thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-11-27 18:12:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into fix-regexp-recompilation-in-parser

Browse Source
This commit is contained in:
Alexey Milovidov 2024-03-21 02:21:09 +01:00
commit 99ed82d561
264 changed files with 10527 additions and 2629 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
CMakeLists.txt
View File

@ -61,11 +61,16 @@ if (ENABLE_CHECK_HEAVY_BUILDS)
# set CPU time limit to 1000 seconds
set (RLIMIT_CPU 1000)
# -fsanitize=memory and address are too heavy
# Sanitizers are too heavy
if (SANITIZE OR SANITIZE_COVERAGE OR WITH_COVERAGE)
set (RLIMIT_DATA 10000000000) # 10G
endif()
# For some files currently building RISCV64 might be too slow. TODO: Improve compilation times per file
if (ARCH_RISCV64)
set (RLIMIT_CPU 1800)
endif()
set (CMAKE_CXX_COMPILER_LAUNCHER prlimit --as=${RLIMIT_AS} --data=${RLIMIT_DATA} --cpu=${RLIMIT_CPU} ${CMAKE_CXX_COMPILER_LAUNCHER})
endif ()

1
base/base/CMakeLists.txt
View File

@ -20,6 +20,7 @@ set (SRCS
getPageSize.cpp
getThreadId.cpp
int8_to_string.cpp
itoa.cpp
JSON.cpp
mremap.cpp
phdr_cache.cpp

6
base/base/IPv4andIPv6.h
View File

@ -1,8 +1,7 @@
#pragma once
#include <base/strong_typedef.h>
#include <base/extended_types.h>
#include <Common/formatIPv6.h>
#include <base/strong_typedef.h>
#include <Common/memcmpSmall.h>
namespace DB
@ -62,7 +61,8 @@ namespace std
{
size_t operator()(const DB::IPv6 & x) const
{
return std::hash<std::string_view>{}(std::string_view(reinterpret_cast<const char*>(&x.toUnderType()), IPV6_BINARY_LENGTH));
return std::hash<std::string_view>{}(
std::string_view(reinterpret_cast<const char *>(&x.toUnderType()), sizeof(DB::IPv6::UnderlyingType)));
}
};

