row has `a` and `c` columns filled by the passed values, and `b` filled with value by default. + +If a list of columns doesn't include all existing columns, the rest of the columns are filled with: - The values calculated from the `DEFAULT` expressions specified in the table definition. - Zeros and empty strings, if `DEFAULT` expressions are not defined. -If [strict_insert_defaults=1](../../operations/settings/settings.md), columns that do not have `DEFAULT` defined must be listed in the query. +If [strict\_insert\_defaults=1](../../operations/settings/settings.md), columns that do not have `DEFAULT` defined must be listed in the query. Data can be passed to the INSERT in any [format](../../interfaces/formats.md#formats) supported by ClickHouse. The format must be specified explicitly in the query: diff --git a/docs/en/sql-reference/statements/kill.md b/docs/en/sql-reference/statements/kill.md index 1e2d2e925f7..d3f2d9bb5c6 100644 --- a/docs/en/sql-reference/statements/kill.md +++ b/docs/en/sql-reference/statements/kill.md @@ -1,5 +1,5 @@ --- -toc_priority: 48 +toc_priority: 46 toc_title: KILL --- diff --git a/docs/en/sql-reference/statements/optimize.md b/docs/en/sql-reference/statements/optimize.md index 374ef926d6c..a67f282e793 100644 --- a/docs/en/sql-reference/statements/optimize.md +++ b/docs/en/sql-reference/statements/optimize.md @@ -1,5 +1,5 @@ --- -toc_priority: 49 +toc_priority: 47 toc_title: OPTIMIZE --- diff --git a/docs/en/sql-reference/statements/rename.md b/docs/en/sql-reference/statements/rename.md index 315aaac40b5..4f14ad016a3 100644 --- a/docs/en/sql-reference/statements/rename.md +++ b/docs/en/sql-reference/statements/rename.md @@ -1,5 +1,5 @@ --- -toc_priority: 50 +toc_priority: 48 toc_title: RENAME --- diff --git a/docs/en/sql-reference/statements/revoke.md b/docs/en/sql-reference/statements/revoke.md index 8269c1f34a9..71a76546096 100644 --- a/docs/en/sql-reference/statements/revoke.md +++ b/docs/en/sql-reference/statements/revoke.md @@ -1,5 +1,5 @@ --- -toc_priority: 40 +toc_priority: 39 toc_title: REVOKE --- diff --git a/docs/en/sql-reference/statements/select/index.md b/docs/en/sql-reference/statements/select/index.md index 07be8c2bf45..3107f791eb9 100644 --- a/docs/en/sql-reference/statements/select/index.md +++ b/docs/en/sql-reference/statements/select/index.md @@ -1,7 +1,7 @@ --- title: SELECT Query toc_folder_title: SELECT -toc_priority: 33 +toc_priority: 32 toc_title: Overview --- diff --git a/docs/en/sql-reference/statements/select/with.md b/docs/en/sql-reference/statements/select/with.md index a507d5224aa..6a0564a8ede 100644 --- a/docs/en/sql-reference/statements/select/with.md +++ b/docs/en/sql-reference/statements/select/with.md @@ -4,13 +4,17 @@ toc_title: WITH # WITH Clause {#with-clause} -This section provides support for Common Table Expressions ([CTE](https://en.wikipedia.org/wiki/Hierarchical_and_recursive_queries_in_SQL)), so the results of `WITH` clause can be used in the rest of `SELECT` query. +Clickhouse supports Common Table Expressions ([CTE](https://en.wikipedia.org/wiki/Hierarchical_and_recursive_queries_in_SQL)), that is provides to use results of `WITH` clause in the rest of `SELECT` query. Named subqueries can be included to the current and child query context in places where table objects are allowed. Recursion is prevented by hiding the current level CTEs from the WITH expression. -## Limitations {#limitations} +## Syntax -1. Recursive queries are not supported. -2. When subquery is used inside WITH section, it’s result should be scalar with exactly one row. -3. Expression’s results are not available in subqueries. +``` sql +WITH AS +``` +or +``` sql +WITH AS +``` ## Examples {#examples} @@ -22,10 +26,10 @@ SELECT * FROM hits WHERE EventDate = toDate(ts_upper_bound) AND - EventTime <= ts_upper_bound + EventTime <= ts_upper_bound; ``` -**Example 2:** Evicting sum(bytes) expression result from SELECT clause column list +**Example 2:** Evicting a sum(bytes) expression result from the SELECT clause column list ``` sql WITH sum(bytes) as s @@ -34,10 +38,10 @@ SELECT table FROM system.parts GROUP BY table -ORDER BY s +ORDER BY s; ``` -**Example 3:** Using results of scalar subquery +**Example 3:** Using results of a scalar subquery ``` sql /* this example would return TOP 10 of most huge tables */ @@ -53,27 +57,14 @@ SELECT FROM system.parts GROUP BY table ORDER BY table_disk_usage DESC -LIMIT 10 +LIMIT 10; ``` -**Example 4:** Re-using expression in subquery - -As a workaround for current limitation for expression usage in subqueries, you may duplicate it. +**Example 4:** Reusing expression in a subquery ``` sql -WITH ['hello'] AS hello -SELECT - hello, - * -FROM -( - WITH ['hello'] AS hello - SELECT hello -) +WITH test1 AS (SELECT i + 1, j + 1 FROM test1) +SELECT * FROM test1; ``` -``` text -┌─hello─────┬─hello─────┐ -│ ['hello'] │ ['hello'] │ -└───────────┴───────────┘ -``` +[Original article](https://clickhouse.tech/docs/en/sql-reference/statements/select/with/) diff --git a/docs/en/sql-reference/statements/set-role.md b/docs/en/sql-reference/statements/set-role.md index 564c8ec859f..cf14a9c6d75 100644 --- a/docs/en/sql-reference/statements/set-role.md +++ b/docs/en/sql-reference/statements/set-role.md @@ -1,5 +1,5 @@ --- -toc_priority: 52 +toc_priority: 51 toc_title: SET ROLE --- diff --git a/docs/en/sql-reference/statements/set.md b/docs/en/sql-reference/statements/set.md index 4a5bbf0baf6..c6c8d28257d 100644 --- a/docs/en/sql-reference/statements/set.md +++ b/docs/en/sql-reference/statements/set.md @@ -1,5 +1,5 @@ --- -toc_priority: 51 +toc_priority: 49 toc_title: SET --- diff --git a/docs/en/sql-reference/statements/show.md b/docs/en/sql-reference/statements/show.md index a18e99d7b11..e1f2ef3488a 100644 --- a/docs/en/sql-reference/statements/show.md +++ b/docs/en/sql-reference/statements/show.md @@ -1,5 +1,5 @@ --- -toc_priority: 38 +toc_priority: 37 toc_title: SHOW --- @@ -15,12 +15,83 @@ Returns a single `String`-type ‘statement’ column, which contains a single v ## SHOW DATABASES {#show-databases} -``` sql -SHOW DATABASES [INTO OUTFILE filename] [FORMAT format] +Prints a list of all databases. + +```sql +SHOW DATABASES [LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE filename] [FORMAT format] ``` -Prints a list of all databases. -This query is identical to `SELECT name FROM system.databases [INTO OUTFILE filename] [FORMAT format]`. +This statement is identical to the query: + +```sql +SELECT name FROM system.databases [WHERE name LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE filename] [FORMAT format] +``` + +### Examples {#examples} + +Getting database names, containing the symbols sequence 'de' in their names: + +``` sql +SHOW DATABASES LIKE '%de%' +``` + +Result: + +``` text +┌─name────┐ +│ default │ +└─────────┘ +``` + +Getting database names, containing symbols sequence 'de' in their names, in the case insensitive manner: + +``` sql +SHOW DATABASES ILIKE '%DE%' +``` + +Result: + +``` text +┌─name────┐ +│ default │ +└─────────┘ +``` + +Getting database names, not containing the symbols sequence 'de' in their names: + +``` sql +SHOW DATABASES NOT LIKE '%de%' +``` + +Result: + +``` text +┌─name───────────────────────────┐ +│ _temporary_and_external_tables │ +│ system │ +│ test │ +│ tutorial │ +└────────────────────────────────┘ +``` + +Getting the first two rows from database names: + +``` sql +SHOW DATABASES LIMIT 2 +``` + +Result: + +``` text +┌─name───────────────────────────┐ +│ _temporary_and_external_tables │ +│ default │ +└────────────────────────────────┘ +``` + +### See Also {#see-also} + +- [CREATE DATABASE](https://clickhouse.tech/docs/en/sql-reference/statements/create/database/#query-language-create-database) ## SHOW PROCESSLIST {#show-processlist} @@ -42,33 +113,86 @@ $ watch -n1 "clickhouse-client --query='SHOW PROCESSLIST'" Displays a list of tables. -``` sql -SHOW [TEMPORARY] TABLES [{FROM | IN} ] [LIKE '' | WHERE expr] [LIMIT ] [INTO OUTFILE ] [FORMAT ] +```sql +SHOW [TEMPORARY] TABLES [{FROM | IN} ] [LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE ] [FORMAT ] ``` If the `FROM` clause is not specified, the query returns the list of tables from the current database. -You can get the same results as the `SHOW TABLES` query in the following way: +This statement is identical to the query: -``` sql -SELECT name FROM system.tables WHERE database = [AND name LIKE ] [LIMIT ] [INTO OUTFILE ] [FORMAT ] +```sql +SELECT name FROM system.tables [WHERE name LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE ] [FORMAT ] ``` -**Example** +### Examples {#examples} -The following query selects the first two rows from the list of tables in the `system` database, whose names contain `co`. +Getting table names, containing the symbols sequence 'user' in their names: ``` sql -SHOW TABLES FROM system LIKE '%co%' LIMIT 2 +SHOW TABLES FROM system LIKE '%user%' ``` +Result: + +``` text +┌─name─────────────┐ +│ user_directories │ +│ users │ +└──────────────────┘ +``` + +Getting table names, containing sequence 'user' in their names, in the case insensitive manner: + +``` sql +SHOW TABLES FROM system ILIKE '%USER%' +``` + +Result: + +``` text +┌─name─────────────┐ +│ user_directories │ +│ users │ +└──────────────────┘ +``` + +Getting table names, not containing the symbol sequence 's' in their names: + +``` sql +SHOW TABLES FROM system NOT LIKE '%s%' +``` + +Result: + +``` text +┌─name─────────┐ +│ metric_log │ +│ metric_log_0 │ +│ metric_log_1 │ +└──────────────┘ +``` + +Getting the first two rows from table names: + +``` sql +SHOW TABLES FROM system LIMIT 2 +``` + +Result: + ``` text ┌─name───────────────────────────┐ │ aggregate_function_combinators │ -│ collations │ +│ asynchronous_metric_log │ └────────────────────────────────┘ ``` +### See Also {#see-also} + +- [Create Tables](https://clickhouse.tech/docs/en/getting-started/tutorial/#create-tables) +- [SHOW CREATE TABLE](https://clickhouse.tech/docs/en/sql-reference/statements/show/#show-create-table) + ## SHOW DICTIONARIES {#show-dictionaries} Displays a list of [external dictionaries](../../sql-reference/dictionaries/external-dictionaries/external-dicts.md). diff --git a/docs/en/sql-reference/statements/system.md b/docs/en/sql-reference/statements/system.md index 8230177f249..75303fde19e 100644 --- a/docs/en/sql-reference/statements/system.md +++ b/docs/en/sql-reference/statements/system.md @@ -1,5 +1,5 @@ --- -toc_priority: 37 +toc_priority: 36 toc_title: SYSTEM --- diff --git a/docs/en/sql-reference/statements/truncate.md b/docs/en/sql-reference/statements/truncate.md index a8cf802616f..a13936c88ab 100644 --- a/docs/en/sql-reference/statements/truncate.md +++ b/docs/en/sql-reference/statements/truncate.md @@ -1,5 +1,5 @@ --- -toc_priority: 53 +toc_priority: 52 toc_title: TRUNCATE --- diff --git a/docs/en/sql-reference/statements/use.md b/docs/en/sql-reference/statements/use.md index 2932542f052..841c23d333d 100644 --- a/docs/en/sql-reference/statements/use.md +++ b/docs/en/sql-reference/statements/use.md @@ -1,5 +1,5 @@ --- -toc_priority: 54 +toc_priority: 53 toc_title: USE --- diff --git a/docs/en/sql-reference/syntax.md b/docs/en/sql-reference/syntax.md index fecea3ab30b..70994f3d882 100644 --- a/docs/en/sql-reference/syntax.md +++ b/docs/en/sql-reference/syntax.md @@ -87,7 +87,7 @@ In string literals, you need to escape at least `'` and `\`. Single quotes can b ### Compound {#compound} -Arrays are constructed with square brackets `[1, 2, 3]`. Nuples are constructed with round brackets `(1, 'Hello, world!', 2)`. +Arrays are constructed with square brackets `[1, 2, 3]`. Tuples are constructed with round brackets `(1, 'Hello, world!', 2)`. Technically these are not literals, but expressions with the array creation operator and the tuple creation operator, respectively. An array must consist of at least one item, and a tuple must have at least two items. There’s a separate case when tuples appear in the `IN` clause of a `SELECT` query. Query results can include tuples, but tuples can’t be saved to a database (except of tables with [Memory](../engines/table-engines/special/memory.md) engine). diff --git a/docs/en/sql-reference/table-functions/null.md b/docs/en/sql-reference/table-functions/null.md new file mode 100644 index 00000000000..6edec61add7 --- /dev/null +++ b/docs/en/sql-reference/table-functions/null.md @@ -0,0 +1,41 @@ +--- +toc_priority: 53 +toc_title: null function +--- + +# null {#null-function} + +Accepts an inserted data of the specified structure and immediately drops it away. The function is used for convenience writing tests and demonstrations. + +**Syntax** + +``` sql +null('structure') +``` + +**Parameter** + +- `structure` — A list of columns and column types. [String](../../sql-reference/data-types/string.md). + +**Returned value** + +A table with the specified structure, which is dropped right after the query execution. + +**Example** + +Query with the `null` function: + +``` sql +INSERT INTO function null('x UInt64') SELECT * FROM numbers_mt(1000000000); +``` +can replace three queries: + +```sql +CREATE TABLE t (x UInt64) ENGINE = Null; +INSERT INTO t SELECT * FROM numbers_mt(1000000000); +DROP TABLE IF EXISTS t; +``` + +See also: format **Null**. + +[Original article](https://clickhouse.tech/docs/en/sql-reference/table-functions/null/) diff --git a/docs/en/sql-reference/table-functions/view.md b/docs/en/sql-reference/table-functions/view.md new file mode 100644 index 00000000000..9997971af65 --- /dev/null +++ b/docs/en/sql-reference/table-functions/view.md @@ -0,0 +1,67 @@ +--- +toc_priority: 51 +toc_title: view +--- + +## view {#view} + +Turns a subquery into a table. The function implements views (see [CREATE VIEW](https://clickhouse.tech/docs/en/sql-reference/statements/create/view/#create-view)). The resulting table doesn't store data, but only stores the specified `SELECT` query. When reading from the table, ClickHouse executes the query and deletes all unnecessary columns from the result. + +**Syntax** + +``` sql +view(subquery) +``` + +**Parameters** + +- `subquery` — `SELECT` query. + +**Returned value** + +- A table. + +**Example** + +Input table: + +``` text +┌─id─┬─name─────┬─days─┐ +│ 1 │ January │ 31 │ +│ 2 │ February │ 29 │ +│ 3 │ March │ 31 │ +│ 4 │ April │ 30 │ +└────┴──────────┴──────┘ +``` + +Query: + +``` sql +SELECT * FROM view(SELECT name FROM months) +``` + +Result: + +``` text +┌─name─────┐ +│ January │ +│ February │ +│ March │ +│ April │ +└──────────┘ +``` + +You can use the `view` function as a parameter of the [remote](https://clickhouse.tech/docs/en/sql-reference/table-functions/remote/#remote-remotesecure) and [cluster](https://clickhouse.tech/docs/en/sql-reference/table-functions/cluster/#cluster-clusterallreplicas) table functions: + +``` sql +SELECT * FROM remote(`127.0.0.1`, view(SELECT a, b, c FROM table_name)) +``` + +``` sql +SELECT * FROM cluster(`cluster_name`, view(SELECT a, b, c FROM table_name)) +``` + +**See Also** + +- [View Table Engine](https://clickhouse.tech/docs/en/engines/table-engines/special/view/) +[Original article](https://clickhouse.tech/docs/en/query_language/table_functions/view/) \ No newline at end of file diff --git a/docs/ru/commercial/cloud.md b/docs/ru/commercial/cloud.md index 6cbc0e6743b..e6b0309c456 100644 --- a/docs/ru/commercial/cloud.md +++ b/docs/ru/commercial/cloud.md @@ -1,3 +1,8 @@ +--- +toc_priority: 1 +toc_title: "\u041f\u043e\u0441\u0442\u0430\u0432\u0449\u0438\u043a\u0438\u0020\u043e\u0431\u043b\u0430\u0447\u043d\u044b\u0445\u0020\u0443\u0441\u043b\u0443\u0433\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + # Поставщики облачных услуг ClickHouse {#clickhouse-cloud-service-providers} !!! info "Инфо" diff --git a/docs/ru/development/architecture.md b/docs/ru/development/architecture.md index e66b20f764a..53c007e078f 100644 --- a/docs/ru/development/architecture.md +++ b/docs/ru/development/architecture.md @@ -1,3 +1,8 @@ +--- +toc_priority: 62 +toc_title: "\u041e\u0431\u0437\u043e\u0440\u0020\u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u044b\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + # Обзор архитектуры ClickHouse {#overview-of-clickhouse-architecture} ClickHouse - полноценная колоночная СУБД. Данные хранятся в колонках, а в процессе обработки - в массивах (векторах или фрагментах (chunk’ах) колонок). По возможности операции выполняются на массивах, а не на индивидуальных значениях. Это называется “векторизованное выполнения запросов” (vectorized query execution), и помогает снизить стоимость фактической обработки данных. @@ -185,7 +190,7 @@ ClickHouse имеет сильную типизацию, поэтому нет Репликация использует асинхронную multi-master схему. Вы можете вставить данные в любую реплику, которая имеет открытую сессию в `ZooKeeper`, и данные реплицируются на все другие реплики асинхронно. Поскольку ClickHouse не поддерживает UPDATE, репликация исключает конфликты (conflict-free replication). Поскольку подтверждение вставок кворумом не реализовано, только что вставленные данные могут быть потеряны в случае сбоя одного узла. -Метаданные для репликации хранятся в `ZooKeeper`. Существует журнал репликации, в котором перечислены действия, которые необходимо выполнить. Среди этих действий: получить часть (get the part); объединить части (merge parts); удалить партицию (drop a partition) и так далее. Каждая реплика копирует журнал репликации в свою очередь, а затем выполняет действия из очереди. Например, при вставке в журнале создается действие «получить часть» (get the part), и каждая реплика загружает эту часть. Слияния координируются между репликами, чтобы получить идентичные до байта результаты. Все части объединяются одинаково на всех репликах. Это достигается путем выбора одной реплики в качестве лидера, и эта реплика инициирует слияния и записывает действия «слияния частей» в журнал. +Метаданные для репликации хранятся в `ZooKeeper`. Существует журнал репликации, в котором перечислены действия, которые необходимо выполнить. Среди этих действий: получить часть (get the part); объединить части (merge parts); удалить партицию (drop a partition) и так далее. Каждая реплика копирует журнал репликации в свою очередь, а затем выполняет действия из очереди. Например, при вставке в журнале создается действие «получить часть» (get the part), и каждая реплика загружает эту часть. Слияния координируются между репликами, чтобы получить идентичные до байта результаты. Все части объединяются одинаково на всех репликах. Одна из реплик-лидеров инициирует новое слияние кусков первой и записывает действия «слияния частей» в журнал. Несколько реплик (или все) могут быть лидерами одновременно. Реплике можно запретить быть лидером с помощью `merge_tree` настройки `replicated_can_become_leader`. Репликация является физической: между узлами передаются только сжатые части, а не запросы. Слияния обрабатываются на каждой реплике независимо, в большинстве случаев, чтобы снизить затраты на сеть, во избежание усиления роли сети. Крупные объединенные части отправляются по сети только в случае значительной задержки репликации. diff --git a/docs/ru/development/browse-code.md b/docs/ru/development/browse-code.md index 814b213a6a7..ac17cf0e6f5 100644 --- a/docs/ru/development/browse-code.md +++ b/docs/ru/development/browse-code.md @@ -1,3 +1,9 @@ +--- +toc_priority: 71 +toc_title: "\u041d\u0430\u0432\u0438\u0433\u0430\u0446\u0438\u044f\u0020\u043f\u043e\u0020\u043a\u043e\u0434\u0443\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + + # Навигация по коду ClickHouse {#navigatsiia-po-kodu-clickhouse} Для навигации по коду онлайн доступен **Woboq**, он расположен [здесь](https://clickhouse.tech/codebrowser/html_report///ClickHouse/src/index.html). В нём реализовано удобное перемещение между исходными файлами, семантическая подсветка, подсказки, индексация и поиск. Слепок кода обновляется ежедневно. diff --git a/docs/ru/development/contrib.md b/docs/ru/development/contrib.md index c640ecee79d..e65ab4819e8 100644 --- a/docs/ru/development/contrib.md +++ b/docs/ru/development/contrib.md @@ -1,3 +1,9 @@ +--- +toc_priority: 70 +toc_title: "\u0418\u0441\u043f\u043e\u043b\u044c\u0437\u0443\u0435\u043c\u044b\u0435\u0020\u0441\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0435\u0020\u0431\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0438" +--- + + # Используемые сторонние библиотеки {#ispolzuemye-storonnie-biblioteki} | Библиотека | Лицензия | diff --git a/docs/ru/development/developer-instruction.md b/docs/ru/development/developer-instruction.md index 9edcd4cf9f9..b0c84d9de7c 100644 --- a/docs/ru/development/developer-instruction.md +++ b/docs/ru/development/developer-instruction.md @@ -1,3 +1,8 @@ +--- +toc_priority: 61 +toc_title: "\u0418\u043d\u0441\u0442\u0440\u0443\u043a\u0446\u0438\u044f\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432" +--- + # Инструкция для разработчиков Сборка ClickHouse поддерживается на Linux, FreeBSD, Mac OS X. diff --git a/docs/ru/development/style.md b/docs/ru/development/style.md index 951537e8113..671293a7bbd 100644 --- a/docs/ru/development/style.md +++ b/docs/ru/development/style.md @@ -1,3 +1,9 @@ +--- +toc_priority: 68 +toc_title: "\u041a\u0430\u043a\u0020\u043f\u0438\u0441\u0430\u0442\u044c\u0020\u043a\u043e\u0434\u0020\u043d\u0430\u0020\u0043\u002b\u002b" +--- + + # Как писать код на C++ {#kak-pisat-kod-na-c} ## Общее {#obshchee} diff --git a/docs/ru/engines/database-engines/index.md b/docs/ru/engines/database-engines/index.md index 79be85518e2..d3dd729e302 100644 --- a/docs/ru/engines/database-engines/index.md +++ b/docs/ru/engines/database-engines/index.md @@ -1,3 +1,10 @@ +--- +toc_folder_title: "\u0414\u0432\u0438\u0436\u043a\u0438\u0020\u0431\u0430\u0437\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +toc_priority: 27 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" +--- + + # Движки баз данных {#dvizhki-baz-dannykh} Движки баз данных обеспечивают работу с таблицами. diff --git a/docs/ru/engines/database-engines/lazy.md b/docs/ru/engines/database-engines/lazy.md index c1538ec91bc..c01aae0284e 100644 --- a/docs/ru/engines/database-engines/lazy.md +++ b/docs/ru/engines/database-engines/lazy.md @@ -1,3 +1,8 @@ +--- +toc_priority: 31 +toc_title: Lazy +--- + # Lazy {#lazy} Сохраняет таблицы только в оперативной памяти `expiration_time_in_seconds` через несколько секунд после последнего доступа. Может использоваться только с таблицами \*Log. diff --git a/docs/ru/engines/database-engines/mysql.md b/docs/ru/engines/database-engines/mysql.md index e293cb6ec77..2f8335de73a 100644 --- a/docs/ru/engines/database-engines/mysql.md +++ b/docs/ru/engines/database-engines/mysql.md @@ -1,3 +1,8 @@ +--- +toc_priority: 30 +toc_title: MySQL +--- + # MySQL {#mysql} Позволяет подключаться к базам данных на удалённом MySQL сервере и выполнять запросы `INSERT` и `SELECT` для обмена данными между ClickHouse и MySQL. diff --git a/docs/ru/engines/index.md b/docs/ru/engines/index.md index 333e65dcb7b..28ccc8bcfe6 100644 --- a/docs/ru/engines/index.md +++ b/docs/ru/engines/index.md @@ -1,6 +1,6 @@ --- -toc_folder_title: "\u0414\u0432\u0438\u0436\u043A\u0438" +toc_folder_title: "\u0045\u006e\u0067\u0069\u006e\u0065\u0073" +toc_hidden: true toc_priority: 25 +toc_title: hidden --- - - diff --git a/docs/ru/engines/table-engines/index.md b/docs/ru/engines/table-engines/index.md index 6a954313c60..740588c50a4 100644 --- a/docs/ru/engines/table-engines/index.md +++ b/docs/ru/engines/table-engines/index.md @@ -1,3 +1,10 @@ +--- +toc_folder_title: "\u0414\u0432\u0438\u0436\u043a\u0438\u0020\u0442\u0430\u0431\u043b\u0438\u0446" +toc_priority: 26 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" +--- + + # Движки таблиц {#table_engines} Движок таблицы (тип таблицы) определяет: diff --git a/docs/ru/engines/table-engines/integrations/hdfs.md b/docs/ru/engines/table-engines/integrations/hdfs.md index b5234b38d7d..bd8e760fce4 100644 --- a/docs/ru/engines/table-engines/integrations/hdfs.md +++ b/docs/ru/engines/table-engines/integrations/hdfs.md @@ -1,3 +1,8 @@ +--- +toc_priority: 4 +toc_title: HDFS +--- + # HDFS {#table_engines-hdfs} Управляет данными в HDFS. Данный движок похож на движки [File](../special/file.md#table_engines-file) и [URL](../special/url.md#table_engines-url). diff --git a/docs/ru/engines/table-engines/integrations/index.md b/docs/ru/engines/table-engines/integrations/index.md index 0186f6cf19e..02189cf9e55 100644 --- a/docs/ru/engines/table-engines/integrations/index.md +++ b/docs/ru/engines/table-engines/integrations/index.md @@ -1,5 +1,5 @@ --- -toc_folder_title: Integrations +toc_folder_title: "\u0414\u0432\u0438\u0436\u043a\u0438\u0020\u0442\u0430\u0431\u043b\u0438\u0446\u0020\u0434\u043b\u044f\u0020\u0438\u043d\u0442\u0435\u0433\u0440\u0430\u0446\u0438\u0438" toc_priority: 30 --- diff --git a/docs/ru/engines/table-engines/integrations/jdbc.md b/docs/ru/engines/table-engines/integrations/jdbc.md index cded0c51fa0..d7d438e0633 100644 --- a/docs/ru/engines/table-engines/integrations/jdbc.md +++ b/docs/ru/engines/table-engines/integrations/jdbc.md @@ -1,3 +1,8 @@ +--- +toc_priority: 2 +toc_title: JDBC +--- + # JDBC {#table-engine-jdbc} Позволяет ClickHouse подключаться к внешним базам данных с помощью [JDBC](https://en.wikipedia.org/wiki/Java_Database_Connectivity). diff --git a/docs/ru/engines/table-engines/integrations/kafka.md b/docs/ru/engines/table-engines/integrations/kafka.md index 8c47e97f497..940fee2452b 100644 --- a/docs/ru/engines/table-engines/integrations/kafka.md +++ b/docs/ru/engines/table-engines/integrations/kafka.md @@ -1,3 +1,8 @@ +--- +toc_priority: 5 +toc_title: Kafka +--- + # Kafka {#kafka} Движок работает с [Apache Kafka](http://kafka.apache.org/). @@ -154,6 +159,22 @@ Kafka(kafka_broker_list, kafka_topic_list, kafka_group_name, kafka_format В документе [librdkafka configuration reference](https://github.com/edenhill/librdkafka/blob/master/CONFIGURATION.md) можно увидеть список возможных опций конфигурации. Используйте подчеркивание (`_`) вместо точки в конфигурации ClickHouse. Например, `check.crcs=true` будет соответствовать `true`. +### Поддержка Kerberos {#kafka-kerberos-support} + +Чтобы начать работу с Kafka с поддержкой Kerberos, добавьте дочерний элемент `security_protocol` со значением `sasl_plaintext`. Этого будет достаточно, если получен тикет на получение тикета (ticket-granting ticket) Kerberos и он кэшируется средствами ОС. +ClickHouse может поддерживать учетные данные Kerberos с помощью файла keytab. Рассмотрим дочерние элементы `sasl_kerberos_service_name`, `sasl_kerberos_keytab`, `sasl_kerberos_principal` и `sasl.kerberos.kinit.cmd`. + +Пример: + +``` xml + + + SASL_PLAINTEXT + /home/kafkauser/kafkauser.keytab + kafkauser/kafkahost@EXAMPLE.COM + +``` + ## Виртуальные столбцы {#virtualnye-stolbtsy} - `_topic` — топик Kafka. diff --git a/docs/ru/engines/table-engines/integrations/mysql.md b/docs/ru/engines/table-engines/integrations/mysql.md index 9891fe79959..3370e9b06d0 100644 --- a/docs/ru/engines/table-engines/integrations/mysql.md +++ b/docs/ru/engines/table-engines/integrations/mysql.md @@ -1,3 +1,8 @@ +--- +toc_priority: 3 +toc_title: MySQL +--- + # MySQL {#mysql} Движок MySQL позволяет выполнять запросы `SELECT` над данными, хранящимися на удалённом MySQL сервере. diff --git a/docs/ru/engines/table-engines/integrations/odbc.md b/docs/ru/engines/table-engines/integrations/odbc.md index aecdbbf038d..97317d647c8 100644 --- a/docs/ru/engines/table-engines/integrations/odbc.md +++ b/docs/ru/engines/table-engines/integrations/odbc.md @@ -1,3 +1,8 @@ +--- +toc_priority: 1 +toc_title: ODBC +--- + # ODBC {#table-engine-odbc} Позволяет ClickHouse подключаться к внешним базам данных с помощью [ODBC](https://en.wikipedia.org/wiki/Open_Database_Connectivity). diff --git a/docs/ru/engines/table-engines/integrations/rabbitmq.md b/docs/ru/engines/table-engines/integrations/rabbitmq.md index ef7b811e295..dedb5842d68 100644 --- a/docs/ru/engines/table-engines/integrations/rabbitmq.md +++ b/docs/ru/engines/table-engines/integrations/rabbitmq.md @@ -45,7 +45,7 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] - `rabbitmq_row_delimiter` – символ-разделитель, который завершает сообщение. - `rabbitmq_schema` – опциональный параметр, необходимый, если используется формат, требующий определения схемы. Например, [Cap’n Proto](https://capnproto.org/) требует путь к файлу со схемой и название корневого объекта `schema.capnp:Message`. - `rabbitmq_num_consumers` – количество потребителей на таблицу. По умолчанию: `1`. Укажите больше потребителей, если пропускная способность одного потребителя недостаточна. -- `rabbitmq_num_queues` – количество очередей на потребителя. По умолчанию: `1`. Укажите больше потребителей, если пропускная способность одной очереди на потребителя недостаточна. +- `rabbitmq_num_queues` – количество очередей. По умолчанию: `1`. Большее число очередей может сильно увеличить пропускную способность. - `rabbitmq_queue_base` - настройка для имен очередей. Сценарии использования описаны ниже. - `rabbitmq_persistent` - флаг, от которого зависит настройка 'durable' для сообщений при запросах `INSERT`. По умолчанию: `0`. - `rabbitmq_skip_broken_messages` – максимальное количество некорректных сообщений в блоке. Если `rabbitmq_skip_broken_messages = N`, то движок отбрасывает `N` сообщений, которые не получилось обработать. Одно сообщение в точности соответствует одной записи (строке). Значение по умолчанию – 0. @@ -140,4 +140,5 @@ Example: - `_channel_id` - идентификатор канала `ChannelID`, на котором было получено сообщение. - `_delivery_tag` - значение `DeliveryTag` полученного сообщения. Уникально в рамках одного канала. - `_redelivered` - флаг `redelivered`. (Не равно нулю, если есть возможность, что сообщение было получено более, чем одним каналом.) -- `_message_id` - значение `MessageID` полученного сообщения. Данное поле непусто, если указано в параметрах при отправке сообщения. +- `_message_id` - значение поля `messageID` полученного сообщения. Данное поле непусто, если указано в параметрах при отправке сообщения. +- `_timestamp` - значение поля `timestamp` полученного сообщения. Данное поле непусто, если указано в параметрах при отправке сообщения. diff --git a/docs/ru/engines/table-engines/log-family/index.md b/docs/ru/engines/table-engines/log-family/index.md index d2cb0df5a8a..7c6d2f81d7c 100644 --- a/docs/ru/engines/table-engines/log-family/index.md +++ b/docs/ru/engines/table-engines/log-family/index.md @@ -1,6 +1,6 @@ --- -toc_folder_title: Семейство Log -toc_title: Введение +toc_folder_title: "\u0421\u0435\u043c\u0435\u0439\u0441\u0442\u0432\u043e\u0020\u004c\u006f\u0067" +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" toc_priority: 29 --- diff --git a/docs/ru/engines/table-engines/log-family/log.md b/docs/ru/engines/table-engines/log-family/log.md index 826738d6a3d..fad331454c7 100644 --- a/docs/ru/engines/table-engines/log-family/log.md +++ b/docs/ru/engines/table-engines/log-family/log.md @@ -1,3 +1,8 @@ +--- +toc_priority: 33 +toc_title: Log +--- + # Log {#log} Движок относится к семейству движков Log. Смотрите общие свойства и различия движков в статье [Семейство Log](index.md). diff --git a/docs/ru/engines/table-engines/log-family/stripelog.md b/docs/ru/engines/table-engines/log-family/stripelog.md index 9523a354ee7..e505aae4c52 100644 --- a/docs/ru/engines/table-engines/log-family/stripelog.md +++ b/docs/ru/engines/table-engines/log-family/stripelog.md @@ -1,3 +1,8 @@ +--- +toc_priority: 32 +toc_title: StripeLog +--- + # StripeLog {#stripelog} Движок относится к семейству движков Log. Смотрите общие свойства и различия движков в статье [Семейство Log](index.md). diff --git a/docs/ru/engines/table-engines/log-family/tinylog.md b/docs/ru/engines/table-engines/log-family/tinylog.md index 69bc88b133e..d5c24d41ca4 100644 --- a/docs/ru/engines/table-engines/log-family/tinylog.md +++ b/docs/ru/engines/table-engines/log-family/tinylog.md @@ -1,3 +1,8 @@ +--- +toc_priority: 34 +toc_title: TinyLog +--- + # TinyLog {#tinylog} Движок относится к семейству движков Log. Смотрите общие свойства и различия движков в статье [Семейство Log](index.md). diff --git a/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md index 3e18f890af7..99b4ec06765 100644 --- a/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 35 +toc_title: AggregatingMergeTree +--- + # AggregatingMergeTree {#aggregatingmergetree} Движок наследует функциональность [MergeTree](mergetree.md#table_engines-mergetree), изменяя логику слияния кусков данных. Все строки с одинаковым первичным ключом (точнее, с одинаковым [ключом сортировки](mergetree.md)) ClickHouse заменяет на одну (в пределах одного куска данных), которая хранит объединение состояний агрегатных функций. diff --git a/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md index f7bc14cc647..8ea3a5a7c92 100644 --- a/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 36 +toc_title: CollapsingMergeTree +--- + # CollapsingMergeTree {#table_engine-collapsingmergetree} Движок наследует функциональность от [MergeTree](mergetree.md) и добавляет в алгоритм слияния кусков данных логику сворачивания (удаления) строк. diff --git a/docs/ru/engines/table-engines/mergetree-family/custom-partitioning-key.md b/docs/ru/engines/table-engines/mergetree-family/custom-partitioning-key.md index 32b5eee7fab..2d26528d964 100644 --- a/docs/ru/engines/table-engines/mergetree-family/custom-partitioning-key.md +++ b/docs/ru/engines/table-engines/mergetree-family/custom-partitioning-key.md @@ -1,3 +1,9 @@ +--- +toc_priority: 32 +toc_title: "\u041f\u0440\u043e\u0438\u0437\u0432\u043e\u043b\u044c\u043d\u044b\u0439\u0020\u043a\u043b\u044e\u0447\u0020\u043f\u0430\u0440\u0442\u0438\u0446\u0438\u043e\u043d\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f" +--- + + # Произвольный ключ партиционирования {#proizvolnyi-kliuch-partitsionirovaniia} Партиционирование данных доступно для таблиц семейства [MergeTree](mergetree.md) (включая [реплицированные таблицы](replication.md)). Таблицы [MaterializedView](../special/materializedview.md#materializedview), созданные на основе таблиц MergeTree, также поддерживают партиционирование. diff --git a/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md b/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md index 861a8b6e633..e47c9127711 100644 --- a/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: GraphiteMergeTree +--- + # GraphiteMergeTree {#graphitemergetree} Движок предназначен для прореживания и агрегирования/усреднения (rollup) данных [Graphite](http://graphite.readthedocs.io/en/latest/index.html). Он может быть интересен разработчикам, которые хотят использовать ClickHouse как хранилище данных для Graphite. diff --git a/docs/ru/engines/table-engines/mergetree-family/index.md b/docs/ru/engines/table-engines/mergetree-family/index.md index 9e989d807da..abdfdd77d7f 100644 --- a/docs/ru/engines/table-engines/mergetree-family/index.md +++ b/docs/ru/engines/table-engines/mergetree-family/index.md @@ -1,6 +1,5 @@ --- toc_folder_title: MergeTree Family toc_priority: 28 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" --- - - diff --git a/docs/ru/engines/table-engines/mergetree-family/mergetree.md b/docs/ru/engines/table-engines/mergetree-family/mergetree.md index ef1e236a95c..bb076d480cd 100644 --- a/docs/ru/engines/table-engines/mergetree-family/mergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/mergetree.md @@ -79,6 +79,7 @@ ORDER BY expr - `index_granularity` — максимальное количество строк данных между засечками индекса. По умолчанию — 8192. Смотрите [Хранение данных](#mergetree-data-storage). - `index_granularity_bytes` — максимальный размер гранул данных в байтах. По умолчанию — 10Mb. Чтобы ограничить размер гранул только количеством строк, установите значение 0 (не рекомендовано). Смотрите [Хранение данных](#mergetree-data-storage). + - `min_index_granularity_bytes` — минимально допустимый размер гранул данных в байтах. Значение по умолчанию — 1024b. Для обеспечения защиты от случайного создания таблиц с очень низким значением `index_granularity_bytes`. Смотрите [Хранение данных](#mergetree-data-storage). - `enable_mixed_granularity_parts` — включает или выключает переход к ограничению размера гранул с помощью настройки `index_granularity_bytes`. Настройка `index_granularity_bytes` улучшает производительность ClickHouse при выборке данных из таблиц с большими (десятки и сотни мегабайтов) строками. Если у вас есть таблицы с большими строками, можно включить эту настройку, чтобы повысить эффективность запросов `SELECT`. - `use_minimalistic_part_header_in_zookeeper` — Способ хранения заголовков кусков данных в ZooKeeper. Если `use_minimalistic_part_header_in_zookeeper = 1`, то ZooKeeper хранит меньше данных. Подробнее читайте в [описании настройки](../../../operations/server-configuration-parameters/settings.md#server-settings-use_minimalistic_part_header_in_zookeeper) в разделе "Конфигурационные параметры сервера". - `min_merge_bytes_to_use_direct_io` — минимальный объём данных при слиянии, необходимый для прямого (небуферизованного) чтения/записи (direct I/O) на диск. При слиянии частей данных ClickHouse вычисляет общий объём хранения всех данных, подлежащих слиянию. Если общий объём хранения всех данных для чтения превышает `min_bytes_to_use_direct_io` байт, тогда ClickHouse использует флаг `O_DIRECT` при чтении данных с диска. Если `min_merge_bytes_to_use_direct_io = 0`, тогда прямой ввод-вывод отключен. Значение по умолчанию: `10 * 1024 * 1024 * 1024` байтов. diff --git a/docs/ru/engines/table-engines/mergetree-family/replacingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/replacingmergetree.md index 4aa1eb556f3..1228371e8ea 100644 --- a/docs/ru/engines/table-engines/mergetree-family/replacingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/replacingmergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 33 +toc_title: ReplacingMergeTree +--- + # ReplacingMergeTree {#replacingmergetree} Движок отличается от [MergeTree](mergetree.md#table_engines-mergetree) тем, что выполняет удаление дублирующихся записей с одинаковым значением [ключа сортировки](mergetree.md)). diff --git a/docs/ru/engines/table-engines/mergetree-family/replication.md b/docs/ru/engines/table-engines/mergetree-family/replication.md index 73291020aa3..6d3930e33ce 100644 --- a/docs/ru/engines/table-engines/mergetree-family/replication.md +++ b/docs/ru/engines/table-engines/mergetree-family/replication.md @@ -1,3 +1,8 @@ +--- +toc_priority: 31 +toc_title: "\u0420\u0435\u043f\u043b\u0438\u043a\u0430\u0446\u0438\u044f\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +--- + # Репликация данных {#table_engines-replication} Репликация поддерживается только для таблиц семейства MergeTree: @@ -14,7 +19,7 @@ Репликация не зависит от шардирования. На каждом шарде репликация работает независимо. -Реплицируются сжатые данные запросов `INSERT`, `ALTER` (см. подробности в описании запроса [ALTER](../../../engines/table-engines/mergetree-family/replication.md#query_language_queries_alter)). +Реплицируются сжатые данные запросов `INSERT`, `ALTER` (см. подробности в описании запроса [ALTER](../../../sql-reference/statements/alter/index.md#query_language_queries_alter)). Запросы `CREATE`, `DROP`, `ATTACH`, `DETACH` и `RENAME` выполняются на одном сервере и не реплицируются: @@ -113,7 +118,9 @@ CREATE TABLE table_name -Как видно в примере, эти параметры могут содержать подстановки в фигурных скобках. Подставляемые значения достаются из конфигурационного файла, из секции `macros`. Пример: +Как видно в примере, эти параметры могут содержать подстановки в фигурных скобках. Подставляемые значения достаются из конфигурационного файла, из секции «[macros](../../../operations/server-configuration-parameters/settings/#macros)». + +Пример: ``` xml @@ -133,12 +140,40 @@ CREATE TABLE table_name `table_name` - имя узла для таблицы в ZooKeeper. Разумно делать его таким же, как имя таблицы. Оно указывается явно, так как, в отличие от имени таблицы, оно не меняется после запроса RENAME. *Подсказка*: можно также указать имя базы данных перед `table_name`, например `db_name.table_name` +Можно использовать две встроенных подстановки `{database}` и `{table}`, они раскрываются в имя таблицы и в имя базы данных соответственно (если эти подстановки не переопределены в секции `macros`). Т.о. Zookeeper путь можно задать как `'/clickhouse/tables/{layer}-{shard}/{database}/{table}'`. +Будьте осторожны с переименованиями таблицы при использовании этих автоматических подстановок. Путь в Zookeeper-е нельзя изменить, а подстановка при переименовании таблицы раскроется в другой путь, таблица будет обращаться к несуществующему в Zookeeper-е пути и перейдет в режим только для чтения. + Имя реплики — то, что идентифицирует разные реплики одной и той же таблицы. Можно использовать для него имя сервера, как показано в примере. Впрочем, достаточно, чтобы имя было уникально лишь в пределах каждого шарда. Можно не использовать подстановки, а указать соответствующие параметры явно. Это может быть удобным для тестирования и при настройке маленьких кластеров. Однако в этом случае нельзя пользоваться распределенными DDL-запросами (`ON CLUSTER`). При работе с большими кластерами мы рекомендуем использовать подстановки, они уменьшают вероятность ошибки. +Можно указать аргументы по умолчанию для движка реплицируемых таблиц в файле конфигурации сервера. + +```xml +/clickhouse/tables/{shard}/{database}/{table} +{replica} +``` + +В этом случае можно опустить аргументы при создании таблиц: + +``` sql +CREATE TABLE table_name ( + x UInt32 +) ENGINE = ReplicatedMergeTree +ORDER BY x; +``` + +Это будет эквивалентно следующему запросу: + +``` sql +CREATE TABLE table_name ( + x UInt32 +) ENGINE = ReplicatedMergeTree('/clickhouse/tables/{shard}/{database}/table_name', '{replica}') +ORDER BY x; +``` + Выполните запрос `CREATE TABLE` на каждой реплике. Запрос создаёт новую реплицируемую таблицу, или добавляет новую реплику к имеющимся. Если вы добавляете новую реплику после того, как таблица на других репликах уже содержит некоторые данные, то после выполнения запроса, данные на новую реплику будут скачаны с других реплик. То есть, новая реплика синхронизирует себя с остальными. diff --git a/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md index 070cd109c30..7b9c11adc2e 100644 --- a/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 34 +toc_title: SummingMergeTree +--- + # SummingMergeTree {#summingmergetree} Движок наследует функциональность [MergeTree](mergetree.md#table_engines-mergetree). Отличие заключается в том, что для таблиц `SummingMergeTree` при слиянии кусков данных ClickHouse все строки с одинаковым первичным ключом (точнее, с одинаковым [ключом сортировки](mergetree.md)) заменяет на одну, которая хранит только суммы значений из столбцов с цифровым типом данных. Если ключ сортировки подобран таким образом, что одному значению ключа соответствует много строк, это значительно уменьшает объём хранения и ускоряет последующую выборку данных. diff --git a/docs/ru/engines/table-engines/mergetree-family/versionedcollapsingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/versionedcollapsingmergetree.md index bf280eb52bc..2adb8cc0d77 100644 --- a/docs/ru/engines/table-engines/mergetree-family/versionedcollapsingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/versionedcollapsingmergetree.md @@ -1,3 +1,8 @@ +--- +toc_priority: 37 +toc_title: VersionedCollapsingMergeTree +--- + # VersionedCollapsingMergeTree {#versionedcollapsingmergetree} Движок: diff --git a/docs/ru/engines/table-engines/special/buffer.md b/docs/ru/engines/table-engines/special/buffer.md index 72d2822af98..986fe9adbb9 100644 --- a/docs/ru/engines/table-engines/special/buffer.md +++ b/docs/ru/engines/table-engines/special/buffer.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: Buffer +--- + # Buffer {#buffer} Буферизует записываемые данные в оперативке, периодически сбрасывая их в другую таблицу. При чтении, производится чтение данных одновременно из буфера и из другой таблицы. diff --git a/docs/ru/engines/table-engines/special/dictionary.md b/docs/ru/engines/table-engines/special/dictionary.md index b35c8cecc8f..048da157b2d 100644 --- a/docs/ru/engines/table-engines/special/dictionary.md +++ b/docs/ru/engines/table-engines/special/dictionary.md @@ -1,3 +1,8 @@ +--- +toc_priority: 35 +toc_title: Dictionary +--- + # Dictionary {#dictionary} Движок `Dictionary` отображает данные [словаря](../../../sql-reference/dictionaries/external-dictionaries/external-dicts.md) как таблицу ClickHouse. diff --git a/docs/ru/engines/table-engines/special/distributed.md b/docs/ru/engines/table-engines/special/distributed.md index c8f7fe6eba7..7ab0b916337 100644 --- a/docs/ru/engines/table-engines/special/distributed.md +++ b/docs/ru/engines/table-engines/special/distributed.md @@ -1,3 +1,8 @@ +--- +toc_priority: 33 +toc_title: Distributed +--- + # Distributed {#distributed} **Движок Distributed не хранит данные самостоятельно**, а позволяет обрабатывать запросы распределённо, на нескольких серверах. Чтение автоматически распараллеливается. При чтении будут использованы индексы таблиц на удалённых серверах, если есть. diff --git a/docs/ru/engines/table-engines/special/external-data.md b/docs/ru/engines/table-engines/special/external-data.md index 41708978367..7e383c0c12d 100644 --- a/docs/ru/engines/table-engines/special/external-data.md +++ b/docs/ru/engines/table-engines/special/external-data.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: "\u0412\u043d\u0435\u0448\u043d\u0438\u0435\u0020\u0434\u0430\u043d\u043d\u044b\u0435\u0020\u0434\u043b\u044f\u0020\u043e\u0431\u0440\u0430\u0431\u043e\u0442\u043a\u0438\u0020\u0437\u0430\u043f\u0440\u043e\u0441\u0430" +--- + # Внешние данные для обработки запроса {#vneshnie-dannye-dlia-obrabotki-zaprosa} ClickHouse позволяет отправить на сервер данные, необходимые для обработки одного запроса, вместе с запросом SELECT. Такие данные будут положены во временную таблицу (см. раздел «Временные таблицы») и смогут использоваться в запросе (например, в операторах IN). diff --git a/docs/ru/engines/table-engines/special/file.md b/docs/ru/engines/table-engines/special/file.md index 907988a47aa..6a55ef31732 100644 --- a/docs/ru/engines/table-engines/special/file.md +++ b/docs/ru/engines/table-engines/special/file.md @@ -1,3 +1,8 @@ +--- +toc_priority: 37 +toc_title: File +--- + # File(Format) {#table_engines-file} Управляет данными в одном файле на диске в указанном формате. diff --git a/docs/ru/engines/table-engines/special/index.md b/docs/ru/engines/table-engines/special/index.md index dcf46c503b2..0d86461dd2d 100644 --- a/docs/ru/engines/table-engines/special/index.md +++ b/docs/ru/engines/table-engines/special/index.md @@ -1,5 +1,5 @@ --- -toc_folder_title: Special +toc_folder_title: "\u0421\u043f\u0435\u0446\u0438\u0430\u043b\u044c\u043d\u044b\u0435\u0020\u0434\u0432\u0438\u0436\u043a\u0438\u0020\u0442\u0430\u0431\u043b\u0438\u0446" toc_priority: 31 --- diff --git a/docs/ru/engines/table-engines/special/join.md b/docs/ru/engines/table-engines/special/join.md index aa734f5ca55..65bece4724c 100644 --- a/docs/ru/engines/table-engines/special/join.md +++ b/docs/ru/engines/table-engines/special/join.md @@ -1,3 +1,8 @@ +--- +toc_priority: 40 +toc_title: Join +--- + # Join {#join} Подготовленная структура данных для использования в операциях [JOIN](../../../engines/table-engines/special/join.md#select-join). diff --git a/docs/ru/engines/table-engines/special/materializedview.md b/docs/ru/engines/table-engines/special/materializedview.md index ae7a9965a24..1281d1db9ab 100644 --- a/docs/ru/engines/table-engines/special/materializedview.md +++ b/docs/ru/engines/table-engines/special/materializedview.md @@ -1,3 +1,8 @@ +--- +toc_priority: 43 +toc_title: MaterializedView +--- + # MaterializedView {#materializedview} Используется для реализации материализованных представлений (подробнее см. запрос [CREATE TABLE](../../../sql-reference/statements/create/table.md#create-table-query)). Для хранения данных, использует другой движок, который был указан при создании представления. При чтении из таблицы, просто использует этот движок. diff --git a/docs/ru/engines/table-engines/special/memory.md b/docs/ru/engines/table-engines/special/memory.md index 6de7978d134..9ca189ef3b2 100644 --- a/docs/ru/engines/table-engines/special/memory.md +++ b/docs/ru/engines/table-engines/special/memory.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: Memory +--- + # Memory {#memory} Хранит данные в оперативке, в несжатом виде. Данные хранятся именно в таком виде, в каком они получаются при чтении. То есть, само чтение из этой таблицы полностью бесплатно. diff --git a/docs/ru/engines/table-engines/special/merge.md b/docs/ru/engines/table-engines/special/merge.md index 4f6e1751591..656aa7cfd6b 100644 --- a/docs/ru/engines/table-engines/special/merge.md +++ b/docs/ru/engines/table-engines/special/merge.md @@ -1,3 +1,8 @@ +--- +toc_priority: 36 +toc_title: Merge +--- + # Merge {#merge} Движок `Merge` (не путайте с движком `MergeTree`) не хранит данные самостоятельно, а позволяет читать одновременно из произвольного количества других таблиц. diff --git a/docs/ru/engines/table-engines/special/null.md b/docs/ru/engines/table-engines/special/null.md index ac8c7942f79..2c3af1ce11e 100644 --- a/docs/ru/engines/table-engines/special/null.md +++ b/docs/ru/engines/table-engines/special/null.md @@ -1,4 +1,7 @@ -# Null {#null} +--- +toc_priority: 38 +toc_title: 'Null' +--- При записи в таблицу типа Null, данные игнорируются. При чтении из таблицы типа Null, возвращается пустота. diff --git a/docs/ru/engines/table-engines/special/set.md b/docs/ru/engines/table-engines/special/set.md index 497d52821d1..73fe3c3cf50 100644 --- a/docs/ru/engines/table-engines/special/set.md +++ b/docs/ru/engines/table-engines/special/set.md @@ -1,3 +1,8 @@ +--- +toc_priority: 39 +toc_title: Set +--- + # Set {#set} Представляет собой множество, постоянно находящееся в оперативке. Предназначено для использования в правой части оператора IN (смотрите раздел «Операторы IN»). diff --git a/docs/ru/engines/table-engines/special/url.md b/docs/ru/engines/table-engines/special/url.md index 69c1567a1c2..cdb5afddf75 100644 --- a/docs/ru/engines/table-engines/special/url.md +++ b/docs/ru/engines/table-engines/special/url.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: URL +--- + # URL(URL, Format) {#table_engines-url} Управляет данными на удаленном HTTP/HTTPS сервере. Данный движок похож diff --git a/docs/ru/engines/table-engines/special/view.md b/docs/ru/engines/table-engines/special/view.md index a732a675d4c..18813a55da2 100644 --- a/docs/ru/engines/table-engines/special/view.md +++ b/docs/ru/engines/table-engines/special/view.md @@ -1,3 +1,8 @@ +--- +toc_priority: 42 +toc_title: View +--- + # View {#table_engines-view} Используется для реализации представлений (подробнее см. запрос `CREATE VIEW`). Не хранит данные, а хранит только указанный запрос `SELECT`. При чтении из таблицы, выполняет его (с удалением из запроса всех ненужных столбцов). diff --git a/docs/ru/faq/index.md b/docs/ru/faq/index.md index 2ee9d51e83b..d9eccf59c5c 100644 --- a/docs/ru/faq/index.md +++ b/docs/ru/faq/index.md @@ -1,6 +1,6 @@ --- toc_folder_title: F.A.Q. +toc_hidden: true toc_priority: 76 --- - diff --git a/docs/ru/getting-started/example-datasets/amplab-benchmark.md b/docs/ru/getting-started/example-datasets/amplab-benchmark.md index b503fc909f3..bc59672ab26 100644 --- a/docs/ru/getting-started/example-datasets/amplab-benchmark.md +++ b/docs/ru/getting-started/example-datasets/amplab-benchmark.md @@ -1,3 +1,8 @@ +--- +toc_priority: 19 +toc_title: AMPLab Big Data Benchmark +--- + # AMPLab Big Data Benchmark {#amplab-big-data-benchmark} См. https://amplab.cs.berkeley.edu/benchmark/ diff --git a/docs/ru/getting-started/example-datasets/criteo.md b/docs/ru/getting-started/example-datasets/criteo.md index 4abb8c43f35..7a58da5b695 100644 --- a/docs/ru/getting-started/example-datasets/criteo.md +++ b/docs/ru/getting-started/example-datasets/criteo.md @@ -1,3 +1,8 @@ +--- +toc_priority: 18 +toc_title: "\u0422\u0435\u0440\u0430\u0431\u0430\u0439\u0442\u0020\u043b\u043e\u0433\u043e\u0432\u0020\u043a\u043b\u0438\u043a\u043e\u0432\u0020\u043e\u0442\u0020\u0043\u0072\u0069\u0074\u0065\u006f" +--- + # Терабайт логов кликов от Criteo {#terabait-logov-klikov-ot-criteo} Скачайте данные с http://labs.criteo.com/downloads/download-terabyte-click-logs/ diff --git a/docs/ru/getting-started/example-datasets/index.md b/docs/ru/getting-started/example-datasets/index.md index c349e369b1d..eff944a7980 100644 --- a/docs/ru/getting-started/example-datasets/index.md +++ b/docs/ru/getting-started/example-datasets/index.md @@ -1,8 +1,7 @@ --- -toc_folder_title: "\u0422\u0435\u0441\u0442\u043E\u0432\u044B\u0435 \u043C\u0430\u0441\ - \u0441\u0438\u0432\u044B \u0434\u0430\u043D\u043D\u044B\u0445" -toc_priority: 12 -toc_title: "\u041E\u0431\u0437\u043E\u0440" +toc_folder_title: "\u0422\u0435\u0441\u0442\u043e\u0432\u044b\u0435\u0020\u043c\u0430\u0441\u0441\u0438\u0432\u044b\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +toc_priority: 14 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" --- # Тестовые массивы данных {#testovye-massivy-dannykh} diff --git a/docs/ru/getting-started/example-datasets/metrica.md b/docs/ru/getting-started/example-datasets/metrica.md index ab2fe3d6c64..e8a3163376c 100644 --- a/docs/ru/getting-started/example-datasets/metrica.md +++ b/docs/ru/getting-started/example-datasets/metrica.md @@ -1,3 +1,8 @@ +--- +toc_priority: 15 +toc_title: "\u0410\u043d\u043e\u043d\u0438\u043c\u0438\u0437\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0435\u0020\u0434\u0430\u043d\u043d\u044b\u0435\u0020\u042f\u043d\u0434\u0435\u043a\u0441\u002e\u041c\u0435\u0442\u0440\u0438\u043a\u0438" +--- + # Анонимизированные данные Яндекс.Метрики {#anonimizirovannye-dannye-iandeks-metriki} Датасет состоит из двух таблиц, содержащих анонимизированные данные о хитах (`hits_v1`) и визитах (`visits_v1`) Яндекс.Метрики. Каждую из таблиц можно скачать в виде сжатого `.tsv.xz`-файла или в виде уже готовых партиций. Также можно скачать расширенную версию таблицы `hits`, содержащую 100 миллионов строк в виде [архива c файлами TSV](https://clickhouse-datasets.s3.yandex.net/hits/tsv/hits_100m_obfuscated_v1.tsv.xz) и в виде [готовых партиций](https://clickhouse-datasets.s3.yandex.net/hits/partitions/hits_100m_obfuscated_v1.tar.xz). diff --git a/docs/ru/getting-started/example-datasets/nyc-taxi.md b/docs/ru/getting-started/example-datasets/nyc-taxi.md index 64202b2e765..1f981324261 100644 --- a/docs/ru/getting-started/example-datasets/nyc-taxi.md +++ b/docs/ru/getting-started/example-datasets/nyc-taxi.md @@ -1,3 +1,8 @@ +--- +toc_priority: 20 +toc_title: "\u0414\u0430\u043d\u043d\u044b\u0435\u0020\u043e\u0020\u0442\u0430\u043a\u0441\u0438\u0020\u0432\u0020\u041d\u044c\u044e\u002d\u0419\u043e\u0440\u043a\u0435" +--- + # Данные о такси в Нью-Йорке {#dannye-o-taksi-v-niu-iorke} Этот датасет может быть получен двумя способами: diff --git a/docs/ru/getting-started/example-datasets/ontime.md b/docs/ru/getting-started/example-datasets/ontime.md index 7e7a6be2196..4d3eea14da6 100644 --- a/docs/ru/getting-started/example-datasets/ontime.md +++ b/docs/ru/getting-started/example-datasets/ontime.md @@ -1,3 +1,8 @@ +--- +toc_priority: 21 +toc_title: OnTime +--- + # OnTime {#ontime} Этот датасет может быть получен двумя способами: diff --git a/docs/ru/getting-started/example-datasets/star-schema.md b/docs/ru/getting-started/example-datasets/star-schema.md index d938b1e8620..f7502e8409d 100644 --- a/docs/ru/getting-started/example-datasets/star-schema.md +++ b/docs/ru/getting-started/example-datasets/star-schema.md @@ -1,3 +1,8 @@ +--- +toc_priority: 16 +toc_title: Star Schema Benchmark +--- + # Star Schema Benchmark {#star-schema-benchmark} Компиляция dbgen: diff --git a/docs/ru/getting-started/example-datasets/wikistat.md b/docs/ru/getting-started/example-datasets/wikistat.md index 0e3e269fe9f..c5a877ff8fd 100644 --- a/docs/ru/getting-started/example-datasets/wikistat.md +++ b/docs/ru/getting-started/example-datasets/wikistat.md @@ -1,3 +1,8 @@ +--- +toc_priority: 17 +toc_title: WikiStat +--- + # WikiStat {#wikistat} См: http://dumps.wikimedia.org/other/pagecounts-raw/ diff --git a/docs/ru/getting-started/index.md b/docs/ru/getting-started/index.md index b2b292c4725..ab72ce4a1d2 100644 --- a/docs/ru/getting-started/index.md +++ b/docs/ru/getting-started/index.md @@ -1,6 +1,5 @@ --- -toc_folder_title: "\u041D\u0430\u0447\u0430\u043B\u043E \u0440\u0430\u0431\u043E\u0442\ - \u044B" +toc_folder_title: "\u041d\u0430\u0447\u0430\u043b\u043e\u0020\u0440\u0430\u0431\u043e\u0442\u044b" toc_hidden: true toc_priority: 8 toc_title: hidden diff --git a/docs/ru/getting-started/install.md b/docs/ru/getting-started/install.md index 3e9270ef3bc..fb14ecfe599 100644 --- a/docs/ru/getting-started/install.md +++ b/docs/ru/getting-started/install.md @@ -1,3 +1,8 @@ +--- +toc_priority: 11 +toc_title: "\u0423\u0441\u0442\u0430\u043d\u043e\u0432\u043a\u0430" +--- + # Установка {#ustanovka} ## Системные требования {#sistemnye-trebovaniia} diff --git a/docs/ru/getting-started/playground.md b/docs/ru/getting-started/playground.md index e3bae5c9756..86a5cd5272c 100644 --- a/docs/ru/getting-started/playground.md +++ b/docs/ru/getting-started/playground.md @@ -1,3 +1,8 @@ +--- +toc_priority: 14 +toc_title: Playground +--- + # ClickHouse Playground {#clickhouse-playground} [ClickHouse Playground](https://play.clickhouse.tech) позволяет пользователям экспериментировать с ClickHouse, мгновенно выполняя запросы без настройки своего сервера или кластера. diff --git a/docs/ru/guides/apply-catboost-model.md b/docs/ru/guides/apply-catboost-model.md index 3515d2731c2..026b4d9d75e 100644 --- a/docs/ru/guides/apply-catboost-model.md +++ b/docs/ru/guides/apply-catboost-model.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: "\u041f\u0440\u0438\u043c\u0435\u043d\u0435\u043d\u0438\u0435\u0020\u043c\u043e\u0434\u0435\u043b\u0438\u0020\u0043\u0061\u0074\u0042\u006f\u006f\u0073\u0074\u0020\u0432\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + # Применение модели CatBoost в ClickHouse {#applying-catboost-model-in-clickhouse} [CatBoost](https://catboost.ai) — открытая программная библиотека разработанная компанией [Яндекс](https://yandex.ru/company/) для машинного обучения, которая использует схему градиентного бустинга. diff --git a/docs/ru/interfaces/cli.md b/docs/ru/interfaces/cli.md index 816b5962280..222af33f952 100644 --- a/docs/ru/interfaces/cli.md +++ b/docs/ru/interfaces/cli.md @@ -1,3 +1,8 @@ +--- +toc_priority: 17 +toc_title: "\u041a\u043b\u0438\u0435\u043d\u0442\u0020\u043a\u043e\u043c\u0430\u043d\u0434\u043d\u043e\u0439\u0020\u0441\u0442\u0440\u043e\u043a\u0438" +--- + # Клиент командной строки {#klient-komandnoi-stroki} ClickHouse предоставляет собственный клиент командной строки: `clickhouse-client`. Клиент поддерживает запуск с аргументами командной строки и с конфигурационными файлами. Подробнее читайте в разделе [Конфигурирование](#interfaces_cli_configuration). diff --git a/docs/ru/interfaces/cpp.md b/docs/ru/interfaces/cpp.md index 7136af3197b..264b4f82500 100644 --- a/docs/ru/interfaces/cpp.md +++ b/docs/ru/interfaces/cpp.md @@ -1,3 +1,8 @@ +--- +toc_priority: 24 +toc_title: "\u0043\u002b\u002b\u0020\u043a\u043b\u0438\u0435\u043d\u0442\u0441\u043a\u0430\u044f\u0020\u0431\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0430" +--- + # C++ клиентская библиотека {#c-klientskaia-biblioteka} См. README в репозитории [clickhouse-cpp](https://github.com/ClickHouse/clickhouse-cpp). diff --git a/docs/ru/interfaces/formats.md b/docs/ru/interfaces/formats.md index 2745139998f..042c62e310c 100644 --- a/docs/ru/interfaces/formats.md +++ b/docs/ru/interfaces/formats.md @@ -1,44 +1,61 @@ +--- +toc_priority: 21 +toc_title: "\u0424\u043e\u0440\u043c\u0430\u0442\u044b\u0020\u0432\u0445\u043e\u0434\u043d\u044b\u0445\u0020\u0438\u0020\u0432\u044b\u0445\u043e\u0434\u043d\u044b\u0445\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +--- + # Форматы входных и выходных данных {#formats} ClickHouse может принимать (`INSERT`) и отдавать (`SELECT`) данные в различных форматах. Поддерживаемые форматы и возможность использовать их в запросах `INSERT` и `SELECT` перечислены в таблице ниже. -| Формат | INSERT | SELECT | -|-----------------------------------------------------------------|--------|--------| -| [TabSeparated](#tabseparated) | ✔ | ✔ | -| [TabSeparatedRaw](#tabseparatedraw) | ✔ | ✔ | -| [TabSeparatedWithNames](#tabseparatedwithnames) | ✔ | ✔ | -| [TabSeparatedWithNamesAndTypes](#tabseparatedwithnamesandtypes) | ✔ | ✔ | -| [Template](#format-template) | ✔ | ✔ | -| [TemplateIgnoreSpaces](#templateignorespaces) | ✔ | ✗ | -| [CSV](#csv) | ✔ | ✔ | -| [CSVWithNames](#csvwithnames) | ✔ | ✔ | -| [CustomSeparated](#format-customseparated) | ✔ | ✔ | -| [Values](#data-format-values) | ✔ | ✔ | -| [Vertical](#vertical) | ✗ | ✔ | -| [JSON](#json) | ✗ | ✔ | -| [JSONCompact](#jsoncompact) | ✗ | ✔ | -| [JSONEachRow](#jsoneachrow) | ✔ | ✔ | -| [TSKV](#tskv) | ✔ | ✔ | -| [Pretty](#pretty) | ✗ | ✔ | -| [PrettyCompact](#prettycompact) | ✗ | ✔ | -| [PrettyCompactMonoBlock](#prettycompactmonoblock) | ✗ | ✔ | -| [PrettyNoEscapes](#prettynoescapes) | ✗ | ✔ | -| [PrettySpace](#prettyspace) | ✗ | ✔ | -| [Protobuf](#protobuf) | ✔ | ✔ | -| [ProtobufSingle](#protobufsingle) | ✔ | ✔ | -| [Parquet](#data-format-parquet) | ✔ | ✔ | -| [Arrow](#data-format-arrow) | ✔ | ✔ | -| [ArrowStream](#data-format-arrow-stream) | ✔ | ✔ | -| [ORC](#data-format-orc) | ✔ | ✗ | -| [RowBinary](#rowbinary) | ✔ | ✔ | -| [RowBinaryWithNamesAndTypes](#rowbinarywithnamesandtypes) | ✔ | ✔ | -| [Native](#native) | ✔ | ✔ | -| [Null](#null) | ✗ | ✔ | -| [XML](#xml) | ✗ | ✔ | -| [CapnProto](#capnproto) | ✔ | ✗ | -| [LineAsString](#lineasstring) | ✔ | ✗ | +| Формат | INSERT | SELECT | +|-----------------------------------------------------------------------------------------|--------|--------| +| [TabSeparated](#tabseparated) | ✔ | ✔ | +| [TabSeparatedRaw](#tabseparatedraw) | ✔ | ✔ | +| [TabSeparatedWithNames](#tabseparatedwithnames) | ✔ | ✔ | +| [TabSeparatedWithNamesAndTypes](#tabseparatedwithnamesandtypes) | ✔ | ✔ | +| [Template](#format-template) | ✔ | ✔ | +| [TemplateIgnoreSpaces](#templateignorespaces) | ✔ | ✗ | +| [CSV](#csv) | ✔ | ✔ | +| [CSVWithNames](#csvwithnames) | ✔ | ✔ | +| [CustomSeparated](#format-customseparated) | ✔ | ✔ | +| [Values](#data-format-values) | ✔ | ✔ | +| [Vertical](#vertical) | ✗ | ✔ | +| [VerticalRaw](#verticalraw) | ✗ | ✔ | +| [JSON](#json) | ✗ | ✔ | +| [JSONString](#jsonstring) | ✗ | ✔ | +| [JSONCompact](#jsoncompact) | ✗ | ✔ | +| [JSONCompactString](#jsoncompactstring) | ✗ | ✔ | +| [JSONEachRow](#jsoneachrow) | ✔ | ✔ | +| [JSONEachRowWithProgress](#jsoneachrowwithprogress) | ✗ | ✔ | +| [JSONStringEachRow](#jsonstringeachrow) | ✔ | ✔ | +| [JSONStringEachRowWithProgress](#jsonstringeachrowwithprogress) | ✗ | ✔ | +| [JSONCompactEachRow](#jsoncompacteachrow) | ✔ | ✔ | +| [JSONCompactEachRowWithNamesAndTypes](#jsoncompacteachrowwithnamesandtypes) | ✔ | ✔ | +| [JSONCompactStringEachRow](#jsoncompactstringeachrow) | ✔ | ✔ | +| [JSONCompactStringEachRowWithNamesAndTypes](#jsoncompactstringeachrowwithnamesandtypes) | ✔ | ✔ | +| [TSKV](#tskv) | ✔ | ✔ | +| [Pretty](#pretty) | ✗ | ✔ | +| [PrettyCompact](#prettycompact) | ✗ | ✔ | +| [PrettyCompactMonoBlock](#prettycompactmonoblock) | ✗ | ✔ | +| [PrettyNoEscapes](#prettynoescapes) | ✗ | ✔ | +| [PrettySpace](#prettyspace) | ✗ | ✔ | +| [Protobuf](#protobuf) | ✔ | ✔ | +| [ProtobufSingle](#protobufsingle) | ✔ | ✔ | +| [Avro](#data-format-avro) | ✔ | ✔ | +| [AvroConfluent](#data-format-avro-confluent) | ✔ | ✗ | +| [Parquet](#data-format-parquet) | ✔ | ✔ | +| [Arrow](#data-format-arrow) | ✔ | ✔ | +| [ArrowStream](#data-format-arrow-stream) | ✔ | ✔ | +| [ORC](#data-format-orc) | ✔ | ✗ | +| [RowBinary](#rowbinary) | ✔ | ✔ | +| [RowBinaryWithNamesAndTypes](#rowbinarywithnamesandtypes) | ✔ | ✔ | +| [Native](#native) | ✔ | ✔ | +| [Null](#null) | ✗ | ✔ | +| [XML](#xml) | ✗ | ✔ | +| [CapnProto](#capnproto) | ✔ | ✗ | +| [LineAsString](#lineasstring) | ✔ | ✗ | Вы можете регулировать некоторые параметры работы с форматами с помощью настроек ClickHouse. За дополнительной информацией обращайтесь к разделу [Настройки](../operations/settings/settings.md). @@ -364,62 +381,41 @@ SELECT SearchPhrase, count() AS c FROM test.hits GROUP BY SearchPhrase WITH TOTA "meta": [ { - "name": "SearchPhrase", + "name": "'hello'", "type": "String" }, { - "name": "c", + "name": "multiply(42, number)", "type": "UInt64" + }, + { + "name": "range(5)", + "type": "Array(UInt8)" } ], "data": [ { - "SearchPhrase": "", - "c": "8267016" + "'hello'": "hello", + "multiply(42, number)": "0", + "range(5)": [0,1,2,3,4] }, { - "SearchPhrase": "bathroom interior design", - "c": "2166" + "'hello'": "hello", + "multiply(42, number)": "42", + "range(5)": [0,1,2,3,4] }, { - "SearchPhrase": "yandex", - "c": "1655" - }, - { - "SearchPhrase": "spring 2014 fashion", - "c": "1549" - }, - { - "SearchPhrase": "freeform photos", - "c": "1480" + "'hello'": "hello", + "multiply(42, number)": "84", + "range(5)": [0,1,2,3,4] } ], - "totals": - { - "SearchPhrase": "", - "c": "8873898" - }, + "rows": 3, - "extremes": - { - "min": - { - "SearchPhrase": "", - "c": "1480" - }, - "max": - { - "SearchPhrase": "", - "c": "8267016" - } - }, - - "rows": 5, - - "rows_before_limit_at_least": 141137 + "rows_before_limit_at_least": 3 } ``` @@ -438,65 +434,167 @@ JSON совместим с JavaScript. Для этого, дополнитель ClickHouse поддерживает [NULL](../sql-reference/syntax.md), который при выводе JSON будет отображен как `null`. Чтобы включить отображение в результате значений `+nan`, `-nan`, `+inf`, `-inf`, установите параметр [output_format_json_quote_denormals](../operations/settings/settings.md#settings-output_format_json_quote_denormals) равным 1. -Смотрите также формат [JSONEachRow](#jsoneachrow) . +Смотрите также формат [JSONEachRow](#jsoneachrow). + +## JSONString {#jsonstring} + +Отличается от JSON только тем, что поля данных выводятся в строках, а не в типизированных значениях JSON. + +Пример: + +```json +{ + "meta": + [ + { + "name": "'hello'", + "type": "String" + }, + { + "name": "multiply(42, number)", + "type": "UInt64" + }, + { + "name": "range(5)", + "type": "Array(UInt8)" + } + ], + + "data": + [ + { + "'hello'": "hello", + "multiply(42, number)": "0", + "range(5)": "[0,1,2,3,4]" + }, + { + "'hello'": "hello", + "multiply(42, number)": "42", + "range(5)": "[0,1,2,3,4]" + }, + { + "'hello'": "hello", + "multiply(42, number)": "84", + "range(5)": "[0,1,2,3,4]" + } + ], + + "rows": 3, + + "rows_before_limit_at_least": 3 +} +``` ## JSONCompact {#jsoncompact} +## JSONCompactString {#jsoncompactstring} Отличается от JSON только тем, что строчки данных выводятся в массивах, а не в object-ах. Пример: ``` json +// JSONCompact { "meta": [ { - "name": "SearchPhrase", + "name": "'hello'", "type": "String" }, { - "name": "c", + "name": "multiply(42, number)", "type": "UInt64" + }, + { + "name": "range(5)", + "type": "Array(UInt8)" } ], "data": [ - ["", "8267016"], - ["интерьер ванной комнаты", "2166"], - ["яндекс", "1655"], - ["весна 2014 мода", "1549"], - ["фриформ фото", "1480"] + ["hello", "0", [0,1,2,3,4]], + ["hello", "42", [0,1,2,3,4]], + ["hello", "84", [0,1,2,3,4]] ], - "totals": ["","8873898"], + "rows": 3, - "extremes": - { - "min": ["","1480"], - "max": ["","8267016"] - }, - - "rows": 5, - - "rows_before_limit_at_least": 141137 + "rows_before_limit_at_least": 3 } ``` -Этот формат подходит только для вывода результата выполнения запроса, но не для парсинга (приёма данных для вставки в таблицу). -Смотрите также формат `JSONEachRow`. +```json +// JSONCompactString +{ + "meta": + [ + { + "name": "'hello'", + "type": "String" + }, + { + "name": "multiply(42, number)", + "type": "UInt64" + }, + { + "name": "range(5)", + "type": "Array(UInt8)" + } + ], -## JSONEachRow {#jsoneachrow} + "data": + [ + ["hello", "0", "[0,1,2,3,4]"], + ["hello", "42", "[0,1,2,3,4]"], + ["hello", "84", "[0,1,2,3,4]"] + ], -При использовании этого формата, ClickHouse выводит каждую запись как объект JSON (каждый объект отдельной строкой), при этом данные в целом — невалидный JSON. + "rows": 3, -``` json -{"SearchPhrase":"дизайн штор","count()":"1064"} -{"SearchPhrase":"баку","count()":"1000"} -{"SearchPhrase":"","count":"8267016"} + "rows_before_limit_at_least": 3 +} ``` -При вставке данных необходимо каждую запись передавать как отдельный объект JSON. +## JSONEachRow {#jsoneachrow} +## JSONStringEachRow {#jsonstringeachrow} +## JSONCompactEachRow {#jsoncompacteachrow} +## JSONCompactStringEachRow {#jsoncompactstringeachrow} + +При использовании этих форматов ClickHouse выводит каждую запись как значения JSON (каждое значение отдельной строкой), при этом данные в целом — невалидный JSON. + +``` json +{"some_int":42,"some_str":"hello","some_tuple":[1,"a"]} // JSONEachRow +[42,"hello",[1,"a"]] // JSONCompactEachRow +["42","hello","(2,'a')"] // JSONCompactStringsEachRow +``` + +При вставке данных вы должны предоставить отдельное значение JSON для каждой строки. + +## JSONEachRowWithProgress {#jsoneachrowwithprogress} +## JSONStringEachRowWithProgress {#jsonstringeachrowwithprogress} + +Отличается от `JSONEachRow`/`JSONStringEachRow` тем, что ClickHouse будет выдавать информацию о ходе выполнения в виде значений JSON. + +```json +{"row":{"'hello'":"hello","multiply(42, number)":"0","range(5)":[0,1,2,3,4]}} +{"row":{"'hello'":"hello","multiply(42, number)":"42","range(5)":[0,1,2,3,4]}} +{"row":{"'hello'":"hello","multiply(42, number)":"84","range(5)":[0,1,2,3,4]}} +{"progress":{"read_rows":"3","read_bytes":"24","written_rows":"0","written_bytes":"0","total_rows_to_read":"3"}} +``` + +## JSONCompactEachRowWithNamesAndTypes {#jsoncompacteachrowwithnamesandtypes} +## JSONCompactStringEachRowWithNamesAndTypes {#jsoncompactstringeachrowwithnamesandtypes} + +Отличается от `JSONCompactEachRow`/`JSONCompactStringEachRow` тем, что имена и типы столбцов записываются как первые две строки. + +```json +["'hello'", "multiply(42, number)", "range(5)"] +["String", "UInt64", "Array(UInt8)"] +["hello", "0", [0,1,2,3,4]] +["hello", "42", [0,1,2,3,4]] +["hello", "84", [0,1,2,3,4]] +``` ### Вставка данных {#vstavka-dannykh} @@ -784,6 +882,10 @@ test: string with 'quotes' and with some special Этот формат подходит только для вывода результата выполнения запроса, но не для парсинга (приёма данных для вставки в таблицу). +## VerticalRaw {#verticalraw} + +Аналогичен [Vertical](#vertical), но с отключенным выходом. Этот формат подходит только для вывода результата выполнения запроса, но не для парсинга (приёма данных для вставки в таблицу). + ## XML {#xml} Формат XML подходит только для вывода данных, не для парсинга. Пример: diff --git a/docs/ru/interfaces/http.md b/docs/ru/interfaces/http.md index ffe4b2e5276..096fb6c05bc 100644 --- a/docs/ru/interfaces/http.md +++ b/docs/ru/interfaces/http.md @@ -1,3 +1,8 @@ +--- +toc_priority: 19 +toc_title: "\u0048\u0054\u0054\u0050\u002d\u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441" +--- + # HTTP-интерфейс {#http-interface} HTTP интерфейс позволяет использовать ClickHouse на любой платформе, из любого языка программирования. У нас он используется для работы из Java и Perl, а также из shell-скриптов. В других отделах, HTTP интерфейс используется из Perl, Python и Go. HTTP интерфейс более ограничен по сравнению с родным интерфейсом, но является более совместимым. @@ -71,8 +76,11 @@ ECT 1 ``` По умолчанию, данные возвращаются в формате TabSeparated (подробнее смотри раздел «Форматы»). + Можно попросить любой другой формат - с помощью секции FORMAT запроса. +Кроме того, вы можете использовать параметр URL-адреса `default_format` или заголовок `X-ClickHouse-Format`, чтобы указать формат по умолчанию, отличный от `TabSeparated`. + ``` bash $ echo 'SELECT 1 FORMAT Pretty' | curl 'http://localhost:8123/?' --data-binary @- ┏━━━┓ @@ -163,7 +171,7 @@ $ echo "SELECT 1" | gzip -c | curl -sS --data-binary @- -H 'Content-Encoding: gz !!! note "Примечание" Некоторые HTTP-клиенты могут по умолчанию распаковывать данные (`gzip` и `deflate`) с сервера в фоновом режиме и вы можете получить распакованные данные, даже если правильно используете настройки сжатия. -В параметре URL database может быть указана БД по умолчанию. +Вы можете использовать параметр URL `database` или заголовок `X-ClickHouse-Database`, чтобы указать БД по умолчанию. ``` bash $ echo 'SELECT number FROM numbers LIMIT 10' | curl 'http://localhost:8123/?database=system' --data-binary @- diff --git a/docs/ru/interfaces/jdbc.md b/docs/ru/interfaces/jdbc.md index e83cd2cfacf..196dba64933 100644 --- a/docs/ru/interfaces/jdbc.md +++ b/docs/ru/interfaces/jdbc.md @@ -1,3 +1,8 @@ +--- +toc_priority: 22 +toc_title: "\u004a\u0044\u0042\u0043\u002d\u0434\u0440\u0430\u0439\u0432\u0435\u0440" +--- + # JDBC-драйвер {#jdbc-draiver} - **[Официальный драйвер](https://github.com/ClickHouse/clickhouse-jdbc)** diff --git a/docs/ru/interfaces/mysql.md b/docs/ru/interfaces/mysql.md index 79368aa1ba2..fa0003e0bea 100644 --- a/docs/ru/interfaces/mysql.md +++ b/docs/ru/interfaces/mysql.md @@ -1,3 +1,8 @@ +--- +toc_priority: 20 +toc_title: "\u004d\u0079\u0053\u0051\u004c\u002d\u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441" +--- + # MySQL-интерфейс {#mysql-interface} ClickHouse поддерживает взаимодействие по протоколу MySQL. Данная функция включается настройкой [mysql_port](../operations/server-configuration-parameters/settings.md#server_configuration_parameters-mysql_port) в конфигурационном файле: diff --git a/docs/ru/interfaces/odbc.md b/docs/ru/interfaces/odbc.md index 0f4a6279539..728c4bd6979 100644 --- a/docs/ru/interfaces/odbc.md +++ b/docs/ru/interfaces/odbc.md @@ -1,3 +1,9 @@ +--- +toc_priority: 23 +toc_title: "\u004f\u0044\u0042\u0043\u002d\u0434\u0440\u0430\u0439\u0432\u0435\u0440" +--- + + # ODBC-драйвер {#odbc-draiver} - [Официальный драйвер](https://github.com/ClickHouse/clickhouse-odbc). diff --git a/docs/ru/interfaces/tcp.md b/docs/ru/interfaces/tcp.md index 6bdfb286ac2..d89646f15b7 100644 --- a/docs/ru/interfaces/tcp.md +++ b/docs/ru/interfaces/tcp.md @@ -1,3 +1,8 @@ +--- +toc_priority: 18 +toc_title: "\u0420\u043e\u0434\u043d\u043e\u0439\u0020\u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441\u0020\u0028\u0054\u0043\u0050\u0029" +--- + # Родной интерфейс (TCP) {#rodnoi-interfeis-tcp} Нативный протокол используется в [клиенте командной строки](cli.md), для взаимодействия между серверами во время обработки распределенных запросов, а также в других программах на C++. К сожалению, у родного протокола ClickHouse пока нет формальной спецификации, но в нем можно разобраться с использованием исходного кода ClickHouse (начиная с [примерно этого места](https://github.com/ClickHouse/ClickHouse/tree/master/src/Client)) и/или путем перехвата и анализа TCP трафика. diff --git a/docs/ru/interfaces/third-party/client-libraries.md b/docs/ru/interfaces/third-party/client-libraries.md index 66a2252b23a..f35acb9e968 100644 --- a/docs/ru/interfaces/third-party/client-libraries.md +++ b/docs/ru/interfaces/third-party/client-libraries.md @@ -1,3 +1,8 @@ +--- +toc_priority: 26 +toc_title: "\u041a\u043b\u0438\u0435\u043d\u0442\u0441\u043a\u0438\u0435\u0020\u0431\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0438\u0020\u043e\u0442\u0020\u0441\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0445\u0020\u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432" +--- + # Клиентские библиотеки от сторонних разработчиков {#klientskie-biblioteki-ot-storonnikh-razrabotchikov} !!! warning "Disclaimer" diff --git a/docs/ru/interfaces/third-party/gui.md b/docs/ru/interfaces/third-party/gui.md index f7eaa5cc77f..1fabdb8a31c 100644 --- a/docs/ru/interfaces/third-party/gui.md +++ b/docs/ru/interfaces/third-party/gui.md @@ -1,3 +1,9 @@ +--- +toc_priority: 28 +toc_title: "\u0412\u0438\u0437\u0443\u0430\u043b\u044c\u043d\u044b\u0435\u0020\u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441\u044b\u0020\u043e\u0442\u0020\u0441\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0445\u0020\u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432" +--- + + # Визуальные интерфейсы от сторонних разработчиков {#vizualnye-interfeisy-ot-storonnikh-razrabotchikov} ## С открытым исходным кодом {#s-otkrytym-iskhodnym-kodom} diff --git a/docs/ru/interfaces/third-party/index.md b/docs/ru/interfaces/third-party/index.md index bfdd554aebd..a57169df73b 100644 --- a/docs/ru/interfaces/third-party/index.md +++ b/docs/ru/interfaces/third-party/index.md @@ -1,6 +1,5 @@ --- -toc_folder_title: "\u041E\u0442 \u0441\u0442\u043E\u0440\u043E\u043D\u043D\u0438\u0445\ - \ \u0440\u0430\u0437\u0440\u0430\u0431\u043E\u0442\u0447\u0438\u043A\u043E\u0432" +toc_folder_title: "\u0421\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0435\u0020\u0438\u043d\u0442\u0435\u0440\u0444\u0435\u0439\u0441\u044b" toc_priority: 24 --- diff --git a/docs/ru/interfaces/third-party/integrations.md b/docs/ru/interfaces/third-party/integrations.md index 527f65343db..60d6181ab3f 100644 --- a/docs/ru/interfaces/third-party/integrations.md +++ b/docs/ru/interfaces/third-party/integrations.md @@ -1,3 +1,8 @@ +--- +toc_priority: 27 +toc_title: "\u0411\u0438\u0431\u043b\u0438\u043e\u0442\u0435\u043a\u0438\u0020\u0434\u043b\u044f\u0020\u0438\u043d\u0442\u0435\u0433\u0440\u0430\u0446\u0438\u0438\u0020\u043e\u0442\u0020\u0441\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0445\u0020\u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432" +--- + # Библиотеки для интеграции от сторонних разработчиков {#biblioteki-dlia-integratsii-ot-storonnikh-razrabotchikov} !!! warning "Disclaimer" diff --git a/docs/ru/interfaces/third-party/proxy.md b/docs/ru/interfaces/third-party/proxy.md index 225c3fee150..fc66ecde293 100644 --- a/docs/ru/interfaces/third-party/proxy.md +++ b/docs/ru/interfaces/third-party/proxy.md @@ -1,3 +1,8 @@ +--- +toc_priority: 29 +toc_title: "\u041f\u0440\u043e\u043a\u0441\u0438\u002d\u0441\u0435\u0440\u0432\u0435\u0440\u044b\u0020\u043e\u0442\u0020\u0441\u0442\u043e\u0440\u043e\u043d\u043d\u0438\u0445\u0020\u0440\u0430\u0437\u0440\u0430\u0431\u043e\u0442\u0447\u0438\u043a\u043e\u0432" +--- + # Прокси-серверы от сторонних разработчиков {#proksi-servery-ot-storonnikh-razrabotchikov} ## chproxy {#chproxy} diff --git a/docs/ru/introduction/distinctive-features.md b/docs/ru/introduction/distinctive-features.md index 98dbe6df134..4eeeef4a443 100644 --- a/docs/ru/introduction/distinctive-features.md +++ b/docs/ru/introduction/distinctive-features.md @@ -1,3 +1,8 @@ +--- +toc_priority: 4 +toc_title: "\u041e\u0442\u043b\u0438\u0447\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0435\u0020\u0432\u043e\u0437\u043c\u043e\u0436\u043d\u043e\u0441\u0442\u0438\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + # Отличительные возможности ClickHouse {#otlichitelnye-vozmozhnosti-clickhouse} ## По-настоящему столбцовая СУБД {#po-nastoiashchemu-stolbtsovaia-subd} diff --git a/docs/ru/introduction/history.md b/docs/ru/introduction/history.md index 65254b0f4f0..ab740954bbe 100644 --- a/docs/ru/introduction/history.md +++ b/docs/ru/introduction/history.md @@ -1,3 +1,9 @@ +--- +toc_priority: 7 +toc_title: "\u0418\u0441\u0442\u043e\u0440\u0438\u044f\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + + # История ClickHouse {#istoriia-clickhouse} ClickHouse изначально разрабатывался для обеспечения работы [Яндекс.Метрики](https://metrika.yandex.ru/), [второй крупнейшей в мире](http://w3techs.com/technologies/overview/traffic_analysis/all) платформы для веб аналитики, и продолжает быть её ключевым компонентом. При более 13 триллионах записей в базе данных и более 20 миллиардах событий в сутки, ClickHouse позволяет генерировать индивидуально настроенные отчёты на лету напрямую из неагрегированных данных. Данная статья вкратце демонстрирует какие цели исторически стояли перед ClickHouse на ранних этапах его развития. diff --git a/docs/ru/introduction/performance.md b/docs/ru/introduction/performance.md index cbefa10cf7c..c449e76a6ea 100644 --- a/docs/ru/introduction/performance.md +++ b/docs/ru/introduction/performance.md @@ -1,3 +1,8 @@ +--- +toc_priority: 6 +toc_title: "\u041f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u044c" +--- + # Производительность {#proizvoditelnost} По результатам внутреннего тестирования в Яндексе, ClickHouse обладает наиболее высокой производительностью (как наиболее высокой пропускной способностью на длинных запросах, так и наиболее низкой задержкой на коротких запросах), при соответствующем сценарии работы, среди доступных для тестирования систем подобного класса. Результаты тестирования можно посмотреть на [отдельной странице](https://clickhouse.tech/benchmark/dbms/). diff --git a/docs/ru/operations/access-rights.md b/docs/ru/operations/access-rights.md index 27dbc2fbf62..9973de91161 100644 --- a/docs/ru/operations/access-rights.md +++ b/docs/ru/operations/access-rights.md @@ -1,3 +1,8 @@ +--- +toc_priority: 48 +toc_title: "\u0423\u043f\u0440\u0430\u0432\u043b\u0435\u043d\u0438\u0435\u0020\u0434\u043e\u0441\u0442\u0443\u043f\u043e\u043c" +--- + # Управление доступом {#access-control} ClickHouse поддерживает управление доступом на основе подхода [RBAC](https://ru.wikipedia.org/wiki/Управление_доступом_на_основе_ролей). diff --git a/docs/ru/operations/backup.md b/docs/ru/operations/backup.md index 89fb0403543..0dcb6fd307d 100644 --- a/docs/ru/operations/backup.md +++ b/docs/ru/operations/backup.md @@ -1,3 +1,8 @@ +--- +toc_priority: 49 +toc_title: "\u0420\u0435\u0437\u0435\u0440\u0432\u043d\u043e\u0435\u0020\u043a\u043e\u043f\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u0435\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +--- + # Резервное копирование данных {#rezervnoe-kopirovanie-dannykh} [Репликация](../engines/table-engines/mergetree-family/replication.md) обеспечивает защиту от аппаратных сбоев, но не защищает от человеческих ошибок: случайного удаления данных, удаления не той таблицы, которую надо было, или таблицы на не том кластере, а также программных ошибок, которые приводят к неправильной обработке данных или их повреждению. Во многих случаях подобные ошибки влияют на все реплики. ClickHouse имеет встроенные средства защиты для предотвращения некоторых типов ошибок — например, по умолчанию [не получится удалить таблицы \*MergeTree, содержащие более 50 Гб данных, одной командой](https://github.com/ClickHouse/ClickHouse/blob/v18.14.18-stable/programs/server/config.xml#L322-L330). Однако эти средства защиты не охватывают все возможные случаи и могут быть обойдены. diff --git a/docs/ru/operations/configuration-files.md b/docs/ru/operations/configuration-files.md index df56ab9ac7d..a4cc9182427 100644 --- a/docs/ru/operations/configuration-files.md +++ b/docs/ru/operations/configuration-files.md @@ -1,3 +1,9 @@ +--- +toc_priority: 50 +toc_title: "\u041a\u043e\u043d\u0444\u0438\u0433\u0443\u0440\u0430\u0446\u0438\u043e\u043d\u043d\u044b\u0435\u0020\u0444\u0430\u0439\u043b\u044b" +--- + + # Конфигурационные файлы {#configuration_files} Основной конфигурационный файл сервера - `config.xml`. Он расположен в директории `/etc/clickhouse-server/`. diff --git a/docs/ru/operations/index.md b/docs/ru/operations/index.md index 0a15959d652..74a1d135967 100644 --- a/docs/ru/operations/index.md +++ b/docs/ru/operations/index.md @@ -1,7 +1,7 @@ --- toc_folder_title: "\u042d\u043a\u0441\u043f\u043b\u0443\u0430\u0442\u0430\u0446\u0438\u044f" toc_priority: 41 -toc_title: intro +toc_title: "\u042d\u043a\u0441\u043f\u043b\u0443\u0430\u0442\u0430\u0446\u0438\u044f" --- # Эксплуатация {#operations} diff --git a/docs/ru/operations/monitoring.md b/docs/ru/operations/monitoring.md index a1013f5eae5..52d0b5ecc8a 100644 --- a/docs/ru/operations/monitoring.md +++ b/docs/ru/operations/monitoring.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: "\u041c\u043e\u043d\u0438\u0442\u043e\u0440\u0438\u043d\u0433" +--- + # Мониторинг {#monitoring} Вы можете отслеживать: diff --git a/docs/ru/operations/quotas.md b/docs/ru/operations/quotas.md index 399e80d2011..92533eef0c1 100644 --- a/docs/ru/operations/quotas.md +++ b/docs/ru/operations/quotas.md @@ -1,3 +1,8 @@ +--- +toc_priority: 51 +toc_title: "\u041a\u0432\u043e\u0442\u044b" +--- + # Квоты {#quotas} Квоты позволяют ограничить использование ресурсов за некоторый интервал времени, или просто подсчитывать использование ресурсов. diff --git a/docs/ru/operations/requirements.md b/docs/ru/operations/requirements.md index e1ecafd9f3b..36a7dd30b34 100644 --- a/docs/ru/operations/requirements.md +++ b/docs/ru/operations/requirements.