503
base/base/itoa.cpp Normal file
View File

@ -0,0 +1,503 @@
// Based on https://github.com/amdn/itoa and combined with our optimizations
//
//=== itoa.cpp - Fast integer to ascii conversion --*- C++ -*-//
//
// The MIT License (MIT)
// Copyright (c) 2016 Arturo Martin-de-Nicolas
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
// SOFTWARE.
//===----------------------------------------------------------------------===//
#include <cstddef>
#include <cstdint>
#include <cstring>
#include <type_traits>
#include <base/defines.h>
#include <base/extended_types.h>
#include <base/itoa.h>
namespace
{
template <typename T>
ALWAYS_INLINE inline constexpr T pow10(size_t x)
{
return x ? 10 * pow10<T>(x - 1) : 1;
}
// Division by a power of 10 is implemented using a multiplicative inverse.
// This strength reduction is also done by optimizing compilers, but
// presently the fastest results are produced by using the values
// for the multiplication and the shift as given by the algorithm
// described by Agner Fog in "Optimizing Subroutines in Assembly Language"
//
// http://www.agner.org/optimize/optimizing_assembly.pdf
//
// "Integer division by a constant (all processors)
// A floating point number can be divided by a constant by multiplying
// with the reciprocal. If we want to do the same with integers, we have
// to scale the reciprocal by 2n and then shift the product to the right
// by n. There are various algorithms for finding a suitable value of n
// and compensating for rounding errors. The algorithm described below
// was invented by Terje Mathisen, Norway, and not published elsewhere."
/// Division by constant is performed by:
/// 1. Adding 1 if needed;
/// 2. Multiplying by another constant;
/// 3. Shifting right by another constant.
template <typename UInt, bool add_, UInt multiplier_, unsigned shift_>
struct Division
{
static constexpr bool add{add_};
static constexpr UInt multiplier{multiplier_};
static constexpr unsigned shift{shift_};
};
/// Select a type with appropriate number of bytes from the list of types.
/// First parameter is the number of bytes requested. Then goes a list of types with 1, 2, 4, ... number of bytes.
/// Example: SelectType<4, uint8_t, uint16_t, uint32_t, uint64_t> will select uint32_t.
template <size_t N, typename T, typename... Ts>
struct SelectType
{
using Result = typename SelectType<N / 2, Ts...>::Result;
};
template <typename T, typename... Ts>
struct SelectType<1, T, Ts...>
{
using Result = T;
};
/// Division by 10^N where N is the size of the type.
template <size_t N>
using DivisionBy10PowN = typename SelectType<
N,
Division<uint8_t, false, 205U, 11>, /// divide by 10
Division<uint16_t, true, 41943U, 22>, /// divide by 100
Division<uint32_t, false, 3518437209U, 45>, /// divide by 10000
Division<uint64_t, false, 12379400392853802749ULL, 90> /// divide by 100000000
>::Result;
template <size_t N>
using UnsignedOfSize = typename SelectType<N, uint8_t, uint16_t, uint32_t, uint64_t, __uint128_t>::Result;
/// Holds the result of dividing an unsigned N-byte variable by 10^N resulting in
template <size_t N>
struct QuotientAndRemainder
{
UnsignedOfSize<N> quotient; // quotient with fewer than 2*N decimal digits
UnsignedOfSize<N / 2> remainder; // remainder with at most N decimal digits
};
template <size_t N>
QuotientAndRemainder<N> inline split(UnsignedOfSize<N> value)
{
constexpr DivisionBy10PowN<N> division;
UnsignedOfSize<N> quotient = (division.multiplier * (UnsignedOfSize<2 * N>(value) + division.add)) >> division.shift;
UnsignedOfSize<N / 2> remainder = static_cast<UnsignedOfSize<N / 2>>(value - quotient * pow10<UnsignedOfSize<N / 2>>(N));
return {quotient, remainder};
}
ALWAYS_INLINE inline char * outDigit(char * p, uint8_t value)
{
*p = '0' + value;
++p;
return p;
}
// Using a lookup table to convert binary numbers from 0 to 99
// into ascii characters as described by Andrei Alexandrescu in
// https://www.facebook.com/notes/facebook-engineering/three-optimization-tips-for-c/10151361643253920/
const char digits[201] = "00010203040506070809"
"10111213141516171819"
"20212223242526272829"
"30313233343536373839"
"40414243444546474849"
"50515253545556575859"
"60616263646566676869"
"70717273747576777879"
"80818283848586878889"
"90919293949596979899";
ALWAYS_INLINE inline char * outTwoDigits(char * p, uint8_t value)
{
memcpy(p, &digits[value * 2], 2);
p += 2;
return p;
}
namespace convert
{
template <typename UInt, size_t N = sizeof(UInt)>
char * head(char * p, UInt u);
template <typename UInt, size_t N = sizeof(UInt)>
char * tail(char * p, UInt u);
//===----------------------------------------------------------===//
// head: find most significant digit, skip leading zeros
//===----------------------------------------------------------===//
// "x" contains quotient and remainder after division by 10^N
// quotient is less than 10^N
template <size_t N>
ALWAYS_INLINE inline char * head(char * p, QuotientAndRemainder<N> x)
{
p = head(p, UnsignedOfSize<N / 2>(x.quotient));
p = tail(p, x.remainder);
return p;
}
// "u" is less than 10^2*N
template <typename UInt, size_t N>
ALWAYS_INLINE inline char * head(char * p, UInt u)
{
return u < pow10<UnsignedOfSize<N>>(N) ? head(p, UnsignedOfSize<N / 2>(u)) : head<N>(p, split<N>(u));
}
// recursion base case, selected when "u" is one byte
template <>
ALWAYS_INLINE inline char * head<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
return u < 10 ? outDigit(p, u) : outTwoDigits(p, u);
}
//===----------------------------------------------------------===//
// tail: produce all digits including leading zeros
//===----------------------------------------------------------===//
// recursive step, "u" is less than 10^2*N
template <typename UInt, size_t N>
ALWAYS_INLINE inline char * tail(char * p, UInt u)
{
QuotientAndRemainder<N> x = split<N>(u);
p = tail(p, UnsignedOfSize<N / 2>(x.quotient));
p = tail(p, x.remainder);
return p;
}
// recursion base case, selected when "u" is one byte
template <>
ALWAYS_INLINE inline char * tail<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
return outTwoDigits(p, u);
}
//===----------------------------------------------------------===//
// large values are >= 10^2*N
// where x contains quotient and remainder after division by 10^N
//===----------------------------------------------------------===//
template <size_t N>
ALWAYS_INLINE inline char * large(char * p, QuotientAndRemainder<N> x)
{
QuotientAndRemainder<N> y = split<N>(x.quotient);
p = head(p, UnsignedOfSize<N / 2>(y.quotient));
p = tail(p, y.remainder);
p = tail(p, x.remainder);
return p;
}
//===----------------------------------------------------------===//
// handle values of "u" that might be >= 10^2*N
// where N is the size of "u" in bytes
//===----------------------------------------------------------===//
template <typename UInt, size_t N = sizeof(UInt)>
ALWAYS_INLINE inline char * uitoa(char * p, UInt u)
{
if (u < pow10<UnsignedOfSize<N>>(N))
return head(p, UnsignedOfSize<N / 2>(u));
QuotientAndRemainder<N> x = split<N>(u);
return u < pow10<UnsignedOfSize<N>>(2 * N) ? head<N>(p, x) : large<N>(p, x);
}
// selected when "u" is one byte
template <>
ALWAYS_INLINE inline char * uitoa<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
if (u < 10)
return outDigit(p, u);
else if (u < 100)
return outTwoDigits(p, u);
else
{
p = outDigit(p, u / 100);
p = outTwoDigits(p, u % 100);
return p;
}
}
//===----------------------------------------------------------===//
// handle unsigned and signed integral operands
//===----------------------------------------------------------===//
// itoa: handle unsigned integral operands (selected by SFINAE)
template <typename U, std::enable_if_t<!std::is_signed_v<U> && std::is_integral_v<U>> * = nullptr>
ALWAYS_INLINE inline char * itoa(U u, char * p)
{
return convert::uitoa(p, u);
}
// itoa: handle signed integral operands (selected by SFINAE)
template <typename I, size_t N = sizeof(I), std::enable_if_t<std::is_signed_v<I> && std::is_integral_v<I>> * = nullptr>
ALWAYS_INLINE inline char * itoa(I i, char * p)
{
// Need "mask" to be filled with a copy of the sign bit.
// If "i" is a negative value, then the result of "operator >>"
// is implementation-defined, though usually it is an arithmetic
// right shift that replicates the sign bit.
// Use a conditional expression to be portable,
// a good optimizing compiler generates an arithmetic right shift
// and avoids the conditional branch.
UnsignedOfSize<N> mask = i < 0 ? ~UnsignedOfSize<N>(0) : 0;
// Now get the absolute value of "i" and cast to unsigned type UnsignedOfSize<N>.
// Cannot use std::abs() because the result is undefined
// in 2's complement systems for the most-negative value.
// Want to avoid conditional branch for performance reasons since
// CPU branch prediction will be ineffective when negative values
// occur randomly.
// Let "u" be "i" cast to unsigned type UnsignedOfSize<N>.
// Subtract "u" from 2*u if "i" is positive or 0 if "i" is negative.
// This yields the absolute value with the desired type without
// using a conditional branch and without invoking undefined or
// implementation defined behavior:
UnsignedOfSize<N> u = ((2 * UnsignedOfSize<N>(i)) & ~mask) - UnsignedOfSize<N>(i);
// Unconditionally store a minus sign when producing digits
// in a forward direction and increment the pointer only if
// the value is in fact negative.
// This avoids a conditional branch and is safe because we will
// always produce at least one digit and it will overwrite the
// minus sign when the value is not negative.
*p = '-';
p += (mask & 1);
p = convert::uitoa(p, u);
return p;
}
}
const uint64_t max_multiple_of_hundred_that_fits_in_64_bits = 1'00'00'00'00'00'00'00'00'00ull;
const int max_multiple_of_hundred_blocks = 9;
static_assert(max_multiple_of_hundred_that_fits_in_64_bits % 100 == 0);
ALWAYS_INLINE inline char * writeUIntText(UInt128 _x, char * p)
{
/// If we the highest 64bit item is empty, we can print just the lowest item as u64
if (_x.items[UInt128::_impl::little(1)] == 0)
return convert::itoa(_x.items[UInt128::_impl::little(0)], p);
/// Doing operations using __int128 is faster and we already rely on this feature
using T = unsigned __int128;
T x = (T(_x.items[UInt128::_impl::little(1)]) << 64) + T(_x.items[UInt128::_impl::little(0)]);
/// We are going to accumulate blocks of 2 digits to print until the number is small enough to be printed as u64
/// To do this we could do: x / 100, x % 100
/// But these would mean doing many iterations with long integers, so instead we divide by a much longer integer
/// multiple of 100 (100^9) and then get the blocks out of it (as u64)
/// Once we reach u64::max we can stop and use the fast method to print that in the front
static const T large_divisor = max_multiple_of_hundred_that_fits_in_64_bits;
static const T largest_uint64 = std::numeric_limits<uint64_t>::max();
uint8_t two_values[20] = {0}; // 39 Max characters / 2
int current_block = 0;
while (x > largest_uint64)
{
uint64_t u64_remainder = uint64_t(x % large_divisor);
x /= large_divisor;
int pos = current_block;
while (u64_remainder)
{
two_values[pos] = uint8_t(u64_remainder % 100);
pos++;
u64_remainder /= 100;
}
current_block += max_multiple_of_hundred_blocks;
}
char * highest_part_print = convert::itoa(uint64_t(x), p);
for (int i = 0; i < current_block; i++)
{
outTwoDigits(highest_part_print, two_values[current_block - 1 - i]);
highest_part_print += 2;
}
return highest_part_print;
}
ALWAYS_INLINE inline char * writeUIntText(UInt256 _x, char * p)
{
/// If possible, treat it as a smaller integer as they are much faster to print
if (_x.items[UInt256::_impl::little(3)] == 0 && _x.items[UInt256::_impl::little(2)] == 0)
return writeUIntText(UInt128{_x.items[UInt256::_impl::little(0)], _x.items[UInt256::_impl::little(1)]}, p);
/// If available (x86) we transform from our custom class to _BitInt(256) which has better support in the compiler
/// and produces better code
using T =
#if defined(__x86_64__)
# pragma clang diagnostic push
# pragma clang diagnostic ignored "-Wbit-int-extension"
unsigned _BitInt(256)
# pragma clang diagnostic pop
#else
UInt256
#endif
;
#if defined(__x86_64__)
T x = (T(_x.items[UInt256::_impl::little(3)]) << 192) + (T(_x.items[UInt256::_impl::little(2)]) << 128)
+ (T(_x.items[UInt256::_impl::little(1)]) << 64) + T(_x.items[UInt256::_impl::little(0)]);
#else
T x = _x;
#endif
/// Similar to writeUIntText(UInt128) only that in this case we will stop as soon as we reach the largest u128
/// and switch to that function
uint8_t two_values[39] = {0}; // 78 Max characters / 2
int current_pos = 0;
static const T large_divisor = max_multiple_of_hundred_that_fits_in_64_bits;
static const T largest_uint128 = T(std::numeric_limits<uint64_t>::max()) << 64 | T(std::numeric_limits<uint64_t>::max());
while (x > largest_uint128)
{
uint64_t u64_remainder = uint64_t(x % large_divisor);
x /= large_divisor;
int pos = current_pos;
while (u64_remainder)
{
two_values[pos] = uint8_t(u64_remainder % 100);
pos++;
u64_remainder /= 100;
}
current_pos += max_multiple_of_hundred_blocks;
}
#if defined(__x86_64__)
UInt128 pending{uint64_t(x), uint64_t(x >> 64)};
#else
UInt128 pending{x.items[UInt256::_impl::little(0)], x.items[UInt256::_impl::little(1)]};
#endif
char * highest_part_print = writeUIntText(pending, p);
for (int i = 0; i < current_pos; i++)
{
outTwoDigits(highest_part_print, two_values[current_pos - 1 - i]);
highest_part_print += 2;
}
return highest_part_print;
}
ALWAYS_INLINE inline char * writeLeadingMinus(char * pos)
{
*pos = '-';
return pos + 1;
}
template <typename T>
ALWAYS_INLINE inline char * writeSIntText(T x, char * pos)
{
static_assert(std::is_same_v<T, Int128> || std::is_same_v<T, Int256>);
using UnsignedT = make_unsigned_t<T>;
static constexpr T min_int = UnsignedT(1) << (sizeof(T) * 8 - 1);
if (unlikely(x == min_int))
{
if constexpr (std::is_same_v<T, Int128>)
{
const char * res = "-170141183460469231731687303715884105728";
memcpy(pos, res, strlen(res));
return pos + strlen(res);
}
else if constexpr (std::is_same_v<T, Int256>)
{
const char * res = "-57896044618658097711785492504343953926634992332820282019728792003956564819968";
memcpy(pos, res, strlen(res));
return pos + strlen(res);
}
}
if (x < 0)
{
x = -x;
pos = writeLeadingMinus(pos);
}
return writeUIntText(UnsignedT(x), pos);
}
}
char * itoa(UInt8 i, char * p)
{
return convert::itoa(uint8_t(i), p);
}
char * itoa(Int8 i, char * p)
{
return convert::itoa(int8_t(i), p);
}
char * itoa(UInt128 i, char * p)
{
return writeUIntText(i, p);
}
char * itoa(Int128 i, char * p)
{
return writeSIntText(i, p);
}
char * itoa(UInt256 i, char * p)
{
return writeUIntText(i, p);
}
char * itoa(Int256 i, char * p)
{
return writeSIntText(i, p);
}
#define DEFAULT_ITOA(T) \
char * itoa(T i, char * p) \
{ \
return convert::itoa(i, p); \
}
#define FOR_MISSING_INTEGER_TYPES(M) \
M(uint8_t) \
M(UInt16) \
M(UInt32) \
M(UInt64) \
M(int8_t) \
M(Int16) \
M(Int32) \
M(Int64)
FOR_MISSING_INTEGER_TYPES(DEFAULT_ITOA)
#if defined(OS_DARWIN)
DEFAULT_ITOA(unsigned long)
DEFAULT_ITOA(long)
#endif
#undef FOR_MISSING_INTEGER_TYPES
#undef DEFAULT_ITOA

462
base/base/itoa.h
View File

@ -1,446 +1,30 @@
#pragma once
// Based on https://github.com/amdn/itoa and combined with our optimizations
//
//=== itoa.h - Fast integer to ascii conversion --*- C++ -*-//
//
// The MIT License (MIT)
// Copyright (c) 2016 Arturo Martin-de-Nicolas
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
// SOFTWARE.
//===----------------------------------------------------------------------===//
#include <cstdint>
#include <cstddef>
#include <cstring>
#include <type_traits>
#include <base/extended_types.h>
#define FOR_INTEGER_TYPES(M) \
M(uint8_t) \
M(UInt8) \
M(UInt16) \
M(UInt32) \
M(UInt64) \
M(UInt128) \
M(UInt256) \
M(int8_t) \
M(Int8) \
M(Int16) \
M(Int32) \
M(Int64) \
M(Int128) \
M(Int256)
template <typename T>
inline int digits10(T x)
{
if (x < 10ULL)
return 1;
if (x < 100ULL)
return 2;
if (x < 1000ULL)
return 3;
#define INSTANTIATION(T) char * itoa(T i, char * p);
FOR_INTEGER_TYPES(INSTANTIATION)
if (x < 1000000000000ULL)
{
if (x < 100000000ULL)
{
if (x < 1000000ULL)
{
if (x < 10000ULL)
return 4;
else
return 5 + (x >= 100000ULL);
}
#if defined(OS_DARWIN)
INSTANTIATION(unsigned long)
INSTANTIATION(long)
#endif
return 7 + (x >= 10000000ULL);
}
if (x < 10000000000ULL)
return 9 + (x >= 1000000000ULL);
return 11 + (x >= 100000000000ULL);
}
return 12 + digits10(x / 1000000000000ULL);
}
namespace impl
{
template <typename T>
static constexpr T pow10(size_t x)
{
return x ? 10 * pow10<T>(x - 1) : 1;
}
// Division by a power of 10 is implemented using a multiplicative inverse.
// This strength reduction is also done by optimizing compilers, but
// presently the fastest results are produced by using the values
// for the multiplication and the shift as given by the algorithm
// described by Agner Fog in "Optimizing Subroutines in Assembly Language"
//
// http://www.agner.org/optimize/optimizing_assembly.pdf
//
// "Integer division by a constant (all processors)
// A floating point number can be divided by a constant by multiplying
// with the reciprocal. If we want to do the same with integers, we have
// to scale the reciprocal by 2n and then shift the product to the right
// by n. There are various algorithms for finding a suitable value of n
// and compensating for rounding errors. The algorithm described below
// was invented by Terje Mathisen, Norway, and not published elsewhere."
/// Division by constant is performed by:
/// 1. Adding 1 if needed;
/// 2. Multiplying by another constant;
/// 3. Shifting right by another constant.
template <typename UInt, bool add_, UInt multiplier_, unsigned shift_>
struct Division
{
static constexpr bool add{add_};
static constexpr UInt multiplier{multiplier_};
static constexpr unsigned shift{shift_};
};
/// Select a type with appropriate number of bytes from the list of types.
/// First parameter is the number of bytes requested. Then goes a list of types with 1, 2, 4, ... number of bytes.
/// Example: SelectType<4, uint8_t, uint16_t, uint32_t, uint64_t> will select uint32_t.
template <size_t N, typename T, typename... Ts>
struct SelectType
{
using Result = typename SelectType<N / 2, Ts...>::Result;
};
template <typename T, typename... Ts>
struct SelectType<1, T, Ts...>
{
using Result = T;
};
/// Division by 10^N where N is the size of the type.
template <size_t N>
using DivisionBy10PowN = typename SelectType
<
N,
Division<uint8_t, false, 205U, 11>, /// divide by 10
Division<uint16_t, true, 41943U, 22>, /// divide by 100
Division<uint32_t, false, 3518437209U, 45>, /// divide by 10000
Division<uint64_t, false, 12379400392853802749ULL, 90> /// divide by 100000000
>::Result;
template <size_t N>
using UnsignedOfSize = typename SelectType
<
N,
uint8_t,
uint16_t,
uint32_t,
uint64_t,
__uint128_t
>::Result;
/// Holds the result of dividing an unsigned N-byte variable by 10^N resulting in
template <size_t N>
struct QuotientAndRemainder
{
UnsignedOfSize<N> quotient; // quotient with fewer than 2*N decimal digits
UnsignedOfSize<N / 2> remainder; // remainder with at most N decimal digits
};
template <size_t N>
QuotientAndRemainder<N> static inline split(UnsignedOfSize<N> value)
{
constexpr DivisionBy10PowN<N> division;
UnsignedOfSize<N> quotient = (division.multiplier * (UnsignedOfSize<2 * N>(value) + division.add)) >> division.shift;
UnsignedOfSize<N / 2> remainder = static_cast<UnsignedOfSize<N / 2>>(value - quotient * pow10<UnsignedOfSize<N / 2>>(N));
return {quotient, remainder};
}
static inline char * outDigit(char * p, uint8_t value)
{
*p = '0' + value;
++p;
return p;
}
// Using a lookup table to convert binary numbers from 0 to 99
// into ascii characters as described by Andrei Alexandrescu in
// https://www.facebook.com/notes/facebook-engineering/three-optimization-tips-for-c/10151361643253920/
static const char digits[201] = "00010203040506070809"
"10111213141516171819"
"20212223242526272829"
"30313233343536373839"
"40414243444546474849"
"50515253545556575859"
"60616263646566676869"
"70717273747576777879"
"80818283848586878889"
"90919293949596979899";
static inline char * outTwoDigits(char * p, uint8_t value)
{
memcpy(p, &digits[value * 2], 2);
p += 2;
return p;
}
namespace convert
{
template <typename UInt, size_t N = sizeof(UInt)> static char * head(char * p, UInt u);
template <typename UInt, size_t N = sizeof(UInt)> static char * tail(char * p, UInt u);
//===----------------------------------------------------------===//
// head: find most significant digit, skip leading zeros
//===----------------------------------------------------------===//
// "x" contains quotient and remainder after division by 10^N
// quotient is less than 10^N
template <size_t N>
static inline char * head(char * p, QuotientAndRemainder<N> x)
{
p = head(p, UnsignedOfSize<N / 2>(x.quotient));
p = tail(p, x.remainder);
return p;
}
// "u" is less than 10^2*N
template <typename UInt, size_t N>
static inline char * head(char * p, UInt u)
{
return u < pow10<UnsignedOfSize<N>>(N)
? head(p, UnsignedOfSize<N / 2>(u))
: head<N>(p, split<N>(u));
}
// recursion base case, selected when "u" is one byte
template <>
inline char * head<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
return u < 10
? outDigit(p, u)
: outTwoDigits(p, u);
}
//===----------------------------------------------------------===//
// tail: produce all digits including leading zeros
//===----------------------------------------------------------===//
// recursive step, "u" is less than 10^2*N
template <typename UInt, size_t N>
static inline char * tail(char * p, UInt u)
{
QuotientAndRemainder<N> x = split<N>(u);
p = tail(p, UnsignedOfSize<N / 2>(x.quotient));
p = tail(p, x.remainder);
return p;
}
// recursion base case, selected when "u" is one byte
template <>
inline char * tail<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
return outTwoDigits(p, u);
}
//===----------------------------------------------------------===//
// large values are >= 10^2*N
// where x contains quotient and remainder after division by 10^N
//===----------------------------------------------------------===//
template <size_t N>
static inline char * large(char * p, QuotientAndRemainder<N> x)
{
QuotientAndRemainder<N> y = split<N>(x.quotient);
p = head(p, UnsignedOfSize<N / 2>(y.quotient));
p = tail(p, y.remainder);
p = tail(p, x.remainder);
return p;
}
//===----------------------------------------------------------===//
// handle values of "u" that might be >= 10^2*N
// where N is the size of "u" in bytes
//===----------------------------------------------------------===//
template <typename UInt, size_t N = sizeof(UInt)>
static inline char * uitoa(char * p, UInt u)
{
if (u < pow10<UnsignedOfSize<N>>(N))
return head(p, UnsignedOfSize<N / 2>(u));
QuotientAndRemainder<N> x = split<N>(u);
return u < pow10<UnsignedOfSize<N>>(2 * N)
? head<N>(p, x)
: large<N>(p, x);
}
// selected when "u" is one byte
template <>
inline char * uitoa<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
{
if (u < 10)
return outDigit(p, u);
else if (u < 100)
return outTwoDigits(p, u);
else
{
p = outDigit(p, u / 100);
p = outTwoDigits(p, u % 100);
return p;
}
}
//===----------------------------------------------------------===//
// handle unsigned and signed integral operands
//===----------------------------------------------------------===//
// itoa: handle unsigned integral operands (selected by SFINAE)
template <typename U, std::enable_if_t<!std::is_signed_v<U> && std::is_integral_v<U>> * = nullptr>
static inline char * itoa(U u, char * p)
{
return convert::uitoa(p, u);
}
// itoa: handle signed integral operands (selected by SFINAE)
template <typename I, size_t N = sizeof(I), std::enable_if_t<std::is_signed_v<I> && std::is_integral_v<I>> * = nullptr>
static inline char * itoa(I i, char * p)
{
// Need "mask" to be filled with a copy of the sign bit.
// If "i" is a negative value, then the result of "operator >>"
// is implementation-defined, though usually it is an arithmetic
// right shift that replicates the sign bit.
// Use a conditional expression to be portable,
// a good optimizing compiler generates an arithmetic right shift
// and avoids the conditional branch.
UnsignedOfSize<N> mask = i < 0 ? ~UnsignedOfSize<N>(0) : 0;
// Now get the absolute value of "i" and cast to unsigned type UnsignedOfSize<N>.
// Cannot use std::abs() because the result is undefined
// in 2's complement systems for the most-negative value.
// Want to avoid conditional branch for performance reasons since
// CPU branch prediction will be ineffective when negative values
// occur randomly.
// Let "u" be "i" cast to unsigned type UnsignedOfSize<N>.
// Subtract "u" from 2*u if "i" is positive or 0 if "i" is negative.
// This yields the absolute value with the desired type without
// using a conditional branch and without invoking undefined or
// implementation defined behavior:
UnsignedOfSize<N> u = ((2 * UnsignedOfSize<N>(i)) & ~mask) - UnsignedOfSize<N>(i);
// Unconditionally store a minus sign when producing digits
// in a forward direction and increment the pointer only if
// the value is in fact negative.
// This avoids a conditional branch and is safe because we will
// always produce at least one digit and it will overwrite the
// minus sign when the value is not negative.
*p = '-';
p += (mask & 1);
p = convert::uitoa(p, u);
return p;
}
}
template <typename T>
static inline char * writeUIntText(T x, char * p)
{
static_assert(is_unsigned_v<T>);
int len = digits10(x);
auto * pp = p + len;
while (x >= 100)
{
const auto i = x % 100;
x /= 100;
pp -= 2;
outTwoDigits(pp, i);
}
if (x < 10)
*p = '0' + x;
else
outTwoDigits(p, x);
return p + len;
}
static inline char * writeLeadingMinus(char * pos)
{
*pos = '-';
return pos + 1;
}
template <typename T>
static inline char * writeSIntText(T x, char * pos)
{
static_assert(std::is_same_v<T, Int128> || std::is_same_v<T, Int256>);
using UnsignedT = make_unsigned_t<T>;
static constexpr T min_int = UnsignedT(1) << (sizeof(T) * 8 - 1);
if (unlikely(x == min_int))
{
if constexpr (std::is_same_v<T, Int128>)
{
const char * res = "-170141183460469231731687303715884105728";
memcpy(pos, res, strlen(res));
return pos + strlen(res);
}
else if constexpr (std::is_same_v<T, Int256>)
{
const char * res = "-57896044618658097711785492504343953926634992332820282019728792003956564819968";
memcpy(pos, res, strlen(res));
return pos + strlen(res);
}
}
if (x < 0)
{
x = -x;
pos = writeLeadingMinus(pos);
}
return writeUIntText(UnsignedT(x), pos);
}
}
template <typename I>
char * itoa(I i, char * p)
{
return impl::convert::itoa(i, p);
}
template <>
inline char * itoa(char8_t i, char * p)
{
return impl::convert::itoa(uint8_t(i), p);
}
template <>
inline char * itoa(UInt128 i, char * p)
{
return impl::writeUIntText(i, p);
}
template <>
inline char * itoa(Int128 i, char * p)
{
return impl::writeSIntText(i, p);
}
template <>
inline char * itoa(UInt256 i, char * p)
{
return impl::writeUIntText(i, p);
}
template <>
inline char * itoa(Int256 i, char * p)
{
return impl::writeSIntText(i, p);
}
#undef FOR_INTEGER_TYPES
#undef INSTANTIATION