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: "\u0422\u0440\u0435\u0431\u043e\u0432\u0430\u043d\u0438\u044f" +--- + # Требования {#trebovaniia} ## Процессор {#protsessor} diff --git a/docs/ru/operations/server-configuration-parameters/settings.md b/docs/ru/operations/server-configuration-parameters/settings.md index e8da92c1662..17966ef0547 100644 --- a/docs/ru/operations/server-configuration-parameters/settings.md +++ b/docs/ru/operations/server-configuration-parameters/settings.md @@ -1,3 +1,8 @@ +--- +toc_priority: 57 +toc_title: "\u041a\u043e\u043d\u0444\u0438\u0433\u0443\u0440\u0430\u0446\u0438\u043e\u043d\u043d\u044b\u0435\u0020\u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u044b\u0020\u0441\u0435\u0440\u0432\u0435\u0440\u0430" +--- + # Конфигурационные параметры сервера {#server-configuration-parameters-reference} ## builtin_dictionaries_reload_interval {#builtin-dictionaries-reload-interval} @@ -387,7 +392,7 @@ ClickHouse проверяет условия для `min_part_size` и `min_part Можно не указывать, если реплицируемых таблицы не используются. -Подробнее смотрите в разделе «[Создание реплицируемых таблиц](../../operations/server-configuration-parameters/settings.md)». +Подробнее смотрите в разделе «[Создание реплицируемых таблиц](../../engines/table-engines/mergetree-family/replication.md)». **Пример** diff --git a/docs/ru/operations/settings/constraints-on-settings.md b/docs/ru/operations/settings/constraints-on-settings.md index dd324caecc6..b23be22958c 100644 --- a/docs/ru/operations/settings/constraints-on-settings.md +++ b/docs/ru/operations/settings/constraints-on-settings.md @@ -1,3 +1,8 @@ +--- +toc_priority: 62 +toc_title: "\u041e\u0433\u0440\u0430\u043d\u0438\u0447\u0435\u043d\u0438\u044f\u0020\u043d\u0430\u0020\u0438\u0437\u043c\u0435\u043d\u0435\u043d\u0438\u0435\u0020\u043d\u0430\u0441\u0442\u0440\u043e\u0435\u043a" +--- + # Ограничения на изменение настроек {#constraints-on-settings} Ограничения на изменение настроек могут находиться внутри секции `profiles` файла `user.xml` и запрещают пользователю менять некоторые настройки с помощью запроса `SET`. diff --git a/docs/ru/operations/settings/index.md b/docs/ru/operations/settings/index.md index edc8d2d3014..c24b7053c46 100644 --- a/docs/ru/operations/settings/index.md +++ b/docs/ru/operations/settings/index.md @@ -1,7 +1,7 @@ --- -toc_folder_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" +toc_folder_title: Settings toc_priority: 55 -toc_title: "\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0438" +toc_title: Introduction --- # Настройки {#session-settings-intro} diff --git a/docs/ru/operations/settings/permissions-for-queries.md b/docs/ru/operations/settings/permissions-for-queries.md index cb65e652331..ae896dac77c 100644 --- a/docs/ru/operations/settings/permissions-for-queries.md +++ b/docs/ru/operations/settings/permissions-for-queries.md @@ -1,3 +1,8 @@ +--- +toc_priority: 58 +toc_title: "\u0420\u0430\u0437\u0440\u0435\u0448\u0435\u043d\u0438\u044f\u0020\u0434\u043b\u044f\u0020\u0437\u0430\u043f\u0440\u043e\u0441\u043e\u0432" +--- + # Разрешения для запросов {#permissions_for_queries} Запросы в ClickHouse можно разделить на несколько типов: diff --git a/docs/ru/operations/settings/query-complexity.md b/docs/ru/operations/settings/query-complexity.md index aeb5445a054..a62e7523207 100644 --- a/docs/ru/operations/settings/query-complexity.md +++ b/docs/ru/operations/settings/query-complexity.md @@ -1,3 +1,8 @@ +--- +toc_priority: 59 +toc_title: "\u041e\u0433\u0440\u0430\u043d\u0438\u0447\u0435\u043d\u0438\u044f\u0020\u043d\u0430\u0020\u0441\u043b\u043e\u0436\u043d\u043e\u0441\u0442\u044c\u0020\u0437\u0430\u043f\u0440\u043e\u0441\u0430" +--- + # Ограничения на сложность запроса {#restrictions-on-query-complexity} Ограничения на сложность запроса - часть настроек. diff --git a/docs/ru/operations/settings/settings-profiles.md b/docs/ru/operations/settings/settings-profiles.md index d1e24490120..10feda01850 100644 --- a/docs/ru/operations/settings/settings-profiles.md +++ b/docs/ru/operations/settings/settings-profiles.md @@ -1,3 +1,8 @@ +--- +toc_priority: 61 +toc_title: "\u041f\u0440\u043e\u0444\u0438\u043b\u0438\u0020\u043d\u0430\u0441\u0442\u0440\u043e\u0435\u043a" +--- + # Профили настроек {#settings-profiles} Профиль настроек — это набор настроек, сгруппированных под одним именем. diff --git a/docs/ru/operations/settings/settings-users.md b/docs/ru/operations/settings/settings-users.md index 7c12780823a..2069922d0ea 100644 --- a/docs/ru/operations/settings/settings-users.md +++ b/docs/ru/operations/settings/settings-users.md @@ -1,3 +1,8 @@ +--- +toc_priority: 63 +toc_title: "\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0438\u0020\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u0442\u0435\u043b\u0435\u0439" +--- + # Настройки пользователей {#nastroiki-polzovatelei} Раздел `users` конфигурационного файла `user.xml` содержит настройки для пользователей. diff --git a/docs/ru/operations/settings/settings.md b/docs/ru/operations/settings/settings.md index da933e30300..700203e36e1 100644 --- a/docs/ru/operations/settings/settings.md +++ b/docs/ru/operations/settings/settings.md @@ -1,6 +1,6 @@ --- toc_priority: 60 -toc_title: Settings +toc_title: "\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0438" --- # Настройки {#settings} @@ -1164,9 +1164,9 @@ ClickHouse генерирует исключение ## insert_quorum_timeout {#settings-insert_quorum_timeout} -Время ожидания кворумной записи в секундах. Если время прошло, а запись так не состоялась, то ClickHouse сгенерирует исключение и клиент должен повторить запрос на запись того же блока на эту же или любую другую реплику. +Время ожидания кворумной записи в миллисекундах. Если время прошло, а запись так не состоялась, то ClickHouse сгенерирует исключение и клиент должен повторить запрос на запись того же блока на эту же или любую другую реплику. -Значение по умолчанию: 60 секунд. +Значение по умолчанию: 600000 миллисекунд (10 минут). См. также: @@ -1977,6 +1977,48 @@ SELECT range(number) FROM system.numbers LIMIT 5 FORMAT PrettyCompactNoEscapes; └───────────────┘ ``` +## output_format_pretty_row_numbers {#output_format_pretty_row_numbers} + +Включает режим отображения номеров строк для запросов, выводимых в формате [Pretty](../../interfaces/formats.md#pretty). + +Возможные значения: + +- 0 — номера строк не выводятся. +- 1 — номера строк выводятся. + +Значение по умолчанию: `0`. + +**Пример** + +Запрос: + +```sql +SET output_format_pretty_row_numbers = 1; +SELECT TOP 3 name, value FROM system.settings; +``` + +Результат: + +```text + ┌─name────────────────────┬─value───┐ +1. │ min_compress_block_size │ 65536 │ +2. │ max_compress_block_size │ 1048576 │ +3. │ max_block_size │ 65505 │ + └─────────────────────────┴─────────┘ +``` + +## allow_experimental_bigint_types {#allow_experimental_bigint_types} + +Включает или отключает поддержку целочисленных значений, превышающих максимальное значение, допустимое для типа `int`. + +Возможные значения: + +- 1 — большие целочисленные значения поддерживаются. +- 0 — большие целочисленные значения не поддерживаются. + +Значение по умолчанию: `0`. + + ## lock_acquire_timeout {#lock_acquire_timeout} Устанавливает, сколько секунд сервер ожидает возможности выполнить блокировку таблицы. diff --git a/docs/ru/operations/system-tables/asynchronous_metric_log.md b/docs/ru/operations/system-tables/asynchronous_metric_log.md index 719e965ebe4..2fe617e48af 100644 --- a/docs/ru/operations/system-tables/asynchronous_metric_log.md +++ b/docs/ru/operations/system-tables/asynchronous_metric_log.md @@ -5,6 +5,7 @@ Столбцы: - `event_date` ([Date](../../sql-reference/data-types/date.md)) — дата события. - `event_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время события. +- `event_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — время события в микросекундах. - `name` ([String](../../sql-reference/data-types/string.md)) — название метрики. - `value` ([Float64](../../sql-reference/data-types/float.md)) — значение метрики. diff --git a/docs/ru/operations/system-tables/columns.md b/docs/ru/operations/system-tables/columns.md index b476d6907f8..8cb9408e7d8 100644 --- a/docs/ru/operations/system-tables/columns.md +++ b/docs/ru/operations/system-tables/columns.md @@ -6,19 +6,21 @@ Таблица `system.columns` содержит столбцы (тип столбца указан в скобках): -- `database` (String) — имя базы данных. -- `table` (String) — имя таблицы. -- `name` (String) — имя столбца. -- `type` (String) — тип столбца. -- `default_kind` (String) — тип выражения (`DEFAULT`, `MATERIALIZED`, `ALIAS`) значения по умолчанию, или пустая строка. -- `default_expression` (String) — выражение для значения по умолчанию или пустая строка. -- `data_compressed_bytes` (UInt64) — размер сжатых данных в байтах. -- `data_uncompressed_bytes` (UInt64) — размер распакованных данных в байтах. -- `marks_bytes` (UInt64) — размер засечек в байтах. -- `comment` (String) — комментарий к столбцу или пустая строка. -- `is_in_partition_key` (UInt8) — флаг, показывающий включение столбца в ключ партиционирования. -- `is_in_sorting_key` (UInt8) — флаг, показывающий включение столбца в ключ сортировки. -- `is_in_primary_key` (UInt8) — флаг, показывающий включение столбца в первичный ключ. -- `is_in_sampling_key` (UInt8) — флаг, показывающий включение столбца в ключ выборки. +- `database` ([String](../../sql-reference/data-types/string.md)) — имя базы данных. +- `table` ([String](../../sql-reference/data-types/string.md)) — имя таблицы. +- `name` ([String](../../sql-reference/data-types/string.md)) — имя столбца. +- `type` ([String](../../sql-reference/data-types/string.md)) — тип столбца. +- `position` ([UInt64](../../sql-reference/data-types/int-uint.md)) — порядковый номер столбца в таблице (нумерация начинается с 1). +- `default_kind` ([String](../../sql-reference/data-types/string.md)) — тип выражения (`DEFAULT`, `MATERIALIZED`, `ALIAS`) для значения по умолчанию или пустая строка. +- `default_expression` ([String](../../sql-reference/data-types/string.md)) — выражение для значения по умолчанию или пустая строка. +- `data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — размер сжатых данных в байтах. +- `data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — размер распакованных данных в байтах. +- `marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — размер засечек в байтах. +- `comment` ([String](../../sql-reference/data-types/string.md)) — комментарий к столбцу или пустая строка. +- `is_in_partition_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — флаг, показывающий включение столбца в ключ партиционирования. +- `is_in_sorting_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — флаг, показывающий включение столбца в ключ сортировки. +- `is_in_primary_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — флаг, показывающий включение столбца в первичный ключ. +- `is_in_sampling_key` ([UInt8](../../sql-reference/data-types/int-uint.md)) — флаг, показывающий включение столбца в ключ выборки. +- `compression_codec` ([String](../../sql-reference/data-types/string.md)) — имя кодека сжатия. [Оригинальная статья](https://clickhouse.tech/docs/ru/operations/system_tables/columns) diff --git a/docs/ru/operations/system-tables/crash-log.md b/docs/ru/operations/system-tables/crash-log.md new file mode 100644 index 00000000000..d2b3ae5c6f5 --- /dev/null +++ b/docs/ru/operations/system-tables/crash-log.md @@ -0,0 +1,48 @@ +# system.crash_log {#system-tables_crash_log} + +Содержит информацию о трассировках стека для фатальных ошибок. Таблица не содержится в базе данных по умолчанию, а создается только при возникновении фатальных ошибок. + +Колонки: + +- `event_date` ([Datetime](../../sql-reference/data-types/datetime.md)) — Дата события. +- `event_time` ([Datetime](../../sql-reference/data-types/datetime.md)) — Время события. +- `timestamp_ns` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Время события с наносекундами. +- `signal` ([Int32](../../sql-reference/data-types/int-uint.md)) — Номер сигнала, пришедшего в поток. +- `thread_id` ([UInt64](../../sql-reference/data-types/int-uint.md)) — Идентификатор треда. +- `query_id` ([String](../../sql-reference/data-types/string.md)) — Идентификатор запроса. +- `trace` ([Array](../../sql-reference/data-types/array.md)([UInt64](../../sql-reference/data-types/int-uint.md))) — Трассировка стека в момент ошибки. Представляет собой список физических адресов, по которым расположены вызываемые методы. +- `trace_full` ([Array](../../sql-reference/data-types/array.md)([String](../../sql-reference/data-types/string.md))) — Трассировка стека в момент ошибки. Содержит вызываемые методы. +- `version` ([String](../../sql-reference/data-types/string.md)) — Версия сервера ClickHouse. +- `revision` ([UInt32](../../sql-reference/data-types/int-uint.md)) — Ревизия сборки сервера ClickHouse. +- `build_id` ([String](../../sql-reference/data-types/string.md)) — BuildID, сгенерированный компилятором. + +**Пример** + +Запрос: + +``` sql +SELECT * FROM system.crash_log ORDER BY event_time DESC LIMIT 1; +``` + +Результат (приведён не полностью): + +``` text +Row 1: +────── +event_date: 2020-10-14 +event_time: 2020-10-14 15:47:40 +timestamp_ns: 1602679660271312710 +signal: 11 +thread_id: 23624 +query_id: 428aab7c-8f5c-44e9-9607-d16b44467e69 +trace: [188531193,...] +trace_full: ['3. DB::(anonymous namespace)::FunctionFormatReadableTimeDelta::executeImpl(std::__1::vector >&, std::__1::vector > const&, unsigned long, unsigned long) const @ 0xb3cc1f9 in /home/username/work/ClickHouse/build/programs/clickhouse',...] +version: ClickHouse 20.11.1.1 +revision: 54442 +build_id: +``` + +**См. также** +- Системная таблица [trace_log](../../operations/system-tables/trace_log.md) + +[Original article](https://clickhouse.tech/docs/en/operations/system-tables/crash-log) diff --git a/docs/ru/operations/system-tables/index.md b/docs/ru/operations/system-tables/index.md index 6fa989d3d0d..93ea1c92068 100644 --- a/docs/ru/operations/system-tables/index.md +++ b/docs/ru/operations/system-tables/index.md @@ -1,6 +1,6 @@ --- toc_priority: 52 -toc_title: Системные таблицы +toc_title: "\u0421\u0438\u0441\u0442\u0435\u043c\u043d\u044b\u0435\u0020\u0442\u0430\u0431\u043b\u0438\u0446\u044b" --- # Системные таблицы {#system-tables} diff --git a/docs/ru/operations/system-tables/metric_log.md b/docs/ru/operations/system-tables/metric_log.md index 32ebe91dd91..ad25e8d5cfd 100644 --- a/docs/ru/operations/system-tables/metric_log.md +++ b/docs/ru/operations/system-tables/metric_log.md @@ -1,6 +1,7 @@ # system.metric_log {#system_tables-metric_log} Содержит историю значений метрик из таблиц `system.metrics` и `system.events`, периодически сбрасываемую на диск. + Для включения сбора истории метрик в таблице `system.metric_log` создайте `/etc/clickhouse-server/config.d/metric_log.xml` следующего содержания: ``` xml @@ -14,6 +15,11 @@ ``` +Столбцы: +- `event_date` ([Date](../../sql-reference/data-types/date.md)) — дата события. +- `event_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время события. +- `event_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — время события в микросекундах. + **Пример** ``` sql diff --git a/docs/ru/operations/system-tables/parts_columns.md b/docs/ru/operations/system-tables/parts_columns.md new file mode 100644 index 00000000000..db4d453e8f1 --- /dev/null +++ b/docs/ru/operations/system-tables/parts_columns.md @@ -0,0 +1,148 @@ +# system.parts_columns {#system_tables-parts_columns} + +Содержит информацию о кусках данных и столбцах таблиц семейства [MergeTree](../../engines/table-engines/mergetree-family/mergetree.md). + +Каждая строка описывает один кусок данных. + +Столбцы: + +- `partition` ([String](../../sql-reference/data-types/string.md)) — имя партиции. Что такое партиция вы можете узнать из описания запроса [ALTER](../../sql-reference/statements/alter/index.md#query_language_queries_alter). + + Форматы: + + - `YYYYMM` для автоматической схемы партиционирования по месяцам. + - `any_string` при партиционировании вручную. + +- `name` ([String](../../sql-reference/data-types/string.md)) — имя куска данных. + +- `part_type` ([String](../../sql-reference/data-types/string.md)) — формат хранения данных. + + Возможные значения: + + - `Wide` — каждая колонка хранится в отдельном файле. + - `Compact` — все колонки хранятся в одном файле. + + Формат хранения данных определяется настройками `min_bytes_for_wide_part` и `min_rows_for_wide_part` таблицы [MergeTree](../../engines/table-engines/mergetree-family/mergetree.md). + +- `active` ([UInt8](../../sql-reference/data-types/int-uint.md)) — признак активности. Если кусок данных активен, то он используется таблицей, в противном случае он будет удален. Неактивные куски остаются после слияний. + +- `marks` ([UInt64](../../sql-reference/data-types/int-uint.md)) — количество засечек. Чтобы получить примерное количество строк в куске данных, умножьте `marks` на гранулированность индекса (обычно 8192). + +- `rows` ([UInt64](../../sql-reference/data-types/int-uint.md)) — количество строк. + +- `bytes_on_disk` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер всех файлов кусков данных в байтах. + +- `data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер сжатой информации в куске данных. Размер всех дополнительных файлов (например, файлов с засечками) не учитывается. + +- `data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер распакованной информации в куске данных. Размер всех дополнительных файлов (например, файлов с засечками) не учитывается. + +- `marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — размер файла с засечками. + +- `modification_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время модификации директории с куском данных. Обычно соответствует времени создания куска. + +- `remove_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время, когда кусок данных стал неактивным. + +- `refcount` ([UInt32](../../sql-reference/data-types/int-uint.md)) — количество мест, в котором кусок данных используется. Значение больше 2 говорит о том, что кусок участвует в запросах или в слияниях. + +- `min_date` ([Date](../../sql-reference/data-types/date.md)) — минимальное значение ключа даты в куске данных. + +- `max_date` ([Date](../../sql-reference/data-types/date.md)) — максимальное значение ключа даты в куске данных. + +- `partition_id` ([String](../../sql-reference/data-types/string.md)) — ID партиции. + +- `min_block_number` ([UInt64](../../sql-reference/data-types/int-uint.md)) — минимальное число кусков данных, из которых состоит текущий после слияния. + +- `max_block_number` ([UInt64](../../sql-reference/data-types/int-uint.md)) — максимальное число кусков данных, из которых состоит текущий после слияния. + +- `level` ([UInt32](../../sql-reference/data-types/int-uint.md)) — глубина дерева слияний. Если слияний не было, то `level=0`. + +- `data_version` ([UInt64](../../sql-reference/data-types/int-uint.md)) — число, которое используется для определения того, какие мутации необходимо применить к куску данных (мутации с версией большей, чем `data_version`). + +- `primary_key_bytes_in_memory` ([UInt64](../../sql-reference/data-types/int-uint.md)) — объём памяти в байтах, занимаемой значениями первичных ключей. + +- `primary_key_bytes_in_memory_allocated` ([UInt64](../../sql-reference/data-types/int-uint.md)) — объём памяти в байтах, выделенный для размещения первичных ключей. + +- `database` ([String](../../sql-reference/data-types/string.md)) — имя базы данных. + +- `table` ([String](../../sql-reference/data-types/string.md)) — имя таблицы. + +- `engine` ([String](../../sql-reference/data-types/string.md)) — имя движка таблицы, без параметров. + +- `disk_name` ([String](../../sql-reference/data-types/string.md)) — имя диска, на котором находится кусок данных. + +- `path` ([String](../../sql-reference/data-types/string.md)) — абсолютный путь к папке с файлами кусков данных. + +- `column` ([String](../../sql-reference/data-types/string.md)) — имя столбца. + +- `type` ([String](../../sql-reference/data-types/string.md)) — тип столбца. + +- `column_position` ([UInt64](../../sql-reference/data-types/int-uint.md)) — порядковый номер столбца (нумерация начинается с 1). + +- `default_kind` ([String](../../sql-reference/data-types/string.md)) — тип выражения (`DEFAULT`, `MATERIALIZED`, `ALIAS`) для значения по умолчанию или пустая строка. + +- `default_expression` ([String](../../sql-reference/data-types/string.md)) — выражение для значения по умолчанию или пустая строка. + +- `column_bytes_on_disk` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер столбца в байтах. + +- `column_data_compressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер сжатой информации в столбце в байтах. + +- `column_data_uncompressed_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — общий размер распакованной информации в столбце в байтах. + +- `column_marks_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — размер столбца с засечками в байтах. + +- `bytes` ([UInt64](../../sql-reference/data-types/int-uint.md)) — алиас для `bytes_on_disk`. + +- `marks_size` ([UInt64](../../sql-reference/data-types/int-uint.md)) — алиас для `marks_bytes`. + +**Пример** + +``` sql +SELECT * FROM system.parts_columns LIMIT 1 FORMAT Vertical; +``` + +``` text +Row 1: +────── +partition: tuple() +name: all_1_2_1 +part_type: Wide +active: 1 +marks: 2 +rows: 2 +bytes_on_disk: 155 +data_compressed_bytes: 56 +data_uncompressed_bytes: 4 +marks_bytes: 96 +modification_time: 2020-09-23 10:13:36 +remove_time: 2106-02-07 06:28:15 +refcount: 1 +min_date: 1970-01-01 +max_date: 1970-01-01 +partition_id: all +min_block_number: 1 +max_block_number: 2 +level: 1 +data_version: 1 +primary_key_bytes_in_memory: 2 +primary_key_bytes_in_memory_allocated: 64 +database: default +table: 53r93yleapyears +engine: MergeTree +disk_name: default +path: /var/lib/clickhouse/data/default/53r93yleapyears/all_1_2_1/ +column: id +type: Int8 +column_position: 1 +default_kind: +default_expression: +column_bytes_on_disk: 76 +column_data_compressed_bytes: 28 +column_data_uncompressed_bytes: 2 +column_marks_bytes: 48 +``` + +**Смотрите также** + +- [Движок MergeTree](../../engines/table-engines/mergetree-family/mergetree.md) + +[Оригинальная статья](https://clickhouse.tech/docs/en/operations/system_tables/parts_columns) diff --git a/docs/ru/operations/system-tables/query_log.md b/docs/ru/operations/system-tables/query_log.md index 340e77d85da..39f685288d8 100644 --- a/docs/ru/operations/system-tables/query_log.md +++ b/docs/ru/operations/system-tables/query_log.md @@ -33,10 +33,12 @@ ClickHouse не удаляет данные из таблица автомати - `'ExceptionWhileProcessing' = 4` — исключение во время обработки запроса. - `event_date` ([Date](../../sql-reference/data-types/date.md)) — дата начала запроса. - `event_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время начала запроса. +- `event_time_microseconds` ([DateTime](../../sql-reference/data-types/datetime.md)) — время начала запроса с точностью до микросекунд. - `query_start_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время начала обработки запроса. +- `query_start_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — время начала обработки запроса с точностью до микросекунд. - `query_duration_ms` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — длительность выполнения запроса в миллисекундах. -- `read_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Общее количество строк, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_rows` включает в себя общее количество строк, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_rows`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся. -- `read_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — Общее количество байтов, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_bytes` включает в себя общее количество байтов, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_bytes`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся. +- `read_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — общее количество строк, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_rows` включает в себя общее количество строк, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_rows`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся. +- `read_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — общее количество байтов, считанных из всех таблиц и табличных функций, участвующих в запросе. Включает в себя обычные подзапросы, подзапросы для `IN` и `JOIN`. Для распределенных запросов `read_bytes` включает в себя общее количество байтов, прочитанных на всех репликах. Каждая реплика передает собственное значение `read_bytes`, а сервер-инициатор запроса суммирует все полученные и локальные значения. Объемы кэша не учитываюся. - `written_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — количество записанных строк для запросов `INSERT`. Для других запросов, значение столбца 0. - `written_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — объём записанных данных в байтах для запросов `INSERT`. Для других запросов, значение столбца 0. - `result_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — количество строк в результате запроса `SELECT` или количество строк в запросе `INSERT`. @@ -75,64 +77,67 @@ ClickHouse не удаляет данные из таблица автомати - `quota_key` ([String](../../sql-reference/data-types/string.md)) — «ключ квоты» из настроек [квот](quotas.md) (см. `keyed`). - `revision` ([UInt32](../../sql-reference/data-types/int-uint.md)) — ревизия ClickHouse. - `thread_numbers` ([Array(UInt32)](../../sql-reference/data-types/array.md)) — количество потоков, участвующих в обработке запросов. -- `ProfileEvents.Names` ([Array(String)](../../sql-reference/data-types/array.md)) — Счетчики для изменения различных метрик. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(#system_tables-events +- `ProfileEvents.Names` ([Array(String)](../../sql-reference/data-types/array.md)) — счетчики для изменения различных метрик. Описание метрик можно получить из таблицы [system.events](#system_tables-events)(#system_tables-events - `ProfileEvents.Values` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — метрики, перечисленные в столбце `ProfileEvents.Names`. - `Settings.Names` ([Array(String)](../../sql-reference/data-types/array.md)) — имена настроек, которые меняются, когда клиент выполняет запрос. Чтобы разрешить логирование изменений настроек, установите параметр `log_query_settings` равным 1. -- `Settings.Values` ([Array(String)](../../sql-reference/data-types/array.md)) — Значения настроек, которые перечислены в столбце `Settings.Names`. +- `Settings.Values` ([Array(String)](../../sql-reference/data-types/array.md)) — значения настроек, которые перечислены в столбце `Settings.Names`. **Пример** ``` sql -SELECT * FROM system.query_log LIMIT 1 FORMAT Vertical; +SELECT * FROM system.query_log LIMIT 1 \G ``` ``` text Row 1: ────── -type: QueryStart -event_date: 2020-05-13 -event_time: 2020-05-13 14:02:28 -query_start_time: 2020-05-13 14:02:28 -query_duration_ms: 0 -read_rows: 0 -read_bytes: 0 -written_rows: 0 -written_bytes: 0 -result_rows: 0 -result_bytes: 0 -memory_usage: 0 -query: SELECT 1 -exception_code: 0 -exception: -stack_trace: -is_initial_query: 1 -user: default -query_id: 5e834082-6f6d-4e34-b47b-cd1934f4002a -address: ::ffff:127.0.0.1 -port: 57720 -initial_user: default -initial_query_id: 5e834082-6f6d-4e34-b47b-cd1934f4002a -initial_address: ::ffff:127.0.0.1 -initial_port: 57720 -interface: 1 -os_user: bayonet -client_hostname: clickhouse.ru-central1.internal -client_name: ClickHouse client -client_revision: 54434 -client_version_major: 20 -client_version_minor: 4 -client_version_patch: 1 -http_method: 0 -http_user_agent: -quota_key: -revision: 54434 -thread_ids: [] -ProfileEvents.Names: [] -ProfileEvents.Values: [] -Settings.Names: ['use_uncompressed_cache','load_balancing','log_queries','max_memory_usage'] -Settings.Values: ['0','random','1','10000000000'] - +type: QueryStart +event_date: 2020-09-11 +event_time: 2020-09-11 10:08:17 +event_time_microseconds: 2020-09-11 10:08:17.063321 +query_start_time: 2020-09-11 10:08:17 +query_start_time_microseconds: 2020-09-11 10:08:17.063321 +query_duration_ms: 0 +read_rows: 0 +read_bytes: 0 +written_rows: 0 +written_bytes: 0 +result_rows: 0 +result_bytes: 0 +memory_usage: 0 +current_database: default +query: INSERT INTO test1 VALUES +exception_code: 0 +exception: +stack_trace: +is_initial_query: 1 +user: default +query_id: 50a320fd-85a8-49b8-8761-98a86bcbacef +address: ::ffff:127.0.0.1 +port: 33452 +initial_user: default +initial_query_id: 50a320fd-85a8-49b8-8761-98a86bcbacef +initial_address: ::ffff:127.0.0.1 +initial_port: 33452 +interface: 1 +os_user: bharatnc +client_hostname: tower +client_name: ClickHouse +client_revision: 54437 +client_version_major: 20 +client_version_minor: 7 +client_version_patch: 2 +http_method: 0 +http_user_agent: +quota_key: +revision: 54440 +thread_ids: [] +ProfileEvents.Names: [] +ProfileEvents.Values: [] +Settings.Names: ['use_uncompressed_cache','load_balancing','log_queries','max_memory_usage','allow_introspection_functions'] +Settings.Values: ['0','random','1','10000000000','1'] ``` + **Смотрите также** - [system.query_thread_log](../../operations/system-tables/query_thread_log.md#system_tables-query_thread_log) — в этой таблице содержится информация о цепочке каждого выполненного запроса. diff --git a/docs/ru/operations/system-tables/query_thread_log.md b/docs/ru/operations/system-tables/query_thread_log.md index acee3063323..052baf98035 100644 --- a/docs/ru/operations/system-tables/query_thread_log.md +++ b/docs/ru/operations/system-tables/query_thread_log.md @@ -15,7 +15,9 @@ ClickHouse не удаляет данные из таблицы автомати - `event_date` ([Date](../../sql-reference/data-types/date.md)) — дата завершения выполнения запроса потоком. - `event_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — дата и время завершения выполнения запроса потоком. +- `event_time_microseconds` ([DateTime](../../sql-reference/data-types/datetime.md)) — дата и время завершения выполнения запроса потоком с точностью до микросекунд. - `query_start_time` ([DateTime](../../sql-reference/data-types/datetime.md)) — время начала обработки запроса. +- `query_start_time_microseconds` ([DateTime64](../../sql-reference/data-types/datetime64.md)) — время начала обработки запроса с точностью до микросекунд. - `query_duration_ms` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — длительность обработки запроса в миллисекундах. - `read_rows` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — количество прочитанных строк. - `read_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — количество прочитанных байтов. @@ -23,7 +25,7 @@ ClickHouse не удаляет данные из таблицы автомати - `written_bytes` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — объём записанных данных в байтах для запросов `INSERT`. Для других запросов, значение столбца 0. - `memory_usage` ([Int64](../../sql-reference/data-types/int-uint.md)) — разница между выделенной и освобождённой памятью в контексте потока. - `peak_memory_usage` ([Int64](../../sql-reference/data-types/int-uint.md)) — максимальная разница между выделенной и освобождённой памятью в контексте потока. -- `thread_name` ([String](../../sql-reference/data-types/string.md)) — Имя потока. +- `thread_name` ([String](../../sql-reference/data-types/string.md)) — имя потока. - `thread_id` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — tid (ID потока операционной системы). - `master_thread_id` ([UInt64](../../sql-reference/data-types/int-uint.md#uint-ranges)) — tid (ID потока операционной системы) главного потока. - `query` ([String](../../sql-reference/data-types/string.md)) — текст запроса. @@ -55,56 +57,57 @@ ClickHouse не удаляет данные из таблицы автомати - `http_user_agent` ([String](../../sql-reference/data-types/string.md)) — HTTP заголовок `UserAgent`. - `quota_key` ([String](../../sql-reference/data-types/string.md)) — «ключ квоты» из настроек [квот](quotas.md) (см. `keyed`). - `revision` ([UInt32](../../sql-reference/data-types/int-uint.md)) — ревизия ClickHouse. -- `ProfileEvents.Names` ([Array(String)](../../sql-reference/data-types/array.md)) — Счетчики для изменения различных метрик для данного потока. Описание метрик можно получить из таблицы [system.events](#system_tables-events). +- `ProfileEvents.Names` ([Array(String)](../../sql-reference/data-types/array.md)) — счетчики для изменения различных метрик для данного потока. Описание метрик можно получить из таблицы [system.events](#system_tables-events). - `ProfileEvents.Values` ([Array(UInt64)](../../sql-reference/data-types/array.md)) — метрики для данного потока, перечисленные в столбце `ProfileEvents.Names`. **Пример** ``` sql - SELECT * FROM system.query_thread_log LIMIT 1 FORMAT Vertical + SELECT * FROM system.query_thread_log LIMIT 1 \G ``` ``` text Row 1: ────── -event_date: 2020-05-13 -event_time: 2020-05-13 14:02:28 -query_start_time: 2020-05-13 14:02:28 -query_duration_ms: 0 -read_rows: 1 -read_bytes: 1 -written_rows: 0 -written_bytes: 0 -memory_usage: 0 -peak_memory_usage: 0 -thread_name: QueryPipelineEx -thread_id: 28952 -master_thread_id: 28924 -query: SELECT 1 -is_initial_query: 1 -user: default -query_id: 5e834082-6f6d-4e34-b47b-cd1934f4002a -address: ::ffff:127.0.0.1 -port: 57720 -initial_user: default -initial_query_id: 5e834082-6f6d-4e34-b47b-cd1934f4002a -initial_address: ::ffff:127.0.0.1 -initial_port: 57720 -interface: 1 -os_user: bayonet -client_hostname: clickhouse.ru-central1.internal -client_name: ClickHouse client -client_revision: 54434 -client_version_major: 20 -client_version_minor: 4 -client_version_patch: 1 -http_method: 0 -http_user_agent: -quota_key: -revision: 54434 -ProfileEvents.Names: ['ContextLock','RealTimeMicroseconds','UserTimeMicroseconds','OSCPUWaitMicroseconds','OSCPUVirtualTimeMicroseconds'] -ProfileEvents.Values: [1,97,81,5,81] -... +event_date: 2020-09-11 +event_time: 2020-09-11 10:08:17 +event_time_microseconds: 2020-09-11 10:08:17.134042 +query_start_time: 2020-09-11 10:08:17 +query_start_time_microseconds: 2020-09-11 10:08:17.063150 +query_duration_ms: 70 +read_rows: 0 +read_bytes: 0 +written_rows: 1 +written_bytes: 12 +memory_usage: 4300844 +peak_memory_usage: 4300844 +thread_name: TCPHandler +thread_id: 638133 +master_thread_id: 638133 +query: INSERT INTO test1 VALUES +is_initial_query: 1 +user: default +query_id: 50a320fd-85a8-49b8-8761-98a86bcbacef +address: ::ffff:127.0.0.1 +port: 33452 +initial_user: default +initial_query_id: 50a320fd-85a8-49b8-8761-98a86bcbacef +initial_address: ::ffff:127.0.0.1 +initial_port: 33452 +interface: 1 +os_user: bharatnc +client_hostname: tower +client_name: ClickHouse +client_revision: 54437 +client_version_major: 20 +client_version_minor: 7 +client_version_patch: 2 +http_method: 0 +http_user_agent: +quota_key: +revision: 54440 +ProfileEvents.Names: ['Query','InsertQuery','FileOpen','WriteBufferFromFileDescriptorWrite','WriteBufferFromFileDescriptorWriteBytes','ReadCompressedBytes','CompressedReadBufferBlocks','CompressedReadBufferBytes','IOBufferAllocs','IOBufferAllocBytes','FunctionExecute','CreatedWriteBufferOrdinary','DiskWriteElapsedMicroseconds','NetworkReceiveElapsedMicroseconds','NetworkSendElapsedMicroseconds','InsertedRows','InsertedBytes','SelectedRows','SelectedBytes','MergeTreeDataWriterRows','MergeTreeDataWriterUncompressedBytes','MergeTreeDataWriterCompressedBytes','MergeTreeDataWriterBlocks','MergeTreeDataWriterBlocksAlreadySorted','ContextLock','RWLockAcquiredReadLocks','RealTimeMicroseconds','UserTimeMicroseconds','SoftPageFaults','OSCPUVirtualTimeMicroseconds','OSWriteBytes','OSReadChars','OSWriteChars'] +ProfileEvents.Values: [1,1,11,11,591,148,3,71,29,6533808,1,11,72,18,47,1,12,1,12,1,12,189,1,1,10,2,70853,2748,49,2747,45056,422,1520] ``` **Смотрите также** diff --git a/docs/ru/operations/system-tables/replicas.md b/docs/ru/operations/system-tables/replicas.md index bb2e6a550a0..8d4eb60c56a 100644 --- a/docs/ru/operations/system-tables/replicas.md +++ b/docs/ru/operations/system-tables/replicas.md @@ -53,9 +53,9 @@ active_replicas: 2 - `table` (`String`) - имя таблицы. - `engine` (`String`) - имя движка таблицы. - `is_leader` (`UInt8`) - является ли реплика лидером. - В один момент времени, не более одной из реплик является лидером. Лидер отвечает за выбор фоновых слияний, которые следует произвести. + Несколько реплик могут быть лидерами одновременно. Реплике можно запретить быть лидером с помощью `merge_tree` настройки `replicated_can_become_leader`. Лидеры назначают фоновые слияния, которые следует произвести. Замечу, что запись можно осуществлять на любую реплику (доступную и имеющую сессию в ZK), независимо от лидерства. -- `can_become_leader` (`UInt8`) - может ли реплика быть выбрана лидером. +- `can_become_leader` (`UInt8`) - может ли реплика быть лидером. - `is_readonly` (`UInt8`) - находится ли реплика в режиме «только для чтения» Этот режим включается, если в конфиге нет секции с ZK; если при переинициализации сессии в ZK произошла неизвестная ошибка; во время переинициализации сессии с ZK. - `is_session_expired` (`UInt8`) - истекла ли сессия с ZK. В основном, то же самое, что и `is_readonly`. diff --git a/docs/ru/operations/system-tables/text_log.md b/docs/ru/operations/system-tables/text_log.md index 01e34c914e6..141c3680c07 100644 --- a/docs/ru/operations/system-tables/text_log.md +++ b/docs/ru/operations/system-tables/text_log.md @@ -4,12 +4,13 @@ Столбцы: -- `event_date` (Date) — Дата создания записи. -- `event_time` (DateTime) — Время создания записи. -- `microseconds` (UInt32) — Время создания записи в микросекундах. -- `thread_name` (String) — Название потока, из которого была сделана запись. -- `thread_id` (UInt64) — Идентификатор потока ОС. -- `level` (Enum8) — Уровень логирования записи. Возможные значения: +- `event_date` (Date) — дата создания записи. +- `event_time` (DateTime) — время создания записи. +- `event_time_microseconds` (DateTime) — время создания записи с точностью до микросекунд. +- `microseconds` (UInt32) — время создания записи в микросекундах. +- `thread_name` (String) — название потока, из которого была сделана запись. +- `thread_id` (UInt64) — идентификатор потока ОС. +- `level` (Enum8) — уровень логирования записи. Возможные значения: - `1` или `'Fatal'`. - `2` или `'Critical'`. - `3` или `'Error'`. @@ -18,11 +19,35 @@ - `6` или `'Information'`. - `7` или `'Debug'`. - `8` или `'Trace'`. -- `query_id` (String) — Идентификатор запроса. -- `logger_name` (LowCardinality(String)) — Название логгера (`DDLWorker`). -- `message` (String) — Само тело записи. -- `revision` (UInt32) — Ревизия ClickHouse. -- `source_file` (LowCardinality(String)) — Исходный файл, из которого была сделана запись. -- `source_line` (UInt64) — Исходная строка, из которой была сделана запись. +- `query_id` (String) — идентификатор запроса. +- `logger_name` (LowCardinality(String)) — название логгера (`DDLWorker`). +- `message` (String) — само тело записи. +- `revision` (UInt32) — ревизия ClickHouse. +- `source_file` (LowCardinality(String)) — исходный файл, из которого была сделана запись. +- `source_line` (UInt64) — исходная строка, из которой была сделана запись. + +**Пример** + +``` sql +SELECT * FROM system.text_log LIMIT 1 \G +``` + +``` text +Row 1: +────── +event_date: 2020-09-10 +event_time: 2020-09-10 11:23:07 +event_time_microseconds: 2020-09-10 11:23:07.871397 +microseconds: 871397 +thread_name: clickhouse-serv +thread_id: 564917 +level: Information +query_id: +logger_name: DNSCacheUpdater +message: Update period 15 seconds +revision: 54440 +source_file: /ClickHouse/src/Interpreters/DNSCacheUpdater.cpp; void DB::DNSCacheUpdater::start() +source_line: 45 +``` [Оригинальная статья](https://clickhouse.tech/docs/ru/operations/system_tables/text_log) diff --git a/docs/ru/operations/system-tables/trace_log.md b/docs/ru/operations/system-tables/trace_log.md index 8bab8ff646c..3f0a16199d5 100644 --- a/docs/ru/operations/system-tables/trace_log.md +++ b/docs/ru/operations/system-tables/trace_log.md @@ -6,26 +6,28 @@ ClickHouse создает эту таблицу когда утсановлен Для анализа stack traces, используйте функции интроспекции `addressToLine`, `addressToSymbol` и `demangle`. -Колонки: +Столбцы: -- `event_date`([Date](../../sql-reference/data-types/date.md)) — Дата в момент снятия экземпляра стэка адресов вызова. +- `event_date`([Date](../../sql-reference/data-types/date.md)) — дата в момент снятия экземпляра стэка адресов вызова. -- `event_time`([DateTime](../../sql-reference/data-types/datetime.md)) — Дата и время в момент снятия экземпляра стэка адресов вызова. +- `event_time`([DateTime](../../sql-reference/data-types/datetime.md)) — дата и время в момент снятия экземпляра стэка адресов вызова. + +- `event_time_microseconds` ([DateTime](../../sql-reference/data-types/datetime.md)) — дата и время в момент снятия экземпляра стэка адресов вызова с точностью до микросекунд. - `revision`([UInt32](../../sql-reference/data-types/int-uint.md)) — ревизия сборки сервера ClickHouse. Во время соединения с сервером через `clickhouse-client`, вы видите строку похожую на `Connected to ClickHouse server version 19.18.1 revision 54429.`. Это поле содержит номер после `revision`, но не содержит строку после `version`. -- `timer_type`([Enum8](../../sql-reference/data-types/enum.md)) — Тип таймера: +- `timer_type`([Enum8](../../sql-reference/data-types/enum.md)) — тип таймера: - `Real` означает wall-clock время. - `CPU` означает относительное CPU время. -- `thread_number`([UInt32](../../sql-reference/data-types/int-uint.md)) — Идентификатор треда. +- `thread_number`([UInt32](../../sql-reference/data-types/int-uint.md)) — идентификатор треда. -- `query_id`([String](../../sql-reference/data-types/string.md)) — Идентификатор запроса который может быть использован для получения деталей о запросе из таблицы [query_log](query_log.md#system_tables-query_log) system table. +- `query_id`([String](../../sql-reference/data-types/string.md)) — идентификатор запроса который может быть использован для получения деталей о запросе из таблицы [query_log](query_log.md#system_tables-query_log) system table. -- `trace`([Array(UInt64)](../../sql-reference/data-types/array.md)) — Трассировка стека адресов вызова в момент семплирования. Каждый элемент массива это адрес виртуальной памяти внутри процесса сервера ClickHouse. +- `trace`([Array(UInt64)](../../sql-reference/data-types/array.md)) — трассировка стека адресов вызова в момент семплирования. Каждый элемент массива — это адрес виртуальной памяти внутри процесса сервера ClickHouse. **Пример** @@ -36,13 +38,16 @@ SELECT * FROM system.trace_log LIMIT 1 \G ``` text Row 1: ────── -event_date: 2019-11-15 -event_time: 2019-11-15 15:09:38 -revision: 54428 -timer_type: Real -thread_number: 48 -query_id: acc4d61f-5bd1-4a3e-bc91-2180be37c915 -trace: [94222141367858,94222152240175,94222152325351,94222152329944,94222152330796,94222151449980,94222144088167,94222151682763,94222144088167,94222151682763,94222144088167,94222144058283,94222144059248,94222091840750,94222091842302,94222091831228,94222189631488,140509950166747,140509942945935] +event_date: 2020-09-10 +event_time: 2020-09-10 11:23:09 +event_time_microseconds: 2020-09-10 11:23:09.872924 +timestamp_ns: 1599762189872924510 +revision: 54440 +trace_type: Memory +thread_id: 564963 +query_id: +trace: [371912858,371912789,371798468,371799717,371801313,371790250,624462773,566365041,566440261,566445834,566460071,566459914,566459842,566459580,566459469,566459389,566459341,566455774,371993941,371988245,372158848,372187428,372187309,372187093,372185478,140222123165193,140222122205443] +size: 5244400 ``` [Оригинальная статья](https://clickhouse.tech/docs/ru/operations/system_tables/trace_log) diff --git a/docs/ru/operations/tips.md b/docs/ru/operations/tips.md index f076fab1311..40035309c03 100644 --- a/docs/ru/operations/tips.md +++ b/docs/ru/operations/tips.md @@ -1,3 +1,8 @@ +--- +toc_priority: 58 +toc_title: "\u0421\u043e\u0432\u0435\u0442\u044b\u0020\u043f\u043e\u0020\u044d\u043a\u0441\u043f\u043b\u0443\u0430\u0442\u0430\u0446\u0438\u0438" +--- + # Советы по эксплуатации {#sovety-po-ekspluatatsii} ## CPU Scaling Governor {#cpu-scaling-governor} diff --git a/docs/ru/operations/troubleshooting.md b/docs/ru/operations/troubleshooting.md index cb86dfef5e8..3df2a1dd46c 100644 --- a/docs/ru/operations/troubleshooting.md +++ b/docs/ru/operations/troubleshooting.md @@ -1,3 +1,8 @@ +--- +toc_priority: 46 +toc_title: "\u0423\u0441\u0442\u0440\u0430\u043d\u0435\u043d\u0438\u0435\u0020\u043d\u0435\u0438\u0441\u043f\u0440\u0430\u0432\u043d\u043e\u0441\u0442\u0435\u0439" +--- + # Устранение неисправностей {#ustranenie-neispravnostei} - [Установка дистрибутива](#troubleshooting-installation-errors) diff --git a/docs/ru/operations/update.md b/docs/ru/operations/update.md index f1998864f40..c74b28b3fd7 100644 --- a/docs/ru/operations/update.md +++ b/docs/ru/operations/update.md @@ -1,3 +1,8 @@ +--- +toc_priority: 47 +toc_title: "\u041e\u0431\u043d\u043e\u0432\u043b\u0435\u043d\u0438\u0435\u0020\u0043\u006c\u0069\u0063\u006b\u0048\u006f\u0075\u0073\u0065" +--- + # Обновление ClickHouse {#obnovlenie-clickhouse} Если ClickHouse установлен с помощью deb-пакетов, выполните следующие команды на сервере: diff --git a/docs/ru/operations/utilities/clickhouse-copier.md b/docs/ru/operations/utilities/clickhouse-copier.md index b43f5ccaf7a..64e3c1eee12 100644 --- a/docs/ru/operations/utilities/clickhouse-copier.md +++ b/docs/ru/operations/utilities/clickhouse-copier.md @@ -1,3 +1,8 @@ +--- +toc_priority: 59 +toc_title: clickhouse-copier +--- + # clickhouse-copier {#clickhouse-copier} Копирует данные из таблиц одного кластера в таблицы другого (или этого же) кластера. diff --git a/docs/ru/operations/utilities/clickhouse-local.md b/docs/ru/operations/utilities/clickhouse-local.md index 7dfa9587686..2b5c9b119e2 100644 --- a/docs/ru/operations/utilities/clickhouse-local.md +++ b/docs/ru/operations/utilities/clickhouse-local.md @@ -1,3 +1,8 @@ +--- +toc_priority: 60 +toc_title: clickhouse-local +--- + # clickhouse-local {#clickhouse-local} Принимает на вход данные, которые можно представить в табличном виде и выполняет над ними операции, заданные на [языке запросов](../../operations/utilities/clickhouse-local.md) ClickHouse. @@ -9,6 +14,8 @@ !!! warning "Warning" Мы не рекомендуем подключать серверную конфигурацию к `clickhouse-local`, поскольку данные можно легко повредить неосторожными действиями. +Для временных данных по умолчанию создается специальный каталог. Если вы хотите обойти это действие, каталог данных можно указать с помощью опции `-- --path`. + ## Вызов программы {#vyzov-programmy} Основной формат вызова: @@ -34,25 +41,51 @@ $ clickhouse-local --structure "table_structure" --input-format "format_of_incom ## Примеры вызова {#primery-vyzova} ``` bash -$ echo -e "1,2\n3,4" | clickhouse-local -S "a Int64, b Int64" -if "CSV" -q "SELECT * FROM table" +$ echo -e "1,2\n3,4" | clickhouse-local --structure "a Int64, b Int64" \ + --input-format "CSV" --query "SELECT * FROM table" Read 2 rows, 32.00 B in 0.000 sec., 5182 rows/sec., 80.97 KiB/sec. -1 2 -3 4 +1 2 +3 4 ``` Вызов выше эквивалентен следующему: ``` bash -$ echo -e "1,2\n3,4" | clickhouse-local -q "CREATE TABLE table (a Int64, b Int64) ENGINE = File(CSV, stdin); SELECT a, b FROM table; DROP TABLE table" +$ echo -e "1,2\n3,4" | clickhouse-local --query " + CREATE TABLE table (a Int64, b Int64) ENGINE = File(CSV, stdin); + SELECT a, b FROM table; + DROP TABLE table" Read 2 rows, 32.00 B in 0.000 sec., 4987 rows/sec., 77.93 KiB/sec. -1 2 -3 4 +1 2 +3 4 +``` + + +Необязательно использовать ключи `stdin` или `--file`. Вы можете открывать любое количество файлов с помощью [табличной функции `file`](../../sql-reference/table-functions/file.md): + +``` bash +$ echo 1 | tee 1.tsv +1 + +$ echo 2 | tee 2.tsv +2 + +$ clickhouse-local --query " + select * from file('1.tsv', TSV, 'a int') t1 + cross join file('2.tsv', TSV, 'b int') t2" +1 2 ``` А теперь давайте выведем на экран объём оперативной памяти, занимаемой пользователями (Unix): ``` bash -$ ps aux | tail -n +2 | awk '{ printf("%s\t%s\n", $1, $4) }' | clickhouse-local -S "user String, mem Float64" -q "SELECT user, round(sum(mem), 2) as memTotal FROM table GROUP BY user ORDER BY memTotal DESC FORMAT Pretty" +$ ps aux | tail -n +2 | awk '{ printf("%s\t%s\n", $1, $4) }' \ + | clickhouse-local --structure "user String, mem Float64" \ + --query "SELECT user, round(sum(mem), 2) as memTotal + FROM table GROUP BY user ORDER BY memTotal DESC FORMAT Pretty" +``` + +``` text Read 186 rows, 4.15 KiB in 0.035 sec., 5302 rows/sec., 118.34 KiB/sec. ┏━━━━━━━━━━┳━━━━━━━━━━┓ ┃ user ┃ memTotal ┃ diff --git a/docs/ru/operations/utilities/index.md b/docs/ru/operations/utilities/index.md index 91ee649ee3a..5b55ebd798d 100644 --- a/docs/ru/operations/utilities/index.md +++ b/docs/ru/operations/utilities/index.md @@ -1,3 +1,9 @@ +--- +toc_folder_title: "\u0423\u0442\u0438\u043b\u0438\u0442\u044b" +toc_priority: 56 +toc_title: "\u041e\u0431\u0437\u043e\u0440" +--- + # Утилиты ClickHouse {#utility-clickhouse} - [clickhouse-local](clickhouse-local.md) diff --git a/docs/ru/sql-reference/aggregate-functions/combinators.md b/docs/ru/sql-reference/aggregate-functions/combinators.md index ca65f6ac615..ef014906423 100644 --- a/docs/ru/sql-reference/aggregate-functions/combinators.md +++ b/docs/ru/sql-reference/aggregate-functions/combinators.md @@ -1,3 +1,9 @@ +--- +toc_priority: 37 +toc_title: "\u041a\u043e\u043c\u0431\u0438\u043d\u0430\u0442\u043e\u0440\u044b\u0020\u0430\u0433\u0440\u0435\u0433\u0430\u0442\u043d\u044b\u0445\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0439" +--- + + # Комбинаторы агрегатных функций {#aggregate_functions_combinators} К имени агрегатной функции может быть приписан некоторый суффикс. При этом, работа агрегатной функции некоторым образом модифицируется. diff --git a/docs/ru/sql-reference/aggregate-functions/parametric-functions.md b/docs/ru/sql-reference/aggregate-functions/parametric-functions.md index 7cc0bdef581..f20acaa45c3 100644 --- a/docs/ru/sql-reference/aggregate-functions/parametric-functions.md +++ b/docs/ru/sql-reference/aggregate-functions/parametric-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: "\u041f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u0438\u0447\u0435\u0441\u043a\u0438\u0435\u0020\u0430\u0433\u0440\u0435\u0433\u0430\u0442\u043d\u044b\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Параметрические агрегатные функции {#aggregate_functions_parametric} Некоторые агрегатные функции могут принимать не только столбцы-аргументы (по которым производится свёртка), но и набор параметров - констант для инициализации. Синтаксис - две пары круглых скобок вместо одной. Первая - для параметров, вторая - для аргументов. diff --git a/docs/ru/sql-reference/data-types/boolean.md b/docs/ru/sql-reference/data-types/boolean.md index f868ebf6d14..bb0cd50c739 100644 --- a/docs/ru/sql-reference/data-types/boolean.md +++ b/docs/ru/sql-reference/data-types/boolean.md @@ -1,3 +1,8 @@ +--- +toc_priority: 43 +toc_title: "\u0411\u0443\u043b\u0435\u0432\u044b\u0020\u0437\u043d\u0430\u0447\u0435\u043d\u0438\u044f" +--- + # Булевы значения {#bulevy-znacheniia} Отдельного типа для булевых значений нет. Для них используется тип UInt8, в котором используются только значения 0 и 1. diff --git a/docs/ru/sql-reference/data-types/date.md b/docs/ru/sql-reference/data-types/date.md index 97f58ec30f6..9bcae2c1d72 100644 --- a/docs/ru/sql-reference/data-types/date.md +++ b/docs/ru/sql-reference/data-types/date.md @@ -1,3 +1,8 @@ +--- +toc_priority: 47 +toc_title: Date +--- + # Date {#data-type-date} Дата. Хранится в двух байтах в виде (беззнакового) числа дней, прошедших от 1970-01-01. Позволяет хранить значения от чуть больше, чем начала unix-эпохи до верхнего порога, определяющегося константой на этапе компиляции (сейчас - до 2106 года, последний полностью поддерживаемый год - 2105). diff --git a/docs/ru/sql-reference/data-types/datetime.md b/docs/ru/sql-reference/data-types/datetime.md index d60e81c74d5..87c5da68f35 100644 --- a/docs/ru/sql-reference/data-types/datetime.md +++ b/docs/ru/sql-reference/data-types/datetime.md @@ -1,3 +1,8 @@ +--- +toc_priority: 48 +toc_title: DateTime +--- + # DateTime {#data_type-datetime} Позволяет хранить момент времени, который может быть представлен как календарная дата и время. diff --git a/docs/ru/sql-reference/data-types/datetime64.md b/docs/ru/sql-reference/data-types/datetime64.md index bc9394276ee..0a602e44636 100644 --- a/docs/ru/sql-reference/data-types/datetime64.md +++ b/docs/ru/sql-reference/data-types/datetime64.md @@ -1,3 +1,8 @@ +--- +toc_priority: 49 +toc_title: DateTime64 +--- + # DateTime64 {#data_type-datetime64} Позволяет хранить момент времени, который может быть представлен как календарная дата и время, с заданной суб-секундной точностью. diff --git a/docs/ru/sql-reference/data-types/decimal.md b/docs/ru/sql-reference/data-types/decimal.md index 29192cb8819..bdcd3c767b9 100644 --- a/docs/ru/sql-reference/data-types/decimal.md +++ b/docs/ru/sql-reference/data-types/decimal.md @@ -1,3 +1,8 @@ +--- +toc_priority: 42 +toc_title: Decimal +--- + # Decimal(P, S), Decimal32(S), Decimal64(S), Decimal128(S), Decimal256(S) {#decimalp-s-decimal32s-decimal64s-decimal128s} Знаковые дробные числа с сохранением точности операций сложения, умножения и вычитания. Для деления осуществляется отбрасывание (не округление) знаков, не попадающих в младший десятичный разряд. diff --git a/docs/ru/sql-reference/data-types/domains/index.md b/docs/ru/sql-reference/data-types/domains/index.md index fe5c7ab7349..4449469b1bc 100644 --- a/docs/ru/sql-reference/data-types/domains/index.md +++ b/docs/ru/sql-reference/data-types/domains/index.md @@ -1,6 +1,6 @@ --- -toc_folder_title: Домены -toc_title_title: Обзор +toc_folder_title: "\u0414\u043e\u043c\u0435\u043d\u044b" +toc_title_title: "\u041e\u0431\u0437\u043e\u0440" toc_priority: 56 --- diff --git a/docs/ru/sql-reference/data-types/domains/ipv4.md b/docs/ru/sql-reference/data-types/domains/ipv4.md index 68b67bcca60..57d6f12ab17 100644 --- a/docs/ru/sql-reference/data-types/domains/ipv4.md +++ b/docs/ru/sql-reference/data-types/domains/ipv4.md @@ -1,3 +1,8 @@ +--- +toc_priority: 59 +toc_title: IPv4 +--- + ## IPv4 {#ipv4} `IPv4` — это домен, базирующийся на типе данных `UInt32` предназначенный для хранения адресов IPv4. Он обеспечивает компактное хранение данных с удобным для человека форматом ввода-вывода, и явно отображаемым типом данных в структуре таблицы. diff --git a/docs/ru/sql-reference/data-types/domains/ipv6.md b/docs/ru/sql-reference/data-types/domains/ipv6.md index c88ee74adea..04c5fd0d491 100644 --- a/docs/ru/sql-reference/data-types/domains/ipv6.md +++ b/docs/ru/sql-reference/data-types/domains/ipv6.md @@ -1,3 +1,8 @@ +--- +toc_priority: 60 +toc_title: IPv6 +--- + ## IPv6 {#ipv6} `IPv6` — это домен, базирующийся на типе данных `FixedString(16)`, предназначенный для хранения адресов IPv6. Он обеспечивает компактное хранение данных с удобным для человека форматом ввода-вывода, и явно отображаемым типом данных в структуре таблицы. diff --git a/docs/ru/sql-reference/data-types/enum.md b/docs/ru/sql-reference/data-types/enum.md index 792c82f7410..b86d15c19a8 100644 --- a/docs/ru/sql-reference/data-types/enum.md +++ b/docs/ru/sql-reference/data-types/enum.md @@ -1,3 +1,8 @@ +--- +toc_priority: 50 +toc_title: Enum +--- + # Enum {#enum} Перечисляемый тип данных, содержащий именованные значения. diff --git a/docs/ru/sql-reference/data-types/fixedstring.md b/docs/ru/sql-reference/data-types/fixedstring.md index 4a26b2f76af..21115418e30 100644 --- a/docs/ru/sql-reference/data-types/fixedstring.md +++ b/docs/ru/sql-reference/data-types/fixedstring.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: FixedString(N) +--- + # FixedString {#fixedstring} Строка фиксированной длины `N` байт (не символов, не кодовых точек). diff --git a/docs/ru/sql-reference/data-types/float.md b/docs/ru/sql-reference/data-types/float.md index 91d4b655e2a..f2e85f35041 100644 --- a/docs/ru/sql-reference/data-types/float.md +++ b/docs/ru/sql-reference/data-types/float.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: Float32, Float64 +--- + # Float32, Float64 {#float32-float64} [Числа с плавающей запятой](https://en.wikipedia.org/wiki/IEEE_754). diff --git a/docs/ru/sql-reference/data-types/int-uint.md b/docs/ru/sql-reference/data-types/int-uint.md index 3a33c95e4c3..c45c639aace 100644 --- a/docs/ru/sql-reference/data-types/int-uint.md +++ b/docs/ru/sql-reference/data-types/int-uint.md @@ -1,3 +1,8 @@ +--- +toc_priority: 40 +toc_title: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64 +--- + # UInt8, UInt16, UInt32, UInt64, UInt256, Int8, Int16, Int32, Int64, Int128, Int256 {#uint8-uint16-uint32-uint64-int8-int16-int32-int64} Целые числа фиксированной длины, без знака или со знаком. diff --git a/docs/ru/sql-reference/data-types/nested-data-structures/index.md b/docs/ru/sql-reference/data-types/nested-data-structures/index.md index 73d9019c96a..d53cabc6652 100644 --- a/docs/ru/sql-reference/data-types/nested-data-structures/index.md +++ b/docs/ru/sql-reference/data-types/nested-data-structures/index.md @@ -1,3 +1,10 @@ +--- +toc_folder_title: "\u0412\u043b\u043e\u0436\u0435\u043d\u043d\u044b\u0435\u0020\u0441\u0442\u0440\u0443\u043a\u0442\u0443\u0440\u044b\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +toc_hidden: true +toc_priority: 54 +toc_title: hidden +--- + # Вложенные структуры данных {#vlozhennye-struktury-dannykh} [Оригинальная статья](https://clickhouse.tech/docs/ru/data_types/nested_data_structures/) diff --git a/docs/ru/sql-reference/data-types/special-data-types/expression.md b/docs/ru/sql-reference/data-types/special-data-types/expression.md index 1f4b960fb10..718fcc886a6 100644 --- a/docs/ru/sql-reference/data-types/special-data-types/expression.md +++ b/docs/ru/sql-reference/data-types/special-data-types/expression.md @@ -1,3 +1,8 @@ +--- +toc_priority: 58 +toc_title: Expression +--- + # Expression {#expression} Используется для представления лямбда-выражений в функциях высшего порядка. diff --git a/docs/ru/sql-reference/data-types/special-data-types/index.md b/docs/ru/sql-reference/data-types/special-data-types/index.md index d5aff1501db..29c057472ea 100644 --- a/docs/ru/sql-reference/data-types/special-data-types/index.md +++ b/docs/ru/sql-reference/data-types/special-data-types/index.md @@ -1,3 +1,10 @@ +--- +toc_folder_title: "\u0421\u043b\u0443\u0436\u0435\u0431\u043d\u044b\u0435\u0020\u0442\u0438\u043f\u044b\u0020\u0434\u0430\u043d\u043d\u044b\u0445" +toc_hidden: true +toc_priority: 55 +toc_title: hidden +--- + # Служебные типы данных {#sluzhebnye-tipy-dannykh} Значения служебных типов данных не могут сохраняться в таблицу и выводиться в качестве результата, а возникают как промежуточный результат выполнения запроса. diff --git a/docs/ru/sql-reference/data-types/special-data-types/interval.md b/docs/ru/sql-reference/data-types/special-data-types/interval.md index a77d05ab8be..31240b49c97 100644 --- a/docs/ru/sql-reference/data-types/special-data-types/interval.md +++ b/docs/ru/sql-reference/data-types/special-data-types/interval.md @@ -1,3 +1,8 @@ +--- +toc_priority: 61 +toc_title: Interval +--- + # Interval {#data-type-interval} Семейство типов данных, представляющих интервалы дат и времени. Оператор [INTERVAL](../../../sql-reference/data-types/special-data-types/interval.md#operator-interval) возвращает значения этих типов. diff --git a/docs/ru/sql-reference/data-types/special-data-types/nothing.md b/docs/ru/sql-reference/data-types/special-data-types/nothing.md index 9644f102522..c6a9cb868d8 100644 --- a/docs/ru/sql-reference/data-types/special-data-types/nothing.md +++ b/docs/ru/sql-reference/data-types/special-data-types/nothing.md @@ -1,3 +1,8 @@ +--- +toc_priority: 60 +toc_title: Nothing +--- + # Nothing {#nothing} Этот тип данных предназначен только для того, чтобы представлять [NULL](../../../sql-reference/data-types/special-data-types/nothing.md), т.е. отсутствие значения. diff --git a/docs/ru/sql-reference/data-types/special-data-types/set.md b/docs/ru/sql-reference/data-types/special-data-types/set.md index 9a99ed0ca9b..4c2f4ed2c66 100644 --- a/docs/ru/sql-reference/data-types/special-data-types/set.md +++ b/docs/ru/sql-reference/data-types/special-data-types/set.md @@ -1,3 +1,8 @@ +--- +toc_priority: 59 +toc_title: Set +--- + # Set {#set} Используется для представления правой части выражения IN. diff --git a/docs/ru/sql-reference/data-types/string.md b/docs/ru/sql-reference/data-types/string.md index 4669d154df2..798caec4d62 100644 --- a/docs/ru/sql-reference/data-types/string.md +++ b/docs/ru/sql-reference/data-types/string.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: String +--- + # String {#string} Строки произвольной длины. Длина не ограничена. Значение может содержать произвольный набор байт, включая нулевые байты. diff --git a/docs/ru/sql-reference/data-types/uuid.md b/docs/ru/sql-reference/data-types/uuid.md index 9d667a50526..b780190f6f4 100644 --- a/docs/ru/sql-reference/data-types/uuid.md +++ b/docs/ru/sql-reference/data-types/uuid.md @@ -1,3 +1,9 @@ +--- +toc_priority: 46 +toc_title: UUID +--- + + # UUID {#uuid-data-type} Универсальный уникальный идентификатор (UUID) - это 16-байтовое число, используемое для идентификации записей. Подробнее про UUID читайте на [Википедии](https://en.wikipedia.org/wiki/Universally_unique_identifier). diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-hierarchical.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-hierarchical.md index 0869b409b0b..350e391dbed 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-hierarchical.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-hierarchical.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: "\u0418\u0435\u0440\u0430\u0440\u0445\u0438\u0447\u0435\u0441\u043a\u0438\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0438" +--- + # Иерархические словари {#ierarkhicheskie-slovari} ClickHouse поддерживает иерархические словари с [числовыми ключом](external-dicts-dict-structure.md#ext_dict-numeric-key). diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md index 61275cc6652..e97b1e421a4 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: "\u0425\u0440\u0430\u043d\u0435\u043d\u0438\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0435\u0439\u0020\u0432\u0020\u043f\u0430\u043c\u044f\u0442\u0438" +--- + # Хранение словарей в памяти {#dicts-external-dicts-dict-layout} Словари можно размещать в памяти множеством способов. @@ -49,6 +54,8 @@ LAYOUT(LAYOUT_TYPE(param value)) -- layout settings - [hashed](#dicts-external_dicts_dict_layout-hashed) - [sparse_hashed](#dicts-external_dicts_dict_layout-sparse_hashed) - [cache](#cache) +- [ssd_cache](#ssd-cache) +- [ssd_complex_key_cache](#complex-key-ssd-cache) - [direct](#direct) - [range_hashed](#range-hashed) - [complex_key_hashed](#complex-key-hashed) @@ -294,6 +301,40 @@ LAYOUT(CACHE(SIZE_IN_CELLS 1000000000)) Тип размещения предназначен для использования с составными [ключами](external-dicts-dict-structure.md). Аналогичен `cache`. +### ssd_cache {#ssd-cache} + +Похож на `cache`, но хранит данные на SSD и индекс в оперативной памяти. + +``` xml + + + + 4096 + + 16777216 + + 131072 + + 1048576 + + /var/lib/clickhouse/clickhouse_dictionaries/test_dict + + 1048576 + + +``` + +или + +``` sql +LAYOUT(CACHE(BLOCK_SIZE 4096 FILE_SIZE 16777216 READ_BUFFER_SIZE 1048576 + PATH /var/lib/clickhouse/clickhouse_dictionaries/test_dict MAX_STORED_KEYS 1048576)) +``` + +### complex_key_ssd_cache {#complex-key-ssd-cache} + +Тип размещения предназначен для использования с составными [ключами](../../../sql-reference/dictionaries/external-dictionaries/external-dicts-dict-structure.md). Похож на `ssd_cache`. + ### direct {#direct} Словарь не хранит данные локально и взаимодействует с источником непосредственно в момент запроса. diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md index 2f287795296..4dbf4be9f96 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-lifetime.md @@ -1,3 +1,8 @@ +--- +toc_priority: 42 +toc_title: "\u041e\u0431\u043d\u043e\u0432\u043b\u0435\u043d\u0438\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0435\u0439" +--- + # Обновление словарей {#obnovlenie-slovarei} ClickHouse периодически обновляет словари. Интервал обновления для полностью загружаемых словарей и интервал инвалидации для кэшируемых словарей определяется в теге `` в секундах. diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-sources.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-sources.md index 61d98f37cfd..0015edfdf72 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-sources.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-sources.md @@ -1,3 +1,8 @@ +--- +toc_priority: 43 +toc_title: "\u0418\u0441\u0442\u043e\u0447\u043d\u0438\u043a\u0438\u0020\u0432\u043d\u0435\u0448\u043d\u0438\u0445\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0435\u0439" +--- + # Источники внешних словарей {#dicts-external-dicts-dict-sources} Внешний словарь можно подключить из множества источников. diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-structure.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-structure.md index 4c3b4eb22e4..bf87ce61b9e 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-structure.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-structure.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: "\u041a\u043b\u044e\u0447\u0020\u0438\u0020\u043f\u043e\u043b\u044f\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u044f" +--- + # Ключ и поля словаря {#kliuch-i-polia-slovaria} Секция `` описывает ключ словаря и поля, доступные для запросов. diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict.md index 4ecc9b6c093..ff18f906926 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts-dict.md @@ -1,3 +1,8 @@ +--- +toc_priority: 40 +toc_title: "\u041d\u0430\u0441\u0442\u0440\u043e\u0439\u043a\u0430\u0020\u0432\u043d\u0435\u0448\u043d\u0435\u0433\u043e\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u044f" +--- + # Настройка внешнего словаря {#dicts-external-dicts-dict} XML-конфигурация словаря имеет следующую структуру: diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts.md index 734de8cffdf..c18af68c15e 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/external-dicts.md @@ -1,3 +1,9 @@ +--- +toc_priority: 39 +toc_title: "\u0412\u043d\u0435\u0448\u043d\u0438\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0438" +--- + + # Внешние словари {#dicts-external-dicts} Существует возможность подключать собственные словари из различных источников данных. Источником данных для словаря может быть локальный текстовый/исполняемый файл, HTTP(s) ресурс или другая СУБД. Подробнее смотрите в разделе «[Источники внешних словарей](external-dicts-dict-sources.md)». diff --git a/docs/ru/sql-reference/dictionaries/external-dictionaries/index.md b/docs/ru/sql-reference/dictionaries/external-dictionaries/index.md index 4098ac38060..b448858b1fa 100644 --- a/docs/ru/sql-reference/dictionaries/external-dictionaries/index.md +++ b/docs/ru/sql-reference/dictionaries/external-dictionaries/index.md @@ -1,5 +1,5 @@ --- -toc_folder_title: External Dictionaries +toc_folder_title: "\u0412\u043d\u0435\u0448\u043d\u0438\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0438" toc_priority: 37 --- diff --git a/docs/ru/sql-reference/dictionaries/index.md b/docs/ru/sql-reference/dictionaries/index.md index 8625a29b1e1..5a4119b4dd5 100644 --- a/docs/ru/sql-reference/dictionaries/index.md +++ b/docs/ru/sql-reference/dictionaries/index.md @@ -1,3 +1,9 @@ +--- +toc_folder_title: "\u0421\u043b\u043e\u0432\u0430\u0440\u0438" +toc_priority: 35 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" +--- + # Словари {#slovari} Словарь — это отображение (`ключ -> атрибуты`), которое удобно использовать для различного вида справочников. diff --git a/docs/ru/sql-reference/dictionaries/internal-dicts.md b/docs/ru/sql-reference/dictionaries/internal-dicts.md index 0e8e7c82a6e..d8103efa6ae 100644 --- a/docs/ru/sql-reference/dictionaries/internal-dicts.md +++ b/docs/ru/sql-reference/dictionaries/internal-dicts.md @@ -1,3 +1,8 @@ +--- +toc_priority: 39 +toc_title: "\u0412\u0441\u0442\u0440\u043e\u0435\u043d\u043d\u044b\u0435\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u0438" +--- + # Встроенные словари {#internal_dicts} ClickHouse содержит встроенную возможность работы с геобазой. diff --git a/docs/ru/sql-reference/distributed-ddl.md b/docs/ru/sql-reference/distributed-ddl.md index a06a931680a..275709320f6 100644 --- a/docs/ru/sql-reference/distributed-ddl.md +++ b/docs/ru/sql-reference/distributed-ddl.md @@ -1,6 +1,6 @@ --- toc_priority: 32 -toc_title: Распределенные DDL запросы +toc_title: "\u0420\u0430\u0441\u043f\u0440\u0435\u0434\u0435\u043b\u0435\u043d\u043d\u044b\u0435\u0020\u0044\u0044\u004c\u0020\u0437\u0430\u043f\u0440\u043e\u0441\u044b\u000a" --- # Распределенные DDL запросы (секция ON CLUSTER) {#raspredelennye-ddl-zaprosy-sektsiia-on-cluster} diff --git a/docs/ru/sql-reference/functions/arithmetic-functions.md b/docs/ru/sql-reference/functions/arithmetic-functions.md index 8513737f025..16c3e8fd8f0 100644 --- a/docs/ru/sql-reference/functions/arithmetic-functions.md +++ b/docs/ru/sql-reference/functions/arithmetic-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 34 +toc_title: "\u0410\u0440\u0438\u0444\u043c\u0435\u0442\u0438\u0447\u0435\u0441\u043a\u0438\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Арифметические функции {#arifmeticheskie-funktsii} Для всех арифметических функций, тип результата вычисляется, как минимальный числовой тип, который может вместить результат, если такой тип есть. Минимум берётся одновременно по числу бит, знаковости и «плавучести». Если бит не хватает, то берётся тип максимальной битности. diff --git a/docs/ru/sql-reference/functions/array-functions.md b/docs/ru/sql-reference/functions/array-functions.md index 91c0443c85d..62181eebb4b 100644 --- a/docs/ru/sql-reference/functions/array-functions.md +++ b/docs/ru/sql-reference/functions/array-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 35 +toc_title: "\u041c\u0430\u0441\u0441\u0438\u0432\u044b" +--- + # Массивы {#functions-for-working-with-arrays} ## empty {#function-empty} diff --git a/docs/ru/sql-reference/functions/array-join.md b/docs/ru/sql-reference/functions/array-join.md index afbc7bfe3e8..2ed3d25fa92 100644 --- a/docs/ru/sql-reference/functions/array-join.md +++ b/docs/ru/sql-reference/functions/array-join.md @@ -1,3 +1,8 @@ +--- +toc_priority: 61 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u044f\u0020\u0041\u0072\u0072\u0061\u0079\u004a\u006f\u0069\u006e" +--- + # Функция ArrayJoin {#functions_arrayjoin} Это совсем необычная функция. diff --git a/docs/ru/sql-reference/functions/bit-functions.md b/docs/ru/sql-reference/functions/bit-functions.md index 9d78dd861e9..8c7808437a5 100644 --- a/docs/ru/sql-reference/functions/bit-functions.md +++ b/docs/ru/sql-reference/functions/bit-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 48 +toc_title: "\u0411\u0438\u0442\u043e\u0432\u044b\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Битовые функции {#bitovye-funktsii} Битовые функции работают для любой пары типов из UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64. diff --git a/docs/ru/sql-reference/functions/bitmap-functions.md b/docs/ru/sql-reference/functions/bitmap-functions.md index aa154831b48..b21ddea94e4 100644 --- a/docs/ru/sql-reference/functions/bitmap-functions.md +++ b/docs/ru/sql-reference/functions/bitmap-functions.md @@ -1,4 +1,9 @@ -# Функции для битмапов {#bitmap-functions} +--- +toc_priority: 49 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0431\u0438\u0442\u043c\u0430\u043f\u043e\u0432" +--- + +# Функции для битовых масок {#bitmap-functions} ## bitmapBuild {#bitmap_functions-bitmapbuild} diff --git a/docs/ru/sql-reference/functions/comparison-functions.md b/docs/ru/sql-reference/functions/comparison-functions.md index 9b921d6fe55..a98c97ec96c 100644 --- a/docs/ru/sql-reference/functions/comparison-functions.md +++ b/docs/ru/sql-reference/functions/comparison-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 36 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0441\u0440\u0430\u0432\u043d\u0435\u043d\u0438\u044f" +--- + # Функции сравнения {#funktsii-sravneniia} Функции сравнения возвращают всегда 0 или 1 (UInt8). diff --git a/docs/ru/sql-reference/functions/conditional-functions.md b/docs/ru/sql-reference/functions/conditional-functions.md index 7efb6f7bfc5..83268b68959 100644 --- a/docs/ru/sql-reference/functions/conditional-functions.md +++ b/docs/ru/sql-reference/functions/conditional-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 43 +toc_title: "\u0423\u0441\u043b\u043e\u0432\u043d\u044b\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Условные функции {#uslovnye-funktsii} ## if {#if} diff --git a/docs/ru/sql-reference/functions/date-time-functions.md b/docs/ru/sql-reference/functions/date-time-functions.md index aa9fdee478d..d24de2faae1 100644 --- a/docs/ru/sql-reference/functions/date-time-functions.md +++ b/docs/ru/sql-reference/functions/date-time-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 39 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0434\u0430\u0442\u0430\u043c\u0438\u0020\u0438\u0020\u0432\u0440\u0435\u043c\u0435\u043d\u0435\u043c" +--- + # Функции для работы с датами и временем {#funktsii-dlia-raboty-s-datami-i-vremenem} Поддержка часовых поясов diff --git a/docs/ru/sql-reference/functions/encoding-functions.md b/docs/ru/sql-reference/functions/encoding-functions.md index b8ea481364d..6f1c2aad6cb 100644 --- a/docs/ru/sql-reference/functions/encoding-functions.md +++ b/docs/ru/sql-reference/functions/encoding-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 52 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043a\u043e\u0434\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f" +--- + # Функции кодирования {#funktsii-kodirovaniia} ## char {#char} diff --git a/docs/ru/sql-reference/functions/ext-dict-functions.md b/docs/ru/sql-reference/functions/ext-dict-functions.md index 792afd1775d..6054ed141d4 100644 --- a/docs/ru/sql-reference/functions/ext-dict-functions.md +++ b/docs/ru/sql-reference/functions/ext-dict-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 58 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0432\u043d\u0435\u0448\u043d\u0438\u043c\u0438\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u044f\u043c\u0438" +--- + # Функции для работы с внешними словарями {#ext_dict_functions} Информацию о подключении и настройке внешних словарей смотрите в разделе [Внешние словари](../../sql-reference/dictionaries/external-dictionaries/external-dicts.md). diff --git a/docs/ru/sql-reference/functions/functions-for-nulls.md b/docs/ru/sql-reference/functions/functions-for-nulls.md index 5fbfb4bd27b..17da1ea9194 100644 --- a/docs/ru/sql-reference/functions/functions-for-nulls.md +++ b/docs/ru/sql-reference/functions/functions-for-nulls.md @@ -1,3 +1,8 @@ +--- +toc_priority: 63 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u004e\u0075\u006c\u006c\u0061\u0062\u006c\u0065\u002d\u0430\u0440\u0433\u0443\u043c\u0435\u043d\u0442\u0430\u043c\u0438" +--- + # Функции для работы с Nullable-аргументами {#funktsii-dlia-raboty-s-nullable-argumentami} ## isNull {#isnull} diff --git a/docs/ru/sql-reference/functions/geo/coordinates.md b/docs/ru/sql-reference/functions/geo/coordinates.md index 9d65139e69e..1931a9b932f 100644 --- a/docs/ru/sql-reference/functions/geo/coordinates.md +++ b/docs/ru/sql-reference/functions/geo/coordinates.md @@ -1,5 +1,6 @@ --- -toc_title: Функции для работы с географическими координатами +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0433\u0435\u043e\u0433\u0440\u0430\u0444\u0438\u0447\u0435\u0441\u043a\u0438\u043c\u0438\u0020\u043a\u043e\u043e\u0440\u0434\u0438\u043d\u0430\u0442\u0430\u043c\u0438" +toc_priority: 62 --- # Функции для работы с географическими координатами {#geographical-coordinates} diff --git a/docs/ru/sql-reference/functions/geo/geohash.md b/docs/ru/sql-reference/functions/geo/geohash.md index 7273d58b1d9..38c64f11b10 100644 --- a/docs/ru/sql-reference/functions/geo/geohash.md +++ b/docs/ru/sql-reference/functions/geo/geohash.md @@ -1,5 +1,5 @@ --- -toc_title: Geohash +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0441\u0438\u0441\u0442\u0435\u043c\u043e\u0439\u0020\u0047\u0065\u006f\u0068\u0061\u0073\u0068" --- # Функции для работы с системой Geohash {#geohash} diff --git a/docs/ru/sql-reference/functions/geo/h3.md b/docs/ru/sql-reference/functions/geo/h3.md index d9a5e72721b..69d06b5dfa6 100644 --- a/docs/ru/sql-reference/functions/geo/h3.md +++ b/docs/ru/sql-reference/functions/geo/h3.md @@ -1,5 +1,5 @@ --- -toc_title: Индексы H3 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0438\u043d\u0434\u0435\u043a\u0441\u0430\u043c\u0438\u0020\u0048\u0033" --- # Функции для работы с индексами H3 {#h3index} diff --git a/docs/ru/sql-reference/functions/geo/index.md b/docs/ru/sql-reference/functions/geo/index.md index bfea32a245b..cedaafaa31d 100644 --- a/docs/ru/sql-reference/functions/geo/index.md +++ b/docs/ru/sql-reference/functions/geo/index.md @@ -1,6 +1,6 @@ --- toc_priority: 62 -toc_folder_title: Гео-данные +toc_folder_title: "\u0413\u0435\u043e\u002d\u0434\u0430\u043d\u043d\u044b\u0435" toc_title: hidden --- diff --git a/docs/ru/sql-reference/functions/hash-functions.md b/docs/ru/sql-reference/functions/hash-functions.md index 201fc934cea..92fc69227f4 100644 --- a/docs/ru/sql-reference/functions/hash-functions.md +++ b/docs/ru/sql-reference/functions/hash-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 50 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0445\u044d\u0448\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f" +--- + # Функции хэширования {#funktsii-kheshirovaniia} Функции хэширования могут использоваться для детерминированного псевдослучайного разбрасывания элементов. diff --git a/docs/ru/sql-reference/functions/in-functions.md b/docs/ru/sql-reference/functions/in-functions.md index 679fcbccc21..e137187a36b 100644 --- a/docs/ru/sql-reference/functions/in-functions.md +++ b/docs/ru/sql-reference/functions/in-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 60 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0435\u0430\u043b\u0438\u0437\u0430\u0446\u0438\u0438\u0020\u043e\u043f\u0435\u0440\u0430\u0442\u043e\u0440\u0430\u0020\u0049\u004e" +--- + # Функции для реализации оператора IN {#funktsii-dlia-realizatsii-operatora-in} ## in, notIn, globalIn, globalNotIn {#in-functions} diff --git a/docs/ru/sql-reference/functions/index.md b/docs/ru/sql-reference/functions/index.md index 9c1c0c5ca9d..25d3b6de067 100644 --- a/docs/ru/sql-reference/functions/index.md +++ b/docs/ru/sql-reference/functions/index.md @@ -1,3 +1,9 @@ +--- +toc_folder_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438" +toc_priority: 32 +toc_title: "\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435" +--- + # Функции {#funktsii} Функции бывают как минимум\* двух видов - обычные функции (называются просто, функциями) и агрегатные функции. Это совершенно разные вещи. Обычные функции работают так, как будто применяются к каждой строке по отдельности (для каждой строки, результат вычисления функции не зависит от других строк). Агрегатные функции аккумулируют множество значений из разных строк (то есть, зависят от целого множества строк). diff --git a/docs/ru/sql-reference/functions/introspection.md b/docs/ru/sql-reference/functions/introspection.md index c09efd35093..9f4f2ebd1e9 100644 --- a/docs/ru/sql-reference/functions/introspection.md +++ b/docs/ru/sql-reference/functions/introspection.md @@ -1,3 +1,8 @@ +--- +toc_priority: 65 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0438\u043d\u0442\u0440\u043e\u0441\u043f\u0435\u043a\u0446\u0438\u0438" +--- + # Функции интроспекции {#introspection-functions} Функции из этого раздела могут использоваться для интроспекции [ELF](https://en.wikipedia.org/wiki/Executable_and_Linkable_Format) и [DWARF](https://en.wikipedia.org/wiki/DWARF) в целях профилирования запросов. diff --git a/docs/ru/sql-reference/functions/ip-address-functions.md b/docs/ru/sql-reference/functions/ip-address-functions.md index a9a0a7f919a..724fb97c0d5 100644 --- a/docs/ru/sql-reference/functions/ip-address-functions.md +++ b/docs/ru/sql-reference/functions/ip-address-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 55 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0049\u0050\u002d\u0430\u0434\u0440\u0435\u0441\u0430\u043c\u0438" +--- + # Функции для работы с IP-адресами {#funktsii-dlia-raboty-s-ip-adresami} ## IPv4NumToString(num) {#ipv4numtostringnum} diff --git a/docs/ru/sql-reference/functions/json-functions.md b/docs/ru/sql-reference/functions/json-functions.md index 752b70b7c5f..69b8f8f98f5 100644 --- a/docs/ru/sql-reference/functions/json-functions.md +++ b/docs/ru/sql-reference/functions/json-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 56 +toc_title: JSON +--- + # Функции для работы с JSON {#funktsii-dlia-raboty-s-json} В Яндекс.Метрике пользователями передаётся JSON в качестве параметров визитов. Для работы с таким JSON-ом, реализованы некоторые функции. (Хотя в большинстве случаев, JSON-ы дополнительно обрабатываются заранее, и полученные значения кладутся в отдельные столбцы в уже обработанном виде.) Все эти функции исходят из сильных допущений о том, каким может быть JSON, и при этом стараются почти ничего не делать. diff --git a/docs/ru/sql-reference/functions/logical-functions.md b/docs/ru/sql-reference/functions/logical-functions.md index de2a3d51729..9b1ee6a66a7 100644 --- a/docs/ru/sql-reference/functions/logical-functions.md +++ b/docs/ru/sql-reference/functions/logical-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 37 +toc_title: "\u041b\u043e\u0433\u0438\u0447\u0435\u0441\u043a\u0438\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Логические функции {#logicheskie-funktsii} Логические функции принимают любые числовые типы, а возвращают число типа UInt8, равное 0 или 1. diff --git a/docs/ru/sql-reference/functions/machine-learning-functions.md b/docs/ru/sql-reference/functions/machine-learning-functions.md index e9b0e8c9bc9..2ffdfd05613 100644 --- a/docs/ru/sql-reference/functions/machine-learning-functions.md +++ b/docs/ru/sql-reference/functions/machine-learning-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 64 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e\u0020\u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f" +--- + # Функции машинного обучения {#funktsii-mashinnogo-obucheniia} ## evalMLMethod (prediction) {#machine_learning_methods-evalmlmethod} diff --git a/docs/ru/sql-reference/functions/math-functions.md b/docs/ru/sql-reference/functions/math-functions.md index e52b14d26c6..6df366d129f 100644 --- a/docs/ru/sql-reference/functions/math-functions.md +++ b/docs/ru/sql-reference/functions/math-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: "\u041c\u0430\u0442\u0435\u043c\u0430\u0442\u0438\u0447\u0435\u0441\u043a\u0438\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Математические функции {#matematicheskie-funktsii} Все функции возвращают число типа Float64. Точность результата близка к максимально возможной, но результат может не совпадать с наиболее близким к соответствующему вещественному числу машинно представимым числом. diff --git a/docs/ru/sql-reference/functions/other-functions.md b/docs/ru/sql-reference/functions/other-functions.md index bb5371ff449..9367f3be00c 100644 --- a/docs/ru/sql-reference/functions/other-functions.md +++ b/docs/ru/sql-reference/functions/other-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 66 +toc_title: "\u041f\u0440\u043e\u0447\u0438\u0435\u0020\u0444\u0443\u043d\u043a\u0446\u0438\u0438" +--- + # Прочие функции {#other-functions} ## hostName() {#hostname} @@ -1468,7 +1473,7 @@ SELECT getSetting('custom_a'); ## isDecimalOverflow {#is-decimal-overflow} -Проверяет, находится ли число [Decimal](../../sql-reference/data-types/decimal.md#decimalp-s-decimal32s-decimal64s-decimal128s) вне собственной (или заданной) области значений. +Проверяет, находится ли число [Decimal](../../sql-reference/data-types/decimal.md) вне собственной (или заданной) области значений. **Синтаксис** @@ -1478,7 +1483,7 @@ isDecimalOverflow(d, [p]) **Параметры** -- `d` — число. [Decimal](../../sql-reference/data-types/decimal.md#decimalp-s-decimal32s-decimal64s-decimal128s). +- `d` — число. [Decimal](../../sql-reference/data-types/decimal.md). - `p` — точность. Необязательный параметр. Если опущен, используется исходная точность первого аргумента. Использование этого параметра может быть полезно для извлечения данных в другую СУБД или файл. [UInt8](../../sql-reference/data-types/int-uint.md#uint-ranges). **Возвращаемое значение** @@ -1515,7 +1520,7 @@ countDigits(x) **Параметры** -- `x` — [целое](../../sql-reference/data-types/int-uint.md#uint8-uint16-uint32-uint64-int8-int16-int32-int64) или [дробное](../../sql-reference/data-types/decimal.md#decimalp-s-decimal32s-decimal64s-decimal128s) число. +- `x` — [целое](../../sql-reference/data-types/int-uint.md#uint8-uint16-uint32-uint64-int8-int16-int32-int64) или [дробное](../../sql-reference/data-types/decimal.md) число. **Возвращаемое значение** diff --git a/docs/ru/sql-reference/functions/random-functions.md b/docs/ru/sql-reference/functions/random-functions.md index 21dcfeeb3c0..f3889504fa6 100644 --- a/docs/ru/sql-reference/functions/random-functions.md +++ b/docs/ru/sql-reference/functions/random-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 51 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0433\u0435\u043d\u0435\u0440\u0430\u0446\u0438\u0438\u0020\u043f\u0441\u0435\u0432\u0434\u043e\u0441\u043b\u0443\u0447\u0430\u0439\u043d\u044b\u0445\u0020\u0447\u0438\u0441\u0435\u043b" +--- + # Функции генерации псевдослучайных чисел {#functions-for-generating-pseudo-random-numbers} Используются не криптографические генераторы псевдослучайных чисел. diff --git a/docs/ru/sql-reference/functions/rounding-functions.md b/docs/ru/sql-reference/functions/rounding-functions.md index c643bccf3e2..78033160396 100644 --- a/docs/ru/sql-reference/functions/rounding-functions.md +++ b/docs/ru/sql-reference/functions/rounding-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043e\u043a\u0440\u0443\u0433\u043b\u0435\u043d\u0438\u044f" +--- + # Функции округления {#funktsii-okrugleniia} ## floor(x\[, N\]) {#floorx-n} diff --git a/docs/ru/sql-reference/functions/splitting-merging-functions.md b/docs/ru/sql-reference/functions/splitting-merging-functions.md index bf4e76c3bb1..d451eabc407 100644 --- a/docs/ru/sql-reference/functions/splitting-merging-functions.md +++ b/docs/ru/sql-reference/functions/splitting-merging-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 47 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0440\u0430\u0437\u0431\u0438\u0435\u043d\u0438\u044f\u0020\u0438\u0020\u0441\u043b\u0438\u044f\u043d\u0438\u044f\u0020\u0441\u0442\u0440\u043e\u043a\u0020\u0438\u0020\u043c\u0430\u0441\u0441\u0438\u0432\u043e\u0432" +--- + # Функции разбиения и слияния строк и массивов {#funktsii-razbieniia-i-sliianiia-strok-i-massivov} ## splitByChar(separator, s) {#splitbycharseparator-s} diff --git a/docs/ru/sql-reference/functions/string-functions.md b/docs/ru/sql-reference/functions/string-functions.md index 58ed582b399..cc488fb2d9c 100644 --- a/docs/ru/sql-reference/functions/string-functions.md +++ b/docs/ru/sql-reference/functions/string-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 40 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u043e\u0020\u0441\u0442\u0440\u043e\u043a\u0430\u043c\u0438" +--- + # Функции для работы со строками {#funktsii-dlia-raboty-so-strokami} ## empty {#empty} diff --git a/docs/ru/sql-reference/functions/string-replace-functions.md b/docs/ru/sql-reference/functions/string-replace-functions.md index 8a2db9bf472..f334d6804f9 100644 --- a/docs/ru/sql-reference/functions/string-replace-functions.md +++ b/docs/ru/sql-reference/functions/string-replace-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 42 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043f\u043e\u0438\u0441\u043a\u0430\u0020\u0438\u0020\u0437\u0430\u043c\u0435\u043d\u044b\u0020\u0432\u0020\u0441\u0442\u0440\u043e\u043a\u0430\u0445" +--- + # Функции поиска и замены в строках {#funktsii-poiska-i-zameny-v-strokakh} ## replaceOne(haystack, pattern, replacement) {#replaceonehaystack-pattern-replacement} diff --git a/docs/ru/sql-reference/functions/string-search-functions.md b/docs/ru/sql-reference/functions/string-search-functions.md index 29dd67fd0eb..d2f1119783b 100644 --- a/docs/ru/sql-reference/functions/string-search-functions.md +++ b/docs/ru/sql-reference/functions/string-search-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043f\u043e\u0438\u0441\u043a\u0430\u0020\u0432\u0020\u0441\u0442\u0440\u043e\u043a\u0430\u0445" +--- + # Функции поиска в строках {#funktsii-poiska-v-strokakh} Во всех функциях, поиск регистрозависимый по умолчанию. Существуют варианты функций для регистронезависимого поиска. @@ -442,6 +447,66 @@ SELECT extractAllGroupsVertical('abc=111, def=222, ghi=333', '("[^"]+"|\\w+)=("[ То же, что like, но с отрицанием. +## ilike {#ilike} + +Нечувствительный к регистру вариант функции [like](https://clickhouse.tech/docs/ru/sql-reference/functions/string-search-functions/#function-like). Вы можете использовать оператор `ILIKE` вместо функции `ilike`. + +**Синтаксис** + +``` sql +ilike(haystack, pattern) +``` + +**Параметры** + +- `haystack` — Входная строка. [String](../../sql-reference/syntax.md#syntax-string-literal). +- `pattern` — Если `pattern` не содержит процента или нижнего подчеркивания, тогда `pattern` представляет саму строку. Нижнее подчеркивание (`_`) в `pattern` обозначает любой отдельный символ. Знак процента (`%`) соответствует последовательности из любого количества символов: от нуля и более. + +Некоторые примеры `pattern`: + +``` text +'abc' ILIKE 'abc' true +'abc' ILIKE 'a%' true +'abc' ILIKE '_b_' true +'abc' ILIKE 'c' false +``` + +**Возвращаемые значения** + +- Правда, если строка соответствует `pattern`. +- Ложь, если строка не соответствует `pattern`. + +**Пример** + +Входная таблица: + +``` text +┌─id─┬─name─────┬─days─┐ +│ 1 │ January │ 31 │ +│ 2 │ February │ 29 │ +│ 3 │ March │ 31 │ +│ 4 │ April │ 30 │ +└────┴──────────┴──────┘ +``` + +Запрос: + +``` sql +SELECT * FROM Months WHERE ilike(name, '%j%') +``` + +Результат: + +``` text +┌─id─┬─name────┬─days─┐ +│ 1 │ January │ 31 │ +└────┴─────────┴──────┘ +``` + +**Смотрите также** + +- [like](https://clickhouse.tech/docs/ru/sql-reference/functions/string-search-functions/#function-like) + ## ngramDistance(haystack, needle) {#ngramdistancehaystack-needle} Вычисление 4-граммного расстояния между `haystack` и `needle`: считается симметрическая разность между двумя мультимножествами 4-грамм и нормализуется на сумму их мощностей. Возвращает число float от 0 до 1 – чем ближе к нулю, тем больше строки похожи друг на друга. Если константный `needle` или `haystack` больше чем 32КБ, кидается исключение. Если некоторые строки из неконстантного `haystack` или `needle` больше 32КБ, расстояние всегда равно единице. diff --git a/docs/ru/sql-reference/functions/type-conversion-functions.md b/docs/ru/sql-reference/functions/type-conversion-functions.md index 3bee1a3656f..3733e570f10 100644 --- a/docs/ru/sql-reference/functions/type-conversion-functions.md +++ b/docs/ru/sql-reference/functions/type-conversion-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u043f\u0440\u0435\u043e\u0431\u0440\u0430\u0437\u043e\u0432\u0430\u043d\u0438\u044f\u0020\u0442\u0438\u043f\u043e\u0432" +--- + # Функции преобразования типов {#funktsii-preobrazovaniia-tipov} ## Общие проблемы преобразования чисел {#numeric-conversion-issues} @@ -767,4 +772,43 @@ FROM numbers(3) │ └──────────────────────────────────┘ ``` + +## formatRowNoNewline {#formatrownonewline} + +Преобразует произвольные выражения в строку заданного формата. При этом удаляет лишние переводы строк `\n`, если они появились. + +**Синтаксис** + +``` sql +formatRowNoNewline(format, x, y, ...) +``` + +**Параметры** + +- `format` — Текстовый формат. Например, [CSV](../../interfaces/formats.md#csv), [TSV](../../interfaces/formats.md#tabseparated). +- `x`,`y`, ... — Выражения. + +**Возвращаемое значение** + +- Отформатированная строка (в текстовых форматах без завершающего перевода строки). + +**Пример** + +Запрос: + +``` sql +SELECT formatRowNoNewline('CSV', number, 'good') +FROM numbers(3) +``` + +Ответ: + +``` text +┌─formatRowNoNewline('CSV', number, 'good')─┐ +│ 0,"good" │ +│ 1,"good" │ +│ 2,"good" │ +└───────────────────────────────────────────┘ +``` + [Оригинальная статья](https://clickhouse.tech/docs/ru/query_language/functions/type_conversion_functions/) diff --git a/docs/ru/sql-reference/functions/url-functions.md b/docs/ru/sql-reference/functions/url-functions.md index 54b752d7c68..8f10a1ebd2b 100644 --- a/docs/ru/sql-reference/functions/url-functions.md +++ b/docs/ru/sql-reference/functions/url-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 54 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0055\u0052\u004c" +--- + # Функции для работы с URL {#funktsii-dlia-raboty-s-url} Все функции работают не по RFC - то есть, максимально упрощены ради производительности. diff --git a/docs/ru/sql-reference/functions/uuid-functions.md b/docs/ru/sql-reference/functions/uuid-functions.md index b2b567c6a06..6082fcaa712 100644 --- a/docs/ru/sql-reference/functions/uuid-functions.md +++ b/docs/ru/sql-reference/functions/uuid-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 53 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u0020\u0055\u0055\u0049\u0044" +--- + # Функции для работы с UUID {#funktsii-dlia-raboty-s-uuid} ## generateUUIDv4 {#uuid-function-generate} @@ -54,6 +59,54 @@ SELECT toUUID('61f0c404-5cb3-11e7-907b-a6006ad3dba0') AS uuid └──────────────────────────────────────┘ ``` +## toUUIDOrNull (x) {#touuidornull-x} + +Принимает строку, и пытается преобразовать в тип UUID. При неудаче возвращает NULL. + +``` sql +toUUIDOrNull(String) +``` + +**Возвращаемое значение** + +Значение типа Nullable(UUID). + +**Пример использования** + +``` sql +SELECT toUUIDOrNull('61f0c404-5cb3-11e7-907b-a6006ad3dba0T') AS uuid +``` + +``` text +┌─uuid─┐ +│ ᴺᵁᴸᴸ │ +└──────┘ +``` + +## toUUIDOrZero (x) {#touuidorzero-x} + +Принимает строку, и пытается преобразовать в тип UUID. При неудаче возвращает нулевой UUID. + +``` sql +toUUIDOrZero(String) +``` + +**Возвращаемое значение** + +Значение типа UUID. + +**Пример использования** + +``` sql +SELECT toUUIDOrZero('61f0c404-5cb3-11e7-907b-a6006ad3dba0T') AS uuid +``` + +``` text +┌─────────────────────────────────uuid─┐ +│ 00000000-0000-0000-0000-000000000000 │ +└──────────────────────────────────────┘ +``` + ## UUIDStringToNum {#uuidstringtonum} Принимает строку, содержащую 36 символов в формате `xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx`, и возвращает в виде набора байт в [FixedString(16)](../../sql-reference/functions/uuid-functions.md). diff --git a/docs/ru/sql-reference/functions/ym-dict-functions.md b/docs/ru/sql-reference/functions/ym-dict-functions.md index 7ac27c0d285..c3b04e4ab66 100644 --- a/docs/ru/sql-reference/functions/ym-dict-functions.md +++ b/docs/ru/sql-reference/functions/ym-dict-functions.md @@ -1,3 +1,8 @@ +--- +toc_priority: 59 +toc_title: "\u0424\u0443\u043d\u043a\u0446\u0438\u0438\u0020\u0434\u043b\u044f\u0020\u0440\u0430\u0431\u043e\u0442\u044b\u0020\u0441\u043e\u0020\u0441\u043b\u043e\u0432\u0430\u0440\u044f\u043c\u0438\u0020\u042f\u043d\u0434\u0435\u043a\u0441\u002e\u041c\u0435\u0442\u0440\u0438\u043a\u0438" +--- + # Функции для работы со словарями Яндекс.Метрики {#ym-dict-functions} Чтобы указанные ниже функции работали, в конфиге сервера должны быть указаны пути и адреса для получения всех словарей Яндекс.Метрики. Словари загружаются при первом вызове любой из этих функций. Если справочники не удаётся загрузить - будет выкинуто исключение. diff --git a/docs/ru/sql-reference/operators/index.md b/docs/ru/sql-reference/operators/index.md index ef4d1ae048e..3befb18687d 100644 --- a/docs/ru/sql-reference/operators/index.md +++ b/docs/ru/sql-reference/operators/index.