11
docker/test/fuzzer/run-fuzzer.sh
View File

@ -173,16 +173,15 @@ function fuzz
mkdir -p /var/run/clickhouse-server
# NOTE: we use process substitution here to preserve keep $! as a pid of clickhouse-server
# server.log -> CH logs
# stderr.log -> Process logs (sanitizer)
# server.log -> All server logs, including sanitizer
# stderr.log -> Process logs (sanitizer) only
clickhouse-server \
--config-file db/config.xml \
--pid-file /var/run/clickhouse-server/clickhouse-server.pid \
-- --path db \
--logger.console=0 \
--logger.log=server.log > stderr.log 2>&1 &
server_pid=$!
--logger.log=server.log 2>&1 | tee -a stderr.log >> server.log 2>&1 &
server_pid=$(pidof clickhouse-server)
kill -0 $server_pid
@ -310,7 +309,7 @@ quit
if [ "$server_died" == 1 ]
then
# The server has died.
if ! rg --text -o 'Received signal.*|Logical error.*|Assertion.*failed|Failed assertion.*|.*runtime error: .*|.*is located.*|(SUMMARY|ERROR): [a-zA-Z]+Sanitizer:.*|.*_LIBCPP_ASSERT.*|.*Child process was terminated by signal 9.*' server.log stderr.log > description.txt
if ! rg --text -o 'Received signal.*|Logical error.*|Assertion.*failed|Failed assertion.*|.*runtime error: .*|.*is located.*|(SUMMARY|ERROR): [a-zA-Z]+Sanitizer:.*|.*_LIBCPP_ASSERT.*|.*Child process was terminated by signal 9.*' server.log > description.txt
then
echo "Lost connection to server. See the logs." > description.txt
fi

1
docker/test/integration/runner/Dockerfile
View File

@ -126,7 +126,6 @@ RUN set -x \
COPY modprobe.sh /usr/local/bin/modprobe
COPY dockerd-entrypoint.sh /usr/local/bin/
COPY compose/ /compose/
COPY misc/ /misc/

40
docker/test/stateless/run.sh
View File

@ -51,10 +51,7 @@ fi
config_logs_export_cluster /etc/clickhouse-server/config.d/system_logs_export.yaml
if [[ -n "$BUGFIX_VALIDATE_CHECK" ]] && [[ "$BUGFIX_VALIDATE_CHECK" -eq 1 ]]; then
sudo cat /etc/clickhouse-server/config.d/zookeeper.xml \
| sed "/<use_compression>1<\/use_compression>/d" \
> /etc/clickhouse-server/config.d/zookeeper.xml.tmp
sudo mv /etc/clickhouse-server/config.d/zookeeper.xml.tmp /etc/clickhouse-server/config.d/zookeeper.xml
sudo sed -i "/<use_compression>1<\/use_compression>/d" /etc/clickhouse-server/config.d/zookeeper.xml
# it contains some new settings, but we can safely remove it
rm /etc/clickhouse-server/config.d/handlers.yaml
@ -62,23 +59,14 @@ if [[ -n "$BUGFIX_VALIDATE_CHECK" ]] && [[ "$BUGFIX_VALIDATE_CHECK" -eq 1 ]]; th
rm /etc/clickhouse-server/config.d/zero_copy_destructive_operations.xml
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/azure_storage_conf.xml \
| sed "s|<object_storage_type>azure|<object_storage_type>azure_blob_storage|" \
> /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp /etc/clickhouse-server/config.d/azure_storage_conf.xml
sudo sed -i "s|<object_storage_type>azure<|<object_storage_type>azure_blob_storage<|" /etc/clickhouse-server/config.d/azure_storage_conf.xml
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/storage_conf.xml \
| sed "s|<object_storage_type>local|<object_storage_type>local_blob_storage|" \
> /etc/clickhouse-server/config.d/storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/storage_conf.xml.tmp /etc/clickhouse-server/config.d/storage_conf.xml
sudo sed -i "s|<object_storage_type>local<|<object_storage_type>local_blob_storage<|" /etc/clickhouse-server/config.d/storage_conf.xml
function remove_keeper_config()
{
sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \
| sed "/<$1>$2<\/$1>/d" \
> /etc/clickhouse-server/config.d/keeper_port.xml.tmp
sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml
sudo sed -i "/<$1>$2<\/$1>/d" /etc/clickhouse-server/config.d/keeper_port.xml
}
# commit_logs_cache_size_threshold setting doesn't exist on some older versions
remove_keeper_config "commit_logs_cache_size_threshold" "[[:digit:]]\+"
@ -113,25 +101,13 @@ else
fi
if [[ -n "$USE_DATABASE_REPLICATED" ]] && [[ "$USE_DATABASE_REPLICATED" -eq 1 ]]; then
sudo cat /etc/clickhouse-server1/config.d/filesystem_caches_path.xml \
| sed "s|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches/</filesystem_caches_path>|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches_1/</filesystem_caches_path>|" \
> /etc/clickhouse-server1/config.d/filesystem_caches_path.xml.tmp
mv /etc/clickhouse-server1/config.d/filesystem_caches_path.xml.tmp /etc/clickhouse-server1/config.d/filesystem_caches_path.xml
sudo sed -i "s|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches/</filesystem_caches_path>|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches_1/</filesystem_caches_path>|" /etc/clickhouse-server1/config.d/filesystem_caches_path.xml
sudo cat /etc/clickhouse-server2/config.d/filesystem_caches_path.xml \
| sed "s|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches/</filesystem_caches_path>|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches_2/</filesystem_caches_path>|" \
> /etc/clickhouse-server2/config.d/filesystem_caches_path.xml.tmp
mv /etc/clickhouse-server2/config.d/filesystem_caches_path.xml.tmp /etc/clickhouse-server2/config.d/filesystem_caches_path.xml
sudo sed -i "s|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches/</filesystem_caches_path>|<filesystem_caches_path>/var/lib/clickhouse/filesystem_caches_2/</filesystem_caches_path>|" /etc/clickhouse-server2/config.d/filesystem_caches_path.xml
sudo cat /etc/clickhouse-server1/config.d/filesystem_caches_path.xml \
| sed "s|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches/</custom_cached_disks_base_directory>|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches_1/</custom_cached_disks_base_directory>|" \
> /etc/clickhouse-server1/config.d/filesystem_caches_path.xml.tmp
mv /etc/clickhouse-server1/config.d/filesystem_caches_path.xml.tmp /etc/clickhouse-server1/config.d/filesystem_caches_path.xml
sudo sed -i "s|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches/</custom_cached_disks_base_directory>|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches_1/</custom_cached_disks_base_directory>|" /etc/clickhouse-server1/config.d/filesystem_caches_path.xml
sudo cat /etc/clickhouse-server2/config.d/filesystem_caches_path.xml \
| sed "s|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches/</custom_cached_disks_base_directory>|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches_2/</custom_cached_disks_base_directory>|" \
> /etc/clickhouse-server2/config.d/filesystem_caches_path.xml.tmp
mv /etc/clickhouse-server2/config.d/filesystem_caches_path.xml.tmp /etc/clickhouse-server2/config.d/filesystem_caches_path.xml
sudo sed -i "s|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches/</custom_cached_disks_base_directory>|<custom_cached_disks_base_directory replace=\"replace\">/var/lib/clickhouse/filesystem_caches_2/</custom_cached_disks_base_directory>|" /etc/clickhouse-server2/config.d/filesystem_caches_path.xml
mkdir -p /var/run/clickhouse-server1
sudo chown clickhouse:clickhouse /var/run/clickhouse-server1

40
docker/test/upgrade/run.sh
View File

@ -67,10 +67,7 @@ configure
function remove_keeper_config()
{
sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \
| sed "/<$1>$2<\/$1>/d" \
> /etc/clickhouse-server/config.d/keeper_port.xml.tmp
sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml
sudo sed -i "/<$1>$2<\/$1>/d" /etc/clickhouse-server/config.d/keeper_port.xml
}
# async_replication setting doesn't exist on some older versions
@ -80,16 +77,10 @@ remove_keeper_config "async_replication" "1"
remove_keeper_config "create_if_not_exists" "[01]"
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/azure_storage_conf.xml \
| sed "s|<object_storage_type>azure|<object_storage_type>azure_blob_storage|" \
> /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp /etc/clickhouse-server/config.d/azure_storage_conf.xml
sudo sed -i "s|<object_storage_type>azure<|<object_storage_type>azure_blob_storage<|" /etc/clickhouse-server/config.d/azure_storage_conf.xml
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/storage_conf.xml \
| sed "s|<object_storage_type>local|<object_storage_type>local_blob_storage|" \
> /etc/clickhouse-server/config.d/storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/storage_conf.xml.tmp /etc/clickhouse-server/config.d/storage_conf.xml
sudo sed -i "s|<object_storage_type>local<|<object_storage_type>local_blob_storage<|" /etc/clickhouse-server/config.d/storage_conf.xml
# latest_logs_cache_size_threshold setting doesn't exist on some older versions
remove_keeper_config "latest_logs_cache_size_threshold" "[[:digit:]]\+"
@ -120,22 +111,13 @@ export ZOOKEEPER_FAULT_INJECTION=0
configure
# force_sync=false doesn't work correctly on some older versions
sudo cat /etc/clickhouse-server/config.d/keeper_port.xml \
| sed "s|<force_sync>false</force_sync>|<force_sync>true</force_sync>|" \
> /etc/clickhouse-server/config.d/keeper_port.xml.tmp
sudo mv /etc/clickhouse-server/config.d/keeper_port.xml.tmp /etc/clickhouse-server/config.d/keeper_port.xml
sudo sed -i "s|<force_sync>false</force_sync>|<force_sync>true</force_sync>|" /etc/clickhouse-server/config.d/keeper_port.xml
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/azure_storage_conf.xml \
| sed "s|<object_storage_type>azure|<object_storage_type>azure_blob_storage|" \
> /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/azure_storage_conf.xml.tmp /etc/clickhouse-server/config.d/azure_storage_conf.xml
sudo sed -i "s|<object_storage_type>azure<|<object_storage_type>azure_blob_storage<|" /etc/clickhouse-server/config.d/azure_storage_conf.xml
#todo: remove these after 24.3 released.
sudo cat /etc/clickhouse-server/config.d/storage_conf.xml \
| sed "s|<object_storage_type>local|<object_storage_type>local_blob_storage|" \
> /etc/clickhouse-server/config.d/storage_conf.xml.tmp
sudo mv /etc/clickhouse-server/config.d/storage_conf.xml.tmp /etc/clickhouse-server/config.d/storage_conf.xml
sudo sed -i "s|<object_storage_type>local<|<object_storage_type>local_blob_storage<|" /etc/clickhouse-server/config.d/storage_conf.xml
# async_replication setting doesn't exist on some older versions
remove_keeper_config "async_replication" "1"
@ -150,10 +132,7 @@ remove_keeper_config "latest_logs_cache_size_threshold" "[[:digit:]]\+"
remove_keeper_config "commit_logs_cache_size_threshold" "[[:digit:]]\+"
# But we still need default disk because some tables loaded only into it
sudo cat /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml \
| sed "s|<main><disk>s3</disk></main>|<main><disk>s3</disk></main><default><disk>default</disk></default>|" \
> /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp
mv /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
sudo sed -i "s|<main><disk>s3</disk></main>|<main><disk>s3</disk></main><default><disk>default</disk></default>|" /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
sudo chown clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
sudo chgrp clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
@ -256,10 +235,7 @@ then
fi
# Just in case previous version left some garbage in zk
sudo cat /etc/clickhouse-server/config.d/lost_forever_check.xml \
| sed "s|>1<|>0<|g" \
> /etc/clickhouse-server/config.d/lost_forever_check.xml.tmp
sudo mv /etc/clickhouse-server/config.d/lost_forever_check.xml.tmp /etc/clickhouse-server/config.d/lost_forever_check.xml
sudo sed -i "s|>1<|>0<|g" /etc/clickhouse-server/config.d/lost_forever_check.xml \
rm /etc/clickhouse-server/config.d/filesystem_caches_path.xml
start 500

4
docs/en/operations/server-configuration-parameters/settings.md
View File

@ -933,9 +933,9 @@ Hard limit is configured via system tools
## database_atomic_delay_before_drop_table_sec {#database_atomic_delay_before_drop_table_sec}
The delay before a table data is dropped in seconds. If the `DROP TABLE` query has a `SYNC` modifier, this setting is ignored.
Sets the delay before remove table data in seconds. If the query has `SYNC` modifier, this setting is ignored.
Default value: `480` (8 minutes).
Default value: `480` (8 minute).
## database_catalog_unused_dir_hide_timeout_sec {#database_catalog_unused_dir_hide_timeout_sec}