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: "\u041e\u043f\u0435\u0440\u0430\u0442\u043e\u0440\u044b" +--- + # Операторы {#operatory} Все операторы преобразуются в соответствующие функции на этапе парсинга запроса, с учётом их приоритетов и ассоциативности. @@ -49,6 +54,8 @@ `a NOT LIKE s` - функция `notLike(a, b)` +`a ILIKE s` – функция `ilike(a, b)` + `a BETWEEN b AND c` - равнозначно `a >= b AND a <= c` `a NOT BETWEEN b AND c` - равнозначно `a < b OR a > c` diff --git a/docs/ru/sql-reference/statements/alter/column.md b/docs/ru/sql-reference/statements/alter/column.md index 811539d60d3..a8fe0ccf642 100644 --- a/docs/ru/sql-reference/statements/alter/column.md +++ b/docs/ru/sql-reference/statements/alter/column.md @@ -1,6 +1,6 @@ --- toc_priority: 37 -toc_title: COLUMN +toc_title: "\u041c\u0430\u043d\u0438\u043f\u0443\u043b\u044f\u0446\u0438\u0438\u0020\u0441\u043e\u0020\u0441\u0442\u043e\u043b\u0431\u0446\u0430\u043c\u0438" --- # Манипуляции со столбцами {#manipuliatsii-so-stolbtsami} diff --git a/docs/ru/sql-reference/statements/alter/constraint.md b/docs/ru/sql-reference/statements/alter/constraint.md index bacdff9ff57..e26db208493 100644 --- a/docs/ru/sql-reference/statements/alter/constraint.md +++ b/docs/ru/sql-reference/statements/alter/constraint.md @@ -1,6 +1,6 @@ --- toc_priority: 43 -toc_title: CONSTRAINT +toc_title: "\u041c\u0430\u043d\u0438\u043f\u0443\u043b\u044f\u0446\u0438\u0438\u0020\u0441\u0020\u043e\u0433\u0440\u0430\u043d\u0438\u0447\u0435\u043d\u0438\u044f\u043c\u0438" --- # Манипуляции с ограничениями (constraints) {#manipuliatsii-s-ogranicheniiami-constraints} diff --git a/docs/ru/sql-reference/statements/alter/index.md b/docs/ru/sql-reference/statements/alter/index.md index 035be934eb4..830c4a5745b 100644 --- a/docs/ru/sql-reference/statements/alter/index.md +++ b/docs/ru/sql-reference/statements/alter/index.md @@ -1,5 +1,5 @@ --- -toc_priority: 36 +toc_priority: 35 toc_title: ALTER --- diff --git a/docs/ru/sql-reference/statements/alter/index/index.md b/docs/ru/sql-reference/statements/alter/index/index.md index 863f8b875dd..2cadbbe065e 100644 --- a/docs/ru/sql-reference/statements/alter/index/index.md +++ b/docs/ru/sql-reference/statements/alter/index/index.md @@ -1,7 +1,7 @@ --- toc_hidden_folder: true toc_priority: 42 -toc_title: INDEX +toc_title: "\u041c\u0430\u043d\u0438\u043f\u0443\u043b\u044f\u0446\u0438\u0438\u0020\u0441\u0020\u0438\u043d\u0434\u0435\u043a\u0441\u0430\u043c\u0438" --- # Манипуляции с индексами {#manipuliatsii-s-indeksami} diff --git a/docs/ru/sql-reference/statements/alter/sample-by.md b/docs/ru/sql-reference/statements/alter/sample-by.md new file mode 100644 index 00000000000..2235e345a45 --- /dev/null +++ b/docs/ru/sql-reference/statements/alter/sample-by.md @@ -0,0 +1,20 @@ +--- +toc_priority: 41 +toc_title: SAMPLE BY +--- + +# Manipulating Sampling-Key Expressions {#manipulations-with-sampling-key-expressions} + +Синтаксис: + +``` sql +ALTER TABLE [db].name [ON CLUSTER cluster] MODIFY SAMPLE BY new_expression +``` + +Команда меняет [ключ сэмплирования](../../../engines/table-engines/mergetree-family/mergetree.md) таблицы на `new_expression` (выражение или ряд выражений). + +Эта команда является упрощенной в том смысле, что она изменяет только метаданные. Первичный ключ должен содержать новый ключ сэмплирования. + +!!! note "Note" + Это работает только для таблиц в семействе [MergeTree](../../../engines/table-engines/mergetree-family/mergetree.md) (включая +[реплицируемые](../../../engines/table-engines/mergetree-family/replication.md) таблицы). \ No newline at end of file diff --git a/docs/ru/sql-reference/statements/attach.md b/docs/ru/sql-reference/statements/attach.md index 2aaf133130b..259ab893e63 100644 --- a/docs/ru/sql-reference/statements/attach.md +++ b/docs/ru/sql-reference/statements/attach.md @@ -1,5 +1,5 @@ --- -toc_priority: 42 +toc_priority: 40 toc_title: ATTACH --- diff --git a/docs/ru/sql-reference/statements/check-table.md b/docs/ru/sql-reference/statements/check-table.md index fa37b24851f..3dc135d87c6 100644 --- a/docs/ru/sql-reference/statements/check-table.md +++ b/docs/ru/sql-reference/statements/check-table.md @@ -1,5 +1,5 @@ --- -toc_priority: 43 +toc_priority: 41 toc_title: CHECK --- diff --git a/docs/ru/sql-reference/statements/create/database.md b/docs/ru/sql-reference/statements/create/database.md index a6c8b81c8b8..e6c561f8e0b 100644 --- a/docs/ru/sql-reference/statements/create/database.md +++ b/docs/ru/sql-reference/statements/create/database.md @@ -1,6 +1,6 @@ --- -toc_priority: 1 -toc_title: База данных +toc_priority: 35 +toc_title: "\u0411\u0430\u0437\u0430\u0020\u0434\u0430\u043d\u043d\u044b\u0445" --- # CREATE DATABASE {#query-language-create-database} diff --git a/docs/ru/sql-reference/statements/create/dictionary.md b/docs/ru/sql-reference/statements/create/dictionary.md index da9443e1b3a..a20dc812e02 100644 --- a/docs/ru/sql-reference/statements/create/dictionary.md +++ b/docs/ru/sql-reference/statements/create/dictionary.md @@ -1,6 +1,6 @@ --- -toc_priority: 4 -toc_title: Словарь +toc_priority: 38 +toc_title: "\u0421\u043b\u043e\u0432\u0430\u0440\u044c" --- # CREATE DICTIONARY {#create-dictionary-query} diff --git a/docs/ru/sql-reference/statements/create/index.md b/docs/ru/sql-reference/statements/create/index.md index f6399cfba11..28ddce2afe3 100644 --- a/docs/ru/sql-reference/statements/create/index.md +++ b/docs/ru/sql-reference/statements/create/index.md @@ -1,7 +1,7 @@ --- toc_folder_title: CREATE -toc_priority: 35 -toc_title: Обзор +toc_priority: 34 +toc_title: "\u041e\u0431\u0437\u043e\u0440" --- # Запросы CREATE {#create-queries} diff --git a/docs/ru/sql-reference/statements/create/quota.md b/docs/ru/sql-reference/statements/create/quota.md index 6351de2d38a..fe18869bf2e 100644 --- a/docs/ru/sql-reference/statements/create/quota.md +++ b/docs/ru/sql-reference/statements/create/quota.md @@ -1,6 +1,6 @@ --- -toc_priority: 8 -toc_title: Квота +toc_priority: 42 +toc_title: "\u041a\u0432\u043e\u0442\u0430" --- # CREATE QUOTA {#create-quota-statement} diff --git a/docs/ru/sql-reference/statements/create/role.md b/docs/ru/sql-reference/statements/create/role.md index b8c0fc2b453..b9e529fb213 100644 --- a/docs/ru/sql-reference/statements/create/role.md +++ b/docs/ru/sql-reference/statements/create/role.md @@ -1,6 +1,6 @@ --- -toc_priority: 6 -toc_title: Роль +toc_priority: 40 +toc_title: "\u0420\u043e\u043b\u044c" --- # CREATE ROLE {#create-role-statement} diff --git a/docs/ru/sql-reference/statements/create/row-policy.md b/docs/ru/sql-reference/statements/create/row-policy.md index 7ec28761452..a62e275a046 100644 --- a/docs/ru/sql-reference/statements/create/row-policy.md +++ b/docs/ru/sql-reference/statements/create/row-policy.md @@ -1,6 +1,6 @@ --- -toc_priority: 7 -toc_title: Политика доступа +toc_priority: 41 +toc_title: "\u041f\u043e\u043b\u0438\u0442\u0438\u043a\u0430\u0020\u0434\u043e\u0441\u0442\u0443\u043f\u0430" --- # CREATE ROW POLICY {#create-row-policy-statement} diff --git a/docs/ru/sql-reference/statements/create/settings-profile.md b/docs/ru/sql-reference/statements/create/settings-profile.md index 70c90b53565..9d525023af2 100644 --- a/docs/ru/sql-reference/statements/create/settings-profile.md +++ b/docs/ru/sql-reference/statements/create/settings-profile.md @@ -1,6 +1,6 @@ --- -toc_priority: 9 -toc_title: Профиль настроек +toc_priority: 43 +toc_title: "\u041f\u0440\u043e\u0444\u0438\u043b\u044c\u0020\u043d\u0430\u0441\u0442\u0440\u043e\u0435\u043a" --- # CREATE SETTINGS PROFILE {#create-settings-profile-statement} diff --git a/docs/ru/sql-reference/statements/create/table.md b/docs/ru/sql-reference/statements/create/table.md index 74eb0ff56e7..d54ec189a1a 100644 --- a/docs/ru/sql-reference/statements/create/table.md +++ b/docs/ru/sql-reference/statements/create/table.md @@ -1,6 +1,6 @@ --- -toc_priority: 2 -toc_title: Таблица +toc_priority: 36 +toc_title: "\u0422\u0430\u0431\u043b\u0438\u0446\u0430" --- # CREATE TABLE {#create-table-query} @@ -119,7 +119,18 @@ ENGINE = ... ``` -Если задать кодек для столбца, то кодек по умолчанию не применяется. Кодеки можно последовательно комбинировать, например, `CODEC(Delta, ZSTD)`. Чтобы выбрать наиболее подходящую для вашего проекта комбинацию кодеков, необходимо провести сравнительные тесты, подобные тем, что описаны в статье Altinity [New Encodings to Improve ClickHouse Efficiency](https://www.altinity.com/blog/2019/7/new-encodings-to-improve-clickhouse). +Если кодек `Default` задан для столбца, используется сжатие по умолчанию, которое может зависеть от различных настроек (и свойств данных) во время выполнения. +Пример: `value UInt64 CODEC(Default)` — то же самое, что не указать кодек. + +Также можно подменить кодек столбца сжатием по умолчанию, определенным в config.xml: + +``` sql +ALTER TABLE codec_example MODIFY COLUMN float_value CODEC(Default); +``` + +Кодеки можно последовательно комбинировать, например, `CODEC(Delta, Default)`. + +Чтобы выбрать наиболее подходящую для вашего проекта комбинацию кодеков, необходимо провести сравнительные тесты, подобные тем, что описаны в статье Altinity [New Encodings to Improve ClickHouse Efficiency](https://www.altinity.com/blog/2019/7/new-encodings-to-improve-clickhouse). Для столбцов типа `ALIAS` кодеки не применяются. !!! warning "Предупреждение" Нельзя распаковать базу данных ClickHouse с помощью сторонних утилит наподобие `lz4`. Необходимо использовать специальную утилиту [clickhouse-compressor](https://github.com/ClickHouse/ClickHouse/tree/master/programs/compressor). @@ -195,4 +206,4 @@ CREATE TEMPORARY TABLE [IF NOT EXISTS] table_name [Оригинальная статья](https://clickhouse.tech/docs/ru/sql-reference/statements/create/table) - \ No newline at end of file + diff --git a/docs/ru/sql-reference/statements/create/user.md b/docs/ru/sql-reference/statements/create/user.md index 3c04b4df86c..e7af1659a1b 100644 --- a/docs/ru/sql-reference/statements/create/user.md +++ b/docs/ru/sql-reference/statements/create/user.md @@ -1,6 +1,6 @@ --- -toc_priority: 5 -toc_title: Пользователь +toc_priority: 39 +toc_title: "\u041f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u0442\u0435\u043b\u044c" --- # CREATE USER {#create-user-statement} diff --git a/docs/ru/sql-reference/statements/create/view.md b/docs/ru/sql-reference/statements/create/view.md index caa3d04659e..09026874948 100644 --- a/docs/ru/sql-reference/statements/create/view.md +++ b/docs/ru/sql-reference/statements/create/view.md @@ -1,6 +1,6 @@ --- -toc_priority: 3 -toc_title: Представление +toc_priority: 37 +toc_title: "\u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u043b\u0435\u043d\u0438\u0435" --- # CREATE VIEW {#create-view} @@ -13,7 +13,7 @@ toc_title: Представление CREATE [OR REPLACE] VIEW [IF NOT EXISTS] [db.]table_name [ON CLUSTER] AS SELECT ... ``` -Normal views don’t store any data, they just perform a read from another table on each access. In other words, a normal view is nothing more than a saved query. When reading from a view, this saved query is used as a subquery in the [FROM](../../../sql-reference/statements/select/from.md) clause. +Обычные представления не хранят никаких данных, они выполняют чтение данных из другой таблицы при каждом доступе. Другими словами, обычное представление - это не что иное, как сохраненный запрос. При чтении данных из представления этот сохраненный запрос используется как подзапрос в секции [FROM](../../../sql-reference/statements/select/from.md). Для примера, пусть вы создали представление: diff --git a/docs/ru/sql-reference/statements/describe-table.md b/docs/ru/sql-reference/statements/describe-table.md index 5b38dca5833..64ed61de232 100644 --- a/docs/ru/sql-reference/statements/describe-table.md +++ b/docs/ru/sql-reference/statements/describe-table.md @@ -1,5 +1,5 @@ --- -toc_priority: 44 +toc_priority: 42 toc_title: DESCRIBE --- diff --git a/docs/ru/sql-reference/statements/detach.md b/docs/ru/sql-reference/statements/detach.md index 172e86179c8..00d0a4b20c6 100644 --- a/docs/ru/sql-reference/statements/detach.md +++ b/docs/ru/sql-reference/statements/detach.md @@ -1,5 +1,5 @@ --- -toc_priority: 45 +toc_priority: 43 toc_title: DETACH --- diff --git a/docs/ru/sql-reference/statements/drop.md b/docs/ru/sql-reference/statements/drop.md index 22e553cfdac..514a92db91f 100644 --- a/docs/ru/sql-reference/statements/drop.md +++ b/docs/ru/sql-reference/statements/drop.md @@ -1,110 +1,100 @@ --- -toc_priority: 46 +toc_priority: 44 toc_title: DROP --- # DROP {#drop} -Удаляет существующий объект. -Если указано `IF EXISTS` - не выдавать ошибку, если объекта не существует. +Удаляет существующий объект. Если указано `IF EXISTS` - не выдавать ошибку, если объекта не существует. ## DROP DATABASE {#drop-database} +Удаляет все таблицы в базе данных `db`, затем удаляет саму базу данных `db`. + +Синтаксис: + ``` sql DROP DATABASE [IF EXISTS] db [ON CLUSTER cluster] ``` -Удаляет все таблицы в базе данных db, затем удаляет саму базу данных db. - - ## DROP TABLE {#drop-table} +Удаляет таблицу. + +Синтаксис: + ``` sql DROP [TEMPORARY] TABLE [IF EXISTS] [db.]name [ON CLUSTER cluster] ``` -Удаляет таблицу. - - ## DROP DICTIONARY {#drop-dictionary} +Удаляет словарь. + +Синтаксис: + ``` sql DROP DICTIONARY [IF EXISTS] [db.]name ``` -Удаляет словарь. - - ## DROP USER {#drop-user-statement} Удаляет пользователя. -### Синтаксис {#drop-user-syntax} +Синтаксис: ```sql DROP USER [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` - ## DROP ROLE {#drop-role-statement} -Удаляет роль. +Удаляет роль. При удалении роль отзывается у всех объектов системы доступа, которым она присвоена. -При удалении роль отзывается у всех объектов системы доступа, которым она присвоена. - -### Синтаксис {#drop-role-syntax} +Синтаксис: ```sql DROP ROLE [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` - ## DROP ROW POLICY {#drop-row-policy-statement} -Удаляет политику доступа к строкам. +Удаляет политику доступа к строкам. При удалении политика отзывается у всех объектов системы доступа, которым она присвоена. -При удалении политика отзывается у всех объектов системы доступа, которым она присвоена. - -### Синтаксис {#drop-row-policy-syntax} +Синтаксис: ``` sql DROP [ROW] POLICY [IF EXISTS] name [,...] ON [database.]table [,...] [ON CLUSTER cluster_name] ``` - ## DROP QUOTA {#drop-quota-statement} -Удаляет квоту. +Удаляет квоту. При удалении квота отзывается у всех объектов системы доступа, которым она присвоена. -При удалении квота отзывается у всех объектов системы доступа, которым она присвоена. - -### Синтаксис {#drop-quota-syntax} +Синтаксис: ``` sql DROP QUOTA [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` - ## DROP SETTINGS PROFILE {#drop-settings-profile-statement} -Удаляет профиль настроек. +Удаляет профиль настроек. При удалении профиль отзывается у всех объектов системы доступа, которым он присвоен. -При удалении профиль отзывается у всех объектов системы доступа, которым он присвоен. - -### Синтаксис {#drop-settings-profile-syntax} +Синтаксис: ``` sql DROP [SETTINGS] PROFILE [IF EXISTS] name [,...] [ON CLUSTER cluster_name] ``` - ## DROP VIEW {#drop-view} +Удаляет представление. Представления могут быть удалены и командой `DROP TABLE`, но команда `DROP VIEW` проверяет, что `[db.]name` является представлением. + +Синтаксис: + ``` sql DROP VIEW [IF EXISTS] [db.]name [ON CLUSTER cluster] ``` -Удаляет представление. Представления могут быть удалены и командой `DROP TABLE`, но команда `DROP VIEW` проверяет, что `[db.]name` является представлением. - - -[Оригинальная статья](https://clickhouse.tech/docs/ru/sql-reference/statements/drop/) +[Оригинальная статья](https://clickhouse.tech/docs/ru/sql-reference/statements/drop/) \ No newline at end of file diff --git a/docs/ru/sql-reference/statements/exists.md b/docs/ru/sql-reference/statements/exists.md index 865d23ad622..0b2fd69273c 100644 --- a/docs/ru/sql-reference/statements/exists.md +++ b/docs/ru/sql-reference/statements/exists.md @@ -1,5 +1,5 @@ --- -toc_priority: 47 +toc_priority: 45 toc_title: EXISTS --- diff --git a/docs/ru/sql-reference/statements/grant.md b/docs/ru/sql-reference/statements/grant.md index 8eea84ac594..d38e2ea38a0 100644 --- a/docs/ru/sql-reference/statements/grant.md +++ b/docs/ru/sql-reference/statements/grant.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: GRANT +--- + # GRANT - Присваивает [привилегии](#grant-privileges) пользователям или ролям ClickHouse. @@ -79,6 +84,7 @@ GRANT SELECT(x,y) ON db.table TO john WITH GRANT OPTION - `ALTER RENAME COLUMN` - `ALTER INDEX` - `ALTER ORDER BY` + - `ALTER SAMPLE BY` - `ALTER ADD INDEX` - `ALTER DROP INDEX` - `ALTER MATERIALIZE INDEX` @@ -264,6 +270,7 @@ GRANT INSERT(x,y) ON db.table TO john - `ALTER RENAME COLUMN`. Уровень: `COLUMN`. Алиасы: `RENAME COLUMN` - `ALTER INDEX`. Уровень: `GROUP`. Алиасы: `INDEX` - `ALTER ORDER BY`. Уровень: `TABLE`. Алиасы: `ALTER MODIFY ORDER BY`, `MODIFY ORDER BY` + - `ALTER SAMPLE BY`. Уровень: `TABLE`. Алиасы: `ALTER MODIFY SAMPLE BY`, `MODIFY SAMPLE BY` - `ALTER ADD INDEX`. Уровень: `TABLE`. Алиасы: `ADD INDEX` - `ALTER DROP INDEX`. Уровень: `TABLE`. Алиасы: `DROP INDEX` - `ALTER MATERIALIZE INDEX`. Уровень: `TABLE`. Алиасы: `MATERIALIZE INDEX` diff --git a/docs/ru/sql-reference/statements/insert-into.md b/docs/ru/sql-reference/statements/insert-into.md index 281b708f3f9..0d38be81ac6 100644 --- a/docs/ru/sql-reference/statements/insert-into.md +++ b/docs/ru/sql-reference/statements/insert-into.md @@ -1,5 +1,5 @@ --- -toc_priority: 34 +toc_priority: 33 toc_title: INSERT INTO --- @@ -13,7 +13,55 @@ toc_title: INSERT INTO INSERT INTO [db.]table [(c1, c2, c3)] VALUES (v11, v12, v13), (v21, v22, v23), ... ``` -В запросе можно указать список столбцов для вставки `[(c1, c2, c3)]`. В этом случае, в остальные столбцы записываются: +Вы можете указать список столбцов для вставки, используя следующий синтаксис: `(c1, c2, c3)` или `COLUMNS(c1,c2,c3)`. + +Можно не перечислять все необходимые столбцы, а использовать синтаксис `(* EXCEPT(column_list))`. + +В качестве примера рассмотрим таблицу: + +``` sql +SHOW CREATE insert_select_testtable +``` + +``` +┌─statement────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐ +│ CREATE TABLE insert_select_testtable +( + `a` Int8, + `b` String, + `c` Int8 +) +ENGINE = MergeTree() +ORDER BY a +SETTINGS index_granularity = 8192 │ +└──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘ +``` + +``` sql +INSERT INTO insert_select_testtable (*) VALUES (1, 'a', 1) +``` + +Если вы хотите вставить данные во все столбцы, кроме 'b', вам нужно передать столько значений, сколько столбцов вы указали в скобках: + +``` sql +INSERT INTO insert_select_testtable (* EXCEPT(b)) Values (2, 2) +``` + +``` sql +SELECT * FROM insert_select_testtable +``` + +``` +┌─a─┬─b─┬─c─┐ +│ 2 │ │ 2 │ +└───┴───┴───┘ +┌─a─┬─b─┬─c─┐ +│ 1 │ a │ 1 │ +└───┴───┴───┘ +``` + +В этом примере мы видим, что вторая строка содержит столбцы `a` и `c`, заполненные переданными значениями и `b`, заполненный значением по умолчанию. +Если список столбцов не включает все существующие столбцы, то все остальные столбцы заполняются следующим образом: - Значения, вычисляемые из `DEFAULT` выражений, указанных в определении таблицы. - Нули и пустые строки, если `DEFAULT` не определены. diff --git a/docs/ru/sql-reference/statements/kill.md b/docs/ru/sql-reference/statements/kill.md index dd2f24ffa27..e2556a7f782 100644 --- a/docs/ru/sql-reference/statements/kill.md +++ b/docs/ru/sql-reference/statements/kill.md @@ -1,5 +1,5 @@ --- -toc_priority: 48 +toc_priority: 46 toc_title: KILL --- diff --git a/docs/ru/sql-reference/statements/optimize.md b/docs/ru/sql-reference/statements/optimize.md index dc70d86a1a0..9b94c31a8f7 100644 --- a/docs/ru/sql-reference/statements/optimize.md +++ b/docs/ru/sql-reference/statements/optimize.md @@ -1,5 +1,5 @@ --- -toc_priority: 49 +toc_priority: 47 toc_title: OPTIMIZE --- diff --git a/docs/ru/sql-reference/statements/rename.md b/docs/ru/sql-reference/statements/rename.md index 9f6069d8a94..94bf3c682a1 100644 --- a/docs/ru/sql-reference/statements/rename.md +++ b/docs/ru/sql-reference/statements/rename.md @@ -1,5 +1,5 @@ --- -toc_priority: 50 +toc_priority: 48 toc_title: RENAME --- diff --git a/docs/ru/sql-reference/statements/revoke.md b/docs/ru/sql-reference/statements/revoke.md index 1d2928bb76e..339746b8591 100644 --- a/docs/ru/sql-reference/statements/revoke.md +++ b/docs/ru/sql-reference/statements/revoke.md @@ -1,3 +1,8 @@ +--- +toc_priority: 39 +toc_title: REVOKE +--- + # REVOKE Отзывает привилегии у пользователей или ролей. diff --git a/docs/ru/sql-reference/statements/select/array-join.md b/docs/ru/sql-reference/statements/select/array-join.md index f8f11ba1b17..a3abf9e5e2e 100644 --- a/docs/ru/sql-reference/statements/select/array-join.md +++ b/docs/ru/sql-reference/statements/select/array-join.md @@ -1,3 +1,7 @@ +--- +toc_title: ARRAY JOIN +--- + # Секция ARRAY JOIN {#select-array-join-clause} Типовая операция для таблиц, содержащих столбец-массив — произвести новую таблицу, которая будет иметь столбец с каждым отдельным элементом массивов из изначального столбца, в то время как значения других столбцов дублируются. Это основной сценарий использования секции `ARRAY JOIN`. diff --git a/docs/ru/sql-reference/statements/select/distinct.md b/docs/ru/sql-reference/statements/select/distinct.md index 62e2e25b7e5..9d620079f6b 100644 --- a/docs/ru/sql-reference/statements/select/distinct.md +++ b/docs/ru/sql-reference/statements/select/distinct.md @@ -1,3 +1,7 @@ +--- +toc_title: DISTINCT +--- + # Секция DISTINCT {#select-distinct} Если указан `SELECT DISTINCT`, то в результате запроса останутся только уникальные строки. Таким образом, из всех наборов полностью совпадающих строк в результате останется только одна строка. diff --git a/docs/ru/sql-reference/statements/select/format.md b/docs/ru/sql-reference/statements/select/format.md index dad0ef0d62c..18972ddd82b 100644 --- a/docs/ru/sql-reference/statements/select/format.md +++ b/docs/ru/sql-reference/statements/select/format.md @@ -1,3 +1,7 @@ +--- +toc_title: FORMAT +--- + # Секция FORMAT {#format-clause} ClickHouse поддерживает широкий спектр [форматов сериализации](../../../interfaces/formats.md) это может быть использовано, в частности, для результатов запросов. Существует несколько способов выбора формата для `SELECT`, один из них заключается в том, чтобы указать `FORMAT format` в конце запроса, чтобы получить результирующие данные в любом конкретном формате. diff --git a/docs/ru/sql-reference/statements/select/from.md b/docs/ru/sql-reference/statements/select/from.md index ac0ab1dcd3f..491bbfe892b 100644 --- a/docs/ru/sql-reference/statements/select/from.md +++ b/docs/ru/sql-reference/statements/select/from.md @@ -1,3 +1,7 @@ +--- +toc_title: FROM +--- + # Секция FROM {#select-from} В секции `FROM` указывается источник, из которого будут читаться данные: diff --git a/docs/ru/sql-reference/statements/select/group-by.md b/docs/ru/sql-reference/statements/select/group-by.md index 9581f477af5..a0454ef1d91 100644 --- a/docs/ru/sql-reference/statements/select/group-by.md +++ b/docs/ru/sql-reference/statements/select/group-by.md @@ -1,3 +1,7 @@ +--- +toc_title: GROUP BY +--- + # Секция GROUP BY {#select-group-by-clause} Секция `GROUP BY` переключает `SELECT` запрос в режим агрегации, который работает следующим образом: diff --git a/docs/ru/sql-reference/statements/select/having.md b/docs/ru/sql-reference/statements/select/having.md index 83f58c5566f..dc701df906f 100644 --- a/docs/ru/sql-reference/statements/select/having.md +++ b/docs/ru/sql-reference/statements/select/having.md @@ -1,3 +1,7 @@ +--- +toc_title: HAVING +--- + # Секция HAVING {#having-clause} Позволяет фильтровать результаты агрегации, полученные с помощью [GROUP BY](group-by.md). Разница с [WHERE](where.md) в том, что `WHERE` выполняется перед агрегацией, в то время как `HAVING` выполняется после него. diff --git a/docs/ru/sql-reference/statements/select/index.md b/docs/ru/sql-reference/statements/select/index.md index b735d37118c..f5fe2788370 100644 --- a/docs/ru/sql-reference/statements/select/index.md +++ b/docs/ru/sql-reference/statements/select/index.md @@ -1,6 +1,8 @@ --- -toc_priority: 33 -toc_title: SELECT +title: "\u0421\u0438\u043d\u0442\u0430\u043a\u0441\u0438\u0441\u0020\u0437\u0430\u043f\u0440\u043e\u0441\u043e\u0432\u0020\u0053\u0045\u004c\u0045\u0043\u0054" +toc_folder_title: SELECT +toc_priority: 32 +toc_title: "\u041e\u0431\u0437\u043e\u0440" --- # Синтаксис запросов SELECT {#select-queries-syntax} diff --git a/docs/ru/sql-reference/statements/select/into-outfile.md b/docs/ru/sql-reference/statements/select/into-outfile.md index 0f5cf01e9d1..f956903f8b4 100644 --- a/docs/ru/sql-reference/statements/select/into-outfile.md +++ b/docs/ru/sql-reference/statements/select/into-outfile.md @@ -1,3 +1,7 @@ +--- +toc_title: INTO OUTFILE +--- + # Секция INTO OUTFILE {#into-outfile-clause} Чтобы перенаправить вывод `SELECT` запроса в указанный файл на стороне клиента, добавьте к нему секцию `INTO OUTFILE filename` (где filename — строковый литерал). diff --git a/docs/ru/sql-reference/statements/select/join.md b/docs/ru/sql-reference/statements/select/join.md index 6f1b5e2cde6..c5548d74156 100644 --- a/docs/ru/sql-reference/statements/select/join.md +++ b/docs/ru/sql-reference/statements/select/join.md @@ -1,3 +1,7 @@ +--- +toc_title: JOIN +--- + # Секция JOIN {#select-join} Join создаёт новую таблицу путем объединения столбцов из одной или нескольких таблиц с использованием общих для каждой из них значений. Это обычная операция в базах данных с поддержкой SQL, которая соответствует join из [реляционной алгебры](https://en.wikipedia.org/wiki/Relational_algebra#Joins_and_join-like_operators). Частный случай соединения одной таблицы часто называют «self-join». diff --git a/docs/ru/sql-reference/statements/select/limit-by.md b/docs/ru/sql-reference/statements/select/limit-by.md index ea5d467ae4f..fba81c023b5 100644 --- a/docs/ru/sql-reference/statements/select/limit-by.md +++ b/docs/ru/sql-reference/statements/select/limit-by.md @@ -1,3 +1,7 @@ +--- +toc_title: LIMIT BY +--- + # Секция LIMIT BY {#limit-by-clause} Запрос с секцией `LIMIT n BY expressions` выбирает первые `n` строк для каждого отличного значения `expressions`. Ключ `LIMIT BY` может содержать любое количество [выражений](../../syntax.md#syntax-expressions). diff --git a/docs/ru/sql-reference/statements/select/prewhere.md b/docs/ru/sql-reference/statements/select/prewhere.md index 1c8595d8e0c..c2a02b1a436 100644 --- a/docs/ru/sql-reference/statements/select/prewhere.md +++ b/docs/ru/sql-reference/statements/select/prewhere.md @@ -1,3 +1,7 @@ +--- +toc_title: PREWHERE +--- + # Секция PREWHERE {#prewhere-clause} Prewhere — это оптимизация для более эффективного применения фильтрации. Она включена по умолчанию, даже если секция `PREWHERE` явно не указана. В этом случае работает автоматическое перемещение части выражения из [WHERE](where.md) до стадии prewhere. Роль секции `PREWHERE` только для управления этой оптимизацией, если вы думаете, что знаете, как сделать перемещение условия лучше, чем это происходит по умолчанию. diff --git a/docs/ru/sql-reference/statements/select/sample.md b/docs/ru/sql-reference/statements/select/sample.md index ca6b49c9ad6..a71e8256f0f 100644 --- a/docs/ru/sql-reference/statements/select/sample.md +++ b/docs/ru/sql-reference/statements/select/sample.md @@ -1,3 +1,7 @@ +--- +toc_title: SAMPLE +--- + # Секция SAMPLE {#select-sample-clause} Секция `SAMPLE` позволяет выполнять запросы приближённо. Например, чтобы посчитать статистику по всем визитам, можно обработать 1/10 всех визитов и результат домножить на 10. diff --git a/docs/ru/sql-reference/statements/select/union-all.md b/docs/ru/sql-reference/statements/select/union-all.md index bffd667fa1d..b9d1f485a7b 100644 --- a/docs/ru/sql-reference/statements/select/union-all.md +++ b/docs/ru/sql-reference/statements/select/union-all.md @@ -1,3 +1,7 @@ +--- +toc_title: UNION ALL +--- + # Секция UNION ALL {#union-all-clause} Вы можете использовать `UNION ALL` чтобы объединить любое количество `SELECT` запросы путем расширения их результатов. Пример: diff --git a/docs/ru/sql-reference/statements/select/where.md b/docs/ru/sql-reference/statements/select/where.md index 63d081db43d..8cb8e634303 100644 --- a/docs/ru/sql-reference/statements/select/where.md +++ b/docs/ru/sql-reference/statements/select/where.md @@ -1,3 +1,7 @@ +--- +toc_title: WHERE +--- + # Секция WHERE {#select-where} Позволяет задать выражение, которое ClickHouse использует для фильтрации данных перед всеми другими действиями в запросе кроме выражений, содержащихся в секции [PREWHERE](prewhere.md#prewhere-clause). Обычно, это выражение с логическими операторами. diff --git a/docs/ru/sql-reference/statements/select/with.md b/docs/ru/sql-reference/statements/select/with.md index a5be733866f..328b28c27ef 100644 --- a/docs/ru/sql-reference/statements/select/with.md +++ b/docs/ru/sql-reference/statements/select/with.md @@ -1,15 +1,22 @@ -# Секция WITH {#sektsiia-with} +--- +toc_title: WITH +--- -Данная секция представляет собой [Common Table Expressions](https://ru.wikipedia.org/wiki/Иерархические_и_рекурсивные_запросы_в_SQL), то есть позволяет использовать результаты выражений из секции `WITH` в остальной части `SELECT` запроса. +# Секция WITH {#with-clause} + +Clickhouse поддерживает [Общие табличные выражения](https://ru.wikipedia.org/wiki/Иерархические_и_рекурсивные_запросы_в_SQL), то есть позволяет использовать результаты выражений из секции `WITH` в остальной части `SELECT` запроса. Именованные подзапросы могут быть включены в текущий и дочерний контекст запроса в тех местах, где разрешены табличные объекты. Рекурсия предотвращается путем скрытия общего табличного выражения текущего уровня из выражения `WITH`. + +## Синтаксис + +``` sql +WITH AS +``` +или +``` sql +WITH AS +``` -### Ограничения - -1. Рекурсивные запросы не поддерживаются -2. Если в качестве выражения используется подзапрос, то результат должен содержать ровно одну строку -3. Результаты выражений нельзя переиспользовать во вложенных запросах -В дальнейшем, результаты выражений можно использовать в секции SELECT. - -### Примеры +## Примеры **Пример 1:** Использование константного выражения как «переменной» @@ -19,7 +26,7 @@ SELECT * FROM hits WHERE EventDate = toDate(ts_upper_bound) AND - EventTime <= ts_upper_bound + EventTime <= ts_upper_bound; ``` **Пример 2:** Выкидывание выражения sum(bytes) из списка колонок в SELECT @@ -31,7 +38,7 @@ SELECT table FROM system.parts GROUP BY table -ORDER BY s +ORDER BY s; ``` **Пример 3:** Использование результатов скалярного подзапроса @@ -50,27 +57,14 @@ SELECT FROM system.parts GROUP BY table ORDER BY table_disk_usage DESC -LIMIT 10 +LIMIT 10; ``` **Пример 4:** Переиспользование выражения -В настоящий момент, переиспользование выражения из секции WITH внутри подзапроса возможно только через дублирование. - ``` sql -WITH ['hello'] AS hello -SELECT - hello, - * -FROM -( - WITH ['hello'] AS hello - SELECT hello -) +WITH test1 AS (SELECT i + 1, j + 1 FROM test1) +SELECT * FROM test1; ``` -``` text -┌─hello─────┬─hello─────┐ -│ ['hello'] │ ['hello'] │ -└───────────┴───────────┘ -``` +[Оригинальная статья](https://clickhouse.tech/docs/en/sql-reference/statements/select/with/) diff --git a/docs/ru/sql-reference/statements/set-role.md b/docs/ru/sql-reference/statements/set-role.md index 5d840fcddb7..ccbef41aa9b 100644 --- a/docs/ru/sql-reference/statements/set-role.md +++ b/docs/ru/sql-reference/statements/set-role.md @@ -1,5 +1,5 @@ --- -toc_priority: 52 +toc_priority: 50 toc_title: SET ROLE --- diff --git a/docs/ru/sql-reference/statements/set.md b/docs/ru/sql-reference/statements/set.md index 0e12e2ee7bc..b60dfcf8324 100644 --- a/docs/ru/sql-reference/statements/set.md +++ b/docs/ru/sql-reference/statements/set.md @@ -1,5 +1,5 @@ --- -toc_priority: 51 +toc_priority: 49 toc_title: SET --- diff --git a/docs/ru/sql-reference/statements/show.md b/docs/ru/sql-reference/statements/show.md index 575742568cb..4516a401de9 100644 --- a/docs/ru/sql-reference/statements/show.md +++ b/docs/ru/sql-reference/statements/show.md @@ -1,3 +1,8 @@ +--- +toc_priority: 37 +toc_title: SHOW +--- + # SHOW Queries {#show-queries} ## SHOW CREATE TABLE {#show-create-table} @@ -10,12 +15,83 @@ SHOW CREATE [TEMPORARY] [TABLE|DICTIONARY] [db.]table [INTO OUTFILE filename] [F ## SHOW DATABASES {#show-databases} -``` sql -SHOW DATABASES [INTO OUTFILE filename] [FORMAT format] +Выводит список всех баз данных. + +```sql +SHOW DATABASES [LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE filename] [FORMAT format] ``` -Выводит список всех баз данных. -Запрос полностью аналогичен запросу `SELECT name FROM system.databases [INTO OUTFILE filename] [FORMAT format]`. +Этот запрос идентичен запросу: + +```sql +SELECT name FROM system.databases [WHERE name LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE filename] [FORMAT format] +``` + +### Примеры {#examples} + +Получение списка баз данных, имена которых содержат последовательность символов 'de': + +``` sql +SHOW DATABASES LIKE '%de%' +``` + +Результат: + +``` text +┌─name────┐ +│ default │ +└─────────┘ +``` + +Получение списка баз данных, имена которых содержат последовательность символов 'de' независимо от регистра: + +``` sql +SHOW DATABASES ILIKE '%DE%' +``` + +Результат: + +``` text +┌─name────┐ +│ default │ +└─────────┘ +``` + +Получение списка баз данных, имена которых не содержат последовательность символов 'de': + +``` sql +SHOW DATABASES NOT LIKE '%de%' +``` + +Результат: + +``` text +┌─name───────────────────────────┐ +│ _temporary_and_external_tables │ +│ system │ +│ test │ +│ tutorial │ +└────────────────────────────────┘ +``` + +Получение первых двух строк из списка имен баз данных: + +``` sql +SHOW DATABASES LIMIT 2 +``` + +Результат: + +``` text +┌─name───────────────────────────┐ +│ _temporary_and_external_tables │ +│ default │ +└────────────────────────────────┘ +``` + +### Смотрите также {#see-also} + +- [CREATE DATABASE](https://clickhouse.tech/docs/ru/sql-reference/statements/create/database/#query-language-create-database) ## SHOW PROCESSLIST {#show-processlist} @@ -37,33 +113,86 @@ $ watch -n1 "clickhouse-client --query='SHOW PROCESSLIST'" Выводит список таблиц. -``` sql -SHOW [TEMPORARY] TABLES [{FROM | IN} ] [LIKE '' | WHERE expr] [LIMIT ] [INTO OUTFILE ] [FORMAT ] +```sql +SHOW [TEMPORARY] TABLES [{FROM | IN} ] [LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE ] [FORMAT ] ``` -Если секция `FROM` не используется, то запрос возвращает список таблиц из текущей базы данных. +Если условие `FROM` не указано, запрос возвращает список таблиц из текущей базы данных. -Результат, идентичный тому, что выдаёт запрос `SHOW TABLES` можно получить также запросом следующего вида: +Этот запрос идентичен запросу: -``` sql -SELECT name FROM system.tables WHERE database = [AND name LIKE ] [LIMIT ] [INTO OUTFILE ] [FORMAT ] +```sql +SELECT name FROM system.tables [WHERE name LIKE | ILIKE | NOT LIKE ''] [LIMIT ] [INTO OUTFILE ] [FORMAT ] ``` -**Пример** +### Примеры {#examples} -Следующий запрос выбирает первые две строки из списка таблиц в базе данных `system`, чьи имена содержат `co`. +Получение списка таблиц, имена которых содержат последовательность символов 'user': ``` sql -SHOW TABLES FROM system LIKE '%co%' LIMIT 2 +SHOW TABLES FROM system LIKE '%user%' ``` +Результат: + +``` text +┌─name─────────────┐ +│ user_directories │ +│ users │ +└──────────────────┘ +``` + +Получение списка таблиц, имена которых содержат последовательность символов 'user' без учета регистра: + +``` sql +SHOW TABLES FROM system ILIKE '%USER%' +``` + +Результат: + +``` text +┌─name─────────────┐ +│ user_directories │ +│ users │ +└──────────────────┘ +``` + +Получение списка таблиц, имена которых не содержат символ 's': + +``` sql +SHOW TABLES FROM system NOT LIKE '%s%' +``` + +Результат: + +``` text +┌─name─────────┐ +│ metric_log │ +│ metric_log_0 │ +│ metric_log_1 │ +└──────────────┘ +``` + +Получение первых двух строк из списка таблиц: + +``` sql +SHOW TABLES FROM system LIMIT 2 +``` + +Результат: + ``` text ┌─name───────────────────────────┐ │ aggregate_function_combinators │ -│ collations │ +│ asynchronous_metric_log │ └────────────────────────────────┘ ``` +### Смотрите также {#see-also} + +- [Create Tables](https://clickhouse.tech/docs/ru/getting-started/tutorial/#create-tables) +- [SHOW CREATE TABLE](https://clickhouse.tech/docs/ru/sql-reference/statements/show/#show-create-table) + ## SHOW DICTIONARIES {#show-dictionaries} Выводит список [внешних словарей](../../sql-reference/statements/show.md). diff --git a/docs/ru/sql-reference/statements/system.md b/docs/ru/sql-reference/statements/system.md index 1ba3a7555b9..4780e9b613f 100644 --- a/docs/ru/sql-reference/statements/system.md +++ b/docs/ru/sql-reference/statements/system.md @@ -1,3 +1,8 @@ +--- +toc_priority: 36 +toc_title: SYSTEM +--- + # Запросы SYSTEM {#query-language-system} - [RELOAD EMBEDDED DICTIONARIES](#query_language-system-reload-emdedded-dictionaries) diff --git a/docs/ru/sql-reference/statements/truncate.md b/docs/ru/sql-reference/statements/truncate.md index f8806496e48..4909d349658 100644 --- a/docs/ru/sql-reference/statements/truncate.md +++ b/docs/ru/sql-reference/statements/truncate.md @@ -1,5 +1,5 @@ --- -toc_priority: 53 +toc_priority: 51 toc_title: TRUNCATE --- diff --git a/docs/ru/sql-reference/statements/use.md b/docs/ru/sql-reference/statements/use.md index 84b0f5ed13f..c84329ea5ff 100644 --- a/docs/ru/sql-reference/statements/use.md +++ b/docs/ru/sql-reference/statements/use.md @@ -1,5 +1,5 @@ --- -toc_priority: 54 +toc_priority: 52 toc_title: USE --- diff --git a/docs/ru/sql-reference/syntax.md b/docs/ru/sql-reference/syntax.md index 5c819002cbe..ca73d3a137e 100644 --- a/docs/ru/sql-reference/syntax.md +++ b/docs/ru/sql-reference/syntax.md @@ -1,3 +1,8 @@ +--- +toc_priority: 31 +toc_title: "\u0421\u0438\u043d\u0442\u0430\u043a\u0441\u0438\u0441" +--- + # Синтаксис {#sintaksis} В системе есть два вида парсеров: полноценный парсер SQL (recursive descent parser) и парсер форматов данных (быстрый потоковый парсер). diff --git a/docs/ru/sql-reference/table-functions/file.md b/docs/ru/sql-reference/table-functions/file.md index 4581fa081d7..d3e6e106125 100644 --- a/docs/ru/sql-reference/table-functions/file.md +++ b/docs/ru/sql-reference/table-functions/file.md @@ -1,3 +1,8 @@ +--- +toc_priority: 37 +toc_title: file +--- + # file {#file} Создаёт таблицу из файла. Данная табличная функция похожа на табличные функции [file](file.md) и [hdfs](hdfs.md). diff --git a/docs/ru/sql-reference/table-functions/generate.md b/docs/ru/sql-reference/table-functions/generate.md index 9e6d36b2a4b..47b7e43bc86 100644 --- a/docs/ru/sql-reference/table-functions/generate.md +++ b/docs/ru/sql-reference/table-functions/generate.md @@ -1,3 +1,8 @@ +--- +toc_priority: 47 +toc_title: generateRandom +--- + # generateRandom {#generaterandom} Генерирует случайные данные с заданной схемой. diff --git a/docs/ru/sql-reference/table-functions/hdfs.md b/docs/ru/sql-reference/table-functions/hdfs.md index e90f27a2eca..6edd70b7b1b 100644 --- a/docs/ru/sql-reference/table-functions/hdfs.md +++ b/docs/ru/sql-reference/table-functions/hdfs.md @@ -1,3 +1,8 @@ +--- +toc_priority: 45 +toc_title: hdfs +--- + # hdfs {#hdfs} Создаёт таблицу из файла в HDFS. Данная табличная функция похожа на табличные функции [url](url.md) и [file](file.md). diff --git a/docs/ru/sql-reference/table-functions/input.md b/docs/ru/sql-reference/table-functions/input.md index 399268f9af6..96cf7515d52 100644 --- a/docs/ru/sql-reference/table-functions/input.md +++ b/docs/ru/sql-reference/table-functions/input.md @@ -1,3 +1,8 @@ +--- +toc_priority: 46 +toc_title: input +--- + # input {#input} `input(structure)` - табличная функция, позволяющая эффективно преобразовывать и вставлять отправленные на сервер данные, diff --git a/docs/ru/sql-reference/table-functions/jdbc.md b/docs/ru/sql-reference/table-functions/jdbc.md index 20ef0d1f107..d388262606f 100644 --- a/docs/ru/sql-reference/table-functions/jdbc.md +++ b/docs/ru/sql-reference/table-functions/jdbc.md @@ -1,3 +1,8 @@ +--- +toc_priority: 43 +toc_title: jdbc +--- + # jdbc {#jdbc} `jdbc(jdbc_connection_uri, schema, table)` - возвращает таблицу, соединение с которой происходит через JDBC-драйвер. diff --git a/docs/ru/sql-reference/table-functions/merge.md b/docs/ru/sql-reference/table-functions/merge.md index d4e66391382..0822fdfe535 100644 --- a/docs/ru/sql-reference/table-functions/merge.md +++ b/docs/ru/sql-reference/table-functions/merge.md @@ -1,3 +1,8 @@ +--- +toc_priority: 38 +toc_title: merge +--- + # merge {#merge} `merge(db_name, 'tables_regexp')` - создаёт временную таблицу типа Merge. Подробнее смотрите раздел «Движки таблиц, Merge». diff --git a/docs/ru/sql-reference/table-functions/mysql.md b/docs/ru/sql-reference/table-functions/mysql.md index 99d82022df4..21841eee67a 100644 --- a/docs/ru/sql-reference/table-functions/mysql.md +++ b/docs/ru/sql-reference/table-functions/mysql.md @@ -1,3 +1,8 @@ +--- +toc_priority: 42 +toc_title: mysql +--- + # mysql {#mysql} Позволяет выполнять запросы `SELECT` над данными, хранящимися на удалённом MySQL сервере. diff --git a/docs/ru/sql-reference/table-functions/numbers.md b/docs/ru/sql-reference/table-functions/numbers.md index 79d01dd0b92..005f400e082 100644 --- a/docs/ru/sql-reference/table-functions/numbers.md +++ b/docs/ru/sql-reference/table-functions/numbers.md @@ -1,3 +1,8 @@ +--- +toc_priority: 39 +toc_title: numbers +--- + # numbers {#numbers} `numbers(N)` - возвращает таблицу с единственным столбцом `number` (UInt64), содержащим натуральные числа от `0` до `N-1`. diff --git a/docs/ru/sql-reference/table-functions/odbc.md b/docs/ru/sql-reference/table-functions/odbc.md index 38da5066cbd..19203123840 100644 --- a/docs/ru/sql-reference/table-functions/odbc.md +++ b/docs/ru/sql-reference/table-functions/odbc.md @@ -1,3 +1,8 @@ +--- +toc_priority: 44 +toc_title: odbc +--- + # odbc {#table-functions-odbc} Возвращает таблицу, подключенную через [ODBC](https://en.wikipedia.org/wiki/Open_Database_Connectivity). diff --git a/docs/ru/sql-reference/table-functions/remote.md b/docs/ru/sql-reference/table-functions/remote.md index 944500b57a0..901317a805d 100644 --- a/docs/ru/sql-reference/table-functions/remote.md +++ b/docs/ru/sql-reference/table-functions/remote.md @@ -1,3 +1,8 @@ +--- +toc_priority: 40 +toc_title: remote +--- + # remote, remoteSecure {#remote-remotesecure} Позволяет обратиться к удалённым серверам без создания таблицы типа `Distributed`. diff --git a/docs/ru/sql-reference/table-functions/url.md b/docs/ru/sql-reference/table-functions/url.md index 4c49bc76751..0cd7c24c663 100644 --- a/docs/ru/sql-reference/table-functions/url.md +++ b/docs/ru/sql-reference/table-functions/url.md @@ -1,3 +1,8 @@ +--- +toc_priority: 41 +toc_title: url +--- + # url {#url} `url(URL, format, structure)` - возвращает таблицу со столбцами, указанными в diff --git a/docs/ru/sql-reference/table-functions/view.md b/docs/ru/sql-reference/table-functions/view.md new file mode 100644 index 00000000000..8a97253d048 --- /dev/null +++ b/docs/ru/sql-reference/table-functions/view.md @@ -0,0 +1,62 @@ +## view {#view} + +Преобразовывает подзапрос в таблицу. Функция реализовывает представления (смотрите [CREATE VIEW](https://clickhouse.tech/docs/ru/sql-reference/statements/create/view/#create-view)). Результирующая таблица не хранит данные, а только сохраняет указанный запрос `SELECT`. При чтении из таблицы, ClickHouse выполняет запрос и удаляет все ненужные столбцы из результата. + +**Синтаксис** + +``` sql +view(subquery) +``` + +**Входные параметры** + +- `subquery` — запрос `SELECT`. + +**Возвращаемое значение** + +- Таблица. + +**Пример** + +Входная таблица: + +``` text +┌─id─┬─name─────┬─days─┐ +│ 1 │ January │ 31 │ +│ 2 │ February │ 29 │ +│ 3 │ March │ 31 │ +│ 4 │ April │ 30 │ +└────┴──────────┴──────┘ +``` + +Запрос: + +``` sql +SELECT * FROM view(SELECT name FROM months) +``` + +Результат: + +``` text +┌─name─────┐ +│ January │ +│ February │ +│ March │ +│ April │ +└──────────┘ +``` + +Вы можете использовать функцию `view` как параметр табличных функций [remote](https://clickhouse.tech/docs/ru/sql-reference/table-functions/remote/#remote-remotesecure) и [cluster](https://clickhouse.tech/docs/ru/sql-reference/table-functions/cluster/#cluster-clusterallreplicas): + +``` sql +SELECT * FROM remote(`127.0.0.1`, view(SELECT a, b, c FROM table_name)) +``` + +``` sql +SELECT * FROM cluster(`cluster_name`, view(SELECT a, b, c FROM table_name)) +``` + +**Смотрите также** + +- [view](https://clickhouse.tech/docs/ru/engines/table-engines/special/view/#table_engines-view) +[Оригинальная статья](https://clickhouse.tech/docs/ru/query_language/table_functions/view/) \ No newline at end of file diff --git a/docs/ru/whats-new/index.md b/docs/ru/whats-new/index.md index a8c908088d0..b8ba6133454 100644 --- a/docs/ru/whats-new/index.md +++ b/docs/ru/whats-new/index.md @@ -1,6 +1,6 @@ --- toc_folder_title: "\u0427\u0442\u043E \u043D\u043E\u0432\u043E\u0433\u043E?" -toc_priority: 72 +toc_priority: 82 --- diff --git a/docs/ru/whats-new/security-changelog.md b/docs/ru/whats-new/security-changelog.md index ad55e1eeb51..1f46535833d 100644 --- a/docs/ru/whats-new/security-changelog.md +++ b/docs/ru/whats-new/security-changelog.md @@ -1,3 +1,8 @@ +--- +toc_priority: 76 +toc_title: Security Changelog +--- + # Security Changelog {#security-changelog} ## Исправлено в релизе 19.14.3.3, 2019-09-10 {#ispravleno-v-relize-19-14-3-3-2019-09-10} diff --git a/docs/tools/requirements.txt b/docs/tools/requirements.txt index 68a72a6a785..3c59a601ff2 100644 --- a/docs/tools/requirements.txt +++ b/docs/tools/requirements.txt @@ -14,7 +14,7 @@ Jinja2==2.11.2 jinja2-highlight==0.6.1 jsmin==2.2.2 livereload==2.6.2 -Markdown==3.2.1 +Markdown==3.3.2 MarkupSafe==1.1.1 mkdocs==1.1.2 mkdocs-htmlproofer-plugin==0.0.3 @@ -33,6 +33,6 @@ singledispatch==3.4.0.3 six==1.15.0 soupsieve==2.0.1 termcolor==1.1.0 -tornado==5.1.1 +tornado==6.1 Unidecode==1.1.1 urllib3==1.25.10 diff --git a/docs/zh/operations/system-tables/query_log.md b/docs/zh/operations/system-tables/query_log.md index bf3aa063a83..6d8d7a39699 100644 --- a/docs/zh/operations/system-tables/query_log.md +++ b/docs/zh/operations/system-tables/query_log.md @@ -3,7 +3,7 @@ machine_translated: true machine_translated_rev: 5decc73b5dc60054f19087d3690c4eb99446a6c3 --- -# 系统。query_log {#system_tables-query_log} +# system.query_log {#system_tables-query_log} 包含有关已执行查询的信息,例如,开始时间、处理持续时间、错误消息。 @@ -140,4 +140,4 @@ Settings.Values: ['0','random','1','10000000000'] **另请参阅** -- [系统。query_thread_log](../../operations/system-tables/query_thread_log.md#system_tables-query_thread_log) — This table contains information about each query execution thread. +- [system.query_thread_log](../../operations/system-tables/query_thread_log.md#system_tables-query_thread_log) — This table contains information about each query execution thread. diff --git a/programs/client/Client.cpp b/programs/client/Client.cpp index 07f2a231afe..005fece3277 100644 --- a/programs/client/Client.cpp +++ b/programs/client/Client.cpp @@ -218,6 +218,8 @@ private: QueryFuzzer fuzzer; int query_fuzzer_runs = 0; + std::optional suggest; + /// We will format query_id in interactive mode in various ways, the default is just to print Query id: ... std::vector> query_id_formats; @@ -577,10 +579,11 @@ private: if (print_time_to_stderr) throw Exception("time option could be specified only in non-interactive mode", ErrorCodes::BAD_ARGUMENTS); + suggest.emplace(); if (server_revision >= Suggest::MIN_SERVER_REVISION && !config().getBool("disable_suggestion", false)) { /// Load suggestion data from the server. - Suggest::instance().load(connection_parameters, config().getInt("suggestion_limit")); + suggest->load(connection_parameters, config().getInt("suggestion_limit")); } /// Load command history if present. @@ -607,7 +610,7 @@ private: highlight_callback = highlight; ReplxxLineReader lr( - Suggest::instance(), + *suggest, history_file, config().has("multiline"), query_extenders, @@ -615,7 +618,7 @@ private: highlight_callback); #elif defined(USE_READLINE) && USE_READLINE - ReadlineLineReader lr(Suggest::instance(), history_file, config().has("multiline"), query_extenders, query_delimiters); + ReadlineLineReader lr(*suggest, history_file, config().has("multiline"), query_extenders, query_delimiters); #else LineReader lr(history_file, config().has("multiline"), query_extenders, query_delimiters); #endif @@ -1499,7 +1502,7 @@ private: ASTPtr parseQuery(const char * & pos, const char * end, bool allow_multi_statements) { - ParserQuery parser(end, true); + ParserQuery parser(end); ASTPtr res; const auto & settings = context.getSettingsRef(); @@ -1896,7 +1899,7 @@ private: if (has_vertical_output_suffix) throw Exception("Output format already specified", ErrorCodes::CLIENT_OUTPUT_FORMAT_SPECIFIED); const auto & id = query_with_output->format->as(); - current_format = id.name; + current_format = id.name(); } } @@ -2324,6 +2327,8 @@ public: ("log-level", po::value(), "client log level") ("server_logs_file", po::value(), "put server logs into specified file") ("query-fuzzer-runs", po::value()->default_value(0), "query fuzzer runs") + ("opentelemetry-traceparent", po::value(), "OpenTelemetry traceparent header as described by W3C Trace Context recommendation") + ("opentelemetry-tracestate", po::value(), "OpenTelemetry tracestate header as described by W3C Trace Context recommendation") ; Settings cmd_settings; @@ -2492,6 +2497,25 @@ public: ignore_error = true; } + if (options.count("opentelemetry-traceparent")) + { + std::string traceparent = options["opentelemetry-traceparent"].as(); + std::string error; + if (!context.getClientInfo().parseTraceparentHeader( + traceparent, error)) + { + throw Exception(ErrorCodes::BAD_ARGUMENTS, + "Cannot parse OpenTelemetry traceparent '{}': {}", + traceparent, error); + } + } + + if (options.count("opentelemetry-tracestate")) + { + context.getClientInfo().opentelemetry_tracestate = + options["opentelemetry-tracestate"].as(); + } + argsToConfig(common_arguments, config(), 100); clearPasswordFromCommandLine(argc, argv); diff --git a/programs/client/Suggest.h b/programs/client/Suggest.h index b13289ac322..03332088cbe 100644 --- a/programs/client/Suggest.h +++ b/programs/client/Suggest.h @@ -18,10 +18,11 @@ namespace ErrorCodes class Suggest : public LineReader::Suggest, boost::noncopyable { public: - static Suggest & instance() + Suggest(); + ~Suggest() { - static Suggest instance; - return instance; + if (loading_thread.joinable()) + loading_thread.join(); } void load(const ConnectionParameters & connection_parameters, size_t suggestion_limit); @@ -30,12 +31,6 @@ public: static constexpr int MIN_SERVER_REVISION = 54406; private: - Suggest(); - ~Suggest() - { - if (loading_thread.joinable()) - loading_thread.join(); - } void loadImpl(Connection & connection, const ConnectionTimeouts & timeouts, size_t suggestion_limit); void fetch(Connection & connection, const ConnectionTimeouts & timeouts, const std::string & query); diff --git a/programs/copier/Internals.cpp b/programs/copier/Internals.cpp index 24b5d616af4..0f607ea5faf 100644 --- a/programs/copier/Internals.cpp +++ b/programs/copier/Internals.cpp @@ -168,11 +168,11 @@ ASTPtr extractOrderBy(const ASTPtr & storage_ast) throw Exception("ORDER BY cannot be empty", ErrorCodes::BAD_ARGUMENTS); } -/// Wraps only identifiers with backticks. +/// Wraps only identifiers with backticks. std::string wrapIdentifiersWithBackticks(const ASTPtr & root) { if (auto identifier = std::dynamic_pointer_cast(root)) - return backQuote(identifier->name); + return backQuote(identifier->name()); if (auto function = std::dynamic_pointer_cast(root)) return function->name + '(' + wrapIdentifiersWithBackticks(function->arguments) + ')'; @@ -214,7 +214,7 @@ Names extractPrimaryKeyColumnNames(const ASTPtr & storage_ast) for (size_t i = 0; i < sorting_key_size; ++i) { /// Column name could be represented as a f_1(f_2(...f_n(column_name))). - /// Each f_i could take one or more parameters. + /// Each f_i could take one or more parameters. /// We will wrap identifiers with backticks to allow non-standart identifier names. String sorting_key_column = sorting_key_expr_list->children[i]->getColumnName(); diff --git a/programs/install/Install.cpp b/programs/install/Install.cpp index ae0c22c8fcc..8290118089c 100644 --- a/programs/install/Install.cpp +++ b/programs/install/Install.cpp @@ -548,11 +548,27 @@ int mainEntryClickHouseInstall(int argc, char ** argv) users_config_file.string(), users_d.string()); } - /// Set capabilities for the binary. + /** Set capabilities for the binary. + * + * 1. Check that "setcap" tool exists. + * 2. Check that an arbitrary program with installed capabilities can run. + * 3. Set the capabilities. + * + * The second is important for Docker and systemd-nspawn. + * When the container has no capabilities, + * but the executable file inside the container has capabilities, + * then attempt to run this file will end up with a cryptic "Operation not permitted" message. + */ #if defined(__linux__) fmt::print("Setting capabilities for clickhouse binary. This is optional.\n"); - std::string command = fmt::format("command -v setcap && setcap 'cap_net_admin,cap_ipc_lock,cap_sys_nice+ep' {}", main_bin_path.string()); + std::string command = fmt::format("command -v setcap >/dev/null" + " && echo > {0} && chmod a+x {0} && {0} && setcap 'cap_net_admin,cap_ipc_lock,cap_sys_nice+ep' {0} && {0} && rm {0}" + " && setcap 'cap_net_admin,cap_ipc_lock,cap_sys_nice+ep' {1}" + " || echo \"Cannot set 'net_admin' or 'ipc_lock' or 'sys_nice' capability for clickhouse binary." + " This is optional. Taskstats accounting will be disabled." + " To enable taskstats accounting you may add the required capability later manually.\"", + "/tmp/test_setcap.sh", main_bin_path.string()); fmt::print(" {}\n", command); executeScript(command); #endif diff --git a/programs/local/LocalServer.cpp b/programs/local/LocalServer.cpp index 41da477152c..b9dde555788 100644 --- a/programs/local/LocalServer.cpp +++ b/programs/local/LocalServer.cpp @@ -37,6 +37,7 @@ #include #include #include +#include #include @@ -47,9 +48,9 @@ namespace DB namespace ErrorCodes { extern const int BAD_ARGUMENTS; - extern const int LOGICAL_ERROR; extern const int SYNTAX_ERROR; extern const int CANNOT_LOAD_CONFIG; + extern const int FILE_ALREADY_EXISTS; } @@ -57,8 +58,8 @@ LocalServer::LocalServer() = default; LocalServer::~LocalServer() { - if (context) - context->shutdown(); /// required for properly exception handling + if (global_context) + global_context->shutdown(); /// required for properly exception handling } @@ -95,9 +96,9 @@ void LocalServer::initialize(Poco::Util::Application & self) } } -void LocalServer::applyCmdSettings() +void LocalServer::applyCmdSettings(Context & context) { - context->applySettingsChanges(cmd_settings.changes()); + context.applySettingsChanges(cmd_settings.changes()); } /// If path is specified and not empty, will try to setup server environment and load existing metadata @@ -121,38 +122,54 @@ void LocalServer::tryInitPath() } else { - // Default unique path in the system temporary directory. - const auto tmp = std::filesystem::temp_directory_path(); - const auto default_path = tmp - / fmt::format("clickhouse-local-{}", getpid()); + // The path is not provided explicitly - use a unique path in the system temporary directory + // (or in the current dir if temporary don't exist) + Poco::Logger * log = &logger(); + std::filesystem::path parent_folder; + std::filesystem::path default_path; + + try + { + // try to guess a tmp folder name, and check if it's a directory (throw exception otherwise) + parent_folder = std::filesystem::temp_directory_path(); + + } + catch (const std::filesystem::filesystem_error& e) + { + // tmp folder don't exists? misconfiguration? chroot? + LOG_DEBUG(log, "Can not get temporary folder: {}", e.what()); + parent_folder = std::filesystem::current_path(); + + std::filesystem::is_directory(parent_folder); // that will throw an exception if it's not a directory + LOG_DEBUG(log, "Will create working directory inside current directory: {}", parent_folder.string()); + } + + /// we can have another clickhouse-local running simultaneously, even with the same PID (for ex. - several dockers mounting the same folder) + /// or it can be some leftovers from other clickhouse-local runs + /// as we can't accurately distinguish those situations we don't touch any existent folders + /// we just try to pick some free name for our working folder + + default_path = parent_folder / fmt::format("clickhouse-local-{}-{}-{}", getpid(), time(nullptr), randomSeed()); if (exists(default_path)) - { - // This is a directory that is left by a previous run of - // clickhouse-local that had the same pid and did not complete - // correctly. Remove it, with an additional sanity check. - if (!std::filesystem::equivalent(default_path.parent_path(), tmp)) - { - throw Exception(ErrorCodes::LOGICAL_ERROR, - "The temporary directory of clickhouse-local '{}' is not" - " inside the system temporary directory '{}'. Will not delete" - " it", default_path.string(), tmp.string()); - } - - remove_all(default_path); - } + throw Exception(ErrorCodes::FILE_ALREADY_EXISTS, "Unsuccessful attempt to create working directory: {} exist!", default_path.string()); create_directory(default_path); temporary_directory_to_delete = default_path; path = default_path.string(); + LOG_DEBUG(log, "Working directory created: {}", path); } if (path.back() != '/') path += '/'; - context->setPath(path); - context->setUserFilesPath(""); // user's files are everywhere + global_context->setPath(path); + + global_context->setTemporaryStorage(path + "tmp"); + global_context->setFlagsPath(path + "flags"); + + global_context->setUserFilesPath(""); // user's files are everywhere } @@ -186,9 +203,9 @@ try } shared_context = Context::createShared(); - context = std::make_unique(Context::createGlobal(shared_context.get())); - context->makeGlobalContext(); - context->setApplicationType(Context::ApplicationType::LOCAL); + global_context = std::make_unique(Context::createGlobal(shared_context.get())); + global_context->makeGlobalContext(); + global_context->setApplicationType(Context::ApplicationType::LOCAL); tryInitPath(); std::optional status; @@ -210,32 +227,32 @@ try /// Maybe useless if (config().has("macros")) - context->setMacros(std::make_unique(config(), "macros", log)); + global_context->setMacros(std::make_unique(config(), "macros", log)); /// Skip networking /// Sets external authenticators config (LDAP). - context->setExternalAuthenticatorsConfig(config()); + global_context->setExternalAuthenticatorsConfig(config()); setupUsers(); /// Limit on total number of concurrently executing queries. /// There is no need for concurrent queries, override max_concurrent_queries. - context->getProcessList().setMaxSize(0); + global_context->getProcessList().setMaxSize(0); /// Size of cache for uncompressed blocks. Zero means disabled. size_t uncompressed_cache_size = config().getUInt64("uncompressed_cache_size", 0); if (uncompressed_cache_size) - context->setUncompressedCache(uncompressed_cache_size); + global_context->setUncompressedCache(uncompressed_cache_size); /// Size of cache for marks (index of MergeTree family of tables). It is necessary. /// Specify default value for mark_cache_size explicitly! size_t mark_cache_size = config().getUInt64("mark_cache_size", 5368709120); if (mark_cache_size) - context->setMarkCache(mark_cache_size); + global_context->setMarkCache(mark_cache_size); /// Load global settings from default_profile and system_profile. - context->setDefaultProfiles(config()); + global_context->setDefaultProfiles(config()); /** Init dummy default DB * NOTE: We force using isolated default database to avoid conflicts with default database from server environment @@ -243,34 +260,34 @@ try * if such tables will not be dropped, clickhouse-server will not be able to load them due to security reasons. */ std::string default_database = config().getString("default_database", "_local"); - DatabaseCatalog::instance().attachDatabase(default_database, std::make_shared(default_database, *context)); - context->setCurrentDatabase(default_database); - applyCmdOptions(); + DatabaseCatalog::instance().attachDatabase(default_database, std::make_shared(default_database, *global_context)); + global_context->setCurrentDatabase(default_database); + applyCmdOptions(*global_context); - String path = context->getPath(); + String path = global_context->getPath(); if (!path.empty()) { /// Lock path directory before read - status.emplace(context->getPath() + "status", StatusFile::write_full_info); + status.emplace(global_context->getPath() + "status", StatusFile::write_full_info); LOG_DEBUG(log, "Loading metadata from {}", path); Poco::File(path + "data/").createDirectories(); Poco::File(path + "metadata/").createDirectories(); - loadMetadataSystem(*context); - attachSystemTables(*context); - loadMetadata(*context); + loadMetadataSystem(*global_context); + attachSystemTables(*global_context); + loadMetadata(*global_context); DatabaseCatalog::instance().loadDatabases(); LOG_DEBUG(log, "Loaded metadata."); } else { - attachSystemTables(*context); + attachSystemTables(*global_context); } processQueries(); - context->shutdown(); - context.reset(); + global_context->shutdown(); + global_context.reset(); status.reset(); cleanup(); @@ -323,7 +340,7 @@ void LocalServer::processQueries() String initial_create_query = getInitialCreateTableQuery(); String queries_str = initial_create_query + config().getRawString("query"); - const auto & settings = context->getSettingsRef(); + const auto & settings = global_context->getSettingsRef(); std::vector queries; auto parse_res = splitMultipartQuery(queries_str, queries, settings.max_query_size, settings.max_parser_depth); @@ -331,15 +348,19 @@ void LocalServer::processQueries() if (!parse_res.second) throw Exception("Cannot parse and execute the following part of query: " + String(parse_res.first), ErrorCodes::SYNTAX_ERROR); - context->makeSessionContext(); - context->makeQueryContext(); + /// we can't mutate global global_context (can lead to races, as it was already passed to some background threads) + /// so we can't reuse it safely as a query context and need a copy here + auto context = Context(*global_context); - context->setUser("default", "", Poco::Net::SocketAddress{}); - context->setCurrentQueryId(""); - applyCmdSettings(); + context.makeSessionContext(); + context.makeQueryContext(); + + context.setUser("default", "", Poco::Net::SocketAddress{}); + context.setCurrentQueryId(""); + applyCmdSettings(context); /// Use the same query_id (and thread group) for all queries - CurrentThread::QueryScope query_scope_holder(*context); + CurrentThread::QueryScope query_scope_holder(context); bool echo_queries = config().hasOption("echo") || config().hasOption("verbose"); std::exception_ptr exception; @@ -358,7 +379,7 @@ void LocalServer::processQueries() try { - executeQuery(read_buf, write_buf, /* allow_into_outfile = */ true, *context, {}); + executeQuery(read_buf, write_buf, /* allow_into_outfile = */ true, context, {}); } catch (...) { @@ -423,30 +444,19 @@ void LocalServer::setupUsers() } if (users_config) - context->setUsersConfig(users_config); + global_context->setUsersConfig(users_config); else throw Exception("Can't load config for users", ErrorCodes::CANNOT_LOAD_CONFIG); } void LocalServer::cleanup() { - // Delete the temporary directory if needed. Just in case, check that it is - // in the system temporary directory, not to delete user data if there is a - // bug. + // Delete the temporary directory if needed. if (temporary_directory_to_delete) { - const auto tmp = std::filesystem::temp_directory_path(); const auto dir = *temporary_directory_to_delete; temporary_directory_to_delete.reset(); - - if (!std::filesystem::equivalent(dir.parent_path(), tmp)) - { - throw Exception(ErrorCodes::LOGICAL_ERROR, - "The temporary directory of clickhouse-local '{}' is not inside" - " the system temporary directory '{}'. Will not delete it", - dir.string(), tmp.string()); - } - + LOG_DEBUG(&logger(), "Removing temporary directory: {}", dir.string()); remove_all(dir); } } @@ -577,10 +587,10 @@ void LocalServer::init(int argc, char ** argv) argsToConfig(arguments, config(), 100); } -void LocalServer::applyCmdOptions() +void LocalServer::applyCmdOptions(Context & context) { - context->setDefaultFormat(config().getString("output-format", config().getString("format", "TSV"))); - applyCmdSettings(); + context.setDefaultFormat(config().getString("output-format", config().getString("format", "TSV"))); + applyCmdSettings(context); } } diff --git a/programs/local/LocalServer.h b/programs/local/LocalServer.h index a8908754369..02778bd86cb 100644 --- a/programs/local/LocalServer.h +++ b/programs/local/LocalServer.h @@ -36,15 +36,15 @@ private: std::string getInitialCreateTableQuery(); void tryInitPath(); - void applyCmdOptions(); - void applyCmdSettings(); + void applyCmdOptions(Context & context); + void applyCmdSettings(Context & context); void processQueries(); void setupUsers(); void cleanup(); protected: SharedContextHolder shared_context; - std::unique_ptr context; + std::unique_ptr global_context; /// Settings specified via command line args Settings cmd_settings; diff --git a/programs/server/CMakeLists.txt b/programs/server/CMakeLists.txt index b3dcf1955fe..198d9081168 100644 --- a/programs/server/CMakeLists.txt +++ b/programs/server/CMakeLists.txt @@ -4,7 +4,7 @@ set(CLICKHOUSE_SERVER_SOURCES ) if (OS_LINUX) - set (LINK_CONFIG_LIB INTERFACE "-Wl,${WHOLE_ARCHIVE} $ -Wl,${NO_WHOLE_ARCHIVE}") + set (LINK_RESOURCE_LIB INTERFACE "-Wl,${WHOLE_ARCHIVE} $ -Wl,${NO_WHOLE_ARCHIVE}") endif () set (CLICKHOUSE_SERVER_LINK @@ -20,7 +20,7 @@ set (CLICKHOUSE_SERVER_LINK clickhouse_table_functions string_utils - ${LINK_CONFIG_LIB} + ${LINK_RESOURCE_LIB} PUBLIC daemon @@ -37,20 +37,20 @@ if (OS_LINUX) # 1. Allow to run the binary without download of any other files. # 2. Allow to implement "sudo clickhouse install" tool. - foreach(CONFIG_FILE config users embedded) - set(CONFIG_OBJ ${CONFIG_FILE}.o) - set(CONFIG_OBJS ${CONFIG_OBJS} ${CONFIG_OBJ}) + foreach(RESOURCE_FILE config.xml users.xml embedded.xml play.html) + set(RESOURCE_OBJ ${RESOURCE_FILE}.o) + set(RESOURCE_OBJS ${RESOURCE_OBJS} ${RESOURCE_OBJ}) # https://stackoverflow.com/questions/14776463/compile-and-add-an-object-file-from-a-binary-with-cmake - add_custom_command(OUTPUT ${CONFIG_OBJ} - COMMAND cd ${CMAKE_CURRENT_SOURCE_DIR} && ${OBJCOPY_PATH} -I binary ${OBJCOPY_ARCH_OPTIONS} ${CONFIG_FILE}.xml ${CMAKE_CURRENT_BINARY_DIR}/${CONFIG_OBJ} + add_custom_command(OUTPUT ${RESOURCE_OBJ} + COMMAND cd ${CMAKE_CURRENT_SOURCE_DIR} && ${OBJCOPY_PATH} -I binary ${OBJCOPY_ARCH_OPTIONS} ${RESOURCE_FILE} ${CMAKE_CURRENT_BINARY_DIR}/${RESOURCE_OBJ} COMMAND ${OBJCOPY_PATH} --rename-section .data=.rodata,alloc,load,readonly,data,contents - ${CMAKE_CURRENT_BINARY_DIR}/${CONFIG_OBJ} ${CMAKE_CURRENT_BINARY_DIR}/${CONFIG_OBJ}) + ${CMAKE_CURRENT_BINARY_DIR}/${RESOURCE_OBJ} ${CMAKE_CURRENT_BINARY_DIR}/${RESOURCE_OBJ}) - set_source_files_properties(${CONFIG_OBJ} PROPERTIES EXTERNAL_OBJECT true GENERATED true) - endforeach(CONFIG_FILE) + set_source_files_properties(${RESOURCE_OBJ} PROPERTIES EXTERNAL_OBJECT true GENERATED true) + endforeach(RESOURCE_FILE) - add_library(clickhouse_server_configs STATIC ${CONFIG_OBJS}) + add_library(clickhouse_server_configs STATIC ${RESOURCE_OBJS}) set_target_properties(clickhouse_server_configs PROPERTIES LINKER_LANGUAGE C) # whole-archive prevents symbols from being discarded for unknown reason diff --git a/programs/server/Server.cpp b/programs/server/Server.cpp index da5760acc09..b85cb5e75f2 100644 --- a/programs/server/Server.cpp +++ b/programs/server/Server.cpp @@ -258,7 +258,7 @@ int Server::main(const std::vector & /*args*/) Poco::Logger * log = &logger(); UseSSL use_ssl; - ThreadStatus thread_status; + MainThreadStatus::getInstance(); registerFunctions(); registerAggregateFunctions(); diff --git a/programs/server/config.xml b/programs/server/config.xml index 83921b12382..e17b59671af 100644 --- a/programs/server/config.xml +++ b/programs/server/config.xml @@ -212,22 +212,10 @@ /var/lib/clickhouse/user_files/ - - - - - users.xml - - - - /var/lib/clickhouse/access/ - - - - + - + + + + + users.xml + + + + /var/lib/clickhouse/access/ + + + + + default @@ -374,6 +392,22 @@ + + + true + + 127.0.0.1 + 9000 + + + + true + + 127.0.0.2 + 9000 + + + @@ -594,6 +628,31 @@ 60000 + + + + + engine MergeTree + partition by toYYYYMM(finish_date) + order by (finish_date, finish_time_us, trace_id) + + system + opentelemetry_span_log
+ 7500 +
+ + @@ -704,18 +763,22 @@ --> /var/lib/clickhouse/format_schemas/ - - hide SSN - \b\d{3}-\d{2}-\d{4}\b - 000-00-0000 + hide encrypt/decrypt arguments + ((?:aes_)?(?:encrypt|decrypt)(?:_mysql)?)\s*\(\s*(?:'(?:\\'|.)+'|.*?)\s*\) + + \1(???) - --> + + + ClickHouse Query + + + + + + + +
+ +
+
+ +
+
+ +  (Ctrl+Enter) + + 🌑🌞 +
+
+
+

+    
+

+

+ + + + diff --git a/src/Access/AccessControlManager.cpp b/src/Access/AccessControlManager.cpp index ecbb02966f0..a95d65ebb59 100644 --- a/src/Access/AccessControlManager.cpp +++ b/src/Access/AccessControlManager.cpp @@ -3,6 +3,7 @@ #include #include #include +#include #include #include #include @@ -136,7 +137,6 @@ AccessControlManager::AccessControlManager() AccessControlManager::~AccessControlManager() = default; - void AccessControlManager::setUsersConfig(const Poco::Util::AbstractConfiguration & users_config_) { auto storages = getStoragesPtr(); @@ -162,6 +162,7 @@ void AccessControlManager::addUsersConfigStorage(const String & storage_name_, c auto new_storage = std::make_shared(storage_name_, check_setting_name_function); new_storage->setConfig(users_config_); addStorage(new_storage); + LOG_DEBUG(getLogger(), "Added {} access storage '{}', path: {}", String(new_storage->getStorageType()), new_storage->getStorageName(), new_storage->getPath()); } void AccessControlManager::addUsersConfigStorage( @@ -194,6 +195,7 @@ void AccessControlManager::addUsersConfigStorage( auto new_storage = std::make_shared(storage_name_, check_setting_name_function); new_storage->load(users_config_path_, include_from_path_, preprocessed_dir_, get_zookeeper_function_); addStorage(new_storage); + LOG_DEBUG(getLogger(), "Added {} access storage '{}', path: {}", String(new_storage->getStorageType()), new_storage->getStorageName(), new_storage->getPath()); } void AccessControlManager::reloadUsersConfigs() @@ -237,7 +239,9 @@ void AccessControlManager::addDiskStorage(const String & storage_name_, const St } } } - addStorage(std::make_shared(storage_name_, directory_, readonly_)); + auto new_storage = std::make_shared(storage_name_, directory_, readonly_); + addStorage(new_storage); + LOG_DEBUG(getLogger(), "Added {} access storage '{}', path: {}", String(new_storage->getStorageType()), new_storage->getStorageName(), new_storage->getPath()); } @@ -249,7 +253,17 @@ void AccessControlManager::addMemoryStorage(const String & storage_name_) if (auto memory_storage = typeid_cast>(storage)) return; } - addStorage(std::make_shared(storage_name_)); + auto new_storage = std::make_shared(storage_name_); + addStorage(new_storage); + LOG_DEBUG(getLogger(), "Added {} access storage '{}'", String(new_storage->getStorageType()), new_storage->getStorageName()); +} + + +void AccessControlManager::addLDAPStorage(const String & storage_name_, const Poco::Util::AbstractConfiguration & config_, const String & prefix_) +{ + auto new_storage = std::make_shared(storage_name_, this, config_, prefix_); + addStorage(new_storage); + LOG_DEBUG(getLogger(), "Added {} access storage '{}', LDAP server name: {}", String(new_storage->getStorageType()), new_storage->getStorageName(), new_storage->getLDAPServerName()); } @@ -275,6 +289,8 @@ void AccessControlManager::addStoragesFromUserDirectoriesConfig( type = UsersConfigAccessStorage::STORAGE_TYPE; else if ((type == "local") || (type == "local_directory")) type = DiskAccessStorage::STORAGE_TYPE; + else if (type == "ldap") + type = LDAPAccessStorage::STORAGE_TYPE; String name = config.getString(prefix + ".name", type); @@ -295,6 +311,10 @@ void AccessControlManager::addStoragesFromUserDirectoriesConfig( bool readonly = config.getBool(prefix + ".readonly", false); addDiskStorage(name, path, readonly); } + else if (type == LDAPAccessStorage::STORAGE_TYPE) + { + addLDAPStorage(name, config, prefix); + } else throw Exception("Unknown storage type '" + type + "' at " + prefix + " in config", ErrorCodes::UNKNOWN_ELEMENT_IN_CONFIG); } @@ -346,7 +366,7 @@ UUID AccessControlManager::login(const String & user_name, const String & passwo void AccessControlManager::setExternalAuthenticatorsConfig(const Poco::Util::AbstractConfiguration & config) { - external_authenticators->setConfig(config, getLogger()); + external_authenticators->setConfiguration(config, getLogger()); } diff --git a/src/Access/AccessControlManager.h b/src/Access/AccessControlManager.h index 81a66ce8f1d..c960b330ee6 100644 --- a/src/Access/AccessControlManager.h +++ b/src/Access/AccessControlManager.h @@ -82,6 +82,9 @@ public: void addMemoryStorage(); void addMemoryStorage(const String & storage_name_); + /// Adds LDAPAccessStorage which allows querying remote LDAP server for user info. + void addLDAPStorage(const String & storage_name_, const Poco::Util::AbstractConfiguration & config_, const String & prefix_); + /// Adds storages from config. void addStoragesFromUserDirectoriesConfig(const Poco::Util::AbstractConfiguration & config, const String & key, diff --git a/src/Access/ContextAccess.cpp b/src/Access/ContextAccess.cpp index 61ab4c8002d..0459022cb1a 100644 --- a/src/Access/ContextAccess.cpp +++ b/src/Access/ContextAccess.cpp @@ -100,7 +100,7 @@ namespace if (res & alter_table) res |= alter_view; - /// CREATE TABLE (on any database/table) => CREATE_TEMPORARY_TABLE (global) + /// CREATE TABLE (on any database/table) => CREATE_TEMPORARY_TABLE (global) static const AccessFlags create_temporary_table = AccessType::CREATE_TEMPORARY_TABLE; if ((level == 0) && (max_flags_with_children & create_table)) res |= create_temporary_table; diff --git a/src/Access/ExternalAuthenticators.cpp b/src/Access/ExternalAuthenticators.cpp index a0c5fbf1a79..3ed1b21c3c2 100644 --- a/src/Access/ExternalAuthenticators.cpp +++ b/src/Access/ExternalAuthenticators.cpp @@ -156,7 +156,7 @@ void ExternalAuthenticators::reset() ldap_server_params.clear(); } -void ExternalAuthenticators::setConfig(const Poco::Util::AbstractConfiguration & config, Poco::Logger * log) +void ExternalAuthenticators::setConfiguration(const Poco::Util::AbstractConfiguration & config, Poco::Logger * log) { std::scoped_lock lock(mutex); reset(); diff --git a/src/Access/ExternalAuthenticators.h b/src/Access/ExternalAuthenticators.h index 7484996c472..7502409d817 100644 --- a/src/Access/ExternalAuthenticators.h +++ b/src/Access/ExternalAuthenticators.h @@ -26,7 +26,7 @@ class ExternalAuthenticators { public: void reset(); - void setConfig(const Poco::Util::AbstractConfiguration & config, Poco::Logger * log); + void setConfiguration(const Poco::Util::AbstractConfiguration & config, Poco::Logger * log); void setLDAPServerParams(const String & server, const LDAPServerParams & params); LDAPServerParams getLDAPServerParams(const String & server) const; diff --git a/src/Access/IAccessStorage.cpp b/src/Access/IAccessStorage.cpp index e5170221e18..58821e7de4b 100644 --- a/src/Access/IAccessStorage.cpp +++ b/src/Access/IAccessStorage.cpp @@ -14,6 +14,8 @@ namespace ErrorCodes extern const int ACCESS_ENTITY_ALREADY_EXISTS; extern const int ACCESS_ENTITY_NOT_FOUND; extern const int ACCESS_STORAGE_READONLY; + extern const int WRONG_PASSWORD; + extern const int IP_ADDRESS_NOT_ALLOWED; extern const int AUTHENTICATION_FAILED; extern const int LOGICAL_ERROR; } @@ -418,9 +420,21 @@ UUID IAccessStorage::login( const String & user_name, const String & password, const Poco::Net::IPAddress & address, - const ExternalAuthenticators & external_authenticators) const + const ExternalAuthenticators & external_authenticators, + bool replace_exception_with_cannot_authenticate) const { - return loginImpl(user_name, password, address, external_authenticators); + try + { + return loginImpl(user_name, password, address, external_authenticators); + } + catch (...) + { + if (!replace_exception_with_cannot_authenticate) + throw; + + tryLogCurrentException(getLogger(), user_name + ": Authentication failed"); + throwCannotAuthenticate(user_name); + } } @@ -434,11 +448,16 @@ UUID IAccessStorage::loginImpl( { if (auto user = tryRead(*id)) { - if (isPasswordCorrectImpl(*user, password, external_authenticators) && isAddressAllowedImpl(*user, address)) - return *id; + if (!isPasswordCorrectImpl(*user, password, external_authenticators)) + throwInvalidPassword(); + + if (!isAddressAllowedImpl(*user, address)) + throwAddressNotAllowed(address); + + return *id; } } - throwCannotAuthenticate(user_name); + throwNotFound(EntityType::USER, user_name); } @@ -554,10 +573,19 @@ void IAccessStorage::throwReadonlyCannotRemove(EntityType type, const String & n ErrorCodes::ACCESS_STORAGE_READONLY); } +void IAccessStorage::throwAddressNotAllowed(const Poco::Net::IPAddress & address) +{ + throw Exception("Connections from " + address.toString() + " are not allowed", ErrorCodes::IP_ADDRESS_NOT_ALLOWED); +} + +void IAccessStorage::throwInvalidPassword() +{ + throw Exception("Invalid password", ErrorCodes::WRONG_PASSWORD); +} void IAccessStorage::throwCannotAuthenticate(const String & user_name) { - /// We use the same message for all authentification failures because we don't want to give away any unnecessary information for security reasons, + /// We use the same message for all authentication failures because we don't want to give away any unnecessary information for security reasons, /// only the log will show the exact reason. throw Exception(user_name + ": Authentication failed: password is incorrect or there is no user with such name", ErrorCodes::AUTHENTICATION_FAILED); } diff --git a/src/Access/IAccessStorage.h b/src/Access/IAccessStorage.h index 5a86e817fb2..ecf6b260712 100644 --- a/src/Access/IAccessStorage.h +++ b/src/Access/IAccessStorage.h @@ -144,7 +144,7 @@ public: /// Finds an user, check its password and returns the ID of the user. /// Throws an exception if no such user or password is incorrect. - UUID login(const String & user_name, const String & password, const Poco::Net::IPAddress & address, const ExternalAuthenticators & external_authenticators) const; + UUID login(const String & user_name, const String & password, const Poco::Net::IPAddress & address, const ExternalAuthenticators & external_authenticators, bool replace_exception_with_cannot_authenticate = true) const; /// Returns the ID of an user who has logged in (maybe on another node). /// The function assumes that the password has been already checked somehow, so we can skip checking it now. @@ -182,6 +182,8 @@ protected: [[noreturn]] void throwReadonlyCannotInsert(EntityType type, const String & name) const; [[noreturn]] void throwReadonlyCannotUpdate(EntityType type, const String & name) const; [[noreturn]] void throwReadonlyCannotRemove(EntityType type, const String & name) const; + [[noreturn]] static void throwAddressNotAllowed(const Poco::Net::IPAddress & address); + [[noreturn]] static void throwInvalidPassword(); [[noreturn]] static void throwCannotAuthenticate(const String & user_name); using Notification = std::tuple; diff --git a/src/Access/LDAPAccessStorage.cpp b/src/Access/LDAPAccessStorage.cpp new file mode 100644 index 00000000000..b20ef3a39a9 --- /dev/null +++ b/src/Access/LDAPAccessStorage.cpp @@ -0,0 +1,319 @@ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + + +namespace DB +{ +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} + + +LDAPAccessStorage::LDAPAccessStorage(const String & storage_name_, AccessControlManager * access_control_manager_, const Poco::Util::AbstractConfiguration & config, const String & prefix) + : IAccessStorage(storage_name_) +{ + setConfiguration(access_control_manager_, config, prefix); +} + + +String LDAPAccessStorage::getLDAPServerName() const +{ + return ldap_server; +} + + +void LDAPAccessStorage::setConfiguration(AccessControlManager * access_control_manager_, const Poco::Util::AbstractConfiguration & config, const String & prefix) +{ + std::scoped_lock lock(mutex); + + // TODO: switch to passing config as a ConfigurationView and remove this extra prefix once a version of Poco with proper implementation is available. + const String prefix_str = (prefix.empty() ? "" : prefix + "."); + + const bool has_server = config.has(prefix_str + "server"); + const bool has_roles = config.has(prefix_str + "roles"); + + if (!has_server) + throw Exception("Missing 'server' field for LDAP user directory.", ErrorCodes::BAD_ARGUMENTS); + + const auto ldap_server_cfg = config.getString(prefix_str + "server"); + if (ldap_server_cfg.empty()) + throw Exception("Empty 'server' field for LDAP user directory.", ErrorCodes::BAD_ARGUMENTS); + + std::set roles_cfg; + if (has_roles) + { + Poco::Util::AbstractConfiguration::Keys role_names; + config.keys(prefix_str + "roles", role_names); + + // Currently, we only extract names of roles from the section names and assign them directly and unconditionally. + roles_cfg.insert(role_names.begin(), role_names.end()); + } + + access_control_manager = access_control_manager_; + ldap_server = ldap_server_cfg; + default_role_names.swap(roles_cfg); + roles_of_interest.clear(); + role_change_subscription = access_control_manager->subscribeForChanges( + [this] (const UUID & id, const AccessEntityPtr & entity) + { + return this->processRoleChange(id, entity); + } + ); + + /// Update `roles_of_interests` with initial values. + for (const auto & role_name : default_role_names) + { + if (auto role_id = access_control_manager->find(role_name)) + roles_of_interest.emplace(*role_id, role_name); + } +} + + +void LDAPAccessStorage::processRoleChange(const UUID & id, const AccessEntityPtr & entity) +{ + std::scoped_lock lock(mutex); + + /// Update `roles_of_interests`. + auto role = typeid_cast>(entity); + bool need_to_update_users = false; + + if (role && default_role_names.count(role->getName())) + { + /// If a role was created with one of the `default_role_names` or renamed to one of the `default_role_names`, + /// then set `need_to_update_users`. + need_to_update_users = roles_of_interest.insert_or_assign(id, role->getName()).second; + } + else + { + /// If a role was removed or renamed to a name which isn't contained in the `default_role_names`, + /// then set `need_to_update_users`. + need_to_update_users = roles_of_interest.erase(id) > 0; + } + + /// Update users which have been created. + if (need_to_update_users) + { + auto update_func = [this] (const AccessEntityPtr & entity_) -> AccessEntityPtr + { + if (auto user = typeid_cast>(entity_)) + { + auto changed_user = typeid_cast>(user->clone()); + auto & granted_roles = changed_user->granted_roles.roles; + granted_roles.clear(); + boost::range::copy(roles_of_interest | boost::adaptors::map_keys, std::inserter(granted_roles, granted_roles.end())); + return changed_user; + } + return entity_; + }; + memory_storage.update(memory_storage.findAll(), update_func); + } +} + + +void LDAPAccessStorage::checkAllDefaultRoleNamesFoundNoLock() const +{ + boost::container::flat_set role_names_of_interest; + boost::range::copy(roles_of_interest | boost::adaptors::map_values, std::inserter(role_names_of_interest, role_names_of_interest.end())); + + for (const auto & role_name : default_role_names) + { + if (!role_names_of_interest.count(role_name)) + throwDefaultRoleNotFound(role_name); + } +} + + +const char * LDAPAccessStorage::getStorageType() const +{ + return STORAGE_TYPE; +} + + +String LDAPAccessStorage::getStorageParamsJSON() const +{ + std::scoped_lock lock(mutex); + Poco::JSON::Object params_json; + + params_json.set("server", ldap_server); + params_json.set("roles", default_role_names); + + std::ostringstream oss; + Poco::JSON::Stringifier::stringify(params_json, oss); + + return oss.str(); +} + + +std::optional LDAPAccessStorage::findImpl(EntityType type, const String & name) const +{ + std::scoped_lock lock(mutex); + return memory_storage.find(type, name); +} + + +std::vector LDAPAccessStorage::findAllImpl(EntityType type) const +{ + std::scoped_lock lock(mutex); + return memory_storage.findAll(type); +} + + +bool LDAPAccessStorage::existsImpl(const UUID & id) const +{ + std::scoped_lock lock(mutex); + return memory_storage.exists(id); +} + + +AccessEntityPtr LDAPAccessStorage::readImpl(const UUID & id) const +{ + std::scoped_lock lock(mutex); + return memory_storage.read(id); +} + + +String LDAPAccessStorage::readNameImpl(const UUID & id) const +{ + std::scoped_lock lock(mutex); + return memory_storage.readName(id); +} + + +bool LDAPAccessStorage::canInsertImpl(const AccessEntityPtr &) const +{ + return false; +} + + +UUID LDAPAccessStorage::insertImpl(const AccessEntityPtr & entity, bool) +{ + throwReadonlyCannotInsert(entity->getType(), entity->getName()); +} + + +void LDAPAccessStorage::removeImpl(const UUID & id) +{ + std::scoped_lock lock(mutex); + auto entity = read(id); + throwReadonlyCannotRemove(entity->getType(), entity->getName()); +} + + +void LDAPAccessStorage::updateImpl(const UUID & id, const UpdateFunc &) +{ + std::scoped_lock lock(mutex); + auto entity = read(id); + throwReadonlyCannotUpdate(entity->getType(), entity->getName()); +} + + +ext::scope_guard LDAPAccessStorage::subscribeForChangesImpl(const UUID & id, const OnChangedHandler & handler) const +{ + std::scoped_lock lock(mutex); + return memory_storage.subscribeForChanges(id, handler); +} + + +ext::scope_guard LDAPAccessStorage::subscribeForChangesImpl(EntityType type, const OnChangedHandler & handler) const +{ + std::scoped_lock lock(mutex); + return memory_storage.subscribeForChanges(type, handler); +} + + +bool LDAPAccessStorage::hasSubscriptionImpl(const UUID & id) const +{ + std::scoped_lock lock(mutex); + return memory_storage.hasSubscription(id); +} + + +bool LDAPAccessStorage::hasSubscriptionImpl(EntityType type) const +{ + std::scoped_lock lock(mutex); + return memory_storage.hasSubscription(type); +} + +UUID LDAPAccessStorage::loginImpl(const String & user_name, const String & password, const Poco::Net::IPAddress & address, const ExternalAuthenticators & external_authenticators) const +{ + std::scoped_lock lock(mutex); + auto id = memory_storage.find(user_name); + if (id) + { + auto user = memory_storage.read(*id); + + if (!isPasswordCorrectImpl(*user, password, external_authenticators)) + throwInvalidPassword(); + + if (!isAddressAllowedImpl(*user, address)) + throwAddressNotAllowed(address); + + return *id; + } + else + { + // User does not exist, so we create one, and will add it if authentication is successful. + auto user = std::make_shared(); + user->setName(user_name); + user->authentication = Authentication(Authentication::Type::LDAP_SERVER); + user->authentication.setServerName(ldap_server); + + if (!isPasswordCorrectImpl(*user, password, external_authenticators)) + throwInvalidPassword(); + + if (!isAddressAllowedImpl(*user, address)) + throwAddressNotAllowed(address); + + checkAllDefaultRoleNamesFoundNoLock(); + + auto & granted_roles = user->granted_roles.roles; + boost::range::copy(roles_of_interest | boost::adaptors::map_keys, std::inserter(granted_roles, granted_roles.end())); + + return memory_storage.insert(user); + } +} + +UUID LDAPAccessStorage::getIDOfLoggedUserImpl(const String & user_name) const +{ + std::scoped_lock lock(mutex); + auto id = memory_storage.find(user_name); + if (id) + { + return *id; + } + else + { + // User does not exist, so we create one, and add it pretending that the authentication is successful. + auto user = std::make_shared(); + user->setName(user_name); + user->authentication = Authentication(Authentication::Type::LDAP_SERVER); + user->authentication.setServerName(ldap_server); + + checkAllDefaultRoleNamesFoundNoLock(); + + auto & granted_roles = user->granted_roles.roles; + boost::range::copy(roles_of_interest | boost::adaptors::map_keys, std::inserter(granted_roles, granted_roles.end())); + + return memory_storage.insert(user); + } +} + +void LDAPAccessStorage::throwDefaultRoleNotFound(const String & role_name) +{ + throw Exception("One of the default roles, the role '" + role_name + "', is not found", IAccessEntity::TypeInfo::get(IAccessEntity::Type::ROLE).not_found_error_code); +} + +} diff --git a/src/Access/LDAPAccessStorage.h b/src/Access/LDAPAccessStorage.h new file mode 100644 index 00000000000..b1b0001d9bc --- /dev/null +++ b/src/Access/LDAPAccessStorage.h @@ -0,0 +1,73 @@ +#pragma once + +#include +#include +#include +#include +#include +#include + + +namespace Poco +{ + namespace Util + { + class AbstractConfiguration; + } +} + + +namespace DB +{ +class AccessControlManager; + +/// Implementation of IAccessStorage which allows attaching users from a remote LDAP server. +/// Currently, any user name will be treated as a name of an existing remote user, +/// a user info entity will be created, with LDAP_SERVER authentication type. +class LDAPAccessStorage : public IAccessStorage +{ +public: + static constexpr char STORAGE_TYPE[] = "ldap"; + + explicit LDAPAccessStorage(const String & storage_name_, AccessControlManager * access_control_manager_, const Poco::Util::AbstractConfiguration & config, const String & prefix); + virtual ~LDAPAccessStorage() override = default; + + String getLDAPServerName() const; + +public: // IAccessStorage implementations. + virtual const char * getStorageType() const override; + virtual String getStorageParamsJSON() const override; + +private: // IAccessStorage implementations. + virtual std::optional findImpl(EntityType type, const String & name) const override; + virtual std::vector findAllImpl(EntityType type) const override; + virtual bool existsImpl(const UUID & id) const override; + virtual AccessEntityPtr readImpl(const UUID & id) const override; + virtual String readNameImpl(const UUID & id) const override; + virtual bool canInsertImpl(const AccessEntityPtr &) const override; + virtual UUID insertImpl(const AccessEntityPtr & entity, bool replace_if_exists) override; + virtual void removeImpl(const UUID & id) override; + virtual void updateImpl(const UUID & id, const UpdateFunc & update_func) override; + virtual ext::scope_guard subscribeForChangesImpl(const UUID & id, const OnChangedHandler & handler) const override; + virtual ext::scope_guard subscribeForChangesImpl(EntityType type, const OnChangedHandler & handler) const override; + virtual bool hasSubscriptionImpl(const UUID & id) const override; + virtual bool hasSubscriptionImpl(EntityType type) const override; + virtual UUID loginImpl(const String & user_name, const String & password, const Poco::Net::IPAddress & address, const ExternalAuthenticators & external_authenticators) const override; + virtual UUID getIDOfLoggedUserImpl(const String & user_name) const override; + +private: + void setConfiguration(AccessControlManager * access_control_manager_, const Poco::Util::AbstractConfiguration & config, const String & prefix); + void processRoleChange(const UUID & id, const AccessEntityPtr & entity); + void checkAllDefaultRoleNamesFoundNoLock() const; + + [[noreturn]] static void throwDefaultRoleNotFound(const String & role_name); + + mutable std::recursive_mutex mutex; + AccessControlManager * access_control_manager = nullptr; + String ldap_server; + std::set default_role_names; + std::map roles_of_interest; + ext::scope_guard role_change_subscription; + mutable MemoryAccessStorage memory_storage; +}; +} diff --git a/src/Access/LDAPClient.cpp b/src/Access/LDAPClient.cpp index a85e96ab86c..a3223902361 100644 --- a/src/Access/LDAPClient.cpp +++ b/src/Access/LDAPClient.cpp @@ -2,6 +2,8 @@ #include #include +#include + #include #include @@ -27,16 +29,13 @@ LDAPClient::~LDAPClient() closeConnection(); } -void LDAPClient::openConnection() -{ - const bool graceful_bind_failure = false; - diag(openConnection(graceful_bind_failure)); -} - #if USE_LDAP namespace { + + std::recursive_mutex ldap_global_mutex; + auto escapeForLDAP(const String & src) { String dest; @@ -63,10 +62,13 @@ namespace return dest; } + } void LDAPClient::diag(const int rc) { + std::scoped_lock lock(ldap_global_mutex); + if (rc != LDAP_SUCCESS) { String text; @@ -100,8 +102,10 @@ void LDAPClient::diag(const int rc) } } -int LDAPClient::openConnection(const bool graceful_bind_failure) +void LDAPClient::openConnection() { + std::scoped_lock lock(ldap_global_mutex); + closeConnection(); { @@ -232,8 +236,6 @@ int LDAPClient::openConnection(const bool graceful_bind_failure) if (params.enable_tls == LDAPServerParams::TLSEnable::YES_STARTTLS) diag(ldap_start_tls_s(handle, nullptr, nullptr)); - int rc = LDAP_OTHER; - switch (params.sasl_mechanism) { case LDAPServerParams::SASLMechanism::SIMPLE: @@ -244,20 +246,21 @@ int LDAPClient::openConnection(const bool graceful_bind_failure) cred.bv_val = const_cast(params.password.c_str()); cred.bv_len = params.password.size(); - rc = ldap_sasl_bind_s(handle, dn.c_str(), LDAP_SASL_SIMPLE, &cred, nullptr, nullptr, nullptr); - - if (!graceful_bind_failure) - diag(rc); + diag(ldap_sasl_bind_s(handle, dn.c_str(), LDAP_SASL_SIMPLE, &cred, nullptr, nullptr, nullptr)); break; } + default: + { + throw Exception("Unknown SASL mechanism", ErrorCodes::LDAP_ERROR); + } } - - return rc; } void LDAPClient::closeConnection() noexcept { + std::scoped_lock lock(ldap_global_mutex); + if (!handle) return; @@ -267,42 +270,21 @@ void LDAPClient::closeConnection() noexcept bool LDAPSimpleAuthClient::check() { - if (params.user.empty()) - throw Exception("LDAP authentication of a user with an empty name is not allowed", ErrorCodes::BAD_ARGUMENTS); + std::scoped_lock lock(ldap_global_mutex); + if (params.user.empty()) + throw Exception("LDAP authentication of a user with empty name is not allowed", ErrorCodes::BAD_ARGUMENTS); + + // Silently reject authentication attempt if the password is empty as if it didn't match. if (params.password.empty()) - return false; // Silently reject authentication attempt if the password is empty as if it didn't match. + return false; SCOPE_EXIT({ closeConnection(); }); - const bool graceful_bind_failure = true; - const auto rc = openConnection(graceful_bind_failure); + // Will throw on any error, including invalid credentials. + openConnection(); - bool result = false; - - switch (rc) - { - case LDAP_SUCCESS: - { - result = true; - break; - } - - case LDAP_INVALID_CREDENTIALS: - { - result = false; - break; - } - - default: - { - result = false; - diag(rc); - break; - } - } - - return result; + return true; } #else // USE_LDAP @@ -312,7 +294,7 @@ void LDAPClient::diag(const int) throw Exception("ClickHouse was built without LDAP support", ErrorCodes::FEATURE_IS_NOT_ENABLED_AT_BUILD_TIME); } -int LDAPClient::openConnection(const bool) +void LDAPClient::openConnection() { throw Exception("ClickHouse was built without LDAP support", ErrorCodes::FEATURE_IS_NOT_ENABLED_AT_BUILD_TIME); } diff --git a/src/Access/LDAPClient.h b/src/Access/LDAPClient.h index b117ed9a026..777c87c5b94 100644 --- a/src/Access/LDAPClient.h +++ b/src/Access/LDAPClient.h @@ -32,7 +32,6 @@ public: protected: MAYBE_NORETURN void diag(const int rc); MAYBE_NORETURN void openConnection(); - int openConnection(const bool graceful_bind_failure = false); void closeConnection() noexcept; protected: diff --git