52
docs/en/sql-reference/statements/undrop.md
View File

@ -13,6 +13,13 @@ a system table called `system.dropped_tables`.
If you have a materialized view without a `TO` clause associated with the dropped table, then you will also have to UNDROP the inner table of that view.
:::note
UNDROP TABLE is experimental. To use it add this setting:
```sql
set allow_experimental_undrop_table_query = 1;
```
:::
:::tip
Also see [DROP TABLE](/docs/en/sql-reference/statements/drop.md)
:::
@ -25,53 +32,60 @@ UNDROP TABLE [db.]name [UUID '<uuid>'] [ON CLUSTER cluster]
**Example**
``` sql
set allow_experimental_undrop_table_query = 1;
```
```sql
CREATE TABLE tab
CREATE TABLE undropMe
(
`id` UInt8
)
ENGINE = MergeTree
ORDER BY id;
DROP TABLE tab;
SELECT *
FROM system.dropped_tables
FORMAT Vertical;
ORDER BY id
```
```sql
DROP TABLE undropMe
```
```sql
SELECT *
FROM system.dropped_tables
FORMAT Vertical
```
```response
Row 1:
──────
index: 0
database: default
table: tab
table: undropMe
uuid: aa696a1a-1d70-4e60-a841-4c80827706cc
engine: MergeTree
metadata_dropped_path: /var/lib/clickhouse/metadata_dropped/default.tab.aa696a1a-1d70-4e60-a841-4c80827706cc.sql
metadata_dropped_path: /var/lib/clickhouse/metadata_dropped/default.undropMe.aa696a1a-1d70-4e60-a841-4c80827706cc.sql
table_dropped_time: 2023-04-05 14:12:12
1 row in set. Elapsed: 0.001 sec.
```
```sql
UNDROP TABLE tab;
UNDROP TABLE undropMe
```
```response
Ok.
```
```sql
SELECT *
FROM system.dropped_tables
FORMAT Vertical;
FORMAT Vertical
```
```response
Ok.
0 rows in set. Elapsed: 0.001 sec.
```
```sql
DESCRIBE TABLE tab
FORMAT Vertical;
DESCRIBE TABLE undropMe
FORMAT Vertical
```
```response
Row 1:
──────

3
programs/benchmark/Benchmark.cpp
View File

@ -34,6 +34,7 @@
#include <Common/StudentTTest.h>
#include <Common/CurrentMetrics.h>
#include <Common/ErrorCodes.h>
#include <Core/BaseSettingsProgramOptions.h>
/** A tool for evaluating ClickHouse performance.
@ -623,7 +624,7 @@ int mainEntryClickHouseBenchmark(int argc, char ** argv)
;
Settings settings;
settings.addProgramOptions(desc);
addProgramOptions(settings, desc);
boost::program_options::variables_map options;
boost::program_options::store(boost::program_options::parse_command_line(argc, argv, desc), options);

3
programs/format/Format.cpp
View File

@ -17,6 +17,7 @@
#include <Common/ErrorCodes.h>
#include <Common/StringUtils/StringUtils.h>
#include <Common/TerminalSize.h>
#include <Core/BaseSettingsProgramOptions.h>
#include <Interpreters/Context.h>
#include <Functions/FunctionFactory.h>
@ -102,7 +103,7 @@ int mainEntryClickHouseFormat(int argc, char ** argv)
{
std::string_view name = field.getName();
if (name == "max_parser_depth" || name == "max_query_size")
cmd_settings.addProgramOption(desc, name, field);
addProgramOption(cmd_settings, desc, name, field);
}
boost::program_options::variables_map options;

1
programs/server/Server.cpp
View File

@ -36,6 +36,7 @@
#include <Common/getNumberOfPhysicalCPUCores.h>
#include <Common/getExecutablePath.h>
#include <Common/ProfileEvents.h>
#include <Common/Scheduler/IResourceManager.h>
#include <Common/ThreadProfileEvents.h>
#include <Common/ThreadStatus.h>
#include <Common/getMappedArea.h>

1
src/Access/Common/QuotaDefs.cpp
View File

@ -2,6 +2,7 @@
#include <Common/Exception.h>
#include <IO/ReadHelpers.h>
#include <IO/WriteHelpers.h>
#include <base/range.h>
#include <boost/algorithm/string/case_conv.hpp>
#include <boost/algorithm/string/classification.hpp>

2
src/Access/IAccessStorage.h
View File

@ -13,6 +13,8 @@
#include <optional>
#include <vector>
#include <boost/noncopyable.hpp>
namespace Poco { class Logger; }
namespace Poco::Net { class IPAddress; }

1
src/Access/User.cpp
View File

@ -1,4 +1,5 @@
#include <Access/User.h>
#include <Common/StringUtils/StringUtils.h>
#include <Core/Protocol.h>
#include <base/insertAtEnd.h>

8
src/AggregateFunctions/AggregateFunctionRetention.cpp
View File

@ -1,16 +1,16 @@
#include <AggregateFunctions/AggregateFunctionFactory.h>
#include <AggregateFunctions/IAggregateFunction.h>
#include <AggregateFunctions/FactoryHelpers.h>
#include <unordered_set>
#include <Columns/ColumnArray.h>
#include <Common/assert_cast.h>
#include <DataTypes/DataTypesNumber.h>
#include <DataTypes/DataTypeArray.h>
#include <IO/ReadHelpers.h>
#include <IO/WriteHelpers.h>
#include <bitset>
#include <base/range.h>
#include <AggregateFunctions/IAggregateFunction.h>
#include <bitset>
#include <unordered_set>
namespace DB

2
src/AggregateFunctions/AggregateFunctionSequenceMatch.cpp
View File

@ -10,6 +10,8 @@
#include <Common/assert_cast.h>
#include <IO/ReadHelpers.h>
#include <IO/WriteHelpers.h>
#include <base/range.h>
#include <bitset>
#include <stack>

219
src/Analyzer/Passes/QueryAnalysisPass.cpp
View File

@ -48,6 +48,7 @@
#include <Interpreters/SelectQueryOptions.h>
#include <Interpreters/Set.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/ExternalDictionariesLoader.h>
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
@ -776,6 +777,7 @@ struct IdentifierResolveScope
std::unordered_map<QueryTreeNodePtr, TableExpressionData> table_expression_node_to_data;
QueryTreeNodePtrWithHashSet nullable_group_by_keys;
QueryTreeNodePtrWithHashMap<QueryTreeNodePtr> nullable_join_columns;
/// Use identifier lookup to result cache
bool use_identifier_lookup_to_result_cache = true;
@ -1276,7 +1278,11 @@ private:
return {};
}
static void convertJoinedColumnTypeToNullIfNeeded(QueryTreeNodePtr & resolved_identifier, const JoinKind & join_kind, std::optional<JoinTableSide> resolved_side)
static QueryTreeNodePtr convertJoinedColumnTypeToNullIfNeeded(
const QueryTreeNodePtr & resolved_identifier,
const JoinKind & join_kind,
std::optional<JoinTableSide> resolved_side,
IdentifierResolveScope & scope)
{
if (resolved_identifier->getNodeType() == QueryTreeNodeType::COLUMN &&
JoinCommon::canBecomeNullable(resolved_identifier->getResultType()) &&
@ -1284,9 +1290,20 @@ private:
(isLeft(join_kind) && resolved_side && *resolved_side == JoinTableSide::Right) ||
(isRight(join_kind) && resolved_side && *resolved_side == JoinTableSide::Left)))
{
auto & resolved_column = resolved_identifier->as<ColumnNode &>();
resolved_column.setColumnType(makeNullableOrLowCardinalityNullable(resolved_column.getColumnType()));
auto nullable_resolved_identifier = resolved_identifier->clone();
auto & resolved_column = nullable_resolved_identifier->as<ColumnNode &>();
auto new_result_type = makeNullableOrLowCardinalityNullable(resolved_column.getColumnType());
resolved_column.setColumnType(new_result_type);
if (resolved_column.hasExpression())
{
auto & resolved_expression = resolved_column.getExpression();
if (!resolved_expression->getResultType()->equals(*new_result_type))
resolved_expression = buildCastFunction(resolved_expression, new_result_type, scope.context, true);
}
scope.nullable_join_columns[nullable_resolved_identifier] = resolved_identifier;
return nullable_resolved_identifier;
}
return nullptr;
}
/// Resolve identifier functions
@ -3258,6 +3275,32 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromTableExpression(const Id
return {};
}
QueryTreeNodePtr checkIsMissedObjectJSONSubcolumn(const QueryTreeNodePtr & left_resolved_identifier,
const QueryTreeNodePtr & right_resolved_identifier)
{
if (left_resolved_identifier && right_resolved_identifier && left_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT
&& right_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
auto & left_resolved_column = left_resolved_identifier->as<ConstantNode &>();
auto & right_resolved_column = right_resolved_identifier->as<ConstantNode &>();
if (left_resolved_column.getValueStringRepresentation() == "NULL" && right_resolved_column.getValueStringRepresentation() == "NULL")
return left_resolved_identifier;
}
else if (left_resolved_identifier && left_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
auto & left_resolved_column = left_resolved_identifier->as<ConstantNode &>();
if (left_resolved_column.getValueStringRepresentation() == "NULL")
return left_resolved_identifier;
}
else if (right_resolved_identifier && right_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
auto & right_resolved_column = right_resolved_identifier->as<ConstantNode &>();
if (right_resolved_column.getValueStringRepresentation() == "NULL")
return right_resolved_identifier;
}
return {};
}
QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromJoin(const IdentifierLookup & identifier_lookup,
const QueryTreeNodePtr & table_expression_node,
IdentifierResolveScope & scope)
@ -3358,28 +3401,8 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromJoin(const IdentifierLoo
/// If columns from left or right table were missed Object(Nullable('json')) subcolumns, they will be replaced
/// to ConstantNode(NULL), which can't be cast to ColumnNode, so we resolve it here.
if (left_resolved_identifier && right_resolved_identifier && left_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT
&& right_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
auto & left_resolved_column = left_resolved_identifier->as<ConstantNode &>();
auto & right_resolved_column = right_resolved_identifier->as<ConstantNode &>();
if (left_resolved_column.getValueStringRepresentation() == "NULL" && right_resolved_column.getValueStringRepresentation() == "NULL")
return left_resolved_identifier;
}
else if (left_resolved_identifier && left_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
resolved_side = JoinTableSide::Left;
auto & left_resolved_column = left_resolved_identifier->as<ConstantNode &>();
if (left_resolved_column.getValueStringRepresentation() == "NULL")
return left_resolved_identifier;
}
else if (right_resolved_identifier && right_resolved_identifier->getNodeType() == QueryTreeNodeType::CONSTANT)
{
resolved_side = JoinTableSide::Right;
auto & right_resolved_column = right_resolved_identifier->as<ConstantNode &>();
if (right_resolved_column.getValueStringRepresentation() == "NULL")
return right_resolved_identifier;
}
if (auto missed_subcolumn_identifier = checkIsMissedObjectJSONSubcolumn(left_resolved_identifier, right_resolved_identifier))
return missed_subcolumn_identifier;
if (left_resolved_identifier && right_resolved_identifier)
{
@ -3521,8 +3544,9 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromJoin(const IdentifierLoo
if (scope.join_use_nulls)
{
resolved_identifier = resolved_identifier->clone();
convertJoinedColumnTypeToNullIfNeeded(resolved_identifier, join_kind, resolved_side);
auto nullable_resolved_identifier = convertJoinedColumnTypeToNullIfNeeded(resolved_identifier, join_kind, resolved_side, scope);
if (nullable_resolved_identifier)
resolved_identifier = nullable_resolved_identifier;
}
return resolved_identifier;
@ -4402,6 +4426,28 @@ QueryAnalyzer::QueryTreeNodesWithNames QueryAnalyzer::resolveUnqualifiedMatcher(
const auto & join_using_column_nodes_list = join_using_column_node.getExpressionOrThrow()->as<ListNode &>();
const auto & join_using_column_nodes = join_using_column_nodes_list.getNodes();
/** If column doesn't exists in the table, then do not match column from USING clause.
* Example: SELECT a + 1 AS id, * FROM (SELECT 1 AS a) AS t1 JOIN (SELECT 2 AS id) AS t2 USING (id);
* In this case `id` is not present in the left table expression,
* so asterisk should return `id` from the right table expression.
*/
auto is_column_from_parent_scope = [&scope](const QueryTreeNodePtr & using_node_from_table)
{
const auto & using_column_from_table = using_node_from_table->as<ColumnNode &>();