a/src/Access/LDAPParams.h b/src/Access/LDAPParams.h index 2168ce45203..eeadba6bc01 100644 --- a/src/Access/LDAPParams.h +++ b/src/Access/LDAPParams.h @@ -42,6 +42,7 @@ struct LDAPServerParams enum class SASLMechanism { + UNKNOWN, SIMPLE }; diff --git a/src/Access/MemoryAccessStorage.cpp b/src/Access/MemoryAccessStorage.cpp index 114c8eb8bee..58280c8b1d7 100644 --- a/src/Access/MemoryAccessStorage.cpp +++ b/src/Access/MemoryAccessStorage.cpp @@ -69,7 +69,7 @@ UUID MemoryAccessStorage::insertImpl(const AccessEntityPtr & new_entity, bool re UUID id = generateRandomID(); std::lock_guard lock{mutex}; - insertNoLock(generateRandomID(), new_entity, replace_if_exists, notifications); + insertNoLock(id, new_entity, replace_if_exists, notifications); return id; } diff --git a/src/Access/MultipleAccessStorage.cpp b/src/Access/MultipleAccessStorage.cpp index 8ddc7410d8d..a8ce3f602ed 100644 --- a/src/Access/MultipleAccessStorage.cpp +++ b/src/Access/MultipleAccessStorage.cpp @@ -2,6 +2,7 @@ #include #include #include +#include #include #include @@ -27,6 +28,15 @@ MultipleAccessStorage::MultipleAccessStorage(const String & storage_name_) { } +MultipleAccessStorage::~MultipleAccessStorage() +{ + /// It's better to remove the storages in the reverse order because they could depend on each other somehow. + const auto storages = getStoragesPtr(); + for (const auto & storage : *storages | boost::adaptors::reversed) + { + removeStorage(storage); + } +} void MultipleAccessStorage::setStorages(const std::vector & storages) { @@ -400,7 +410,7 @@ UUID MultipleAccessStorage::loginImpl(const String & user_name, const String & p { try { - auto id = storage->login(user_name, password, address, external_authenticators); + auto id = storage->login(user_name, password, address, external_authenticators, /* replace_exception_with_cannot_authenticate = */ false); std::lock_guard lock{mutex}; ids_cache.set(id, storage); return id; @@ -416,7 +426,7 @@ UUID MultipleAccessStorage::loginImpl(const String & user_name, const String & p throw; } } - throwCannotAuthenticate(user_name); + throwNotFound(EntityType::USER, user_name); } diff --git a/src/Access/MultipleAccessStorage.h b/src/Access/MultipleAccessStorage.h index 36551f1cbc8..8844de8c029 100644 --- a/src/Access/MultipleAccessStorage.h +++ b/src/Access/MultipleAccessStorage.h @@ -18,6 +18,7 @@ public: using ConstStoragePtr = std::shared_ptr; MultipleAccessStorage(const String & storage_name_ = STORAGE_TYPE); + ~MultipleAccessStorage() override; const char * getStorageType() const override { return STORAGE_TYPE; } diff --git a/src/Access/ya.make b/src/Access/ya.make index e5fa73f107c..1ec8cb32c97 100644 --- a/src/Access/ya.make +++ b/src/Access/ya.make @@ -5,7 +5,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( AccessControlManager.cpp @@ -24,6 +23,7 @@ SRCS( GrantedRoles.cpp IAccessEntity.cpp IAccessStorage.cpp + LDAPAccessStorage.cpp LDAPClient.cpp MemoryAccessStorage.cpp MultipleAccessStorage.cpp diff --git a/src/Access/ya.make.in b/src/Access/ya.make.in index e48d0d1bda7..ce7cd88b272 100644 --- a/src/Access/ya.make.in +++ b/src/Access/ya.make.in @@ -4,7 +4,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( diff --git a/src/AggregateFunctions/AggregateFunctionBoundingRatio.cpp b/src/AggregateFunctions/AggregateFunctionBoundingRatio.cpp index e338b060b12..6f4f254ae8f 100644 --- a/src/AggregateFunctions/AggregateFunctionBoundingRatio.cpp +++ b/src/AggregateFunctions/AggregateFunctionBoundingRatio.cpp @@ -31,7 +31,7 @@ AggregateFunctionPtr createAggregateFunctionRate(const std::string & name, const void registerAggregateFunctionRate(AggregateFunctionFactory & factory) { - factory.registerFunction("boundingRatio", createAggregateFunctionRate, AggregateFunctionFactory::CaseInsensitive); + factory.registerFunction("boundingRatio", createAggregateFunctionRate); } } diff --git a/src/AggregateFunctions/AggregateFunctionCount.cpp b/src/AggregateFunctions/AggregateFunctionCount.cpp index 05824947b87..6ea63bedaf0 100644 --- a/src/AggregateFunctions/AggregateFunctionCount.cpp +++ b/src/AggregateFunctions/AggregateFunctionCount.cpp @@ -8,7 +8,7 @@ namespace DB { AggregateFunctionPtr AggregateFunctionCount::getOwnNullAdapter( - const AggregateFunctionPtr &, const DataTypes & types, const Array & params, const AggregateFunctionProperties & /*properties*/) const + const AggregateFunctionPtr &, const DataTypes & types, const Array & params) const { return std::make_shared(types[0], params); } diff --git a/src/AggregateFunctions/AggregateFunctionCount.h b/src/AggregateFunctions/AggregateFunctionCount.h index eb1583df92a..29c5de0021c 100644 --- a/src/AggregateFunctions/AggregateFunctionCount.h +++ b/src/AggregateFunctions/AggregateFunctionCount.h @@ -69,7 +69,7 @@ public: } AggregateFunctionPtr getOwnNullAdapter( - const AggregateFunctionPtr &, const DataTypes & types, const Array & params, const AggregateFunctionProperties & /*properties*/) const override; + const AggregateFunctionPtr &, const DataTypes & types, const Array & params) const override; }; diff --git a/src/AggregateFunctions/AggregateFunctionGroupArray.h b/src/AggregateFunctions/AggregateFunctionGroupArray.h index 83e096c797b..02b9003eb96 100644 --- a/src/AggregateFunctions/AggregateFunctionGroupArray.h +++ b/src/AggregateFunctions/AggregateFunctionGroupArray.h @@ -296,7 +296,7 @@ public: { typename ColumnVector::Container & data_to = assert_cast &>(arr_to.getData()).getData(); if constexpr (is_big_int_v) - // is data_to empty? we should probaly use std::vector::insert then + // is data_to empty? we should probably use std::vector::insert then for (auto it = this->data(place).value.begin(); it != this->data(place).value.end(); it++) data_to.push_back(*it); else diff --git a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h index d457d01f523..d80e5e81f19 100644 --- a/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h +++ b/src/AggregateFunctions/AggregateFunctionGroupBitmapData.h @@ -13,7 +13,6 @@ // this one: https://github.com/RoaringBitmap/CRoaring/blob/master/include/roaring/roaring.h #include - namespace DB { /** @@ -599,128 +598,6 @@ public: } } -private: - /// To read and write the DB Buffer directly, migrate code from CRoaring - void db_roaring_bitmap_add_many(DB::ReadBuffer & db_buf, roaring_bitmap_t * r, size_t n_args) - { - void * container = nullptr; // hold value of last container touched - uint8_t typecode = 0; // typecode of last container touched - uint32_t prev = 0; // previous valued inserted - size_t i = 0; // index of value - int containerindex = 0; - if (n_args == 0) - return; - uint32_t val; - readBinary(val, db_buf); - container = containerptr_roaring_bitmap_add(r, val, &typecode, &containerindex); - prev = val; - ++i; - for (; i < n_args; ++i) - { - readBinary(val, db_buf); - if (((prev ^ val) >> 16) == 0) - { // no need to seek the container, it is at hand - // because we already have the container at hand, we can do the - // insertion - // automatically, bypassing the roaring_bitmap_add call - uint8_t newtypecode = typecode; - void * container2 = container_add(container, val & 0xFFFF, typecode, &newtypecode); - // rare instance when we need to - if (container2 != container) - { - // change the container type - container_free(container, typecode); - ra_set_container_at_index(&r->high_low_container, containerindex, container2, newtypecode); - typecode = newtypecode; - container = container2; - } - } - else - { - container = containerptr_roaring_bitmap_add(r, val, &typecode, &containerindex); - } - prev = val; - } - } - - void db_ra_to_uint32_array(DB::WriteBuffer & db_buf, roaring_array_t * ra) const - { - size_t ctr = 0; - for (Int32 i = 0; i < ra->size; ++i) - { - Int32 num_added = db_container_to_uint32_array(db_buf, ra->containers[i], ra->typecodes[i], (static_cast(ra->keys[i])) << 16); - ctr += num_added; - } - } - - UInt32 db_container_to_uint32_array(DB::WriteBuffer & db_buf, const void * container, uint8_t typecode, UInt32 base) const - { - container = container_unwrap_shared(container, &typecode); - switch (typecode) - { - case BITSET_CONTAINER_TYPE_CODE: - return db_bitset_container_to_uint32_array(db_buf, static_cast(container), base); - case ARRAY_CONTAINER_TYPE_CODE: - return db_array_container_to_uint32_array(db_buf, static_cast(container), base); - case RUN_CONTAINER_TYPE_CODE: - return db_run_container_to_uint32_array(db_buf, static_cast(container), base); - } - return 0; - } - - UInt32 db_bitset_container_to_uint32_array(DB::WriteBuffer & db_buf, const bitset_container_t * cont, UInt32 base) const - { - return static_cast(db_bitset_extract_setbits(db_buf, cont->array, BITSET_CONTAINER_SIZE_IN_WORDS, base)); - } - - size_t db_bitset_extract_setbits(DB::WriteBuffer & db_buf, UInt64 * bitset, size_t length, UInt32 base) const - { - UInt32 outpos = 0; - for (size_t i = 0; i < length; ++i) - { - UInt64 w = bitset[i]; - while (w != 0) - { - UInt64 t = w & (~w + 1); // on x64, should compile to BLSI (careful: the Intel compiler seems to fail) - UInt32 r = __builtin_ctzll(w); // on x64, should compile to TZCNT - UInt32 val = r + base; - writePODBinary(val, db_buf); - outpos++; - w ^= t; - } - base += 64; - } - return outpos; - } - - int db_array_container_to_uint32_array(DB::WriteBuffer & db_buf, const array_container_t * cont, UInt32 base) const - { - UInt32 outpos = 0; - for (Int32 i = 0; i < cont->cardinality; ++i) - { - const UInt32 val = base + cont->array[i]; - writePODBinary(val, db_buf); - outpos++; - } - return outpos; - } - - int db_run_container_to_uint32_array(DB::WriteBuffer & db_buf, const run_container_t * cont, UInt32 base) const - { - UInt32 outpos = 0; - for (Int32 i = 0; i < cont->n_runs; ++i) - { - UInt32 run_start = base + cont->runs[i].value; - UInt16 le = cont->runs[i].length; - for (Int32 j = 0; j <= le; ++j) - { - UInt32 val = run_start + j; - writePODBinary(val, db_buf); - outpos++; - } - } - return outpos; - } }; template diff --git a/src/AggregateFunctions/AggregateFunctionIf.cpp b/src/AggregateFunctions/AggregateFunctionIf.cpp index 47afddaf7ff..19a175de911 100644 --- a/src/AggregateFunctions/AggregateFunctionIf.cpp +++ b/src/AggregateFunctions/AggregateFunctionIf.cpp @@ -1,7 +1,6 @@ #include #include #include "registerAggregateFunctions.h" -#include "AggregateFunctionNull.h" namespace DB @@ -9,7 +8,6 @@ namespace DB namespace ErrorCodes { - extern const int LOGICAL_ERROR; extern const int ILLEGAL_TYPE_OF_ARGUMENT; extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH; } @@ -42,164 +40,6 @@ public: } }; -/** There are two cases: for single argument and variadic. - * Code for single argument is much more efficient. - */ -template -class AggregateFunctionIfNullUnary final - : public AggregateFunctionNullBase> -{ -private: - size_t num_arguments; - - using Base = AggregateFunctionNullBase>; -public: - - String getName() const override - { - return Base::getName() + "If"; - } - - AggregateFunctionIfNullUnary(AggregateFunctionPtr nested_function_, const DataTypes & arguments, const Array & params) - : Base(std::move(nested_function_), arguments, params), num_arguments(arguments.size()) - { - if (num_arguments == 0) - throw Exception("Aggregate function " + getName() + " require at least one argument", - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); - } - - static inline bool singleFilter(const IColumn ** columns, size_t row_num, size_t num_arguments) - { - const IColumn * filter_column = columns[num_arguments - 1]; - if (const ColumnNullable * nullable_column = typeid_cast(filter_column)) - filter_column = nullable_column->getNestedColumnPtr().get(); - - return assert_cast(*filter_column).getData()[row_num]; - } - - void add(AggregateDataPtr place, const IColumn ** columns, size_t row_num, Arena * arena) const override - { - const ColumnNullable * column = assert_cast(columns[0]); - const IColumn * nested_column = &column->getNestedColumn(); - if (!column->isNullAt(row_num) && singleFilter(columns, row_num, num_arguments)) - { - this->setFlag(place); - this->nested_function->add(this->nestedPlace(place), &nested_column, row_num, arena); - } - } -}; - -template -class AggregateFunctionIfNullVariadic final - : public AggregateFunctionNullBase> -{ -public: - - String getName() const override - { - return Base::getName() + "If"; - } - - AggregateFunctionIfNullVariadic(AggregateFunctionPtr nested_function_, const DataTypes & arguments, const Array & params) - : Base(std::move(nested_function_), arguments, params), number_of_arguments(arguments.size()) - { - if (number_of_arguments == 1) - throw Exception("Logical error: single argument is passed to AggregateFunctionIfNullVariadic", ErrorCodes::LOGICAL_ERROR); - - if (number_of_arguments > MAX_ARGS) - throw Exception("Maximum number of arguments for aggregate function with Nullable types is " + toString(size_t(MAX_ARGS)), - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); - - for (size_t i = 0; i < number_of_arguments; ++i) - is_nullable[i] = arguments[i]->isNullable(); - } - - static inline bool singleFilter(const IColumn ** columns, size_t row_num, size_t num_arguments) - { - return assert_cast(*columns[num_arguments - 1]).getData()[row_num]; - } - - void add(AggregateDataPtr place, const IColumn ** columns, size_t row_num, Arena * arena) const override - { - /// This container stores the columns we really pass to the nested function. - const IColumn * nested_columns[number_of_arguments]; - - for (size_t i = 0; i < number_of_arguments; ++i) - { - if (is_nullable[i]) - { - const ColumnNullable & nullable_col = assert_cast(*columns[i]); - if (null_is_skipped && nullable_col.isNullAt(row_num)) - { - /// If at least one column has a null value in the current row, - /// we don't process this row. - return; - } - nested_columns[i] = &nullable_col.getNestedColumn(); - } - else - nested_columns[i] = columns[i]; - } - - if (singleFilter(nested_columns, row_num, number_of_arguments)) - { - this->setFlag(place); - this->nested_function->add(this->nestedPlace(place), nested_columns, row_num, arena); - } - } - -private: - using Base = AggregateFunctionNullBase>; - - enum { MAX_ARGS = 8 }; - size_t number_of_arguments = 0; - std::array is_nullable; /// Plain array is better than std::vector due to one indirection less. -}; - - -AggregateFunctionPtr AggregateFunctionIf::getOwnNullAdapter( - const AggregateFunctionPtr & nested_function, const DataTypes & arguments, - const Array & params, const AggregateFunctionProperties & properties) const -{ - bool return_type_is_nullable = !properties.returns_default_when_only_null && getReturnType()->canBeInsideNullable(); - size_t nullable_size = std::count_if(arguments.begin(), arguments.end(), [](const auto & element) { return element->isNullable(); }); - return_type_is_nullable &= nullable_size != 1 || !arguments.back()->isNullable(); /// If only condition is nullable. we should non-nullable type. - bool serialize_flag = return_type_is_nullable || properties.returns_default_when_only_null; - - if (arguments.size() <= 2 && arguments.front()->isNullable()) - { - if (return_type_is_nullable) - { - return std::make_shared>(nested_func, arguments, params); - } - else - { - if (serialize_flag) - return std::make_shared>(nested_func, arguments, params); - else - return std::make_shared>(nested_func, arguments, params); - } - } - else - { - if (return_type_is_nullable) - { - return std::make_shared>(nested_function, arguments, params); - } - else - { - if (serialize_flag) - return std::make_shared>(nested_function, arguments, params); - else - return std::make_shared>(nested_function, arguments, params); - } - } -} - void registerAggregateFunctionCombinatorIf(AggregateFunctionCombinatorFactory & factory) { factory.registerCombinator(std::make_shared()); diff --git a/src/AggregateFunctions/AggregateFunctionIf.h b/src/AggregateFunctions/AggregateFunctionIf.h index d5d2b9be0dd..f04450c9142 100644 --- a/src/AggregateFunctions/AggregateFunctionIf.h +++ b/src/AggregateFunctions/AggregateFunctionIf.h @@ -109,10 +109,6 @@ public: { return nested_func->isState(); } - - AggregateFunctionPtr getOwnNullAdapter( - const AggregateFunctionPtr & nested_function, const DataTypes & arguments, - const Array & params, const AggregateFunctionProperties & properties) const override; }; } diff --git a/src/AggregateFunctions/AggregateFunctionMLMethod.cpp b/src/AggregateFunctions/AggregateFunctionMLMethod.cpp index 5cbdf4c3b45..812794902df 100644 --- a/src/AggregateFunctions/AggregateFunctionMLMethod.cpp +++ b/src/AggregateFunctions/AggregateFunctionMLMethod.cpp @@ -143,13 +143,12 @@ void LinearModelData::updateState() void LinearModelData::predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const Context & context) const { - gradient_computer->predict(container, columns, offset, limit, arguments, weights, bias, context); + gradient_computer->predict(container, arguments, offset, limit, weights, bias, context); } void LinearModelData::returnWeights(IColumn & to) const @@ -449,15 +448,14 @@ void IWeightsUpdater::addToBatch( void LogisticRegression::predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const std::vector & weights, Float64 bias, const Context & /*context*/) const { - size_t rows_num = columns[arguments.front()].column->size(); + size_t rows_num = arguments.front().column->size(); if (offset > rows_num || offset + limit > rows_num) throw Exception("Invalid offset and limit for LogisticRegression::predict. " @@ -468,7 +466,7 @@ void LogisticRegression::predict( for (size_t i = 1; i < arguments.size(); ++i) { - const ColumnWithTypeAndName & cur_col = columns[arguments[i]]; + const ColumnWithTypeAndName & cur_col = arguments[i]; if (!isNativeNumber(cur_col.type)) throw Exception("Prediction arguments must have numeric type", ErrorCodes::BAD_ARGUMENTS); @@ -518,10 +516,9 @@ void LogisticRegression::compute( void LinearRegression::predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const std::vector & weights, Float64 bias, const Context & /*context*/) const @@ -531,7 +528,7 @@ void LinearRegression::predict( throw Exception("In predict function number of arguments differs from the size of weights vector", ErrorCodes::LOGICAL_ERROR); } - size_t rows_num = columns[arguments.front()].column->size(); + size_t rows_num = arguments.front().column->size(); if (offset > rows_num || offset + limit > rows_num) throw Exception("Invalid offset and limit for LogisticRegression::predict. " @@ -542,7 +539,7 @@ void LinearRegression::predict( for (size_t i = 1; i < arguments.size(); ++i) { - const ColumnWithTypeAndName & cur_col = columns[arguments[i]]; + const ColumnWithTypeAndName & cur_col = arguments[i]; if (!isNativeNumber(cur_col.type)) throw Exception("Prediction arguments must have numeric type", ErrorCodes::BAD_ARGUMENTS); diff --git a/src/AggregateFunctions/AggregateFunctionMLMethod.h b/src/AggregateFunctions/AggregateFunctionMLMethod.h index 6f7c1232f4a..494907c4002 100644 --- a/src/AggregateFunctions/AggregateFunctionMLMethod.h +++ b/src/AggregateFunctions/AggregateFunctionMLMethod.h @@ -39,10 +39,9 @@ public: virtual void predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const std::vector & weights, Float64 bias, const Context & context) const = 0; @@ -65,10 +64,9 @@ public: void predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const std::vector & weights, Float64 bias, const Context & context) const override; @@ -91,10 +89,9 @@ public: void predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const std::vector & weights, Float64 bias, const Context & context) const override; @@ -264,10 +261,9 @@ public: void predict( ColumnVector::Container & container, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const Context & context) const; void returnWeights(IColumn & to) const; @@ -364,10 +360,9 @@ public: void predictValues( ConstAggregateDataPtr place, IColumn & to, - ColumnsWithTypeAndName & columns, + ColumnsWithTypeAndName & arguments, size_t offset, size_t limit, - const ColumnNumbers & arguments, const Context & context) const override { if (arguments.size() != param_num + 1) @@ -382,7 +377,7 @@ public: throw Exception("Cast of column of predictions is incorrect. getReturnTypeToPredict must return same value as it is casted to", ErrorCodes::LOGICAL_ERROR); - this->data(place).predict(column->getData(), columns, offset, limit, arguments, context); + this->data(place).predict(column->getData(), arguments, offset, limit, context); } /** This function is called if aggregate function without State modifier is selected in a query. diff --git a/src/AggregateFunctions/AggregateFunctionNull.cpp b/src/AggregateFunctions/AggregateFunctionNull.cpp index f584ae1f34c..5e0d6ee6e21 100644 --- a/src/AggregateFunctions/AggregateFunctionNull.cpp +++ b/src/AggregateFunctions/AggregateFunctionNull.cpp @@ -72,7 +72,7 @@ public: assert(nested_function); - if (auto adapter = nested_function->getOwnNullAdapter(nested_function, arguments, params, properties)) + if (auto adapter = nested_function->getOwnNullAdapter(nested_function, arguments, params)) return adapter; /// If applied to aggregate function with -State combinator, we apply -Null combinator to it's nested_function instead of itself. diff --git a/src/AggregateFunctions/AggregateFunctionRankCorrelation.cpp b/src/AggregateFunctions/AggregateFunctionRankCorrelation.cpp index 20472279dba..796ff028424 100644 --- a/src/AggregateFunctions/AggregateFunctionRankCorrelation.cpp +++ b/src/AggregateFunctions/AggregateFunctionRankCorrelation.cpp @@ -45,7 +45,7 @@ AggregateFunctionPtr createAggregateFunctionRankCorrelation(const std::string & void registerAggregateFunctionRankCorrelation(AggregateFunctionFactory & factory) { - factory.registerFunction("rankCorr", createAggregateFunctionRankCorrelation, AggregateFunctionFactory::CaseInsensitive); + factory.registerFunction("rankCorr", createAggregateFunctionRankCorrelation); } } diff --git a/src/AggregateFunctions/AggregateFunctionRankCorrelation.h b/src/AggregateFunctions/AggregateFunctionRankCorrelation.h index 15057940ebd..75592cf5c9b 100644 --- a/src/AggregateFunctions/AggregateFunctionRankCorrelation.h +++ b/src/AggregateFunctions/AggregateFunctionRankCorrelation.h @@ -21,10 +21,6 @@ #include -namespace ErrorCodes -{ -extern const int BAD_ARGUMENTS; -} namespace DB { @@ -138,23 +134,18 @@ public: const auto & value = this->data(place).values; size_t size = this->data(place).size_x; - if (size < 2) - { - throw Exception("Aggregate function " + getName() + " requires samples to be of size > 1", ErrorCodes::BAD_ARGUMENTS); - } - - //create a copy of values not to format data + // create a copy of values not to format data PODArrayWithStackMemory, 32> tmp_values; tmp_values.resize(size); for (size_t j = 0; j < size; ++ j) tmp_values[j] = static_cast>(value[j]); - //sort x_values + // sort x_values std::sort(std::begin(tmp_values), std::end(tmp_values), ComparePairFirst{}); for (size_t j = 0; j < size;) { - //replace x_values with their ranks + // replace x_values with their ranks size_t rank = j + 1; size_t same = 1; size_t cur_sum = rank; @@ -166,9 +157,9 @@ public: { // rank of (j + 1)th number rank += 1; - same++; + ++same; cur_sum += rank; - j++; + ++j; } else break; @@ -178,16 +169,16 @@ public: Float64 insert_rank = static_cast(cur_sum) / same; for (size_t i = cur_start; i <= j; ++i) tmp_values[i].first = insert_rank; - j++; + ++j; } - //sort y_values + // sort y_values std::sort(std::begin(tmp_values), std::end(tmp_values), ComparePairSecond{}); - //replace y_values with their ranks + // replace y_values with their ranks for (size_t j = 0; j < size;) { - //replace x_values with their ranks + // replace x_values with their ranks size_t rank = j + 1; size_t same = 1; size_t cur_sum = rank; @@ -199,9 +190,9 @@ public: { // rank of (j + 1)th number rank += 1; - same++; + ++same; cur_sum += rank; - j++; + ++j; } else { @@ -213,10 +204,10 @@ public: Float64 insert_rank = static_cast(cur_sum) / same; for (size_t i = cur_start; i <= j; ++i) tmp_values[i].second = insert_rank; - j++; + ++j; } - //count d^2 sum + // count d^2 sum Float64 answer = static_cast(0); for (size_t j = 0; j < size; ++ j) answer += (tmp_values[j].first - tmp_values[j].second) * (tmp_values[j].first - tmp_values[j].second); diff --git a/src/AggregateFunctions/AggregateFunctionRetention.cpp b/src/AggregateFunctions/AggregateFunctionRetention.cpp index 4497703c550..c9d475c78f3 100644 --- a/src/AggregateFunctions/AggregateFunctionRetention.cpp +++ b/src/AggregateFunctions/AggregateFunctionRetention.cpp @@ -32,7 +32,7 @@ AggregateFunctionPtr createAggregateFunctionRetention(const std::string & name, void registerAggregateFunctionRetention(AggregateFunctionFactory & factory) { - factory.registerFunction("retention", createAggregateFunctionRetention, AggregateFunctionFactory::CaseInsensitive); + factory.registerFunction("retention", createAggregateFunctionRetention); } } diff --git a/src/AggregateFunctions/AggregateFunctionStudentTTest.cpp b/src/AggregateFunctions/AggregateFunctionStudentTTest.cpp new file mode 100644 index 00000000000..58fc9e5b5b9 --- /dev/null +++ b/src/AggregateFunctions/AggregateFunctionStudentTTest.cpp @@ -0,0 +1,52 @@ +#include +#include +#include +#include "registerAggregateFunctions.h" + +#include +#include + + +// the return type is boolean (we use UInt8 as we do not have boolean in clickhouse) + +namespace ErrorCodes +{ +extern const int NOT_IMPLEMENTED; +} + +namespace DB +{ + +namespace +{ + +AggregateFunctionPtr createAggregateFunctionStudentTTest(const std::string & name, const DataTypes & argument_types, const Array & parameters) +{ + assertBinary(name, argument_types); + assertNoParameters(name, parameters); + + AggregateFunctionPtr res; + + if (isDecimal(argument_types[0]) || isDecimal(argument_types[1])) + { + throw Exception("Aggregate function " + name + " only supports numerical types", ErrorCodes::NOT_IMPLEMENTED); + } + else + { + res.reset(createWithTwoNumericTypes(*argument_types[0], *argument_types[1], argument_types)); + } + + if (!res) + { + throw Exception("Aggregate function " + name + " only supports numerical types", ErrorCodes::NOT_IMPLEMENTED); + } + + return res; +} +} + +void registerAggregateFunctionStudentTTest(AggregateFunctionFactory & factory) +{ + factory.registerFunction("studentTTest", createAggregateFunctionStudentTTest); +} +} diff --git a/src/AggregateFunctions/AggregateFunctionStudentTTest.h b/src/AggregateFunctions/AggregateFunctionStudentTTest.h new file mode 100644 index 00000000000..0aef8f3ee2a --- /dev/null +++ b/src/AggregateFunctions/AggregateFunctionStudentTTest.h @@ -0,0 +1,262 @@ +#pragma once + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} + +#if defined(OS_DARWIN) +extern "C" +{ + double lgammal_r(double x, int * signgamp); +} +#endif + + +namespace DB +{ + +template +struct AggregateFunctionStudentTTestData final +{ + size_t size_x = 0; + size_t size_y = 0; + X sum_x = static_cast(0); + Y sum_y = static_cast(0); + X square_sum_x = static_cast(0); + Y square_sum_y = static_cast(0); + Float64 mean_x = static_cast(0); + Float64 mean_y = static_cast(0); + + void add(X x, Y y) + { + sum_x += x; + sum_y += y; + size_x++; + size_y++; + mean_x = static_cast(sum_x) / size_x; + mean_y = static_cast(sum_y) / size_y; + square_sum_x += x * x; + square_sum_y += y * y; + } + + void merge(const AggregateFunctionStudentTTestData &other) + { + sum_x += other.sum_x; + sum_y += other.sum_y; + size_x += other.size_x; + size_y += other.size_y; + mean_x = static_cast(sum_x) / size_x; + mean_y = static_cast(sum_y) / size_y; + square_sum_x += other.square_sum_x; + square_sum_y += other.square_sum_y; + } + + void serialize(WriteBuffer &buf) const + { + writeBinary(mean_x, buf); + writeBinary(mean_y, buf); + writeBinary(sum_x, buf); + writeBinary(sum_y, buf); + writeBinary(square_sum_x, buf); + writeBinary(square_sum_y, buf); + writeBinary(size_x, buf); + writeBinary(size_y, buf); + } + + void deserialize(ReadBuffer &buf) + { + readBinary(mean_x, buf); + readBinary(mean_y, buf); + readBinary(sum_x, buf); + readBinary(sum_y, buf); + readBinary(square_sum_x, buf); + readBinary(square_sum_y, buf); + readBinary(size_x, buf); + readBinary(size_y, buf); + } + + size_t getSizeY() const + { + return size_y; + } + + size_t getSizeX() const + { + return size_x; + } + + Float64 getSSquared() const + { + /// The original formulae looks like + /// \frac{\sum_{i = 1}^{n_x}{(x_i - \bar{x}) ^ 2} + \sum_{i = 1}^{n_y}{(y_i - \bar{y}) ^ 2}}{n_x + n_y - 2} + /// But we made some mathematical transformations not to store original sequences. + /// Also we dropped sqrt, because later it will be squared later. + const Float64 all_x = square_sum_x + size_x * std::pow(mean_x, 2) - 2 * mean_x * sum_x; + const Float64 all_y = square_sum_y + size_y * std::pow(mean_y, 2) - 2 * mean_y * sum_y; + return static_cast(all_x + all_y) / (size_x + size_y - 2); + } + + + Float64 getTStatisticSquared() const + { + return std::pow(mean_x - mean_y, 2) / getStandartErrorSquared(); + } + + Float64 getTStatistic() const + { + return (mean_x - mean_y) / std::sqrt(getStandartErrorSquared()); + } + + Float64 getStandartErrorSquared() const + { + if (size_x == 0 || size_y == 0) + throw Exception("Division by zero encountered in Aggregate function StudentTTest", ErrorCodes::BAD_ARGUMENTS); + + return getSSquared() * (1.0 / static_cast(size_x) + 1.0 / static_cast(size_y)); + } + + Float64 getDegreesOfFreedom() const + { + return static_cast(size_x + size_y - 2); + } + + static Float64 integrateSimpson(Float64 a, Float64 b, std::function func) + { + const size_t iterations = std::max(1e6, 1e4 * std::abs(std::round(b))); + const long double h = (b - a) / iterations; + Float64 sum_odds = 0.0; + for (size_t i = 1; i < iterations; i += 2) + sum_odds += func(a + i * h); + Float64 sum_evens = 0.0; + for (size_t i = 2; i < iterations; i += 2) + sum_evens += func(a + i * h); + return (func(a) + func(b) + 2 * sum_evens + 4 * sum_odds) * h / 3; + } + + Float64 getPValue() const + { + const Float64 v = getDegreesOfFreedom(); + const Float64 t = getTStatisticSquared(); + auto f = [&v] (double x) { return std::pow(x, v/2 - 1) / std::sqrt(1 - x); }; + Float64 numenator = integrateSimpson(0, v / (t + v), f); + int unused; + Float64 denominator = std::exp(lgammal_r(v / 2, &unused) + lgammal_r(0.5, &unused) - lgammal_r(v / 2 + 0.5, &unused)); + return numenator / denominator; + } + + std::pair getResult() const + { + return std::make_pair(getTStatistic(), getPValue()); + } +}; + +/// Returns tuple of (t-statistic, p-value) +/// https://cpb-us-w2.wpmucdn.com/voices.uchicago.edu/dist/9/1193/files/2016/01/05b-TandP.pdf +template +class AggregateFunctionStudentTTest : + public IAggregateFunctionDataHelper,AggregateFunctionStudentTTest> +{ + +public: + AggregateFunctionStudentTTest(const DataTypes & arguments) + : IAggregateFunctionDataHelper, AggregateFunctionStudentTTest> ({arguments}, {}) + {} + + String getName() const override + { + return "studentTTest"; + } + + DataTypePtr getReturnType() const override + { + DataTypes types + { + std::make_shared>(), + std::make_shared>(), + }; + + Strings names + { + "t-statistic", + "p-value" + }; + + return std::make_shared( + std::move(types), + std::move(names) + ); + } + + void add(AggregateDataPtr place, const IColumn ** columns, size_t row_num, Arena *) const override + { + auto col_x = assert_cast *>(columns[0]); + auto col_y = assert_cast *>(columns[1]); + + X x = col_x->getData()[row_num]; + Y y = col_y->getData()[row_num]; + + this->data(place).add(x, y); + } + + void merge(AggregateDataPtr place, ConstAggregateDataPtr rhs, Arena *) const override + { + this->data(place).merge(this->data(rhs)); + } + + void serialize(ConstAggregateDataPtr place, WriteBuffer & buf) const override + { + this->data(place).serialize(buf); + } + + void deserialize(AggregateDataPtr place, ReadBuffer & buf, Arena *) const override + { + this->data(place).deserialize(buf); + } + + void insertResultInto(AggregateDataPtr place, IColumn & to, Arena * /*arena*/) const override + { + size_t size_x = this->data(place).getSizeX(); + size_t size_y = this->data(place).getSizeY(); + + if (size_x < 2 || size_y < 2) + { + throw Exception("Aggregate function " + getName() + " requires samples to be of size > 1", ErrorCodes::BAD_ARGUMENTS); + } + + Float64 t_statistic = 0.0; + Float64 p_value = 0.0; + std::tie(t_statistic, p_value) = this->data(place).getResult(); + + /// Because p-value is a probability. + p_value = std::min(1.0, std::max(0.0, p_value)); + + auto & column_tuple = assert_cast(to); + auto & column_stat = assert_cast &>(column_tuple.getColumn(0)); + auto & column_value = assert_cast &>(column_tuple.getColumn(1)); + + column_stat.getData().push_back(t_statistic); + column_value.getData().push_back(p_value); + } + +}; + +}; diff --git a/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.cpp b/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.cpp index cd0599729b9..c8711c257f8 100644 --- a/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.cpp +++ b/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.cpp @@ -28,8 +28,8 @@ namespace void registerAggregateFunctionTimeSeriesGroupSum(AggregateFunctionFactory & factory) { - factory.registerFunction("timeSeriesGroupSum", createAggregateFunctionTimeSeriesGroupSum, AggregateFunctionFactory::CaseInsensitive); - factory.registerFunction("timeSeriesGroupRateSum", createAggregateFunctionTimeSeriesGroupSum, AggregateFunctionFactory::CaseInsensitive); + factory.registerFunction("timeSeriesGroupSum", createAggregateFunctionTimeSeriesGroupSum); + factory.registerFunction("timeSeriesGroupRateSum", createAggregateFunctionTimeSeriesGroupSum); } } diff --git a/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.h b/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.h index be0a3eb4af5..b755fbf081b 100644 --- a/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.h +++ b/src/AggregateFunctions/AggregateFunctionTimeSeriesGroupSum.h @@ -92,7 +92,7 @@ struct AggregateFunctionTimeSeriesGroupSumData it_ss->second.add(t, v); } if (result.size() > 0 && t < result.back().first) - throw Exception{"timeSeriesGroupSum or timeSeriesGroupRateSum must order by timestamp asc!!!", ErrorCodes::LOGICAL_ERROR}; + throw Exception{"timeSeriesGroupSum or timeSeriesGroupRateSum must order by timestamp asc.", ErrorCodes::LOGICAL_ERROR}; if (result.size() > 0 && t == result.back().first) { //do not add new point diff --git a/src/AggregateFunctions/AggregateFunctionWelchTTest.cpp b/src/AggregateFunctions/AggregateFunctionWelchTTest.cpp new file mode 100644 index 00000000000..0dcb125305d --- /dev/null +++ b/src/AggregateFunctions/AggregateFunctionWelchTTest.cpp @@ -0,0 +1,49 @@ +#include +#include +#include +#include "registerAggregateFunctions.h" + +#include +#include + +namespace ErrorCodes +{ +extern const int NOT_IMPLEMENTED; +} + +namespace DB +{ + +namespace +{ + +AggregateFunctionPtr createAggregateFunctionWelchTTest(const std::string & name, const DataTypes & argument_types, const Array & parameters) +{ + assertBinary(name, argument_types); + assertNoParameters(name, parameters); + + AggregateFunctionPtr res; + + if (isDecimal(argument_types[0]) || isDecimal(argument_types[1])) + { + throw Exception("Aggregate function " + name + " only supports numerical types", ErrorCodes::NOT_IMPLEMENTED); + } + else + { + res.reset(createWithTwoNumericTypes(*argument_types[0], *argument_types[1], argument_types)); + } + + if (!res) + { + throw Exception("Aggregate function " + name + " only supports numerical types", ErrorCodes::NOT_IMPLEMENTED); + } + + return res; +} +} + +void registerAggregateFunctionWelchTTest(AggregateFunctionFactory & factory) +{ + factory.registerFunction("welchTTest", createAggregateFunctionWelchTTest); +} +} diff --git a/src/AggregateFunctions/AggregateFunctionWelchTTest.h b/src/AggregateFunctions/AggregateFunctionWelchTTest.h new file mode 100644 index 00000000000..b598f25162e --- /dev/null +++ b/src/AggregateFunctions/AggregateFunctionWelchTTest.h @@ -0,0 +1,274 @@ +#pragma once + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + + +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} + +#if defined(OS_DARWIN) +extern "C" +{ + double lgammal_r(double x, int * signgamp); +} +#endif + + +namespace DB +{ + +template +struct AggregateFunctionWelchTTestData final +{ + size_t size_x = 0; + size_t size_y = 0; + X sum_x = static_cast(0); + Y sum_y = static_cast(0); + X square_sum_x = static_cast(0); + Y square_sum_y = static_cast(0); + Float64 mean_x = static_cast(0); + Float64 mean_y = static_cast(0); + + void add(X x, Y y) + { + sum_x += x; + sum_y += y; + size_x++; + size_y++; + mean_x = static_cast(sum_x) / size_x; + mean_y = static_cast(sum_y) / size_y; + square_sum_x += x * x; + square_sum_y += y * y; + } + + void merge(const AggregateFunctionWelchTTestData &other) + { + sum_x += other.sum_x; + sum_y += other.sum_y; + size_x += other.size_x; + size_y += other.size_y; + mean_x = static_cast(sum_x) / size_x; + mean_y = static_cast(sum_y) / size_y; + square_sum_x += other.square_sum_x; + square_sum_y += other.square_sum_y; + } + + void serialize(WriteBuffer &buf) const + { + writeBinary(mean_x, buf); + writeBinary(mean_y, buf); + writeBinary(sum_x, buf); + writeBinary(sum_y, buf); + writeBinary(square_sum_x, buf); + writeBinary(square_sum_y, buf); + writeBinary(size_x, buf); + writeBinary(size_y, buf); + } + + void deserialize(ReadBuffer &buf) + { + readBinary(mean_x, buf); + readBinary(mean_y, buf); + readBinary(sum_x, buf); + readBinary(sum_y, buf); + readBinary(square_sum_x, buf); + readBinary(square_sum_y, buf); + readBinary(size_x, buf); + readBinary(size_y, buf); + } + + size_t getSizeY() const + { + return size_y; + } + + size_t getSizeX() const + { + return size_x; + } + + Float64 getSxSquared() const + { + /// The original formulae looks like \frac{1}{size_x - 1} \sum_{i = 1}^{size_x}{(x_i - \bar{x}) ^ 2} + /// But we made some mathematical transformations not to store original sequences. + /// Also we dropped sqrt, because later it will be squared later. + return static_cast(square_sum_x + size_x * std::pow(mean_x, 2) - 2 * mean_x * sum_x) / (size_x - 1); + } + + Float64 getSySquared() const + { + /// The original formulae looks like \frac{1}{size_y - 1} \sum_{i = 1}^{size_y}{(y_i - \bar{y}) ^ 2} + /// But we made some mathematical transformations not to store original sequences. + /// Also we dropped sqrt, because later it will be squared later. + return static_cast(square_sum_y + size_y * std::pow(mean_y, 2) - 2 * mean_y * sum_y) / (size_y - 1); + } + + Float64 getTStatisticSquared() const + { + if (size_x == 0 || size_y == 0) + { + throw Exception("Division by zero encountered in Aggregate function WelchTTest", ErrorCodes::BAD_ARGUMENTS); + } + + return std::pow(mean_x - mean_y, 2) / (getSxSquared() / size_x + getSySquared() / size_y); + } + + Float64 getTStatistic() const + { + if (size_x == 0 || size_y == 0) + { + throw Exception("Division by zero encountered in Aggregate function WelchTTest", ErrorCodes::BAD_ARGUMENTS); + } + + return (mean_x - mean_y) / std::sqrt(getSxSquared() / size_x + getSySquared() / size_y); + } + + Float64 getDegreesOfFreedom() const + { + auto sx = getSxSquared(); + auto sy = getSySquared(); + Float64 numerator = std::pow(sx / size_x + sy / size_y, 2); + Float64 denominator_first = std::pow(sx, 2) / (std::pow(size_x, 2) * (size_x - 1)); + Float64 denominator_second = std::pow(sy, 2) / (std::pow(size_y, 2) * (size_y - 1)); + return numerator / (denominator_first + denominator_second); + } + + static Float64 integrateSimpson(Float64 a, Float64 b, std::function func) + { + size_t iterations = std::max(1e6, 1e4 * std::abs(std::round(b))); + double h = (b - a) / iterations; + Float64 sum_odds = 0.0; + for (size_t i = 1; i < iterations; i += 2) + sum_odds += func(a + i * h); + Float64 sum_evens = 0.0; + for (size_t i = 2; i < iterations; i += 2) + sum_evens += func(a + i * h); + return (func(a) + func(b) + 2 * sum_evens + 4 * sum_odds) * h / 3; + } + + Float64 getPValue() const + { + const Float64 v = getDegreesOfFreedom(); + const Float64 t = getTStatisticSquared(); + auto f = [&v] (double x) { return std::pow(x, v / 2 - 1) / std::sqrt(1 - x); }; + Float64 numenator = integrateSimpson(0, v / (t + v), f); + int unused; + Float64 denominator = std::exp(lgammal_r(v / 2, &unused) + lgammal_r(0.5, &unused) - lgammal_r(v / 2 + 0.5, &unused)); + return numenator / denominator; + } + + std::pair getResult() const + { + return std::make_pair(getTStatistic(), getPValue()); + } +}; + +/// Returns tuple of (t-statistic, p-value) +/// https://cpb-us-w2.wpmucdn.com/voices.uchicago.edu/dist/9/1193/files/2016/01/05b-TandP.pdf +template +class AggregateFunctionWelchTTest : + public IAggregateFunctionDataHelper,AggregateFunctionWelchTTest> +{ + +public: + AggregateFunctionWelchTTest(const DataTypes & arguments) + : IAggregateFunctionDataHelper, AggregateFunctionWelchTTest> ({arguments}, {}) + {} + + String getName() const override + { + return "welchTTest"; + } + + DataTypePtr getReturnType() const override + { + DataTypes types + { + std::make_shared>(), + std::make_shared>(), + }; + + Strings names + { + "t-statistic", + "p-value" + }; + + return std::make_shared( + std::move(types), + std::move(names) + ); + } + + void add(AggregateDataPtr place, const IColumn ** columns, size_t row_num, Arena *) const override + { + auto col_x = assert_cast *>(columns[0]); + auto col_y = assert_cast *>(columns[1]); + + X x = col_x->getData()[row_num]; + Y y = col_y->getData()[row_num]; + + this->data(place).add(x, y); + } + + void merge(AggregateDataPtr place, ConstAggregateDataPtr rhs, Arena *) const override + { + this->data(place).merge(this->data(rhs)); + } + + void serialize(ConstAggregateDataPtr place, WriteBuffer & buf) const override + { + this->data(place).serialize(buf); + } + + void deserialize(AggregateDataPtr place, ReadBuffer & buf, Arena *) const override + { + this->data(place).deserialize(buf); + } + + void insertResultInto(AggregateDataPtr place, IColumn & to, Arena * /*arena*/) const override + { + size_t size_x = this->data(place).getSizeX(); + size_t size_y = this->data(place).getSizeY(); + + if (size_x < 2 || size_y < 2) + { + throw Exception("Aggregate function " + getName() + " requires samples to be of size > 1", ErrorCodes::BAD_ARGUMENTS); + } + + Float64 t_statistic = 0.0; + Float64 p_value = 0.0; + std::tie(t_statistic, p_value) = this->data(place).getResult(); + + /// Because p-value is a probability. + p_value = std::min(1.0, std::max(0.0, p_value)); + + auto & column_tuple = assert_cast(to); + auto & column_stat = assert_cast &>(column_tuple.getColumn(0)); + auto & column_value = assert_cast &>(column_tuple.getColumn(1)); + + column_stat.getData().push_back(t_statistic); + column_value.getData().push_back(p_value); + } + +}; + +}; diff --git a/src/AggregateFunctions/AggregateFunctionWindowFunnel.cpp b/src/AggregateFunctions/AggregateFunctionWindowFunnel.cpp index 872c70c2b98..1e9f2782d95 100644 --- a/src/AggregateFunctions/AggregateFunctionWindowFunnel.cpp +++ b/src/AggregateFunctions/AggregateFunctionWindowFunnel.cpp @@ -58,7 +58,7 @@ AggregateFunctionPtr createAggregateFunctionWindowFunnel(const std::string & nam void registerAggregateFunctionWindowFunnel(AggregateFunctionFactory & factory) { - factory.registerFunction("windowFunnel", createAggregateFunctionWindowFunnel, AggregateFunctionFactory::CaseInsensitive); + factory.registerFunction("windowFunnel", createAggregateFunctionWindowFunnel); } } diff --git a/src/AggregateFunctions/AggregateFunctionWindowFunnel.h b/src/AggregateFunctions/AggregateFunctionWindowFunnel.h index fe45fec4b76..3297819a9ff 100644 --- a/src/AggregateFunctions/AggregateFunctionWindowFunnel.h +++ b/src/AggregateFunctions/AggregateFunctionWindowFunnel.h @@ -241,8 +241,7 @@ public: } AggregateFunctionPtr getOwnNullAdapter( - const AggregateFunctionPtr & nested_function, const DataTypes & arguments, const Array & params, - const AggregateFunctionProperties & /*properties*/) const override + const AggregateFunctionPtr & nested_function, const DataTypes & arguments, const Array & params) const override { return std::make_shared>(nested_function, arguments, params); } diff --git a/src/AggregateFunctions/IAggregateFunction.h b/src/AggregateFunctions/IAggregateFunction.h index 498a5303ac4..4f9552d2345 100644 --- a/src/AggregateFunctions/IAggregateFunction.h +++ b/src/AggregateFunctions/IAggregateFunction.h @@ -33,7 +33,6 @@ using ConstAggregateDataPtr = const char *; class IAggregateFunction; using AggregateFunctionPtr = std::shared_ptr; -struct AggregateFunctionProperties; /** Aggregate functions interface. * Instances of classes with this interface do not contain the data itself for aggregation, @@ -114,10 +113,9 @@ public: virtual void predictValues( ConstAggregateDataPtr /* place */, IColumn & /*to*/, - ColumnsWithTypeAndName & /*block*/, + ColumnsWithTypeAndName & /*arguments*/, size_t /*offset*/, size_t /*limit*/, - const ColumnNumbers & /*arguments*/, const Context & /*context*/) const { throw Exception("Method predictValues is not supported for " + getName(), ErrorCodes::NOT_IMPLEMENTED); @@ -187,8 +185,7 @@ public: * arguments and params are for nested_function. */ virtual AggregateFunctionPtr getOwnNullAdapter( - const AggregateFunctionPtr & /*nested_function*/, const DataTypes & /*arguments*/, - const Array & /*params*/, const AggregateFunctionProperties & /*properties*/) const + const AggregateFunctionPtr & /*nested_function*/, const DataTypes & /*arguments*/, const Array & /*params*/) const { return nullptr; } diff --git a/src/AggregateFunctions/ReservoirSampler.h b/src/AggregateFunctions/ReservoirSampler.h index 2ff5ab863af..bbb8172b209 100644 --- a/src/AggregateFunctions/ReservoirSampler.h +++ b/src/AggregateFunctions/ReservoirSampler.h @@ -158,12 +158,25 @@ public: } else { - randomShuffle(samples); + /// Replace every element in our reservoir to the b's reservoir + /// with the probability of b.total_values / (a.total_values + b.total_values) + /// Do it more roughly than true random sampling to save performance. + total_values += b.total_values; - for (size_t i = 0; i < sample_count; ++i) + + /// Will replace every frequency'th element in a to element from b. + double frequency = static_cast(total_values) / b.total_values; + + /// When frequency is too low, replace just one random element with the corresponding probability. + if (frequency * 2 >= sample_count) { - UInt64 rnd = genRandom(total_values); - if (rnd < b.total_values) + UInt64 rnd = genRandom(frequency); + if (rnd < sample_count) + samples[rnd] = b.samples[rnd]; + } + else + { + for (double i = 0; i < sample_count; i += frequency) samples[i] = b.samples[i]; } } @@ -200,9 +213,6 @@ public: } private: - friend void qdigest_test(int normal_size, UInt64 value_limit, const std::vector & values, int queries_count, bool verbose); - friend void rs_perf_test(); - /// We allocate a little memory on the stack - to avoid allocations when there are many objects with a small number of elements. using Array = DB::PODArrayWithStackMemory; @@ -222,15 +232,6 @@ private: return (static_cast(rng()) * (static_cast(rng.max()) + 1ULL) + static_cast(rng())) % lim; } - void randomShuffle(Array & v) - { - for (size_t i = 1; i < v.size(); ++i) - { - size_t j = genRandom(i + 1); - std::swap(v[i], v[j]); - } - } - void sortIfNeeded() { if (sorted) diff --git a/src/AggregateFunctions/ReservoirSamplerDeterministic.h b/src/AggregateFunctions/ReservoirSamplerDeterministic.h index 3097070c651..eae24c1f3e9 100644 --- a/src/AggregateFunctions/ReservoirSamplerDeterministic.h +++ b/src/AggregateFunctions/ReservoirSamplerDeterministic.h @@ -39,8 +39,8 @@ namespace ErrorCodes namespace detail { -const size_t DEFAULT_SAMPLE_COUNT = 8192; -const auto MAX_SKIP_DEGREE = sizeof(UInt32) * 8; + const size_t DEFAULT_MAX_SAMPLE_SIZE = 8192; + const auto MAX_SKIP_DEGREE = sizeof(UInt32) * 8; } /// What if there is not a single value - throw an exception, or return 0 or NaN in the case of double? @@ -50,6 +50,7 @@ enum class ReservoirSamplerDeterministicOnEmpty RETURN_NAN_OR_ZERO, }; + template class ReservoirSamplerDeterministic @@ -60,8 +61,8 @@ class ReservoirSamplerDeterministic } public: - ReservoirSamplerDeterministic(const size_t sample_count_ = DEFAULT_SAMPLE_COUNT) - : sample_count{sample_count_} + ReservoirSamplerDeterministic(const size_t max_sample_size_ = detail::DEFAULT_MAX_SAMPLE_SIZE) + : max_sample_size{max_sample_size_} { } @@ -131,8 +132,8 @@ public: void merge(const ReservoirSamplerDeterministic & b) { - if (sample_count != b.sample_count) - throw Poco::Exception("Cannot merge ReservoirSamplerDeterministic's with different sample_count"); + if (max_sample_size != b.max_sample_size) + throw Poco::Exception("Cannot merge ReservoirSamplerDeterministic's with different max sample size"); sorted = false; if (b.skip_degree > skip_degree) @@ -150,11 +151,16 @@ public: void read(DB::ReadBuffer & buf) { - DB::readIntBinary(sample_count, buf); + size_t size = 0; + DB::readIntBinary(size, buf); DB::readIntBinary(total_values, buf); - samples.resize(std::min(total_values, sample_count)); - for (size_t i = 0; i < samples.size(); ++i) + /// Compatibility with old versions. + if (size > total_values) + size = total_values; + + samples.resize(size); + for (size_t i = 0; i < size; ++i) DB::readPODBinary(samples[i], buf); sorted = false; @@ -162,10 +168,11 @@ public: void write(DB::WriteBuffer & buf) const { - DB::writeIntBinary(sample_count, buf); + size_t size = samples.size(); + DB::writeIntBinary(size, buf); DB::writeIntBinary(total_values, buf); - for (size_t i = 0; i < std::min(sample_count, total_values); ++i) + for (size_t i = 0; i < size; ++i) DB::writePODBinary(samples[i], buf); } @@ -174,18 +181,19 @@ private: using Element = std::pair; using Array = DB::PODArray; - size_t sample_count; - size_t total_values{}; - bool sorted{}; + const size_t max_sample_size; /// Maximum amount of stored values. + size_t total_values = 0; /// How many values were inserted (regardless if they remain in sample or not). + bool sorted = false; Array samples; - UInt8 skip_degree{}; + UInt8 skip_degree = 0; /// The number N determining that we save only one per 2^N elements in average. void insertImpl(const T & v, const UInt32 hash) { - /// @todo why + 1? I don't quite recall - while (samples.size() + 1 >= sample_count) + /// Make a room for plus one element. + while (samples.size() >= max_sample_size) { - if (++skip_degree > detail::MAX_SKIP_DEGREE) + ++skip_degree; + if (skip_degree > detail::MAX_SKIP_DEGREE) throw DB::Exception{"skip_degree exceeds maximum value", DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED}; thinOut(); } @@ -195,35 +203,17 @@ private: void thinOut() { - auto size = samples.size(); - for (size_t i = 0; i < size;) - { - if (!good(samples[i].second)) - { - /// swap current element with the last one - std::swap(samples[size - 1], samples[i]); - --size; - } - else - ++i; - } - - if (size != samples.size()) - { - samples.resize(size); - sorted = false; - } + samples.resize(std::distance(samples.begin(), + std::remove_if(samples.begin(), samples.end(), [this](const auto & elem){ return !good(elem.second); }))); + sorted = false; } void sortIfNeeded() { if (sorted) return; + std::sort(samples.begin(), samples.end(), [](const auto & lhs, const auto & rhs) { return lhs.first < rhs.first; }); sorted = true; - std::sort(samples.begin(), samples.end(), [] (const std::pair & lhs, const std::pair & rhs) - { - return lhs.first < rhs.first; - }); } template diff --git a/src/AggregateFunctions/registerAggregateFunctions.cpp b/src/AggregateFunctions/registerAggregateFunctions.cpp index b8af252eefa..9fd02ba9d6c 100644 --- a/src/AggregateFunctions/registerAggregateFunctions.cpp +++ b/src/AggregateFunctions/registerAggregateFunctions.cpp @@ -45,6 +45,8 @@ void registerAggregateFunctions() registerAggregateFunctionMoving(factory); registerAggregateFunctionCategoricalIV(factory); registerAggregateFunctionAggThrow(factory); + registerAggregateFunctionWelchTTest(factory); + registerAggregateFunctionStudentTTest(factory); registerAggregateFunctionRankCorrelation(factory); } diff --git a/src/AggregateFunctions/registerAggregateFunctions.h b/src/AggregateFunctions/registerAggregateFunctions.h index 90054788613..abbba56ed32 100644 --- a/src/AggregateFunctions/registerAggregateFunctions.h +++ b/src/AggregateFunctions/registerAggregateFunctions.h @@ -35,6 +35,8 @@ void registerAggregateFunctionSimpleLinearRegression(AggregateFunctionFactory &) void registerAggregateFunctionMoving(AggregateFunctionFactory &); void registerAggregateFunctionCategoricalIV(AggregateFunctionFactory &); void registerAggregateFunctionAggThrow(AggregateFunctionFactory &); +void registerAggregateFunctionWelchTTest(AggregateFunctionFactory &); +void registerAggregateFunctionStudentTTest(AggregateFunctionFactory &); void registerAggregateFunctionRankCorrelation(AggregateFunctionFactory &); class AggregateFunctionCombinatorFactory; diff --git a/src/AggregateFunctions/ya.make b/src/AggregateFunctions/ya.make index fe0574ce80b..8d27cf986d0 100644 --- a/src/AggregateFunctions/ya.make +++ b/src/AggregateFunctions/ya.make @@ -5,7 +5,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( AggregateFunctionAggThrow.cpp @@ -42,6 +41,7 @@ SRCS( AggregateFunctionState.cpp AggregateFunctionStatistics.cpp AggregateFunctionStatisticsSimple.cpp + AggregateFunctionStudentTTest.cpp AggregateFunctionSum.cpp AggregateFunctionSumMap.cpp AggregateFunctionTimeSeriesGroupSum.cpp @@ -49,6 +49,7 @@ SRCS( AggregateFunctionUniqCombined.cpp AggregateFunctionUniq.cpp AggregateFunctionUniqUpTo.cpp + AggregateFunctionWelchTTest.cpp AggregateFunctionWindowFunnel.cpp parseAggregateFunctionParameters.cpp registerAggregateFunctions.cpp diff --git a/src/AggregateFunctions/ya.make.in b/src/AggregateFunctions/ya.make.in index 4c2943b0539..dd49b679d28 100644 --- a/src/AggregateFunctions/ya.make.in +++ b/src/AggregateFunctions/ya.make.in @@ -4,7 +4,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index 0016c51b7f8..085269847e4 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -173,14 +173,20 @@ add_object_library(clickhouse_processors_merges Processors/Merges) add_object_library(clickhouse_processors_merges_algorithms Processors/Merges/Algorithms) add_object_library(clickhouse_processors_queryplan Processors/QueryPlan) +set (DBMS_COMMON_LIBRARIES) +# libgcc_s does not provide an implementation of an atomics library. Instead, +# GCC’s libatomic library can be used to supply these when using libgcc_s. +if ((NOT USE_LIBCXX) AND COMPILER_CLANG AND OS_LINUX) + list (APPEND DBMS_COMMON_LIBRARIES atomic) +endif() if (MAKE_STATIC_LIBRARIES OR NOT SPLIT_SHARED_LIBRARIES) add_library (dbms STATIC ${dbms_headers} ${dbms_sources}) - target_link_libraries (dbms PRIVATE jemalloc libdivide) + target_link_libraries (dbms PRIVATE jemalloc libdivide ${DBMS_COMMON_LIBRARIES}) set (all_modules dbms) else() add_library (dbms SHARED ${dbms_headers} ${dbms_sources}) - target_link_libraries (dbms PUBLIC ${all_modules}) + target_link_libraries (dbms PUBLIC ${all_modules} ${DBMS_COMMON_LIBRARIES}) target_link_libraries (clickhouse_interpreters PRIVATE jemalloc libdivide) list (APPEND all_modules dbms) # force all split libs to be linked diff --git a/src/Client/ya.make b/src/Client/ya.make index b07e5afc0fb..07cc6725308 100644 --- a/src/Client/ya.make +++ b/src/Client/ya.make @@ -6,7 +6,6 @@ PEERDIR( contrib/libs/poco/NetSSL_OpenSSL ) -CFLAGS(-g0) SRCS( Connection.cpp diff --git a/src/Client/ya.make.in b/src/Client/ya.make.in index 704a05d8f3b..d8faff9ae1a 100644 --- a/src/Client/ya.make.in +++ b/src/Client/ya.make.in @@ -5,7 +5,6 @@ PEERDIR( contrib/libs/poco/NetSSL_OpenSSL ) -CFLAGS(-g0) SRCS( diff --git a/src/Columns/ColumnAggregateFunction.cpp b/src/Columns/ColumnAggregateFunction.cpp index 189ca12b5f7..f5b266b6983 100644 --- a/src/Columns/ColumnAggregateFunction.cpp +++ b/src/Columns/ColumnAggregateFunction.cpp @@ -161,7 +161,7 @@ MutableColumnPtr ColumnAggregateFunction::convertToValues(MutableColumnPtr colum return res; } -MutableColumnPtr ColumnAggregateFunction::predictValues(ColumnsWithTypeAndName & block, const ColumnNumbers & arguments, const Context & context) const +MutableColumnPtr ColumnAggregateFunction::predictValues(ColumnsWithTypeAndName & arguments, const Context & context) const { MutableColumnPtr res = func->getReturnTypeToPredict()->createColumn(); res->reserve(data.size()); @@ -172,7 +172,7 @@ MutableColumnPtr ColumnAggregateFunction::predictValues(ColumnsWithTypeAndName & if (data.size() == 1) { /// Case for const column. Predict using single model. - machine_learning_function->predictValues(data[0], *res, block, 0, block[arguments.front()].column->size(), arguments, context); + machine_learning_function->predictValues(data[0], *res, arguments, 0, arguments.front().column->size(), context); } else { @@ -180,7 +180,7 @@ MutableColumnPtr ColumnAggregateFunction::predictValues(ColumnsWithTypeAndName & size_t row_num = 0; for (auto * val : data) { - machine_learning_function->predictValues(val, *res, block, row_num, 1, arguments, context); + machine_learning_function->predictValues(val, *res, arguments, row_num, 1, context); ++row_num; } } diff --git a/src/Columns/ColumnAggregateFunction.h b/src/Columns/ColumnAggregateFunction.h index 19b9bd09a7a..79e52e5769a 100644 --- a/src/Columns/ColumnAggregateFunction.h +++ b/src/Columns/ColumnAggregateFunction.h @@ -119,7 +119,7 @@ public: const char * getFamilyName() const override { return "AggregateFunction"; } TypeIndex getDataType() const override { return TypeIndex::AggregateFunction; } - MutableColumnPtr predictValues(ColumnsWithTypeAndName & block, const ColumnNumbers & arguments, const Context & context) const; + MutableColumnPtr predictValues(ColumnsWithTypeAndName & arguments, const Context & context) const; size_t size() const override { diff --git a/src/Columns/ColumnConst.cpp b/src/Columns/ColumnConst.cpp index b7fb22aeb0e..550a44a23a2 100644 --- a/src/Columns/ColumnConst.cpp +++ b/src/Columns/ColumnConst.cpp @@ -6,6 +6,8 @@ #include #include +#include + #if defined(MEMORY_SANITIZER) #include #endif diff --git a/src/Columns/ColumnFunction.cpp b/src/Columns/ColumnFunction.cpp index 25d1a75b0e2..292e10e8e14 100644 --- a/src/Columns/ColumnFunction.cpp +++ b/src/Columns/ColumnFunction.cpp @@ -188,15 +188,10 @@ ColumnWithTypeAndName ColumnFunction::reduce() const "arguments but " + toString(captured) + " columns were captured.", ErrorCodes::LOGICAL_ERROR); auto columns = captured_columns; - columns.emplace_back(ColumnWithTypeAndName {nullptr, function->getReturnType(), ""}); + ColumnWithTypeAndName res{nullptr, function->getResultType(), ""}; - ColumnNumbers arguments(captured_columns.size()); - for (size_t i = 0; i < captured_columns.size(); ++i) - arguments[i] = i; - - function->execute(columns, arguments, captured_columns.size(), size_); - - return columns[captured_columns.size()]; + res.column = function->execute(columns, res.type, size_); + return res; } } diff --git a/src/Columns/ColumnNullable.cpp b/src/Columns/ColumnNullable.cpp index bdbc941c1e7..51248a598af 100644 --- a/src/Columns/ColumnNullable.cpp +++ b/src/Columns/ColumnNullable.cpp @@ -344,7 +344,7 @@ void ColumnNullable::updatePermutation(bool reverse, size_t limit, int null_dire /// Shift all NULL values to the end. for (const auto & [first, last] : equal_ranges) { - /// Current interval is righter than limit. + /// Current interval is righter than limit. if (limit && first > limit) break; diff --git a/src/Columns/ColumnString.cpp b/src/Columns/ColumnString.cpp index 23414626a59..cd06ea20f83 100644 --- a/src/Columns/ColumnString.cpp +++ b/src/Columns/ColumnString.cpp @@ -634,4 +634,10 @@ void ColumnString::protect() getOffsets().protect(); } +void ColumnString::validate() const +{ + if (!offsets.empty() && offsets.back() != chars.size()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "ColumnString validation failed: size mismatch (internal logical error) {} != {}", offsets.back(), chars.size()); +} + } diff --git a/src/Columns/ColumnString.h b/src/Columns/ColumnString.h index 128e1efe146..19398e07b83 100644 --- a/src/Columns/ColumnString.h +++ b/src/Columns/ColumnString.h @@ -267,6 +267,9 @@ public: Offsets & getOffsets() { return offsets; } const Offsets & getOffsets() const { return offsets; } + + // Throws an exception if offsets/chars are messed up + void validate() const; }; diff --git a/src/Columns/IColumnUnique.h b/src/Columns/IColumnUnique.h index 99facea6055..99e134675f6 100644 --- a/src/Columns/IColumnUnique.h +++ b/src/Columns/IColumnUnique.h @@ -82,7 +82,7 @@ public: * @see DB::ColumnUnique * * The most common example uses https://clickhouse.tech/docs/en/sql-reference/data-types/lowcardinality/ columns. - * Consider data type @e LC(String). The inner type here is @e String which is more or less a contigous memory + * Consider data type @e LC(String). The inner type here is @e String which is more or less a contiguous memory * region, so it can be easily represented as a @e StringRef. So we pass that ref to this function and get its * index in the dictionary, which can be used to operate with the indices column. */ diff --git a/src/Columns/ya.make b/src/Columns/ya.make index 78c0e1b992d..1463bbc69e2 100644 --- a/src/Columns/ya.make +++ b/src/Columns/ya.make @@ -13,7 +13,6 @@ PEERDIR( contrib/libs/pdqsort ) -CFLAGS(-g0) SRCS( Collator.cpp diff --git a/src/Common/CurrentMetrics.cpp b/src/Common/CurrentMetrics.cpp index ef0b82666dd..3d6a2d6f99c 100644 --- a/src/Common/CurrentMetrics.cpp +++ b/src/Common/CurrentMetrics.cpp @@ -54,6 +54,7 @@ M(LocalThread, "Number of threads in local thread pools. Should be similar to GlobalThreadActive.") \ M(LocalThreadActive, "Number of threads in local thread pools running a task.") \ M(DistributedFilesToInsert, "Number of pending files to process for asynchronous insertion into Distributed tables. Number of files for every shard is summed.") \ + M(TablesToDropQueueSize, "Number of dropped tables, that are waiting for background data removal.") \ namespace CurrentMetrics { diff --git a/src/Common/ErrorCodes.cpp b/src/Common/ErrorCodes.cpp index b841368f662..f069b27827e 100644 --- a/src/Common/ErrorCodes.cpp +++ b/src/Common/ErrorCodes.cpp @@ -1,524 +1,563 @@ +#include + +/** Previously, these constants were located in one enum. + * But in this case there is a problem: when you add a new constant, you need to recompile + * all translation units that use at least one constant (almost the whole project). + * Therefore it is made so that definitions of constants are located here, in one file, + * and their declaration are in different files, at the place of use. + * + * Later it was converted to the lookup table, to provide: + * - errorCodeToName() + * - system.errors table + */ + +#define APPLY_FOR_ERROR_CODES(M) \ + M(0, OK) \ + M(1, UNSUPPORTED_METHOD) \ + M(2, UNSUPPORTED_PARAMETER) \ + M(3, UNEXPECTED_END_OF_FILE) \ + M(4, EXPECTED_END_OF_FILE) \ + M(6, CANNOT_PARSE_TEXT) \ + M(7, INCORRECT_NUMBER_OF_COLUMNS) \ + M(8, THERE_IS_NO_COLUMN) \ + M(9, SIZES_OF_COLUMNS_DOESNT_MATCH) \ + M(10, NOT_FOUND_COLUMN_IN_BLOCK) \ + M(11, POSITION_OUT_OF_BOUND) \ + M(12, PARAMETER_OUT_OF_BOUND) \ + M(13, SIZES_OF_COLUMNS_IN_TUPLE_DOESNT_MATCH) \ + M(15, DUPLICATE_COLUMN) \ + M(16, NO_SUCH_COLUMN_IN_TABLE) \ + M(17, DELIMITER_IN_STRING_LITERAL_DOESNT_MATCH) \ + M(18, CANNOT_INSERT_ELEMENT_INTO_CONSTANT_COLUMN) \ + M(19, SIZE_OF_FIXED_STRING_DOESNT_MATCH) \ + M(20, NUMBER_OF_COLUMNS_DOESNT_MATCH) \ + M(21, CANNOT_READ_ALL_DATA_FROM_TAB_SEPARATED_INPUT) \ + M(22, CANNOT_PARSE_ALL_VALUE_FROM_TAB_SEPARATED_INPUT) \ + M(23, CANNOT_READ_FROM_ISTREAM) \ + M(24, CANNOT_WRITE_TO_OSTREAM) \ + M(25, CANNOT_PARSE_ESCAPE_SEQUENCE) \ + M(26, CANNOT_PARSE_QUOTED_STRING) \ + M(27, CANNOT_PARSE_INPUT_ASSERTION_FAILED) \ + M(28, CANNOT_PRINT_FLOAT_OR_DOUBLE_NUMBER) \ + M(29, CANNOT_PRINT_INTEGER) \ + M(30, CANNOT_READ_SIZE_OF_COMPRESSED_CHUNK) \ + M(31, CANNOT_READ_COMPRESSED_CHUNK) \ + M(32, ATTEMPT_TO_READ_AFTER_EOF) \ + M(33, CANNOT_READ_ALL_DATA) \ + M(34, TOO_MANY_ARGUMENTS_FOR_FUNCTION) \ + M(35, TOO_FEW_ARGUMENTS_FOR_FUNCTION) \ + M(36, BAD_ARGUMENTS) \ + M(37, UNKNOWN_ELEMENT_IN_AST) \ + M(38, CANNOT_PARSE_DATE) \ + M(39, TOO_LARGE_SIZE_COMPRESSED) \ + M(40, CHECKSUM_DOESNT_MATCH) \ + M(41, CANNOT_PARSE_DATETIME) \ + M(42, NUMBER_OF_ARGUMENTS_DOESNT_MATCH) \ + M(43, ILLEGAL_TYPE_OF_ARGUMENT) \ + M(44, ILLEGAL_COLUMN) \ + M(45, ILLEGAL_NUMBER_OF_RESULT_COLUMNS) \ + M(46, UNKNOWN_FUNCTION) \ + M(47, UNKNOWN_IDENTIFIER) \ + M(48, NOT_IMPLEMENTED) \ + M(49, LOGICAL_ERROR) \ + M(50, UNKNOWN_TYPE) \ + M(51, EMPTY_LIST_OF_COLUMNS_QUERIED) \ + M(52, COLUMN_QUERIED_MORE_THAN_ONCE) \ + M(53, TYPE_MISMATCH) \ + M(54, STORAGE_DOESNT_ALLOW_PARAMETERS) \ + M(55, STORAGE_REQUIRES_PARAMETER) \ + M(56, UNKNOWN_STORAGE) \ + M(57, TABLE_ALREADY_EXISTS) \ + M(58, TABLE_METADATA_ALREADY_EXISTS) \ + M(59, ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER) \ + M(60, UNKNOWN_TABLE) \ + M(61, ONLY_FILTER_COLUMN_IN_BLOCK) \ + M(62, SYNTAX_ERROR) \ + M(63, UNKNOWN_AGGREGATE_FUNCTION) \ + M(64, CANNOT_READ_AGGREGATE_FUNCTION_FROM_TEXT) \ + M(65, CANNOT_WRITE_AGGREGATE_FUNCTION_AS_TEXT) \ + M(66, NOT_A_COLUMN) \ + M(67, ILLEGAL_KEY_OF_AGGREGATION) \ + M(68, CANNOT_GET_SIZE_OF_FIELD) \ + M(69, ARGUMENT_OUT_OF_BOUND) \ + M(70, CANNOT_CONVERT_TYPE) \ + M(71, CANNOT_WRITE_AFTER_END_OF_BUFFER) \ + M(72, CANNOT_PARSE_NUMBER) \ + M(73, UNKNOWN_FORMAT) \ + M(74, CANNOT_READ_FROM_FILE_DESCRIPTOR) \ + M(75, CANNOT_WRITE_TO_FILE_DESCRIPTOR) \ + M(76, CANNOT_OPEN_FILE) \ + M(77, CANNOT_CLOSE_FILE) \ + M(78, UNKNOWN_TYPE_OF_QUERY) \ + M(79, INCORRECT_FILE_NAME) \ + M(80, INCORRECT_QUERY) \ + M(81, UNKNOWN_DATABASE) \ + M(82, DATABASE_ALREADY_EXISTS) \ + M(83, DIRECTORY_DOESNT_EXIST) \ + M(84, DIRECTORY_ALREADY_EXISTS) \ + M(85, FORMAT_IS_NOT_SUITABLE_FOR_INPUT) \ + M(86, RECEIVED_ERROR_FROM_REMOTE_IO_SERVER) \ + M(87, CANNOT_SEEK_THROUGH_FILE) \ + M(88, CANNOT_TRUNCATE_FILE) \ + M(89, UNKNOWN_COMPRESSION_METHOD) \ + M(90, EMPTY_LIST_OF_COLUMNS_PASSED) \ + M(91, SIZES_OF_MARKS_FILES_ARE_INCONSISTENT) \ + M(92, EMPTY_DATA_PASSED) \ + M(93, UNKNOWN_AGGREGATED_DATA_VARIANT) \ + M(94, CANNOT_MERGE_DIFFERENT_AGGREGATED_DATA_VARIANTS) \ + M(95, CANNOT_READ_FROM_SOCKET) \ + M(96, CANNOT_WRITE_TO_SOCKET) \ + M(97, CANNOT_READ_ALL_DATA_FROM_CHUNKED_INPUT) \ + M(98, CANNOT_WRITE_TO_EMPTY_BLOCK_OUTPUT_STREAM) \ + M(99, UNKNOWN_PACKET_FROM_CLIENT) \ + M(100, UNKNOWN_PACKET_FROM_SERVER) \ + M(101, UNEXPECTED_PACKET_FROM_CLIENT) \ + M(102, UNEXPECTED_PACKET_FROM_SERVER) \ + M(103, RECEIVED_DATA_FOR_WRONG_QUERY_ID) \ + M(104, TOO_SMALL_BUFFER_SIZE) \ + M(105, CANNOT_READ_HISTORY) \ + M(106, CANNOT_APPEND_HISTORY) \ + M(107, FILE_DOESNT_EXIST) \ + M(108, NO_DATA_TO_INSERT) \ + M(109, CANNOT_BLOCK_SIGNAL) \ + M(110, CANNOT_UNBLOCK_SIGNAL) \ + M(111, CANNOT_MANIPULATE_SIGSET) \ + M(112, CANNOT_WAIT_FOR_SIGNAL) \ + M(113, THERE_IS_NO_SESSION) \ + M(114, CANNOT_CLOCK_GETTIME) \ + M(115, UNKNOWN_SETTING) \ + M(116, THERE_IS_NO_DEFAULT_VALUE) \ + M(117, INCORRECT_DATA) \ + M(119, ENGINE_REQUIRED) \ + M(120, CANNOT_INSERT_VALUE_OF_DIFFERENT_SIZE_INTO_TUPLE) \ + M(121, UNSUPPORTED_JOIN_KEYS) \ + M(122, INCOMPATIBLE_COLUMNS) \ + M(123, UNKNOWN_TYPE_OF_AST_NODE) \ + M(124, INCORRECT_ELEMENT_OF_SET) \ + M(125, INCORRECT_RESULT_OF_SCALAR_SUBQUERY) \ + M(126, CANNOT_GET_RETURN_TYPE) \ + M(127, ILLEGAL_INDEX) \ + M(128, TOO_LARGE_ARRAY_SIZE) \ + M(129, FUNCTION_IS_SPECIAL) \ + M(130, CANNOT_READ_ARRAY_FROM_TEXT) \ + M(131, TOO_LARGE_STRING_SIZE) \ + M(133, AGGREGATE_FUNCTION_DOESNT_ALLOW_PARAMETERS) \ + M(134, PARAMETERS_TO_AGGREGATE_FUNCTIONS_MUST_BE_LITERALS) \ + M(135, ZERO_ARRAY_OR_TUPLE_INDEX) \ + M(137, UNKNOWN_ELEMENT_IN_CONFIG) \ + M(138, EXCESSIVE_ELEMENT_IN_CONFIG) \ + M(139, NO_ELEMENTS_IN_CONFIG) \ + M(140, ALL_REQUESTED_COLUMNS_ARE_MISSING) \ + M(141, SAMPLING_NOT_SUPPORTED) \ + M(142, NOT_FOUND_NODE) \ + M(143, FOUND_MORE_THAN_ONE_NODE) \ + M(144, FIRST_DATE_IS_BIGGER_THAN_LAST_DATE) \ + M(145, UNKNOWN_OVERFLOW_MODE) \ + M(146, QUERY_SECTION_DOESNT_MAKE_SENSE) \ + M(147, NOT_FOUND_FUNCTION_ELEMENT_FOR_AGGREGATE) \ + M(148, NOT_FOUND_RELATION_ELEMENT_FOR_CONDITION) \ + M(149, NOT_FOUND_RHS_ELEMENT_FOR_CONDITION) \ + M(150, EMPTY_LIST_OF_ATTRIBUTES_PASSED) \ + M(151, INDEX_OF_COLUMN_IN_SORT_CLAUSE_IS_OUT_OF_RANGE) \ + M(152, UNKNOWN_DIRECTION_OF_SORTING) \ + M(153, ILLEGAL_DIVISION) \ + M(154, AGGREGATE_FUNCTION_NOT_APPLICABLE) \ + M(155, UNKNOWN_RELATION) \ + M(156, DICTIONARIES_WAS_NOT_LOADED) \ + M(157, ILLEGAL_OVERFLOW_MODE) \ + M(158, TOO_MANY_ROWS) \ + M(159, TIMEOUT_EXCEEDED) \ + M(160, TOO_SLOW) \ + M(161, TOO_MANY_COLUMNS) \ + M(162, TOO_DEEP_SUBQUERIES) \ + M(163, TOO_DEEP_PIPELINE) \ + M(164, READONLY) \ + M(165, TOO_MANY_TEMPORARY_COLUMNS) \ + M(166, TOO_MANY_TEMPORARY_NON_CONST_COLUMNS) \ + M(167, TOO_DEEP_AST) \ + M(168, TOO_BIG_AST) \ + M(169, BAD_TYPE_OF_FIELD) \ + M(170, BAD_GET) \ + M(172, CANNOT_CREATE_DIRECTORY) \ + M(173, CANNOT_ALLOCATE_MEMORY) \ + M(174, CYCLIC_ALIASES) \ + M(176, CHUNK_NOT_FOUND) \ + M(177, DUPLICATE_CHUNK_NAME) \ + M(178, MULTIPLE_ALIASES_FOR_EXPRESSION) \ + M(179, MULTIPLE_EXPRESSIONS_FOR_ALIAS) \ + M(180, THERE_IS_NO_PROFILE) \ + M(181, ILLEGAL_FINAL) \ + M(182, ILLEGAL_PREWHERE) \ + M(183, UNEXPECTED_EXPRESSION) \ + M(184, ILLEGAL_AGGREGATION) \ + M(185, UNSUPPORTED_MYISAM_BLOCK_TYPE) \ + M(186, UNSUPPORTED_COLLATION_LOCALE) \ + M(187, COLLATION_COMPARISON_FAILED) \ + M(188, UNKNOWN_ACTION) \ + M(189, TABLE_MUST_NOT_BE_CREATED_MANUALLY) \ + M(190, SIZES_OF_ARRAYS_DOESNT_MATCH) \ + M(191, SET_SIZE_LIMIT_EXCEEDED) \ + M(192, UNKNOWN_USER) \ + M(193, WRONG_PASSWORD) \ + M(194, REQUIRED_PASSWORD) \ + M(195, IP_ADDRESS_NOT_ALLOWED) \ + M(196, UNKNOWN_ADDRESS_PATTERN_TYPE) \ + M(197, SERVER_REVISION_IS_TOO_OLD) \ + M(198, DNS_ERROR) \ + M(199, UNKNOWN_QUOTA) \ + M(200, QUOTA_DOESNT_ALLOW_KEYS) \ + M(201, QUOTA_EXPIRED) \ + M(202, TOO_MANY_SIMULTANEOUS_QUERIES) \ + M(203, NO_FREE_CONNECTION) \ + M(204, CANNOT_FSYNC) \ + M(205, NESTED_TYPE_TOO_DEEP) \ + M(206, ALIAS_REQUIRED) \ + M(207, AMBIGUOUS_IDENTIFIER) \ + M(208, EMPTY_NESTED_TABLE) \ + M(209, SOCKET_TIMEOUT) \ + M(210, NETWORK_ERROR) \ + M(211, EMPTY_QUERY) \ + M(212, UNKNOWN_LOAD_BALANCING) \ + M(213, UNKNOWN_TOTALS_MODE) \ + M(214, CANNOT_STATVFS) \ + M(215, NOT_AN_AGGREGATE) \ + M(216, QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING) \ + M(217, CLIENT_HAS_CONNECTED_TO_WRONG_PORT) \ + M(218, TABLE_IS_DROPPED) \ + M(219, DATABASE_NOT_EMPTY) \ + M(220, DUPLICATE_INTERSERVER_IO_ENDPOINT) \ + M(221, NO_SUCH_INTERSERVER_IO_ENDPOINT) \ + M(222, ADDING_REPLICA_TO_NON_EMPTY_TABLE) \ + M(223, UNEXPECTED_AST_STRUCTURE) \ + M(224, REPLICA_IS_ALREADY_ACTIVE) \ + M(225, NO_ZOOKEEPER) \ + M(226, NO_FILE_IN_DATA_PART) \ + M(227, UNEXPECTED_FILE_IN_DATA_PART) \ + M(228, BAD_SIZE_OF_FILE_IN_DATA_PART) \ + M(229, QUERY_IS_TOO_LARGE) \ + M(230, NOT_FOUND_EXPECTED_DATA_PART) \ + M(231, TOO_MANY_UNEXPECTED_DATA_PARTS) \ + M(232, NO_SUCH_DATA_PART) \ + M(233, BAD_DATA_PART_NAME) \ + M(234, NO_REPLICA_HAS_PART) \ + M(235, DUPLICATE_DATA_PART) \ + M(236, ABORTED) \ + M(237, NO_REPLICA_NAME_GIVEN) \ + M(238, FORMAT_VERSION_TOO_OLD) \ + M(239, CANNOT_MUNMAP) \ + M(240, CANNOT_MREMAP) \ + M(241, MEMORY_LIMIT_EXCEEDED) \ + M(242, TABLE_IS_READ_ONLY) \ + M(243, NOT_ENOUGH_SPACE) \ + M(244, UNEXPECTED_ZOOKEEPER_ERROR) \ + M(246, CORRUPTED_DATA) \ + M(247, INCORRECT_MARK) \ + M(248, INVALID_PARTITION_VALUE) \ + M(250, NOT_ENOUGH_BLOCK_NUMBERS) \ + M(251, NO_SUCH_REPLICA) \ + M(252, TOO_MANY_PARTS) \ + M(253, REPLICA_IS_ALREADY_EXIST) \ + M(254, NO_ACTIVE_REPLICAS) \ + M(255, TOO_MANY_RETRIES_TO_FETCH_PARTS) \ + M(256, PARTITION_ALREADY_EXISTS) \ + M(257, PARTITION_DOESNT_EXIST) \ + M(258, UNION_ALL_RESULT_STRUCTURES_MISMATCH) \ + M(260, CLIENT_OUTPUT_FORMAT_SPECIFIED) \ + M(261, UNKNOWN_BLOCK_INFO_FIELD) \ + M(262, BAD_COLLATION) \ + M(263, CANNOT_COMPILE_CODE) \ + M(264, INCOMPATIBLE_TYPE_OF_JOIN) \ + M(265, NO_AVAILABLE_REPLICA) \ + M(266, MISMATCH_REPLICAS_DATA_SOURCES) \ + M(267, STORAGE_DOESNT_SUPPORT_PARALLEL_REPLICAS) \ + M(268, CPUID_ERROR) \ + M(269, INFINITE_LOOP) \ + M(270, CANNOT_COMPRESS) \ + M(271, CANNOT_DECOMPRESS) \ + M(272, CANNOT_IO_SUBMIT) \ + M(273, CANNOT_IO_GETEVENTS) \ + M(274, AIO_READ_ERROR) \ + M(275, AIO_WRITE_ERROR) \ + M(277, INDEX_NOT_USED) \ + M(279, ALL_CONNECTION_TRIES_FAILED) \ + M(280, NO_AVAILABLE_DATA) \ + M(281, DICTIONARY_IS_EMPTY) \ + M(282, INCORRECT_INDEX) \ + M(283, UNKNOWN_DISTRIBUTED_PRODUCT_MODE) \ + M(284, WRONG_GLOBAL_SUBQUERY) \ + M(285, TOO_FEW_LIVE_REPLICAS) \ + M(286, UNSATISFIED_QUORUM_FOR_PREVIOUS_WRITE) \ + M(287, UNKNOWN_FORMAT_VERSION) \ + M(288, DISTRIBUTED_IN_JOIN_SUBQUERY_DENIED) \ + M(289, REPLICA_IS_NOT_IN_QUORUM) \ + M(290, LIMIT_EXCEEDED) \ + M(291, DATABASE_ACCESS_DENIED) \ + M(293, MONGODB_CANNOT_AUTHENTICATE) \ + M(294, INVALID_BLOCK_EXTRA_INFO) \ + M(295, RECEIVED_EMPTY_DATA) \ + M(296, NO_REMOTE_SHARD_FOUND) \ + M(297, SHARD_HAS_NO_CONNECTIONS) \ + M(298, CANNOT_PIPE) \ + M(299, CANNOT_FORK) \ + M(300, CANNOT_DLSYM) \ + M(301, CANNOT_CREATE_CHILD_PROCESS) \ + M(302, CHILD_WAS_NOT_EXITED_NORMALLY) \ + M(303, CANNOT_SELECT) \ + M(304, CANNOT_WAITPID) \ + M(305, TABLE_WAS_NOT_DROPPED) \ + M(306, TOO_DEEP_RECURSION) \ + M(307, TOO_MANY_BYTES) \ + M(308, UNEXPECTED_NODE_IN_ZOOKEEPER) \ + M(309, FUNCTION_CANNOT_HAVE_PARAMETERS) \ + M(317, INVALID_SHARD_WEIGHT) \ + M(318, INVALID_CONFIG_PARAMETER) \ + M(319, UNKNOWN_STATUS_OF_INSERT) \ + M(321, VALUE_IS_OUT_OF_RANGE_OF_DATA_TYPE) \ + M(335, BARRIER_TIMEOUT) \ + M(336, UNKNOWN_DATABASE_ENGINE) \ + M(337, DDL_GUARD_IS_ACTIVE) \ + M(341, UNFINISHED) \ + M(342, METADATA_MISMATCH) \ + M(344, SUPPORT_IS_DISABLED) \ + M(345, TABLE_DIFFERS_TOO_MUCH) \ + M(346, CANNOT_CONVERT_CHARSET) \ + M(347, CANNOT_LOAD_CONFIG) \ + M(349, CANNOT_INSERT_NULL_IN_ORDINARY_COLUMN) \ + M(350, INCOMPATIBLE_SOURCE_TABLES) \ + M(351, AMBIGUOUS_TABLE_NAME) \ + M(352, AMBIGUOUS_COLUMN_NAME) \ + M(353, INDEX_OF_POSITIONAL_ARGUMENT_IS_OUT_OF_RANGE) \ + M(354, ZLIB_INFLATE_FAILED) \ + M(355, ZLIB_DEFLATE_FAILED) \ + M(356, BAD_LAMBDA) \ + M(357, RESERVED_IDENTIFIER_NAME) \ + M(358, INTO_OUTFILE_NOT_ALLOWED) \ + M(359, TABLE_SIZE_EXCEEDS_MAX_DROP_SIZE_LIMIT) \ + M(360, CANNOT_CREATE_CHARSET_CONVERTER) \ + M(361, SEEK_POSITION_OUT_OF_BOUND) \ + M(362, CURRENT_WRITE_BUFFER_IS_EXHAUSTED) \ + M(363, CANNOT_CREATE_IO_BUFFER) \ + M(364, RECEIVED_ERROR_TOO_MANY_REQUESTS) \ + M(366, SIZES_OF_NESTED_COLUMNS_ARE_INCONSISTENT) \ + M(367, TOO_MANY_FETCHES) \ + M(369, ALL_REPLICAS_ARE_STALE) \ + M(370, DATA_TYPE_CANNOT_BE_USED_IN_TABLES) \ + M(371, INCONSISTENT_CLUSTER_DEFINITION) \ + M(372, SESSION_NOT_FOUND) \ + M(373, SESSION_IS_LOCKED) \ + M(374, INVALID_SESSION_TIMEOUT) \ + M(375, CANNOT_DLOPEN) \ + M(376, CANNOT_PARSE_UUID) \ + M(377, ILLEGAL_SYNTAX_FOR_DATA_TYPE) \ + M(378, DATA_TYPE_CANNOT_HAVE_ARGUMENTS) \ + M(379, UNKNOWN_STATUS_OF_DISTRIBUTED_DDL_TASK) \ + M(380, CANNOT_KILL) \ + M(381, HTTP_LENGTH_REQUIRED) \ + M(382, CANNOT_LOAD_CATBOOST_MODEL) \ + M(383, CANNOT_APPLY_CATBOOST_MODEL) \ + M(384, PART_IS_TEMPORARILY_LOCKED) \ + M(385, MULTIPLE_STREAMS_REQUIRED) \ + M(386, NO_COMMON_TYPE) \ + M(387, DICTIONARY_ALREADY_EXISTS) \ + M(388, CANNOT_ASSIGN_OPTIMIZE) \ + M(389, INSERT_WAS_DEDUPLICATED) \ + M(390, CANNOT_GET_CREATE_TABLE_QUERY) \ + M(391, EXTERNAL_LIBRARY_ERROR) \ + M(392, QUERY_IS_PROHIBITED) \ + M(393, THERE_IS_NO_QUERY) \ + M(394, QUERY_WAS_CANCELLED) \ + M(395, FUNCTION_THROW_IF_VALUE_IS_NON_ZERO) \ + M(396, TOO_MANY_ROWS_OR_BYTES) \ + M(397, QUERY_IS_NOT_SUPPORTED_IN_MATERIALIZED_VIEW) \ + M(398, UNKNOWN_MUTATION_COMMAND) \ + M(399, FORMAT_IS_NOT_SUITABLE_FOR_OUTPUT) \ + M(400, CANNOT_STAT) \ + M(401, FEATURE_IS_NOT_ENABLED_AT_BUILD_TIME) \ + M(402, CANNOT_IOSETUP) \ + M(403, INVALID_JOIN_ON_EXPRESSION) \ + M(404, BAD_ODBC_CONNECTION_STRING) \ + M(405, PARTITION_SIZE_EXCEEDS_MAX_DROP_SIZE_LIMIT) \ + M(406, TOP_AND_LIMIT_TOGETHER) \ + M(407, DECIMAL_OVERFLOW) \ + M(408, BAD_REQUEST_PARAMETER) \ + M(409, EXTERNAL_EXECUTABLE_NOT_FOUND) \ + M(410, EXTERNAL_SERVER_IS_NOT_RESPONDING) \ + M(411, PTHREAD_ERROR) \ + M(412, NETLINK_ERROR) \ + M(413, CANNOT_SET_SIGNAL_HANDLER) \ + M(415, ALL_REPLICAS_LOST) \ + M(416, REPLICA_STATUS_CHANGED) \ + M(417, EXPECTED_ALL_OR_ANY) \ + M(418, UNKNOWN_JOIN) \ + M(419, MULTIPLE_ASSIGNMENTS_TO_COLUMN) \ + M(420, CANNOT_UPDATE_COLUMN) \ + M(421, CANNOT_ADD_DIFFERENT_AGGREGATE_STATES) \ + M(422, UNSUPPORTED_URI_SCHEME) \ + M(423, CANNOT_GETTIMEOFDAY) \ + M(424, CANNOT_LINK) \ + M(425, SYSTEM_ERROR) \ + M(427, CANNOT_COMPILE_REGEXP) \ + M(428, UNKNOWN_LOG_LEVEL) \ + M(429, FAILED_TO_GETPWUID) \ + M(430, MISMATCHING_USERS_FOR_PROCESS_AND_DATA) \ + M(431, ILLEGAL_SYNTAX_FOR_CODEC_TYPE) \ + M(432, UNKNOWN_CODEC) \ + M(433, ILLEGAL_CODEC_PARAMETER) \ + M(434, CANNOT_PARSE_PROTOBUF_SCHEMA) \ + M(435, NO_DATA_FOR_REQUIRED_PROTOBUF_FIELD) \ + M(436, PROTOBUF_BAD_CAST) \ + M(437, PROTOBUF_FIELD_NOT_REPEATED) \ + M(438, DATA_TYPE_CANNOT_BE_PROMOTED) \ + M(439, CANNOT_SCHEDULE_TASK) \ + M(440, INVALID_LIMIT_EXPRESSION) \ + M(441, CANNOT_PARSE_DOMAIN_VALUE_FROM_STRING) \ + M(442, BAD_DATABASE_FOR_TEMPORARY_TABLE) \ + M(443, NO_COMMON_COLUMNS_WITH_PROTOBUF_SCHEMA) \ + M(444, UNKNOWN_PROTOBUF_FORMAT) \ + M(445, CANNOT_MPROTECT) \ + M(446, FUNCTION_NOT_ALLOWED) \ + M(447, HYPERSCAN_CANNOT_SCAN_TEXT) \ + M(448, BROTLI_READ_FAILED) \ + M(449, BROTLI_WRITE_FAILED) \ + M(450, BAD_TTL_EXPRESSION) \ + M(451, BAD_TTL_FILE) \ + M(452, SETTING_CONSTRAINT_VIOLATION) \ + M(453, MYSQL_CLIENT_INSUFFICIENT_CAPABILITIES) \ + M(454, OPENSSL_ERROR) \ + M(455, SUSPICIOUS_TYPE_FOR_LOW_CARDINALITY) \ + M(456, UNKNOWN_QUERY_PARAMETER) \ + M(457, BAD_QUERY_PARAMETER) \ + M(458, CANNOT_UNLINK) \ + M(459, CANNOT_SET_THREAD_PRIORITY) \ + M(460, CANNOT_CREATE_TIMER) \ + M(461, CANNOT_SET_TIMER_PERIOD) \ + M(462, CANNOT_DELETE_TIMER) \ + M(463, CANNOT_FCNTL) \ + M(464, CANNOT_PARSE_ELF) \ + M(465, CANNOT_PARSE_DWARF) \ + M(466, INSECURE_PATH) \ + M(467, CANNOT_PARSE_BOOL) \ + M(468, CANNOT_PTHREAD_ATTR) \ + M(469, VIOLATED_CONSTRAINT) \ + M(470, QUERY_IS_NOT_SUPPORTED_IN_LIVE_VIEW) \ + M(471, INVALID_SETTING_VALUE) \ + M(472, READONLY_SETTING) \ + M(473, DEADLOCK_AVOIDED) \ + M(474, INVALID_TEMPLATE_FORMAT) \ + M(475, INVALID_WITH_FILL_EXPRESSION) \ + M(476, WITH_TIES_WITHOUT_ORDER_BY) \ + M(477, INVALID_USAGE_OF_INPUT) \ + M(478, UNKNOWN_POLICY) \ + M(479, UNKNOWN_DISK) \ + M(480, UNKNOWN_PROTOCOL) \ + M(481, PATH_ACCESS_DENIED) \ + M(482, DICTIONARY_ACCESS_DENIED) \ + M(483, TOO_MANY_REDIRECTS) \ + M(484, INTERNAL_REDIS_ERROR) \ + M(485, SCALAR_ALREADY_EXISTS) \ + M(487, CANNOT_GET_CREATE_DICTIONARY_QUERY) \ + M(488, UNKNOWN_DICTIONARY) \ + M(489, INCORRECT_DICTIONARY_DEFINITION) \ + M(490, CANNOT_FORMAT_DATETIME) \ + M(491, UNACCEPTABLE_URL) \ + M(492, ACCESS_ENTITY_NOT_FOUND) \ + M(493, ACCESS_ENTITY_ALREADY_EXISTS) \ + M(494, ACCESS_ENTITY_FOUND_DUPLICATES) \ + M(495, ACCESS_STORAGE_READONLY) \ + M(496, QUOTA_REQUIRES_CLIENT_KEY) \ + M(497, ACCESS_DENIED) \ + M(498, LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED) \ + M(499, S3_ERROR) \ + M(501, CANNOT_CREATE_DATABASE) \ + M(502, CANNOT_SIGQUEUE) \ + M(503, AGGREGATE_FUNCTION_THROW) \ + M(504, FILE_ALREADY_EXISTS) \ + M(505, CANNOT_DELETE_DIRECTORY) \ + M(506, UNEXPECTED_ERROR_CODE) \ + M(507, UNABLE_TO_SKIP_UNUSED_SHARDS) \ + M(508, UNKNOWN_ACCESS_TYPE) \ + M(509, INVALID_GRANT) \ + M(510, CACHE_DICTIONARY_UPDATE_FAIL) \ + M(511, UNKNOWN_ROLE) \ + M(512, SET_NON_GRANTED_ROLE) \ + M(513, UNKNOWN_PART_TYPE) \ + M(514, ACCESS_STORAGE_FOR_INSERTION_NOT_FOUND) \ + M(515, INCORRECT_ACCESS_ENTITY_DEFINITION) \ + M(516, AUTHENTICATION_FAILED) \ + M(517, CANNOT_ASSIGN_ALTER) \ + M(518, CANNOT_COMMIT_OFFSET) \ + M(519, NO_REMOTE_SHARD_AVAILABLE) \ + M(520, CANNOT_DETACH_DICTIONARY_AS_TABLE) \ + M(521, ATOMIC_RENAME_FAIL) \ + M(523, UNKNOWN_ROW_POLICY) \ + M(524, ALTER_OF_COLUMN_IS_FORBIDDEN) \ + M(525, INCORRECT_DISK_INDEX) \ + M(526, UNKNOWN_VOLUME_TYPE) \ + M(527, NO_SUITABLE_FUNCTION_IMPLEMENTATION) \ + M(528, CASSANDRA_INTERNAL_ERROR) \ + M(529, NOT_A_LEADER) \ + M(530, CANNOT_CONNECT_RABBITMQ) \ + M(531, CANNOT_FSTAT) \ + M(532, LDAP_ERROR) \ + M(533, INCONSISTENT_RESERVATIONS) \ + M(534, NO_RESERVATIONS_PROVIDED) \ + M(535, UNKNOWN_RAID_TYPE) \ + M(536, CANNOT_RESTORE_FROM_FIELD_DUMP) \ + M(537, ILLEGAL_MYSQL_VARIABLE) \ + M(538, MYSQL_SYNTAX_ERROR) \ + M(539, CANNOT_BIND_RABBITMQ_EXCHANGE) \ + M(540, CANNOT_DECLARE_RABBITMQ_EXCHANGE) \ + M(541, CANNOT_CREATE_RABBITMQ_QUEUE_BINDING) \ + M(542, CANNOT_REMOVE_RABBITMQ_EXCHANGE) \ + M(543, UNKNOWN_MYSQL_DATATYPES_SUPPORT_LEVEL) \ + M(544, ROW_AND_ROWS_TOGETHER) \ + M(545, FIRST_AND_NEXT_TOGETHER) \ + M(546, NO_ROW_DELIMITER) \ + M(547, INVALID_RAID_TYPE) \ + M(548, UNKNOWN_VOLUME) \ + M(549, DATA_TYPE_CANNOT_BE_USED_IN_KEY) \ + M(550, CONDITIONAL_TREE_PARENT_NOT_FOUND) \ + M(551, ILLEGAL_PROJECTION_MANIPULATOR) \ + M(552, UNRECOGNIZED_ARGUMENTS) \ + \ + M(999, KEEPER_EXCEPTION) \ + M(1000, POCO_EXCEPTION) \ + M(1001, STD_EXCEPTION) \ + M(1002, UNKNOWN_EXCEPTION) \ + +/* See END */ + namespace DB { namespace ErrorCodes { - /** Previously, these constants were located in one enum. - * But in this case there is a problem: when you add a new constant, you need to recompile - * all translation units that use at least one constant (almost the whole project). - * Therefore it is made so that definitions of constants are located here, in one file, - * and their declaration are in different files, at the place of use. - */ + #define M(VALUE, NAME) extern const Value NAME = VALUE; + APPLY_FOR_ERROR_CODES(M) + #undef M - extern const int UNSUPPORTED_METHOD = 1; - extern const int UNSUPPORTED_PARAMETER = 2; - extern const int UNEXPECTED_END_OF_FILE = 3; - extern const int EXPECTED_END_OF_FILE = 4; - extern const int CANNOT_PARSE_TEXT = 6; - extern const int INCORRECT_NUMBER_OF_COLUMNS = 7; - extern const int THERE_IS_NO_COLUMN = 8; - extern const int SIZES_OF_COLUMNS_DOESNT_MATCH = 9; - extern const int NOT_FOUND_COLUMN_IN_BLOCK = 10; - extern const int POSITION_OUT_OF_BOUND = 11; - extern const int PARAMETER_OUT_OF_BOUND = 12; - extern const int SIZES_OF_COLUMNS_IN_TUPLE_DOESNT_MATCH = 13; - extern const int DUPLICATE_COLUMN = 15; - extern const int NO_SUCH_COLUMN_IN_TABLE = 16; - extern const int DELIMITER_IN_STRING_LITERAL_DOESNT_MATCH = 17; - extern const int CANNOT_INSERT_ELEMENT_INTO_CONSTANT_COLUMN = 18; - extern const int SIZE_OF_FIXED_STRING_DOESNT_MATCH = 19; - extern const int NUMBER_OF_COLUMNS_DOESNT_MATCH = 20; - extern const int CANNOT_READ_ALL_DATA_FROM_TAB_SEPARATED_INPUT = 21; - extern const int CANNOT_PARSE_ALL_VALUE_FROM_TAB_SEPARATED_INPUT = 22; - extern const int CANNOT_READ_FROM_ISTREAM = 23; - extern const int CANNOT_WRITE_TO_OSTREAM = 24; - extern const int CANNOT_PARSE_ESCAPE_SEQUENCE = 25; - extern const int CANNOT_PARSE_QUOTED_STRING = 26; - extern const int CANNOT_PARSE_INPUT_ASSERTION_FAILED = 27; - extern const int CANNOT_PRINT_FLOAT_OR_DOUBLE_NUMBER = 28; - extern const int CANNOT_PRINT_INTEGER = 29; - extern const int CANNOT_READ_SIZE_OF_COMPRESSED_CHUNK = 30; - extern const int CANNOT_READ_COMPRESSED_CHUNK = 31; - extern const int ATTEMPT_TO_READ_AFTER_EOF = 32; - extern const int CANNOT_READ_ALL_DATA = 33; - extern const int TOO_MANY_ARGUMENTS_FOR_FUNCTION = 34; - extern const int TOO_FEW_ARGUMENTS_FOR_FUNCTION = 35; - extern const int BAD_ARGUMENTS = 36; - extern const int UNKNOWN_ELEMENT_IN_AST = 37; - extern const int CANNOT_PARSE_DATE = 38; - extern const int TOO_LARGE_SIZE_COMPRESSED = 39; - extern const int CHECKSUM_DOESNT_MATCH = 40; - extern const int CANNOT_PARSE_DATETIME = 41; - extern const int NUMBER_OF_ARGUMENTS_DOESNT_MATCH = 42; - extern const int ILLEGAL_TYPE_OF_ARGUMENT = 43; - extern const int ILLEGAL_COLUMN = 44; - extern const int ILLEGAL_NUMBER_OF_RESULT_COLUMNS = 45; - extern const int UNKNOWN_FUNCTION = 46; - extern const int UNKNOWN_IDENTIFIER = 47; - extern const int NOT_IMPLEMENTED = 48; - extern const int LOGICAL_ERROR = 49; - extern const int UNKNOWN_TYPE = 50; - extern const int EMPTY_LIST_OF_COLUMNS_QUERIED = 51; - extern const int COLUMN_QUERIED_MORE_THAN_ONCE = 52; - extern const int TYPE_MISMATCH = 53; - extern const int STORAGE_DOESNT_ALLOW_PARAMETERS = 54; - extern const int STORAGE_REQUIRES_PARAMETER = 55; - extern const int UNKNOWN_STORAGE = 56; - extern const int TABLE_ALREADY_EXISTS = 57; - extern const int TABLE_METADATA_ALREADY_EXISTS = 58; - extern const int ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER = 59; - extern const int UNKNOWN_TABLE = 60; - extern const int ONLY_FILTER_COLUMN_IN_BLOCK = 61; - extern const int SYNTAX_ERROR = 62; - extern const int UNKNOWN_AGGREGATE_FUNCTION = 63; - extern const int CANNOT_READ_AGGREGATE_FUNCTION_FROM_TEXT = 64; - extern const int CANNOT_WRITE_AGGREGATE_FUNCTION_AS_TEXT = 65; - extern const int NOT_A_COLUMN = 66; - extern const int ILLEGAL_KEY_OF_AGGREGATION = 67; - extern const int CANNOT_GET_SIZE_OF_FIELD = 68; - extern const int ARGUMENT_OUT_OF_BOUND = 69; - extern const int CANNOT_CONVERT_TYPE = 70; - extern const int CANNOT_WRITE_AFTER_END_OF_BUFFER = 71; - extern const int CANNOT_PARSE_NUMBER = 72; - extern const int UNKNOWN_FORMAT = 73; - extern const int CANNOT_READ_FROM_FILE_DESCRIPTOR = 74; - extern const int CANNOT_WRITE_TO_FILE_DESCRIPTOR = 75; - extern const int CANNOT_OPEN_FILE = 76; - extern const int CANNOT_CLOSE_FILE = 77; - extern const int UNKNOWN_TYPE_OF_QUERY = 78; - extern const int INCORRECT_FILE_NAME = 79; - extern const int INCORRECT_QUERY = 80; - extern const int UNKNOWN_DATABASE = 81; - extern const int DATABASE_ALREADY_EXISTS = 82; - extern const int DIRECTORY_DOESNT_EXIST = 83; - extern const int DIRECTORY_ALREADY_EXISTS = 84; - extern const int FORMAT_IS_NOT_SUITABLE_FOR_INPUT = 85; - extern const int RECEIVED_ERROR_FROM_REMOTE_IO_SERVER = 86; - extern const int CANNOT_SEEK_THROUGH_FILE = 87; - extern const int CANNOT_TRUNCATE_FILE = 88; - extern const int UNKNOWN_COMPRESSION_METHOD = 89; - extern const int EMPTY_LIST_OF_COLUMNS_PASSED = 90; - extern const int SIZES_OF_MARKS_FILES_ARE_INCONSISTENT = 91; - extern const int EMPTY_DATA_PASSED = 92; - extern const int UNKNOWN_AGGREGATED_DATA_VARIANT = 93; - extern const int CANNOT_MERGE_DIFFERENT_AGGREGATED_DATA_VARIANTS = 94; - extern const int CANNOT_READ_FROM_SOCKET = 95; - extern const int CANNOT_WRITE_TO_SOCKET = 96; - extern const int CANNOT_READ_ALL_DATA_FROM_CHUNKED_INPUT = 97; - extern const int CANNOT_WRITE_TO_EMPTY_BLOCK_OUTPUT_STREAM = 98; - extern const int UNKNOWN_PACKET_FROM_CLIENT = 99; - extern const int UNKNOWN_PACKET_FROM_SERVER = 100; - extern const int UNEXPECTED_PACKET_FROM_CLIENT = 101; - extern const int UNEXPECTED_PACKET_FROM_SERVER = 102; - extern const int RECEIVED_DATA_FOR_WRONG_QUERY_ID = 103; - extern const int TOO_SMALL_BUFFER_SIZE = 104; - extern const int CANNOT_READ_HISTORY = 105; - extern const int CANNOT_APPEND_HISTORY = 106; - extern const int FILE_DOESNT_EXIST = 107; - extern const int NO_DATA_TO_INSERT = 108; - extern const int CANNOT_BLOCK_SIGNAL = 109; - extern const int CANNOT_UNBLOCK_SIGNAL = 110; - extern const int CANNOT_MANIPULATE_SIGSET = 111; - extern const int CANNOT_WAIT_FOR_SIGNAL = 112; - extern const int THERE_IS_NO_SESSION = 113; - extern const int CANNOT_CLOCK_GETTIME = 114; - extern const int UNKNOWN_SETTING = 115; - extern const int THERE_IS_NO_DEFAULT_VALUE = 116; - extern const int INCORRECT_DATA = 117; - extern const int ENGINE_REQUIRED = 119; - extern const int CANNOT_INSERT_VALUE_OF_DIFFERENT_SIZE_INTO_TUPLE = 120; - extern const int UNSUPPORTED_JOIN_KEYS = 121; - extern const int INCOMPATIBLE_COLUMNS = 122; - extern const int UNKNOWN_TYPE_OF_AST_NODE = 123; - extern const int INCORRECT_ELEMENT_OF_SET = 124; - extern const int INCORRECT_RESULT_OF_SCALAR_SUBQUERY = 125; - extern const int CANNOT_GET_RETURN_TYPE = 126; - extern const int ILLEGAL_INDEX = 127; - extern const int TOO_LARGE_ARRAY_SIZE = 128; - extern const int FUNCTION_IS_SPECIAL = 129; - extern const int CANNOT_READ_ARRAY_FROM_TEXT = 130; - extern const int TOO_LARGE_STRING_SIZE = 131; - extern const int AGGREGATE_FUNCTION_DOESNT_ALLOW_PARAMETERS = 133; - extern const int PARAMETERS_TO_AGGREGATE_FUNCTIONS_MUST_BE_LITERALS = 134; - extern const int ZERO_ARRAY_OR_TUPLE_INDEX = 135; - extern const int UNKNOWN_ELEMENT_IN_CONFIG = 137; - extern const int EXCESSIVE_ELEMENT_IN_CONFIG = 138; - extern const int NO_ELEMENTS_IN_CONFIG = 139; - extern const int ALL_REQUESTED_COLUMNS_ARE_MISSING = 140; - extern const int SAMPLING_NOT_SUPPORTED = 141; - extern const int NOT_FOUND_NODE = 142; - extern const int FOUND_MORE_THAN_ONE_NODE = 143; - extern const int FIRST_DATE_IS_BIGGER_THAN_LAST_DATE = 144; - extern const int UNKNOWN_OVERFLOW_MODE = 145; - extern const int QUERY_SECTION_DOESNT_MAKE_SENSE = 146; - extern const int NOT_FOUND_FUNCTION_ELEMENT_FOR_AGGREGATE = 147; - extern const int NOT_FOUND_RELATION_ELEMENT_FOR_CONDITION = 148; - extern const int NOT_FOUND_RHS_ELEMENT_FOR_CONDITION = 149; - extern const int EMPTY_LIST_OF_ATTRIBUTES_PASSED = 150; - extern const int INDEX_OF_COLUMN_IN_SORT_CLAUSE_IS_OUT_OF_RANGE = 151; - extern const int UNKNOWN_DIRECTION_OF_SORTING = 152; - extern const int ILLEGAL_DIVISION = 153; - extern const int AGGREGATE_FUNCTION_NOT_APPLICABLE = 154; - extern const int UNKNOWN_RELATION = 155; - extern const int DICTIONARIES_WAS_NOT_LOADED = 156; - extern const int ILLEGAL_OVERFLOW_MODE = 157; - extern const int TOO_MANY_ROWS = 158; - extern const int TIMEOUT_EXCEEDED = 159; - extern const int TOO_SLOW = 160; - extern const int TOO_MANY_COLUMNS = 161; - extern const int TOO_DEEP_SUBQUERIES = 162; - extern const int TOO_DEEP_PIPELINE = 163; - extern const int READONLY = 164; - extern const int TOO_MANY_TEMPORARY_COLUMNS = 165; - extern const int TOO_MANY_TEMPORARY_NON_CONST_COLUMNS = 166; - extern const int TOO_DEEP_AST = 167; - extern const int TOO_BIG_AST = 168; - extern const int BAD_TYPE_OF_FIELD = 169; - extern const int BAD_GET = 170; - extern const int CANNOT_CREATE_DIRECTORY = 172; - extern const int CANNOT_ALLOCATE_MEMORY = 173; - extern const int CYCLIC_ALIASES = 174; - extern const int CHUNK_NOT_FOUND = 176; - extern const int DUPLICATE_CHUNK_NAME = 177; - extern const int MULTIPLE_ALIASES_FOR_EXPRESSION = 178; - extern const int MULTIPLE_EXPRESSIONS_FOR_ALIAS = 179; - extern const int THERE_IS_NO_PROFILE = 180; - extern const int ILLEGAL_FINAL = 181; - extern const int ILLEGAL_PREWHERE = 182; - extern const int UNEXPECTED_EXPRESSION = 183; - extern const int ILLEGAL_AGGREGATION = 184; - extern const int UNSUPPORTED_MYISAM_BLOCK_TYPE = 185; - extern const int UNSUPPORTED_COLLATION_LOCALE = 186; - extern const int COLLATION_COMPARISON_FAILED = 187; - extern const int UNKNOWN_ACTION = 188; - extern const int TABLE_MUST_NOT_BE_CREATED_MANUALLY = 189; - extern const int SIZES_OF_ARRAYS_DOESNT_MATCH = 190; - extern const int SET_SIZE_LIMIT_EXCEEDED = 191; - extern const int UNKNOWN_USER = 192; - extern const int WRONG_PASSWORD = 193; - extern const int REQUIRED_PASSWORD = 194; - extern const int IP_ADDRESS_NOT_ALLOWED = 195; - extern const int UNKNOWN_ADDRESS_PATTERN_TYPE = 196; - extern const int SERVER_REVISION_IS_TOO_OLD = 197; - extern const int DNS_ERROR = 198; - extern const int UNKNOWN_QUOTA = 199; - extern const int QUOTA_DOESNT_ALLOW_KEYS = 200; - extern const int QUOTA_EXPIRED = 201; - extern const int TOO_MANY_SIMULTANEOUS_QUERIES = 202; - extern const int NO_FREE_CONNECTION = 203; - extern const int CANNOT_FSYNC = 204; - extern const int NESTED_TYPE_TOO_DEEP = 205; - extern const int ALIAS_REQUIRED = 206; - extern const int AMBIGUOUS_IDENTIFIER = 207; - extern const int EMPTY_NESTED_TABLE = 208; - extern const int SOCKET_TIMEOUT = 209; - extern const int NETWORK_ERROR = 210; - extern const int EMPTY_QUERY = 211; - extern const int UNKNOWN_LOAD_BALANCING = 212; - extern const int UNKNOWN_TOTALS_MODE = 213; - extern const int CANNOT_STATVFS = 214; - extern const int NOT_AN_AGGREGATE = 215; - extern const int QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING = 216; - extern const int CLIENT_HAS_CONNECTED_TO_WRONG_PORT = 217; - extern const int TABLE_IS_DROPPED = 218; - extern const int DATABASE_NOT_EMPTY = 219; - extern const int DUPLICATE_INTERSERVER_IO_ENDPOINT = 220; - extern const int NO_SUCH_INTERSERVER_IO_ENDPOINT = 221; - extern const int ADDING_REPLICA_TO_NON_EMPTY_TABLE = 222; - extern const int UNEXPECTED_AST_STRUCTURE = 223; - extern const int REPLICA_IS_ALREADY_ACTIVE = 224; - extern const int NO_ZOOKEEPER = 225; - extern const int NO_FILE_IN_DATA_PART = 226; - extern const int UNEXPECTED_FILE_IN_DATA_PART = 227; - extern const int BAD_SIZE_OF_FILE_IN_DATA_PART = 228; - extern const int QUERY_IS_TOO_LARGE = 229; - extern const int NOT_FOUND_EXPECTED_DATA_PART = 230; - extern const int TOO_MANY_UNEXPECTED_DATA_PARTS = 231; - extern const int NO_SUCH_DATA_PART = 232; - extern const int BAD_DATA_PART_NAME = 233; - extern const int NO_REPLICA_HAS_PART = 234; - extern const int DUPLICATE_DATA_PART = 235; - extern const int ABORTED = 236; - extern const int NO_REPLICA_NAME_GIVEN = 237; - extern const int FORMAT_VERSION_TOO_OLD = 238; - extern const int CANNOT_MUNMAP = 239; - extern const int CANNOT_MREMAP = 240; - extern const int MEMORY_LIMIT_EXCEEDED = 241; - extern const int TABLE_IS_READ_ONLY = 242; - extern const int NOT_ENOUGH_SPACE = 243; - extern const int UNEXPECTED_ZOOKEEPER_ERROR = 244; - extern const int CORRUPTED_DATA = 246; - extern const int INCORRECT_MARK = 247; - extern const int INVALID_PARTITION_VALUE = 248; - extern const int NOT_ENOUGH_BLOCK_NUMBERS = 250; - extern const int NO_SUCH_REPLICA = 251; - extern const int TOO_MANY_PARTS = 252; - extern const int REPLICA_IS_ALREADY_EXIST = 253; - extern const int NO_ACTIVE_REPLICAS = 254; - extern const int TOO_MANY_RETRIES_TO_FETCH_PARTS = 255; - extern const int PARTITION_ALREADY_EXISTS = 256; - extern const int PARTITION_DOESNT_EXIST = 257; - extern const int UNION_ALL_RESULT_STRUCTURES_MISMATCH = 258; - extern const int CLIENT_OUTPUT_FORMAT_SPECIFIED = 260; - extern const int UNKNOWN_BLOCK_INFO_FIELD = 261; - extern const int BAD_COLLATION = 262; - extern const int CANNOT_COMPILE_CODE = 263; - extern const int INCOMPATIBLE_TYPE_OF_JOIN = 264; - extern const int NO_AVAILABLE_REPLICA = 265; - extern const int MISMATCH_REPLICAS_DATA_SOURCES = 266; - extern const int STORAGE_DOESNT_SUPPORT_PARALLEL_REPLICAS = 267; - extern const int CPUID_ERROR = 268; - extern const int INFINITE_LOOP = 269; - extern const int CANNOT_COMPRESS = 270; - extern const int CANNOT_DECOMPRESS = 271; - extern const int CANNOT_IO_SUBMIT = 272; - extern const int CANNOT_IO_GETEVENTS = 273; - extern const int AIO_READ_ERROR = 274; - extern const int AIO_WRITE_ERROR = 275; - extern const int INDEX_NOT_USED = 277; - extern const int ALL_CONNECTION_TRIES_FAILED = 279; - extern const int NO_AVAILABLE_DATA = 280; - extern const int DICTIONARY_IS_EMPTY = 281; - extern const int INCORRECT_INDEX = 282; - extern const int UNKNOWN_DISTRIBUTED_PRODUCT_MODE = 283; - extern const int WRONG_GLOBAL_SUBQUERY = 284; - extern const int TOO_FEW_LIVE_REPLICAS = 285; - extern const int UNSATISFIED_QUORUM_FOR_PREVIOUS_WRITE = 286; - extern const int UNKNOWN_FORMAT_VERSION = 287; - extern const int DISTRIBUTED_IN_JOIN_SUBQUERY_DENIED = 288; - extern const int REPLICA_IS_NOT_IN_QUORUM = 289; - extern const int LIMIT_EXCEEDED = 290; - extern const int DATABASE_ACCESS_DENIED = 291; - extern const int MONGODB_CANNOT_AUTHENTICATE = 293; - extern const int INVALID_BLOCK_EXTRA_INFO = 294; - extern const int RECEIVED_EMPTY_DATA = 295; - extern const int NO_REMOTE_SHARD_FOUND = 296; - extern const int SHARD_HAS_NO_CONNECTIONS = 297; - extern const int CANNOT_PIPE = 298; - extern const int CANNOT_FORK = 299; - extern const int CANNOT_DLSYM = 300; - extern const int CANNOT_CREATE_CHILD_PROCESS = 301; - extern const int CHILD_WAS_NOT_EXITED_NORMALLY = 302; - extern const int CANNOT_SELECT = 303; - extern const int CANNOT_WAITPID = 304; - extern const int TABLE_WAS_NOT_DROPPED = 305; - extern const int TOO_DEEP_RECURSION = 306; - extern const int TOO_MANY_BYTES = 307; - extern const int UNEXPECTED_NODE_IN_ZOOKEEPER = 308; - extern const int FUNCTION_CANNOT_HAVE_PARAMETERS = 309; - extern const int INVALID_SHARD_WEIGHT = 317; - extern const int INVALID_CONFIG_PARAMETER = 318; - extern const int UNKNOWN_STATUS_OF_INSERT = 319; - extern const int VALUE_IS_OUT_OF_RANGE_OF_DATA_TYPE = 321; - extern const int BARRIER_TIMEOUT = 335; - extern const int UNKNOWN_DATABASE_ENGINE = 336; - extern const int DDL_GUARD_IS_ACTIVE = 337; - extern const int UNFINISHED = 341; - extern const int METADATA_MISMATCH = 342; - extern const int SUPPORT_IS_DISABLED = 344; - extern const int TABLE_DIFFERS_TOO_MUCH = 345; - extern const int CANNOT_CONVERT_CHARSET = 346; - extern const int CANNOT_LOAD_CONFIG = 347; - extern const int CANNOT_INSERT_NULL_IN_ORDINARY_COLUMN = 349; - extern const int INCOMPATIBLE_SOURCE_TABLES = 350; - extern const int AMBIGUOUS_TABLE_NAME = 351; - extern const int AMBIGUOUS_COLUMN_NAME = 352; - extern const int INDEX_OF_POSITIONAL_ARGUMENT_IS_OUT_OF_RANGE = 353; - extern const int ZLIB_INFLATE_FAILED = 354; - extern const int ZLIB_DEFLATE_FAILED = 355; - extern const int BAD_LAMBDA = 356; - extern const int RESERVED_IDENTIFIER_NAME = 357; - extern const int INTO_OUTFILE_NOT_ALLOWED = 358; - extern const int TABLE_SIZE_EXCEEDS_MAX_DROP_SIZE_LIMIT = 359; - extern const int CANNOT_CREATE_CHARSET_CONVERTER = 360; - extern const int SEEK_POSITION_OUT_OF_BOUND = 361; - extern const int CURRENT_WRITE_BUFFER_IS_EXHAUSTED = 362; - extern const int CANNOT_CREATE_IO_BUFFER = 363; - extern const int RECEIVED_ERROR_TOO_MANY_REQUESTS = 364; - extern const int SIZES_OF_NESTED_COLUMNS_ARE_INCONSISTENT = 366; - extern const int TOO_MANY_FETCHES = 367; - extern const int ALL_REPLICAS_ARE_STALE = 369; - extern const int DATA_TYPE_CANNOT_BE_USED_IN_TABLES = 370; - extern const int INCONSISTENT_CLUSTER_DEFINITION = 371; - extern const int SESSION_NOT_FOUND = 372; - extern const int SESSION_IS_LOCKED = 373; - extern const int INVALID_SESSION_TIMEOUT = 374; - extern const int CANNOT_DLOPEN = 375; - extern const int CANNOT_PARSE_UUID = 376; - extern const int ILLEGAL_SYNTAX_FOR_DATA_TYPE = 377; - extern const int DATA_TYPE_CANNOT_HAVE_ARGUMENTS = 378; - extern const int UNKNOWN_STATUS_OF_DISTRIBUTED_DDL_TASK = 379; - extern const int CANNOT_KILL = 380; - extern const int HTTP_LENGTH_REQUIRED = 381; - extern const int CANNOT_LOAD_CATBOOST_MODEL = 382; - extern const int CANNOT_APPLY_CATBOOST_MODEL = 383; - extern const int PART_IS_TEMPORARILY_LOCKED = 384; - extern const int MULTIPLE_STREAMS_REQUIRED = 385; - extern const int NO_COMMON_TYPE = 386; - extern const int DICTIONARY_ALREADY_EXISTS = 387; - extern const int CANNOT_ASSIGN_OPTIMIZE = 388; - extern const int INSERT_WAS_DEDUPLICATED = 389; - extern const int CANNOT_GET_CREATE_TABLE_QUERY = 390; - extern const int EXTERNAL_LIBRARY_ERROR = 391; - extern const int QUERY_IS_PROHIBITED = 392; - extern const int THERE_IS_NO_QUERY = 393; - extern const int QUERY_WAS_CANCELLED = 394; - extern const int FUNCTION_THROW_IF_VALUE_IS_NON_ZERO = 395; - extern const int TOO_MANY_ROWS_OR_BYTES = 396; - extern const int QUERY_IS_NOT_SUPPORTED_IN_MATERIALIZED_VIEW = 397; - extern const int UNKNOWN_MUTATION_COMMAND = 398; - extern const int FORMAT_IS_NOT_SUITABLE_FOR_OUTPUT = 399; - extern const int CANNOT_STAT = 400; - extern const int FEATURE_IS_NOT_ENABLED_AT_BUILD_TIME = 401; - extern const int CANNOT_IOSETUP = 402; - extern const int INVALID_JOIN_ON_EXPRESSION = 403; - extern const int BAD_ODBC_CONNECTION_STRING = 404; - extern const int PARTITION_SIZE_EXCEEDS_MAX_DROP_SIZE_LIMIT = 405; - extern const int TOP_AND_LIMIT_TOGETHER = 406; - extern const int DECIMAL_OVERFLOW = 407; - extern const int BAD_REQUEST_PARAMETER = 408; - extern const int EXTERNAL_EXECUTABLE_NOT_FOUND = 409; - extern const int EXTERNAL_SERVER_IS_NOT_RESPONDING = 410; - extern const int PTHREAD_ERROR = 411; - extern const int NETLINK_ERROR = 412; - extern const int CANNOT_SET_SIGNAL_HANDLER = 413; - extern const int ALL_REPLICAS_LOST = 415; - extern const int REPLICA_STATUS_CHANGED = 416; - extern const int EXPECTED_ALL_OR_ANY = 417; - extern const int UNKNOWN_JOIN = 418; - extern const int MULTIPLE_ASSIGNMENTS_TO_COLUMN = 419; - extern const int CANNOT_UPDATE_COLUMN = 420; - extern const int CANNOT_ADD_DIFFERENT_AGGREGATE_STATES = 421; - extern const int UNSUPPORTED_URI_SCHEME = 422; - extern const int CANNOT_GETTIMEOFDAY = 423; - extern const int CANNOT_LINK = 424; - extern const int SYSTEM_ERROR = 425; - extern const int CANNOT_COMPILE_REGEXP = 427; - extern const int UNKNOWN_LOG_LEVEL = 428; - extern const int FAILED_TO_GETPWUID = 429; - extern const int MISMATCHING_USERS_FOR_PROCESS_AND_DATA = 430; - extern const int ILLEGAL_SYNTAX_FOR_CODEC_TYPE = 431; - extern const int UNKNOWN_CODEC = 432; - extern const int ILLEGAL_CODEC_PARAMETER = 433; - extern const int CANNOT_PARSE_PROTOBUF_SCHEMA = 434; - extern const int NO_DATA_FOR_REQUIRED_PROTOBUF_FIELD = 435; - extern const int PROTOBUF_BAD_CAST = 436; - extern const int PROTOBUF_FIELD_NOT_REPEATED = 437; - extern const int DATA_TYPE_CANNOT_BE_PROMOTED = 438; - extern const int CANNOT_SCHEDULE_TASK = 439; - extern const int INVALID_LIMIT_EXPRESSION = 440; - extern const int CANNOT_PARSE_DOMAIN_VALUE_FROM_STRING = 441; - extern const int BAD_DATABASE_FOR_TEMPORARY_TABLE = 442; - extern const int NO_COMMON_COLUMNS_WITH_PROTOBUF_SCHEMA = 443; - extern const int UNKNOWN_PROTOBUF_FORMAT = 444; - extern const int CANNOT_MPROTECT = 445; - extern const int FUNCTION_NOT_ALLOWED = 446; - extern const int HYPERSCAN_CANNOT_SCAN_TEXT = 447; - extern const int BROTLI_READ_FAILED = 448; - extern const int BROTLI_WRITE_FAILED = 449; - extern const int BAD_TTL_EXPRESSION = 450; - extern const int BAD_TTL_FILE = 451; - extern const int SETTING_CONSTRAINT_VIOLATION = 452; - extern const int MYSQL_CLIENT_INSUFFICIENT_CAPABILITIES = 453; - extern const int OPENSSL_ERROR = 454; - extern const int SUSPICIOUS_TYPE_FOR_LOW_CARDINALITY = 455; - extern const int UNKNOWN_QUERY_PARAMETER = 456; - extern const int BAD_QUERY_PARAMETER = 457; - extern const int CANNOT_UNLINK = 458; - extern const int CANNOT_SET_THREAD_PRIORITY = 459; - extern const int CANNOT_CREATE_TIMER = 460; - extern const int CANNOT_SET_TIMER_PERIOD = 461; - extern const int CANNOT_DELETE_TIMER = 462; - extern const int CANNOT_FCNTL = 463; - extern const int CANNOT_PARSE_ELF = 464; - extern const int CANNOT_PARSE_DWARF = 465; - extern const int INSECURE_PATH = 466; - extern const int CANNOT_PARSE_BOOL = 467; - extern const int CANNOT_PTHREAD_ATTR = 468; - extern const int VIOLATED_CONSTRAINT = 469; - extern const int QUERY_IS_NOT_SUPPORTED_IN_LIVE_VIEW = 470; - extern const int INVALID_SETTING_VALUE = 471; - extern const int READONLY_SETTING = 472; - extern const int DEADLOCK_AVOIDED = 473; - extern const int INVALID_TEMPLATE_FORMAT = 474; - extern const int INVALID_WITH_FILL_EXPRESSION = 475; - extern const int WITH_TIES_WITHOUT_ORDER_BY = 476; - extern const int INVALID_USAGE_OF_INPUT = 477; - extern const int UNKNOWN_POLICY = 478; - extern const int UNKNOWN_DISK = 479; - extern const int UNKNOWN_PROTOCOL = 480; - extern const int PATH_ACCESS_DENIED = 481; - extern const int DICTIONARY_ACCESS_DENIED = 482; - extern const int TOO_MANY_REDIRECTS = 483; - extern const int INTERNAL_REDIS_ERROR = 484; - extern const int SCALAR_ALREADY_EXISTS = 485; - extern const int CANNOT_GET_CREATE_DICTIONARY_QUERY = 487; - extern const int UNKNOWN_DICTIONARY = 488; - extern const int INCORRECT_DICTIONARY_DEFINITION = 489; - extern const int CANNOT_FORMAT_DATETIME = 490; - extern const int UNACCEPTABLE_URL = 491; - extern const int ACCESS_ENTITY_NOT_FOUND = 492; - extern const int ACCESS_ENTITY_ALREADY_EXISTS = 493; - extern const int ACCESS_ENTITY_FOUND_DUPLICATES = 494; - extern const int ACCESS_STORAGE_READONLY = 495; - extern const int QUOTA_REQUIRES_CLIENT_KEY = 496; - extern const int ACCESS_DENIED = 497; - extern const int LIMIT_BY_WITH_TIES_IS_NOT_SUPPORTED = 498; - extern const int S3_ERROR = 499; - extern const int CANNOT_CREATE_DATABASE = 501; - extern const int CANNOT_SIGQUEUE = 502; - extern const int AGGREGATE_FUNCTION_THROW = 503; - extern const int FILE_ALREADY_EXISTS = 504; - extern const int CANNOT_DELETE_DIRECTORY = 505; - extern const int UNEXPECTED_ERROR_CODE = 506; - extern const int UNABLE_TO_SKIP_UNUSED_SHARDS = 507; - extern const int UNKNOWN_ACCESS_TYPE = 508; - extern const int INVALID_GRANT = 509; - extern const int CACHE_DICTIONARY_UPDATE_FAIL = 510; - extern const int UNKNOWN_ROLE = 511; - extern const int SET_NON_GRANTED_ROLE = 512; - extern const int UNKNOWN_PART_TYPE = 513; - extern const int ACCESS_STORAGE_FOR_INSERTION_NOT_FOUND = 514; - extern const int INCORRECT_ACCESS_ENTITY_DEFINITION = 515; - extern const int AUTHENTICATION_FAILED = 516; - extern const int CANNOT_ASSIGN_ALTER = 517; - extern const int CANNOT_COMMIT_OFFSET = 518; - extern const int NO_REMOTE_SHARD_AVAILABLE = 519; - extern const int CANNOT_DETACH_DICTIONARY_AS_TABLE = 520; - extern const int ATOMIC_RENAME_FAIL = 521; - extern const int UNKNOWN_ROW_POLICY = 523; - extern const int ALTER_OF_COLUMN_IS_FORBIDDEN = 524; - extern const int INCORRECT_DISK_INDEX = 525; - extern const int UNKNOWN_VOLUME_TYPE = 526; - extern const int NO_SUITABLE_FUNCTION_IMPLEMENTATION = 527; - extern const int CASSANDRA_INTERNAL_ERROR = 528; - extern const int NOT_A_LEADER = 529; - extern const int CANNOT_CONNECT_RABBITMQ = 530; - extern const int CANNOT_FSTAT = 531; - extern const int LDAP_ERROR = 532; - extern const int INCONSISTENT_RESERVATIONS = 533; - extern const int NO_RESERVATIONS_PROVIDED = 534; - extern const int UNKNOWN_RAID_TYPE = 535; - extern const int CANNOT_RESTORE_FROM_FIELD_DUMP = 536; - extern const int ILLEGAL_MYSQL_VARIABLE = 537; - extern const int MYSQL_SYNTAX_ERROR = 538; - extern const int CANNOT_BIND_RABBITMQ_EXCHANGE = 539; - extern const int CANNOT_DECLARE_RABBITMQ_EXCHANGE = 540; - extern const int CANNOT_CREATE_RABBITMQ_QUEUE_BINDING = 541; - extern const int CANNOT_REMOVE_RABBITMQ_EXCHANGE = 542; - extern const int UNKNOWN_MYSQL_DATATYPES_SUPPORT_LEVEL = 543; - extern const int ROW_AND_ROWS_TOGETHER = 544; - extern const int FIRST_AND_NEXT_TOGETHER = 545; - extern const int NO_ROW_DELIMITER = 546; + constexpr Value END = 3000; + std::atomic values[END + 1] {}; - extern const int KEEPER_EXCEPTION = 999; - extern const int POCO_EXCEPTION = 1000; - extern const int STD_EXCEPTION = 1001; - extern const int UNKNOWN_EXCEPTION = 1002; + struct ErrorCodesNames + { + std::string_view names[END + 1]; + ErrorCodesNames() + { + #define M(VALUE, NAME) names[VALUE] = std::string_view(#NAME); + APPLY_FOR_ERROR_CODES(M) + #undef M + } + } error_codes_names; - extern const int CONDITIONAL_TREE_PARENT_NOT_FOUND = 2001; - extern const int ILLEGAL_PROJECTION_MANIPULATOR = 2002; - extern const int UNRECOGNIZED_ARGUMENTS = 2003; + std::string_view getName(ErrorCode error_code) + { + if (error_code >= END) + return std::string_view(); + return error_codes_names.names[error_code]; + } + + ErrorCode end() { return END+1; } } } diff --git a/src/Common/ErrorCodes.h b/src/Common/ErrorCodes.h new file mode 100644 index 00000000000..cc610c5d927 --- /dev/null +++ b/src/Common/ErrorCodes.h @@ -0,0 +1,46 @@ +#pragma once + +#include +#include +#include +#include +#include +#include + +/** Allows to count number of simultaneously happening error codes. + * See also Exception.cpp for incrementing part. + */ + +namespace DB +{ + +namespace ErrorCodes +{ + /// ErrorCode identifier (index in array). + using ErrorCode = size_t; + using Value = int; + + /// Get name of error_code by identifier. + /// Returns statically allocated string. + std::string_view getName(ErrorCode error_code); + + /// ErrorCode identifier -> current value of error_code. + extern std::atomic values[]; + + /// Get index just after last error_code identifier. + ErrorCode end(); + + /// Add value for specified error_code. + inline void increment(ErrorCode error_code) + { + if (error_code >= end()) + { + /// For everything outside the range, use END. + /// (end() is the pointer pass the end, while END is the last value that has an element in values array). + error_code = end() - 1; + } + values[error_code].fetch_add(1, std::memory_order_relaxed); + } +} + +} diff --git a/src/Common/Exception.cpp b/src/Common/Exception.cpp index 4dbd685d562..0214fa7b065 100644 --- a/src/Common/Exception.cpp +++ b/src/Common/Exception.cpp @@ -2,6 +2,7 @@ #include #include +#include #include #include #include @@ -13,6 +14,7 @@ #include #include #include +#include #include #if !defined(ARCADIA_BUILD) @@ -36,15 +38,16 @@ namespace ErrorCodes Exception::Exception(const std::string & msg, int code) : Poco::Exception(msg, code) { - // In debug builds, treat LOGICAL_ERROR as an assertion failure. + // In debug builds and builds with sanitizers, treat LOGICAL_ERROR as an assertion failure. // Log the message before we fail. -#ifndef NDEBUG +#ifdef ABORT_ON_LOGICAL_ERROR if (code == ErrorCodes::LOGICAL_ERROR) { - LOG_ERROR(&Poco::Logger::root(), "Logical error: '{}'.", msg); - assert(false); + LOG_FATAL(&Poco::Logger::root(), "Logical error: '{}'.", msg); + abort(); } #endif + ErrorCodes::increment(code); } Exception::Exception(CreateFromPocoTag, const Poco::Exception & exc) diff --git a/src/Common/Exception.h b/src/Common/Exception.h index 314c59cbf51..0096c87d6e5 100644 --- a/src/Common/Exception.h +++ b/src/Common/Exception.h @@ -10,6 +10,10 @@ #include +#if !defined(NDEBUG) || defined(ADDRESS_SANITIZER) || defined(THREAD_SANITIZER) || defined(MEMORY_SANITIZER) || defined(UNDEFINED_BEHAVIOR_SANITIZER) +#define ABORT_ON_LOGICAL_ERROR +#endif + namespace Poco { class Logger; } diff --git a/src/Common/FileSyncGuard.h b/src/Common/FileSyncGuard.h index 6451f6ebf36..486b02d0f24 100644 --- a/src/Common/FileSyncGuard.h +++ b/src/Common/FileSyncGuard.h @@ -5,15 +5,15 @@ namespace DB { -/// Helper class, that recieves file descriptor and does fsync for it in destructor. +/// Helper class, that receives file descriptor and does fsync for it in destructor. /// It's used to keep descriptor open, while doing some operations with it, and do fsync at the end. /// Guaranties of sequence 'close-reopen-fsync' may depend on kernel version. /// Source: linux-fsdevel mailing-list https://marc.info/?l=linux-fsdevel&m=152535409207496 class FileSyncGuard { public: - /// NOTE: If you have already opened descriptor, it's preffered to use - /// this constructor instead of construnctor with path. + /// NOTE: If you have already opened descriptor, it's preferred to use + /// this constructor instead of constructor with path. FileSyncGuard(const DiskPtr & disk_, int fd_) : disk(disk_), fd(fd_) {} FileSyncGuard(const DiskPtr & disk_, const String & path) diff --git a/src/Common/HashTable/HashTable.h b/src/Common/HashTable/HashTable.h index 897c84fe951..a569b1c15db 100644 --- a/src/Common/HashTable/HashTable.h +++ b/src/Common/HashTable/HashTable.h @@ -4,6 +4,7 @@ #include +#include #include #include @@ -314,8 +315,8 @@ public: zeroValue()->~Cell(); } - Cell * zeroValue() { return reinterpret_cast(&zero_value_storage); } - const Cell * zeroValue() const { return reinterpret_cast(&zero_value_storage); } + Cell * zeroValue() { return std::launder(reinterpret_cast(&zero_value_storage)); } + const Cell * zeroValue() const { return std::launder(reinterpret_cast(&zero_value_storage)); } }; template diff --git a/src/Common/HashTable/StringHashTable.h b/src/Common/HashTable/StringHashTable.h index e316f9b6520..06389825e60 100644 --- a/src/Common/HashTable/StringHashTable.h +++ b/src/Common/HashTable/StringHashTable.h @@ -3,8 +3,10 @@ #include #include +#include #include + using StringKey8 = UInt64; using StringKey16 = DB::UInt128; struct StringKey24 @@ -106,8 +108,8 @@ public: zeroValue()->~Cell(); } - Cell * zeroValue() { return reinterpret_cast(&zero_value_storage); } - const Cell * zeroValue() const { return reinterpret_cast(&zero_value_storage); } + Cell * zeroValue() { return std::launder(reinterpret_cast(&zero_value_storage)); } + const Cell * zeroValue() const { return std::launder(reinterpret_cast(&zero_value_storage)); } using LookupResult = Cell *; using ConstLookupResult = const Cell *; diff --git a/src/Common/HashTable/TwoLevelStringHashMap.h b/src/Common/HashTable/TwoLevelStringHashMap.h index 55d54e51b6a..6bd8f74dbd6 100644 --- a/src/Common/HashTable/TwoLevelStringHashMap.h +++ b/src/Common/HashTable/TwoLevelStringHashMap.h @@ -18,7 +18,7 @@ public: void ALWAYS_INLINE forEachMapped(Func && func) { for (auto i = 0u; i < this->NUM_BUCKETS; ++i) - return this->impls[i].forEachMapped(func); + this->impls[i].forEachMapped(func); } TMapped & ALWAYS_INLINE operator[](const Key & x) diff --git a/src/Common/MemoryTracker.cpp b/src/Common/MemoryTracker.cpp index 5d51fc9f301..380fcb1b2b6 100644 --- a/src/Common/MemoryTracker.cpp +++ b/src/Common/MemoryTracker.cpp @@ -13,6 +13,24 @@ #include #include +namespace +{ + +MemoryTracker * getMemoryTracker() +{ + if (auto * thread_memory_tracker = DB::CurrentThread::getMemoryTracker()) + return thread_memory_tracker; + + /// Once the main thread is initialized, + /// total_memory_tracker is initialized too. + /// And can be used, since MainThreadStatus is required for profiling. + if (DB::MainThreadStatus::get()) + return &total_memory_tracker; + + return nullptr; +} + +} namespace DB { @@ -30,6 +48,8 @@ namespace ProfileEvents static constexpr size_t log_peak_memory_usage_every = 1ULL << 30; +thread_local bool MemoryTracker::BlockerInThread::is_blocked = false; + MemoryTracker total_memory_tracker(nullptr, VariableContext::Global); @@ -56,13 +76,15 @@ MemoryTracker::~MemoryTracker() void MemoryTracker::logPeakMemoryUsage() const { const auto * description = description_ptr.load(std::memory_order_relaxed); - LOG_DEBUG(&Poco::Logger::get("MemoryTracker"), "Peak memory usage{}: {}.", (description ? " " + std::string(description) : ""), ReadableSize(peak)); + LOG_DEBUG(&Poco::Logger::get("MemoryTracker"), + "Peak memory usage{}: {}.", (description ? " " + std::string(description) : ""), ReadableSize(peak)); } void MemoryTracker::logMemoryUsage(Int64 current) const { const auto * description = description_ptr.load(std::memory_order_relaxed); - LOG_DEBUG(&Poco::Logger::get("MemoryTracker"), "Current memory usage{}: {}.", (description ? " " + std::string(description) : ""), ReadableSize(current)); + LOG_DEBUG(&Poco::Logger::get("MemoryTracker"), + "Current memory usage{}: {}.", (description ? " " + std::string(description) : ""), ReadableSize(current)); } @@ -71,7 +93,7 @@ void MemoryTracker::alloc(Int64 size) if (size < 0) throw DB::Exception(DB::ErrorCodes::LOGICAL_ERROR, "Negative size ({}) is passed to MemoryTracker. It is a bug.", size); - if (blocker.isCancelled()) + if (BlockerInThread::isBlocked()) return; /** Using memory_order_relaxed means that if allocations are done simultaneously, @@ -86,12 +108,15 @@ void MemoryTracker::alloc(Int64 size) Int64 current_hard_limit = hard_limit.load(std::memory_order_relaxed); Int64 current_profiler_limit = profiler_limit.load(std::memory_order_relaxed); - /// Cap the limit to the total_memory_tracker, since it may include some drift. + /// Cap the limit to the total_memory_tracker, since it may include some drift + /// for user-level memory tracker. /// /// And since total_memory_tracker is reset to the process resident /// memory peridically (in AsynchronousMetrics::update()), any limit can be /// capped to it, to avoid possible drift. - if (unlikely(current_hard_limit && will_be > current_hard_limit)) + if (unlikely(current_hard_limit + && will_be > current_hard_limit + && level == VariableContext::User)) { Int64 total_amount = total_memory_tracker.get(); if (amount > total_amount) @@ -104,10 +129,8 @@ void MemoryTracker::alloc(Int64 size) std::bernoulli_distribution fault(fault_probability); if (unlikely(fault_probability && fault(thread_local_rng))) { - free(size); - /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc - auto untrack_lock = blocker.cancel(); // NOLINT + BlockerInThread untrack_lock; ProfileEvents::increment(ProfileEvents::QueryMemoryLimitExceeded); std::stringstream message; @@ -118,12 +141,13 @@ void MemoryTracker::alloc(Int64 size) << " (attempt to allocate chunk of " << size << " bytes)" << ", maximum: " << formatReadableSizeWithBinarySuffix(current_hard_limit); + amount.fetch_sub(size, std::memory_order_relaxed); throw DB::Exception(message.str(), DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED); } if (unlikely(current_profiler_limit && will_be > current_profiler_limit)) { - auto no_track = blocker.cancel(); + BlockerInThread untrack_lock; DB::TraceCollector::collect(DB::TraceType::Memory, StackTrace(), size); setOrRaiseProfilerLimit((will_be + profiler_step - 1) / profiler_step * profiler_step); } @@ -131,16 +155,14 @@ void MemoryTracker::alloc(Int64 size) std::bernoulli_distribution sample(sample_probability); if (unlikely(sample_probability && sample(thread_local_rng))) { - auto no_track = blocker.cancel(); + BlockerInThread untrack_lock; DB::TraceCollector::collect(DB::TraceType::MemorySample, StackTrace(), size); } if (unlikely(current_hard_limit && will_be > current_hard_limit)) { - free(size); - /// Prevent recursion. Exception::ctor -> std::string -> new[] -> MemoryTracker::alloc - auto no_track = blocker.cancel(); // NOLINT + BlockerInThread untrack_lock; ProfileEvents::increment(ProfileEvents::QueryMemoryLimitExceeded); std::stringstream message; @@ -151,6 +173,7 @@ void MemoryTracker::alloc(Int64 size) << " (attempt to allocate chunk of " << size << " bytes)" << ", maximum: " << formatReadableSizeWithBinarySuffix(current_hard_limit); + amount.fetch_sub(size, std::memory_order_relaxed); throw DB::Exception(message.str(), DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED); } @@ -177,24 +200,25 @@ void MemoryTracker::updatePeak(Int64 will_be) void MemoryTracker::free(Int64 size) { - if (blocker.isCancelled()) + if (BlockerInThread::isBlocked()) return; std::bernoulli_distribution sample(sample_probability); if (unlikely(sample_probability && sample(thread_local_rng))) { - auto no_track = blocker.cancel(); + BlockerInThread untrack_lock; DB::TraceCollector::collect(DB::TraceType::MemorySample, StackTrace(), -size); } + Int64 accounted_size = size; if (level == VariableContext::Thread) { /// Could become negative if memory allocated in this thread is freed in another one - amount.fetch_sub(size, std::memory_order_relaxed); + amount.fetch_sub(accounted_size, std::memory_order_relaxed); } else { - Int64 new_amount = amount.fetch_sub(size, std::memory_order_relaxed) - size; + Int64 new_amount = amount.fetch_sub(accounted_size, std::memory_order_relaxed) - accounted_size; /** Sometimes, query could free some data, that was allocated outside of query context. * Example: cache eviction. @@ -205,7 +229,7 @@ void MemoryTracker::free(Int64 size) if (unlikely(new_amount < 0)) { amount.fetch_sub(new_amount); - size += new_amount; + accounted_size += new_amount; } } @@ -213,7 +237,7 @@ void MemoryTracker::free(Int64 size) loaded_next->free(size); if (metric != CurrentMetrics::end()) - CurrentMetrics::sub(metric, size); + CurrentMetrics::sub(metric, accounted_size); } @@ -265,16 +289,24 @@ namespace CurrentMemoryTracker void alloc(Int64 size) { - if (auto * memory_tracker = DB::CurrentThread::getMemoryTracker()) + if (auto * memory_tracker = getMemoryTracker()) { - current_thread->untracked_memory += size; - if (current_thread->untracked_memory > current_thread->untracked_memory_limit) + if (current_thread) { - /// Zero untracked before track. If tracker throws out-of-limit we would be able to alloc up to untracked_memory_limit bytes - /// more. It could be useful to enlarge Exception message in rethrow logic. - Int64 tmp = current_thread->untracked_memory; - current_thread->untracked_memory = 0; - memory_tracker->alloc(tmp); + current_thread->untracked_memory += size; + if (current_thread->untracked_memory > current_thread->untracked_memory_limit) + { + /// Zero untracked before track. If tracker throws out-of-limit we would be able to alloc up to untracked_memory_limit bytes + /// more. It could be useful to enlarge Exception message in rethrow logic. + Int64 tmp = current_thread->untracked_memory; + current_thread->untracked_memory = 0; + memory_tracker->alloc(tmp); + } + } + /// total_memory_tracker only, ignore untracked_memory + else + { + memory_tracker->alloc(size); } } } @@ -287,22 +319,22 @@ namespace CurrentMemoryTracker void free(Int64 size) { - if (auto * memory_tracker = DB::CurrentThread::getMemoryTracker()) + if (auto * memory_tracker = getMemoryTracker()) { - current_thread->untracked_memory -= size; - if (current_thread->untracked_memory < -current_thread->untracked_memory_limit) + if (current_thread) { - memory_tracker->free(-current_thread->untracked_memory); - current_thread->untracked_memory = 0; + current_thread->untracked_memory -= size; + if (current_thread->untracked_memory < -current_thread->untracked_memory_limit) + { + memory_tracker->free(-current_thread->untracked_memory); + current_thread->untracked_memory = 0; + } + } + /// total_memory_tracker only, ignore untracked_memory + else + { + memory_tracker->free(size); } } } } - -DB::SimpleActionLock getCurrentMemoryTrackerActionLock() -{ - auto * memory_tracker = DB::CurrentThread::getMemoryTracker(); - if (!memory_tracker) - return {}; - return memory_tracker->blocker.cancel(); -} diff --git a/src/Common/MemoryTracker.h b/src/Common/MemoryTracker.h index 8af683ae790..9f4f4357024 100644 --- a/src/Common/MemoryTracker.h +++ b/src/Common/MemoryTracker.h @@ -3,7 +3,6 @@ #include #include #include -#include #include @@ -131,8 +130,18 @@ public: /// Prints info about peak memory consumption into log. void logPeakMemoryUsage() const; - /// To be able to temporarily stop memory tracker - DB::SimpleActionBlocker blocker; + /// To be able to temporarily stop memory tracking from current thread. + struct BlockerInThread + { + private: + BlockerInThread(const BlockerInThread &) = delete; + BlockerInThread & operator=(const BlockerInThread &) = delete; + static thread_local bool is_blocked; + public: + BlockerInThread() { is_blocked = true; } + ~BlockerInThread() { is_blocked = false; } + static bool isBlocked() { return is_blocked; } + }; }; extern MemoryTracker total_memory_tracker; @@ -145,7 +154,3 @@ namespace CurrentMemoryTracker void realloc(Int64 old_size, Int64 new_size); void free(Int64 size); } - - -/// Holding this object will temporarily disable memory tracking. -DB::SimpleActionLock getCurrentMemoryTrackerActionLock(); diff --git a/src/Common/ThreadPool.cpp b/src/Common/ThreadPool.cpp index 21116e9d432..e527e97d608 100644 --- a/src/Common/ThreadPool.cpp +++ b/src/Common/ThreadPool.cpp @@ -234,13 +234,13 @@ void ThreadPoolImpl::worker(typename std::list::iterator thread_ std::is_same_v ? CurrentMetrics::GlobalThreadActive : CurrentMetrics::LocalThreadActive); job(); - /// job should be reseted before decrementing scheduled_jobs to + /// job should be reset before decrementing scheduled_jobs to /// ensure that the Job destroyed before wait() returns. job = {}; } catch (...) { - /// job should be reseted before decrementing scheduled_jobs to + /// job should be reset before decrementing scheduled_jobs to /// ensure that the Job destroyed before wait() returns. job = {}; diff --git a/src/Common/ThreadProfileEvents.h b/src/Common/ThreadProfileEvents.h index 9ebb5fbc2c6..d96339add26 100644 --- a/src/Common/ThreadProfileEvents.h +++ b/src/Common/ThreadProfileEvents.h @@ -75,13 +75,6 @@ inline TUInt safeDiff(TUInt prev, TUInt curr) } -inline UInt64 getCurrentTimeNanoseconds(clockid_t clock_type = CLOCK_MONOTONIC) -{ - struct timespec ts; - clock_gettime(clock_type, &ts); - return ts.tv_sec * 1000000000ULL + ts.tv_nsec; -} - struct RUsageCounters { /// In nanoseconds @@ -108,13 +101,13 @@ struct RUsageCounters hard_page_faults = static_cast(rusage.ru_majflt); } - static RUsageCounters current(UInt64 real_time_ = getCurrentTimeNanoseconds()) + static RUsageCounters current() { ::rusage rusage {}; #if !defined(__APPLE__) ::getrusage(RUSAGE_THREAD, &rusage); #endif - return RUsageCounters(rusage, real_time_); + return RUsageCounters(rusage, getClockMonotonic()); } static void incrementProfileEvents(const RUsageCounters & prev, const RUsageCounters & curr, ProfileEvents::Counters & profile_events) @@ -133,6 +126,14 @@ struct RUsageCounters incrementProfileEvents(last_counters, current_counters, profile_events); last_counters = current_counters; } + +private: + static inline UInt64 getClockMonotonic() + { + struct timespec ts; + clock_gettime(CLOCK_MONOTONIC, &ts); + return ts.tv_sec * 1000000000ULL + ts.tv_nsec; + } }; // thread_local is disabled in Arcadia, so we have to use a dummy implementation diff --git a/src/Common/ThreadStatus.cpp b/src/Common/ThreadStatus.cpp index ea9e507850c..bac0559fc6b 100644 --- a/src/Common/ThreadStatus.cpp +++ b/src/Common/ThreadStatus.cpp @@ -20,6 +20,7 @@ namespace ErrorCodes thread_local ThreadStatus * current_thread = nullptr; +thread_local ThreadStatus * main_thread = nullptr; ThreadStatus::ThreadStatus() @@ -115,4 +116,20 @@ void ThreadStatus::onFatalError() fatal_error_callback(); } +ThreadStatus * MainThreadStatus::main_thread = nullptr; +MainThreadStatus & MainThreadStatus::getInstance() +{ + static MainThreadStatus thread_status; + return thread_status; +} +MainThreadStatus::MainThreadStatus() + : ThreadStatus() +{ + main_thread = current_thread; +} +MainThreadStatus::~MainThreadStatus() +{ + main_thread = nullptr; +} + } diff --git a/src/Common/ThreadStatus.h b/src/Common/ThreadStatus.h index fc3951c8ca4..1b4d20e9721 100644 --- a/src/Common/ThreadStatus.h +++ b/src/Common/ThreadStatus.h @@ -164,13 +164,15 @@ public: void detachQuery(bool exit_if_already_detached = false, bool thread_exits = false); protected: + void applyQuerySettings(); + void initPerformanceCounters(); void initQueryProfiler(); void finalizeQueryProfiler(); - void logToQueryThreadLog(QueryThreadLog & thread_log); + void logToQueryThreadLog(QueryThreadLog & thread_log, const String & current_database, std::chrono::time_point now); void assertState(const std::initializer_list & permitted_states, const char * description = nullptr) const; @@ -213,4 +215,22 @@ private: void setupState(const ThreadGroupStatusPtr & thread_group_); }; +/** + * Creates ThreadStatus for the main thread. + */ +class MainThreadStatus : public ThreadStatus +{ +public: + static MainThreadStatus & getInstance(); + static ThreadStatus * get() { return main_thread; } + static bool isMainThread() { return main_thread == current_thread; } + + ~MainThreadStatus(); + +private: + MainThreadStatus(); + + static ThreadStatus * main_thread; +}; + } diff --git a/src/Common/TraceCollector.cpp b/src/Common/TraceCollector.cpp index 104b747d431..d10d5981d57 100644 --- a/src/Common/TraceCollector.cpp +++ b/src/Common/TraceCollector.cpp @@ -66,10 +66,20 @@ void TraceCollector::collect(TraceType trace_type, const StackTrace & stack_trac char buffer[buf_size]; WriteBufferFromFileDescriptorDiscardOnFailure out(pipe.fds_rw[1], buf_size, buffer); - StringRef query_id = CurrentThread::getQueryId(); - query_id.size = std::min(query_id.size, QUERY_ID_MAX_LEN); + StringRef query_id; + UInt64 thread_id; - auto thread_id = CurrentThread::get().thread_id; + if (CurrentThread::isInitialized()) + { + query_id = CurrentThread::getQueryId(); + query_id.size = std::min(query_id.size, QUERY_ID_MAX_LEN); + + thread_id = CurrentThread::get().thread_id; + } + else + { + thread_id = MainThreadStatus::get()->thread_id; + } writeChar(false, out); /// true if requested to stop the collecting thread. writeStringBinary(query_id, out); @@ -142,7 +152,7 @@ void TraceCollector::run() if (trace_log) { // time and time_in_microseconds are both being constructed from the same timespec so that the - // times will be equal upto the precision of a second. + // times will be equal up to the precision of a second. struct timespec ts; clock_gettime(CLOCK_REALTIME, &ts); diff --git a/src/Common/ZooKeeper/ZooKeeperImpl.cpp b/src/Common/ZooKeeper/ZooKeeperImpl.cpp index abb8158781b..f5c57781eef 100644 --- a/src/Common/ZooKeeper/ZooKeeperImpl.cpp +++ b/src/Common/ZooKeeper/ZooKeeperImpl.cpp @@ -1288,13 +1288,13 @@ void ZooKeeper::receiveEvent() response->removeRootPath(root_path); } - /// Instead of setting the watch in sendEvent, set it in receiveEvent becuase need to check the response. + /// Instead of setting the watch in sendEvent, set it in receiveEvent because need to check the response. /// The watch shouldn't be set if the node does not exist and it will never exist like sequential ephemeral nodes. /// By using getData() instead of exists(), a watch won't be set if the node doesn't exist. if (request_info.watch) { bool add_watch = false; - /// 3 indicates the ZooKeeperExistsRequest. + /// 3 indicates the ZooKeeperExistsRequest. // For exists, we set the watch on both node exist and nonexist case. // For other case like getData, we only set the watch when node exists. if (request_info.request->getOpNum() == 3) diff --git a/src/Common/tests/average.cpp b/src/Common/tests/average.cpp index 5f3b13af8e8..efe78212f06 100644 --- a/src/Common/tests/average.cpp +++ b/src/Common/tests/average.cpp @@ -502,8 +502,8 @@ Float NO_INLINE really_unrolled(const PODArray & keys, const PODArray void add(Float value) @@ -522,13 +522,13 @@ Float NO_INLINE another_unrolled_x4(const PODArray & keys, const PODArray { State4 map[256]{}; - size_t size = keys.size() & ~size_t(3); - for (size_t i = 0; i < size; i+=4) + size_t size = keys.size() / 4 * 4; + for (size_t i = 0; i < size; i += 4) { map[keys[i]].add<0>(values[i]); - map[keys[i+1]].add<1>(values[i]); - map[keys[i+2]].add<2>(values[i]); - map[keys[i+3]].add<3>(values[i]); + map[keys[i + 1]].add<1>(values[i]); + map[keys[i + 2]].add<2>(values[i]); + map[keys[i + 3]].add<3>(values[i]); } /// tail diff --git a/src/Common/tests/gtest_rw_lock.cpp b/src/Common/tests/gtest_rw_lock.cpp index 5f40a6b57d3..c77e405e18b 100644 --- a/src/Common/tests/gtest_rw_lock.cpp +++ b/src/Common/tests/gtest_rw_lock.cpp @@ -131,7 +131,10 @@ TEST(Common, RWLockRecursive) auto lock2 = fifo_lock->getLock(RWLockImpl::Read, "q2"); +#ifndef ABORT_ON_LOGICAL_ERROR + /// It throws LOGICAL_ERROR EXPECT_ANY_THROW({fifo_lock->getLock(RWLockImpl::Write, "q2");}); +#endif } fifo_lock->getLock(RWLockImpl::Write, "q2"); diff --git a/src/Common/ya.make b/src/Common/ya.make index fb04ecaa141..b19a5183201 100644 --- a/src/Common/ya.make +++ b/src/Common/ya.make @@ -21,7 +21,6 @@ PEERDIR( INCLUDE(${ARCADIA_ROOT}/clickhouse/cmake/yandex/ya.make.versions.inc) -CFLAGS(-g0) SRCS( ActionLock.cpp diff --git a/src/Common/ya.make.in b/src/Common/ya.make.in index f8b7601e215..49c8baa5eec 100644 --- a/src/Common/ya.make.in +++ b/src/Common/ya.make.in @@ -20,7 +20,6 @@ PEERDIR( INCLUDE(${ARCADIA_ROOT}/clickhouse/cmake/yandex/ya.make.versions.inc) -CFLAGS(-g0) SRCS( diff --git a/src/Compression/CompressedReadBufferBase.cpp b/src/Compression/CompressedReadBufferBase.cpp index be2f697e1b3..7a6b605d015 100644 --- a/src/Compression/CompressedReadBufferBase.cpp +++ b/src/Compression/CompressedReadBufferBase.cpp @@ -185,9 +185,9 @@ void CompressedReadBufferBase::decompress(char * to, size_t size_decompressed, s } else { - throw Exception("Data compressed with different methods, given method byte " + throw Exception("Data compressed with different methods, given method byte 0x" + getHexUIntLowercase(method) - + ", previous method byte " + + ", previous method byte 0x" + getHexUIntLowercase(codec->getMethodByte()), ErrorCodes::CANNOT_DECOMPRESS); } diff --git a/src/Compression/CompressionFactory.cpp b/src/Compression/CompressionFactory.cpp index 91b4aa4b8de..46d7d7dfcc4 100644 --- a/src/Compression/CompressionFactory.cpp +++ b/src/Compression/CompressionFactory.cpp @@ -76,7 +76,7 @@ ASTPtr CompressionCodecFactory::validateCodecAndGetPreprocessedAST(const ASTPtr ASTPtr codec_arguments; if (const auto * family_name = inner_codec_ast->as()) { - codec_family_name = family_name->name; + codec_family_name = family_name->name(); codec_arguments = {}; } else if (const auto * ast_func = inner_codec_ast->as()) @@ -87,7 +87,7 @@ ASTPtr CompressionCodecFactory::validateCodecAndGetPreprocessedAST(const ASTPtr else throw Exception("Unexpected AST element for compression codec", ErrorCodes::UNEXPECTED_AST_STRUCTURE); - /// Default codec replaced with current default codec which may dependend on different + /// Default codec replaced with current default codec which may depend on different /// settings (and properties of data) in runtime. CompressionCodecPtr result_codec; if (codec_family_name == DEFAULT_CODEC_NAME) @@ -207,7 +207,7 @@ CompressionCodecPtr CompressionCodecFactory::get(const ASTPtr & ast, const IData ASTPtr codec_arguments; if (const auto * family_name = inner_codec_ast->as()) { - codec_family_name = family_name->name; + codec_family_name = family_name->name(); codec_arguments = {}; } else if (const auto * ast_func = inner_codec_ast->as()) diff --git a/src/Compression/ICompressionCodec.cpp b/src/Compression/ICompressionCodec.cpp index baf6e9b2b86..3746753df8b 100644 --- a/src/Compression/ICompressionCodec.cpp +++ b/src/Compression/ICompressionCodec.cpp @@ -26,7 +26,7 @@ void ICompressionCodec::setCodecDescription(const String & codec_name, const AST std::shared_ptr result = std::make_shared(); result->name = "CODEC"; - /// Special case for codec Multiple, which doens't have name. It's just list + /// Special case for codec Multiple, which doesn't have name. It's just list /// of other codecs. if (codec_name.empty()) { diff --git a/src/Compression/ya.make b/src/Compression/ya.make index a17e2029b8f..8ffcb6be547 100644 --- a/src/Compression/ya.make +++ b/src/Compression/ya.make @@ -12,7 +12,6 @@ PEERDIR( contrib/libs/zstd ) -CFLAGS(-g0) SRCS( CachedCompressedReadBuffer.cpp diff --git a/src/Compression/ya.make.in b/src/Compression/ya.make.in index 780ea72b3ec..3c46b036aa0 100644 --- a/src/Compression/ya.make.in +++ b/src/Compression/ya.make.in @@ -11,7 +11,6 @@ PEERDIR( contrib/libs/zstd ) -CFLAGS(-g0) SRCS( diff --git a/src/Core/DecimalComparison.h b/src/Core/DecimalComparison.h index 8c88afb5be9..674ed31683b 100644 --- a/src/Core/DecimalComparison.h +++ b/src/Core/DecimalComparison.h @@ -60,27 +60,17 @@ public: using ArrayA = typename ColVecA::Container; using ArrayB = typename ColVecB::Container; - DecimalComparison(ColumnsWithTypeAndName & data, size_t result, const ColumnWithTypeAndName & col_left, const ColumnWithTypeAndName & col_right) - { - if (!apply(data, result, col_left, col_right)) - throw Exception("Wrong decimal comparison with " + col_left.type->getName() + " and " + col_right.type->getName(), - ErrorCodes::LOGICAL_ERROR); - } - - static bool apply(ColumnsWithTypeAndName & data, size_t result [[maybe_unused]], - const ColumnWithTypeAndName & col_left, const ColumnWithTypeAndName & col_right) + static ColumnPtr apply(const ColumnWithTypeAndName & col_left, const ColumnWithTypeAndName & col_right) { if constexpr (_actual) { ColumnPtr c_res; Shift shift = getScales(col_left.type, col_right.type); - c_res = applyWithScale(col_left.column, col_right.column, shift); - if (c_res) - data[result].column = std::move(c_res); - return true; + return applyWithScale(col_left.column, col_right.column, shift); } - return false; + else + return nullptr; } static bool compare(A a, B b, UInt32 scale_a, UInt32 scale_b) diff --git a/src/Core/Defines.h b/src/Core/Defines.h index ba3d37242fa..4d7d8e08ac3 100644 --- a/src/Core/Defines.h +++ b/src/Core/Defines.h @@ -67,11 +67,14 @@ /// Minimum revision supporting SettingsBinaryFormat::STRINGS. #define DBMS_MIN_REVISION_WITH_SETTINGS_SERIALIZED_AS_STRINGS 54429 +/// Minimum revision supporting OpenTelemetry +#define DBMS_MIN_REVISION_WITH_OPENTELEMETRY 54442 + /// Mininum revision supporting interserver secret. #define DBMS_MIN_REVISION_WITH_INTERSERVER_SECRET 54441 /// Version of ClickHouse TCP protocol. Increment it manually when you change the protocol. -#define DBMS_TCP_PROTOCOL_VERSION 54441 +#define DBMS_TCP_PROTOCOL_VERSION 54442 /// The boundary on which the blocks for asynchronous file operations should be aligned. #define DEFAULT_AIO_FILE_BLOCK_SIZE 4096 diff --git a/src/Core/Field.h b/src/Core/Field.h index 8973d106c0b..66e4f0ac8db 100644 --- a/src/Core/Field.h +++ b/src/Core/Field.h @@ -767,9 +767,9 @@ T & Field::get() #ifndef NDEBUG // Disregard signedness when converting between int64 types. constexpr Field::Types::Which target = TypeToEnum>::value; - assert(target == which - || (isInt64FieldType(target) && isInt64FieldType(which)) - || target == Field::Types::Decimal64 /* DateTime64 fields */); + if (target != which + && (!isInt64FieldType(target) || !isInt64FieldType(which))) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Invalid Field get from type {} to type {}", Types::toString(which), Types::toString(target)); #endif ValueType * MAY_ALIAS ptr = reinterpret_cast(&storage); diff --git a/src/Core/MySQL/MySQLGtid.cpp b/src/Core/MySQL/MySQLGtid.cpp index df26eb7aa08..aac3e01369a 100644 --- a/src/Core/MySQL/MySQLGtid.cpp +++ b/src/Core/MySQL/MySQLGtid.cpp @@ -85,6 +85,9 @@ void GTIDSets::update(const GTID & other) ErrorCodes::LOGICAL_ERROR); } + /// Try to shirnk Sequence interval. + GTIDSet::tryShirnk(set, i, current); + /// Sequence, extend the interval. if (other.seq_no == current.end) { @@ -116,6 +119,16 @@ void GTIDSets::update(const GTID & other) sets.emplace_back(set); } +void GTIDSet::tryShirnk(GTIDSet & set, unsigned int i, GTIDSet::Interval & current) +{ + if (i != set.intervals.size() -1) + { + auto & next = set.intervals[i+1]; + if (current.end == next.start) + set.tryMerge(i); + } +} + String GTIDSets::toString() const { WriteBufferFromOwnString buffer; diff --git a/src/Core/MySQL/MySQLGtid.h b/src/Core/MySQL/MySQLGtid.h index d228e269872..27aabdafc11 100644 --- a/src/Core/MySQL/MySQLGtid.h +++ b/src/Core/MySQL/MySQLGtid.h @@ -26,6 +26,8 @@ public: std::vector intervals; void tryMerge(size_t i); + + static void tryShirnk(GTIDSet & set, unsigned int i, Interval & current); }; class GTIDSets diff --git a/src/Core/MySQL/MySQLReplication.cpp b/src/Core/MySQL/MySQLReplication.cpp index 1179c0eb46b..c09c4b3b034 100644 --- a/src/Core/MySQL/MySQLReplication.cpp +++ b/src/Core/MySQL/MySQLReplication.cpp @@ -705,7 +705,7 @@ namespace MySQLReplication break; } default: - throw ReplicationError("Position update with unsupport event", ErrorCodes::LOGICAL_ERROR); + throw ReplicationError("Position update with unsupported event", ErrorCodes::LOGICAL_ERROR); } } diff --git a/src/Core/Settings.h b/src/Core/Settings.h index ac743768ff8..4406053d889 100644 --- a/src/Core/Settings.h +++ b/src/Core/Settings.h @@ -111,6 +111,7 @@ class IColumn; M(UInt64, distributed_group_by_no_merge, 0, "If 1, Do not merge aggregation states from different servers for distributed query processing - in case it is for certain that there are different keys on different shards. If 2 - same as 1 but also apply ORDER BY and LIMIT stages", 0) \ M(Bool, optimize_distributed_group_by_sharding_key, false, "Optimize GROUP BY sharding_key queries (by avodiing costly aggregation on the initiator server).", 0) \ M(Bool, optimize_skip_unused_shards, false, "Assumes that data is distributed by sharding_key. Optimization to skip unused shards if SELECT query filters by sharding_key.", 0) \ + M(Bool, allow_nondeterministic_optimize_skip_unused_shards, false, "Allow non-deterministic functions (includes dictGet) in sharding_key for optimize_skip_unused_shards", 0) \ M(UInt64, force_optimize_skip_unused_shards, 0, "Throw an exception if unused shards cannot be skipped (1 - throw only if the table has the sharding key, 2 - always throw.", 0) \ M(UInt64, optimize_skip_unused_shards_nesting, 0, "Same as optimize_skip_unused_shards, but accept nesting level until which it will work.", 0) \ M(UInt64, force_optimize_skip_unused_shards_nesting, 0, "Same as force_optimize_skip_unused_shards, but accept nesting level until which it will work.", 0) \ @@ -125,6 +126,7 @@ class IColumn; M(UInt64, merge_tree_coarse_index_granularity, 8, "If the index segment can contain the required keys, divide it into as many parts and recursively check them.", 0) \ M(UInt64, merge_tree_max_rows_to_use_cache, (128 * 8192), "The maximum number of rows per request, to use the cache of uncompressed data. If the request is large, the cache is not used. (For large queries not to flush out the cache.)", 0) \ M(UInt64, merge_tree_max_bytes_to_use_cache, (192 * 10 * 1024 * 1024), "The maximum number of bytes per request, to use the cache of uncompressed data. If the request is large, the cache is not used. (For large queries not to flush out the cache.)", 0) \ + M(Bool, do_not_merge_across_partitions_select_final, false, "Merge parts only in one partition in select final", 0) \ \ M(UInt64, mysql_max_rows_to_insert, 65536, "The maximum number of rows in MySQL batch insertion of the MySQL storage engine", 0) \ \ @@ -149,10 +151,12 @@ class IColumn; \ M(Bool, log_queries, 1, "Log requests and write the log to the system table.", 0) \ M(LogQueriesType, log_queries_min_type, QueryLogElementType::QUERY_START, "Minimal type in query_log to log, possible values (from low to high): QUERY_START, QUERY_FINISH, EXCEPTION_BEFORE_START, EXCEPTION_WHILE_PROCESSING.", 0) \ + M(Milliseconds, log_queries_min_query_duration_ms, 0, "Minimal time for the query to run, to get to the query_log/query_thread_log.", 0) \ M(UInt64, log_queries_cut_to_length, 100000, "If query length is greater than specified threshold (in bytes), then cut query when writing to query log. Also limit length of printed query in ordinary text log.", 0) \ \ M(DistributedProductMode, distributed_product_mode, DistributedProductMode::DENY, "How are distributed subqueries performed inside IN or JOIN sections?", IMPORTANT) \ \ + M(UInt64, max_concurrent_queries_for_all_users, 0, "The maximum number of concurrent requests for all users.", 0) \ M(UInt64, max_concurrent_queries_for_user, 0, "The maximum number of concurrent requests per user.", 0) \ \ M(Bool, insert_deduplicate, true, "For INSERT queries in the replicated table, specifies that deduplication of insertings blocks should be performed", 0) \ @@ -167,6 +171,8 @@ class IColumn; M(Milliseconds, read_backoff_min_interval_between_events_ms, 1000, "Settings to reduce the number of threads in case of slow reads. Do not pay attention to the event, if the previous one has passed less than a certain amount of time.", 0) \ M(UInt64, read_backoff_min_events, 2, "Settings to reduce the number of threads in case of slow reads. The number of events after which the number of threads will be reduced.", 0) \ \ + M(UInt64, read_backoff_min_concurrency, 1, "Settings to try keeping the minimal number of threads in case of slow reads.", 0) \ + \ M(Float, memory_tracker_fault_probability, 0., "For testing of `exception safety` - throw an exception every time you allocate memory with the specified probability.", 0) \ \ M(Bool, enable_http_compression, 0, "Compress the result if the client over HTTP said that it understands data compressed by gzip or deflate.", 0) \ @@ -220,6 +226,7 @@ class IColumn; M(UInt64, query_profiler_cpu_time_period_ns, 1000000000, "Period for CPU clock timer of query profiler (in nanoseconds). Set 0 value to turn off the CPU clock query profiler. Recommended value is at least 10000000 (100 times a second) for single queries or 1000000000 (once a second) for cluster-wide profiling.", 0) \ M(Bool, metrics_perf_events_enabled, false, "If enabled, some of the perf events will be measured throughout queries' execution.", 0) \ M(String, metrics_perf_events_list, "", "Comma separated list of perf metrics that will be measured throughout queries' execution. Empty means all events. See PerfEventInfo in sources for the available events.", 0) \ + M(Float, opentelemetry_start_trace_probability, 0., "Probability to start an OpenTelemetry trace for an incoming query.", 0) \ \ \ /** Limits during query execution are part of the settings. \ @@ -326,7 +333,6 @@ class IColumn; M(Bool, calculate_text_stack_trace, 1, "Calculate text stack trace in case of exceptions during query execution. This is the default. It requires symbol lookups that may slow down fuzzing tests when huge amount of wrong queries are executed. In normal cases you should not disable this option.", 0) \ M(Bool, allow_ddl, true, "If it is set to true, then a user is allowed to executed DDL queries.", 0) \ M(Bool, parallel_view_processing, false, "Enables pushing to attached views concurrently instead of sequentially.", 0) \ - M(Bool, enable_debug_queries, false, "Enables debug queries such as AST.", 0) \ M(Bool, enable_unaligned_array_join, false, "Allow ARRAY JOIN with multiple arrays that have different sizes. When this settings is enabled, arrays will be resized to the longest one.", 0) \ M(Bool, optimize_read_in_order, true, "Enable ORDER BY optimization for reading data in corresponding order in MergeTree tables.", 0) \ M(Bool, optimize_aggregation_in_order, false, "Enable GROUP BY optimization for aggregating data in corresponding order in MergeTree tables.", 0) \ @@ -368,7 +374,6 @@ class IColumn; M(Bool, optimize_monotonous_functions_in_order_by, true, "Replace monotonous function with its argument in ORDER BY", 0) \ M(Bool, allow_experimental_alter_materialized_view_structure, false, "Allow atomic alter on Materialized views. Work in progress.", 0) \ M(Bool, enable_early_constant_folding, true, "Enable query optimization where we analyze function and subqueries results and rewrite query if there're constants there", 0) \ - \ M(Bool, deduplicate_blocks_in_dependent_materialized_views, false, "Should deduplicate blocks for materialized views if the block is not a duplicate for the table. Use true to always deduplicate in dependent tables.", 0) \ M(Bool, use_compact_format_in_distributed_parts_names, false, "Changes format of directories names for distributed table insert parts.", 0) \ M(Bool, validate_polygons, true, "Throw exception if polygon is invalid in function pointInPolygon (e.g. self-tangent, self-intersecting). If the setting is false, the function will accept invalid polygons but may silently return wrong result.", 0) \ @@ -379,7 +384,6 @@ class IColumn; M(Seconds, lock_acquire_timeout, DBMS_DEFAULT_LOCK_ACQUIRE_TIMEOUT_SEC, "How long locking request should wait before failing", 0) \ M(Bool, materialize_ttl_after_modify, true, "Apply TTL for old data, after ALTER MODIFY TTL query", 0) \ M(String, function_implementation, "", "Choose function implementation for specific target or variant (experimental). If empty enable all of them.", 0) \ - \ M(Bool, allow_experimental_geo_types, false, "Allow geo data types such as Point, Ring, Polygon, MultiPolygon", 0) \ M(Bool, allow_experimental_bigint_types, false, "Allow Int128, Int256, UInt256 and Decimal256 types", 0) \ M(Bool, data_type_default_nullable, false, "Data types without NULL or NOT NULL will make Nullable", 0) \ @@ -387,21 +391,19 @@ class IColumn; M(Bool, alter_partition_verbose_result, false, "Output information about affected parts. Currently works only for FREEZE and ATTACH commands.", 0) \ M(Bool, allow_experimental_database_materialize_mysql, false, "Allow to create database with Engine=MaterializeMySQL(...).", 0) \ M(Bool, system_events_show_zero_values, false, "Include all metrics, even with zero values", 0) \ - M(MySQLDataTypesSupport, mysql_datatypes_support_level, 0, "Which MySQL types should be converted to corresponding ClickHouse types (rather than being represented as String). Can be empty or any combination of 'decimal' or 'datetime64'. When empty MySQL's DECIMAL and DATETIME/TIMESTAMP with non-zero precison are seen as String on ClickHouse's side.", 0) \ + M(MySQLDataTypesSupport, mysql_datatypes_support_level, 0, "Which MySQL types should be converted to corresponding ClickHouse types (rather than being represented as String). Can be empty or any combination of 'decimal' or 'datetime64'. When empty MySQL's DECIMAL and DATETIME/TIMESTAMP with non-zero precision are seen as String on ClickHouse's side.", 0) \ + M(Bool, optimize_trivial_insert_select, true, "Optimize trivial 'INSERT INTO table SELECT ... FROM TABLES' query", 0) \ + M(Bool, allow_non_metadata_alters, true, "Allow to execute alters which affects not only tables metadata, but also data on disk", 0) \ + M(Bool, enable_global_with_statement, false, "Propagate WITH statements to UNION queries and all subqueries", 0) \ + M(Bool, aggregate_functions_null_for_empty, false, "Rewrite all aggregate functions in a query, adding -OrNull suffix to them", 0) \ \ /** Obsolete settings that do nothing but left for compatibility reasons. Remove each one after half a year of obsolescence. */ \ \ - M(Bool, partial_merge_join, false, "Obsolete. Use join_algorithm='prefer_partial_merge' instead.", 0) \ - M(UInt64, max_memory_usage_for_all_queries, 0, "Obsolete. Will be removed after 2020-10-20", 0) \ M(UInt64, multiple_joins_rewriter_version, 0, "Obsolete setting, does nothing. Will be removed after 2021-03-31", 0) \ - \ - M(Bool, force_optimize_skip_unused_shards_no_nested, false, "Obsolete setting, does nothing. Will be removed after 2020-12-01. Use force_optimize_skip_unused_shards_nesting instead.", 0) \ M(Bool, experimental_use_processors, true, "Obsolete setting, does nothing. Will be removed after 2020-11-29.", 0) \ - M(Bool, optimize_trivial_insert_select, true, "Optimize trivial 'INSERT INTO table SELECT ... FROM TABLES' query", 0) \ + M(Bool, force_optimize_skip_unused_shards_no_nested, false, "Obsolete setting, does nothing. Will be removed after 2020-12-01. Use force_optimize_skip_unused_shards_nesting instead.", 0) \ + M(Bool, enable_debug_queries, false, "Enabled debug queries, but now is obsolete", 0) \ M(Bool, allow_experimental_database_atomic, true, "Obsolete setting, does nothing. Will be removed after 2021-02-12", 0) \ - M(Bool, allow_non_metadata_alters, true, "Allow to execute alters which affects not only tables metadata, but also data on disk", 0) \ - M(Bool, enable_global_with_statement, false, "Propagate WITH statements to UNION queries and all subqueries", 0) \ - M(Bool, aggregate_functions_null_for_empty, false, "Rewrite all aggregate functions in a query, adding -OrNull suffix to them", 0) // End of COMMON_SETTINGS // Please add settings related to formats into the FORMAT_FACTORY_SETTINGS below. @@ -412,12 +414,14 @@ class IColumn; M(Bool, format_csv_allow_double_quotes, 1, "If it is set to true, allow strings in double quotes.", 0) \ M(Bool, output_format_csv_crlf_end_of_line, false, "If it is set true, end of line in CSV format will be \\r\\n instead of \\n.", 0) \ M(Bool, input_format_csv_unquoted_null_literal_as_null, false, "Consider unquoted NULL literal as \\N", 0) \ + M(Bool, input_format_csv_enum_as_number, false, "Treat inserted enum values in CSV formats as enum indices \\N", 0) \ M(Bool, input_format_skip_unknown_fields, false, "Skip columns with unknown names from input data (it works for JSONEachRow, CSVWithNames, TSVWithNames and TSKV formats).", 0) \ M(Bool, input_format_with_names_use_header, true, "For TSVWithNames and CSVWithNames input formats this controls whether format parser is to assume that column data appear in the input exactly as they are specified in the header.", 0) \ M(Bool, input_format_import_nested_json, false, "Map nested JSON data to nested tables (it works for JSONEachRow format).", 0) \ M(Bool, optimize_aggregators_of_group_by_keys, true, "Eliminates min/max/any/anyLast aggregators of GROUP BY keys in SELECT section", 0) \ M(Bool, input_format_defaults_for_omitted_fields, true, "For input data calculate default expressions for omitted fields (it works for JSONEachRow, CSV and TSV formats).", IMPORTANT) \ M(Bool, input_format_tsv_empty_as_default, false, "Treat empty fields in TSV input as default values.", 0) \ + M(Bool, input_format_tsv_enum_as_number, false, "Treat inserted enum values in TSV formats as enum indices \\N", 0) \ M(Bool, input_format_null_as_default, false, "For text input formats initialize null fields with default values if data type of this field is not nullable", 0) \ \ M(DateTimeInputFormat, date_time_input_format, FormatSettings::DateTimeInputFormat::Basic, "Method to read DateTime from text input formats. Possible values: 'basic' and 'best_effort'.", 0) \ diff --git a/src/Core/tests/CMakeLists.txt b/src/Core/tests/CMakeLists.txt index d609e49f247..cd6450633ff 100644 --- a/src/Core/tests/CMakeLists.txt +++ b/src/Core/tests/CMakeLists.txt @@ -5,9 +5,6 @@ target_include_directories (string_pool SYSTEM BEFORE PRIVATE ${SPARSEHASH_INCLU add_executable (field field.cpp) target_link_libraries (field PRIVATE dbms) -add_executable (move_field move_field.cpp) -target_link_libraries (move_field PRIVATE clickhouse_common_io) - add_executable (string_ref_hash string_ref_hash.cpp) target_link_libraries (string_ref_hash PRIVATE clickhouse_common_io) diff --git a/src/Core/tests/gtest_move_field.cpp b/src/Core/tests/gtest_move_field.cpp new file mode 100644 index 00000000000..9c807039c6a --- /dev/null +++ b/src/Core/tests/gtest_move_field.cpp @@ -0,0 +1,22 @@ +#include +#include + +using namespace DB; + +GTEST_TEST(Field, Move) +{ + Field f; + + f = Field{String{"Hello, world (1)"}}; + ASSERT_EQ(f.get(), "Hello, world (1)"); + f = Field{String{"Hello, world (2)"}}; + ASSERT_EQ(f.get(), "Hello, world (2)"); + f = Field{Array{Field{String{"Hello, world (3)"}}}}; + ASSERT_EQ(f.get()[0].get(), "Hello, world (3)"); + f = String{"Hello, world (4)"}; + ASSERT_EQ(f.get(), "Hello, world (4)"); + f = Array{Field{String{"Hello, world (5)"}}}; + ASSERT_EQ(f.get()[0].get(), "Hello, world (5)"); + f = Array{String{"Hello, world (6)"}}; + ASSERT_EQ(f.get()[0].get(), "Hello, world (6)"); +} diff --git a/src/Core/tests/move_field.cpp b/src/Core/tests/move_field.cpp deleted file mode 100644 index 2780abffc40..00000000000 --- a/src/Core/tests/move_field.cpp +++ /dev/null @@ -1,25 +0,0 @@ -#include -#include - - -int main(int, char **) -{ - using namespace DB; - - Field f; - - f = Field{String{"Hello, world"}}; - std::cerr << f.get() << "\n"; - f = Field{String{"Hello, world!"}}; - std::cerr << f.get() << "\n"; - f = Field{Array{Field{String{"Hello, world!!"}}}}; - std::cerr << f.get()[0].get() << "\n"; - f = String{"Hello, world!!!"}; - std::cerr << f.get() << "\n"; - f = Array{Field{String{"Hello, world!!!!"}}}; - std::cerr << f.get()[0].get() << "\n"; - f = Array{String{"Hello, world!!!!!"}}; - std::cerr << f.get()[0].get() << "\n"; - - return 0; -} diff --git a/src/Core/tests/mysql_protocol.cpp b/src/Core/tests/mysql_protocol.cpp index 6cad095fc85..7e6aae5da23 100644 --- a/src/Core/tests/mysql_protocol.cpp +++ b/src/Core/tests/mysql_protocol.cpp @@ -260,6 +260,17 @@ int main(int argc, char ** argv) "10662d71-9d91-11ea-bbc2-0242ac110003:6-7", "20662d71-9d91-11ea-bbc2-0242ac110003:9", "10662d71-9d91-11ea-bbc2-0242ac110003:6-7,20662d71-9d91-11ea-bbc2-0242ac110003:9"}, + + {"shirnk-sequence", + "10662d71-9d91-11ea-bbc2-0242ac110003:1-3:4-5:7", + "10662d71-9d91-11ea-bbc2-0242ac110003:6", + "10662d71-9d91-11ea-bbc2-0242ac110003:1-7"}, + + {"shirnk-sequence", + "10662d71-9d91-11ea-bbc2-0242ac110003:1-3:4-5:10", + "10662d71-9d91-11ea-bbc2-0242ac110003:8", + "10662d71-9d91-11ea-bbc2-0242ac110003:1-5:8:10" + } }; for (auto & tc : cases) diff --git a/src/Core/ya.make b/src/Core/ya.make index 424566d212a..626662e992c 100644 --- a/src/Core/ya.make +++ b/src/Core/ya.make @@ -7,7 +7,6 @@ PEERDIR( contrib/restricted/boost/libs ) -CFLAGS(-g0) SRCS( BackgroundSchedulePool.cpp diff --git a/src/Core/ya.make.in b/src/Core/ya.make.in index 95c4e32995d..b2e82663c1e 100644 --- a/src/Core/ya.make.in +++ b/src/Core/ya.make.in @@ -6,7 +6,6 @@ PEERDIR( contrib/restricted/boost/libs ) -CFLAGS(-g0) SRCS( diff --git a/src/DataStreams/SquashingTransform.cpp b/src/DataStreams/SquashingTransform.cpp index c57e2351230..1f6ca8a7306 100644 --- a/src/DataStreams/SquashingTransform.cpp +++ b/src/DataStreams/SquashingTransform.cpp @@ -27,7 +27,7 @@ Block SquashingTransform::add(const Block & input_block) /* * To minimize copying, accept two types of argument: const reference for output - * stream, and rvalue reference for input stream, and decide whether to copy + * stream, and rvalue reference for input stream, and decide whether to copy * inside this function. This allows us not to copy Block unless we absolutely * have to. */ diff --git a/src/DataStreams/ya.make b/src/DataStreams/ya.make index 0c46e42d456..adef8246f33 100644 --- a/src/DataStreams/ya.make +++ b/src/DataStreams/ya.make @@ -8,7 +8,6 @@ PEERDIR( NO_COMPILER_WARNINGS() -CFLAGS(-g0) SRCS( AddingDefaultBlockOutputStream.cpp diff --git a/src/DataStreams/ya.make.in b/src/DataStreams/ya.make.in index 268719112ac..7aa2fe4874e 100644 --- a/src/DataStreams/ya.make.in +++ b/src/DataStreams/ya.make.in @@ -7,7 +7,6 @@ PEERDIR( NO_COMPILER_WARNINGS() -CFLAGS(-g0) SRCS( diff --git a/src/DataTypes/DataTypeEnum.cpp b/src/DataTypes/DataTypeEnum.cpp index 9ad6a9cb690..ce61794facd 100644 --- a/src/DataTypes/DataTypeEnum.cpp +++ b/src/DataTypes/DataTypeEnum.cpp @@ -146,12 +146,17 @@ void DataTypeEnum::serializeTextEscaped(const IColumn & column, size_t row } template -void DataTypeEnum::deserializeTextEscaped(IColumn & column, ReadBuffer & istr, const FormatSettings &) const +void DataTypeEnum::deserializeTextEscaped(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const { - /// NOTE It would be nice to do without creating a temporary object - at least extract std::string out. - std::string field_name; - readEscapedString(field_name, istr); - assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + if (settings.tsv.input_format_enum_as_number) + assert_cast(column).getData().push_back(readValue(istr)); + else + { + /// NOTE It would be nice to do without creating a temporary object - at least extract std::string out. + std::string field_name; + readEscapedString(field_name, istr); + assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + } } template @@ -169,11 +174,16 @@ void DataTypeEnum::deserializeTextQuoted(IColumn & column, ReadBuffer & is } template -void DataTypeEnum::deserializeWholeText(IColumn & column, ReadBuffer & istr, const FormatSettings &) const +void DataTypeEnum::deserializeWholeText(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const { - std::string field_name; - readString(field_name, istr); - assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + if (settings.tsv.input_format_enum_as_number) + assert_cast(column).getData().push_back(readValue(istr)); + else + { + std::string field_name; + readString(field_name, istr); + assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + } } template @@ -191,9 +201,14 @@ void DataTypeEnum::serializeTextXML(const IColumn & column, size_t row_num template void DataTypeEnum::deserializeTextJSON(IColumn & column, ReadBuffer & istr, const FormatSettings &) const { - std::string field_name; - readJSONString(field_name, istr); - assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + if (!istr.eof() && *istr.position() != '"') + assert_cast(column).getData().push_back(readValue(istr)); + else + { + std::string field_name; + readJSONString(field_name, istr); + assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + } } template @@ -205,9 +220,14 @@ void DataTypeEnum::serializeTextCSV(const IColumn & column, size_t row_num template void DataTypeEnum::deserializeTextCSV(IColumn & column, ReadBuffer & istr, const FormatSettings & settings) const { - std::string field_name; - readCSVString(field_name, istr, settings.csv); - assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + if (settings.csv.input_format_enum_as_number) + assert_cast(column).getData().push_back(readValue(istr)); + else + { + std::string field_name; + readCSVString(field_name, istr, settings.csv); + assert_cast(column).getData().push_back(getValue(StringRef(field_name))); + } } template diff --git a/src/DataTypes/DataTypeEnum.h b/src/DataTypes/DataTypeEnum.h index 80b41692cdd..a66b36c6a8d 100644 --- a/src/DataTypes/DataTypeEnum.h +++ b/src/DataTypes/DataTypeEnum.h @@ -66,13 +66,18 @@ public: TypeIndex getTypeId() const override { return sizeof(FieldType) == 1 ? TypeIndex::Enum8 : TypeIndex::Enum16; } - const StringRef & getNameForValue(const FieldType & value) const + auto findByValue(const FieldType & value) const { const auto it = value_to_name_map.find(value); if (it == std::end(value_to_name_map)) throw Exception{"Unexpected value " + toString(value) + " for type " + getName(), ErrorCodes::BAD_ARGUMENTS}; - return it->second; + return it; + } + + const StringRef & getNameForValue(const FieldType & value) const + { + return findByValue(value)->second; } FieldType getValue(StringRef field_name) const @@ -84,6 +89,13 @@ public: return it->getMapped(); } + FieldType readValue(ReadBuffer & istr) const + { + FieldType x; + readText(x, istr); + return findByValue(x)->first; + } + Field castToName(const Field & value_or_name) const override; Field castToValue(const Field & value_or_name) const override; diff --git a/src/DataTypes/DataTypeFactory.cpp b/src/DataTypes/DataTypeFactory.cpp index 9386f4b39f1..5052a065163 100644 --- a/src/DataTypes/DataTypeFactory.cpp +++ b/src/DataTypes/DataTypeFactory.cpp @@ -43,7 +43,7 @@ DataTypePtr DataTypeFactory::get(const ASTPtr & ast) const if (const auto * ident = ast->as()) { - return get(ident->name, {}); + return get(ident->name(), {}); } if (const auto * lit = ast->as()) diff --git a/src/DataTypes/DataTypeLowCardinalityHelpers.cpp b/src/DataTypes/DataTypeLowCardinalityHelpers.cpp index 673253500c4..a68dc30d5c2 100644 --- a/src/DataTypes/DataTypeLowCardinalityHelpers.cpp +++ b/src/DataTypes/DataTypeLowCardinalityHelpers.cpp @@ -34,7 +34,7 @@ DataTypePtr recursiveRemoveLowCardinality(const DataTypePtr & type) element = recursiveRemoveLowCardinality(element); if (tuple_type->haveExplicitNames()) - return std::make_shared(elements, tuple_type->getElementNames()); + return std::make_shared(elements, tuple_type->getElementNames(), tuple_type->serializeNames()); else return std::make_shared(elements); } diff --git a/src/DataTypes/DataTypeString.cpp b/src/DataTypes/DataTypeString.cpp index 9d563ee836c..141f896cfc2 100644 --- a/src/DataTypes/DataTypeString.cpp +++ b/src/DataTypes/DataTypeString.cpp @@ -384,7 +384,7 @@ static DataTypePtr create(const ASTPtr & arguments) throw Exception("String data type family mustn't have more than one argument - size in characters", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); const auto * argument = arguments->children[0]->as(); - if (!argument || argument->value.getType() != Field::Types::UInt64 || argument->value.get() == 0) + if (!argument || argument->value.getType() != Field::Types::UInt64) throw Exception("String data type family may have only a number (positive integer) as its argument", ErrorCodes::UNEXPECTED_AST_STRUCTURE); } diff --git a/src/DataTypes/DataTypeTuple.cpp b/src/DataTypes/DataTypeTuple.cpp index b69c4c31ca4..453cb7f37a3 100644 --- a/src/DataTypes/DataTypeTuple.cpp +++ b/src/DataTypes/DataTypeTuple.cpp @@ -44,28 +44,39 @@ DataTypeTuple::DataTypeTuple(const DataTypes & elems_) names[i] = toString(i + 1); } +static std::optional checkTupleNames(const Strings & names) +{ + std::unordered_set names_set; + for (const auto & name : names) + { + if (name.empty()) + return Exception("Names of tuple elements cannot be empty", ErrorCodes::BAD_ARGUMENTS); -DataTypeTuple::DataTypeTuple(const DataTypes & elems_, const Strings & names_) - : elems(elems_), names(names_), have_explicit_names(true) + if (isNumericASCII(name[0])) + return Exception("Explicitly specified names of tuple elements cannot start with digit", ErrorCodes::BAD_ARGUMENTS); + + if (!names_set.insert(name).second) + return Exception("Names of tuple elements must be unique", ErrorCodes::DUPLICATE_COLUMN); + } + + return {}; +} + +DataTypeTuple::DataTypeTuple(const DataTypes & elems_, const Strings & names_, bool serialize_names_) + : elems(elems_), names(names_), have_explicit_names(true), serialize_names(serialize_names_) { size_t size = elems.size(); if (names.size() != size) throw Exception("Wrong number of names passed to constructor of DataTypeTuple", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); - std::unordered_set names_set; - for (size_t i = 0; i < size; ++i) - { - if (names[i].empty()) - throw Exception("Names of tuple elements cannot be empty", ErrorCodes::BAD_ARGUMENTS); - - if (isNumericASCII(names[i][0])) - throw Exception("Explicitly specified names of tuple elements cannot start with digit", ErrorCodes::BAD_ARGUMENTS); - - if (!names_set.insert(names[i]).second) - throw Exception("Names of tuple elements must be unique", ErrorCodes::DUPLICATE_COLUMN); - } + if (auto exception = checkTupleNames(names)) + throw std::move(*exception); } +bool DataTypeTuple::canBeCreatedWithNames(const Strings & names) +{ + return checkTupleNames(names) == std::nullopt; +} std::string DataTypeTuple::doGetName() const { @@ -78,7 +89,7 @@ std::string DataTypeTuple::doGetName() const if (i != 0) s << ", "; - if (have_explicit_names) + if (have_explicit_names && serialize_names) s << backQuoteIfNeed(names[i]) << ' '; s << elems[i]->getName(); diff --git a/src/DataTypes/DataTypeTuple.h b/src/DataTypes/DataTypeTuple.h index a8d16c28fa5..7e4e68651f1 100644 --- a/src/DataTypes/DataTypeTuple.h +++ b/src/DataTypes/DataTypeTuple.h @@ -22,11 +22,14 @@ private: DataTypes elems; Strings names; bool have_explicit_names; + bool serialize_names = true; public: static constexpr bool is_parametric = true; DataTypeTuple(const DataTypes & elems); - DataTypeTuple(const DataTypes & elems, const Strings & names); + DataTypeTuple(const DataTypes & elems, const Strings & names, bool serialize_names_ = true); + + static bool canBeCreatedWithNames(const Strings & names); TypeIndex getTypeId() const override { return TypeIndex::Tuple; } std::string doGetName() const override; @@ -101,6 +104,7 @@ public: size_t getPositionByName(const String & name) const; bool haveExplicitNames() const { return have_explicit_names; } + bool serializeNames() const { return serialize_names; } }; } diff --git a/src/DataTypes/NumberTraits.h b/src/DataTypes/NumberTraits.h index 603449150db..77bd2101f05 100644 --- a/src/DataTypes/NumberTraits.h +++ b/src/DataTypes/NumberTraits.h @@ -29,7 +29,7 @@ constexpr size_t min(size_t x, size_t y) } /// @note There's no auto scale to larger big integer, only for integral ones. -/// It's cause of (U)Int64 backward compatibilty and very big performance penalties. +/// It's cause of (U)Int64 backward compatibility and very big performance penalties. constexpr size_t nextSize(size_t size) { if (size < 8) diff --git a/src/DataTypes/ya.make b/src/DataTypes/ya.make index 20a63bb7727..97b600f70ba 100644 --- a/src/DataTypes/ya.make +++ b/src/DataTypes/ya.make @@ -6,7 +6,6 @@ PEERDIR( clickhouse/src/Formats ) -CFLAGS(-g0) SRCS( convertMySQLDataType.cpp diff --git a/src/DataTypes/ya.make.in b/src/DataTypes/ya.make.in index f1983be1032..05170178925 100644 --- a/src/DataTypes/ya.make.in +++ b/src/DataTypes/ya.make.in @@ -5,7 +5,6 @@ PEERDIR( clickhouse/src/Formats ) -CFLAGS(-g0) SRCS( diff --git a/src/Databases/DatabaseAtomic.cpp b/src/Databases/DatabaseAtomic.cpp index a9dbae8ec92..af8b751e787 100644 --- a/src/Databases/DatabaseAtomic.cpp +++ b/src/Databases/DatabaseAtomic.cpp @@ -114,9 +114,10 @@ void DatabaseAtomic::dropTable(const Context &, const String & table_name, bool DatabaseWithDictionaries::detachTableUnlocked(table_name, lock); /// Should never throw table_name_to_path.erase(table_name); } - tryRemoveSymlink(table_name); + if (table->storesDataOnDisk()) + tryRemoveSymlink(table_name); /// Remove the inner table (if any) to avoid deadlock - /// (due to attemp to execute DROP from the worker thread) + /// (due to attempt to execute DROP from the worker thread) if (auto * mv = dynamic_cast(table.get())) mv->dropInnerTable(no_delay); /// Notify DatabaseCatalog that table was dropped. It will remove table data in background. @@ -145,7 +146,7 @@ void DatabaseAtomic::renameTable(const Context & context, const String & table_n String old_metadata_path = getObjectMetadataPath(table_name); String new_metadata_path = to_database.getObjectMetadataPath(to_table_name); - auto detach = [](DatabaseAtomic & db, const String & table_name_) + auto detach = [](DatabaseAtomic & db, const String & table_name_, bool has_symlink) { auto it = db.table_name_to_path.find(table_name_); String table_data_path_saved; @@ -155,7 +156,7 @@ void DatabaseAtomic::renameTable(const Context & context, const String & table_n assert(!table_data_path_saved.empty() || db.dictionaries.find(table_name_) != db.dictionaries.end()); db.tables.erase(table_name_); db.table_name_to_path.erase(table_name_); - if (!table_data_path_saved.empty()) + if (has_symlink) db.tryRemoveSymlink(table_name_); return table_data_path_saved; }; @@ -166,7 +167,8 @@ void DatabaseAtomic::renameTable(const Context & context, const String & table_n if (table_data_path_.empty()) return; db.table_name_to_path.emplace(table_name_, table_data_path_); - db.tryCreateSymlink(table_name_, table_data_path_); + if (table_->storesDataOnDisk()) + db.tryCreateSymlink(table_name_, table_data_path_); }; auto assert_can_move_mat_view = [inside_database](const StoragePtr & table_) @@ -228,9 +230,9 @@ void DatabaseAtomic::renameTable(const Context & context, const String & table_n renameNoReplace(old_metadata_path, new_metadata_path); /// After metadata was successfully moved, the following methods should not throw (if them do, it's a logical error) - table_data_path = detach(*this, table_name); + table_data_path = detach(*this, table_name, table->storesDataOnDisk()); if (exchange) - other_table_data_path = detach(other_db, to_table_name); + other_table_data_path = detach(other_db, to_table_name, other_table->storesDataOnDisk()); auto old_table_id = table->getStorageID(); @@ -261,24 +263,33 @@ void DatabaseAtomic::commitCreateTable(const ASTCreateQuery & query, const Stora { DetachedTables not_in_use; auto table_data_path = getTableDataPath(query); + bool locked_uuid = false; try { std::unique_lock lock{mutex}; if (query.database != database_name) throw Exception(ErrorCodes::UNKNOWN_DATABASE, "Database was renamed to `{}`, cannot create table in `{}`", database_name, query.database); + /// Do some checks before renaming file from .tmp to .sql not_in_use = cleanupDetachedTables(); assertDetachedTableNotInUse(query.uuid); - renameNoReplace(table_metadata_tmp_path, table_metadata_path); + /// We will get en exception if some table with the same UUID exists (even if it's detached table or table from another database) + DatabaseCatalog::instance().addUUIDMapping(query.uuid); + locked_uuid = true; + /// It throws if `table_metadata_path` already exists (it's possible if table was detached) + renameNoReplace(table_metadata_tmp_path, table_metadata_path); /// Commit point (a sort of) attachTableUnlocked(query.table, table, lock); /// Should never throw table_name_to_path.emplace(query.table, table_data_path); } catch (...) { Poco::File(table_metadata_tmp_path).remove(); + if (locked_uuid) + DatabaseCatalog::instance().removeUUIDMappingFinally(query.uuid); throw; } - tryCreateSymlink(query.table, table_data_path); + if (table->storesDataOnDisk()) + tryCreateSymlink(query.table, table_data_path); } void DatabaseAtomic::commitAlterTable(const StorageID & table_id, const String & table_metadata_tmp_path, const String & table_metadata_path) @@ -375,17 +386,18 @@ void DatabaseAtomic::loadStoredObjects(Context & context, bool has_force_restore Poco::File(path_to_table_symlinks).createDirectories(); for (const auto & table : table_names) - tryCreateSymlink(table.first, table.second); + tryCreateSymlink(table.first, table.second, true); } } -void DatabaseAtomic::tryCreateSymlink(const String & table_name, const String & actual_data_path) +void DatabaseAtomic::tryCreateSymlink(const String & table_name, const String & actual_data_path, bool if_data_path_exist) { try { String link = path_to_table_symlinks + escapeForFileName(table_name); - String data = Poco::Path(global_context.getPath()).makeAbsolute().toString() + actual_data_path; - Poco::File{data}.linkTo(link, Poco::File::LINK_SYMBOLIC); + Poco::File data = Poco::Path(global_context.getPath()).makeAbsolute().toString() + actual_data_path; + if (!if_data_path_exist || data.exists()) + data.linkTo(link, Poco::File::LINK_SYMBOLIC); } catch (...) { diff --git a/src/Databases/DatabaseAtomic.h b/src/Databases/DatabaseAtomic.h index 2d90ed96f1d..82408ff3ab3 100644 --- a/src/Databases/DatabaseAtomic.h +++ b/src/Databases/DatabaseAtomic.h @@ -55,7 +55,7 @@ public: UUID tryGetTableUUID(const String & table_name) const override; - void tryCreateSymlink(const String & table_name, const String & actual_data_path); + void tryCreateSymlink(const String & table_name, const String & actual_data_path, bool if_data_path_exist = false); void tryRemoveSymlink(const String & table_name); void waitDetachedTableNotInUse(const UUID & uuid); diff --git a/src/Databases/DatabaseLazy.h b/src/Databases/DatabaseLazy.h index 13c14863efb..2d091297c91 100644 --- a/src/Databases/DatabaseLazy.h +++ b/src/Databases/DatabaseLazy.h @@ -22,6 +22,10 @@ public: String getEngineName() const override { return "Lazy"; } + bool canContainMergeTreeTables() const override { return false; } + + bool canContainDistributedTables() const override { return false; } + void loadStoredObjects( Context & context, bool has_force_restore_data_flag, bool force_attach) override; diff --git a/src/Databases/DatabaseMemory.cpp b/src/Databases/DatabaseMemory.cpp index 5eacb846d52..357acb32371 100644 --- a/src/Databases/DatabaseMemory.cpp +++ b/src/Databases/DatabaseMemory.cpp @@ -53,6 +53,9 @@ void DatabaseMemory::dropTable( } table->is_dropped = true; create_queries.erase(table_name); + UUID table_uuid = table->getStorageID().uuid; + if (table_uuid != UUIDHelpers::Nil) + DatabaseCatalog::instance().removeUUIDMappingFinally(table_uuid); } ASTPtr DatabaseMemory::getCreateDatabaseQuery() const diff --git a/src/Databases/DatabaseOnDisk.cpp b/src/Databases/DatabaseOnDisk.cpp index 8d9f222bf69..4d7fcd4e248 100644 --- a/src/Databases/DatabaseOnDisk.cpp +++ b/src/Databases/DatabaseOnDisk.cpp @@ -321,7 +321,7 @@ void DatabaseOnDisk::renameTable( /// Special case: usually no actions with symlinks are required when detaching/attaching table, /// but not when moving from Atomic database to Ordinary - if (from_atomic_to_ordinary) + if (from_atomic_to_ordinary && table->storesDataOnDisk()) { auto & atomic_db = assert_cast(*this); atomic_db.tryRemoveSymlink(table_name); diff --git a/src/Databases/DatabaseWithDictionaries.cpp b/src/Databases/DatabaseWithDictionaries.cpp index ed85028d04d..6c5173c986f 100644 --- a/src/Databases/DatabaseWithDictionaries.cpp +++ b/src/Databases/DatabaseWithDictionaries.cpp @@ -223,6 +223,10 @@ void DatabaseWithDictionaries::removeDictionary(const Context &, const String & attachDictionary(dictionary_name, attach_info); throw; } + + UUID dict_uuid = attach_info.create_query->as()->uuid; + if (dict_uuid != UUIDHelpers::Nil) + DatabaseCatalog::instance().removeUUIDMappingFinally(dict_uuid); } DatabaseDictionariesIteratorPtr DatabaseWithDictionaries::getDictionariesIterator(const FilterByNameFunction & filter_by_dictionary_name) diff --git a/src/Databases/IDatabase.h b/src/Databases/IDatabase.h index b28bd5fd599..fadec5fe7a9 100644 --- a/src/Databases/IDatabase.h +++ b/src/Databases/IDatabase.h @@ -147,6 +147,10 @@ public: /// Get name of database engine. virtual String getEngineName() const = 0; + virtual bool canContainMergeTreeTables() const { return true; } + + virtual bool canContainDistributedTables() const { return true; } + /// Load a set of existing tables. /// You can call only once, right after the object is created. virtual void loadStoredObjects(Context & /*context*/, bool /*has_force_restore_data_flag*/, bool /*force_attach*/ = false) {} diff --git a/src/Databases/MySQL/ConnectionMySQLSettings.h b/src/Databases/MySQL/ConnectionMySQLSettings.h index 90279f846a4..ce2773307c5 100644 --- a/src/Databases/MySQL/ConnectionMySQLSettings.h +++ b/src/Databases/MySQL/ConnectionMySQLSettings.h @@ -11,7 +11,7 @@ class Context; class ASTStorage; #define LIST_OF_CONNECTION_MYSQL_SETTINGS(M) \ - M(MySQLDataTypesSupport, mysql_datatypes_support_level, 0, "Which MySQL types should be converted to corresponding ClickHouse types (rather than being represented as String). Can be empty or any combination of 'decimal' or 'datetime64'. When empty MySQL's DECIMAL and DATETIME/TIMESTAMP with non-zero precison are seen as String on ClickHouse's side.", 0) \ + M(MySQLDataTypesSupport, mysql_datatypes_support_level, 0, "Which MySQL types should be converted to corresponding ClickHouse types (rather than being represented as String). Can be empty or any combination of 'decimal' or 'datetime64'. When empty MySQL's DECIMAL and DATETIME/TIMESTAMP with non-zero precision are seen as String on ClickHouse's side.", 0) \ /// Settings that should not change after the creation of a database. #define APPLY_FOR_IMMUTABLE_CONNECTION_MYSQL_SETTINGS(M) \ diff --git a/src/Databases/MySQL/DatabaseConnectionMySQL.h b/src/Databases/MySQL/DatabaseConnectionMySQL.h index 7bf5e8c1d88..d8694e71db2 100644 --- a/src/Databases/MySQL/DatabaseConnectionMySQL.h +++ b/src/Databases/MySQL/DatabaseConnectionMySQL.h @@ -42,6 +42,12 @@ public: String getEngineName() const override { return "MySQL"; } + bool canContainMergeTreeTables() const override { return false; } + + bool canContainDistributedTables() const override { return false; } + + bool shouldBeEmptyOnDetach() const override { return false; } + bool empty() const override; DatabaseTablesIteratorPtr getTablesIterator(const Context & context, const FilterByNameFunction & filter_by_table_name) override; diff --git a/src/Databases/MySQL/MaterializeMySQLSettings.cpp b/src/Databases/MySQL/MaterializeMySQLSettings.cpp index 609ce011f91..a8672bf488e 100644 --- a/src/Databases/MySQL/MaterializeMySQLSettings.cpp +++ b/src/Databases/MySQL/MaterializeMySQLSettings.cpp @@ -8,7 +8,6 @@ namespace DB namespace ErrorCodes { - extern const int BAD_ARGUMENTS; extern const int UNKNOWN_SETTING; } @@ -25,9 +24,8 @@ void MaterializeMySQLSettings::loadFromQuery(ASTStorage & storage_def) catch (Exception & e) { if (e.code() == ErrorCodes::UNKNOWN_SETTING) - throw Exception(e.message() + " for database " + storage_def.engine->name, ErrorCodes::BAD_ARGUMENTS); - else - e.rethrow(); + e.addMessage("for database " + storage_def.engine->name); + throw; } } else diff --git a/src/Databases/ya.make b/src/Databases/ya.make index b4173057e03..e3c5daeb6bc 100644 --- a/src/Databases/ya.make +++ b/src/Databases/ya.make @@ -5,7 +5,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( DatabaseAtomic.cpp diff --git a/src/Databases/ya.make.in b/src/Databases/ya.make.in index e48d0d1bda7..ce7cd88b272 100644 --- a/src/Databases/ya.make.in +++ b/src/Databases/ya.make.in @@ -4,7 +4,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( diff --git a/src/Dictionaries/FileDictionarySource.cpp b/src/Dictionaries/FileDictionarySource.cpp index 18893a99f4e..82aea4cbb98 100644 --- a/src/Dictionaries/FileDictionarySource.cpp +++ b/src/Dictionaries/FileDictionarySource.cpp @@ -32,7 +32,7 @@ FileDictionarySource::FileDictionarySource( { const String user_files_path = context.getUserFilesPath(); if (!startsWith(filepath, user_files_path)) - throw Exception("File path " + filepath + " is not inside " + user_files_path, ErrorCodes::PATH_ACCESS_DENIED); + throw Exception(ErrorCodes::PATH_ACCESS_DENIED, "File path {} is not inside {}", filepath, user_files_path); } } @@ -60,7 +60,7 @@ BlockInputStreamPtr FileDictionarySource::loadAll() std::string FileDictionarySource::toString() const { - return "File: " + filepath + ' ' + format; + return fmt::format("File: {}, {}", filepath, format); } diff --git a/src/Dictionaries/SSDComplexKeyCacheDictionary.cpp b/src/Dictionaries/SSDComplexKeyCacheDictionary.cpp index 44847df48ff..5ac821e5eda 100644 --- a/src/Dictionaries/SSDComplexKeyCacheDictionary.cpp +++ b/src/Dictionaries/SSDComplexKeyCacheDictionary.cpp @@ -1467,7 +1467,6 @@ void SSDComplexKeyCacheDictionary::getItemsNumberImpl( { assert(dict_struct.key); assert(key_columns.size() == key_types.size()); - assert(key_columns.size() == dict_struct.key->size()); dict_struct.validateKeyTypes(key_types); diff --git a/src/Dictionaries/getDictionaryConfigurationFromAST.cpp b/src/Dictionaries/getDictionaryConfigurationFromAST.cpp index b1962e48eea..430c1d591dd 100644 --- a/src/Dictionaries/getDictionaryConfigurationFromAST.cpp +++ b/src/Dictionaries/getDictionaryConfigurationFromAST.cpp @@ -172,7 +172,7 @@ Names getPrimaryKeyColumns(const ASTExpressionList * primary_key) for (size_t index = 0; index != children.size(); ++index) { const ASTIdentifier * key_part = children[index]->as(); - result.push_back(key_part->name); + result.push_back(key_part->name()); } return result; } @@ -367,7 +367,7 @@ void buildConfigurationFromFunctionWithKeyValueArguments( if (const auto * identifier = pair->second->as(); identifier) { - AutoPtr value(doc->createTextNode(identifier->name)); + AutoPtr value(doc->createTextNode(identifier->name())); current_xml_element->appendChild(value); } else if (const auto * literal = pair->second->as(); literal) diff --git a/src/Dictionaries/ya.make b/src/Dictionaries/ya.make index 485d8b0a16d..9edf156c015 100644 --- a/src/Dictionaries/ya.make +++ b/src/Dictionaries/ya.make @@ -12,7 +12,6 @@ PEERDIR( NO_COMPILER_WARNINGS() -CFLAGS(-g0) SRCS( CacheDictionary.cpp diff --git a/src/Dictionaries/ya.make.in b/src/Dictionaries/ya.make.in index 3eb8e728643..2c0735d38a4 100644 --- a/src/Dictionaries/ya.make.in +++ b/src/Dictionaries/ya.make.in @@ -11,7 +11,6 @@ PEERDIR( NO_COMPILER_WARNINGS() -CFLAGS(-g0) SRCS( diff --git a/src/Disks/DiskDecorator.cpp b/src/Disks/DiskDecorator.cpp index 7f2ea58d7cf..aaa54005f6f 100644 --- a/src/Disks/DiskDecorator.cpp +++ b/src/Disks/DiskDecorator.cpp @@ -180,4 +180,9 @@ void DiskDecorator::sync(int fd) const delegate->sync(fd); } +Executor & DiskDecorator::getExecutor() +{ + return delegate->getExecutor(); +} + } diff --git a/src/Disks/DiskDecorator.h b/src/Disks/DiskDecorator.h index f1ddfff4952..1ce3c3ea773 100644 --- a/src/Disks/DiskDecorator.h +++ b/src/Disks/DiskDecorator.h @@ -4,6 +4,10 @@ namespace DB { + +/** Forwards all methods to another disk. + * Methods can be overridden by descendants. + */ class DiskDecorator : public IDisk { public: @@ -46,6 +50,7 @@ public: void close(int fd) const override; void sync(int fd) const override; const String getType() const override { return delegate->getType(); } + Executor & getExecutor() override; protected: DiskPtr delegate; diff --git a/src/Disks/DiskSelector.h b/src/Disks/DiskSelector.h index 3f19dfba381..5d023fe1fbc 100644 --- a/src/Disks/DiskSelector.h +++ b/src/Disks/DiskSelector.h @@ -23,8 +23,11 @@ public: DiskSelector(const Poco::Util::AbstractConfiguration & config, const String & config_prefix, const Context & context); DiskSelector(const DiskSelector & from) : disks(from.disks) { } - DiskSelectorPtr - updateFromConfig(const Poco::Util::AbstractConfiguration & config, const String & config_prefix, const Context & context) const; + DiskSelectorPtr updateFromConfig( + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + const Context & context + ) const; /// Get disk by name DiskPtr get(const String & name) const; diff --git a/src/Disks/IDisk.h b/src/Disks/IDisk.h index 688c1dfad42..ac0f5a2ae8f 100644 --- a/src/Disks/IDisk.h +++ b/src/Disks/IDisk.h @@ -195,10 +195,10 @@ public: /// Invoked when Global Context is shutdown. virtual void shutdown() { } -private: /// Returns executor to perform asynchronous operations. - Executor & getExecutor() { return *executor; } + virtual Executor & getExecutor() { return *executor; } +private: std::unique_ptr executor; }; diff --git a/src/Disks/IVolume.cpp b/src/Disks/IVolume.cpp index 95f03826591..ac277d962ed 100644 --- a/src/Disks/IVolume.cpp +++ b/src/Disks/IVolume.cpp @@ -9,7 +9,7 @@ namespace DB { namespace ErrorCodes { - extern const int EXCESSIVE_ELEMENT_IN_CONFIG; + extern const int NO_ELEMENTS_IN_CONFIG; extern const int INCONSISTENT_RESERVATIONS; extern const int NO_RESERVATIONS_PROVIDED; extern const int UNKNOWN_VOLUME_TYPE; @@ -51,7 +51,7 @@ IVolume::IVolume( } if (disks.empty()) - throw Exception("Volume must contain at least one disk.", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + throw Exception("Volume must contain at least one disk", ErrorCodes::NO_ELEMENTS_IN_CONFIG); } UInt64 IVolume::getMaxUnreservedFreeSpace() const diff --git a/src/Disks/IVolume.h b/src/Disks/IVolume.h index eaf3bf1dbd4..c040d9d58e1 100644 --- a/src/Disks/IVolume.h +++ b/src/Disks/IVolume.h @@ -64,6 +64,12 @@ public: virtual DiskPtr getDisk(size_t i) const { return disks[i]; } const Disks & getDisks() const { return disks; } + /// Returns effective value of whether merges are allowed on this volume (true) or not (false). + virtual bool areMergesAvoided() const { return false; } + + /// User setting for enabling and disabling merges on volume. + virtual void setAvoidMergesUserOverride(bool /*avoid*/) {} + protected: Disks disks; const String name; diff --git a/src/Disks/S3/ya.make b/src/Disks/S3/ya.make index b32adee0d26..17425f6e69a 100644 --- a/src/Disks/S3/ya.make +++ b/src/Disks/S3/ya.make @@ -4,7 +4,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( DiskS3.cpp diff --git a/src/Disks/SingleDiskVolume.h b/src/Disks/SingleDiskVolume.h index c441d4c2dd2..bade6041ea0 100644 --- a/src/Disks/SingleDiskVolume.h +++ b/src/Disks/SingleDiskVolume.h @@ -8,7 +8,7 @@ namespace DB class SingleDiskVolume : public IVolume { public: - SingleDiskVolume(const String & name_, DiskPtr disk): IVolume(name_, {disk}) + SingleDiskVolume(const String & name_, DiskPtr disk, size_t max_data_part_size_ = 0): IVolume(name_, {disk}, max_data_part_size_) { } diff --git a/src/Disks/StoragePolicy.cpp b/src/Disks/StoragePolicy.cpp index 1aa20301bc0..2215615feda 100644 --- a/src/Disks/StoragePolicy.cpp +++ b/src/Disks/StoragePolicy.cpp @@ -11,6 +11,13 @@ #include +namespace +{ + const auto DEFAULT_STORAGE_POLICY_NAME = "default"; + const auto DEFAULT_VOLUME_NAME = "default"; + const auto DEFAULT_DISK_NAME = "default"; +} + namespace DB { @@ -18,11 +25,14 @@ namespace ErrorCodes { extern const int BAD_ARGUMENTS; extern const int EXCESSIVE_ELEMENT_IN_CONFIG; + extern const int NO_ELEMENTS_IN_CONFIG; extern const int UNKNOWN_DISK; extern const int UNKNOWN_POLICY; + extern const int UNKNOWN_VOLUME; extern const int LOGICAL_ERROR; } + StoragePolicy::StoragePolicy( String name_, const Poco::Util::AbstractConfiguration & config, @@ -30,44 +40,42 @@ StoragePolicy::StoragePolicy( DiskSelectorPtr disks) : name(std::move(name_)) { - String volumes_prefix = config_prefix + ".volumes"; - if (!config.has(volumes_prefix)) - throw Exception("StoragePolicy must contain at least one volume (.volumes)", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); - Poco::Util::AbstractConfiguration::Keys keys; - config.keys(volumes_prefix, keys); + String volumes_prefix = config_prefix + ".volumes"; + + if (!config.has(volumes_prefix)) + { + if (name != DEFAULT_STORAGE_POLICY_NAME) + throw Exception("Storage policy " + backQuote(name) + " must contain at least one volume (.volumes)", ErrorCodes::NO_ELEMENTS_IN_CONFIG); + } + else + { + config.keys(volumes_prefix, keys); + } for (const auto & attr_name : keys) { if (!std::all_of(attr_name.begin(), attr_name.end(), isWordCharASCII)) throw Exception( - "Volume name can contain only alphanumeric and '_' (" + attr_name + ")", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); - volumes.push_back(std::make_shared(attr_name, config, volumes_prefix + "." + attr_name, disks)); - if (volumes_names.find(attr_name) != volumes_names.end()) - throw Exception("Volumes names must be unique (" + attr_name + " duplicated)", ErrorCodes::UNKNOWN_POLICY); - volumes_names[attr_name] = volumes.size() - 1; + "Volume name can contain only alphanumeric and '_' in storage policy " + backQuote(name) + " (" + attr_name + ")", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + volumes.emplace_back(createVolumeFromConfig(attr_name, config, volumes_prefix + "." + attr_name, disks)); + } + + if (volumes.empty() && name == DEFAULT_STORAGE_POLICY_NAME) + { + auto default_volume = std::make_shared(DEFAULT_VOLUME_NAME, std::vector{disks->get(DEFAULT_DISK_NAME)}, 0, false); + volumes.emplace_back(std::move(default_volume)); } if (volumes.empty()) - throw Exception("StoragePolicy must contain at least one volume.", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + throw Exception("Storage policy " + backQuote(name) + " must contain at least one volume.", ErrorCodes::NO_ELEMENTS_IN_CONFIG); - /// Check that disks are unique in Policy - std::set disk_names; - for (const auto & volume : volumes) - { - for (const auto & disk : volume->getDisks()) - { - if (disk_names.find(disk->getName()) != disk_names.end()) - throw Exception( - "Duplicate disk '" + disk->getName() + "' in storage policy '" + name + "'", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); - - disk_names.insert(disk->getName()); - } - } - - move_factor = config.getDouble(config_prefix + ".move_factor", 0.1); + const double default_move_factor = volumes.size() > 1 ? 