auto table_expression_data_it = scope.table_expression_node_to_data.find(using_column_from_table.getColumnSource());
if (table_expression_data_it != scope.table_expression_node_to_data.end())
{
const auto & table_expression_data = table_expression_data_it->second;
const auto & column_name = using_column_from_table.getColumnName();
return !table_expression_data.column_name_to_column_node.contains(column_name);
}
return false;
};
if (is_column_from_parent_scope(join_using_column_nodes.at(0)) ||
is_column_from_parent_scope(join_using_column_nodes.at(1)))
continue;
QueryTreeNodePtr matched_column_node;
if (isRight(join_node->getKind()))
@ -4523,7 +4569,15 @@ ProjectionNames QueryAnalyzer::resolveMatcher(QueryTreeNodePtr & matcher_node, I
for (auto & [node, node_name] : matched_expression_nodes_with_names)
{
auto join_identifier_side = getColumnSideFromJoinTree(node, *nearest_scope_join_node);
convertJoinedColumnTypeToNullIfNeeded(node, nearest_scope_join_node->getKind(), join_identifier_side);
auto projection_name_it = node_to_projection_name.find(node);
auto nullable_node = convertJoinedColumnTypeToNullIfNeeded(node, nearest_scope_join_node->getKind(), join_identifier_side, scope);
if (nullable_node)
{
node = nullable_node;
/// Set the same projection name for new nullable node
if (projection_name_it != node_to_projection_name.end())
node_to_projection_name.emplace(node, projection_name_it->second);
}
}
}
}
@ -7304,8 +7358,64 @@ void QueryAnalyzer::resolveJoin(QueryTreeNodePtr & join_node, IdentifierResolveS
join_using_identifiers.insert(identifier_full_name);
const auto & settings = scope.context->getSettingsRef();
/** While resolving JOIN USING identifier, try to resolve identifier from parent subquery projection.
* Example: SELECT a + 1 AS b FROM (SELECT 1 AS a) t1 JOIN (SELECT 2 AS b) USING b
* In this case `b` is not in the left table expression, but it is in the parent subquery projection.
*/
auto try_resolve_identifier_from_query_projection = [this](const String & identifier_full_name_,
const QueryTreeNodePtr & left_table_expression,
const IdentifierResolveScope & scope_) -> QueryTreeNodePtr
{
const QueryNode * query_node = scope_.scope_node ? scope_.scope_node->as<QueryNode>() : nullptr;
if (!query_node)
return nullptr;
const auto & projection_list = query_node->getProjection();
for (const auto & projection_node : projection_list.getNodes())
{
if (projection_node->hasAlias() && identifier_full_name_ == projection_node->getAlias())
{
auto left_subquery = std::make_shared<QueryNode>(query_node->getMutableContext());
left_subquery->getProjection().getNodes().push_back(projection_node->clone());
left_subquery->getJoinTree() = left_table_expression;
IdentifierResolveScope left_subquery_scope(left_subquery, nullptr /*parent_scope*/);
resolveQuery(left_subquery, left_subquery_scope);
const auto & resolved_nodes = left_subquery->getProjection().getNodes();
if (resolved_nodes.size() == 1)
{
/// Create ColumnNode with expression from parent projection
return std::make_shared<ColumnNode>(
NameAndTypePair{identifier_full_name_, resolved_nodes.front()->getResultType()},
resolved_nodes.front(), left_table_expression);
}
}
}
return nullptr;
};
QueryTreeNodePtr result_left_table_expression = nullptr;
/** With `analyzer_compatibility_join_using_top_level_identifier` alias in projection has higher priority than column from left table.
* But if aliased expression cannot be resolved from left table, we get UNKNOW_IDENTIFIER error,
* despite the fact that column from USING could be resolved from left table.
* It's compatibility with a default behavior for old analyzer.
*/
if (settings.analyzer_compatibility_join_using_top_level_identifier)
result_left_table_expression = try_resolve_identifier_from_query_projection(identifier_full_name, join_node_typed.getLeftTableExpression(), scope);
IdentifierLookup identifier_lookup{identifier_node->getIdentifier(), IdentifierLookupContext::EXPRESSION};
auto result_left_table_expression = tryResolveIdentifierFromJoinTreeNode(identifier_lookup, join_node_typed.getLeftTableExpression(), scope);
if (!result_left_table_expression)
result_left_table_expression = tryResolveIdentifierFromJoinTreeNode(identifier_lookup, join_node_typed.getLeftTableExpression(), scope);
/// Here we may try to resolve identifier from projection in case it's not resolved from left table expression
/// and analyzer_compatibility_join_using_top_level_identifier is disabled.
/// For now we do not do this, because not all corner cases are clear.
/// if (!settings.analyzer_compatibility_join_using_top_level_identifier && !result_left_table_expression)
/// result_left_table_expression = try_resolve_identifier_from_query_projection(identifier_full_name, join_node_typed.getLeftTableExpression(), scope);
if (!result_left_table_expression)
throw Exception(ErrorCodes::UNKNOWN_IDENTIFIER,
"JOIN {} using identifier '{}' cannot be resolved from left table expression. In scope {}",
@ -7448,6 +7558,29 @@ void QueryAnalyzer::resolveQueryJoinTreeNode(QueryTreeNodePtr & join_tree_node,
scope.table_expressions_in_resolve_process.erase(join_tree_node.get());
}
class ReplaceColumnsVisitor : public InDepthQueryTreeVisitor<ReplaceColumnsVisitor>
{
public:
explicit ReplaceColumnsVisitor(const QueryTreeNodePtrWithHashMap<QueryTreeNodePtr> & replacement_map_, const ContextPtr & context_)
: replacement_map(replacement_map_)
, context(context_)
{}
void visitImpl(QueryTreeNodePtr & node)
{
if (auto it = replacement_map.find(node); it != replacement_map.end())
node = it->second;
if (auto * function_node = node->as<FunctionNode>())
rerunFunctionResolve(function_node, context);
}
bool shouldTraverseTopToBottom() const { return false; }
private:
const QueryTreeNodePtrWithHashMap<QueryTreeNodePtr> & replacement_map;
const ContextPtr & context;
};
/** Resolve query.
* This function modifies query node during resolve. It is caller responsibility to clone query node before resolve
* if it is needed for later use.
@ -7635,21 +7768,23 @@ void QueryAnalyzer::resolveQuery(const QueryTreeNodePtr & query_node, Identifier
scope.scope_node->formatASTForErrorMessage());
}
if (query_node_typed.getPrewhere())
if (auto & prewhere_node = query_node_typed.getPrewhere())
{
/** Expression in PREWHERE with JOIN should not be modified by join_use_nulls.
* Example: SELECT * FROM t1 JOIN t2 USING (id) PREWHERE a = 1
* Column `a` should be resolved from table and should not change its type to Nullable.
*/
bool join_use_nulls = scope.join_use_nulls;
bool use_identifier_lookup_to_result_cache = scope.use_identifier_lookup_to_result_cache;
scope.join_use_nulls = false;
scope.use_identifier_lookup_to_result_cache = false;
resolveExpressionNode(prewhere_node, scope, false /*allow_lambda_expression*/, false /*allow_table_expression*/);
resolveExpressionNode(query_node_typed.getPrewhere(), scope, false /*allow_lambda_expression*/, false /*allow_table_expression*/);
scope.join_use_nulls = join_use_nulls;
scope.use_identifier_lookup_to_result_cache = use_identifier_lookup_to_result_cache;
if (scope.join_use_nulls)
{
/** Expression in PREWHERE with JOIN should not be modified by join_use_nulls.
* Example: SELECT * FROM t1 JOIN t2 USING (id) PREWHERE b = 1
* Column `a` should be resolved from table and should not change its type to Nullable.
* More complicated example when column is somewhere inside an expression:
* SELECT a + 1 as b FROM t1 JOIN t2 USING (id) PREWHERE b = 1
* expression `a + 1 as b` in projection and in PREWHERE should have different `a`.
*/
prewhere_node = prewhere_node->clone();
ReplaceColumnsVisitor replace_visitor(scope.nullable_join_columns, scope.context);
replace_visitor.visit(prewhere_node);
}
}
if (query_node_typed.getWhere())

1
src/Backups/BackupEntriesCollector.cpp
View File

@ -6,6 +6,7 @@
#include <Backups/DDLAdjustingForBackupVisitor.h>
#include <Databases/IDatabase.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Parsers/ASTCreateQuery.h>
#include <Parsers/formatAST.h>
#include <Storages/IStorage.h>

1
src/Backups/BackupIO_Default.h
View File

@ -1,6 +1,7 @@
#pragma once
#include <Backups/BackupIO.h>
#include <Common/Logger.h>
#include <IO/ReadSettings.h>
#include <IO/WriteSettings.h>

2
src/Backups/BackupIO_Disk.h
View File

@ -1,7 +1,9 @@
#pragma once
#include <Backups/BackupIO_Default.h>
#include <Common/Logger.h>
#include <Disks/DiskType.h>
#include <filesystem>

1
src/Backups/BackupIO_S3.h
View File

@ -4,6 +4,7 @@
#if USE_AWS_S3
#include <Backups/BackupIO_Default.h>
#include <Common/Logger.h>
#include <Disks/DiskType.h>
#include <IO/S3Common.h>
#include <Storages/StorageS3Settings.h>

9
src/Client/ClientBase.cpp
View File

@ -9,6 +9,7 @@
#include <base/safeExit.h>
#include <base/scope_guard.h>
#include <Core/Block.h>
#include <Core/BaseSettingsProgramOptions.h>
#include <Core/Protocol.h>
#include <Common/DateLUT.h>
#include <Common/MemoryTracker.h>
@ -2714,9 +2715,9 @@ private:
void ClientBase::parseAndCheckOptions(OptionsDescription & options_description, po::variables_map & options, Arguments & arguments)
{
if (allow_repeated_settings)
cmd_settings.addProgramOptionsAsMultitokens(options_description.main_description.value());
addProgramOptionsAsMultitokens(cmd_settings, options_description.main_description.value());
else
cmd_settings.addProgramOptions(options_description.main_description.value());
addProgramOptions(cmd_settings, options_description.main_description.value());
if (allow_merge_tree_settings)
{
@ -2737,9 +2738,9 @@ void ClientBase::parseAndCheckOptions(OptionsDescription & options_description,
return;
if (allow_repeated_settings)
cmd_merge_tree_settings.addProgramOptionAsMultitoken(main_options, name, setting);
addProgramOptionAsMultitoken(cmd_merge_tree_settings, main_options, name, setting);
else
cmd_merge_tree_settings.addProgramOption(main_options, name, setting);
addProgramOption(cmd_merge_tree_settings, main_options, name, setting);
};
const auto & setting_name = setting.getName();

5
src/Client/LocalConnection.cpp
View File

@ -1,13 +1,14 @@
#include "LocalConnection.h"
#include <Core/Protocol.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/executeQuery.h>
#include <Processors/Executors/CompletedPipelineExecutor.h>
#include <Processors/Executors/PullingAsyncPipelineExecutor.h>
#include <Processors/Executors/PushingPipelineExecutor.h>
#include <Processors/Executors/PushingAsyncPipelineExecutor.h>
#include <Processors/Executors/PushingPipelineExecutor.h>
#include <Storages/IStorage.h>
#include <Common/ConcurrentBoundedQueue.h>
#include <Common/CurrentThread.h>
#include <Core/Protocol.h>
namespace DB

11
src/Common/HTTPConnectionPool.cpp
View File

@ -10,14 +10,15 @@
#include <Common/MemoryTrackerSwitcher.h>
#include <Common/SipHash.h>
#include <Poco/Net/HTTPClientSession.h>
#include <Poco/Net/HTTPStream.h>
#include <Poco/Net/HTTPFixedLengthStream.h>
#include <Poco/Net/HTTPChunkedStream.h>
#include <Poco/Net/HTTPClientSession.h>
#include <Poco/Net/HTTPFixedLengthStream.h>
#include <Poco/Net/HTTPRequest.h>
#include <Poco/Net/HTTPResponse.h>
#include <Poco/Net/HTTPStream.h>
#include <Poco/Timespan.h>
#include <Poco/Net/HTTPResponse.h>
#include <Poco/Net/HTTPRequest.h>
#include <queue>
#include "config.h"

7
src/Common/IntervalKind.cpp
View File

@ -1,6 +1,8 @@
#include <Common/IntervalKind.h>
#include <Common/Exception.h>
#include <base/EnumReflection.h>
namespace DB
{
@ -10,6 +12,11 @@ namespace ErrorCodes
extern const int BAD_ARGUMENTS;
}
std::string_view IntervalKind::toString() const
{