0.1 : 0.0; + move_factor = config.getDouble(config_prefix + ".move_factor", default_move_factor); if (move_factor > 1) - throw Exception("Disk move factor have to be in [0., 1.] interval, but set to " + toString(move_factor), ErrorCodes::LOGICAL_ERROR); + throw Exception("Disk move factor have to be in [0., 1.] interval, but set to " + toString(move_factor) + " in storage policy " + backQuote(name), ErrorCodes::LOGICAL_ERROR); + + buildVolumeIndices(); } @@ -75,16 +83,43 @@ StoragePolicy::StoragePolicy(String name_, Volumes volumes_, double move_factor_ : volumes(std::move(volumes_)), name(std::move(name_)), move_factor(move_factor_) { if (volumes.empty()) - throw Exception("StoragePolicy must contain at least one Volume.", ErrorCodes::UNKNOWN_POLICY); + throw Exception("Storage policy " + backQuote(name) + " must contain at least one Volume.", ErrorCodes::NO_ELEMENTS_IN_CONFIG); if (move_factor > 1) - throw Exception("Disk move factor have to be in [0., 1.] interval, but set to " + toString(move_factor), ErrorCodes::LOGICAL_ERROR); + throw Exception("Disk move factor have to be in [0., 1.] interval, but set to " + toString(move_factor) + " in storage policy " + backQuote(name), ErrorCodes::LOGICAL_ERROR); - for (size_t i = 0; i < volumes.size(); ++i) + buildVolumeIndices(); +} + + +StoragePolicy::StoragePolicy(const StoragePolicy & storage_policy, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr disks) + : StoragePolicy(storage_policy.getName(), config, config_prefix, disks) +{ + for (auto & volume : volumes) { - if (volumes_names.find(volumes[i]->getName()) != volumes_names.end()) - throw Exception("Volumes names must be unique (" + volumes[i]->getName() + " duplicated).", ErrorCodes::UNKNOWN_POLICY); - volumes_names[volumes[i]->getName()] = i; + if (storage_policy.volume_index_by_volume_name.count(volume->getName()) > 0) + { + auto old_volume = storage_policy.getVolumeByName(volume->getName()); + try + { + auto new_volume = updateVolumeFromConfig(old_volume, config, config_prefix + ".volumes." + volume->getName(), disks); + volume = std::move(new_volume); + } + catch (Exception & e) + { + /// Default policies are allowed to be missed in configuration. + if (e.code() != ErrorCodes::NO_ELEMENTS_IN_CONFIG || storage_policy.getName() != DEFAULT_STORAGE_POLICY_NAME) + throw; + + Poco::Util::AbstractConfiguration::Keys keys; + config.keys(config_prefix, keys); + if (!keys.empty()) + throw; + } + } } } @@ -93,20 +128,20 @@ bool StoragePolicy::isDefaultPolicy() const { /// Guessing if this policy is default, not 100% correct though. - if (getName() != "default") + if (getName() != DEFAULT_STORAGE_POLICY_NAME) return false; if (volumes.size() != 1) return false; - if (volumes[0]->getName() != "default") + if (volumes[0]->getName() != DEFAULT_VOLUME_NAME) return false; const auto & disks = volumes[0]->getDisks(); if (disks.size() != 1) return false; - if (disks[0]->getName() != "default") + if (disks[0]->getName() != DEFAULT_DISK_NAME) return false; return true; @@ -128,10 +163,10 @@ DiskPtr StoragePolicy::getAnyDisk() const /// StoragePolicy must contain at least one Volume /// Volume must contain at least one Disk if (volumes.empty()) - throw Exception("StoragePolicy has no volumes. It's a bug.", ErrorCodes::LOGICAL_ERROR); + throw Exception("Storage policy " + backQuote(name) + " has no volumes. It's a bug.", ErrorCodes::LOGICAL_ERROR); if (volumes[0]->getDisks().empty()) - throw Exception("Volume '" + volumes[0]->getName() + "' has no disks. It's a bug.", ErrorCodes::LOGICAL_ERROR); + throw Exception("Volume " + backQuote(name) + "." + backQuote(volumes[0]->getName()) + " has no disks. It's a bug.", ErrorCodes::LOGICAL_ERROR); return volumes[0]->getDisks()[0]; } @@ -195,6 +230,24 @@ ReservationPtr StoragePolicy::makeEmptyReservationOnLargestDisk() const } +VolumePtr StoragePolicy::getVolume(size_t index) const +{ + if (index < volume_index_by_volume_name.size()) + return volumes[index]; + else + throw Exception("No volume with index " + std::to_string(index) + " in storage policy " + backQuote(name), ErrorCodes::UNKNOWN_VOLUME); +} + + +VolumePtr StoragePolicy::getVolumeByName(const String & volume_name) const +{ + auto it = volume_index_by_volume_name.find(volume_name); + if (it == volume_index_by_volume_name.end()) + throw Exception("No such volume " + backQuote(volume_name) + " in storage policy " + backQuote(name), ErrorCodes::UNKNOWN_VOLUME); + return getVolume(it->second); +} + + void StoragePolicy::checkCompatibleWith(const StoragePolicyPtr & new_storage_policy) const { std::unordered_set new_volume_names; @@ -204,7 +257,7 @@ void StoragePolicy::checkCompatibleWith(const StoragePolicyPtr & new_storage_pol for (const auto & volume : getVolumes()) { if (new_volume_names.count(volume->getName()) == 0) - throw Exception("New storage policy shall contain volumes of old one", ErrorCodes::BAD_ARGUMENTS); + throw Exception("New storage policy " + backQuote(name) + " shall contain volumes of old one", ErrorCodes::BAD_ARGUMENTS); std::unordered_set new_disk_names; for (const auto & disk : new_storage_policy->getVolumeByName(volume->getName())->getDisks()) @@ -212,24 +265,56 @@ void StoragePolicy::checkCompatibleWith(const StoragePolicyPtr & new_storage_pol for (const auto & disk : volume->getDisks()) if (new_disk_names.count(disk->getName()) == 0) - throw Exception("New storage policy shall contain disks of old one", ErrorCodes::BAD_ARGUMENTS); + throw Exception("New storage policy " + backQuote(name) + " shall contain disks of old one", ErrorCodes::BAD_ARGUMENTS); } } size_t StoragePolicy::getVolumeIndexByDisk(const DiskPtr & disk_ptr) const { - for (size_t i = 0; i < volumes.size(); ++i) - { - const auto & volume = volumes[i]; - for (const auto & disk : volume->getDisks()) - if (disk->getName() == disk_ptr->getName()) - return i; - } - throw Exception("No disk " + disk_ptr->getName() + " in policy " + name, ErrorCodes::UNKNOWN_DISK); + auto it = volume_index_by_disk_name.find(disk_ptr->getName()); + if (it != volume_index_by_disk_name.end()) + return it->second; + else + throw Exception("No disk " + backQuote(disk_ptr->getName()) + " in policy " + backQuote(name), ErrorCodes::UNKNOWN_DISK); } +void StoragePolicy::buildVolumeIndices() +{ + for (size_t index = 0; index < volumes.size(); ++index) + { + const VolumePtr & volume = volumes[index]; + + if (volume_index_by_volume_name.find(volume->getName()) != volume_index_by_volume_name.end()) + throw Exception("Volume names must be unique in storage policy " + + backQuote(name) + " (" + backQuote(volume->getName()) + " is duplicated)" + , ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + + volume_index_by_volume_name[volume->getName()] = index; + + for (const auto & disk : volume->getDisks()) + { + const String & disk_name = disk->getName(); + + if (volume_index_by_disk_name.find(disk_name) != volume_index_by_disk_name.end()) + throw Exception("Disk names must be unique in storage policy " + + backQuote(name) + " (" + backQuote(disk_name) + " is duplicated)" + , ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + + volume_index_by_disk_name[disk_name] = index; + } + } +} + +bool StoragePolicy::hasAnyVolumeWithDisabledMerges() const +{ + for (const auto & volume : volumes) + if (volume->areMergesAvoided()) + return true; + return false; +} + StoragePolicySelector::StoragePolicySelector( const Poco::Util::AbstractConfiguration & config, const String & config_prefix, @@ -242,44 +327,40 @@ StoragePolicySelector::StoragePolicySelector( { if (!std::all_of(name.begin(), name.end(), isWordCharASCII)) throw Exception( - "StoragePolicy name can contain only alphanumeric and '_' (" + name + ")", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); + "Storage policy name can contain only alphanumeric and '_' (" + backQuote(name) + ")", ErrorCodes::EXCESSIVE_ELEMENT_IN_CONFIG); policies.emplace(name, std::make_shared(name, config, config_prefix + "." + name, disks)); LOG_INFO(&Poco::Logger::get("StoragePolicySelector"), "Storage policy {} loaded", backQuote(name)); } - constexpr auto default_storage_policy_name = "default"; - constexpr auto default_volume_name = "default"; - constexpr auto default_disk_name = "default"; - - /// Add default policy if it's not specified explicetly - if (policies.find(default_storage_policy_name) == policies.end()) + /// Add default policy if it isn't explicitly specified. + if (policies.find(DEFAULT_STORAGE_POLICY_NAME) == policies.end()) { - auto default_volume = std::make_shared(default_volume_name, std::vector{disks->get(default_disk_name)}, 0); - - auto default_policy = std::make_shared(default_storage_policy_name, Volumes{default_volume}, 0.0); - policies.emplace(default_storage_policy_name, default_policy); + auto default_policy = std::make_shared(DEFAULT_STORAGE_POLICY_NAME, config, config_prefix + "." + DEFAULT_STORAGE_POLICY_NAME, disks); + policies.emplace(DEFAULT_STORAGE_POLICY_NAME, std::move(default_policy)); } } StoragePolicySelectorPtr StoragePolicySelector::updateFromConfig(const Poco::Util::AbstractConfiguration & config, const String & config_prefix, DiskSelectorPtr disks) const { - Poco::Util::AbstractConfiguration::Keys keys; - config.keys(config_prefix, keys); - std::shared_ptr result = std::make_shared(config, config_prefix, disks); - constexpr auto default_storage_policy_name = "default"; - + /// First pass, check. for (const auto & [name, policy] : policies) { - if (name != default_storage_policy_name && result->policies.count(name) == 0) + if (result->policies.count(name) == 0) throw Exception("Storage policy " + backQuote(name) + " is missing in new configuration", ErrorCodes::BAD_ARGUMENTS); policy->checkCompatibleWith(result->policies[name]); } + /// Second pass, load. + for (const auto & [name, policy] : policies) + { + result->policies[name] = std::make_shared(*policy, config, config_prefix + "." + name, disks); + } + return result; } @@ -288,7 +369,7 @@ StoragePolicyPtr StoragePolicySelector::get(const String & name) const { auto it = policies.find(name); if (it == policies.end()) - throw Exception("Unknown StoragePolicy " + name, ErrorCodes::UNKNOWN_POLICY); + throw Exception("Unknown storage policy " + backQuote(name), ErrorCodes::UNKNOWN_POLICY); return it->second; } diff --git a/src/Disks/StoragePolicy.h b/src/Disks/StoragePolicy.h index 0e0795d8bf1..fc45ed3ed06 100644 --- a/src/Disks/StoragePolicy.h +++ b/src/Disks/StoragePolicy.h @@ -14,6 +14,7 @@ #include #include +#include #include #include #include @@ -36,6 +37,13 @@ public: StoragePolicy(String name_, Volumes volumes_, double move_factor_); + StoragePolicy( + const StoragePolicy & storage_policy, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr disks + ); + bool isDefaultPolicy() const; /// Returns disks ordered by volumes priority @@ -72,29 +80,29 @@ public: /// which should be kept with help of background moves double getMoveFactor() const { return move_factor; } - /// Get volume by index from storage_policy - VolumePtr getVolume(size_t i) const { return (i < volumes_names.size() ? volumes[i] : VolumePtr()); } + /// Get volume by index. + VolumePtr getVolume(size_t index) const; - VolumePtr getVolumeByName(const String & volume_name) const - { - auto it = volumes_names.find(volume_name); - if (it == volumes_names.end()) - return {}; - return getVolume(it->second); - } + VolumePtr getVolumeByName(const String & volume_name) const; /// Checks if storage policy can be replaced by another one. void checkCompatibleWith(const StoragePolicyPtr & new_storage_policy) const; + /// Check if we have any volume with stopped merges + bool hasAnyVolumeWithDisabledMerges() const; + private: Volumes volumes; const String name; - std::map volumes_names; + std::unordered_map volume_index_by_volume_name; + std::unordered_map volume_index_by_disk_name; /// move_factor from interval [0., 1.] /// We move something if disk from this policy /// filled more than total_size * move_factor double move_factor = 0.1; /// by default move factor is 10% + + void buildVolumeIndices(); }; diff --git a/src/Disks/VolumeJBOD.cpp b/src/Disks/VolumeJBOD.cpp index 3ac8a50acfb..e5905f77db3 100644 --- a/src/Disks/VolumeJBOD.cpp +++ b/src/Disks/VolumeJBOD.cpp @@ -56,11 +56,23 @@ VolumeJBOD::VolumeJBOD( /// Default value is 'true' due to backward compatibility. perform_ttl_move_on_insert = config.getBool(config_prefix + ".perform_ttl_move_on_insert", true); + + are_merges_avoided = config.getBool(config_prefix + ".prefer_not_to_merge", false); +} + +VolumeJBOD::VolumeJBOD(const VolumeJBOD & volume_jbod, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr disk_selector) + : VolumeJBOD(volume_jbod.name, config, config_prefix, disk_selector) +{ + are_merges_avoided_user_override = volume_jbod.are_merges_avoided_user_override.load(std::memory_order_relaxed); + last_used = volume_jbod.last_used.load(std::memory_order_relaxed); } DiskPtr VolumeJBOD::getDisk(size_t /* index */) const { - size_t start_from = last_used.fetch_add(1u, std::memory_order_relaxed); + size_t start_from = last_used.fetch_add(1u, std::memory_order_acq_rel); size_t index = start_from % disks.size(); return disks[index]; } @@ -73,7 +85,7 @@ ReservationPtr VolumeJBOD::reserve(UInt64 bytes) if (max_data_part_size != 0 && bytes > max_data_part_size) return {}; - size_t start_from = last_used.fetch_add(1u, std::memory_order_relaxed); + size_t start_from = last_used.fetch_add(1u, std::memory_order_acq_rel); size_t disks_num = disks.size(); for (size_t i = 0; i < disks_num; ++i) { @@ -87,4 +99,19 @@ ReservationPtr VolumeJBOD::reserve(UInt64 bytes) return {}; } +bool VolumeJBOD::areMergesAvoided() const +{ + auto are_merges_avoided_user_override_value = are_merges_avoided_user_override.load(std::memory_order_acquire); + if (are_merges_avoided_user_override_value) + return *are_merges_avoided_user_override_value; + else + return are_merges_avoided; +} + +void VolumeJBOD::setAvoidMergesUserOverride(bool avoid) +{ + are_merges_avoided_user_override.store(avoid, std::memory_order_release); +} + + } diff --git a/src/Disks/VolumeJBOD.h b/src/Disks/VolumeJBOD.h index 52eb2f00721..621125f1109 100644 --- a/src/Disks/VolumeJBOD.h +++ b/src/Disks/VolumeJBOD.h @@ -1,10 +1,19 @@ #pragma once +#include +#include + #include + namespace DB { +class VolumeJBOD; + +using VolumeJBODPtr = std::shared_ptr; +using VolumesJBOD = std::vector; + /** * Implements something similar to JBOD (https://en.wikipedia.org/wiki/Non-RAID_drive_architectures#JBOD). * When MergeTree engine wants to write part — it requests VolumeJBOD to reserve space on the next available @@ -13,8 +22,9 @@ namespace DB class VolumeJBOD : public IVolume { public: - VolumeJBOD(String name_, Disks disks_, UInt64 max_data_part_size_) + VolumeJBOD(String name_, Disks disks_, UInt64 max_data_part_size_, bool are_merges_avoided_) : IVolume(name_, disks_, max_data_part_size_) + , are_merges_avoided(are_merges_avoided_) { } @@ -25,6 +35,13 @@ public: DiskSelectorPtr disk_selector ); + VolumeJBOD( + const VolumeJBOD & volume_jbod, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr disk_selector + ); + VolumeType getType() const override { return VolumeType::JBOD; } /// Always returns next disk (round-robin), ignores argument. @@ -38,11 +55,19 @@ public: /// Returns valid reservation or nullptr if there is no space left on any disk. ReservationPtr reserve(UInt64 bytes) override; + bool areMergesAvoided() const override; + + void setAvoidMergesUserOverride(bool avoid) override; + + /// True if parts on this volume participate in merges according to configuration. + bool are_merges_avoided = true; + private: + /// Index of last used disk. mutable std::atomic last_used = 0; + + /// True if parts on this volume participate in merges according to START/STOP MERGES ON VOLUME. + std::atomic> are_merges_avoided_user_override{std::nullopt}; }; -using VolumeJBODPtr = std::shared_ptr; -using VolumesJBOD = std::vector; - } diff --git a/src/Disks/VolumeRAID1.h b/src/Disks/VolumeRAID1.h index 58cb5bd2623..f6f2d245a49 100644 --- a/src/Disks/VolumeRAID1.h +++ b/src/Disks/VolumeRAID1.h @@ -3,18 +3,23 @@ #include #include + namespace DB { -/// Volume which reserserves space on each underlying disk. +class VolumeRAID1; + +using VolumeRAID1Ptr = std::shared_ptr; + +/// Volume which reserves space on each underlying disk. /// /// NOTE: Just interface implementation, doesn't used in codebase, /// also not available for user. class VolumeRAID1 : public VolumeJBOD { public: - VolumeRAID1(String name_, Disks disks_, UInt64 max_data_part_size_) - : VolumeJBOD(name_, disks_, max_data_part_size_) + VolumeRAID1(String name_, Disks disks_, UInt64 max_data_part_size_, bool are_merges_avoided_in_config_) + : VolumeJBOD(name_, disks_, max_data_part_size_, are_merges_avoided_in_config_) { } @@ -27,11 +32,18 @@ public: { } + VolumeRAID1( + VolumeRAID1 & volume_raid1, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr disk_selector) + : VolumeJBOD(volume_raid1, config, config_prefix, disk_selector) + { + } + VolumeType getType() const override { return VolumeType::RAID1; } ReservationPtr reserve(UInt64 bytes) override; }; -using VolumeRAID1Ptr = std::shared_ptr; - } diff --git a/src/Disks/createVolume.cpp b/src/Disks/createVolume.cpp index 90ed333406e..a290a1d3db3 100644 --- a/src/Disks/createVolume.cpp +++ b/src/Disks/createVolume.cpp @@ -12,6 +12,7 @@ namespace DB namespace ErrorCodes { extern const int UNKNOWN_RAID_TYPE; + extern const int INVALID_RAID_TYPE; } VolumePtr createVolumeFromReservation(const ReservationPtr & reservation, VolumePtr other_volume) @@ -20,12 +21,12 @@ VolumePtr createVolumeFromReservation(const ReservationPtr & reservation, Volume { /// Since reservation on JBOD chooses one of disks and makes reservation there, volume /// for such type of reservation will be with one disk. - return std::make_shared(other_volume->getName(), reservation->getDisk()); + return std::make_shared(other_volume->getName(), reservation->getDisk(), other_volume->max_data_part_size); } if (other_volume->getType() == VolumeType::RAID1) { auto volume = std::dynamic_pointer_cast(other_volume); - return std::make_shared(volume->getName(), reservation->getDisks(), volume->max_data_part_size); + return std::make_shared(volume->getName(), reservation->getDisks(), volume->max_data_part_size, volume->are_merges_avoided); } return nullptr; } @@ -37,17 +38,31 @@ VolumePtr createVolumeFromConfig( DiskSelectorPtr disk_selector ) { - auto has_raid_type = config.has(config_prefix + ".raid_type"); - if (!has_raid_type) - { - return std::make_shared(name, config, config_prefix, disk_selector); - } - String raid_type = config.getString(config_prefix + ".raid_type"); + String raid_type = config.getString(config_prefix + ".raid_type", "JBOD"); if (raid_type == "JBOD") { return std::make_shared(name, config, config_prefix, disk_selector); } - throw Exception("Unknown raid type '" + raid_type + "'", ErrorCodes::UNKNOWN_RAID_TYPE); + throw Exception("Unknown RAID type '" + raid_type + "'", ErrorCodes::UNKNOWN_RAID_TYPE); +} + +VolumePtr updateVolumeFromConfig( + VolumePtr volume, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr & disk_selector +) +{ + String raid_type = config.getString(config_prefix + ".raid_type", "JBOD"); + if (raid_type == "JBOD") + { + VolumeJBODPtr volume_jbod = std::dynamic_pointer_cast(volume); + if (!volume_jbod) + throw Exception("Invalid RAID type '" + raid_type + "', shall be JBOD", ErrorCodes::INVALID_RAID_TYPE); + + return std::make_shared(*volume_jbod, config, config_prefix, disk_selector); + } + throw Exception("Unknown RAID type '" + raid_type + "'", ErrorCodes::UNKNOWN_RAID_TYPE); } } diff --git a/src/Disks/createVolume.h b/src/Disks/createVolume.h index 64f5e73181b..479501759d1 100644 --- a/src/Disks/createVolume.h +++ b/src/Disks/createVolume.h @@ -6,6 +6,7 @@ namespace DB { VolumePtr createVolumeFromReservation(const ReservationPtr & reservation, VolumePtr other_volume); + VolumePtr createVolumeFromConfig( String name_, const Poco::Util::AbstractConfiguration & config, @@ -13,4 +14,11 @@ VolumePtr createVolumeFromConfig( DiskSelectorPtr disk_selector ); +VolumePtr updateVolumeFromConfig( + VolumePtr volume, + const Poco::Util::AbstractConfiguration & config, + const String & config_prefix, + DiskSelectorPtr & disk_selector +); + } diff --git a/src/Disks/ya.make b/src/Disks/ya.make index f01348ff945..d14bc0d05c8 100644 --- a/src/Disks/ya.make +++ b/src/Disks/ya.make @@ -5,7 +5,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( createVolume.cpp diff --git a/src/Disks/ya.make.in b/src/Disks/ya.make.in index 9ed04e23f83..ee13bb272cd 100644 --- a/src/Disks/ya.make.in +++ b/src/Disks/ya.make.in @@ -4,7 +4,6 @@ PEERDIR( clickhouse/src/Common ) -CFLAGS(-g0) SRCS( diff --git a/src/Formats/FormatFactory.cpp b/src/Formats/FormatFactory.cpp index 6707653a84a..4dc5b816420 100644 --- a/src/Formats/FormatFactory.cpp +++ b/src/Formats/FormatFactory.cpp @@ -49,6 +49,7 @@ static FormatSettings getInputFormatSetting(const Settings & settings, const Con format_settings.csv.allow_double_quotes = settings.format_csv_allow_double_quotes; format_settings.csv.unquoted_null_literal_as_null = settings.input_format_csv_unquoted_null_literal_as_null; format_settings.csv.empty_as_default = settings.input_format_defaults_for_omitted_fields; + format_settings.csv.input_format_enum_as_number = settings.input_format_csv_enum_as_number; format_settings.null_as_default = settings.input_format_null_as_default; format_settings.values.interpret_expressions = settings.input_format_values_interpret_expressions; format_settings.values.deduce_templates_of_expressions = settings.input_format_values_deduce_templates_of_expressions; @@ -63,6 +64,7 @@ static FormatSettings getInputFormatSetting(const Settings & settings, const Con format_settings.template_settings.row_format = settings.format_template_row; format_settings.template_settings.row_between_delimiter = settings.format_template_rows_between_delimiter; format_settings.tsv.empty_as_default = settings.input_format_tsv_empty_as_default; + format_settings.tsv.input_format_enum_as_number = settings.input_format_tsv_enum_as_number; format_settings.schema.format_schema = settings.format_schema; format_settings.schema.format_schema_path = context.getFormatSchemaPath(); format_settings.schema.is_server = context.hasGlobalContext() && (context.getGlobalContext().getApplicationType() == Context::ApplicationType::SERVER); diff --git a/src/Formats/FormatSettings.h b/src/Formats/FormatSettings.h index 20c8a032231..8d7c3cdb49f 100644 --- a/src/Formats/FormatSettings.h +++ b/src/Formats/FormatSettings.h @@ -34,6 +34,7 @@ struct FormatSettings bool unquoted_null_literal_as_null = false; bool empty_as_default = false; bool crlf_end_of_line = false; + bool input_format_enum_as_number = false; }; CSV csv; @@ -81,6 +82,7 @@ struct FormatSettings bool empty_as_default = false; bool crlf_end_of_line = false; String null_representation = "\\N"; + bool input_format_enum_as_number = false; }; TSV tsv; diff --git a/src/Formats/ya.make b/src/Formats/ya.make index b4f7b073e21..8e797de39f8 100644 --- a/src/Formats/ya.make +++ b/src/Formats/ya.make @@ -7,7 +7,6 @@ PEERDIR( contrib/libs/protoc ) -CFLAGS(-g0) SRCS( FormatFactory.cpp diff --git a/src/Formats/ya.make.in b/src/Formats/ya.make.in index 51c8bfde2f8..f7d03e7b00f 100644 --- a/src/Formats/ya.make.in +++ b/src/Formats/ya.make.in @@ -6,7 +6,6 @@ PEERDIR( contrib/libs/protoc ) -CFLAGS(-g0) SRCS( diff --git a/src/Functions/CustomWeekTransforms.h b/src/Functions/CustomWeekTransforms.h index 18bd8154a0d..53baaff8db9 100644 --- a/src/Functions/CustomWeekTransforms.h +++ b/src/Functions/CustomWeekTransforms.h @@ -92,7 +92,7 @@ struct ToStartOfWeekImpl template struct Transformer { - Transformer(Transform transform_) + explicit Transformer(Transform transform_) : transform(std::move(transform_)) {} @@ -116,29 +116,29 @@ template struct CustomWeekTransformImpl { template - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/, Transform transform = {}) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/, Transform transform = {}) { const auto op = Transformer{std::move(transform)}; UInt8 week_mode = DEFAULT_WEEK_MODE; if (arguments.size() > 1) { - if (const auto week_mode_column = checkAndGetColumnConst(columns[arguments[1]].column.get())) + if (const auto * week_mode_column = checkAndGetColumnConst(arguments[1].column.get())) week_mode = week_mode_column->getValue(); } - const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(columns, arguments, 2, 0); - const ColumnPtr source_col = columns[arguments[0]].column; + const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(arguments, 2, 0); + const ColumnPtr source_col = arguments[0].column; if (const auto * sources = checkAndGetColumn(source_col.get())) { auto col_to = ToDataType::ColumnType::create(); op.vector(sources->getData(), col_to->getData(), week_mode, time_zone); - columns[result].column = std::move(col_to); + return col_to; } else { throw Exception( - "Illegal column " + columns[arguments[0]].column->getName() + " of first argument of function " + "Illegal column " + arguments[0].column->getName() + " of first argument of function " + Transform::name, ErrorCodes::ILLEGAL_COLUMN); } diff --git a/src/Functions/DateTimeTransforms.h b/src/Functions/DateTimeTransforms.h index bfbb39ef8e7..52cc43c3847 100644 --- a/src/Functions/DateTimeTransforms.h +++ b/src/Functions/DateTimeTransforms.h @@ -683,25 +683,25 @@ struct Transformer template struct DateTimeTransformImpl { - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/, const Transform & transform = {}) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /*input_rows_count*/, const Transform & transform = {}) { using Op = Transformer; - const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(columns, arguments, 1, 0); + const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(arguments, 1, 0); - const ColumnPtr source_col = columns[arguments[0]].column; + const ColumnPtr source_col = arguments[0].column; if (const auto * sources = checkAndGetColumn(source_col.get())) { - auto mutable_result_col = columns[result].type->createColumn(); + auto mutable_result_col = result_type->createColumn(); auto * col_to = assert_cast(mutable_result_col.get()); Op::vector(sources->getData(), col_to->getData(), time_zone, transform); - columns[result].column = std::move(mutable_result_col); + return mutable_result_col; } else { - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + Transform::name, ErrorCodes::ILLEGAL_COLUMN); } diff --git a/src/Functions/FunctionBase64Conversion.h b/src/Functions/FunctionBase64Conversion.h index aa336fc5a3a..adc131053e2 100644 --- a/src/Functions/FunctionBase64Conversion.h +++ b/src/Functions/FunctionBase64Conversion.h @@ -91,14 +91,14 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const ColumnPtr column_string = columns[arguments[0]].column; + const ColumnPtr column_string = arguments[0].column; const ColumnString * input = checkAndGetColumn(column_string.get()); if (!input) throw Exception( - "Illegal column " + columns[arguments[0]].column->getName() + " of first argument of function " + getName(), + "Illegal column " + arguments[0].column->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); auto dst_column = ColumnString::create(); @@ -111,9 +111,9 @@ public: const ColumnString::Offsets & src_offsets = input->getOffsets(); - auto source = input->getChars().data(); - auto dst = dst_data.data(); - auto dst_pos = dst; + const auto * source = input->getChars().data(); + auto * dst = dst_data.data(); + auto * dst_pos = dst; size_t src_offset_prev = 0; @@ -141,7 +141,7 @@ public: { // during decoding character array can be partially polluted // if fail, revert back and clean - auto savepoint = dst_pos; + auto * savepoint = dst_pos; outlen = _tb64d(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); if (!outlen) { @@ -166,7 +166,7 @@ public: dst_data.resize(dst_pos - dst); - columns[result].column = std::move(dst_column); + return dst_column; } }; } diff --git a/src/Functions/FunctionBinaryArithmetic.h b/src/Functions/FunctionBinaryArithmetic.h index 750fdd0d507..43ff42956cd 100644 --- a/src/Functions/FunctionBinaryArithmetic.h +++ b/src/Functions/FunctionBinaryArithmetic.h @@ -326,7 +326,7 @@ struct DecimalBinaryOperation } private: - /// there's implicit type convertion here + /// there's implicit type conversion here static NativeResultType apply(NativeResultType a, NativeResultType b) { if constexpr (can_overflow && check_overflow) @@ -613,17 +613,17 @@ class FunctionBinaryArithmetic : public IFunction } /// Multiply aggregation state by integer constant: by merging it with itself specified number of times. - void executeAggregateMultiply(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const + ColumnPtr executeAggregateMultiply(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const { - ColumnNumbers new_arguments = arguments; - if (WhichDataType(columns[new_arguments[1]].type).isAggregateFunction()) + ColumnsWithTypeAndName new_arguments = arguments; + if (WhichDataType(new_arguments[1].type).isAggregateFunction()) std::swap(new_arguments[0], new_arguments[1]); - if (!isColumnConst(*columns[new_arguments[1]].column)) - throw Exception{"Illegal column " + columns[new_arguments[1]].column->getName() + if (!isColumnConst(*new_arguments[1].column)) + throw Exception{"Illegal column " + new_arguments[1].column->getName() + " of argument of aggregation state multiply. Should be integer constant", ErrorCodes::ILLEGAL_COLUMN}; - const IColumn & agg_state_column = *columns[new_arguments[0]].column; + const IColumn & agg_state_column = *new_arguments[0].column; bool agg_state_is_const = isColumnConst(agg_state_column); const ColumnAggregateFunction & column = typeid_cast( agg_state_is_const ? assert_cast(agg_state_column).getDataColumn() : agg_state_column); @@ -647,7 +647,7 @@ class FunctionBinaryArithmetic : public IFunction auto & vec_to = column_to->getData(); auto & vec_from = column_from->getData(); - UInt64 m = typeid_cast(columns[new_arguments[1]].column.get())->getValue(); + UInt64 m = typeid_cast(new_arguments[1].column.get())->getValue(); // Since we merge the function states by ourselves, we have to have an // Arena for this. Pass it to the resulting column so that the arena @@ -674,16 +674,16 @@ class FunctionBinaryArithmetic : public IFunction } if (agg_state_is_const) - columns[result].column = ColumnConst::create(std::move(column_to), input_rows_count); + return ColumnConst::create(std::move(column_to), input_rows_count); else - columns[result].column = std::move(column_to); + return column_to; } /// Merge two aggregation states together. - void executeAggregateAddition(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const + ColumnPtr executeAggregateAddition(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const { - const IColumn & lhs_column = *columns[arguments[0]].column; - const IColumn & rhs_column = *columns[arguments[1]].column; + const IColumn & lhs_column = *arguments[0].column; + const IColumn & rhs_column = *arguments[1].column; bool lhs_is_const = isColumnConst(lhs_column); bool rhs_is_const = isColumnConst(rhs_column); @@ -707,37 +707,33 @@ class FunctionBinaryArithmetic : public IFunction } if (lhs_is_const && rhs_is_const) - columns[result].column = ColumnConst::create(std::move(column_to), input_rows_count); + return ColumnConst::create(std::move(column_to), input_rows_count); else - columns[result].column = std::move(column_to); + return column_to; } - void executeDateTimeIntervalPlusMinus(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, - size_t result, size_t input_rows_count, const FunctionOverloadResolverPtr & function_builder) const + ColumnPtr executeDateTimeIntervalPlusMinus(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, + size_t input_rows_count, const FunctionOverloadResolverPtr & function_builder) const { - ColumnNumbers new_arguments = arguments; + ColumnsWithTypeAndName new_arguments = arguments; /// Interval argument must be second. - if (WhichDataType(columns[arguments[1]].type).isDateOrDateTime()) + if (WhichDataType(arguments[1].type).isDateOrDateTime()) std::swap(new_arguments[0], new_arguments[1]); /// Change interval argument type to its representation - ColumnsWithTypeAndName new_columns = columns; - new_columns[new_arguments[1]].type = std::make_shared>(); + new_arguments[1].type = std::make_shared>(); - ColumnsWithTypeAndName new_arguments_with_type_and_name = - {new_columns[new_arguments[0]], new_columns[new_arguments[1]]}; - auto function = function_builder->build(new_arguments_with_type_and_name); + auto function = function_builder->build(new_arguments); - function->execute(new_columns, new_arguments, result, input_rows_count); - columns[result].column = new_columns[result].column; + return function->execute(new_arguments, result_type, input_rows_count); } public: static constexpr auto name = Name::name; static FunctionPtr create(const Context & context) { return std::make_shared(context); } - FunctionBinaryArithmetic(const Context & context_) + explicit FunctionBinaryArithmetic(const Context & context_) : context(context_), check_decimal_overflow(decimalCheckArithmeticOverflow(context)) {} @@ -790,7 +786,7 @@ public: new_arguments[1].type = std::make_shared>(); auto function = function_builder->build(new_arguments); - return function->getReturnType(); + return function->getResultType(); } DataTypePtr type_res; @@ -851,20 +847,20 @@ public: return type_res; } - bool executeFixedString(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result) const + ColumnPtr executeFixedString(ColumnsWithTypeAndName & arguments) const { using OpImpl = FixedStringOperationImpl>; - auto col_left_raw = columns[arguments[0]].column.get(); - auto col_right_raw = columns[arguments[1]].column.get(); - if (auto col_left_const = checkAndGetColumnConst(col_left_raw)) + const auto * col_left_raw = arguments[0].column.get(); + const auto * col_right_raw = arguments[1].column.get(); + if (const auto * col_left_const = checkAndGetColumnConst(col_left_raw)) { - if (auto col_right_const = checkAndGetColumnConst(col_right_raw)) + if (const auto * col_right_const = checkAndGetColumnConst(col_right_raw)) { - auto col_left = checkAndGetColumn(col_left_const->getDataColumn()); - auto col_right = checkAndGetColumn(col_right_const->getDataColumn()); + const auto * col_left = checkAndGetColumn(col_left_const->getDataColumn()); + const auto * col_right = checkAndGetColumn(col_right_const->getDataColumn()); if (col_left->getN() != col_right->getN()) - return false; + return nullptr; auto col_res = ColumnFixedString::create(col_left->getN()); auto & out_chars = col_res->getChars(); out_chars.resize(col_left->getN()); @@ -872,25 +868,24 @@ public: col_right->getChars().data(), out_chars.data(), out_chars.size()); - columns[result].column = ColumnConst::create(std::move(col_res), col_left_raw->size()); - return true; + return ColumnConst::create(std::move(col_res), col_left_raw->size()); } } bool is_left_column_const = checkAndGetColumnConst(col_left_raw) != nullptr; bool is_right_column_const = checkAndGetColumnConst(col_right_raw) != nullptr; - auto col_left = is_left_column_const + const auto * col_left = is_left_column_const ? checkAndGetColumn(checkAndGetColumnConst(col_left_raw)->getDataColumn()) : checkAndGetColumn(col_left_raw); - auto col_right = is_right_column_const + const auto * col_right = is_right_column_const ? checkAndGetColumn(checkAndGetColumnConst(col_right_raw)->getDataColumn()) : checkAndGetColumn(col_right_raw); if (col_left && col_right) { if (col_left->getN() != col_right->getN()) - return false; + return nullptr; auto col_res = ColumnFixedString::create(col_left->getN()); auto & out_chars = col_res->getChars(); @@ -922,14 +917,13 @@ public: out_chars.size(), col_left->getN()); } - columns[result].column = std::move(col_res); - return true; + return col_res; } - return false; + return nullptr; } template - bool executeNumeric(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result [[maybe_unused]], const A & left, const B & right) const + ColumnPtr executeNumeric(ColumnsWithTypeAndName & arguments, const A & left, const B & right) const { using LeftDataType = std::decay_t; using RightDataType = std::decay_t; @@ -944,8 +938,8 @@ public: using ColVecT1 = std::conditional_t, ColumnDecimal, ColumnVector>; using ColVecResult = std::conditional_t, ColumnDecimal, ColumnVector>; - auto col_left_raw = columns[arguments[0]].column.get(); - auto col_right_raw = columns[arguments[1]].column.get(); + const auto * col_left_raw = arguments[0].column.get(); + const auto * col_right_raw = arguments[1].column.get(); auto col_left_const = checkAndGetColumnConst(col_left_raw); auto col_right_const = checkAndGetColumnConst(col_right_raw); @@ -981,9 +975,8 @@ public: OpImplCheck::template constantConstant(const_a, const_b, scale_a, scale_b) : OpImpl::template constantConstant(const_a, const_b, scale_a, scale_b); - columns[result].column = ResultDataType(type.getPrecision(), type.getScale()).createColumnConst( + return ResultDataType(type.getPrecision(), type.getScale()).createColumnConst( col_left_const->size(), toField(res, type.getScale())); - return true; } col_res = ColVecResult::create(0, type.getScale()); @@ -1016,7 +1009,7 @@ public: OpImpl::template vectorConstant(col_left->getData(), const_b, vec_res, scale_a, scale_b); } else - return false; + return nullptr; } else { @@ -1026,8 +1019,7 @@ public: if (col_left_const && col_right_const) { auto res = OpImpl::constantConstant(col_left_const->template getValue(), col_right_const->template getValue()); - columns[result].column = ResultDataType().createColumnConst(col_left_const->size(), toField(res)); - return true; + return ResultDataType().createColumnConst(col_left_const->size(), toField(res)); } col_res = ColVecResult::create(); @@ -1047,43 +1039,40 @@ public: OpImpl::vectorConstant(col_left->getData().data(), col_right_const->template getValue(), vec_res.data(), vec_res.size()); } else - return false; + return nullptr; } - columns[result].column = std::move(col_res); - return true; + return col_res; } - return false; + return nullptr; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { /// Special case when multiply aggregate function state - if (isAggregateMultiply(columns[arguments[0]].type, columns[arguments[1]].type)) + if (isAggregateMultiply(arguments[0].type, arguments[1].type)) { - executeAggregateMultiply(columns, arguments, result, input_rows_count); - return; + return executeAggregateMultiply(arguments, result_type, input_rows_count); } /// Special case - addition of two aggregate functions states - if (isAggregateAddition(columns[arguments[0]].type, columns[arguments[1]].type)) + if (isAggregateAddition(arguments[0].type, arguments[1].type)) { - executeAggregateAddition(columns, arguments, result, input_rows_count); - return; + return executeAggregateAddition(arguments, result_type, input_rows_count); } /// Special case when the function is plus or minus, one of arguments is Date/DateTime and another is Interval. if (auto function_builder - = getFunctionForIntervalArithmetic(columns[arguments[0]].type, columns[arguments[1]].type, context)) + = getFunctionForIntervalArithmetic(arguments[0].type, arguments[1].type, context)) { - executeDateTimeIntervalPlusMinus(columns, arguments, result, input_rows_count, function_builder); - return; + return executeDateTimeIntervalPlusMinus(arguments, result_type, input_rows_count, function_builder); } - const auto & left_argument = columns[arguments[0]]; - const auto & right_argument = columns[arguments[1]]; - auto * left_generic = left_argument.type.get(); - auto * right_generic = right_argument.type.get(); + const auto & left_argument = arguments[0]; + const auto & right_argument = arguments[1]; + const auto * left_generic = left_argument.type.get(); + const auto * right_generic = right_argument.type.get(); + ColumnPtr res; bool valid = castBothTypes(left_generic, right_generic, [&](const auto & left, const auto & right) { using LeftDataType = std::decay_t; @@ -1093,10 +1082,10 @@ public: if constexpr (!Op::allow_fixed_string) return false; else - return executeFixedString(columns, arguments, result); + return (res = executeFixedString(arguments)) != nullptr; } else - return executeNumeric(columns, arguments, result, left, right); + return (res = executeNumeric(arguments, left, right)) != nullptr; }); if (!valid) @@ -1109,6 +1098,8 @@ public: left_argument.name, left_argument.type->getName(), right_argument.name, right_argument.type->getName()); } + + return res; } #if USE_EMBEDDED_COMPILER @@ -1190,30 +1181,26 @@ public: { } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (left.column && isColumnConst(*left.column) && arguments.size() == 1) { ColumnsWithTypeAndName columns_with_constant = {{left.column->cloneResized(input_rows_count), left.type, left.name}, - columns[arguments[0]], - columns[result]}; + arguments[0]}; - Base::executeImpl(columns_with_constant, {0, 1}, 2, input_rows_count); - columns[result] = columns_with_constant[2]; + return Base::executeImpl(columns_with_constant, result_type, input_rows_count); } else if (right.column && isColumnConst(*right.column) && arguments.size() == 1) { ColumnsWithTypeAndName columns_with_constant - = {columns[arguments[0]], - {right.column->cloneResized(input_rows_count), right.type, right.name}, - columns[result]}; + = {arguments[0], + {right.column->cloneResized(input_rows_count), right.type, right.name}}; - Base::executeImpl(columns_with_constant, {0, 1}, 2, input_rows_count); - columns[result] = columns_with_constant[2]; + return Base::executeImpl(columns_with_constant, result_type, input_rows_count); } else - Base::executeImpl(columns, arguments, result, input_rows_count); + return Base::executeImpl(arguments, result_type, input_rows_count); } bool hasInformationAboutMonotonicity() const override @@ -1246,12 +1233,11 @@ public: { ColumnsWithTypeAndName columns_with_constant = {{left.column->cloneResized(1), left.type, left.name}, - {right.type->createColumnConst(1, point), right.type, right.name}, - {nullptr, return_type, ""}}; + {right.type->createColumnConst(1, point), right.type, right.name}}; - Base::executeImpl(columns_with_constant, {0, 1}, 2, 1); + auto col = Base::executeImpl(columns_with_constant, return_type, 1); Field point_transformed; - columns_with_constant[2].column->get(0, point_transformed); + col->get(0, point_transformed); return point_transformed; }; transform(left_point); @@ -1282,12 +1268,11 @@ public: { ColumnsWithTypeAndName columns_with_constant = {{left.type->createColumnConst(1, point), left.type, left.name}, - {right.column->cloneResized(1), right.type, right.name}, - {nullptr, return_type, ""}}; + {right.column->cloneResized(1), right.type, right.name}}; - Base::executeImpl(columns_with_constant, {0, 1}, 2, 1); + auto col = Base::executeImpl(columns_with_constant, return_type, 1); Field point_transformed; - columns_with_constant[2].column->get(0, point_transformed); + col->get(0, point_transformed); return point_transformed; }; diff --git a/src/Functions/FunctionBitTestMany.h b/src/Functions/FunctionBitTestMany.h index a3c42582ef1..0c8b803bd22 100644 --- a/src/Functions/FunctionBitTestMany.h +++ b/src/Functions/FunctionBitTestMany.h @@ -54,32 +54,35 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns , const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /*input_rows_count*/) const override { - const auto value_col = columns[arguments.front()].column.get(); + const auto * value_col = arguments.front().column.get(); - if (!execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col) - && !execute(columns, arguments, result, value_col)) + ColumnPtr res; + if (!((res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)) + || (res = execute(arguments, result_type, value_col)))) throw Exception{"Illegal column " + value_col->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; + + return res; } private: template - bool execute( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, + ColumnPtr execute( + ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const IColumn * const value_col_untyped) const { if (const auto value_col = checkAndGetColumn>(value_col_untyped)) { const auto size = value_col->size(); bool is_const; - const auto const_mask = createConstMaskIfConst(columns, arguments, is_const); + const auto const_mask = createConstMaskIfConst(arguments, is_const); const auto & val = value_col->getData(); auto out_col = ColumnVector::create(size); @@ -92,29 +95,28 @@ private: } else { - const auto mask = createMask(size, columns, arguments); + const auto mask = createMask(size, arguments); for (const auto i : ext::range(0, size)) out[i] = Impl::apply(val[i], mask[i]); } - columns[result].column = std::move(out_col); - return true; + return out_col; } else if (const auto value_col_const = checkAndGetColumnConst>(value_col_untyped)) { const auto size = value_col_const->size(); bool is_const; - const auto const_mask = createConstMaskIfConst(columns, arguments, is_const); + const auto const_mask = createConstMaskIfConst(arguments, is_const); const auto val = value_col_const->template getValue(); if (is_const) { - columns[result].column = columns[result].type->createColumnConst(size, toField(Impl::apply(val, const_mask))); + return result_type->createColumnConst(size, toField(Impl::apply(val, const_mask))); } else { - const auto mask = createMask(size, columns, arguments); + const auto mask = createMask(size, arguments); auto out_col = ColumnVector::create(size); auto & out = out_col->getData(); @@ -122,24 +124,22 @@ private: for (const auto i : ext::range(0, size)) out[i] = Impl::apply(val, mask[i]); - columns[result].column = std::move(out_col); + return out_col; } - - return true; } - return false; + return nullptr; } template - ValueType createConstMaskIfConst(const ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, bool & out_is_const) const + ValueType createConstMaskIfConst(const ColumnsWithTypeAndName & arguments, bool & out_is_const) const { out_is_const = true; ValueType mask = 0; for (const auto i : ext::range(1, arguments.size())) { - if (auto pos_col_const = checkAndGetColumnConst>(columns[arguments[i]].column.get())) + if (auto pos_col_const = checkAndGetColumnConst>(arguments[i].column.get())) { const auto pos = pos_col_const->getUInt(0); if (pos < 8 * sizeof(ValueType)) @@ -156,13 +156,13 @@ private: } template - PaddedPODArray createMask(const size_t size, const ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments) const + PaddedPODArray createMask(const size_t size, const ColumnsWithTypeAndName & arguments) const { PaddedPODArray mask(size, ValueType{}); for (const auto i : ext::range(1, arguments.size())) { - const auto pos_col = columns[arguments[i]].column.get(); + const auto * pos_col = arguments[i].column.get(); if (!addToMaskImpl(mask, pos_col) && !addToMaskImpl(mask, pos_col) diff --git a/src/Functions/FunctionCustomWeekToSomething.h b/src/Functions/FunctionCustomWeekToSomething.h index ce8a0bf53d6..74d6a2b5182 100644 --- a/src/Functions/FunctionCustomWeekToSomething.h +++ b/src/Functions/FunctionCustomWeekToSomething.h @@ -96,26 +96,26 @@ public: bool useDefaultImplementationForConstants() const override { return true; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1, 2}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); WhichDataType which(from_type); if (which.isDate()) - CustomWeekTransformImpl::execute( - columns, arguments, result, input_rows_count, Transform{}); + return CustomWeekTransformImpl::execute( + arguments, result_type, input_rows_count, Transform{}); else if (which.isDateTime()) - CustomWeekTransformImpl::execute( - columns, arguments, result, input_rows_count, Transform{}); + return CustomWeekTransformImpl::execute( + arguments, result_type, input_rows_count, Transform{}); else if (which.isDateTime64()) { - CustomWeekTransformImpl::execute( - columns, arguments, result, input_rows_count, + return CustomWeekTransformImpl::execute( + arguments, result_type, input_rows_count, TransformDateTime64{assert_cast(from_type)->getScale()}); } else throw Exception( - "Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName(), + "Illegal type " + arguments[0].type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); } diff --git a/src/Functions/FunctionDateOrDateTimeAddInterval.h b/src/Functions/FunctionDateOrDateTimeAddInterval.h index 6f7cec3eaca..bf2d20ceba7 100644 --- a/src/Functions/FunctionDateOrDateTimeAddInterval.h +++ b/src/Functions/FunctionDateOrDateTimeAddInterval.h @@ -305,7 +305,7 @@ private: template struct DateTimeAddIntervalImpl { - static void execute(Transform transform, ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result) + static ColumnPtr execute(Transform transform, ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type) { using FromValueType = typename FromDataType::FieldType; using FromColumnType = typename FromDataType::ColumnType; @@ -313,16 +313,16 @@ struct DateTimeAddIntervalImpl auto op = Adder{std::move(transform)}; - const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(columns, arguments, 2, 0); + const DateLUTImpl & time_zone = extractTimeZoneFromFunctionArguments(arguments, 2, 0); - const ColumnPtr source_col = columns[arguments[0]].column; + const ColumnPtr source_col = arguments[0].column; - auto result_col = columns[result].type->createColumn(); + auto result_col = result_type->createColumn(); auto col_to = assert_cast(result_col.get()); if (const auto * sources = checkAndGetColumn(source_col.get())) { - const IColumn & delta_column = *columns[arguments[1]].column; + const IColumn & delta_column = *arguments[1].column; if (const auto * delta_const_column = typeid_cast(&delta_column)) op.vectorConstant(sources->getData(), col_to->getData(), delta_const_column->getInt(0), time_zone); @@ -334,16 +334,16 @@ struct DateTimeAddIntervalImpl op.constantVector( sources_const->template getValue(), col_to->getData(), - *columns[arguments[1]].column, time_zone); + *arguments[1].column, time_zone); } else { - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + Transform::name, ErrorCodes::ILLEGAL_COLUMN); } - columns[result].column = std::move(result_col); + return result_col; } }; @@ -463,28 +463,28 @@ public: bool useDefaultImplementationForConstants() const override { return true; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {2}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /*input_rows_count*/) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); WhichDataType which(from_type); if (which.isDate()) { - DateTimeAddIntervalImpl, Transform>::execute( - Transform{}, columns, arguments, result); + return DateTimeAddIntervalImpl, Transform>::execute( + Transform{}, arguments, result_type); } else if (which.isDateTime()) { - DateTimeAddIntervalImpl, Transform>::execute( - Transform{}, columns, arguments, result); + return DateTimeAddIntervalImpl, Transform>::execute( + Transform{}, arguments, result_type); } else if (const auto * datetime64_type = assert_cast(from_type)) { - DateTimeAddIntervalImpl, Transform>::execute( - Transform{datetime64_type->getScale()}, columns, arguments, result); + return DateTimeAddIntervalImpl, Transform>::execute( + Transform{datetime64_type->getScale()}, arguments, result_type); } else - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of first argument of function " + getName(), + throw Exception("Illegal type " + arguments[0].type->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); } }; diff --git a/src/Functions/FunctionDateOrDateTimeToSomething.h b/src/Functions/FunctionDateOrDateTimeToSomething.h index 248007c66de..2d2e4a7ad6f 100644 --- a/src/Functions/FunctionDateOrDateTimeToSomething.h +++ b/src/Functions/FunctionDateOrDateTimeToSomething.h @@ -95,23 +95,23 @@ public: bool useDefaultImplementationForConstants() const override { return true; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); WhichDataType which(from_type); if (which.isDate()) - DateTimeTransformImpl::execute(columns, arguments, result, input_rows_count); + return DateTimeTransformImpl::execute(arguments, result_type, input_rows_count); else if (which.isDateTime()) - DateTimeTransformImpl::execute(columns, arguments, result, input_rows_count); + return DateTimeTransformImpl::execute(arguments, result_type, input_rows_count); else if (which.isDateTime64()) { const auto scale = static_cast(from_type)->getScale(); const TransformDateTime64 transformer(scale); - DateTimeTransformImpl::execute(columns, arguments, result, input_rows_count, transformer); + return DateTimeTransformImpl::execute(arguments, result_type, input_rows_count, transformer); } else - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName(), + throw Exception("Illegal type " + arguments[0].type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); } diff --git a/src/Functions/FunctionFQDN.cpp b/src/Functions/FunctionFQDN.cpp index 06048c61ca8..b47675d63b4 100644 --- a/src/Functions/FunctionFQDN.cpp +++ b/src/Functions/FunctionFQDN.cpp @@ -34,9 +34,9 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers &, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName &, const DataTypePtr & result_type, size_t input_rows_count) const override { - columns[result].column = columns[result].type->createColumnConst( + return result_type->createColumnConst( input_rows_count, getFQDNOrHostName())->convertToFullColumnIfConst(); } }; diff --git a/src/Functions/FunctionHelpers.cpp b/src/Functions/FunctionHelpers.cpp index 473feeaca60..f5ace858665 100644 --- a/src/Functions/FunctionHelpers.cpp +++ b/src/Functions/FunctionHelpers.cpp @@ -7,7 +7,6 @@ #include #include #include -#include namespace DB @@ -51,16 +50,12 @@ Columns convertConstTupleToConstantElements(const ColumnConst & column) } -static ColumnsWithTypeAndName createBlockWithNestedColumnsImpl(const ColumnsWithTypeAndName & columns, const std::unordered_set & args) +ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns) { ColumnsWithTypeAndName res; - size_t num_columns = columns.size(); - - for (size_t i = 0; i < num_columns; ++i) + for (const auto & col : columns) { - const auto & col = columns[i]; - - if (args.count(i) && col.type->isNullable()) + if (col.type->isNullable()) { const DataTypePtr & nested_type = static_cast(*col.type).getNestedType(); @@ -88,20 +83,6 @@ static ColumnsWithTypeAndName createBlockWithNestedColumnsImpl(const ColumnsWith return res; } - -ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns, const ColumnNumbers & args) -{ - std::unordered_set args_set(args.begin(), args.end()); - return createBlockWithNestedColumnsImpl(columns, args_set); -} - -ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns, const ColumnNumbers & args, size_t result) -{ - std::unordered_set args_set(args.begin(), args.end()); - args_set.insert(result); - return createBlockWithNestedColumnsImpl(columns, args_set); -} - void validateArgumentType(const IFunction & func, const DataTypes & arguments, size_t argument_index, bool (* validator_func)(const IDataType &), const char * expected_type_description) @@ -137,7 +118,7 @@ void validateArgumentsImpl(const IFunction & func, const auto & arg = arguments[i + argument_offset]; const auto descriptor = descriptors[i]; if (int error_code = descriptor.isValid(arg.type, arg.column); error_code != 0) - throw Exception("Illegal type of argument #" + std::to_string(i) + throw Exception("Illegal type of argument #" + std::to_string(argument_offset + i + 1) // +1 is for human-friendly 1-based indexing + (descriptor.argument_name ? " '" + std::string(descriptor.argument_name) + "'" : String{}) + " of function " + func.getName() + (descriptor.expected_type_description ? String(", expected ") + descriptor.expected_type_description : String{}) diff --git a/src/Functions/FunctionHelpers.h b/src/Functions/FunctionHelpers.h index 64b7817d6e9..ac10fc14d39 100644 --- a/src/Functions/FunctionHelpers.h +++ b/src/Functions/FunctionHelpers.h @@ -82,13 +82,9 @@ inline std::enable_if_t, Field> toField(const T & x, UInt32 s Columns convertConstTupleToConstantElements(const ColumnConst & column); -/// Returns the copy of a given columns in which each column specified in -/// the "arguments" parameter is replaced with its respective nested +/// Returns the copy of a given columns in which each column is replaced with its respective nested /// column if it is nullable. -ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns, const ColumnNumbers & args); - -/// Similar function as above. Additionally transform the result type if needed. -ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns, const ColumnNumbers & args, size_t result); +ColumnsWithTypeAndName createBlockWithNestedColumns(const ColumnsWithTypeAndName & columns); /// Checks argument type at specified index with predicate. /// throws if there is no argument at specified index or if predicate returns false. diff --git a/src/Functions/FunctionJoinGet.cpp b/src/Functions/FunctionJoinGet.cpp index 81df6ca7912..4e97951fbc0 100644 --- a/src/Functions/FunctionJoinGet.cpp +++ b/src/Functions/FunctionJoinGet.cpp @@ -17,19 +17,19 @@ namespace ErrorCodes } template -void ExecutableFunctionJoinGet::execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t) +ColumnPtr ExecutableFunctionJoinGet::execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t) { ColumnsWithTypeAndName keys; for (size_t i = 2; i < arguments.size(); ++i) { - auto key = columns[arguments[i]]; + auto key = arguments[i]; keys.emplace_back(std::move(key)); } - columns[result] = join->joinGet(keys, result_columns); + return join->joinGet(keys, result_columns).column; } template -ExecutableFunctionImplPtr FunctionJoinGet::prepare(const ColumnsWithTypeAndName &, const ColumnNumbers &, size_t) const +ExecutableFunctionImplPtr FunctionJoinGet::prepare(const ColumnsWithTypeAndName &) const { return std::make_unique>(join, DB::Block{{return_type->createColumn(), return_type, attr_name}}); } diff --git a/src/Functions/FunctionJoinGet.h b/src/Functions/FunctionJoinGet.h index 88366aeb0fd..780b59e20f4 100644 --- a/src/Functions/FunctionJoinGet.h +++ b/src/Functions/FunctionJoinGet.h @@ -24,7 +24,7 @@ public: bool useDefaultImplementationForLowCardinalityColumns() const override { return true; } bool useDefaultImplementationForConstants() const override { return true; } - void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override; + ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) override; String getName() const override { return name; } @@ -54,9 +54,9 @@ public: String getName() const override { return name; } const DataTypes & getArgumentTypes() const override { return argument_types; } - const DataTypePtr & getReturnType() const override { return return_type; } + const DataTypePtr & getResultType() const override { return return_type; } - ExecutableFunctionImplPtr prepare(const ColumnsWithTypeAndName & sample_columns, const ColumnNumbers & arguments, size_t result) const override; + ExecutableFunctionImplPtr prepare(const ColumnsWithTypeAndName &) const override; private: TableLockHolder table_lock; diff --git a/src/Functions/FunctionMathBinaryFloat64.h b/src/Functions/FunctionMathBinaryFloat64.h index 7ddb4bfcaf2..ea222379e1c 100644 --- a/src/Functions/FunctionMathBinaryFloat64.h +++ b/src/Functions/FunctionMathBinaryFloat64.h @@ -54,7 +54,7 @@ private: } template - bool executeTyped(ColumnsWithTypeAndName & columns, const size_t result, const ColumnConst * left_arg, const IColumn * right_arg) const + ColumnPtr executeTyped(const ColumnConst * left_arg, const IColumn * right_arg) const { if (const auto right_arg_typed = checkAndGetColumn>(right_arg)) { @@ -95,15 +95,14 @@ private: memcpy(&dst_data[rows_size], dst_remaining, rows_remaining * sizeof(Float64)); } - columns[result].column = std::move(dst); - return true; + return dst; } - return false; + return nullptr; } template - bool executeTyped(ColumnsWithTypeAndName & columns, const size_t result, const ColumnVector * left_arg, const IColumn * right_arg) const + ColumnPtr executeTyped(const ColumnVector * left_arg, const IColumn * right_arg) const { if (const auto right_arg_typed = checkAndGetColumn>(right_arg)) { @@ -157,8 +156,7 @@ private: memcpy(&dst_data[rows_size], dst_remaining, rows_remaining * sizeof(Float64)); } - columns[result].column = std::move(dst); - return true; + return dst; } if (const auto right_arg_typed = checkAndGetColumnConst>(right_arg)) { @@ -200,17 +198,17 @@ private: memcpy(&dst_data[rows_size], dst_remaining, rows_remaining * sizeof(Float64)); } - columns[result].column = std::move(dst); - return true; + return dst; } - return false; + return nullptr; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnWithTypeAndName & col_left = columns[arguments[0]]; - const ColumnWithTypeAndName & col_right = columns[arguments[1]]; + const ColumnWithTypeAndName & col_left = arguments[0]; + const ColumnWithTypeAndName & col_right = arguments[1]; + ColumnPtr res; auto call = [&](const auto & types) -> bool { @@ -224,7 +222,7 @@ private: if (const auto left_arg_typed = checkAndGetColumn(left_arg)) { - if (executeTyped(columns, result, left_arg_typed, right_arg)) + if ((res = executeTyped(left_arg_typed, right_arg))) return true; throw Exception{"Illegal column " + right_arg->getName() + " of second argument of function " + getName(), @@ -232,7 +230,7 @@ private: } if (const auto left_arg_typed = checkAndGetColumnConst(left_arg)) { - if (executeTyped(columns, result, left_arg_typed, right_arg)) + if ((res = executeTyped(left_arg_typed, right_arg))) return true; throw Exception{"Illegal column " + right_arg->getName() + " of second argument of function " + getName(), @@ -248,6 +246,8 @@ private: if (!callOnBasicTypes(left_index, right_index, call)) throw Exception{"Illegal column " + col_left.column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; + + return res; } }; diff --git a/src/Functions/FunctionMathConstFloat64.h b/src/Functions/FunctionMathConstFloat64.h index 56ad4571f38..42729d5e9f6 100644 --- a/src/Functions/FunctionMathConstFloat64.h +++ b/src/Functions/FunctionMathConstFloat64.h @@ -25,9 +25,9 @@ private: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers &, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName &, const DataTypePtr & result_type, size_t input_rows_count) const override { - columns[result].column = columns[result].type->createColumnConst(input_rows_count, Impl::value); + return result_type->createColumnConst(input_rows_count, Impl::value); } }; diff --git a/src/Functions/FunctionMathUnary.h b/src/Functions/FunctionMathUnary.h index bb26efb0555..abf38d277f3 100644 --- a/src/Functions/FunctionMathUnary.h +++ b/src/Functions/FunctionMathUnary.h @@ -113,7 +113,7 @@ private: } template - static bool execute(ColumnsWithTypeAndName & columns, const ColumnVector * col, const size_t result) + static ColumnPtr execute(const ColumnVector * col) { const auto & src_data = col->getData(); const size_t size = src_data.size(); @@ -124,12 +124,11 @@ private: executeInIterations(src_data.data(), dst_data.data(), size); - columns[result].column = std::move(dst); - return true; + return dst; } template - static bool execute(ColumnsWithTypeAndName & columns, const ColumnDecimal * col, const size_t result) + static ColumnPtr execute(const ColumnDecimal * col) { const auto & src_data = col->getData(); const size_t size = src_data.size(); @@ -144,15 +143,15 @@ private: executeInIterations(dst_data.data(), dst_data.data(), size); - columns[result].column = std::move(dst); - return true; + return dst; } bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnWithTypeAndName & col = columns[arguments[0]]; + const ColumnWithTypeAndName & col = arguments[0]; + ColumnPtr res; auto call = [&](const auto & types) -> bool { @@ -162,12 +161,14 @@ private: using ColVecType = std::conditional_t, ColumnDecimal, ColumnVector>; const auto col_vec = checkAndGetColumn(col.column.get()); - return execute(columns, col_vec, result); + return (res = execute(col_vec)) != nullptr; }; if (!callOnBasicType(col.type->getTypeId(), call)) throw Exception{"Illegal column " + col.column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; + + return res; } }; diff --git a/src/Functions/FunctionNumericPredicate.h b/src/Functions/FunctionNumericPredicate.h index 96867c774df..825a8b0de15 100644 --- a/src/Functions/FunctionNumericPredicate.h +++ b/src/Functions/FunctionNumericPredicate.h @@ -46,25 +46,28 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const auto in = columns[arguments.front()].column.get(); + const auto * in = arguments.front().column.get(); - if ( !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result) - && !execute(columns, in, result)) + ColumnPtr res; + if (!((res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)) + || (res = execute(in)))) throw Exception{"Illegal column " + in->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; + + return res; } template - bool execute(ColumnsWithTypeAndName & columns, const IColumn * in_untyped, const size_t result) const + ColumnPtr execute(const IColumn * in_untyped) const { if (const auto in = checkAndGetColumn>(in_untyped)) { @@ -78,11 +81,10 @@ public: for (const auto i : ext::range(0, size)) out_data[i] = Impl::execute(in_data[i]); - columns[result].column = std::move(out); - return true; + return out; } - return false; + return nullptr; } }; diff --git a/src/Functions/FunctionStartsEndsWith.h b/src/Functions/FunctionStartsEndsWith.h index 6833f4554d2..612e0b3b046 100644 --- a/src/Functions/FunctionStartsEndsWith.h +++ b/src/Functions/FunctionStartsEndsWith.h @@ -63,10 +63,10 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const IColumn * haystack_column = columns[arguments[0]].column.get(); - const IColumn * needle_column = columns[arguments[1]].column.get(); + const IColumn * haystack_column = arguments[0].column.get(); + const IColumn * needle_column = arguments[1].column.get(); auto col_res = ColumnVector::create(); typename ColumnVector::Container & vec_res = col_res->getData(); @@ -84,7 +84,7 @@ public: else throw Exception("Illegal combination of columns as arguments of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - columns[result].column = std::move(col_res); + return col_res; } private: @@ -159,9 +159,9 @@ public: #endif } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - selector.selectAndExecute(columns, arguments, result, input_rows_count); + return selector.selectAndExecute(arguments, result_type, input_rows_count); } static FunctionPtr create(const Context & context) diff --git a/src/Functions/FunctionStringOrArrayToT.h b/src/Functions/FunctionStringOrArrayToT.h index 83a860f7175..f806106560c 100644 --- a/src/Functions/FunctionStringOrArrayToT.h +++ b/src/Functions/FunctionStringOrArrayToT.h @@ -50,9 +50,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /*input_rows_count*/) const override { - const ColumnPtr column = columns[arguments[0]].column; + const ColumnPtr column = arguments[0].column; if (const ColumnString * col = checkAndGetColumn(column.get())) { auto col_res = ColumnVector::create(); @@ -61,7 +61,7 @@ public: vec_res.resize(col->size()); Impl::vector(col->getChars(), col->getOffsets(), vec_res); - columns[result].column = std::move(col_res); + return col_res; } else if (const ColumnFixedString * col_fixed = checkAndGetColumn(column.get())) { @@ -70,7 +70,7 @@ public: ResultType res = 0; Impl::vectorFixedToConstant(col_fixed->getChars(), col_fixed->getN(), res); - columns[result].column = columns[result].type->createColumnConst(col_fixed->size(), toField(res)); + return result_type->createColumnConst(col_fixed->size(), toField(res)); } else { @@ -80,7 +80,7 @@ public: vec_res.resize(col_fixed->size()); Impl::vectorFixedToVector(col_fixed->getChars(), col_fixed->getN(), vec_res); - columns[result].column = std::move(col_res); + return col_res; } } else if (const ColumnArray * col_arr = checkAndGetColumn(column.get())) @@ -91,10 +91,10 @@ public: vec_res.resize(col_arr->size()); Impl::array(col_arr->getOffsets(), vec_res); - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + " of argument of function " + getName(), + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } }; diff --git a/src/Functions/FunctionStringReplace.h b/src/Functions/FunctionStringReplace.h index 137c2241b7f..4ec85591726 100644 --- a/src/Functions/FunctionStringReplace.h +++ b/src/Functions/FunctionStringReplace.h @@ -52,17 +52,17 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr column_src = columns[arguments[0]].column; - const ColumnPtr column_needle = columns[arguments[1]].column; - const ColumnPtr column_replacement = columns[arguments[2]].column; + const ColumnPtr column_src = arguments[0].column; + const ColumnPtr column_needle = arguments[1].column; + const ColumnPtr column_replacement = arguments[2].column; if (!isColumnConst(*column_needle) || !isColumnConst(*column_replacement)) throw Exception("2nd and 3rd arguments of function " + getName() + " must be constants.", ErrorCodes::ILLEGAL_COLUMN); - const IColumn * c1 = columns[arguments[1]].column.get(); - const IColumn * c2 = columns[arguments[2]].column.get(); + const IColumn * c1 = arguments[1].column.get(); + const IColumn * c2 = arguments[2].column.get(); const ColumnConst * c1_const = typeid_cast(c1); const ColumnConst * c2_const = typeid_cast(c2); String needle = c1_const->getValue(); @@ -75,17 +75,17 @@ public: { auto col_res = ColumnString::create(); Impl::vector(col->getChars(), col->getOffsets(), needle, replacement, col_res->getChars(), col_res->getOffsets()); - columns[result].column = std::move(col_res); + return col_res; } else if (const ColumnFixedString * col_fixed = checkAndGetColumn(column_src.get())) { auto col_res = ColumnString::create(); Impl::vectorFixed(col_fixed->getChars(), col_fixed->getN(), needle, replacement, col_res->getChars(), col_res->getOffsets()); - columns[result].column = std::move(col_res); + return col_res; } else throw Exception( - "Illegal column " + columns[arguments[0]].column->getName() + " of first argument of function " + getName(), + "Illegal column " + arguments[0].column->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } }; diff --git a/src/Functions/FunctionStringToString.h b/src/Functions/FunctionStringToString.h index 56f434a2ffa..db85e85a053 100644 --- a/src/Functions/FunctionStringToString.h +++ b/src/Functions/FunctionStringToString.h @@ -52,24 +52,24 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr column = columns[arguments[0]].column; + const ColumnPtr column = arguments[0].column; if (const ColumnString * col = checkAndGetColumn(column.get())) { auto col_res = ColumnString::create(); Impl::vector(col->getChars(), col->getOffsets(), col_res->getChars(), col_res->getOffsets()); - columns[result].column = std::move(col_res); + return col_res; } else if (const ColumnFixedString * col_fixed = checkAndGetColumn(column.get())) { auto col_res = ColumnFixedString::create(col_fixed->getN()); Impl::vectorFixed(col_fixed->getChars(), col_fixed->getN(), col_res->getChars()); - columns[result].column = std::move(col_res); + return col_res; } else throw Exception( - "Illegal column " + columns[arguments[0]].column->getName() + " of argument of function " + getName(), + "Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } }; diff --git a/src/Functions/FunctionUnaryArithmetic.h b/src/Functions/FunctionUnaryArithmetic.h index 7d83f18793d..e62781cc3a1 100644 --- a/src/Functions/FunctionUnaryArithmetic.h +++ b/src/Functions/FunctionUnaryArithmetic.h @@ -154,9 +154,10 @@ public: return result; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - bool valid = castType(columns[arguments[0]].type.get(), [&](const auto & type) + ColumnPtr result_column; + bool valid = castType(arguments[0].type.get(), [&](const auto & type) { using DataType = std::decay_t; @@ -164,13 +165,13 @@ public: { if constexpr (allow_fixed_string) { - if (auto col = checkAndGetColumn(columns[arguments[0]].column.get())) + if (const auto * col = checkAndGetColumn(arguments[0].column.get())) { auto col_res = ColumnFixedString::create(col->getN()); auto & vec_res = col_res->getChars(); vec_res.resize(col->size() * col->getN()); FixedStringUnaryOperationImpl>::vector(col->getChars(), vec_res); - columns[result].column = std::move(col_res); + result_column = std::move(col_res); return true; } } @@ -180,13 +181,13 @@ public: using T0 = typename DataType::FieldType; if constexpr (allow_decimal) { - if (auto col = checkAndGetColumn>(columns[arguments[0]].column.get())) + if (auto col = checkAndGetColumn>(arguments[0].column.get())) { auto col_res = ColumnDecimal::ResultType>::create(0, type.getScale()); auto & vec_res = col_res->getData(); vec_res.resize(col->getData().size()); UnaryOperationImpl>::vector(col->getData(), vec_res); - columns[result].column = std::move(col_res); + result_column = std::move(col_res); return true; } } @@ -194,13 +195,13 @@ public: else { using T0 = typename DataType::FieldType; - if (auto col = checkAndGetColumn>(columns[arguments[0]].column.get())) + if (auto col = checkAndGetColumn>(arguments[0].column.get())) { auto col_res = ColumnVector::ResultType>::create(); auto & vec_res = col_res->getData(); vec_res.resize(col->getData().size()); UnaryOperationImpl>::vector(col->getData(), vec_res); - columns[result].column = std::move(col_res); + result_column = std::move(col_res); return true; } } @@ -209,6 +210,8 @@ public: }); if (!valid) throw Exception(getName() + "'s argument does not match the expected data type", ErrorCodes::LOGICAL_ERROR); + + return result_column; } #if USE_EMBEDDED_COMPILER diff --git a/src/Functions/FunctionUnixTimestamp64.h b/src/Functions/FunctionUnixTimestamp64.h index 86161e46c3d..2a5dee7734a 100644 --- a/src/Functions/FunctionUnixTimestamp64.h +++ b/src/Functions/FunctionUnixTimestamp64.h @@ -65,13 +65,12 @@ public: } } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { using SourceColumnType = typename SourceDataType::ColumnType; using ResultColumnType = typename ResultDataType::ColumnType; - const auto & src = columns[arguments[0]]; - auto & res = columns[result]; + const auto & src = arguments[0]; const auto & col = *src.column; const SourceColumnType * source_col_typed = checkAndGetColumn(col); @@ -80,16 +79,16 @@ public: + std::string(SourceDataType::family_name), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - res.column = res.type->createColumn(); + auto res_column = result_type->createColumn(); if (input_rows_count == 0) - return; + return res_column; - auto & result_data = assert_cast(res.column->assumeMutableRef()).getData(); + auto & result_data = assert_cast(res_column->assumeMutableRef()).getData(); result_data.reserve(source_col_typed->size()); const auto & source_data = source_col_typed->getData(); - const auto scale_diff = getScaleDiff(*checkAndGetDataType(src.type.get()), *checkAndGetDataType(res.type.get())); + const auto scale_diff = getScaleDiff(*checkAndGetDataType(src.type.get()), *checkAndGetDataType(result_type.get())); if (scale_diff == 0) { static_assert(sizeof(typename SourceColumnType::Container::value_type) == sizeof(typename ResultColumnType::Container::value_type)); @@ -114,6 +113,8 @@ public: for (const auto & v : source_data) result_data.push_back(static_cast(toDestValue(v) / scale_multiplier)); } + + return res_column; } private: diff --git a/src/Functions/FunctionsAES.cpp b/src/Functions/FunctionsAES.cpp new file mode 100644 index 00000000000..1588006eed8 --- /dev/null +++ b/src/Functions/FunctionsAES.cpp @@ -0,0 +1,63 @@ +#include + +#if USE_SSL + +#include +#include + +#include +#include + + +namespace DB +{ +namespace ErrorCodes +{ + extern const int OPENSSL_ERROR; +} +} + +namespace OpenSSLDetails +{ +void onError(std::string error_message) +{ + error_message += ". OpenSSL error code: " + std::to_string(ERR_get_error()); + throw DB::Exception(error_message, DB::ErrorCodes::OPENSSL_ERROR); +} + +StringRef foldEncryptionKeyInMySQLCompatitableMode(size_t cipher_key_size, const StringRef & key, std::array & folded_key) +{ + assert(cipher_key_size <= EVP_MAX_KEY_LENGTH); + memcpy(folded_key.data(), key.data, cipher_key_size); + + for (size_t i = cipher_key_size; i < key.size; ++i) + { + folded_key[i % cipher_key_size] ^= key.data[i]; + } + + return StringRef(folded_key.data(), cipher_key_size); +} + +const EVP_CIPHER * getCipherByName(const StringRef & cipher_name) +{ + const auto * evp_cipher = EVP_get_cipherbyname(cipher_name.data); + if (evp_cipher == nullptr) + { + // For some reasons following ciphers can't be found by name. + if (cipher_name == "aes-128-cfb128") + evp_cipher = EVP_aes_128_cfb128(); + else if (cipher_name == "aes-192-cfb128") + evp_cipher = EVP_aes_192_cfb128(); + else if (cipher_name == "aes-256-cfb128") + evp_cipher = EVP_aes_256_cfb128(); + } + + // NOTE: cipher obtained not via EVP_CIPHER_fetch() would cause extra work on each context reset + // with EVP_CIPHER_CTX_reset() or EVP_EncryptInit_ex(), but using EVP_CIPHER_fetch() + // causes data race, so we stick to the slower but safer alternative here. + return evp_cipher; +} + +} + +#endif diff --git a/src/Functions/FunctionsAES.h b/src/Functions/FunctionsAES.h new file mode 100644 index 00000000000..68d8b41407d --- /dev/null +++ b/src/Functions/FunctionsAES.h @@ -0,0 +1,688 @@ +#pragma once + +#if !defined(ARCADIA_BUILD) +# include +#endif + +#if USE_SSL +#include +#include +#include +#include +#include + +#include + +#include +#include + +#include +#include +#include + +#include + + +namespace DB +{ +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} +} + +namespace OpenSSLDetails +{ +[[noreturn]] void onError(std::string error_message); +StringRef foldEncryptionKeyInMySQLCompatitableMode(size_t cipher_key_size, const StringRef & key, std::array & folded_key); + +const EVP_CIPHER * getCipherByName(const StringRef & name); + +enum class CompatibilityMode +{ + MySQL, + OpenSSL +}; + +enum class CipherMode +{ + MySQLCompatibility, // with key folding + OpenSSLCompatibility, // just as regular openssl's enc application does (AEAD modes, like GCM and CCM are not supported) + RFC5116_AEAD_AES_GCM // AEAD GCM with custom IV length and tag (HMAC) appended to the ciphertext, see https://tools.ietf.org/html/rfc5116#section-5.1 +}; + + +template +struct KeyHolder +{ + inline StringRef setKey(size_t cipher_key_size, const StringRef & key) const + { + if (key.size != cipher_key_size) + throw DB::Exception(fmt::format("Invalid key size: {} expected {}", key.size, cipher_key_size), + DB::ErrorCodes::BAD_ARGUMENTS); + + return key; + } +}; + +template <> +struct KeyHolder +{ + inline StringRef setKey(size_t cipher_key_size, const StringRef & key) + { + if (key.size < cipher_key_size) + throw DB::Exception(fmt::format("Invalid key size: {} expected {}", key.size, cipher_key_size), + DB::ErrorCodes::BAD_ARGUMENTS); + + // MySQL does something fancy with the keys that are too long, + // ruining compatibility with OpenSSL and not improving security. + // But we have to do the same to be compatitable with MySQL. + // see https://github.com/mysql/mysql-server/blob/8.0/router/src/harness/src/my_aes_openssl.cc#L71 + // (my_aes_create_key function) + return foldEncryptionKeyInMySQLCompatitableMode(cipher_key_size, key, folded_key); + } + + ~KeyHolder() + { + OPENSSL_cleanse(folded_key.data(), folded_key.size()); + } + +private: + std::array folded_key; +}; + +template +inline void validateCipherMode(const EVP_CIPHER * evp_cipher) +{ + if constexpr (compatibility_mode == CompatibilityMode::MySQL) + { + switch (EVP_CIPHER_mode(evp_cipher)) + { + case EVP_CIPH_ECB_MODE: [[fallthrough]]; + case EVP_CIPH_CBC_MODE: [[fallthrough]]; + case EVP_CIPH_CFB_MODE: [[fallthrough]]; + case EVP_CIPH_OFB_MODE: + return; + } + } + else if constexpr (compatibility_mode == CompatibilityMode::OpenSSL) + { + switch (EVP_CIPHER_mode(evp_cipher)) + { + case EVP_CIPH_ECB_MODE: [[fallthrough]]; + case EVP_CIPH_CBC_MODE: [[fallthrough]]; + case EVP_CIPH_CFB_MODE: [[fallthrough]]; + case EVP_CIPH_OFB_MODE: [[fallthrough]]; + case EVP_CIPH_CTR_MODE: [[fallthrough]]; + case EVP_CIPH_GCM_MODE: + return; + } + } + + throw DB::Exception("Unsupported cipher mode " + std::string(EVP_CIPHER_name(evp_cipher)), DB::ErrorCodes::BAD_ARGUMENTS); +} + +template +inline void validateIV(const StringRef & iv_value, const size_t cipher_iv_size) +{ + // In MySQL mode we don't care if IV is longer than expected, only if shorter. + if ((mode == CipherMode::MySQLCompatibility && iv_value.size != 0 && iv_value.size < cipher_iv_size) + || (mode == CipherMode::OpenSSLCompatibility && iv_value.size != 0 && iv_value.size != cipher_iv_size)) + throw DB::Exception(fmt::format("Invalid IV size: {} expected {}", iv_value.size, cipher_iv_size), + DB::ErrorCodes::BAD_ARGUMENTS); +} + +} + +namespace DB +{ +template +class FunctionEncrypt : public IFunction +{ +public: + static constexpr OpenSSLDetails::CompatibilityMode compatibility_mode = Impl::compatibility_mode; + static constexpr auto name = Impl::name; + static FunctionPtr create(const Context &) { return std::make_shared(); } + +private: + using CipherMode = OpenSSLDetails::CipherMode; + + String getName() const override { return name; } + bool isVariadic() const override { return true; } + size_t getNumberOfArguments() const override { return 0; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {0}; } + bool useDefaultImplementationForConstants() const override { return true; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + auto optional_args = FunctionArgumentDescriptors{ + {"IV", isStringOrFixedString, nullptr, "Initialization vector binary string"}, + }; + + if constexpr (compatibility_mode == OpenSSLDetails::CompatibilityMode::OpenSSL) + { + optional_args.emplace_back(FunctionArgumentDescriptor{ + "AAD", isStringOrFixedString, nullptr, "Additional authenticated data binary string for GCM mode" + }); + } + + validateFunctionArgumentTypes(*this, arguments, + FunctionArgumentDescriptors{ + {"mode", isStringOrFixedString, isColumnConst, "encryption mode string"}, + {"input", nullptr, nullptr, "plaintext"}, + {"key", isStringOrFixedString, nullptr, "encryption key binary string"}, + }, + optional_args + ); + + return std::make_shared(); + } + + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + using namespace OpenSSLDetails; + + const auto mode = arguments[0].column->getDataAt(0); + + if (mode.size == 0 || !std::string_view(mode).starts_with("aes-")) + throw Exception("Invalid mode: " + mode.toString(), ErrorCodes::BAD_ARGUMENTS); + + const auto * evp_cipher = getCipherByName(mode); + if (evp_cipher == nullptr) + throw Exception("Invalid mode: " + mode.toString(), ErrorCodes::BAD_ARGUMENTS); + + const auto cipher_mode = EVP_CIPHER_mode(evp_cipher); + + const auto input_column = arguments[1].column; + const auto key_column = arguments[2].column; + + OpenSSLDetails::validateCipherMode(evp_cipher); + + ColumnPtr result_column; + if (arguments.size() <= 3) + result_column = doEncrypt(evp_cipher, input_rows_count, input_column, key_column, nullptr, nullptr); + else + { + const auto iv_column = arguments[3].column; + if (compatibility_mode != OpenSSLDetails::CompatibilityMode::MySQL && EVP_CIPHER_iv_length(evp_cipher) == 0) + throw Exception(mode.toString() + " does not support IV", ErrorCodes::BAD_ARGUMENTS); + + if (arguments.size() <= 4) + { + result_column = doEncrypt(evp_cipher, input_rows_count, input_column, key_column, iv_column, nullptr); + } + else + { + if (cipher_mode != EVP_CIPH_GCM_MODE) + throw Exception("AAD can be only set for GCM-mode", ErrorCodes::BAD_ARGUMENTS); + + const auto aad_column = arguments[4].column; + result_column = doEncrypt(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + } + + return result_column; + } + + template + static ColumnPtr doEncrypt(const EVP_CIPHER * evp_cipher, + size_t input_rows_count, + const InputColumnType & input_column, + const KeyColumnType & key_column, + const IvColumnType & iv_column, + const AadColumnType & aad_column) + { + if constexpr (compatibility_mode == OpenSSLDetails::CompatibilityMode::MySQL) + { + return doEncryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + else + { + if (EVP_CIPHER_mode(evp_cipher) == EVP_CIPH_GCM_MODE) + { + return doEncryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + else + { + return doEncryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + } + + return nullptr; + } + + template + static ColumnPtr doEncryptImpl(const EVP_CIPHER * evp_cipher, + size_t input_rows_count, + const InputColumnType & input_column, + const KeyColumnType & key_column, + [[maybe_unused]] const IvColumnType & iv_column, + [[maybe_unused]] const AadColumnType & aad_column) + { + using namespace OpenSSLDetails; + + auto evp_ctx_ptr = std::unique_ptr(EVP_CIPHER_CTX_new(), &EVP_CIPHER_CTX_free); + auto * evp_ctx = evp_ctx_ptr.get(); + + const auto block_size = static_cast(EVP_CIPHER_block_size(evp_cipher)); + const auto key_size = static_cast(EVP_CIPHER_key_length(evp_cipher)); + [[maybe_unused]] const auto iv_size = static_cast(EVP_CIPHER_iv_length(evp_cipher)); + const auto tag_size = 16; // https://tools.ietf.org/html/rfc5116#section-5.1 + + auto encrypted_result_column = ColumnString::create(); + auto & encrypted_result_column_data = encrypted_result_column->getChars(); + auto & encrypted_result_column_offsets = encrypted_result_column->getOffsets(); + + { + size_t resulting_size = 0; + // for modes with block_size > 1, plaintext is padded up to a block_size, + // which may result in allocating to much for block_size = 1. + // That may lead later to reading unallocated data from underlying PaddedPODArray + // due to assumption that it is safe to read up to 15 bytes past end. + const auto pad_to_next_block = block_size == 1 ? 0 : 1; + for (size_t r = 0; r < input_rows_count; ++r) + { + resulting_size += (input_column->getDataAt(r).size / block_size + pad_to_next_block) * block_size + 1; + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + resulting_size += tag_size; + } +#if defined(MEMORY_SANITIZER) + encrypted_result_column_data.resize_fill(resulting_size, 0xFF); +#else + encrypted_result_column_data.resize(resulting_size); +#endif + } + + auto * encrypted = encrypted_result_column_data.data(); + + KeyHolder key_holder; + + for (size_t r = 0; r < input_rows_count; ++r) + { + const auto key_value = key_holder.setKey(key_size, key_column->getDataAt(r)); + auto iv_value = StringRef{}; + if constexpr (!std::is_same_v>) + { + iv_value = iv_column->getDataAt(r); + } + + const auto input_value = input_column->getDataAt(r); + + if constexpr (mode != CipherMode::MySQLCompatibility) + { + // in GCM mode IV can be of arbitrary size (>0), IV is optional for other modes. + if (mode == CipherMode::RFC5116_AEAD_AES_GCM && iv_value.size == 0) + { + throw Exception("Invalid IV size " + std::to_string(iv_value.size) + " != expected size " + std::to_string(iv_size), + DB::ErrorCodes::BAD_ARGUMENTS); + } + + if (mode != CipherMode::RFC5116_AEAD_AES_GCM && key_value.size != key_size) + { + throw Exception("Invalid key size " + std::to_string(key_value.size) + " != expected size " + std::to_string(key_size), + DB::ErrorCodes::BAD_ARGUMENTS); + } + } + + // Avoid extra work on empty ciphertext/plaintext for some ciphers + if (!(input_value.size == 0 && block_size == 1 && mode != CipherMode::RFC5116_AEAD_AES_GCM)) + { + // 1: Init CTX + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + { + // 1.a.1: Init CTX with custom IV length and optionally with AAD + if (EVP_EncryptInit_ex(evp_ctx, evp_cipher, nullptr, nullptr, nullptr) != 1) + onError("Failed to initialize encryption context with cipher"); + + if (EVP_CIPHER_CTX_ctrl(evp_ctx, EVP_CTRL_AEAD_SET_IVLEN, iv_value.size, nullptr) != 1) + onError("Failed to set custom IV length to " + std::to_string(iv_value.size)); + + if (EVP_EncryptInit_ex(evp_ctx, nullptr, nullptr, + reinterpret_cast(key_value.data), + reinterpret_cast(iv_value.data)) != 1) + onError("Failed to set key and IV"); + + // 1.a.2 Set AAD + if constexpr (!std::is_same_v>) + { + const auto aad_data = aad_column->getDataAt(r); + int tmp_len = 0; + if (aad_data.size != 0 && EVP_EncryptUpdate(evp_ctx, nullptr, &tmp_len, + reinterpret_cast(aad_data.data), aad_data.size) != 1) + onError("Failed to set AAD data"); + } + } + else + { + // 1.b: Init CTX + validateIV(iv_value, iv_size); + + if (EVP_EncryptInit_ex(evp_ctx, evp_cipher, nullptr, + reinterpret_cast(key_value.data), + reinterpret_cast(iv_value.data)) != 1) + onError("Failed to initialize cipher context"); + } + + int output_len = 0; + // 2: Feed the data to the cipher + if (EVP_EncryptUpdate(evp_ctx, + reinterpret_cast(encrypted), &output_len, + reinterpret_cast(input_value.data), static_cast(input_value.size)) != 1) + onError("Failed to encrypt"); + encrypted += output_len; + + // 3: retrieve encrypted data (ciphertext) + if (EVP_EncryptFinal_ex(evp_ctx, + reinterpret_cast(encrypted), &output_len) != 1) + onError("Failed to fetch ciphertext"); + encrypted += output_len; + + // 4: optionally retrieve a tag and append it to the ciphertext (RFC5116): + // https://tools.ietf.org/html/rfc5116#section-5.1 + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + { + if (EVP_CIPHER_CTX_ctrl(evp_ctx, EVP_CTRL_AEAD_GET_TAG, tag_size, encrypted) != 1) + onError("Failed to retrieve GCM tag"); + encrypted += tag_size; + } + } + + *encrypted = '\0'; + ++encrypted; + + encrypted_result_column_offsets.push_back(encrypted - encrypted_result_column_data.data()); + } + + // in case of block size of 1, we overestimate buffer required for encrypted data, fix it up. + if (!encrypted_result_column_offsets.empty() && encrypted_result_column_data.size() > encrypted_result_column_offsets.back()) + { + encrypted_result_column_data.resize(encrypted_result_column_offsets.back()); + } + + encrypted_result_column->validate(); + return encrypted_result_column; + } +}; + + +/// AES_decrypt(string, key, block_mode[, init_vector]) +template +class FunctionDecrypt : public IFunction +{ +public: + static constexpr OpenSSLDetails::CompatibilityMode compatibility_mode = Impl::compatibility_mode; + static constexpr auto name = Impl::name; + static FunctionPtr create(const Context &) { return std::make_shared(); } + +private: + using CipherMode = OpenSSLDetails::CipherMode; + + String getName() const override { return name; } + bool isVariadic() const override { return true; } + size_t getNumberOfArguments() const override { return 0; } + ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {0}; } + bool useDefaultImplementationForConstants() const override { return true; } + + DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override + { + auto optional_args = FunctionArgumentDescriptors{ + {"IV", isStringOrFixedString, nullptr, "Initialization vector binary string"}, + }; + + if constexpr (compatibility_mode == OpenSSLDetails::CompatibilityMode::OpenSSL) + { + optional_args.emplace_back(FunctionArgumentDescriptor{ + "AAD", isStringOrFixedString, nullptr, "Additional authenticated data binary string for GCM mode" + }); + } + + validateFunctionArgumentTypes(*this, arguments, + FunctionArgumentDescriptors{ + {"mode", isStringOrFixedString, isColumnConst, "decryption mode string"}, + {"input", nullptr, nullptr, "ciphertext"}, + {"key", isStringOrFixedString, nullptr, "decryption key binary string"}, + }, + optional_args + ); + + return std::make_shared(); + } + + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + { + using namespace OpenSSLDetails; + + const auto mode = arguments[0].column->getDataAt(0); + if (mode.size == 0 || !std::string_view(mode).starts_with("aes-")) + throw Exception("Invalid mode: " + mode.toString(), ErrorCodes::BAD_ARGUMENTS); + + const auto * evp_cipher = getCipherByName(mode); + if (evp_cipher == nullptr) + throw Exception("Invalid mode: " + mode.toString(), ErrorCodes::BAD_ARGUMENTS); + + OpenSSLDetails::validateCipherMode(evp_cipher); + + const auto input_column = arguments[1].column; + const auto key_column = arguments[2].column; + + ColumnPtr result_column; + if (arguments.size() <= 3) + result_column = doDecrypt(evp_cipher, input_rows_count, input_column, key_column, nullptr, nullptr); + else + { + const auto iv_column = arguments[3].column; + if (compatibility_mode != OpenSSLDetails::CompatibilityMode::MySQL && EVP_CIPHER_iv_length(evp_cipher) == 0) + throw Exception(mode.toString() + " does not support IV", ErrorCodes::BAD_ARGUMENTS); + + if (arguments.size() <= 4) + { + result_column = doDecrypt(evp_cipher, input_rows_count, input_column, key_column, iv_column, nullptr); + } + else + { + if (EVP_CIPHER_mode(evp_cipher) != EVP_CIPH_GCM_MODE) + throw Exception("AAD can be only set for GCM-mode", ErrorCodes::BAD_ARGUMENTS); + + const auto aad_column = arguments[4].column; + result_column = doDecrypt(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + } + + return result_column; + } + + template + static ColumnPtr doDecrypt(const EVP_CIPHER * evp_cipher, + size_t input_rows_count, + const InputColumnType & input_column, + const KeyColumnType & key_column, + const IvColumnType & iv_column, + const AadColumnType & aad_column) + { + if constexpr (compatibility_mode == OpenSSLDetails::CompatibilityMode::MySQL) + { + return doDecryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + else + { + const auto cipher_mode = EVP_CIPHER_mode(evp_cipher); + if (cipher_mode == EVP_CIPH_GCM_MODE) + { + return doDecryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + else + { + return doDecryptImpl(evp_cipher, input_rows_count, input_column, key_column, iv_column, aad_column); + } + } + + return nullptr; + } + + template + static ColumnPtr doDecryptImpl(const EVP_CIPHER * evp_cipher, + size_t input_rows_count, + const InputColumnType & input_column, + const KeyColumnType & key_column, + [[maybe_unused]] const IvColumnType & iv_column, + [[maybe_unused]] const AadColumnType & aad_column) + { + using namespace OpenSSLDetails; + + auto evp_ctx_ptr = std::unique_ptr(EVP_CIPHER_CTX_new(), &EVP_CIPHER_CTX_free); + auto * evp_ctx = evp_ctx_ptr.get(); + + [[maybe_unused]] const auto block_size = static_cast(EVP_CIPHER_block_size(evp_cipher)); + [[maybe_unused]] const auto iv_size = static_cast(EVP_CIPHER_iv_length(evp_cipher)); + const auto key_size = static_cast(EVP_CIPHER_key_length(evp_cipher)); + const auto tag_size = 16; // https://tools.ietf.org/html/rfc5116#section-5.1 + + auto decrypted_result_column = ColumnString::create(); + auto & decrypted_result_column_data = decrypted_result_column->getChars(); + auto & decrypted_result_column_offsets = decrypted_result_column->getOffsets(); + + { + size_t resulting_size = 0; + for (size_t r = 0; r < input_rows_count; ++r) + { + resulting_size += input_column->getDataAt(r).size + 1; + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + resulting_size -= tag_size; + } + +#if defined(MEMORY_SANITIZER) + // Pre-fill result column with values to prevent MSAN from dropping dead on + // aes-X-ecb mode with "WARNING: MemorySanitizer: use-of-uninitialized-value". + // This is most likely to be caused by the underlying assembler implementation: + // see crypto/aes/aesni-x86_64.s, function aesni_ecb_encrypt + // which msan seems to fail instrument correctly. + decrypted_result_column_data.resize_fill(resulting_size, 0xFF); +#else + decrypted_result_column_data.resize(resulting_size); +#endif + } + auto * decrypted = decrypted_result_column_data.data(); + + KeyHolder key_holder; + for (size_t r = 0; r < input_rows_count; ++r) + { + // 0: prepare key if required + auto key_value = key_holder.setKey(key_size, key_column->getDataAt(r)); + auto iv_value = StringRef{}; + if constexpr (!std::is_same_v>) + { + iv_value = iv_column->getDataAt(r); + } + + auto input_value = input_column->getDataAt(r); + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + { + // empty plaintext results in empty ciphertext + tag, means there should be at least tag_size bytes. + if (input_value.size < tag_size) + throw Exception(fmt::format("Encrypted data is too short: only {} bytes, " + "should contain at least {} bytes of a tag.", + input_value.size, block_size, tag_size), ErrorCodes::BAD_ARGUMENTS); + input_value.size -= tag_size; + } + + if constexpr (mode != CipherMode::MySQLCompatibility) + { + // in GCM mode IV can be of arbitrary size (>0), for other modes IV is optional. + if (mode == CipherMode::RFC5116_AEAD_AES_GCM && iv_value.size == 0) + { + throw Exception("Invalid IV size " + std::to_string(iv_value.size) + " != expected size " + std::to_string(iv_size), + DB::ErrorCodes::BAD_ARGUMENTS); + } + + if (key_value.size != key_size) + { + throw Exception("Invalid key size " + std::to_string(key_value.size) + " != expected size " + std::to_string(key_size), + DB::ErrorCodes::BAD_ARGUMENTS); + } + } + + // Avoid extra work on empty ciphertext/plaintext for some ciphers + if (!(input_value.size == 0 && block_size == 1 && mode != CipherMode::RFC5116_AEAD_AES_GCM)) + { + // 1: Init CTX + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + { + if (EVP_DecryptInit_ex(evp_ctx, evp_cipher, nullptr, nullptr, nullptr) != 1) + onError("Failed to initialize cipher context 1"); + + // 1.a.1 : Set custom IV length + if (EVP_CIPHER_CTX_ctrl(evp_ctx, EVP_CTRL_AEAD_SET_IVLEN, iv_value.size, nullptr) != 1) + onError("Failed to set custom IV length to " + std::to_string(iv_value.size)); + + // 1.a.1 : Init CTX with key and IV + if (EVP_DecryptInit_ex(evp_ctx, nullptr, nullptr, + reinterpret_cast(key_value.data), + reinterpret_cast(iv_value.data)) != 1) + onError("Failed to set key and IV"); + + // 1.a.2: Set AAD if present + if constexpr (!std::is_same_v>) + { + const auto aad_data = aad_column->getDataAt(r); + int tmp_len = 0; + if (aad_data.size != 0 && EVP_DecryptUpdate(evp_ctx, nullptr, &tmp_len, + reinterpret_cast(aad_data.data), aad_data.size) != 1) + onError("Failed to sed AAD data"); + } + } + else + { + // 1.b: Init CTX + validateIV(iv_value, iv_size); + + if (EVP_DecryptInit_ex(evp_ctx, evp_cipher, nullptr, + reinterpret_cast(key_value.data), + reinterpret_cast(iv_value.data)) != 1) + onError("Failed to initialize cipher context"); + } + + // 2: Feed the data to the cipher + int output_len = 0; + if (EVP_DecryptUpdate(evp_ctx, + reinterpret_cast(decrypted), &output_len, + reinterpret_cast(input_value.data), static_cast(input_value.size)) != 1) + onError("Failed to decrypt"); + decrypted += output_len; + + // 3: optionally get tag from the ciphertext (RFC5116) and feed it to the context + if constexpr (mode == CipherMode::RFC5116_AEAD_AES_GCM) + { + void * tag = const_cast(reinterpret_cast(input_value.data + input_value.size)); + if (EVP_CIPHER_CTX_ctrl(evp_ctx, EVP_CTRL_AEAD_SET_TAG, tag_size, tag) != 1) + onError("Failed to set tag"); + } + + // 4: retrieve encrypted data (ciphertext) + if (EVP_DecryptFinal_ex(evp_ctx, + reinterpret_cast(decrypted), &output_len) != 1) + onError("Failed to decrypt"); + decrypted += output_len; + } + + *decrypted = '\0'; + ++decrypted; + + decrypted_result_column_offsets.push_back(decrypted - decrypted_result_column_data.data()); + + } + + // in case we overestimate buffer required for decrypted data, fix it up. + if (!decrypted_result_column_offsets.empty() && decrypted_result_column_data.size() > decrypted_result_column_offsets.back()) + { + decrypted_result_column_data.resize(decrypted_result_column_offsets.back()); + } + + decrypted_result_column->validate(); + return decrypted_result_column; + } +}; + +} + + +#endif diff --git a/src/Functions/FunctionsBitmap.h b/src/Functions/FunctionsBitmap.h index 9d15bf2ba28..ec43ae6351f 100644 --- a/src/Functions/FunctionsBitmap.h +++ b/src/Functions/FunctionsBitmap.h @@ -104,7 +104,7 @@ public: if (arguments[0]->onlyNull()) return arguments[0]; - auto array_type = typeid_cast(arguments[0].get()); + const auto * array_type = typeid_cast(arguments[0].get()); if (!array_type) throw Exception( "First argument for function " + getName() + " must be an array but it has type " + arguments[0]->getName() + ".", @@ -122,23 +122,23 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /* input_rows_count */) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /* input_rows_count */) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); - auto array_type = typeid_cast(from_type); - auto nested_type = array_type->getNestedType(); + const IDataType * from_type = arguments[0].type.get(); + const auto * array_type = typeid_cast(from_type); + const auto & nested_type = array_type->getNestedType(); DataTypes argument_types = {nested_type}; WhichDataType which(nested_type); if (which.isUInt8()) - executeBitmapData(columns, argument_types, arguments, result); + return executeBitmapData(argument_types, arguments); else if (which.isUInt16()) - executeBitmapData(columns, argument_types, arguments, result); + return executeBitmapData(argument_types, arguments); else if (which.isUInt32()) - executeBitmapData(columns, argument_types, arguments, result); + return executeBitmapData(argument_types, arguments); else if (which.isUInt64()) - executeBitmapData(columns, argument_types, arguments, result); + return executeBitmapData(argument_types, arguments); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); @@ -146,11 +146,11 @@ public: private: template - void executeBitmapData(ColumnsWithTypeAndName & columns, DataTypes & argument_types, const ColumnNumbers & arguments, size_t result) const + ColumnPtr executeBitmapData(DataTypes & argument_types, ColumnsWithTypeAndName & arguments) const { // input data - const ColumnArray * array = typeid_cast(columns[arguments[0]].column.get()); - ColumnPtr mapped = array->getDataPtr(); + const ColumnArray * array = typeid_cast(arguments[0].column.get()); + const ColumnPtr & mapped = array->getDataPtr(); const ColumnArray::Offsets & offsets = array->getOffsets(); const ColumnVector * column = checkAndGetColumn>(&*mapped); const typename ColumnVector::Container & input_data = column->getData(); @@ -174,7 +174,7 @@ private: bitmap_data.rbs.add(input_data[pos]); } } - columns[result].column = std::move(col_to); + return col_to; } }; @@ -207,32 +207,32 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { // input data - const auto & return_type = columns[result].type; + const auto & return_type = result_type; auto res_ptr = return_type->createColumn(); ColumnArray & res = assert_cast(*res_ptr); IColumn & res_data = res.getData(); ColumnArray::Offsets & res_offsets = res.getOffsets(); - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, input_rows_count, res_data, res_offsets); + executeIntType(arguments, input_rows_count, res_data, res_offsets); else if (which.isUInt16()) - executeIntType(columns, arguments, input_rows_count, res_data, res_offsets); + executeIntType(arguments, input_rows_count, res_data, res_offsets); else if (which.isUInt32()) - executeIntType(columns, arguments, input_rows_count, res_data, res_offsets); + executeIntType(arguments, input_rows_count, res_data, res_offsets); else if (which.isUInt64()) - executeIntType(columns, arguments, input_rows_count, res_data, res_offsets); + executeIntType(arguments, input_rows_count, res_data, res_offsets); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - columns[result].column = std::move(res_ptr); + return res_ptr; } private: @@ -240,11 +240,11 @@ private: template void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t input_rows_count, IColumn & res_data_col, ColumnArray::Offsets & res_offsets) + ColumnsWithTypeAndName & arguments, size_t input_rows_count, IColumn & res_data_col, ColumnArray::Offsets & res_offsets) const { const ColumnAggregateFunction * column - = typeid_cast(columns[arguments[0]].column.get()); + = typeid_cast(arguments[0].column.get()); PaddedPODArray & res_data = typeid_cast &>(res_data_col).getData(); ColumnArray::Offset res_offset = 0; @@ -282,13 +282,13 @@ public: "First argument for function " + getName() + " must be a bitmap but it has type " + arguments[0]->getName() + ".", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - auto arg_type1 = typeid_cast *>(arguments[1].get()); + const auto * arg_type1 = typeid_cast *>(arguments[1].get()); if (!(arg_type1)) throw Exception( "Second argument for function " + getName() + " must be UInt32 but it has type " + arguments[1]->getName() + ".", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - auto arg_type2 = typeid_cast *>(arguments[1].get()); + const auto * arg_type2 = typeid_cast *>(arguments[1].get()); if (!(arg_type2)) throw Exception( "Third argument for function " + getName() + " must be UInt32 but it has type " + arguments[2]->getName() + ".", @@ -299,19 +299,19 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt16()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt32()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt64()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); @@ -321,9 +321,7 @@ private: using ToType = UInt64; template - void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) - const + ColumnPtr executeIntType(ColumnsWithTypeAndName & arguments, size_t input_rows_count) const { const IColumn * column_ptrs[3]; bool is_column_const[3]; @@ -333,7 +331,7 @@ private: for (size_t i = 0; i < 3; ++i) { - column_ptrs[i] = columns[arguments[i]].column.get(); + column_ptrs[i] = arguments[i].column.get(); is_column_const[i] = isColumnConst(*column_ptrs[i]); } if (is_column_const[0]) @@ -367,7 +365,7 @@ private: = *reinterpret_cast *>(col_to->getData()[i]); Impl::apply(bitmap_data_0, range_start, range_end, bitmap_data_2); } - columns[result].column = std::move(col_to); + return col_to; } }; @@ -435,19 +433,19 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt16()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt32()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else if (which.isUInt64()) - executeIntType(columns, arguments, result, input_rows_count); + return executeIntType(arguments, input_rows_count); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); @@ -457,8 +455,7 @@ private: using ToType = UInt64; template - void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const + ColumnPtr executeIntType(ColumnsWithTypeAndName & arguments, size_t input_rows_count) const { const IColumn * column_ptrs[3]; bool is_column_const[3]; @@ -468,7 +465,7 @@ private: for (size_t i = 0; i < 3; ++i) { - column_ptrs[i] = columns[arguments[i]].column.get(); + column_ptrs[i] = arguments[i].column.get(); is_column_const[i] = isColumnConst(*column_ptrs[i]); } if (is_column_const[0]) @@ -485,7 +482,7 @@ private: array = typeid_cast(typeid_cast(column_ptrs[1])->getDataColumnPtr().get()); else { - array = typeid_cast(columns[arguments[1]].column.get()); + array = typeid_cast(arguments[1].column.get()); } const ColumnArray::Offsets & from_offsets = array->getOffsets(); const ColumnVector::Container & from_container = typeid_cast *>(&array->getData())->getData(); @@ -493,7 +490,7 @@ private: if (is_column_const[2]) array = typeid_cast(typeid_cast(column_ptrs[2])->getDataColumnPtr().get()); else - array = typeid_cast(columns[arguments[2]].column.get()); + array = typeid_cast(arguments[2].column.get()); const ColumnArray::Offsets & to_offsets = array->getOffsets(); const ColumnVector::Container & to_container = typeid_cast *>(&array->getData())->getData(); @@ -538,7 +535,7 @@ private: bitmap_data_2.rbs.merge(bitmap_data_0.rbs); bitmap_data_2.rbs.rb_replace(&from_container[from_start], &to_container[to_start], from_end - from_start); } - columns[result].column = std::move(col_to); + return col_to; } }; @@ -558,7 +555,7 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - auto bitmap_type = typeid_cast(arguments[0].get()); + const auto * bitmap_type = typeid_cast(arguments[0].get()); if (!(bitmap_type && bitmap_type->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "First argument for function " + getName() + " must be a bitmap but it has type " + arguments[0]->getName() + ".", @@ -568,27 +565,27 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { auto col_to = ColumnVector::create(input_rows_count); typename ColumnVector::Container & vec_to = col_to->getData(); - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt16()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt32()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt64()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - columns[result].column = std::move(col_to); + return col_to; } private: @@ -596,10 +593,10 @@ private: template void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const + ColumnsWithTypeAndName & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const { const ColumnAggregateFunction * column - = typeid_cast(columns[arguments[0]].column.get()); + = typeid_cast(arguments[0].column.get()); for (size_t i = 0; i < input_rows_count; ++i) { const AggregateFunctionGroupBitmapData & bitmap_data @@ -722,12 +719,12 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - auto bitmap_type0 = typeid_cast(arguments[0].get()); + const auto * bitmap_type0 = typeid_cast(arguments[0].get()); if (!(bitmap_type0 && bitmap_type0->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "First argument for function " + getName() + " must be a bitmap but it has type " + arguments[0]->getName() + ".", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - auto arg_type1 = typeid_cast *>(arguments[1].get()); + const auto * arg_type1 = typeid_cast *>(arguments[1].get()); if (!(arg_type1)) throw Exception( "Second argument for function " + getName() + " must be UInt32 but it has type " + arguments[1]->getName() + ".", @@ -738,33 +735,33 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { auto col_to = ColumnVector::create(input_rows_count); typename ColumnVector::Container & vec_to = col_to->getData(); - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt16()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt32()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt64()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - columns[result].column = std::move(col_to); + return col_to; } private: template void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const + ColumnsWithTypeAndName & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const { const IColumn * column_ptrs[2]; bool is_column_const[2]; @@ -773,7 +770,7 @@ private: for (size_t i = 0; i < 2; ++i) { - column_ptrs[i] = columns[arguments[i]].column.get(); + column_ptrs[i] = arguments[i].column.get(); is_column_const[i] = isColumnConst(*column_ptrs[i]); } if (is_column_const[0]) @@ -812,13 +809,13 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - auto bitmap_type0 = typeid_cast(arguments[0].get()); + const auto * bitmap_type0 = typeid_cast(arguments[0].get()); if (!(bitmap_type0 && bitmap_type0->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "First argument for function " + getName() + " must be a bitmap but it has type " + arguments[0]->getName() + ".", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - auto bitmap_type1 = typeid_cast(arguments[1].get()); + const auto * bitmap_type1 = typeid_cast(arguments[1].get()); if (!(bitmap_type1 && bitmap_type1->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "Second argument for function " + getName() + " must be a bitmap but it has type " + arguments[1]->getName() + ".", @@ -835,46 +832,46 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { auto col_to = ColumnVector::create(input_rows_count); typename ColumnVector::Container & vec_to = col_to->getData(); - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt16()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt32()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else if (which.isUInt64()) - executeIntType(columns, arguments, input_rows_count, vec_to); + executeIntType(arguments, input_rows_count, vec_to); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - columns[result].column = std::move(col_to); + return col_to; } private: template void executeIntType( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const + ColumnsWithTypeAndName & arguments, size_t input_rows_count, typename ColumnVector::Container & vec_to) const { const ColumnAggregateFunction * column_ptrs[2]; bool is_column_const[2]; for (size_t i = 0; i < 2; ++i) { - if (auto argument_column_const = checkAndGetColumn(columns[arguments[i]].column.get())) + if (const auto * argument_column_const = checkAndGetColumn(arguments[i].column.get())) { column_ptrs[i] = typeid_cast(argument_column_const->getDataColumnPtr().get()); is_column_const[i] = true; } else { - column_ptrs[i] = typeid_cast(columns[arguments[i]].column.get()); + column_ptrs[i] = typeid_cast(arguments[i].column.get()); is_column_const[i] = false; } } @@ -947,13 +944,13 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - auto bitmap_type0 = typeid_cast(arguments[0].get()); + const auto * bitmap_type0 = typeid_cast(arguments[0].get()); if (!(bitmap_type0 && bitmap_type0->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "First argument for function " + getName() + " must be a bitmap but it has type " + arguments[0]->getName() + ".", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - auto bitmap_type1 = typeid_cast(arguments[1].get()); + const auto * bitmap_type1 = typeid_cast(arguments[1].get()); if (!(bitmap_type1 && bitmap_type1->getFunctionName() == AggregateFunctionGroupBitmapData::name())) throw Exception( "Second argument for function " + getName() + " must be a bitmap but it has type " + arguments[1]->getName() + ".", @@ -970,19 +967,19 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); const DataTypeAggregateFunction * aggr_type = typeid_cast(from_type); WhichDataType which(aggr_type->getArgumentsDataTypes()[0]); if (which.isUInt8()) - executeBitmapData(columns, arguments, result, input_rows_count); + return executeBitmapData(arguments, input_rows_count); else if (which.isUInt16()) - executeBitmapData(columns, arguments, result, input_rows_count); + return executeBitmapData(arguments, input_rows_count); else if (which.isUInt32()) - executeBitmapData(columns, arguments, result, input_rows_count); + return executeBitmapData(arguments, input_rows_count); else if (which.isUInt64()) - executeBitmapData(columns, arguments, result, input_rows_count); + return executeBitmapData(arguments, input_rows_count); else throw Exception( "Unexpected type " + from_type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); @@ -990,20 +987,20 @@ public: private: template - void executeBitmapData(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const + ColumnPtr executeBitmapData(ColumnsWithTypeAndName & arguments, size_t input_rows_count) const { const ColumnAggregateFunction * column_ptrs[2]; bool is_column_const[2]; for (size_t i = 0; i < 2; ++i) { - if (auto argument_column_const = typeid_cast(columns[arguments[i]].column.get())) + if (const auto * argument_column_const = typeid_cast(arguments[i].column.get())) { column_ptrs[i] = typeid_cast(argument_column_const->getDataColumnPtr().get()); is_column_const[i] = true; } else { - column_ptrs[i] = typeid_cast(columns[arguments[i]].column.get()); + column_ptrs[i] = typeid_cast(arguments[i].column.get()); is_column_const[i] = false; } } @@ -1026,7 +1023,7 @@ private: = *reinterpret_cast *>(data_ptr_1); Impl::apply(bitmap_data_1, bitmap_data_2); } - columns[result].column = std::move(col_to); + return col_to; } }; diff --git a/src/Functions/FunctionsCoding.h b/src/Functions/FunctionsCoding.h index 08ea1c8418a..6ae75318f72 100644 --- a/src/Functions/FunctionsCoding.h +++ b/src/Functions/FunctionsCoding.h @@ -76,7 +76,7 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - const auto ptr = checkAndGetDataType(arguments[0].get()); + const auto * ptr = checkAndGetDataType(arguments[0].get()); if (!ptr || ptr->getN() != IPV6_BINARY_LENGTH) throw Exception("Illegal type " + arguments[0]->getName() + " of argument of function " + getName() + @@ -88,12 +88,12 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const auto & col_type_name = columns[arguments[0]]; + const auto & col_type_name = arguments[0]; const ColumnPtr & column = col_type_name.column; - if (const auto col_in = checkAndGetColumn(column.get())) + if (const auto * col_in = checkAndGetColumn(column.get())) { if (col_in->getN() != IPV6_BINARY_LENGTH) throw Exception("Illegal type " + col_type_name.type->getName() + @@ -112,8 +112,8 @@ public: vec_res.resize(size * (IPV6_MAX_TEXT_LENGTH + 1)); offsets_res.resize(size); - auto begin = reinterpret_cast(vec_res.data()); - auto pos = begin; + auto * begin = reinterpret_cast(vec_res.data()); + auto * pos = begin; for (size_t offset = 0, i = 0; offset < vec_in.size(); offset += IPV6_BINARY_LENGTH, ++i) { @@ -123,10 +123,10 @@ public: vec_res.resize(pos - begin); - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -145,7 +145,7 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - const auto ptr = checkAndGetDataType(arguments[0].get()); + const auto * ptr = checkAndGetDataType(arguments[0].get()); if (!ptr || ptr->getN() != IPV6_BINARY_LENGTH) throw Exception("Illegal type " + arguments[0]->getName() + " of argument 1 of function " + getName() + @@ -168,17 +168,17 @@ public: bool useDefaultImplementationForConstants() const override { return true; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1, 2}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const auto & col_type_name = columns[arguments[0]]; + const auto & col_type_name = arguments[0]; const ColumnPtr & column = col_type_name.column; - const auto & col_ipv6_zeroed_tail_bytes_type = columns[arguments[1]]; + const auto & col_ipv6_zeroed_tail_bytes_type = arguments[1]; const auto & col_ipv6_zeroed_tail_bytes = col_ipv6_zeroed_tail_bytes_type.column; - const auto & col_ipv4_zeroed_tail_bytes_type = columns[arguments[2]]; + const auto & col_ipv4_zeroed_tail_bytes_type = arguments[2]; const auto & col_ipv4_zeroed_tail_bytes = col_ipv4_zeroed_tail_bytes_type.column; - if (const auto col_in = checkAndGetColumn(column.get())) + if (const auto * col_in = checkAndGetColumn(column.get())) { if (col_in->getN() != IPV6_BINARY_LENGTH) throw Exception("Illegal type " + col_type_name.type->getName() + @@ -187,7 +187,7 @@ public: ", expected FixedString(" + toString(IPV6_BINARY_LENGTH) + ")", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - const auto ipv6_zeroed_tail_bytes = checkAndGetColumnConst>(col_ipv6_zeroed_tail_bytes.get()); + const auto * ipv6_zeroed_tail_bytes = checkAndGetColumnConst>(col_ipv6_zeroed_tail_bytes.get()); if (!ipv6_zeroed_tail_bytes) throw Exception("Illegal type " + col_ipv6_zeroed_tail_bytes_type.type->getName() + " of argument 2 of function " + getName(), @@ -199,7 +199,7 @@ public: " of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - const auto ipv4_zeroed_tail_bytes = checkAndGetColumnConst>(col_ipv4_zeroed_tail_bytes.get()); + const auto * ipv4_zeroed_tail_bytes = checkAndGetColumnConst>(col_ipv4_zeroed_tail_bytes.get()); if (!ipv4_zeroed_tail_bytes) throw Exception("Illegal type " + col_ipv4_zeroed_tail_bytes_type.type->getName() + " of argument 3 of function " + getName(), @@ -221,12 +221,12 @@ public: vec_res.resize(size * (IPV6_MAX_TEXT_LENGTH + 1)); offsets_res.resize(size); - auto begin = reinterpret_cast(vec_res.data()); - auto pos = begin; + auto * begin = reinterpret_cast(vec_res.data()); + auto * pos = begin; for (size_t offset = 0, i = 0; offset < vec_in.size(); offset += IPV6_BINARY_LENGTH, ++i) { - const auto address = &vec_in[offset]; + const auto * address = &vec_in[offset]; UInt8 zeroed_tail_bytes_count = isIPv4Mapped(address) ? ipv4_zeroed_tail_bytes_count : ipv6_zeroed_tail_bytes_count; cutAddress(reinterpret_cast(address), pos, zeroed_tail_bytes_count); offsets_res[i] = pos - begin; @@ -234,22 +234,22 @@ public: vec_res.resize(pos - begin); - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } private: - bool isIPv4Mapped(const UInt8 * address) const + static bool isIPv4Mapped(const UInt8 * address) { return (unalignedLoad(address) == 0) && ((unalignedLoad(address + 8) & 0x00000000FFFFFFFFull) == 0x00000000FFFF0000ull); } - void cutAddress(const unsigned char * address, char *& dst, UInt8 zeroed_tail_bytes_count) const + static void cutAddress(const unsigned char * address, char *& dst, UInt8 zeroed_tail_bytes_count) { formatIPv6(address, dst, zeroed_tail_bytes_count); } @@ -277,11 +277,11 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; - if (const auto col_in = checkAndGetColumn(column.get())) + if (const auto * col_in = checkAndGetColumn(column.get())) { auto col_res = ColumnFixedString::create(IPV6_BINARY_LENGTH); @@ -301,10 +301,10 @@ public: src_offset = offsets_src[i]; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -339,9 +339,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; if (const ColumnUInt32 * col = typeid_cast(column.get())) { @@ -365,10 +365,10 @@ public: vec_res.resize(pos - begin); - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -407,9 +407,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; if (const ColumnString * col = checkAndGetColumn(column.get())) { @@ -428,10 +428,10 @@ public: prev_offset = offsets_src[i]; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -460,12 +460,12 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const auto & col_type_name = columns[arguments[0]]; + const auto & col_type_name = arguments[0]; const ColumnPtr & column = col_type_name.column; - if (const auto col_in = typeid_cast(column.get())) + if (const auto * col_in = typeid_cast(column.get())) { auto col_res = ColumnFixedString::create(IPV6_BINARY_LENGTH); @@ -477,16 +477,16 @@ public: for (size_t out_offset = 0, i = 0; out_offset < vec_res.size(); out_offset += IPV6_BINARY_LENGTH, ++i) mapIPv4ToIPv6(vec_in[i], &vec_res[out_offset]); - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } private: - void mapIPv4ToIPv6(UInt32 in, UInt8 * buf) const + static void mapIPv4ToIPv6(UInt32 in, UInt8 * buf) { unalignedStore(buf, 0); unalignedStore(buf + 8, 0x00000000FFFF0000ull | (static_cast(ntohl(in)) << 32)); @@ -578,9 +578,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; if (const ColumnUInt64 * col = typeid_cast(column.get())) { @@ -602,10 +602,10 @@ public: offsets_res[i] = current_offset; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -688,9 +688,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; if (const ColumnString * col = checkAndGetColumn(column.get())) { @@ -716,10 +716,10 @@ public: prev_offset = current_offset; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -743,7 +743,7 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - const auto ptr = checkAndGetDataType(arguments[0].get()); + const auto * ptr = checkAndGetDataType(arguments[0].get()); if (!ptr || ptr->getN() != uuid_bytes_length) throw Exception("Illegal type " + arguments[0]->getName() + " of argument of function " + getName() + @@ -755,12 +755,12 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnWithTypeAndName & col_type_name = columns[arguments[0]]; + const ColumnWithTypeAndName & col_type_name = arguments[0]; const ColumnPtr & column = col_type_name.column; - if (const auto col_in = checkAndGetColumn(column.get())) + if (const auto * col_in = checkAndGetColumn(column.get())) { if (col_in->getN() != uuid_bytes_length) throw Exception("Illegal type " + col_type_name.type->getName() + @@ -792,10 +792,10 @@ public: offsets_res[i] = dst_offset; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -844,7 +844,7 @@ public: /// String or FixedString(36) if (!isString(arguments[0])) { - const auto ptr = checkAndGetDataType(arguments[0].get()); + const auto * ptr = checkAndGetDataType(arguments[0].get()); if (!ptr || ptr->getN() != uuid_text_length) throw Exception("Illegal type " + arguments[0]->getName() + " of argument of function " + getName() + @@ -857,12 +857,12 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnWithTypeAndName & col_type_name = columns[arguments[0]]; + const ColumnWithTypeAndName & col_type_name = arguments[0]; const ColumnPtr & column = col_type_name.column; - if (const auto col_in = checkAndGetColumn(column.get())) + if (const auto * col_in = checkAndGetColumn(column.get())) { const auto & vec_in = col_in->getChars(); const auto & offsets_in = col_in->getOffsets(); @@ -891,9 +891,9 @@ public: src_offset += string_size; } - columns[result].column = std::move(col_res); + return col_res; } - else if (const auto col_in_fixed = checkAndGetColumn(column.get())) + else if (const auto * col_in_fixed = checkAndGetColumn(column.get())) { if (col_in_fixed->getN() != uuid_text_length) throw Exception("Illegal type " + col_type_name.type->getName() + @@ -920,10 +920,10 @@ public: dst_offset += uuid_bytes_length; } - columns[result].column = std::move(col_res); + return col_res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } }; @@ -1083,7 +1083,7 @@ public: } - void executeOneString(const UInt8 * pos, const UInt8 * end, char *& out) const + static void executeOneString(const UInt8 * pos, const UInt8 * end, char *& out) { while (pos < end) { @@ -1095,7 +1095,7 @@ public: ++out; } - bool tryExecuteString(const IColumn * col, ColumnPtr & col_res) const + static bool tryExecuteString(const IColumn * col, ColumnPtr & col_res) { const ColumnString * col_str_in = checkAndGetColumn(col); @@ -1139,7 +1139,7 @@ public: } } - bool tryExecuteFixedString(const IColumn * col, ColumnPtr & col_res) const + static bool tryExecuteFixedString(const IColumn * col, ColumnPtr & col_res) { const ColumnFixedString * col_fstr_in = checkAndGetColumn(col); @@ -1187,10 +1187,10 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const IColumn * column = columns[arguments[0]].column.get(); - ColumnPtr & res_column = columns[result].column; + const IColumn * column = arguments[0].column.get(); + ColumnPtr res_column; if (tryExecuteUInt(column, res_column) || tryExecuteUInt(column, res_column) || @@ -1203,9 +1203,9 @@ public: tryExecuteDecimal(column, res_column) || tryExecuteDecimal(column, res_column) || tryExecuteDecimal(column, res_column)) - return; + return res_column; - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -1235,7 +1235,7 @@ public: return std::make_shared(); } - void unhexOne(const char * pos, const char * end, char *& out) const + static void unhexOne(const char * pos, const char * end, char *& out) { if ((end - pos) & 1) { @@ -1255,9 +1255,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const ColumnPtr & column = columns[arguments[0]].column; + const ColumnPtr & column = arguments[0].column; if (const ColumnString * col = checkAndGetColumn(column.get())) { @@ -1290,11 +1290,11 @@ public: out_vec.resize(pos - begin); - columns[result].column = std::move(col_res); + return col_res; } else { - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -1335,7 +1335,7 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { auto col_str = ColumnString::create(); ColumnString::Chars & out_vec = col_str->getChars(); @@ -1355,7 +1355,7 @@ public: for (size_t idx = 0; idx < arguments.size(); ++idx) { //partial const column - columns_holder[idx] = columns[arguments[idx]].column->convertToFullColumnIfConst(); + columns_holder[idx] = arguments[idx].column->convertToFullColumnIfConst(); const IColumn * column = columns_holder[idx].get(); if (!(executeNumber(*column, out_vec, idx, input_rows_count, size_per_row) @@ -1369,12 +1369,12 @@ public: || executeNumber(*column, out_vec, idx, input_rows_count, size_per_row) || executeNumber(*column, out_vec, idx, input_rows_count, size_per_row))) { - throw Exception{"Illegal column " + columns[arguments[idx]].column->getName() + throw Exception{"Illegal column " + arguments[idx].column->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; } } - columns[result].column = std::move(col_str); + return col_str; } private: @@ -1461,10 +1461,10 @@ public: } } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const IColumn * in_column = columns[arguments[0]].column.get(); - ColumnPtr & out_column = columns[result].column; + const IColumn * in_column = arguments[0].column.get(); + ColumnPtr out_column; if (tryExecute(in_column, out_column) || tryExecute(in_column, out_column) || @@ -1474,9 +1474,9 @@ public: tryExecute(in_column, out_column) || tryExecute(in_column, out_column) || tryExecute(in_column, out_column)) - return; + return out_column; - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -1506,7 +1506,7 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - bool tryExecuteString(const IColumn * col, ColumnPtr & col_res) const + static bool tryExecuteString(const IColumn * col, ColumnPtr & col_res) { const ColumnString * col_str_in = checkAndGetColumn(col); @@ -1553,7 +1553,7 @@ public: } } - bool tryExecuteFixedString(const IColumn * col, ColumnPtr & col_res) const + static bool tryExecuteFixedString(const IColumn * col, ColumnPtr & col_res) { const ColumnFixedString * col_fstr_in = checkAndGetColumn(col); @@ -1599,15 +1599,15 @@ public: } } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const IColumn * column = columns[arguments[0]].column.get(); - ColumnPtr & res_column = columns[result].column; + const IColumn * column = arguments[0].column.get(); + ColumnPtr res_column; if (tryExecuteFixedString(column, res_column) || tryExecuteString(column, res_column)) - return; + return res_column; - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -1648,7 +1648,7 @@ public: DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override { - const auto first_argument = checkAndGetDataType(arguments[0].get()); + const auto * first_argument = checkAndGetDataType(arguments[0].get()); if (!first_argument || first_argument->getN() != IPV6_BINARY_LENGTH) throw Exception("Illegal type " + arguments[0]->getName() + " of first argument of function " + getName() + @@ -1659,7 +1659,7 @@ public: if (!isUInt8(second_argument)) throw Exception{"Illegal type " + second_argument->getName() + " of second argument of function " + getName() - + ", expected numeric type.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; + + ", expected UInt8", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; DataTypePtr element = DataTypeFactory::instance().get("IPv6"); return std::make_shared(DataTypes{element, element}); @@ -1668,38 +1668,38 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const auto & col_type_name_ip = columns[arguments[0]]; + const auto & col_type_name_ip = arguments[0]; const ColumnPtr & column_ip = col_type_name_ip.column; - const auto col_ip_in = checkAndGetColumn(column_ip.get()); + const auto * col_const_ip_in = checkAndGetColumnConst(column_ip.get()); + const auto * col_ip_in = checkAndGetColumn(column_ip.get()); - if (!col_ip_in) - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + if (!col_ip_in && !col_const_ip_in) + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - if (col_ip_in->getN() != IPV6_BINARY_LENGTH) - throw Exception("Illegal type " + col_type_name_ip.type->getName() + - " of column " + col_ip_in->getName() + - " argument of function " + getName() + - ", expected FixedString(" + toString(IPV6_BINARY_LENGTH) + ")", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + if ((col_const_ip_in && col_const_ip_in->getValue().size() != IPV6_BINARY_LENGTH) || + (col_ip_in && col_ip_in->getN() != IPV6_BINARY_LENGTH)) + throw Exception("Illegal type " + col_type_name_ip.type->getName() + + " of column " + column_ip->getName() + + " argument of function " + getName() + + ", expected FixedString(" + toString(IPV6_BINARY_LENGTH) + ")", + ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - const auto & col_type_name_cidr = columns[arguments[1]]; + const auto & col_type_name_cidr = arguments[1]; const ColumnPtr & column_cidr = col_type_name_cidr.column; - const auto col_const_cidr_in = checkAndGetColumnConst(column_cidr.get()); - const auto col_cidr_in = checkAndGetColumn(column_cidr.get()); + const auto * col_const_cidr_in = checkAndGetColumnConst(column_cidr.get()); + const auto * col_cidr_in = checkAndGetColumn(column_cidr.get()); if (!col_const_cidr_in && !col_cidr_in) - throw Exception("Illegal column " + columns[arguments[1]].column->getName() + throw Exception("Illegal column " + arguments[1].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - const auto & vec_in = col_ip_in->getChars(); - auto col_res_lower_range = ColumnFixedString::create(IPV6_BINARY_LENGTH); auto col_res_upper_range = ColumnFixedString::create(IPV6_BINARY_LENGTH); @@ -1711,17 +1711,27 @@ public: static constexpr UInt8 max_cidr_mask = IPV6_BINARY_LENGTH * 8; + const String col_const_ip_str = col_const_ip_in ? col_const_ip_in->getValue() : ""; + const UInt8 * col_const_ip_value = col_const_ip_in ? reinterpret_cast(col_const_ip_str.c_str()) : nullptr; + for (size_t offset = 0; offset < input_rows_count; ++offset) { const size_t offset_ipv6 = offset * IPV6_BINARY_LENGTH; + + const UInt8 * ip = col_const_ip_in + ? col_const_ip_value + : &col_ip_in->getChars()[offset_ipv6]; + UInt8 cidr = col_const_cidr_in ? col_const_cidr_in->getValue() : col_cidr_in->getData()[offset]; + cidr = std::min(cidr, max_cidr_mask); - applyCIDRMask(&vec_in[offset_ipv6], &vec_res_lower_range[offset_ipv6], &vec_res_upper_range[offset_ipv6], cidr); + + applyCIDRMask(ip, &vec_res_lower_range[offset_ipv6], &vec_res_upper_range[offset_ipv6], cidr); } - columns[result].column = ColumnTuple::create(Columns{std::move(col_res_lower_range), std::move(col_res_upper_range)}); + return ColumnTuple::create(Columns{std::move(col_res_lower_range), std::move(col_res_upper_range)}); } }; @@ -1763,7 +1773,7 @@ public: if (!isUInt8(second_argument)) throw Exception{"Illegal type " + second_argument->getName() + " of second argument of function " + getName() - + ", expected numeric type.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; + + ", expected UInt8", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; DataTypePtr element = DataTypeFactory::instance().get("IPv4"); return std::make_shared(DataTypes{element, element}); @@ -1772,30 +1782,29 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { - const auto & col_type_name_ip = columns[arguments[0]]; + const auto & col_type_name_ip = arguments[0]; const ColumnPtr & column_ip = col_type_name_ip.column; - const auto col_ip_in = checkAndGetColumn(column_ip.get()); - if (!col_ip_in) - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + const auto * col_const_ip_in = checkAndGetColumnConst(column_ip.get()); + const auto * col_ip_in = checkAndGetColumn(column_ip.get()); + if (!col_const_ip_in && !col_ip_in) + throw Exception("Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - const auto & col_type_name_cidr = columns[arguments[1]]; + const auto & col_type_name_cidr = arguments[1]; const ColumnPtr & column_cidr = col_type_name_cidr.column; - const auto col_const_cidr_in = checkAndGetColumnConst(column_cidr.get()); - const auto col_cidr_in = checkAndGetColumn(column_cidr.get()); + const auto * col_const_cidr_in = checkAndGetColumnConst(column_cidr.get()); + const auto * col_cidr_in = checkAndGetColumn(column_cidr.get()); if (!col_const_cidr_in && !col_cidr_in) - throw Exception("Illegal column " + columns[arguments[1]].column->getName() + throw Exception("Illegal column " + arguments[1].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - const auto & vec_in = col_ip_in->getData(); - auto col_res_lower_range = ColumnUInt32::create(); auto col_res_upper_range = ColumnUInt32::create(); @@ -1807,14 +1816,18 @@ public: for (size_t i = 0; i < input_rows_count; ++i) { + UInt32 ip = col_const_ip_in + ? col_const_ip_in->getValue() + : col_ip_in->getData()[i]; + UInt8 cidr = col_const_cidr_in ? col_const_cidr_in->getValue() : col_cidr_in->getData()[i]; - std::tie(vec_res_lower_range[i], vec_res_upper_range[i]) = applyCIDRMask(vec_in[i], cidr); + std::tie(vec_res_lower_range[i], vec_res_upper_range[i]) = applyCIDRMask(ip, cidr); } - columns[result].column = ColumnTuple::create(Columns{std::move(col_res_lower_range), std::move(col_res_upper_range)}); + return ColumnTuple::create(Columns{std::move(col_res_lower_range), std::move(col_res_upper_range)}); } }; diff --git a/src/Functions/FunctionsComparison.h b/src/Functions/FunctionsComparison.h index 39593b4a638..057f52501e5 100644 --- a/src/Functions/FunctionsComparison.h +++ b/src/Functions/FunctionsComparison.h @@ -561,7 +561,7 @@ public: static constexpr auto name = Name::name; static FunctionPtr create(const Context & context) { return std::make_shared(context); } - FunctionComparison(const Context & context_) + explicit FunctionComparison(const Context & context_) : context(context_), check_decimal_overflow(decimalCheckComparisonOverflow(context)) {} @@ -571,7 +571,7 @@ private: bool check_decimal_overflow = true; template - bool executeNumRightType(ColumnsWithTypeAndName & columns, size_t result, const ColumnVector * col_left, const IColumn * col_right_untyped) const + ColumnPtr executeNumRightType(const ColumnVector * col_left, const IColumn * col_right_untyped) const { if (const ColumnVector * col_right = checkAndGetColumn>(col_right_untyped)) { @@ -581,8 +581,7 @@ private: vec_res.resize(col_left->getData().size()); NumComparisonImpl>::vectorVector(col_left->getData(), col_right->getData(), vec_res); - columns[result].column = std::move(col_res); - return true; + return col_res; } else if (auto col_right_const = checkAndGetColumnConst>(col_right_untyped)) { @@ -592,15 +591,14 @@ private: vec_res.resize(col_left->size()); NumComparisonImpl>::vectorConstant(col_left->getData(), col_right_const->template getValue(), vec_res); - columns[result].column = std::move(col_res); - return true; + return col_res; } - return false; + return nullptr; } template - bool executeNumConstRightType(ColumnsWithTypeAndName & columns, size_t result, const ColumnConst * col_left, const IColumn * col_right_untyped) const + ColumnPtr executeNumConstRightType(const ColumnConst * col_left, const IColumn * col_right_untyped) const { if (const ColumnVector * col_right = checkAndGetColumn>(col_right_untyped)) { @@ -610,41 +608,40 @@ private: vec_res.resize(col_left->size()); NumComparisonImpl>::constantVector(col_left->template getValue(), col_right->getData(), vec_res); - columns[result].column = std::move(col_res); - return true; + return col_res; } else if (auto col_right_const = checkAndGetColumnConst>(col_right_untyped)) { UInt8 res = 0; NumComparisonImpl>::constantConstant(col_left->template getValue(), col_right_const->template getValue(), res); - columns[result].column = DataTypeUInt8().createColumnConst(col_left->size(), toField(res)); - return true; + return DataTypeUInt8().createColumnConst(col_left->size(), toField(res)); } - return false; + return nullptr; } template - bool executeNumLeftType(ColumnsWithTypeAndName & columns, size_t result, const IColumn * col_left_untyped, const IColumn * col_right_untyped) const + ColumnPtr executeNumLeftType(const IColumn * col_left_untyped, const IColumn * col_right_untyped) const { + ColumnPtr res = nullptr; if (const ColumnVector * col_left = checkAndGetColumn>(col_left_untyped)) { - if ( executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped) - || executeNumRightType(columns, result, col_left, col_right_untyped)) - return true; + if ( (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped)) + || (res = executeNumRightType(col_left, col_right_untyped))) + return res; else throw Exception("Illegal column " + col_right_untyped->getName() + " of second argument of function " + getName(), @@ -652,34 +649,35 @@ private: } else if (auto col_left_const = checkAndGetColumnConst>(col_left_untyped)) { - if ( executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped) - || executeNumConstRightType(columns, result, col_left_const, col_right_untyped)) - return true; + if ( (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped)) + || (res = executeNumConstRightType(col_left_const, col_right_untyped))) + return res; else throw Exception("Illegal column " + col_right_untyped->getName() + " of second argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } - return false; + return nullptr; } - void executeDecimal(ColumnsWithTypeAndName & columns, size_t result, const ColumnWithTypeAndName & col_left, const ColumnWithTypeAndName & col_right) const + ColumnPtr executeDecimal(const ColumnWithTypeAndName & col_left, const ColumnWithTypeAndName & col_right) const { TypeIndex left_number = col_left.type->getTypeId(); TypeIndex right_number = col_right.type->getTypeId(); + ColumnPtr res; auto call = [&](const auto & types) -> bool { @@ -688,18 +686,19 @@ private: using RightDataType = typename Types::RightType; if (check_decimal_overflow) - DecimalComparison(columns, result, col_left, col_right); + return (res = DecimalComparison::apply(col_left, col_right)) != nullptr; else - DecimalComparison(columns, result, col_left, col_right); - return true; + return (res = DecimalComparison::apply(col_left, col_right)) != nullptr; }; if (!callOnBasicTypes(left_number, right_number, call)) throw Exception("Wrong call for " + getName() + " with " + col_left.type->getName() + " and " + col_right.type->getName(), ErrorCodes::LOGICAL_ERROR); + + return res; } - bool executeString(ColumnsWithTypeAndName & columns, size_t result, const IColumn * c0, const IColumn * c1) const + ColumnPtr executeString(const IColumn * c0, const IColumn * c1) const { const ColumnString * c0_string = checkAndGetColumn(c0); const ColumnString * c1_string = checkAndGetColumn(c1); @@ -710,7 +709,7 @@ private: const ColumnConst * c1_const = checkAndGetColumnConstStringOrFixedString(c1); if (!((c0_string || c0_fixed_string || c0_const) && (c1_string || c1_fixed_string || c1_const))) - return false; + return nullptr; const ColumnString::Chars * c0_const_chars = nullptr; const ColumnString::Chars * c1_const_chars = nullptr; @@ -759,12 +758,11 @@ private: if (c0_const && c1_const) { - auto res = executeString(columns, result, &c0_const->getDataColumn(), &c1_const->getDataColumn()); + auto res = executeString(&c0_const->getDataColumn(), &c1_const->getDataColumn()); if (!res) - return false; + return nullptr; - columns[result].column = ColumnConst::create(columns[result].column, c0_const->size()); - return true; + return ColumnConst::create(res, c0_const->size()); } else { @@ -818,13 +816,12 @@ private: + " of arguments of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); - columns[result].column = std::move(c_res); - return true; + return c_res; } } - bool executeWithConstString( - ColumnsWithTypeAndName & columns, size_t result, const IColumn * col_left_untyped, const IColumn * col_right_untyped, + ColumnPtr executeWithConstString( + const DataTypePtr & result_type, const IColumn * col_left_untyped, const IColumn * col_right_untyped, const DataTypePtr & left_type, const DataTypePtr & right_type, size_t input_rows_count) const { /// To compare something with const string, we cast constant to appropriate type and compare as usual. @@ -835,7 +832,7 @@ private: const ColumnConst * right_const = checkAndGetColumnConstStringOrFixedString(col_right_untyped); if (!left_const && !right_const) - return false; + return nullptr; const IDataType * type_string = left_const ? left_type.get() : right_type.get(); const DataTypePtr & type_to_compare = !left_const ? left_type : right_type; @@ -846,29 +843,25 @@ private: /// If not possible to convert, comparison with =, <, >, <=, >= yields to false and comparison with != yields to true. if (converted.isNull()) { - columns[result].column = DataTypeUInt8().createColumnConst(input_rows_count, IsOperation::not_equals); + return DataTypeUInt8().createColumnConst(input_rows_count, IsOperation::not_equals); } else { auto column_converted = type_to_compare->createColumnConst(input_rows_count, converted); - ColumnsWithTypeAndName tmp_columns_columns + ColumnsWithTypeAndName tmp_columns { { left_const ? column_converted : col_left_untyped->getPtr(), type_to_compare, "" }, { !left_const ? column_converted : col_right_untyped->getPtr(), type_to_compare, "" }, - columns[result] }; - executeImpl(tmp_columns_columns, {0, 1}, 2, input_rows_count); - - columns[result].column = std::move(tmp_columns_columns[2].column); + return executeImpl(tmp_columns, result_type, input_rows_count); } - - return true; } - void executeTuple(ColumnsWithTypeAndName & columns, size_t result, const ColumnWithTypeAndName & c0, const ColumnWithTypeAndName & c1, - size_t input_rows_count) const + ColumnPtr executeTuple( + const DataTypePtr & result_type, const ColumnWithTypeAndName & c0, const ColumnWithTypeAndName & c1, + size_t input_rows_count) const { /** We will lexicographically compare the tuples. This is done as follows: * x == y : x1 == y1 && x2 == y2 ... @@ -892,12 +885,8 @@ private: if (tuple_size != typeid_cast(*c1.type).getElements().size()) throw Exception("Cannot compare tuples of different sizes.", ErrorCodes::BAD_ARGUMENTS); - auto & res = columns[result]; - if (res.type->onlyNull()) - { - res.column = res.type->createColumnConstWithDefaultValue(input_rows_count); - return; - } + if (result_type->onlyNull()) + return result_type->createColumnConstWithDefaultValue(input_rows_count); ColumnsWithTypeAndName x(tuple_size); ColumnsWithTypeAndName y(tuple_size); @@ -927,18 +916,16 @@ private: y[i].column = y_columns[i]; } - executeTupleImpl(columns, result, x, y, tuple_size, input_rows_count); + return executeTupleImpl(x, y, tuple_size, input_rows_count); } - void executeTupleImpl(ColumnsWithTypeAndName & columns, size_t result, const ColumnsWithTypeAndName & x, + ColumnPtr executeTupleImpl(const ColumnsWithTypeAndName & x, const ColumnsWithTypeAndName & y, size_t tuple_size, size_t input_rows_count) const; - void executeTupleEqualityImpl( + ColumnPtr executeTupleEqualityImpl( std::shared_ptr func_compare, std::shared_ptr func_convolution, - ColumnsWithTypeAndName & columns, - size_t result, const ColumnsWithTypeAndName & x, const ColumnsWithTypeAndName & y, size_t tuple_size, @@ -947,84 +934,70 @@ private: if (0 == tuple_size) throw Exception("Comparison of zero-sized tuples is not implemented.", ErrorCodes::NOT_IMPLEMENTED); - ColumnsWithTypeAndName convolution_types(tuple_size); + ColumnsWithTypeAndName convolution_columns(tuple_size); + ColumnsWithTypeAndName tmp_columns(2); - ColumnsWithTypeAndName tmp_columns; for (size_t i = 0; i < tuple_size; ++i) { - tmp_columns.emplace_back(x[i]); - tmp_columns.emplace_back(y[i]); + tmp_columns[0] = x[i]; + tmp_columns[1] = y[i]; - auto impl = func_compare->build({x[i], y[i]}); - convolution_types[i].type = impl->getReturnType(); + auto impl = func_compare->build(tmp_columns); + convolution_columns[i].type = impl->getResultType(); /// Comparison of the elements. - tmp_columns.emplace_back(ColumnWithTypeAndName{ nullptr, impl->getReturnType(), "" }); - impl->execute(tmp_columns, {i * 3, i * 3 + 1}, i * 3 + 2, input_rows_count); + convolution_columns[i].column = impl->execute(tmp_columns, impl->getResultType(), input_rows_count); } if (tuple_size == 1) { /// Do not call AND for single-element tuple. - columns[result].column = tmp_columns[2].column; - return; + return convolution_columns[0].column; } /// Logical convolution. - - ColumnNumbers convolution_args(tuple_size); - for (size_t i = 0; i < tuple_size; ++i) - convolution_args[i] = i * 3 + 2; - - auto impl = func_convolution->build(convolution_types); - tmp_columns.emplace_back(ColumnWithTypeAndName{ nullptr, impl->getReturnType(), "" }); - - impl->execute(tmp_columns, convolution_args, tuple_size * 3, input_rows_count); - columns[result].column = tmp_columns[tuple_size * 3].column; + auto impl = func_convolution->build(convolution_columns); + return impl->execute(convolution_columns, impl->getResultType(), input_rows_count); } - void executeTupleLessGreaterImpl( + ColumnPtr executeTupleLessGreaterImpl( std::shared_ptr func_compare_head, std::shared_ptr func_compare_tail, std::shared_ptr func_and, std::shared_ptr func_or, std::shared_ptr func_equals, - ColumnsWithTypeAndName & columns, - size_t result, const ColumnsWithTypeAndName & x, const ColumnsWithTypeAndName & y, size_t tuple_size, size_t input_rows_count) const { - ColumnsWithTypeAndName tmp_columns; + ColumnsWithTypeAndName less_columns(tuple_size); + ColumnsWithTypeAndName equal_columns(tuple_size - 1); + ColumnsWithTypeAndName tmp_columns(2); /// Pairwise comparison of the inequality of all elements; on the equality of all elements except the last. /// (x[i], y[i], x[i] < y[i], x[i] == y[i]) for (size_t i = 0; i < tuple_size; ++i) { - tmp_columns.emplace_back(x[i]); - tmp_columns.emplace_back(y[i]); - - tmp_columns.emplace_back(ColumnWithTypeAndName()); // pos == i * 4 + 2 + tmp_columns[0] = x[i]; + tmp_columns[1] = y[i]; if (i + 1 != tuple_size) { - auto impl_head = func_compare_head->build({x[i], y[i]}); - tmp_columns[i * 4 + 2].type = impl_head->getReturnType(); - impl_head->execute(tmp_columns, {i * 4, i * 4 + 1}, i * 4 + 2, input_rows_count); + auto impl_head = func_compare_head->build(tmp_columns); + less_columns[i].type = impl_head->getResultType(); + less_columns[i].column = impl_head->execute(tmp_columns, less_columns[i].type, input_rows_count); - tmp_columns.emplace_back(ColumnWithTypeAndName()); // i * 4 + 3 - - auto impl_equals = func_equals->build({x[i], y[i]}); - tmp_columns[i * 4 + 3].type = impl_equals->getReturnType(); - impl_equals->execute(tmp_columns, {i * 4, i * 4 + 1}, i * 4 + 3, input_rows_count); + auto impl_equals = func_equals->build(tmp_columns); + equal_columns[i].type = impl_equals->getResultType(); + equal_columns[i].column = impl_equals->execute(tmp_columns, equal_columns[i].type, input_rows_count); } else { - auto impl_tail = func_compare_tail->build({x[i], y[i]}); - tmp_columns[i * 4 + 2].type = impl_tail->getReturnType(); - impl_tail->execute(tmp_columns, {i * 4, i * 4 + 1}, i * 4 + 2, input_rows_count); + auto impl_tail = func_compare_tail->build(tmp_columns); + less_columns[i].type = impl_tail->getResultType(); + less_columns[i].column = impl_tail->execute(tmp_columns, less_columns[i].type, input_rows_count); } } @@ -1035,38 +1008,28 @@ private: /// for (int i = tuple_size - 2; i >= 0; --i) /// res = (res && `x == y`[i]) || `x < y`[i]; size_t i = tuple_size - 1; + tmp_columns[0] = less_columns[i]; while (i > 0) { --i; - size_t and_lhs_pos = tmp_columns.size() - 1; // res - size_t and_rhs_pos = i * 4 + 3; // `x == y`[i] - tmp_columns.emplace_back(ColumnWithTypeAndName()); + tmp_columns[1] = equal_columns[i]; + auto func_and_adaptor = func_and->build(tmp_columns); - ColumnsWithTypeAndName and_args = {{ nullptr, tmp_columns[and_lhs_pos].type, "" }, - { nullptr, tmp_columns[and_rhs_pos].type, "" }}; + tmp_columns[0].column = func_and_adaptor->execute(tmp_columns, func_and_adaptor->getResultType(), input_rows_count); + tmp_columns[0].type = func_and_adaptor->getResultType(); - auto func_and_adaptor = func_and->build(and_args); - tmp_columns[tmp_columns.size() - 1].type = func_and_adaptor->getReturnType(); - func_and_adaptor->execute(tmp_columns, {and_lhs_pos, and_rhs_pos}, tmp_columns.size() - 1, input_rows_count); - - size_t or_lhs_pos = tmp_columns.size() - 1; // (res && `x == y`[i]) - size_t or_rhs_pos = i * 4 + 2; // `x < y`[i] - tmp_columns.emplace_back(ColumnWithTypeAndName()); - - ColumnsWithTypeAndName or_args = {{ nullptr, tmp_columns[or_lhs_pos].type, "" }, - { nullptr, tmp_columns[or_rhs_pos].type, "" }}; - - auto func_or_adaptor = func_or->build(or_args); - tmp_columns[tmp_columns.size() - 1].type = func_or_adaptor->getReturnType(); - func_or_adaptor->execute(tmp_columns, {or_lhs_pos, or_rhs_pos}, tmp_columns.size() - 1, input_rows_count); + tmp_columns[1] = less_columns[i]; + auto func_or_adaptor = func_or->build(tmp_columns); + tmp_columns[0].column = func_or_adaptor->execute(tmp_columns, func_or_adaptor->getResultType(), input_rows_count); + tmp_columns[tmp_columns.size() - 1].type = func_or_adaptor->getResultType(); } - columns[result].column = tmp_columns[tmp_columns.size() - 1].column; + return tmp_columns[0].column; } - void executeGenericIdenticalTypes(ColumnsWithTypeAndName & columns, size_t result, const IColumn * c0, const IColumn * c1) const + ColumnPtr executeGenericIdenticalTypes(const IColumn * c0, const IColumn * c1) const { bool c0_const = isColumnConst(*c0); bool c1_const = isColumnConst(*c1); @@ -1075,7 +1038,7 @@ private: { UInt8 res = 0; GenericComparisonImpl>::constantConstant(*c0, *c1, res); - columns[result].column = DataTypeUInt8().createColumnConst(c0->size(), toField(res)); + return DataTypeUInt8().createColumnConst(c0->size(), toField(res)); } else { @@ -1090,18 +1053,18 @@ private: else GenericComparisonImpl>::vectorVector(*c0, *c1, vec_res); - columns[result].column = std::move(c_res); + return c_res; } } - void executeGeneric(ColumnsWithTypeAndName & columns, size_t result, const ColumnWithTypeAndName & c0, const ColumnWithTypeAndName & c1) const + ColumnPtr executeGeneric(const ColumnWithTypeAndName & c0, const ColumnWithTypeAndName & c1) const { DataTypePtr common_type = getLeastSupertype({c0.type, c1.type}); ColumnPtr c0_converted = castColumn(c0, common_type); ColumnPtr c1_converted = castColumn(c1, common_type); - executeGenericIdenticalTypes(columns, result, c0_converted.get(), c1_converted.get()); + return executeGenericIdenticalTypes(c0_converted.get(), c1_converted.get()); } public: @@ -1157,7 +1120,7 @@ public: { ColumnsWithTypeAndName args = {{nullptr, left_tuple->getElements()[i], ""}, {nullptr, right_tuple->getElements()[i], ""}}; - auto element_type = adaptor.build(args)->getReturnType(); + auto element_type = adaptor.build(args)->getResultType(); has_nullable = has_nullable || element_type->isNullable(); has_null = has_null || element_type->onlyNull(); } @@ -1173,10 +1136,10 @@ public: return std::make_shared(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - const auto & col_with_type_and_name_left = columns[arguments[0]]; - const auto & col_with_type_and_name_right = columns[arguments[1]]; + const auto & col_with_type_and_name_left = arguments[0]; + const auto & col_with_type_and_name_right = arguments[1]; const IColumn * col_left_untyped = col_with_type_and_name_left.column.get(); const IColumn * col_right_untyped = col_with_type_and_name_right.column.get(); @@ -1194,13 +1157,11 @@ public: || IsOperation::less_or_equals || IsOperation::greater_or_equals) { - columns[result].column = DataTypeUInt8().createColumnConst(input_rows_count, 1u); - return; + return DataTypeUInt8().createColumnConst(input_rows_count, 1u); } else { - columns[result].column = DataTypeUInt8().createColumnConst(input_rows_count, 0u); - return; + return DataTypeUInt8().createColumnConst(input_rows_count, 0u); } } @@ -1216,39 +1177,44 @@ public: bool date_and_datetime = (which_left.idx != which_right.idx) && which_left.isDateOrDateTime() && which_right.isDateOrDateTime(); + ColumnPtr res; if (left_is_num && right_is_num && !date_and_datetime) { - if (!(executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped) - || executeNumLeftType(columns, result, col_left_untyped, col_right_untyped))) + if (!((res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)) + || (res = executeNumLeftType(col_left_untyped, col_right_untyped)))) throw Exception("Illegal column " + col_left_untyped->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); + + return res; } else if (checkAndGetDataType(left_type.get()) && checkAndGetDataType(right_type.get())) { - executeTuple(columns, result, col_with_type_and_name_left, col_with_type_and_name_right, input_rows_count); + return executeTuple(result_type, col_with_type_and_name_left, col_with_type_and_name_right, input_rows_count); } - else if (left_is_string && right_is_string && executeString(columns, result, col_left_untyped, col_right_untyped)) + else if (left_is_string && right_is_string && (res = executeString(col_left_untyped, col_right_untyped))) { + return res; } - else if (executeWithConstString( - columns, result, col_left_untyped, col_right_untyped, + else if ((res = executeWithConstString( + result_type, col_left_untyped, col_right_untyped, left_type, right_type, - input_rows_count)) + input_rows_count))) { + return res; } else if (isColumnedAsDecimal(left_type) || isColumnedAsDecimal(right_type)) { @@ -1257,15 +1223,15 @@ public: throw Exception("No operation " + getName() + " between " + left_type->getName() + " and " + right_type->getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - executeDecimal(columns, result, col_with_type_and_name_left, col_with_type_and_name_right); + return executeDecimal(col_with_type_and_name_left, col_with_type_and_name_right); } else if (left_type->equals(*right_type)) { - executeGenericIdenticalTypes(columns, result, col_left_untyped, col_right_untyped); + return executeGenericIdenticalTypes(col_left_untyped, col_right_untyped); } else { - executeGeneric(columns, result, col_with_type_and_name_left, col_with_type_and_name_right); + return executeGeneric(col_with_type_and_name_left, col_with_type_and_name_right); } } diff --git a/src/Functions/FunctionsConsistentHashing.h b/src/Functions/FunctionsConsistentHashing.h index f5a2d3aac4d..edadfd659e2 100644 --- a/src/Functions/FunctionsConsistentHashing.h +++ b/src/Functions/FunctionsConsistentHashing.h @@ -65,10 +65,10 @@ public: return {1}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - if (isColumnConst(*columns[arguments[1]].column)) - executeConstBuckets(columns, arguments, result); + if (isColumnConst(*arguments[1].column)) + return executeConstBuckets(arguments); else throw Exception( "The second argument of function " + getName() + " (number of buckets) must be constant", ErrorCodes::BAD_ARGUMENTS); @@ -93,9 +93,9 @@ private: return static_cast(buckets); } - void executeConstBuckets(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result) const + ColumnPtr executeConstBuckets(ColumnsWithTypeAndName & arguments) const { - Field buckets_field = (*columns[arguments[1]].column)[0]; + Field buckets_field = (*arguments[1].column)[0]; BucketsType num_buckets; if (buckets_field.getType() == Field::Types::Int64) @@ -106,8 +106,8 @@ private: throw Exception("Illegal type " + String(buckets_field.getTypeName()) + " of the second argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - const auto & hash_col = columns[arguments[0]].column; - const IDataType * hash_type = columns[arguments[0]].type.get(); + const auto & hash_col = arguments[0].column; + const IDataType * hash_type = arguments[0].type.get(); auto res_col = ColumnVector::create(); WhichDataType which(hash_type); @@ -132,7 +132,7 @@ private: throw Exception("Illegal type " + hash_type->getName() + " of the first argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - columns[result].column = std::move(res_col); + return res_col; } template diff --git a/src/Functions/FunctionsConversion.cpp b/src/Functions/FunctionsConversion.cpp index 3f38614f584..df962800385 100644 --- a/src/Functions/FunctionsConversion.cpp +++ b/src/Functions/FunctionsConversion.cpp @@ -68,6 +68,8 @@ void registerFunctionsConversion(FunctionFactory & factory) factory.registerFunction(); factory.registerFunction(); + factory.registerFunction(); + factory.registerFunction(); factory.registerFunction(); factory.registerFunction(); @@ -90,6 +92,8 @@ void registerFunctionsConversion(FunctionFactory & factory) factory.registerFunction(); factory.registerFunction(); + factory.registerFunction(); + factory.registerFunction(); factory.registerFunction(); factory.registerFunction(); diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 1cbf8cc3925..70e8904cfc1 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -101,10 +101,11 @@ struct ConvertImpl using ToFieldType = typename ToDataType::FieldType; template - static void NO_SANITIZE_UNDEFINED execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/, - Additions additions [[maybe_unused]] = Additions()) + static ColumnPtr NO_SANITIZE_UNDEFINED execute( + ColumnsWithTypeAndName & arguments, const DataTypePtr & /*result_type*/, size_t /*input_rows_count*/, + Additions additions [[maybe_unused]] = Additions()) { - const ColumnWithTypeAndName & named_from = columns[arguments[0]]; + const ColumnWithTypeAndName & named_from = arguments[0]; using ColVecFrom = typename FromDataType::ColumnType; using ColVecTo = typename ToDataType::ColumnType; @@ -161,7 +162,7 @@ struct ConvertImpl vec_to[i] = static_cast(vec_from[i]); } - columns[result].column = std::move(col_to); + return col_to; } else throw Exception("Illegal column " + named_from.column->getName() + " of first argument of function " + Name::name, @@ -441,9 +442,9 @@ struct FormatImpl> template struct ConvertImpl, DataTypeNumber, Name> { - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) { - columns[result].column = columns[arguments[0]].column; + return arguments[0].column; } }; @@ -454,16 +455,16 @@ struct ConvertImpl, ColumnDecimal, ColumnVector>; - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) { - const auto & col_with_type_and_name = columns[arguments[0]]; + const auto & col_with_type_and_name = arguments[0]; const auto & type = static_cast(*col_with_type_and_name.type); const DateLUTImpl * time_zone = nullptr; /// For argument of DateTime type, second argument with time zone could be specified. if constexpr (std::is_same_v || std::is_same_v) - time_zone = &extractTimeZoneFromFunctionArguments(columns, arguments, 1, 0); + time_zone = &extractTimeZoneFromFunctionArguments(arguments, 1, 0); if (const auto col_from = checkAndGetColumn(col_with_type_and_name.column.get())) { @@ -495,10 +496,10 @@ struct ConvertImplgetName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + Name::name, ErrorCodes::ILLEGAL_COLUMN); } @@ -508,9 +509,9 @@ struct ConvertImpl inline void parseImpl(DataTypeUUID::FieldType & x, ReadBuffer & rb, const DateLUTImpl *) { UUID tmp; - readText(tmp, rb); + readUUIDText(tmp, rb); x = tmp; } @@ -602,12 +603,23 @@ inline bool tryParseImpl(DataTypeDateTime::FieldType & x, Read return true; } +template <> +inline bool tryParseImpl(DataTypeUUID::FieldType & x, ReadBuffer & rb, const DateLUTImpl *) +{ + UUID tmp; + if (!tryReadUUIDText(tmp, rb)) + return false; + + x = tmp; + return true; +} + /** Throw exception with verbose message when string value is not parsed completely. */ -[[noreturn]] inline void throwExceptionForIncompletelyParsedValue(ReadBuffer & read_buffer, ColumnsWithTypeAndName & columns, size_t result) +[[noreturn]] inline void throwExceptionForIncompletelyParsedValue(ReadBuffer & read_buffer, const DataTypePtr result_type) { - const IDataType & to_type = *columns[result].type; + const IDataType & to_type = *result_type; WriteBufferFromOwnString message_buf; message_buf << "Cannot parse string " << quote << String(read_buffer.buffer().begin(), read_buffer.buffer().size()) @@ -650,7 +662,7 @@ struct ConvertThroughParsing static constexpr bool to_datetime64 = std::is_same_v; - using ToFieldType = typename ToDataType::FieldType; + // using ToFieldType = typename ToDataType::FieldType; static bool isAllRead(ReadBuffer & in) { @@ -670,7 +682,7 @@ struct ConvertThroughParsing } template - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count, + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr & res_type, size_t input_rows_count, Additions additions [[maybe_unused]] = Additions()) { using ColVecTo = typename ToDataType::ColumnType; @@ -681,20 +693,20 @@ struct ConvertThroughParsing /// For conversion to DateTime type, second argument with time zone could be specified. if constexpr (std::is_same_v || to_datetime64) { - const auto result_type = removeNullable(columns[result].type); + const auto result_type = removeNullable(res_type); // Time zone is already figured out during result type resolution, no need to do it here. if (const auto dt_col = checkAndGetDataType(result_type.get())) local_time_zone = &dt_col->getTimeZone(); else { - local_time_zone = &extractTimeZoneFromFunctionArguments(columns, arguments, 1, 0); + local_time_zone = &extractTimeZoneFromFunctionArguments(arguments, 1, 0); } if constexpr (parsing_mode == ConvertFromStringParsingMode::BestEffort || parsing_mode == ConvertFromStringParsingMode::BestEffortUS) utc_time_zone = &DateLUT::instance("UTC"); } - const IColumn * col_from = columns[arguments[0]].column.get(); + const IColumn * col_from = arguments[0].column.get(); const ColumnString * col_from_string = checkAndGetColumn(col_from); const ColumnFixedString * col_from_fixed_string = checkAndGetColumn(col_from); @@ -799,7 +811,7 @@ struct ConvertThroughParsing } if (!isAllRead(read_buffer)) - throwExceptionForIncompletelyParsedValue(read_buffer, columns, result); + throwExceptionForIncompletelyParsedValue(read_buffer, res_type); } else { @@ -847,9 +859,9 @@ struct ConvertThroughParsing } if constexpr (exception_mode == ConvertFromStringExceptionMode::Null) - columns[result].column = ColumnNullable::create(std::move(col_to), std::move(col_null_map_to)); + return ColumnNullable::create(std::move(col_to), std::move(col_null_map_to)); else - columns[result].column = std::move(col_to); + return col_to; } }; @@ -865,12 +877,12 @@ struct ConvertImpl(&col_from)) { @@ -892,15 +904,15 @@ struct ConvertImplGenericFromString data_type_to.deserializeAsWholeText(column_to, read_buffer, format_settings); if (!read_buffer.eof()) - throwExceptionForIncompletelyParsedValue(read_buffer, columns, result); + throwExceptionForIncompletelyParsedValue(read_buffer, result_type); current_offset = offsets[i]; } - columns[result].column = std::move(res); + return res; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of conversion function from string", ErrorCodes::ILLEGAL_COLUMN); } @@ -920,9 +932,9 @@ struct ConvertImpl template struct ConvertImpl, T, Name> { - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) { - columns[result].column = columns[arguments[0]].column; + return arguments[0].column; } }; @@ -933,9 +945,9 @@ struct ConvertImpl, T, Name> template struct ConvertImpl { - static void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) + static ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) { - if (const ColumnFixedString * col_from = checkAndGetColumn(columns[arguments[0]].column.get())) + if (const ColumnFixedString * col_from = checkAndGetColumn(arguments[0].column.get())) { auto col_to = ColumnString::create(); @@ -964,10 +976,10 @@ struct ConvertImpl } data_to.resize(offset_to); - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + Name::name, ErrorCodes::ILLEGAL_COLUMN); } @@ -1009,17 +1021,14 @@ struct NameParseDateTimeBestEffortOrZero; struct NameParseDateTimeBestEffortOrNull; template -static inline bool isDateTime64(const ColumnsWithTypeAndName & arguments, const ColumnNumbers & arguments_index = {}) +static inline bool isDateTime64(const ColumnsWithTypeAndName & arguments) { if constexpr (std::is_same_v) return true; else if constexpr (std::is_same_v || std::is_same_v || std::is_same_v || std::is_same_v) { - if (arguments_index.empty()) - return (arguments.size() == 2 && isUnsignedInteger(arguments[1].type)) || arguments.size() == 3; - else - return (arguments_index.size() == 2 && isUnsignedInteger(arguments[arguments_index[1]].type)) || arguments_index.size() == 3; + return (arguments.size() == 2 && isUnsignedInteger(arguments[1].type)) || arguments.size() == 3; } return false; @@ -1066,7 +1075,7 @@ public: } // toString(DateTime or DateTime64, [timezone: String]) - if ((std::is_same_v && arguments.size() > 0 && (isDateTime64(arguments[0].type) || isDateTime(arguments[0].type))) + if ((std::is_same_v && !arguments.empty() && (isDateTime64(arguments[0].type) || isDateTime(arguments[0].type))) // toUnixTimestamp(value[, timezone : String]) || std::is_same_v // toDate(value[, timezone : String]) @@ -1132,11 +1141,11 @@ public: ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } bool canBeExecutedOnDefaultArguments() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { try { - executeInternal(columns, arguments, result, input_rows_count); + return executeInternal(arguments, result_type, input_rows_count); } catch (Exception & e) { @@ -1144,8 +1153,8 @@ public: if (e.code() == ErrorCodes::ATTEMPT_TO_READ_AFTER_EOF) { e.addMessage("Cannot parse " - + columns[result].type->getName() + " from " - + columns[arguments[0]].type->getName() + + result_type->getName() + " from " + + arguments[0].type->getName() + ", because value is too short"); } else if (e.code() == ErrorCodes::CANNOT_PARSE_NUMBER @@ -1158,8 +1167,8 @@ public: || e.code() == ErrorCodes::CANNOT_PARSE_UUID) { e.addMessage("Cannot parse " - + columns[result].type->getName() + " from " - + columns[arguments[0]].type->getName()); + + result_type->getName() + " from " + + arguments[0].type->getName()); } throw; @@ -1177,13 +1186,14 @@ public: } private: - void executeInternal(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const + ColumnPtr executeInternal(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const { - if (!arguments.size()) + if (arguments.empty()) throw Exception{"Function " + getName() + " expects at least 1 arguments", ErrorCodes::TOO_FEW_ARGUMENTS_FOR_FUNCTION}; - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); + ColumnPtr result_column; auto call = [&](const auto & types) -> bool { @@ -1206,15 +1216,15 @@ private: ErrorCodes::TOO_FEW_ARGUMENTS_FOR_FUNCTION}; } - const ColumnWithTypeAndName & scale_column = columns[arguments[1]]; + const ColumnWithTypeAndName & scale_column = arguments[1]; UInt32 scale = extractToDecimalScale(scale_column); - ConvertImpl::execute(columns, arguments, result, input_rows_count, scale); + result_column = ConvertImpl::execute(arguments, result_type, input_rows_count, scale); } else if constexpr (IsDataTypeDateOrDateTime && std::is_same_v) { - const auto * dt64 = assert_cast(columns[arguments[0]].type.get()); - ConvertImpl::execute(columns, arguments, result, input_rows_count, dt64->getScale()); + const auto * dt64 = assert_cast(arguments[0].type.get()); + result_column = ConvertImpl::execute(arguments, result_type, input_rows_count, dt64->getScale()); } else if constexpr (IsDataTypeDecimalOrNumber && IsDataTypeDecimalOrNumber) { @@ -1233,27 +1243,27 @@ private: throw Exception("Wrong UUID conversion", ErrorCodes::CANNOT_CONVERT_TYPE); } else - ConvertImpl::execute(columns, arguments, result, input_rows_count); + result_column = ConvertImpl::execute(arguments, result_type, input_rows_count); } else - ConvertImpl::execute(columns, arguments, result, input_rows_count); + result_column = ConvertImpl::execute(arguments, result_type, input_rows_count); return true; }; - if (isDateTime64(columns, arguments)) + if (isDateTime64(arguments)) { /// For toDateTime('xxxx-xx-xx xx:xx:xx.00', 2[, 'timezone']) we need to it convert to DateTime64 - const ColumnWithTypeAndName & scale_column = columns[arguments[1]]; + const ColumnWithTypeAndName & scale_column = arguments[1]; UInt32 scale = extractToDecimalScale(scale_column); if (to_datetime64 || scale != 0) /// When scale = 0, the data type is DateTime otherwise the data type is DateTime64 { if (!callOnIndexAndDataType(from_type->getTypeId(), call)) - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName(), + throw Exception("Illegal type " + arguments[0].type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - return; + return result_column; } } @@ -1263,12 +1273,14 @@ private: /// Generic conversion of any type to String. if (std::is_same_v) { - ConvertImplGenericToString::execute(columns, arguments, result); + return ConvertImplGenericToString::execute(arguments); } else - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName(), + throw Exception("Illegal type " + arguments[0].type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); } + + return result_column; } }; @@ -1394,59 +1406,59 @@ public: } template - bool executeInternal(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count, UInt32 scale = 0) const + ColumnPtr executeInternal(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count, UInt32 scale = 0) const { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); if (checkAndGetDataType(from_type)) { - ConvertThroughParsing::execute( - columns, arguments, result, input_rows_count, scale); - return true; + return ConvertThroughParsing::execute( + arguments, result_type, input_rows_count, scale); } else if (checkAndGetDataType(from_type)) { - ConvertThroughParsing::execute( - columns, arguments, result, input_rows_count, scale); - return true; + return ConvertThroughParsing::execute( + arguments, result_type, input_rows_count, scale); } - return false; + return nullptr; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - bool ok = true; + ColumnPtr result_column; if constexpr (to_decimal) - ok = executeInternal(columns, arguments, result, input_rows_count, - assert_cast(*removeNullable(columns[result].type)).getScale()); + result_column = executeInternal(arguments, result_type, input_rows_count, + assert_cast(*removeNullable(result_type)).getScale()); else { - if (isDateTime64(columns, arguments)) + if (isDateTime64(arguments)) { UInt64 scale = to_datetime64 ? DataTypeDateTime64::default_scale : 0; if (arguments.size() > 1) - scale = extractToDecimalScale(columns[arguments[1]]); + scale = extractToDecimalScale(arguments[1]); if (scale == 0) - ok = executeInternal(columns, arguments, result, input_rows_count); + result_column = executeInternal(arguments, result_type, input_rows_count); else { - ok = executeInternal(columns, arguments, result, input_rows_count, static_cast(scale)); + result_column = executeInternal(arguments, result_type, input_rows_count, static_cast(scale)); } } else { - ok = executeInternal(columns, arguments, result, input_rows_count); + result_column = executeInternal(arguments, result_type, input_rows_count); } } - if (!ok) - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName() + if (!result_column) + throw Exception("Illegal type " + arguments[0].type->getName() + " of argument of function " + getName() + ". Only String or FixedString argument is accepted for try-conversion function." + " For other arguments, use function without 'orZero' or 'orNull'.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + + return result_column; } }; @@ -1630,8 +1642,8 @@ struct ToStringMonotonicity IFunction::Monotonicity positive(true, true); IFunction::Monotonicity not_monotonic; - auto type_ptr = &type; - if (auto * low_cardinality_type = checkAndGetDataType(type_ptr)) + const auto * type_ptr = &type; + if (const auto * low_cardinality_type = checkAndGetDataType(type_ptr)) type_ptr = low_cardinality_type->getDictionaryType().get(); /// `toString` function is monotonous if the argument is Date or DateTime or String, or non-negative numbers with the same number of symbols. @@ -1754,6 +1766,7 @@ struct NameToDecimal32OrZero { static constexpr auto name = "toDecimal32OrZero"; struct NameToDecimal64OrZero { static constexpr auto name = "toDecimal64OrZero"; }; struct NameToDecimal128OrZero { static constexpr auto name = "toDecimal128OrZero"; }; struct NameToDecimal256OrZero { static constexpr auto name = "toDecimal256OrZero"; }; +struct NameToUUIDOrZero { static constexpr auto name = "toUUIDOrZero"; }; using FunctionToUInt8OrZero = FunctionConvertFromString; using FunctionToUInt16OrZero = FunctionConvertFromString; @@ -1775,6 +1788,7 @@ using FunctionToDecimal32OrZero = FunctionConvertFromString, NameToDecimal64OrZero, ConvertFromStringExceptionMode::Zero>; using FunctionToDecimal128OrZero = FunctionConvertFromString, NameToDecimal128OrZero, ConvertFromStringExceptionMode::Zero>; using FunctionToDecimal256OrZero = FunctionConvertFromString, NameToDecimal256OrZero, ConvertFromStringExceptionMode::Zero>; +using FunctionToUUIDOrZero = FunctionConvertFromString; struct NameToUInt8OrNull { static constexpr auto name = "toUInt8OrNull"; }; struct NameToUInt16OrNull { static constexpr auto name = "toUInt16OrNull"; }; @@ -1796,6 +1810,7 @@ struct NameToDecimal32OrNull { static constexpr auto name = "toDecimal32OrNull"; struct NameToDecimal64OrNull { static constexpr auto name = "toDecimal64OrNull"; }; struct NameToDecimal128OrNull { static constexpr auto name = "toDecimal128OrNull"; }; struct NameToDecimal256OrNull { static constexpr auto name = "toDecimal256OrNull"; }; +struct NameToUUIDOrNull { static constexpr auto name = "toUUIDOrNull"; }; using FunctionToUInt8OrNull = FunctionConvertFromString; using FunctionToUInt16OrNull = FunctionConvertFromString; @@ -1817,6 +1832,7 @@ using FunctionToDecimal32OrNull = FunctionConvertFromString, NameToDecimal64OrNull, ConvertFromStringExceptionMode::Null>; using FunctionToDecimal128OrNull = FunctionConvertFromString, NameToDecimal128OrNull, ConvertFromStringExceptionMode::Null>; using FunctionToDecimal256OrNull = FunctionConvertFromString, NameToDecimal256OrNull, ConvertFromStringExceptionMode::Null>; +using FunctionToUUIDOrNull = FunctionConvertFromString; struct NameParseDateTimeBestEffort { static constexpr auto name = "parseDateTimeBestEffort"; }; struct NameParseDateTimeBestEffortUS { static constexpr auto name = "parseDateTimeBestEffortUS"; }; @@ -1856,7 +1872,7 @@ using FunctionParseDateTime64BestEffortOrNull = FunctionConvertFromString< class ExecutableFunctionCast : public IExecutableFunctionImpl { public: - using WrapperType = std::function; + using WrapperType = std::function; explicit ExecutableFunctionCast(WrapperType && wrapper_function_, const char * name_) : wrapper_function(std::move(wrapper_function_)), name(name_) {} @@ -1864,14 +1880,14 @@ public: String getName() const override { return name; } protected: - void execute(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override + ColumnPtr execute(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) override { /// drop second argument, pass others - ColumnNumbers new_arguments{arguments.front()}; + ColumnsWithTypeAndName new_arguments{arguments.front()}; if (arguments.size() > 2) new_arguments.insert(std::end(new_arguments), std::next(std::begin(arguments), 2), std::end(arguments)); - wrapper_function(columns, new_arguments, result, input_rows_count); + return wrapper_function(new_arguments, result_type, nullptr, input_rows_count); } bool useDefaultImplementationForNulls() const override { return false; } @@ -1890,7 +1906,7 @@ struct NameCast { static constexpr auto name = "CAST"; }; class FunctionCast final : public IFunctionBaseImpl { public: - using WrapperType = std::function; + using WrapperType = std::function; using MonotonicityForRange = std::function; FunctionCast(const char * name_, MonotonicityForRange && monotonicity_for_range_ @@ -1901,12 +1917,12 @@ public: } const DataTypes & getArgumentTypes() const override { return argument_types; } - const DataTypePtr & getReturnType() const override { return return_type; } + const DataTypePtr & getResultType() const override { return return_type; } - ExecutableFunctionImplPtr prepare(const ColumnsWithTypeAndName & /*sample_columns*/, const ColumnNumbers & /*arguments*/, size_t /*result*/) const override + ExecutableFunctionImplPtr prepare(const ColumnsWithTypeAndName & /*sample_columns*/) const override { return std::make_unique( - prepareUnpackDictionaries(getArgumentTypes()[0], getReturnType()), name); + prepareUnpackDictionaries(getArgumentTypes()[0], getResultType()), name); } String getName() const override { return name; } @@ -1950,13 +1966,13 @@ private: FunctionOverloadResolverAdaptor(std::make_unique(function)) .build({ColumnWithTypeAndName{nullptr, from_type, ""}}); - return [function_adaptor] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [function_adaptor] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { - function_adaptor->execute(columns, arguments, result, input_rows_count); + return function_adaptor->execute(arguments, result_type, input_rows_count); }; } - WrapperType createStringWrapper(const DataTypePtr & from_type) const + static WrapperType createStringWrapper(const DataTypePtr & from_type) { FunctionPtr function = FunctionToString::create(); @@ -1964,9 +1980,9 @@ private: FunctionOverloadResolverAdaptor(std::make_unique(function)) .build({ColumnWithTypeAndName{nullptr, from_type, ""}}); - return [function_adaptor] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [function_adaptor] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { - function_adaptor->execute(columns, arguments, result, input_rows_count); + return function_adaptor->execute(arguments, result_type, input_rows_count); }; } @@ -1975,13 +1991,13 @@ private: if (!isStringOrFixedString(from_type)) throw Exception{"CAST AS FixedString is only implemented for types String and FixedString", ErrorCodes::NOT_IMPLEMENTED}; - return [N] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [N] (ColumnsWithTypeAndName & arguments, const DataTypePtr &, const ColumnNullable *, size_t /*input_rows_count*/) { - FunctionToFixedString::executeForN(columns, arguments, result, N); + return FunctionToFixedString::executeForN(arguments, N); }; } - WrapperType createUUIDWrapper(const DataTypePtr & from_type, const DataTypeUUID * const, bool requested_result_is_nullable) const + static WrapperType createUUIDWrapper(const DataTypePtr & from_type, const DataTypeUUID * const, bool requested_result_is_nullable) { if (requested_result_is_nullable) throw Exception{"CAST AS Nullable(UUID) is not implemented", ErrorCodes::NOT_IMPLEMENTED}; @@ -1992,9 +2008,9 @@ private: FunctionOverloadResolverAdaptor(std::make_unique(function)) .build({ColumnWithTypeAndName{nullptr, from_type, ""}}); - return [function_adaptor] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [function_adaptor] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { - function_adaptor->execute(columns, arguments, result, input_rows_count); + return function_adaptor->execute(arguments, result_type, input_rows_count); }; } @@ -2016,15 +2032,16 @@ private: throw Exception{"Conversion from " + from_type->getName() + " to " + to_type->getName() + " is not supported", ErrorCodes::CANNOT_CONVERT_TYPE}; - return [type_index, scale, to_type] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [type_index, scale, to_type] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { + ColumnPtr result_column; auto res = callOnIndexAndDataType(type_index, [&](const auto & types) -> bool { using Types = std::decay_t; using LeftDataType = typename Types::LeftType; using RightDataType = typename Types::RightType; - ConvertImpl::execute(columns, arguments, result, input_rows_count, scale); + result_column = ConvertImpl::execute(arguments, result_type, input_rows_count, scale); return true; }); @@ -2034,17 +2051,19 @@ private: throw Exception{"Conversion from " + std::string(getTypeName(type_index)) + " to " + to_type->getName() + " is not supported", ErrorCodes::CANNOT_CONVERT_TYPE}; } + + return result_column; }; } - WrapperType createAggregateFunctionWrapper(const DataTypePtr & from_type_untyped, const DataTypeAggregateFunction * to_type) const + static WrapperType createAggregateFunctionWrapper(const DataTypePtr & from_type_untyped, const DataTypeAggregateFunction * to_type) { /// Conversion from String through parsing. if (checkAndGetDataType(from_type_untyped.get())) { - return [] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t /*input_rows_count*/) { - ConvertImplGenericFromString::execute(columns, arguments, result); + return ConvertImplGenericFromString::execute(arguments, result_type); }; } else @@ -2057,15 +2076,15 @@ private: /// Conversion from String through parsing. if (checkAndGetDataType(from_type_untyped.get())) { - return [] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t /*input_rows_count*/) { - ConvertImplGenericFromString::execute(columns, arguments, result); + return ConvertImplGenericFromString::execute(arguments, result_type); }; } DataTypePtr from_nested_type; DataTypePtr to_nested_type; - auto from_type = checkAndGetDataType(from_type_untyped.get()); + const auto * from_type = checkAndGetDataType(from_type_untyped.get()); /// get the most nested type if (from_type && to_type) @@ -2086,25 +2105,20 @@ private: const auto nested_function = prepareUnpackDictionaries(from_nested_type, to_nested_type); return [nested_function, from_nested_type, to_nested_type]( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + ColumnsWithTypeAndName & arguments, const DataTypePtr &, const ColumnNullable * nullable_source, size_t /*input_rows_count*/) -> ColumnPtr { - const auto & array_arg = columns[arguments.front()]; + const auto & array_arg = arguments.front(); if (const ColumnArray * col_array = checkAndGetColumn(array_arg.column.get())) { /// create columns for converting nested column containing original and result columns - ColumnsWithTypeAndName nested_columns_columns - { - { col_array->getDataPtr(), from_nested_type, "" }, - { nullptr, to_nested_type, "" } - }; - ColumnsWithTypeAndName nested_columns(nested_columns_columns); + ColumnsWithTypeAndName nested_columns{{ col_array->getDataPtr(), from_nested_type, "" }}; /// convert nested column - nested_function(nested_columns, {0}, 1, nested_columns_columns.front().column->size()); + auto result_column = nested_function(nested_columns, to_nested_type, nullable_source, nested_columns.front().column->size()); /// set converted nested column to result - columns[result].column = ColumnArray::create(nested_columns[1].column, col_array->getOffsetsPtr()); + return ColumnArray::create(result_column, col_array->getOffsetsPtr()); } else throw Exception{"Illegal column " + array_arg.column->getName() + " for function CAST AS Array", ErrorCodes::LOGICAL_ERROR}; @@ -2116,13 +2130,13 @@ private: /// Conversion from String through parsing. if (checkAndGetDataType(from_type_untyped.get())) { - return [] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t /*input_rows_count*/) { - ConvertImplGenericFromString::execute(columns, arguments, result); + return ConvertImplGenericFromString::execute(arguments, result_type); }; } - const auto from_type = checkAndGetDataType(from_type_untyped.get()); + const auto * from_type = checkAndGetDataType(from_type_untyped.get()); if (!from_type) throw Exception{"CAST AS Tuple can only be performed between tuple types or from String.\nLeft type: " + from_type_untyped->getName() + ", right type: " + to_type->getName(), ErrorCodes::TYPE_MISMATCH}; @@ -2141,64 +2155,50 @@ private: element_wrappers.push_back(prepareUnpackDictionaries(idx_type.second, to_element_types[idx_type.first])); return [element_wrappers, from_element_types, to_element_types] - (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + (ColumnsWithTypeAndName & arguments, const DataTypePtr &, const ColumnNullable * nullable_source, size_t input_rows_count) -> ColumnPtr { - const auto col = columns[arguments.front()].column.get(); - - /// copy tuple elements to a separate columns - ColumnsWithTypeAndName element_columns_columns; + const auto * col = arguments.front().column.get(); size_t tuple_size = from_element_types.size(); const ColumnTuple & column_tuple = typeid_cast(*col); - /// create columns for source elements - for (size_t i = 0; i < tuple_size; ++i) - element_columns_columns.emplace_back(ColumnWithTypeAndName{ column_tuple.getColumns()[i], from_element_types[i], "" }); - - /// create columns for converted elements - for (const auto & to_element_type : to_element_types) - element_columns_columns.emplace_back(ColumnWithTypeAndName{ nullptr, to_element_type, "" }); - - /// insert column for converted tuple - element_columns_columns.emplace_back(ColumnWithTypeAndName{ nullptr, std::make_shared(to_element_types), "" }); + Columns converted_columns(tuple_size); /// invoke conversion for each element - for (const auto idx_element_wrapper : ext::enumerate(element_wrappers)) - idx_element_wrapper.second(element_columns_columns, { idx_element_wrapper.first }, - tuple_size + idx_element_wrapper.first, input_rows_count); - - Columns converted_columns(tuple_size); for (size_t i = 0; i < tuple_size; ++i) - converted_columns[i] = element_columns_columns[tuple_size + i].column; + { + ColumnsWithTypeAndName element = {{column_tuple.getColumns()[i], from_element_types[i], "" }}; + converted_columns[i] = element_wrappers[i](element, to_element_types[i], nullable_source, input_rows_count); + } - columns[result].column = ColumnTuple::create(converted_columns); + return ColumnTuple::create(converted_columns); }; } template - WrapperType createEnumWrapper(const DataTypePtr & from_type, const DataTypeEnum * to_type, bool source_is_nullable) const + WrapperType createEnumWrapper(const DataTypePtr & from_type, const DataTypeEnum * to_type) const { using EnumType = DataTypeEnum; using Function = typename FunctionTo::Type; - if (const auto from_enum8 = checkAndGetDataType(from_type.get())) + if (const auto * from_enum8 = checkAndGetDataType(from_type.get())) checkEnumToEnumConversion(from_enum8, to_type); - else if (const auto from_enum16 = checkAndGetDataType(from_type.get())) + else if (const auto * from_enum16 = checkAndGetDataType(from_type.get())) checkEnumToEnumConversion(from_enum16, to_type); if (checkAndGetDataType(from_type.get())) - return createStringToEnumWrapper(source_is_nullable); + return createStringToEnumWrapper(); else if (checkAndGetDataType(from_type.get())) - return createStringToEnumWrapper(source_is_nullable); + return createStringToEnumWrapper(); else if (isNativeNumber(from_type) || isEnum(from_type)) { auto function = Function::create(); auto func_or_adaptor = FunctionOverloadResolverAdaptor(std::make_unique(function)) .build(ColumnsWithTypeAndName{{nullptr, from_type, "" }}); - return [func_or_adaptor] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [func_or_adaptor] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { - func_or_adaptor->execute(columns, arguments, result, input_rows_count); + return func_or_adaptor->execute(arguments, result_type, input_rows_count); }; } else @@ -2232,28 +2232,19 @@ private: } template - WrapperType createStringToEnumWrapper(ssize_t source_is_nullable) const + WrapperType createStringToEnumWrapper() const { const char * function_name = name; - return [function_name, source_is_nullable] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [function_name] ( + ColumnsWithTypeAndName & arguments, const DataTypePtr & res_type, const ColumnNullable * nullable_col, size_t /*input_rows_count*/) { - const auto first_col = columns[arguments.front()].column.get(); - - auto & col_with_type_and_name = columns[result]; - const auto & result_type = typeid_cast(*col_with_type_and_name.type); + const auto & first_col = arguments.front().column.get(); + const auto & result_type = typeid_cast(*res_type); const ColumnStringType * col = typeid_cast(first_col); - const ColumnNullable * nullable_col = nullptr; - if (source_is_nullable) - { - if (columns.size() <= arguments.front() + 1) - throw Exception("Not enough columns", ErrorCodes::LOGICAL_ERROR); - size_t nullable_pos = columns.size() - 1; - nullable_col = typeid_cast(columns[nullable_pos].column.get()); - if (col && nullable_col && nullable_col->size() != col->size()) - throw Exception("ColumnNullable is not compatible with original", ErrorCodes::LOGICAL_ERROR); - } + if (col && nullable_col && nullable_col->size() != col->size()) + throw Exception("ColumnNullable is not compatible with original", ErrorCodes::LOGICAL_ERROR); if (col) { @@ -2277,7 +2268,7 @@ private: out_data[i] = result_type.getValue(col->getDataAt(i)); } - col_with_type_and_name.column = std::move(res); + return res; } else throw Exception{"Unexpected column " + first_col->getName() + " as first argument of function " + function_name, @@ -2285,21 +2276,21 @@ private: }; } - WrapperType createIdentityWrapper(const DataTypePtr &) const + static WrapperType createIdentityWrapper(const DataTypePtr &) { - return [] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t /*input_rows_count*/) + return [] (ColumnsWithTypeAndName & arguments, const DataTypePtr &, const ColumnNullable *, size_t /*input_rows_count*/) { - columns[result].column = columns[arguments.front()].column; + return arguments.front().column; }; } - WrapperType createNothingWrapper(const IDataType * to_type) const + static WrapperType createNothingWrapper(const IDataType * to_type) { ColumnPtr res = to_type->createColumnConstWithDefaultValue(1); - return [res] (ColumnsWithTypeAndName & columns, const ColumnNumbers &, const size_t result, size_t input_rows_count) + return [res] (ColumnsWithTypeAndName &, const DataTypePtr &, const ColumnNullable *, size_t input_rows_count) { /// Column of Nothing type is trivially convertible to any other column - columns[result].column = res->cloneResized(input_rows_count)->convertToFullColumnIfConst(); + return res->cloneResized(input_rows_count)->convertToFullColumnIfConst(); }; } @@ -2315,10 +2306,9 @@ private: if (!to_nested->isNullable()) throw Exception{"Cannot convert NULL to a non-nullable type", ErrorCodes::CANNOT_CONVERT_TYPE}; - return [](ColumnsWithTypeAndName & columns, const ColumnNumbers &, const size_t result, size_t input_rows_count) + return [](ColumnsWithTypeAndName &, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) { - auto & res = columns[result]; - res.column = res.type->createColumnConstWithDefaultValue(input_rows_count)->convertToFullColumnIfConst(); + return result_type->createColumnConstWithDefaultValue(input_rows_count)->convertToFullColumnIfConst(); }; } @@ -2333,30 +2323,27 @@ private: return wrapper; return [wrapper, from_low_cardinality, to_low_cardinality, skip_not_null_check] - (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable * nullable_source, size_t input_rows_count) -> ColumnPtr { - auto & arg = columns[arguments[0]]; - auto & res = columns[result]; + ColumnsWithTypeAndName args = {arguments[0]}; + auto & arg = args.front(); + auto res_type = result_type; + + ColumnPtr converted_column; ColumnPtr res_indexes; /// For some types default can't be casted (for example, String to Int). In that case convert column to full. bool src_converted_to_full_column = false; { - /// Replace argument and result columns (and types) to dictionary key columns (and types). - /// Call nested wrapper in order to cast dictionary keys. Then restore columns. - auto prev_arg_col = arg.column; - auto prev_arg_type = arg.type; - auto prev_res_type = res.type; - auto tmp_rows_count = input_rows_count; if (to_low_cardinality) - res.type = to_low_cardinality->getDictionaryType(); + res_type = to_low_cardinality->getDictionaryType(); if (from_low_cardinality) { - auto * col_low_cardinality = typeid_cast(prev_arg_col.get()); + const auto * col_low_cardinality = typeid_cast(arguments[0].column.get()); if (skip_not_null_check && col_low_cardinality->containsNull()) throw Exception{"Cannot convert NULL value to non-Nullable type", @@ -2366,7 +2353,7 @@ private: arg.type = from_low_cardinality->getDictionaryType(); /// TODO: Make map with defaults conversion. - src_converted_to_full_column = !removeNullable(arg.type)->equals(*removeNullable(res.type)); + src_converted_to_full_column = !removeNullable(arg.type)->equals(*removeNullable(res_type)); if (src_converted_to_full_column) arg.column = arg.column->index(col_low_cardinality->getIndexes(), 0); else @@ -2376,11 +2363,7 @@ private: } /// Perform the requested conversion. - wrapper(columns, arguments, result, tmp_rows_count); - - arg.column = prev_arg_col; - arg.type = prev_arg_type; - res.type = prev_res_type; + converted_column = wrapper(args, res_type, nullable_source, tmp_rows_count); } if (to_low_cardinality) @@ -2390,16 +2373,17 @@ private: if (from_low_cardinality && !src_converted_to_full_column) { - auto res_keys = std::move(res.column); - col_low_cardinality->insertRangeFromDictionaryEncodedColumn(*res_keys, *res_indexes); + col_low_cardinality->insertRangeFromDictionaryEncodedColumn(*converted_column, *res_indexes); } else - col_low_cardinality->insertRangeFromFullColumn(*res.column, 0, res.column->size()); + col_low_cardinality->insertRangeFromFullColumn(*converted_column, 0, converted_column->size()); - res.column = std::move(res_column); + return res_column; } else if (!src_converted_to_full_column) - res.column = res.column->index(*res_indexes, 0); + return converted_column->index(*res_indexes, 0); + else + return converted_column; }; } @@ -2410,62 +2394,60 @@ private: bool source_is_nullable = from_type->isNullable(); bool result_is_nullable = to_type->isNullable(); - auto wrapper = prepareImpl(removeNullable(from_type), removeNullable(to_type), result_is_nullable, source_is_nullable); + auto wrapper = prepareImpl(removeNullable(from_type), removeNullable(to_type), result_is_nullable); if (result_is_nullable) { return [wrapper, source_is_nullable] - (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) -> ColumnPtr { /// Create a temporary columns on which to perform the operation. - auto & res = columns[result]; - const auto & ret_type = res.type; - const auto & nullable_type = static_cast(*ret_type); + const auto & nullable_type = static_cast(*result_type); const auto & nested_type = nullable_type.getNestedType(); - ColumnsWithTypeAndName tmp_columns_columns; + ColumnsWithTypeAndName tmp_args; if (source_is_nullable) - tmp_columns_columns = createBlockWithNestedColumns(columns, arguments); + tmp_args = createBlockWithNestedColumns(arguments); else - tmp_columns_columns = columns; + tmp_args = arguments; + + const ColumnNullable * nullable_source = nullptr; - size_t tmp_res_index = columns.size(); - tmp_columns_columns.emplace_back(ColumnWithTypeAndName {nullptr, nested_type, ""}); /// Add original ColumnNullable for createStringToEnumWrapper() if (source_is_nullable) { if (arguments.size() != 1) throw Exception("Invalid number of arguments", ErrorCodes::LOGICAL_ERROR); - tmp_columns_columns.emplace_back(columns[arguments.front()]); + nullable_source = typeid_cast(arguments.front().column.get()); } /// Perform the requested conversion. - wrapper(tmp_columns_columns, arguments, tmp_res_index, input_rows_count); - - const auto & tmp_res = tmp_columns_columns[tmp_res_index]; + auto tmp_res = wrapper(tmp_args, nested_type, nullable_source, input_rows_count); /// May happen in fuzzy tests. For debug purpose. - if (!tmp_res.column) - throw Exception("Couldn't convert " + columns[arguments[0]].type->getName() + " to " + if (!tmp_res) + throw Exception("Couldn't convert " + arguments[0].type->getName() + " to " + nested_type->getName() + " in " + " prepareRemoveNullable wrapper.", ErrorCodes::LOGICAL_ERROR); - res.column = wrapInNullable(tmp_res.column, {columns[arguments[0]], tmp_res}, {0}, 1, input_rows_count); + return wrapInNullable(tmp_res, arguments, nested_type, input_rows_count); }; } else if (source_is_nullable) { /// Conversion from Nullable to non-Nullable. - return [wrapper, skip_not_null_check] (ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, size_t input_rows_count) + return [wrapper, skip_not_null_check] + (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable *, size_t input_rows_count) -> ColumnPtr { - auto tmp_columns_columns = createBlockWithNestedColumns(columns, arguments, result); + auto tmp_args = createBlockWithNestedColumns(arguments); + auto nested_type = removeNullable(result_type); /// Check that all values are not-NULL. /// Check can be skipped in case if LowCardinality dictionary is transformed. /// In that case, correctness will be checked beforehand. if (!skip_not_null_check) { - const auto & col = columns[arguments[0]].column; + const auto & col = arguments[0].column; const auto & nullable_col = assert_cast(*col); const auto & null_map = nullable_col.getNullMapData(); @@ -2473,9 +2455,8 @@ private: throw Exception{"Cannot convert NULL value to non-Nullable type", ErrorCodes::CANNOT_INSERT_NULL_IN_ORDINARY_COLUMN}; } - - wrapper(tmp_columns_columns, arguments, result, input_rows_count); - columns[result].column = tmp_columns_columns[result].column; + const ColumnNullable * nullable_source = typeid_cast(arguments.front().column.get()); + return wrapper(tmp_args, nested_type, nullable_source, input_rows_count); }; } else @@ -2484,7 +2465,7 @@ private: /// 'from_type' and 'to_type' are nested types in case of Nullable. /// 'requested_result_is_nullable' is true if CAST to Nullable type is requested. - WrapperType prepareImpl(const DataTypePtr & from_type, const DataTypePtr & to_type, bool requested_result_is_nullable, bool source_is_nullable) const + WrapperType prepareImpl(const DataTypePtr & from_type, const DataTypePtr & to_type, bool requested_result_is_nullable) const { if (from_type->equals(*to_type)) return createIdentityWrapper(from_type); @@ -2522,7 +2503,7 @@ private: std::is_same_v || std::is_same_v) { - ret = createEnumWrapper(from_type, checkAndGetDataType(to_type.get()), source_is_nullable); + ret = createEnumWrapper(from_type, checkAndGetDataType(to_type.get())); return true; } if constexpr ( @@ -2583,7 +2564,7 @@ public: static FunctionOverloadResolverImplPtr create(const Context & context); static FunctionOverloadResolverImplPtr createImpl(bool keep_nullable) { return std::make_unique(keep_nullable); } - CastOverloadResolver(bool keep_nullable_) + explicit CastOverloadResolver(bool keep_nullable_) : keep_nullable(keep_nullable_) {} @@ -2614,7 +2595,7 @@ protected: " Instead there is non-constant column of type " + arguments.back().type->getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - const auto type_col = checkAndGetColumnConst(column.get()); + const auto * type_col = checkAndGetColumnConst(column.get()); if (!type_col) throw Exception("Second argument to " + getName() + " must be a constant string describing type." " Instead there is a column with the following structure: " + column->dumpStructure(), @@ -2638,45 +2619,45 @@ private: return FunctionTo::Type::Monotonic::get; } - MonotonicityForRange getMonotonicityInformation(const DataTypePtr & from_type, const IDataType * to_type) const + static MonotonicityForRange getMonotonicityInformation(const DataTypePtr & from_type, const IDataType * to_type) { - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); if (isEnum(from_type)) { - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); - if (const auto type = checkAndGetDataType(to_type)) + if (const auto * type = checkAndGetDataType(to_type)) return monotonicityForType(type); } /// other types like Null, FixedString, Array and Tuple have no monotonicity defined diff --git a/src/Functions/FunctionsEmbeddedDictionaries.h b/src/Functions/FunctionsEmbeddedDictionaries.h index e74b9c9f5f0..7c1221601f6 100644 --- a/src/Functions/FunctionsEmbeddedDictionaries.h +++ b/src/Functions/FunctionsEmbeddedDictionaries.h @@ -183,17 +183,17 @@ public: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { /// The dictionary key that defines the "point of view". std::string dict_key; if (arguments.size() == 2) { - const ColumnConst * key_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const ColumnConst * key_col = checkAndGetColumnConst(arguments[1].column.get()); if (!key_col) - throw Exception("Illegal column " + columns[arguments[1]].column->getName() + throw Exception("Illegal column " + arguments[1].column->getName() + " of second ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); @@ -203,7 +203,7 @@ public: const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); - if (const ColumnVector * col_from = checkAndGetColumn>(columns[arguments[0]].column.get())) + if (const ColumnVector * col_from = checkAndGetColumn>(arguments[0].column.get())) { auto col_to = ColumnVector::create(); @@ -215,10 +215,10 @@ public: for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from[i], dict); - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + name, ErrorCodes::ILLEGAL_COLUMN); } @@ -279,17 +279,17 @@ public: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { /// The dictionary key that defines the "point of view". std::string dict_key; if (arguments.size() == 3) { - const ColumnConst * key_col = checkAndGetColumnConst(columns[arguments[2]].column.get()); + const ColumnConst * key_col = checkAndGetColumnConst(arguments[2].column.get()); if (!key_col) - throw Exception("Illegal column " + columns[arguments[2]].column->getName() + throw Exception("Illegal column " + arguments[2].column->getName() + " of third ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); @@ -299,10 +299,10 @@ public: const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); - const ColumnVector * col_vec1 = checkAndGetColumn>(columns[arguments[0]].column.get()); - const ColumnVector * col_vec2 = checkAndGetColumn>(columns[arguments[1]].column.get()); - const ColumnConst * col_const1 = checkAndGetColumnConst>(columns[arguments[0]].column.get()); - const ColumnConst * col_const2 = checkAndGetColumnConst>(columns[arguments[1]].column.get()); + const ColumnVector * col_vec1 = checkAndGetColumn>(arguments[0].column.get()); + const ColumnVector * col_vec2 = checkAndGetColumn>(arguments[1].column.get()); + const ColumnConst * col_const1 = checkAndGetColumnConst>(arguments[0].column.get()); + const ColumnConst * col_const2 = checkAndGetColumnConst>(arguments[1].column.get()); if (col_vec1 && col_vec2) { @@ -317,7 +317,7 @@ public: for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from1[i], vec_from2[i], dict); - columns[result].column = std::move(col_to); + return col_to; } else if (col_vec1 && col_const2) { @@ -332,7 +332,7 @@ public: for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(vec_from1[i], const_from2, dict); - columns[result].column = std::move(col_to); + return col_to; } else if (col_const1 && col_vec2) { @@ -347,16 +347,16 @@ public: for (size_t i = 0; i < size; ++i) vec_to[i] = Transform::apply(const_from1, vec_from2[i], dict); - columns[result].column = std::move(col_to); + return col_to; } else if (col_const1 && col_const2) { - columns[result].column = DataTypeUInt8().createColumnConst(col_const1->size(), + return DataTypeUInt8().createColumnConst(col_const1->size(), toField(Transform::apply(col_const1->template getValue(), col_const2->template getValue(), dict))); } else - throw Exception("Illegal columns " + columns[arguments[0]].column->getName() - + " and " + columns[arguments[1]].column->getName() + throw Exception("Illegal columns " + arguments[0].column->getName() + + " and " + arguments[1].column->getName() + " of arguments of function " + name, ErrorCodes::ILLEGAL_COLUMN); } @@ -415,17 +415,17 @@ public: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { /// The dictionary key that defines the "point of view". std::string dict_key; if (arguments.size() == 2) { - const ColumnConst * key_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const ColumnConst * key_col = checkAndGetColumnConst(arguments[1].column.get()); if (!key_col) - throw Exception("Illegal column " + columns[arguments[1]].column->getName() + throw Exception("Illegal column " + arguments[1].column->getName() + " of second ('point of view') argument of function " + name + ". Must be constant string.", ErrorCodes::ILLEGAL_COLUMN); @@ -435,7 +435,7 @@ public: const typename DictGetter::Dst & dict = DictGetter::get(*owned_dict, dict_key); - if (const ColumnVector * col_from = checkAndGetColumn>(columns[arguments[0]].column.get())) + if (const ColumnVector * col_from = checkAndGetColumn>(arguments[0].column.get())) { auto col_values = ColumnVector::create(); auto col_offsets = ColumnArray::ColumnOffsets::create(); @@ -459,10 +459,10 @@ public: res_offsets[i] = res_values.size(); } - columns[result].column = ColumnArray::create(std::move(col_values), std::move(col_offsets)); + return ColumnArray::create(std::move(col_values), std::move(col_offsets)); } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + name, ErrorCodes::ILLEGAL_COLUMN); } @@ -620,24 +620,24 @@ public: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { RegionsNames::Language language = RegionsNames::Language::ru; /// If the result language is specified if (arguments.size() == 2) { - if (const ColumnConst * col_language = checkAndGetColumnConst(columns[arguments[1]].column.get())) + if (const ColumnConst * col_language = checkAndGetColumnConst(arguments[1].column.get())) language = RegionsNames::getLanguageEnum(col_language->getValue()); else - throw Exception("Illegal column " + columns[arguments[1]].column->getName() + throw Exception("Illegal column " + arguments[1].column->getName() + " of the second argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } const RegionsNames & dict = *owned_dict; - if (const ColumnUInt32 * col_from = typeid_cast(columns[arguments[0]].column.get())) + if (const ColumnUInt32 * col_from = typeid_cast(arguments[0].column.get())) { auto col_to = ColumnString::create(); @@ -649,10 +649,10 @@ public: col_to->insertDataWithTerminatingZero(name_ref.data, name_ref.size + 1); } - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of the first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } diff --git a/src/Functions/FunctionsExternalDictionaries.h b/src/Functions/FunctionsExternalDictionaries.h index 4affcdfa7e0..0fae3de1fb2 100644 --- a/src/Functions/FunctionsExternalDictionaries.h +++ b/src/Functions/FunctionsExternalDictionaries.h @@ -78,23 +78,23 @@ namespace ErrorCodes class FunctionDictHelper { public: - FunctionDictHelper(const Context & context_) : context(context_), external_loader(context.getExternalDictionariesLoader()) {} + explicit FunctionDictHelper(const Context & context_) : context(context_), external_loader(context.getExternalDictionariesLoader()) {} std::shared_ptr getDictionary(const String & dictionary_name) { - auto dict = std::atomic_load(&dictionary); - if (dict) - return dict; String resolved_name = DatabaseCatalog::instance().resolveDictionaryName(dictionary_name); - dict = external_loader.getDictionary(resolved_name); - context.checkAccess(AccessType::dictGet, dict->getDatabaseOrNoDatabaseTag(), dict->getDictionaryID().getTableName()); - std::atomic_store(&dictionary, dict); + auto dict = external_loader.getDictionary(resolved_name); + if (!access_checked) + { + context.checkAccess(AccessType::dictGet, dict->getDatabaseOrNoDatabaseTag(), dict->getDictionaryID().getTableName()); + access_checked = true; + } return dict; } std::shared_ptr getDictionary(const ColumnWithTypeAndName & column) { - const auto dict_name_col = checkAndGetColumnConst(column.column.get()); + const auto * dict_name_col = checkAndGetColumnConst(column.column.get()); return getDictionary(dict_name_col->getValue()); } @@ -107,11 +107,11 @@ public: if (sample_columns.columns() != 3 && sample_columns.columns() != 4) throw Exception{"Function dictGet... takes 3 or 4 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto dict_name_col = checkAndGetColumnConst(sample_columns.getByPosition(0).column.get()); + const auto * dict_name_col = checkAndGetColumnConst(sample_columns.getByPosition(0).column.get()); if (!dict_name_col) throw Exception{"First argument of function dictGet... must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; - const auto attr_name_col = checkAndGetColumnConst(sample_columns.getByPosition(1).column.get()); + const auto * attr_name_col = checkAndGetColumnConst(sample_columns.getByPosition(1).column.get()); if (!attr_name_col) throw Exception{"Second argument of function dictGet... must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; @@ -122,6 +122,8 @@ private: const Context & context; const ExternalDictionariesLoader & external_loader; mutable std::shared_ptr dictionary; + /// Access cannot be not granted, since in this case checkAccess() will throw and access_checked will not be updated. + std::atomic access_checked = false; }; @@ -135,7 +137,7 @@ public: return std::make_shared(context); } - FunctionDictHas(const Context & context_) : helper(context_) {} + explicit FunctionDictHas(const Context & context_) : helper(context_) {} String getName() const override { return name; } @@ -161,7 +163,7 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { /** Do not require existence of the dictionary if the function is called for empty columns. * This is needed to allow successful query analysis on a server, @@ -171,68 +173,67 @@ private: * for backward compatibility with the case in Yandex Banner System. */ if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); + ColumnPtr res; - if (!executeDispatchSimple(columns, arguments, result, dict) && - !executeDispatchSimple(columns, arguments, result, dict) && - !executeDispatchSimple(columns, arguments, result, dict) && - !executeDispatchSimple(columns, arguments, result, dict) && + if (!((res = executeDispatchSimple(arguments, dict)) + || (res = executeDispatchSimple(arguments, dict)) + || (res = executeDispatchSimple(arguments, dict)) + || (res = executeDispatchSimple(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchSimple(columns, arguments, result, dict) && + || (res = executeDispatchSimple(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif #if !defined(ARCADIA_BUILD) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatchSimple( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchSimple( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const auto id_col_untyped = columns[arguments[1]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) + const auto * id_col_untyped = arguments[1].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) { const auto & ids = id_col->getData(); auto out = ColumnUInt8::create(ext::size(ids)); dict->has(ids, out->getData()); - columns[result].column = std::move(out); + return out; } else throw Exception{"Second argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - return true; + return nullptr; } template - bool executeDispatchComplex( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchComplex( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const ColumnWithTypeAndName & key_col_with_type = columns[arguments[1]]; + const ColumnWithTypeAndName & key_col_with_type = arguments[1]; const ColumnPtr & key_col = key_col_with_type.column; if (checkColumn(key_col.get())) @@ -242,15 +243,12 @@ private: auto out = ColumnUInt8::create(key_col_with_type.column->size()); dict->has(key_columns, key_types, out->getData()); - columns[result].column = std::move(out); + return out; } else throw Exception{"Second argument of function " + getName() + " must be " + dict->getKeyDescription(), ErrorCodes::TYPE_MISMATCH}; - - return true; } -private: mutable FunctionDictHelper helper; }; @@ -272,7 +270,7 @@ public: return std::make_shared(context); } - FunctionDictGetString(const Context & context_) : helper(context_) {} + explicit FunctionDictGetString(const Context & context_) : helper(context_) {} String getName() const override { return name; } @@ -326,90 +324,87 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); + ColumnPtr res; - if (!executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && + if (!((res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatch(columns, arguments, result, dict) && + || (res = executeDispatch(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif #if !defined(ARCADIA_BUILD) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchRange(columns, arguments, result, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchRange(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 3) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 3 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto id_col_untyped = columns[arguments[2]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) + const auto * id_col_untyped = arguments[2].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) { auto out = ColumnString::create(); dict->getString(attr_name, id_col->getData(), out.get()); - columns[result].column = std::move(out); + return out; } else throw Exception{"Third argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - - return true; } template - bool executeDispatchComplex( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchComplex( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 3) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 3 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const ColumnWithTypeAndName & key_col_with_type = columns[arguments[2]]; + const ColumnWithTypeAndName & key_col_with_type = arguments[2]; /// Functions in external dictionaries_loader only support full-value (not constant) columns with keys. ColumnPtr key_col = key_col_with_type.column->convertToFullColumnIfConst(); @@ -420,34 +415,32 @@ private: auto out = ColumnString::create(); dict->getString(attr_name, key_columns, key_types, out.get()); - columns[result].column = std::move(out); + return out; } else throw Exception{"Third argument of function " + getName() + " must be " + dict->getKeyDescription(), ErrorCodes::TYPE_MISMATCH}; - - return true; } template - bool executeDispatchRange( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchRange( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 4) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 4 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto & id_col_untyped = columns[arguments[2]].column; - const auto & range_col_untyped = columns[arguments[3]].column; + const auto & id_col_untyped = arguments[2].column; + const auto & range_col_untyped = arguments[3].column; PaddedPODArray id_col_values_storage; PaddedPODArray range_col_values_storage; @@ -456,12 +449,9 @@ private: auto out = ColumnString::create(); dict->getString(attr_name, id_col_values, range_col_values, out.get()); - columns[result].column = std::move(out); - - return true; + return out; } -private: mutable FunctionDictHelper helper; }; @@ -476,7 +466,7 @@ public: return std::make_shared(context); } - FunctionDictGetStringOrDefault(const Context & context_) : helper(context_) {} + explicit FunctionDictGetStringOrDefault(const Context & context_) : helper(context_) {} String getName() const override { return name; } @@ -512,100 +502,96 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); - if (!executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && + ColumnPtr res; + if (!((res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatch(columns, arguments, result, dict) && + || (res = executeDispatch(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif #if !defined(ARCADIA_BUILD) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch(const ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto id_col_untyped = columns[arguments[2]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) - executeDispatch(columns, arguments, result, dict, attr_name, id_col); - else if (const auto id_col_const = checkAndGetColumnConst>(id_col_untyped)) - executeDispatch(columns, arguments, result, dict, attr_name, id_col_const); + const auto * id_col_untyped = arguments[2].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) + return executeDispatch(arguments, dict, attr_name, id_col); + else if (const auto * id_col_const = checkAndGetColumnConst>(id_col_untyped)) + return executeDispatch(arguments, dict, attr_name, id_col_const); else throw Exception{"Third argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - - return true; } template - void executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const DictionaryType * dict, + ColumnPtr executeDispatch( + const ColumnsWithTypeAndName & arguments, const DictionaryType * dict, const std::string & attr_name, const ColumnUInt64 * id_col) const { - const auto default_col_untyped = columns[arguments[3]].column.get(); + const auto * default_col_untyped = arguments[3].column.get(); - if (const auto default_col = checkAndGetColumn(default_col_untyped)) + if (const auto * default_col = checkAndGetColumn(default_col_untyped)) { /// vector ids, vector defaults auto out = ColumnString::create(); const auto & ids = id_col->getData(); dict->getString(attr_name, ids, default_col, out.get()); - columns[result].column = std::move(out); + return out; } - else if (const auto default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) + else if (const auto * default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) { /// vector ids, const defaults auto out = ColumnString::create(); const auto & ids = id_col->getData(); String def = default_col_const->getValue(); dict->getString(attr_name, ids, def, out.get()); - columns[result].column = std::move(out); + return out; } else throw Exception{"Fourth argument of function " + getName() + " must be String", ErrorCodes::ILLEGAL_COLUMN}; } template - void executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const DictionaryType * dict, + ColumnPtr executeDispatch( + const ColumnsWithTypeAndName & arguments, const DictionaryType * dict, const std::string & attr_name, const ColumnConst * id_col) const { - const auto default_col_untyped = columns[arguments[3]].column.get(); + const auto * default_col_untyped = arguments[3].column.get(); - if (const auto default_col = checkAndGetColumn(default_col_untyped)) + if (const auto * default_col = checkAndGetColumn(default_col_untyped)) { /// const ids, vector defaults const PaddedPODArray ids(1, id_col->getValue()); @@ -615,39 +601,39 @@ private: { auto out = ColumnString::create(); dict->getString(attr_name, ids, String(), out.get()); - columns[result].column = DataTypeString().createColumnConst(id_col->size(), out->getDataAt(0).toString()); + return DataTypeString().createColumnConst(id_col->size(), out->getDataAt(0).toString()); } else - columns[result].column = columns[arguments[3]].column; // reuse the default column + return arguments[3].column; // reuse the default column } - else if (const auto default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) + else if (const auto * default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) { /// const ids, const defaults const PaddedPODArray ids(1, id_col->getValue()); auto out = ColumnString::create(); String def = default_col_const->getValue(); dict->getString(attr_name, ids, def, out.get()); - columns[result].column = DataTypeString().createColumnConst(id_col->size(), out->getDataAt(0).toString()); + return DataTypeString().createColumnConst(id_col->size(), out->getDataAt(0).toString()); } else throw Exception{"Fourth argument of function " + getName() + " must be String", ErrorCodes::ILLEGAL_COLUMN}; } template - bool executeDispatchComplex( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchComplex( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const ColumnWithTypeAndName & key_col_with_type = columns[arguments[2]]; + const ColumnWithTypeAndName & key_col_with_type = arguments[2]; /// Functions in external dictionaries_loader only support full-value (not constant) columns with keys. ColumnPtr key_col = key_col_with_type.column->convertToFullColumnIfConst(); @@ -656,12 +642,12 @@ private: auto out = ColumnString::create(); - const auto default_col_untyped = columns[arguments[3]].column.get(); - if (const auto default_col = checkAndGetColumn(default_col_untyped)) + const auto * default_col_untyped = arguments[3].column.get(); + if (const auto * default_col = checkAndGetColumn(default_col_untyped)) { dict->getString(attr_name, key_columns, key_types, default_col, out.get()); } - else if (const auto default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) + else if (const auto * default_col_const = checkAndGetColumnConstStringOrFixedString(default_col_untyped)) { String def = default_col_const->getValue(); dict->getString(attr_name, key_columns, key_types, def, out.get()); @@ -669,8 +655,7 @@ private: else throw Exception{"Fourth argument of function " + getName() + " must be String", ErrorCodes::ILLEGAL_COLUMN}; - columns[result].column = std::move(out); - return true; + return out; } mutable FunctionDictHelper helper; @@ -799,7 +784,7 @@ public: return std::make_shared(context, dec_scale); } - FunctionDictGet(const Context & context_, UInt32 dec_scale = 0) + explicit FunctionDictGet(const Context & context_, UInt32 dec_scale = 0) : helper(context_) , decimal_scale(dec_scale) {} @@ -838,7 +823,7 @@ private: if (arguments.size() == 4) { - const auto range_argument = arguments[3].get(); + const auto * range_argument = arguments[3].get(); if (!(range_argument->isValueRepresentedByInteger() && range_argument->getSizeOfValueInMemory() <= sizeof(Int64))) throw Exception{"Illegal type " + range_argument->getName() + " of fourth argument of function " + getName() @@ -854,59 +839,58 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); - if (!executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && + ColumnPtr res; + if (!((res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatch(columns, arguments, result, dict) && + || (res = executeDispatch(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif #if !defined(ARCADIA_BUILD) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchRange(columns, arguments, result, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchRange(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch(ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 3) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 3 arguments.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto id_col_untyped = columns[arguments[2]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) + const auto * id_col_untyped = arguments[2].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) { typename ColVec::MutablePtr out; if constexpr (IsDataTypeDecimal) @@ -916,9 +900,9 @@ private: const auto & ids = id_col->getData(); auto & data = out->getData(); DictGetTraits::get(dict, attr_name, ids, data); - columns[result].column = std::move(out); + return out; } - else if (const auto id_col_const = checkAndGetColumnConst>(id_col_untyped)) + else if (const auto * id_col_const = checkAndGetColumnConst>(id_col_untyped)) { const PaddedPODArray ids(1, id_col_const->getValue()); @@ -926,42 +910,39 @@ private: { DecimalPaddedPODArray data(1, decimal_scale); DictGetTraits::get(dict, attr_name, ids, data); - columns[result].column = - DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( + return DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( id_col_const->size(), toField(data.front(), decimal_scale)); } else { PaddedPODArray data(1); DictGetTraits::get(dict, attr_name, ids, data); - columns[result].column = DataTypeNumber().createColumnConst(id_col_const->size(), toField(data.front())); + return DataTypeNumber().createColumnConst(id_col_const->size(), toField(data.front())); } } else throw Exception{"Third argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - - return true; } template - bool executeDispatchComplex( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchComplex( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 3) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 3 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const ColumnWithTypeAndName & key_col_with_type = columns[arguments[2]]; + const ColumnWithTypeAndName & key_col_with_type = arguments[2]; /// Functions in external dictionaries_loader only support full-value (not constant) columns with keys. ColumnPtr key_col = key_col_with_type.column->convertToFullColumnIfConst(); @@ -978,34 +959,32 @@ private: out = ColVec::create(key_columns.front()->size()); auto & data = out->getData(); DictGetTraits::get(dict, attr_name, key_columns, key_types, data); - columns[result].column = std::move(out); + return out; } else throw Exception{"Third argument of function " + getName() + " must be " + dict->getKeyDescription(), ErrorCodes::TYPE_MISMATCH}; - - return true; } template - bool executeDispatchRange( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchRange( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (arguments.size() != 4) throw Exception{"Function " + getName() + " for dictionary of type " + dict->getTypeName() + " requires exactly 4 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto & id_col_untyped = columns[arguments[2]].column; - const auto & range_col_untyped = columns[arguments[3]].column; + const auto & id_col_untyped = arguments[2].column; + const auto & range_col_untyped = arguments[3].column; PaddedPODArray id_col_values_storage; PaddedPODArray range_col_values_storage; @@ -1019,9 +998,7 @@ private: out = ColVec::create(id_col_untyped->size()); auto & data = out->getData(); DictGetTraits::get(dict, attr_name, id_col_values, range_col_values, data); - columns[result].column = std::move(out); - - return true; + return out; } mutable FunctionDictHelper helper; @@ -1077,7 +1054,7 @@ public: return std::make_shared(context, dec_scale); } - FunctionDictGetOrDefault(const Context & context_, UInt32 dec_scale = 0) + explicit FunctionDictGetOrDefault(const Context & context_, UInt32 dec_scale = 0) : helper(context_) , decimal_scale(dec_scale) {} @@ -1117,69 +1094,66 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); + ColumnPtr res; - if (!executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && + if (!((res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatch(columns, arguments, result, dict) && + || (res = executeDispatch(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) #if defined(OS_LINUX) || defined(__FreeBSD__) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif #if !defined(ARCADIA_BUILD) - !executeDispatchComplex(columns, arguments, result, dict) && + || (res = executeDispatchComplex(arguments, dict)) #endif - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict) && - !executeDispatchComplex(columns, arguments, result, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)) + || (res = executeDispatchComplex(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch(ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const auto id_col_untyped = columns[arguments[2]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) - executeDispatch(columns, arguments, result, dict, attr_name, id_col); - else if (const auto id_col_const = checkAndGetColumnConst>(id_col_untyped)) - executeDispatch(columns, arguments, result, dict, attr_name, id_col_const); + const auto * id_col_untyped = arguments[2].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) + return executeDispatch(arguments, dict, attr_name, id_col); + else if (const auto * id_col_const = checkAndGetColumnConst>(id_col_untyped)) + return executeDispatch(arguments, dict, attr_name, id_col_const); else throw Exception{"Third argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - - return true; } template - void executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const DictionaryType * dict, + ColumnPtr executeDispatch( + ColumnsWithTypeAndName & arguments, const DictionaryType * dict, const std::string & attr_name, const ColumnUInt64 * id_col) const { - const auto default_col_untyped = columns[arguments[3]].column.get(); + const auto * default_col_untyped = arguments[3].column.get(); if (const auto default_col = checkAndGetColumn(default_col_untyped)) { @@ -1193,7 +1167,7 @@ private: auto & data = out->getData(); const auto & defs = default_col->getData(); DictGetTraits::getOrDefault(dict, attr_name, ids, defs, data); - columns[result].column = std::move(out); + return out; } else if (const auto default_col_const = checkAndGetColumnConst(default_col_untyped)) { @@ -1207,18 +1181,18 @@ private: auto & data = out->getData(); const auto def = default_col_const->template getValue(); DictGetTraits::getOrDefault(dict, attr_name, ids, def, data); - columns[result].column = std::move(out); + return out; } else throw Exception{"Fourth argument of function " + getName() + " must be " + TypeName::get(), ErrorCodes::ILLEGAL_COLUMN}; } template - void executeDispatch( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const DictionaryType * dict, + ColumnPtr executeDispatch( + ColumnsWithTypeAndName & arguments, const DictionaryType * dict, const std::string & attr_name, const ColumnConst * id_col) const { - const auto default_col_untyped = columns[arguments[3]].column.get(); + const auto * default_col_untyped = arguments[3].column.get(); if (const auto default_col = checkAndGetColumn(default_col_untyped)) { @@ -1232,19 +1206,18 @@ private: { DecimalPaddedPODArray data(1, decimal_scale); DictGetTraits::getOrDefault(dict, attr_name, ids, Type(), data); - columns[result].column = - DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( + return DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( id_col->size(), toField(data.front(), decimal_scale)); } else { PaddedPODArray data(1); DictGetTraits::getOrDefault(dict, attr_name, ids, Type(), data); - columns[result].column = DataType().createColumnConst(id_col->size(), toField(data.front())); + return DataType().createColumnConst(id_col->size(), toField(data.front())); } } else - columns[result].column = columns[arguments[3]].column; // reuse the default column + return arguments[3].column; // reuse the default column } else if (const auto default_col_const = checkAndGetColumnConst(default_col_untyped)) { @@ -1256,8 +1229,7 @@ private: DecimalPaddedPODArray data(1, decimal_scale); const auto & def = default_col_const->template getValue(); DictGetTraits::getOrDefault(dict, attr_name, ids, def, data); - columns[result].column = - DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( + return DataType(DataType::maxPrecision(), decimal_scale).createColumnConst( id_col->size(), toField(data.front(), decimal_scale)); } else @@ -1265,7 +1237,7 @@ private: PaddedPODArray data(1); const auto & def = default_col_const->template getValue(); DictGetTraits::getOrDefault(dict, attr_name, ids, def, data); - columns[result].column = DataType().createColumnConst(id_col->size(), toField(data.front())); + return DataType().createColumnConst(id_col->size(), toField(data.front())); } } else @@ -1273,20 +1245,20 @@ private: } template - bool executeDispatchComplex( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatchComplex( + ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; - const auto attr_name_col = checkAndGetColumnConst(columns[arguments[1]].column.get()); + const auto * attr_name_col = checkAndGetColumnConst(arguments[1].column.get()); if (!attr_name_col) throw Exception{"Second argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN}; String attr_name = attr_name_col->getValue(); - const ColumnWithTypeAndName & key_col_with_type = columns[arguments[2]]; + const ColumnWithTypeAndName & key_col_with_type = arguments[2]; /// Functions in external dictionaries_loader only support full-value (not constant) columns with keys. ColumnPtr key_col = key_col_with_type.column->convertToFullColumnIfConst(); @@ -1303,7 +1275,7 @@ private: out = ColVec::create(rows); auto & data = out->getData(); - const auto default_col_untyped = columns[arguments[3]].column.get(); + const auto * default_col_untyped = arguments[3].column.get(); if (const auto default_col = checkAndGetColumn(default_col_untyped)) { /// const defaults @@ -1320,8 +1292,7 @@ private: else throw Exception{"Fourth argument of function " + getName() + " must be " + TypeName::get(), ErrorCodes::ILLEGAL_COLUMN}; - columns[result].column = std::move(out); - return true; + return out; } mutable FunctionDictHelper helper; @@ -1374,7 +1345,7 @@ public: return std::make_shared(context); } - FunctionDictGetNoType(const Context & context_) : context(context_), helper(context_) {} + explicit FunctionDictGetNoType(const Context & context_) : context(context_), helper(context_) {} String getName() const override { return name; } @@ -1396,7 +1367,7 @@ private: throw Exception{"Function " + getName() + " takes 3 or 4 arguments", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; String dict_name; - if (auto name_col = checkAndGetColumnConst(arguments[0].column.get())) + if (const auto * name_col = checkAndGetColumnConst(arguments[0].column.get())) { dict_name = name_col->getValue(); } @@ -1405,7 +1376,7 @@ private: + ", expected a const string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; String attr_name; - if (auto name_col = checkAndGetColumnConst(arguments[1].column.get())) + if (const auto * name_col = checkAndGetColumnConst(arguments[1].column.get())) { attr_name = name_col->getValue(); } @@ -1420,7 +1391,7 @@ private: if (arguments.size() == 4) { - const auto range_argument = arguments[3].type.get(); + const auto * range_argument = arguments[3].type.get(); if (!(range_argument->isValueRepresentedByInteger() && range_argument->getSizeOfValueInMemory() <= sizeof(Int64))) throw Exception{"Illegal type " + range_argument->getName() + " of fourth argument of function " + getName() @@ -1501,12 +1472,11 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - impl->executeImpl(columns, arguments, result, input_rows_count); + return impl->executeImpl(arguments, result_type, input_rows_count); } -private: const Context & context; mutable FunctionDictHelper helper; mutable FunctionPtr impl; // underlying function used by dictGet function without explicit type info @@ -1523,7 +1493,7 @@ public: return std::make_shared(context); } - FunctionDictGetNoTypeOrDefault(const Context & context_) : context(context_), helper(context_) {} + explicit FunctionDictGetNoTypeOrDefault(const Context & context_) : context(context_), helper(context_) {} String getName() const override { return name; } @@ -1541,7 +1511,7 @@ private: DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override { String dict_name; - if (auto name_col = checkAndGetColumnConst(arguments[0].column.get())) + if (const auto * name_col = checkAndGetColumnConst(arguments[0].column.get())) { dict_name = name_col->getValue(); } @@ -1550,7 +1520,7 @@ private: + ", expected a const string.", ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; String attr_name; - if (auto name_col = checkAndGetColumnConst(arguments[1].column.get())) + if (const auto * name_col = checkAndGetColumnConst(arguments[1].column.get())) { attr_name = name_col->getValue(); } @@ -1643,12 +1613,11 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - impl->executeImpl(columns, arguments, result, input_rows_count); + return impl->executeImpl(arguments, result_type, input_rows_count); } -private: const Context & context; mutable FunctionDictHelper helper; mutable FunctionPtr impl; // underlying function used by dictGet function without explicit type info @@ -1666,7 +1635,7 @@ public: return std::make_shared(context); } - FunctionDictGetHierarchy(const Context & context_) : helper(context_) {} + explicit FunctionDictGetHierarchy(const Context & context_) : helper(context_) {} String getName() const override { return name; } @@ -1692,31 +1661,29 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); + ColumnPtr res; - if (!executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict) && - !executeDispatch(columns, arguments, result, dict)) + if (!((res = executeDispatch(arguments, result_type, dict)) + || (res = executeDispatch(arguments, result_type, dict)) + || (res = executeDispatch(arguments, result_type, dict)) + || (res = executeDispatch(arguments, result_type, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, - const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (!dict->hasHierarchy()) throw Exception{"Dictionary does not have a hierarchy", ErrorCodes::UNSUPPORTED_METHOD}; @@ -1781,28 +1748,26 @@ private: } }; - const auto id_col_untyped = columns[arguments[1]].column.get(); - if (const auto id_col = checkAndGetColumn(id_col_untyped)) + const auto * id_col_untyped = arguments[1].column.get(); + if (const auto * id_col = checkAndGetColumn(id_col_untyped)) { const auto & in = id_col->getData(); auto backend = ColumnUInt64::create(); auto offsets = ColumnArray::ColumnOffsets::create(); get_hierarchies(in, backend->getData(), offsets->getData()); - columns[result].column = ColumnArray::create(std::move(backend), std::move(offsets)); + return ColumnArray::create(std::move(backend), std::move(offsets)); } - else if (const auto id_col_const = checkAndGetColumnConst>(id_col_untyped)) + else if (const auto * id_col_const = checkAndGetColumnConst>(id_col_untyped)) { const PaddedPODArray in(1, id_col_const->getValue()); auto backend = ColumnUInt64::create(); auto offsets = ColumnArray::ColumnOffsets::create(); get_hierarchies(in, backend->getData(), offsets->getData()); auto array = ColumnArray::create(std::move(backend), std::move(offsets)); - columns[result].column = columns[result].type->createColumnConst(id_col_const->size(), (*array)[0].get()); + return result_type->createColumnConst(id_col_const->size(), (*array)[0].get()); } else throw Exception{"Second argument of function " + getName() + " must be UInt64", ErrorCodes::ILLEGAL_COLUMN}; - - return true; } mutable FunctionDictHelper helper; @@ -1819,7 +1784,7 @@ public: return std::make_shared(context); } - FunctionDictIsIn(const Context & context_) + explicit FunctionDictIsIn(const Context & context_) : helper(context_) {} String getName() const override { return name; } @@ -1849,54 +1814,50 @@ private: bool isDeterministic() const override { return false; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { if (input_rows_count == 0) - { - auto & elem = columns[result]; - elem.column = elem.type->createColumn(); - return; - } + return result_type->createColumn(); - auto dict = helper.getDictionary(columns[arguments[0]]); + auto dict = helper.getDictionary(arguments[0]); - if (!executeDispatch(columns, arguments, result, dict) - && !executeDispatch(columns, arguments, result, dict) - && !executeDispatch(columns, arguments, result, dict) - && !executeDispatch(columns, arguments, result, dict)) + ColumnPtr res; + if (!((res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)) + || (res = executeDispatch(arguments, dict)))) throw Exception{"Unsupported dictionary type " + dict->getTypeName(), ErrorCodes::UNKNOWN_TYPE}; + + return res; } template - bool executeDispatch(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result, - const std::shared_ptr & dict_ptr) const + ColumnPtr executeDispatch(ColumnsWithTypeAndName & arguments, const std::shared_ptr & dict_ptr) const { - const auto dict = typeid_cast(dict_ptr.get()); + const auto * dict = typeid_cast(dict_ptr.get()); if (!dict) - return false; + return nullptr; if (!dict->hasHierarchy()) throw Exception{"Dictionary does not have a hierarchy", ErrorCodes::UNSUPPORTED_METHOD}; - const auto child_id_col_untyped = columns[arguments[1]].column.get(); - const auto ancestor_id_col_untyped = columns[arguments[2]].column.get(); + const auto * child_id_col_untyped = arguments[1].column.get(); + const auto * ancestor_id_col_untyped = arguments[2].column.get(); - if (const auto child_id_col = checkAndGetColumn(child_id_col_untyped)) - execute(columns, result, dict, child_id_col, ancestor_id_col_untyped); - else if (const auto child_id_col_const = checkAndGetColumnConst>(child_id_col_untyped)) - execute(columns, result, dict, child_id_col_const, ancestor_id_col_untyped); + if (const auto * child_id_col = checkAndGetColumn(child_id_col_untyped)) + return execute(dict, child_id_col, ancestor_id_col_untyped); + else if (const auto * child_id_col_const = checkAndGetColumnConst>(child_id_col_untyped)) + return execute(dict, child_id_col_const, ancestor_id_col_untyped); else throw Exception{"Illegal column " + child_id_col_untyped->getName() + " of second argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; - - return true; } template - bool execute(ColumnsWithTypeAndName & columns, const size_t result, const DictionaryType * dict, + ColumnPtr execute(const DictionaryType * dict, const ColumnUInt64 * child_id_col, const IColumn * ancestor_id_col_untyped) const { - if (const auto ancestor_id_col = checkAndGetColumn(ancestor_id_col_untyped)) + if (const auto * ancestor_id_col = checkAndGetColumn(ancestor_id_col_untyped)) { auto out = ColumnUInt8::create(); @@ -1907,9 +1868,9 @@ private: data.resize(size); dict->isInVectorVector(child_ids, ancestor_ids, data); - columns[result].column = std::move(out); + return out; } - else if (const auto ancestor_id_col_const = checkAndGetColumnConst>(ancestor_id_col_untyped)) + else if (const auto * ancestor_id_col_const = checkAndGetColumnConst>(ancestor_id_col_untyped)) { auto out = ColumnUInt8::create(); @@ -1920,22 +1881,19 @@ private: data.resize(size); dict->isInVectorConstant(child_ids, ancestor_id, data); - columns[result].column = std::move(out); + return out; } else { throw Exception{"Illegal column " + ancestor_id_col_untyped->getName() + " of third argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; } - - return true; } template - bool execute(ColumnsWithTypeAndName & columns, const size_t result, const DictionaryType * dict, - const ColumnConst * child_id_col, const IColumn * ancestor_id_col_untyped) const + ColumnPtr execute(const DictionaryType * dict, const ColumnConst * child_id_col, const IColumn * ancestor_id_col_untyped) const { - if (const auto ancestor_id_col = checkAndGetColumn(ancestor_id_col_untyped)) + if (const auto * ancestor_id_col = checkAndGetColumn(ancestor_id_col_untyped)) { auto out = ColumnUInt8::create(); @@ -1946,22 +1904,20 @@ private: data.resize(size); dict->isInConstantVector(child_id, ancestor_ids, data); - columns[result].column = std::move(out); + return out; } - else if (const auto ancestor_id_col_const = checkAndGetColumnConst>(ancestor_id_col_untyped)) + else if (const auto * ancestor_id_col_const = checkAndGetColumnConst>(ancestor_id_col_untyped)) { const auto child_id = child_id_col->getValue(); const auto ancestor_id = ancestor_id_col_const->getValue(); UInt8 res = 0; dict->isInConstantConstant(child_id, ancestor_id, res); - columns[result].column = DataTypeUInt8().createColumnConst(child_id_col->size(), res); + return DataTypeUInt8().createColumnConst(child_id_col->size(), res); } else throw Exception{"Illegal column " + ancestor_id_col_untyped->getName() + " of third argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; - - return true; } mutable FunctionDictHelper helper; diff --git a/src/Functions/FunctionsExternalModels.cpp b/src/Functions/FunctionsExternalModels.cpp index 14ced517e6a..9c1892012e1 100644 --- a/src/Functions/FunctionsExternalModels.cpp +++ b/src/Functions/FunctionsExternalModels.cpp @@ -69,9 +69,9 @@ DataTypePtr FunctionModelEvaluate::getReturnTypeImpl(const ColumnsWithTypeAndNam return type; } -void FunctionModelEvaluate::executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const +ColumnPtr FunctionModelEvaluate::executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const { - const auto * name_col = checkAndGetColumnConst(columns[arguments[0]].column.get()); + const auto * name_col = checkAndGetColumnConst(arguments[0].column.get()); if (!name_col) throw Exception("First argument of function " + getName() + " must be a constant string", ErrorCodes::ILLEGAL_COLUMN); @@ -85,7 +85,7 @@ void FunctionModelEvaluate::executeImpl(ColumnsWithTypeAndName & columns, const column_ptrs.reserve(arguments.size()); for (auto arg : ext::range(1, arguments.size())) { - auto & column = columns[arguments[arg]].column; + auto & column = arguments[arg].column; column_ptrs.push_back(column.get()); if (auto full_column = column->convertToFullColumnIfConst()) { @@ -130,7 +130,7 @@ void FunctionModelEvaluate::executeImpl(ColumnsWithTypeAndName & columns, const res = ColumnNullable::create(res, null_map); } - columns[result].column = res; + return res; } void registerFunctionsExternalModels(FunctionFactory & factory) diff --git a/src/Functions/FunctionsExternalModels.h b/src/Functions/FunctionsExternalModels.h index 40dad72bfc4..336dc164248 100644 --- a/src/Functions/FunctionsExternalModels.h +++ b/src/Functions/FunctionsExternalModels.h @@ -32,7 +32,7 @@ public: DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override; - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override; + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override; private: const ExternalModelsLoader & models_loader; diff --git a/src/Functions/FunctionsHashing.h b/src/Functions/FunctionsHashing.h index 3bed85e30bc..678214cdbad 100644 --- a/src/Functions/FunctionsHashing.h +++ b/src/Functions/FunctionsHashing.h @@ -543,9 +543,9 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - if (const ColumnString * col_from = checkAndGetColumn(columns[arguments[0]].column.get())) + if (const ColumnString * col_from = checkAndGetColumn(arguments[0].column.get())) { auto col_to = ColumnFixedString::create(Impl::length); @@ -566,10 +566,10 @@ public: current_offset = offsets[i]; } - columns[result].column = std::move(col_to); + return col_to; } else if ( - const ColumnFixedString * col_from_fix = checkAndGetColumn(columns[arguments[0]].column.get())) + const ColumnFixedString * col_from_fix = checkAndGetColumn(arguments[0].column.get())) { auto col_to = ColumnFixedString::create(Impl::length); const typename ColumnFixedString::Chars & data = col_from_fix->getChars(); @@ -582,10 +582,10 @@ public: Impl::apply( reinterpret_cast(&data[i * length]), length, reinterpret_cast(&chars_to[i * Impl::length])); } - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN); } @@ -604,11 +604,11 @@ private: using ToType = typename Impl::ReturnType; template - void executeType(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result) const + ColumnPtr executeType(ColumnsWithTypeAndName & arguments) const { using ColVecType = std::conditional_t, ColumnDecimal, ColumnVector>; - if (const ColVecType * col_from = checkAndGetColumn(columns[arguments[0]].column.get())) + if (const ColVecType * col_from = checkAndGetColumn(arguments[0].column.get())) { auto col_to = ColumnVector::create(); @@ -620,10 +620,10 @@ private: for (size_t i = 0; i < size; ++i) vec_to[i] = Impl::apply(vec_from[i]); - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception("Illegal column " + columns[arguments[0]].column->getName() + throw Exception("Illegal column " + arguments[0].column->getName() + " of first argument of function " + Name::name, ErrorCodes::ILLEGAL_COLUMN); } @@ -647,25 +647,37 @@ public: bool useDefaultImplementationForConstants() const override { return true; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { - const IDataType * from_type = columns[arguments[0]].type.get(); + const IDataType * from_type = arguments[0].type.get(); WhichDataType which(from_type); - if (which.isUInt8()) executeType(columns, arguments, result); - else if (which.isUInt16()) executeType(columns, arguments, result); - else if (which.isUInt32()) executeType(columns, arguments, result); - else if (which.isUInt64()) executeType(columns, arguments, result); - else if (which.isInt8()) executeType(columns, arguments, result); - else if (which.isInt16()) executeType(columns, arguments, result); - else if (which.isInt32()) executeType(columns, arguments, result); - else if (which.isInt64()) executeType(columns, arguments, result); - else if (which.isDate()) executeType(columns, arguments, result); - else if (which.isDateTime()) executeType(columns, arguments, result); - else if (which.isDecimal32()) executeType(columns, arguments, result); - else if (which.isDecimal64()) executeType(columns, arguments, result); + if (which.isUInt8()) + return executeType(arguments); + else if (which.isUInt16()) + return executeType(arguments); + else if (which.isUInt32()) + return executeType(arguments); + else if (which.isUInt64()) + return executeType(arguments); + else if (which.isInt8()) + return executeType(arguments); + else if (which.isInt16()) + return executeType(arguments); + else if (which.isInt32()) + return executeType(arguments); + else if (which.isInt64()) + return executeType(arguments); + else if (which.isDate()) + return executeType(arguments); + else if (which.isDateTime()) + return executeType(arguments); + else if (which.isDecimal32()) + return executeType(arguments); + else if (which.isDecimal64()) + return executeType(arguments); else - throw Exception("Illegal type " + columns[arguments[0]].type->getName() + " of argument of function " + getName(), + throw Exception("Illegal type " + arguments[0].type->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); } }; @@ -689,9 +701,9 @@ public: #endif } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - selector.selectAndExecute(columns, arguments, result, input_rows_count); + return selector.selectAndExecute(arguments, result_type, input_rows_count); } static FunctionPtr create(const Context & context) @@ -886,7 +898,7 @@ private: } else if (const ColumnConst * col_from_const = checkAndGetColumnConstStringOrFixedString(column)) { - String value = col_from_const->getValue().data(); + String value = col_from_const->getValue(); const ToType hash = Impl::apply(value.data(), value.size()); const size_t size = vec_to.size(); @@ -1041,7 +1053,7 @@ public: return std::make_shared>(); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override { size_t rows = input_rows_count; auto col_to = ColumnVector::create(rows); @@ -1057,13 +1069,10 @@ public: /// The function supports arbitrary number of arguments of arbitrary types. bool is_first_argument = true; - for (size_t i = 0; i < arguments.size(); ++i) - { - const ColumnWithTypeAndName & col = columns[arguments[i]]; + for (const auto & col : arguments) executeForArgument(col.type.get(), col.column.get(), vec_to, is_first_argument); - } - columns[result].column = std::move(col_to); + return col_to; } }; @@ -1086,9 +1095,9 @@ public: #endif } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { - selector.selectAndExecute(columns, arguments, result, input_rows_count); + return selector.selectAndExecute(arguments, result_type, input_rows_count); } static FunctionPtr create(const Context & context) @@ -1118,7 +1127,7 @@ struct URLHierarchyHashImpl { static size_t findLevelLength(const UInt64 level, const char * begin, const char * end) { - auto pos = begin; + const auto * pos = begin; /// Let's parse everything that goes before the path @@ -1192,7 +1201,7 @@ public: throw Exception{"Number of arguments for function " + getName() + " doesn't match: passed " + toString(arg_count) + ", should be 1 or 2.", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto first_arg = arguments.front().get(); + const auto * first_arg = arguments.front().get(); if (!WhichDataType(first_arg).isString()) throw Exception{"Illegal type " + first_arg->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; @@ -1209,24 +1218,24 @@ public: bool useDefaultImplementationForConstants() const override { return true; } ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result, size_t /*input_rows_count*/) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t /*input_rows_count*/) const override { const auto arg_count = arguments.size(); if (arg_count == 1) - executeSingleArg(columns, arguments, result); + return executeSingleArg(arguments); else if (arg_count == 2) - executeTwoArgs(columns, arguments, result); + return executeTwoArgs(arguments); else throw Exception{"got into IFunction::execute with unexpected number of arguments", ErrorCodes::LOGICAL_ERROR}; } private: - void executeSingleArg(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result) const + ColumnPtr executeSingleArg(ColumnsWithTypeAndName & arguments) const { - const auto col_untyped = columns[arguments.front()].column.get(); + const auto * col_untyped = arguments.front().column.get(); - if (const auto col_from = checkAndGetColumn(col_untyped)) + if (const auto * col_from = checkAndGetColumn(col_untyped)) { const auto size = col_from->size(); auto col_to = ColumnUInt64::create(size); @@ -1245,23 +1254,23 @@ private: current_offset = offsets[i]; } - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception{"Illegal column " + columns[arguments[0]].column->getName() + + throw Exception{"Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; } - void executeTwoArgs(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, const size_t result) const + ColumnPtr executeTwoArgs(ColumnsWithTypeAndName & arguments) const { - const auto level_col = columns[arguments.back()].column.get(); + const auto * level_col = arguments.back().column.get(); if (!isColumnConst(*level_col)) throw Exception{"Second argument of function " + getName() + " must be an integral constant", ErrorCodes::ILLEGAL_COLUMN}; const auto level = level_col->get64(0); - const auto col_untyped = columns[arguments.front()].column.get(); - if (const auto col_from = checkAndGetColumn(col_untyped)) + const auto * col_untyped = arguments.front().column.get(); + if (const auto * col_from = checkAndGetColumn(col_untyped)) { const auto size = col_from->size(); auto col_to = ColumnUInt64::create(size); @@ -1281,10 +1290,10 @@ private: current_offset = offsets[i]; } - columns[result].column = std::move(col_to); + return col_to; } else - throw Exception{"Illegal column " + columns[arguments[0]].column->getName() + + throw Exception{"Illegal column " + arguments[0].column->getName() + " of argument of function " + getName(), ErrorCodes::ILLEGAL_COLUMN}; } }; diff --git a/src/Functions/FunctionsJSON.cpp b/src/Functions/FunctionsJSON.cpp index 6f83bc401de..7478c1627af 100644 --- a/src/Functions/FunctionsJSON.cpp +++ b/src/Functions/FunctionsJSON.cpp @@ -10,13 +10,13 @@ namespace ErrorCodes } -std::vector FunctionJSONHelpers::prepareMoves(const char * function_name, ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t first_index_argument, size_t num_index_arguments) +std::vector FunctionJSONHelpers::prepareMoves(const char * function_name, ColumnsWithTypeAndName & columns, size_t first_index_argument, size_t num_index_arguments) { std::vector moves; moves.reserve(num_index_arguments); for (const auto i : ext::range(first_index_argument, first_index_argument + num_index_arguments)) { - const auto & column = columns[arguments[i]]; + const auto & column = columns[i]; if (!isString(column.type) && !isInteger(column.type)) throw Exception{"The argument " + std::to_string(i + 1) + " of function " + String(function_name) + " should be a string specifying key or an integer specifying index, illegal type: " + column.type->getName(), diff --git a/src/Functions/FunctionsJSON.h b/src/Functions/FunctionsJSON.h index e55288a2367..0fcf1f57f82 100644 --- a/src/Functions/FunctionsJSON.h +++ b/src/Functions/FunctionsJSON.h @@ -55,22 +55,22 @@ public: class Executor { public: - static void run(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result_pos, size_t input_rows_count) + static ColumnPtr run(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) { - MutableColumnPtr to{columns[result_pos].type->createColumn()}; + MutableColumnPtr to{result_type->createColumn()}; to->reserve(input_rows_count); - if (arguments.size() < 1) + if (arguments.empty()) throw Exception{"Function " + String(Name::name) + " requires at least one argument", ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH}; - const auto & first_column = columns[arguments[0]]; + const auto & first_column = arguments[0]; if (!isString(first_column.type)) throw Exception{"The first argument of function " + String(Name::name) + " should be a string containing JSON, illegal type: " + first_column.type->getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; const ColumnPtr & arg_json = first_column.column; - auto col_json_const = typeid_cast(arg_json.get()); - auto col_json_string + const auto * col_json_const = typeid_cast(arg_json.get()); + const auto * col_json_string = typeid_cast(col_json_const ? col_json_const->getDataColumnPtr().get() : arg_json.get()); if (!col_json_string) @@ -79,8 +79,8 @@ public: const ColumnString::Chars & chars = col_json_string->getChars(); const ColumnString::Offsets & offsets = col_json_string->getOffsets(); - size_t num_index_arguments = Impl::getNumberOfIndexArguments(columns, arguments); - std::vector moves = prepareMoves(Name::name, columns, arguments, 1, num_index_arguments); + size_t num_index_arguments = Impl::getNumberOfIndexArguments(arguments); + std::vector moves = prepareMoves(Name::name, arguments, 1, num_index_arguments); /// Preallocate memory in parser if necessary. JSONParser parser; @@ -94,8 +94,8 @@ public: Impl impl; /// prepare() does Impl-specific preparation before handling each row. - if constexpr (has_member_function_prepare::*)(const char *, const ColumnsWithTypeAndName &, const ColumnNumbers &, size_t)>::value) - impl.prepare(Name::name, columns, arguments, result_pos); + if constexpr (has_member_function_prepare::*)(const char *, const ColumnsWithTypeAndName &, const DataTypePtr &)>::value) + impl.prepare(Name::name, arguments, result_type); using Element = typename JSONParser::Element; @@ -121,7 +121,7 @@ public: /// Perform moves. Element element; std::string_view last_key; - bool moves_ok = performMoves(columns, arguments, i, document, moves, element, last_key); + bool moves_ok = performMoves(arguments, i, document, moves, element, last_key); if (moves_ok) added_to_column = impl.insertResultToColumn(*to, element, last_key); @@ -131,7 +131,7 @@ public: if (!added_to_column) to->insertDefault(); } - columns[result_pos].column = std::move(to); + return to; } }; @@ -166,11 +166,11 @@ private: String key; }; - static std::vector prepareMoves(const char * function_name, ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t first_index_argument, size_t num_index_arguments); + static std::vector prepareMoves(const char * function_name, ColumnsWithTypeAndName & columns, size_t first_index_argument, size_t num_index_arguments); /// Performs moves of types MoveType::Index and MoveType::ConstIndex. template - static bool performMoves(const ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t row, + static bool performMoves(const ColumnsWithTypeAndName & arguments, size_t row, const typename JSONParser::Element & document, const std::vector & moves, typename JSONParser::Element & element, std::string_view & last_key) { @@ -196,14 +196,14 @@ private: } case MoveType::Index: { - Int64 index = (*columns[arguments[j + 1]].column)[row].get(); + Int64 index = (*arguments[j + 1].column)[row].get(); if (!moveToElementByIndex(res_element, index, key)) return false; break; } case MoveType::Key: { - key = std::string_view{(*columns[arguments[j + 1]].column).getDataAt(row)}; + key = std::string_view{(*arguments[j + 1].column).getDataAt(row)}; if (!moveToElementByKey(res_element, key)) return false; break; @@ -286,21 +286,18 @@ public: return Impl::getReturnType(Name::name, arguments); } - void executeImpl(ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result_pos, size_t input_rows_count) const override + ColumnPtr executeImpl(ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override { /// Choose JSONParser. #if USE_SIMDJSON if (context.getSettingsRef().allow_simdjson) - { - FunctionJSONHelpers::Executor::run(columns, arguments, result_pos, input_rows_count); - return; - } + return FunctionJSONHelpers::Executor::run(arguments, result_type, input_rows_count); #endif #if USE_RAPIDJSON - FunctionJSONHelpers::Executor::run(columns, arguments, result_pos, input_rows_count); + return FunctionJSONHelpers::Executor::run(arguments, result_type, input_rows_count); #else - FunctionJSONHelpers::Executor::run(columns, arguments, result_pos, input_rows_count); + return FunctionJSONHelpers::Executor::run(arguments, result_type, input_rows_count); #endif } @@ -334,7 +331,7 @@ public: static DataTypePtr getReturnType(const char *, const ColumnsWithTypeAndName &) { return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element &, const std::string_view &) { @@ -362,7 +359,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers &) { return 0; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &) { return 0; } static bool insertResultToColumn(IColumn & dest, const Element &, const std::string_view &) { @@ -386,7 +383,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -416,7 +413,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element &, const std::string_view & last_key) { @@ -450,7 +447,7 @@ public: return std::make_shared>(values); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -492,7 +489,7 @@ public: return std::make_shared>(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -557,7 +554,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -582,7 +579,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -909,11 +906,11 @@ public: return DataTypeFactory::instance().get(col_type_const->getValue()); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 2; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 2; } - void prepare(const char * function_name, const ColumnsWithTypeAndName & columns, const ColumnNumbers &, size_t result_pos) + void prepare(const char * function_name, const ColumnsWithTypeAndName &, const DataTypePtr & result_type) { - extract_tree = JSONExtractTree::build(function_name, columns[result_pos].type); + extract_tree = JSONExtractTree::build(function_name, result_type); } bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) @@ -950,11 +947,10 @@ public: return std::make_unique(tuple_type); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 2; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 2; } - void prepare(const char * function_name, const ColumnsWithTypeAndName & columns, const ColumnNumbers &, size_t result_pos) + void prepare(const char * function_name, const ColumnsWithTypeAndName &, const DataTypePtr & result_type) { - const auto & result_type = columns[result_pos].type; const auto tuple_type = typeid_cast(result_type.get())->getNestedType(); const auto value_type = typeid_cast(tuple_type.get())->getElements()[1]; extract_tree = JSONExtractTree::build(function_name, value_type); @@ -1002,7 +998,7 @@ public: return std::make_shared(); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -1106,7 +1102,7 @@ public: return std::make_shared(std::make_shared()); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } static bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { @@ -1138,7 +1134,7 @@ public: return std::make_unique(tuple_type); } - static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName &, const ColumnNumbers & arguments) { return arguments.size() - 1; } + static size_t getNumberOfIndexArguments(const ColumnsWithTypeAndName & arguments) { return arguments.size() - 1; } bool insertResultToColumn(IColumn & dest, const Element & element, const std::string_view &) { diff --git a/src/Functions/FunctionsLogical.cpp b/src/Functions/FunctionsLogical.cpp index e921471497d..3e19516daaa 100644 --- a/src/Functions/FunctionsLogical.cpp +++ b/src/Functions/FunctionsLogical.cpp @@ -290,38 +290,38 @@ private: /// Apply target function by feeding it "batches" of N columns -/// Combining 10 columns per pass is the fastest for large columns sizes. -/// For small columns sizes - more columns is faster. +/// Combining 8 columns per pass is the fastest method, because it's the maximum when clang vectorizes a loop. template < - typename Op, template typename OperationApplierImpl, size_t N = 10> + typename Op, template typename OperationApplierImpl, size_t N = 8> struct OperationApplier { template static void apply(Columns & in, ResultData & result_data, bool use_result_data_as_input = false) { if (!use_result_data_as_input) - doBatchedApply(in, result_data); + doBatchedApply(in, result_data.data(), result_data.size()); while (!in.empty()) - doBatchedApply(in, result_data); + doBatchedApply(in, result_data.data(), result_data.size()); } - template - static void NO_INLINE doBatchedApply(Columns & in, ResultData & result_data) + template + static void NO_INLINE doBatchedApply(Columns & in, Result * __restrict result_data, size_t size) { if (N > in.size()) { OperationApplier - ::template doBatchedApply(in, result_data); + ::template doBatchedApply(in, result_data, size); return; } const OperationApplierImpl operation_applier_impl(in); - size_t i = 0; - for (auto & res : result_data) + for (size_t i = 0; i < size; ++i) + { if constexpr (CarryResult) - res = Op::apply(res, operation_applier_impl.apply(i++)); + result_data[i] = Op::apply(result_data[i], operation_applier_impl.apply(i)); else - res = operation_applier_impl.apply(i++); + result_data[i] = operation_applier_impl.apply(i); + } in.erase(in.end() - N, in.end()); } @@ -332,7 +332,7 @@ template < struct OperationApplier { template - static void NO_INLINE doBatchedApply(Columns &, Result &) + static void NO_INLINE doBatchedApply(Columns &, Result &, size_t) { throw Exception( "OperationApplier<...>::apply(...): not enough arguments to run this method", @@ -342,7 +342,7 @@ struct OperationApplier template -static void executeForTernaryLogicImpl(ColumnRawPtrs arguments, ColumnWithTypeAndName & result_info, size_t input_rows_count) +static ColumnPtr executeForTernaryLogicImpl(ColumnRawPtrs arguments, const DataTypePtr & result_type, size_t input_rows_count) { /// Combine all constant columns into a single constant value. UInt8 const_3v_value = 0; @@ -351,11 +351,10 @@ static void executeForTernaryLogicImpl(ColumnRawPtrs arguments, ColumnWithTypeAn /// If the constant value uniquely determines the result, return it. if (has_consts && (arguments.empty() || Op::isSaturatedValueTernary(const_3v_value))) { - result_info.column = ColumnConst::create( - buildColumnFromTernaryData(UInt8Container({const_3v_value}), result_info.type->isNullable()), + return ColumnConst::create( + buildColumnFromTernaryData(UInt8Container({const_3v_value}), result_type->isNullable()), input_rows_count ); - return; } const auto result_column = has_consts ? @@ -363,7 +362,7 @@ static void executeForTernaryLogicImpl(ColumnRawPtrs arguments, ColumnWithTypeAn OperationApplier::apply(arguments, result_column->getData(), has_consts); - result_info.column = buildColumnFromTernaryData(result_column->getData(), result_info.type->isNullable()); + return buildColumnFromTernaryData(result_column->getData(), result_type->isNullable()); } @@ -418,7 +417,7 @@ struct TypedExecutorInvoker /// Types of all of the arguments are guaranteed to be non-nullable here template -static void basicExecuteImpl(ColumnRawPtrs arguments, ColumnWithTypeAndName & result_info, size_t input_rows_count) +static ColumnPtr basicExecuteImpl(ColumnRawPtrs arguments, size_t input_rows_count) { /// Combine all constant columns into a single constant value. UInt8 const_val = 0; @@ -429,8 +428,7 @@ static void basicExecuteImpl(ColumnRawPtrs arguments, ColumnWithTypeAndName & re { if (!arguments.empty()) const_val = Op::apply(const_val, 0); - result_info.column = DataTypeUInt8().createColumnConst(input_rows_count, toField(const_val)); - return; + return DataTypeUInt8().createColumnConst(input_rows_count, toField(const_val)); } /// If the constant value is a neutral element, let's forget about it. @@ -448,8 +446,7 @@ static void basicExecuteImpl(ColumnRawPtrs arguments, ColumnWithTypeAndName & re else FastApplierImpl::apply(*arguments[0], *arguments[1], col_res->getData()); - result_info.column = std::move(col_res); - return; + return col_res; } /// Convert all columns to UInt8 @@ -470,7 +467,7 @@ static void basicExecuteImpl(ColumnRawPtrs arguments, ColumnWithTypeAndName & re OperationApplier::apply(uint8_args, col_res->getData(), has_consts); - result_info.column = std::move(col_res); + return col_res; } } @@ -511,18 +508,17 @@ DataTypePtr FunctionAnyArityLogical::getReturnTypeImpl(const DataTyp } template -void FunctionAnyArityLogical::executeImpl( - ColumnsWithTypeAndName & columns, const ColumnNumbers & arguments, size_t result_index, size_t input_rows_count) const +ColumnPtr FunctionAnyArityLogical::executeImpl( + ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const { ColumnRawPtrs args_in; - for (const auto arg_index : arguments) - args_in.push_back(columns[arg_index].column.get()); + for (const auto & arg_index : arguments) + args_in.push_back(arg_index.column.get()); - auto & result_info = columns[result_index]; - if (result_info.type->isNullable()) - executeForTernaryLogicImpl(std::move(args_in), result_info, input_rows_count); + if (result_type->isNullable()) + return executeForTernaryLogicImpl(std::move(args_in), result_type, input_rows_count); else - basicExecuteImpl(std::move(args_in), result_info, input_rows_count); + return basicExecuteImpl(std::move(args_in), input_rows_count); } @@ -554,9 +550,9 @@ DataTypePtr FunctionUnaryLogical::getReturnTypeImpl(const DataTypes } template