return magic_enum::enum_name(kind);
}
Int64 IntervalKind::toAvgNanoseconds() const
{
static constexpr Int64 NANOSECONDS_PER_MICROSECOND = 1000;

3
src/Common/IntervalKind.h
View File

@ -1,7 +1,6 @@
#pragma once
#include <base/types.h>
#include <base/EnumReflection.h>
namespace DB
{
@ -27,7 +26,7 @@ struct IntervalKind
IntervalKind(Kind kind_ = Kind::Second) : kind(kind_) {} /// NOLINT
operator Kind() const { return kind; } /// NOLINT
constexpr std::string_view toString() const { return magic_enum::enum_name(kind); }
std::string_view toString() const;
/// Returns number of nanoseconds in one interval.
/// For `Month`, `Quarter` and `Year` the function returns an average number of nanoseconds.

3
src/Common/Scheduler/IResourceManager.h
View File

@ -12,6 +12,9 @@
namespace DB
{
class ISchedulerNode;
using SchedulerNodePtr = std::shared_ptr<ISchedulerNode>;
/*
* Instance of derived class holds everything required for resource consumption,
* including resources currently registered at `SchedulerRoot`. This is required to avoid

25
src/Common/Scheduler/ResouceLink.cpp Normal file
View File

@ -0,0 +1,25 @@
#include <Common/Scheduler/ISchedulerQueue.h>
#include <Common/Scheduler/ResourceLink.h>
#include <Common/Scheduler/ResourceRequest.h>
namespace DB
{
void ResourceLink::adjust(ResourceCost estimated_cost, ResourceCost real_cost) const
{
if (queue)
queue->adjustBudget(estimated_cost, real_cost);
}
void ResourceLink::consumed(ResourceCost cost) const
{
if (queue)
queue->consumeBudget(cost);
}
void ResourceLink::accumulate(DB::ResourceCost cost) const
{
if (queue)
queue->accumulateBudget(cost);
}
}

3
src/Common/Scheduler/ResourceGuard.h
View File

@ -2,9 +2,10 @@
#include <base/types.h>
#include <Common/Scheduler/ISchedulerConstraint.h>
#include <Common/Scheduler/ISchedulerQueue.h>
#include <Common/Scheduler/ResourceRequest.h>
#include <Common/Scheduler/ResourceLink.h>
#include <Common/Scheduler/ISchedulerConstraint.h>
#include <condition_variable>
#include <mutex>

24
src/Common/Scheduler/ResourceLink.h
View File

@ -2,12 +2,10 @@
#include <base/types.h>
#include <Common/Scheduler/ResourceRequest.h>
#include <Common/Scheduler/ISchedulerQueue.h>
namespace DB
{
class ISchedulerQueue;
using ResourceCost = Int64;
/*
* Everything required for resource consumption. Connection to a specific resource queue.
@ -17,23 +15,11 @@ struct ResourceLink
ISchedulerQueue * queue = nullptr;
bool operator==(const ResourceLink &) const = default;
void adjust(ResourceCost estimated_cost, ResourceCost real_cost) const
{
if (queue)
queue->adjustBudget(estimated_cost, real_cost);
}
void adjust(ResourceCost estimated_cost, ResourceCost real_cost) const;
void consumed(ResourceCost cost) const
{
if (queue)
queue->consumeBudget(cost);
}
void consumed(ResourceCost cost) const;
void accumulate(ResourceCost cost) const
{
if (queue)
queue->accumulateBudget(cost);
}
void accumulate(ResourceCost cost) const;
};
}

4
src/Common/SymbolIndex.cpp
View File

@ -2,6 +2,8 @@
#include <Common/SymbolIndex.h>
#include <Common/MemorySanitizer.h>
#include <base/hex.h>
#include <base/sort.h>
#include <algorithm>
#include <optional>
@ -11,8 +13,6 @@
#include <filesystem>
#include <base/sort.h>
/**
ELF object can contain three different places with symbol names and addresses:

6
src/Common/ThreadStatus.h
View File

@ -1,12 +1,12 @@
#pragma once
#include <Core/SettingsEnums.h>
#include <Interpreters/Context_fwd.h>
#include <Core/LogsLevel.h>
#include <IO/Progress.h>
#include <Interpreters/Context_fwd.h>
#include <base/StringRef.h>
#include <Common/MemoryTracker.h>
#include <Common/ProfileEvents.h>
#include <Common/Stopwatch.h>
#include <base/StringRef.h>
#include <boost/noncopyable.hpp>

7
src/Common/formatIPv6.h
View File

@ -1,13 +1,12 @@
#pragma once
#include <base/types.h>
#include <cstring>
#include <algorithm>
#include <cstring>
#include <type_traits>
#include <utility>
#include <base/range.h>
#include <base/unaligned.h>
#include <base/hex.h>
#include <base/types.h>
#include <base/unaligned.h>
#include <Common/StringUtils/StringUtils.h>
constexpr size_t IPV4_BINARY_LENGTH = 4;

2
src/Common/tests/gtest_proxy_configuration_resolver_provider.cpp
View File

@ -4,6 +4,8 @@
#include <Common/tests/gtest_global_context.h>
#include <Common/tests/gtest_helper_functions.h>
#include <Poco/Util/MapConfiguration.h>
using ConfigurationPtr = Poco::AutoPtr<Poco::Util::AbstractConfiguration>;
class ProxyConfigurationResolverProviderTests : public ::testing::Test

8
src/Compression/CompressionFactory.cpp
View File

@ -7,6 +7,8 @@
#include <Poco/String.h>
#include <IO/ReadBuffer.h>
#include <Parsers/queryToString.h>
#include <Parsers/parseQuery.h>
#include <Parsers/ExpressionElementParsers.h>
#include <Compression/CompressionCodecMultiple.h>
#include <Compression/CompressionCodecNone.h>
#include <IO/WriteHelpers.h>
@ -44,6 +46,12 @@ CompressionCodecPtr CompressionCodecFactory::get(const String & family_name, std
}
}
CompressionCodecPtr CompressionCodecFactory::get(const String & compression_codec) const
{
ParserCodec codec_parser;
auto ast = parseQuery(codec_parser, "(" + Poco::toUpper(compression_codec) + ")", 0, DBMS_DEFAULT_MAX_PARSER_DEPTH, DBMS_DEFAULT_MAX_PARSER_BACKTRACKS);
return CompressionCodecFactory::instance().get(ast, nullptr);
}
CompressionCodecPtr CompressionCodecFactory::get(
const ASTPtr & ast, const IDataType * column_type, CompressionCodecPtr current_default, bool only_generic) const

3
src/Compression/CompressionFactory.h
View File

@ -68,6 +68,9 @@ public:
/// For backward compatibility with config settings
CompressionCodecPtr get(const String & family_name, std::optional<int> level) const;
/// Get codec by name with optional params. Example: LZ4, ZSTD(3)
CompressionCodecPtr get(const String & compression_codec) const;
/// Register codec with parameters and column type
void registerCompressionCodecWithType(const String & family_name, std::optional<uint8_t> byte_code, CreatorWithType creator);
/// Register codec with parameters

64
src/Core/BaseSettings.h
View File

@ -7,7 +7,6 @@
#include <base/range.h>
#include <boost/blank.hpp>
#include <unordered_map>
#include <boost/program_options/options_description.hpp>
namespace boost::program_options
@ -129,18 +128,6 @@ public:
std::conditional_t<Traits::allow_custom_settings, const CustomSettingMap::mapped_type*, boost::blank> custom_setting;
};
/// Adds program options to set the settings from a command line.
/// (Don't forget to call notify() on the `variables_map` after parsing it!)
void addProgramOptions(boost::program_options::options_description & options);
/// Adds program options as to set the settings from a command line.
/// Allows to set one setting multiple times, the last value will be used.
/// (Don't forget to call notify() on the `variables_map` after parsing it!)
void addProgramOptionsAsMultitokens(boost::program_options::options_description & options);
void addProgramOption(boost::program_options::options_description & options, std::string_view name, const SettingFieldRef & field);
void addProgramOptionAsMultitoken(boost::program_options::options_description & options, std::string_view name, const SettingFieldRef & field);
enum SkipFlags
{
SKIP_NONE = 0,
@ -561,57 +548,6 @@ String BaseSettings<TTraits>::toString() const
return out.str();
}
template <typename TTraits>
void BaseSettings<TTraits>::addProgramOptions(boost::program_options::options_description & options)
{
const auto & settings_to_aliases = TTraits::settingsToAliases();
for (const auto & field : all())
{
std::string_view name = field.getName();
addProgramOption(options, name, field);
if (auto it = settings_to_aliases.find(name); it != settings_to_aliases.end())
{
for (const auto alias : it->second)
addProgramOption(options, alias, field);
}
}
}
template <typename TTraits>
void BaseSettings<TTraits>::addProgramOptionsAsMultitokens(boost::program_options::options_description & options)
{
const auto & settings_to_aliases = TTraits::settingsToAliases();
for (const auto & field : all())
{
std::string_view name = field.getName();
addProgramOptionAsMultitoken(options, name, field);
if (auto it = settings_to_aliases.find(name); it != settings_to_aliases.end())
{
for (const auto alias : it->second)
addProgramOptionAsMultitoken(options, alias, field);
}
}
}
template <typename TTraits>
void BaseSettings<TTraits>::addProgramOption(boost::program_options::options_description & options, std::string_view name, const SettingFieldRef & field)
{
auto on_program_option = boost::function1<void, const std::string &>([this, name](const std::string & value) { set(name, value); });
options.add(boost::shared_ptr<boost::program_options::option_description>(new boost::program_options::option_description(
name.data(), boost::program_options::value<std::string>()->composing()->notifier(on_program_option), field.getDescription())));
}
template <typename TTraits>
void BaseSettings<TTraits>::addProgramOptionAsMultitoken(boost::program_options::options_description & options, std::string_view name, const SettingFieldRef & field)
{
auto on_program_option = boost::function1<void, const Strings &>([this, name](const Strings & values) { set(name, values.back()); });
options.add(boost::shared_ptr<boost::program_options::option_description>(new boost::program_options::option_description(
name.data(), boost::program_options::value<Strings>()->multitoken()->composing()->notifier(on_program_option), field.getDescription())));
}
template <typename TTraits>
bool operator==(const BaseSettings<TTraits> & left, const BaseSettings<TTraits> & right)
{

60
src/Core/BaseSettingsProgramOptions.h Normal file
View File

@ -0,0 +1,60 @@
#pragma once
#include <Core/Settings.h>
#include <Core/Types_fwd.h>
#include <boost/program_options.hpp>
namespace DB
{
template <typename T>
void addProgramOptionAsMultitoken(T &cmd_settings, boost::program_options::options_description & options, std::string_view name, const typename T::SettingFieldRef & field)
{
auto on_program_option = boost::function1<void, const Strings &>([&cmd_settings, name](const Strings & values) { cmd_settings.set(name, values.back()); });
options.add(boost::shared_ptr<boost::program_options::option_description>(new boost::program_options::option_description(
name.data(), boost::program_options::value<Strings>()->multitoken()->composing()->notifier(on_program_option), field.getDescription())));
}
template <typename T>
void addProgramOptionsAsMultitokens(T &cmd_settings, boost::program_options::options_description & options)
{
const auto & settings_to_aliases = T::Traits::settingsToAliases();
for (const auto & field : cmd_settings.all())
{
std::string_view name = field.getName();
addProgramOptionAsMultitoken(cmd_settings, options, name, field);
if (auto it = settings_to_aliases.find(name); it != settings_to_aliases.end())
for (const auto alias : it->second)
addProgramOptionAsMultitoken(cmd_settings, options, alias, field);
}
}
/// Adds program options to set the settings from a command line.
/// (Don't forget to call notify() on the `variables_map` after parsing it!)
template <typename T>
void addProgramOption(T &cmd_settings, boost::program_options::options_description & options, std::string_view name, const typename T::SettingFieldRef & field)
{
auto on_program_option = boost::function1<void, const std::string &>([&cmd_settings, name](const std::string & value) { cmd_settings.set(name, value); });
options.add(boost::shared_ptr<boost::program_options::option_description>(new boost::program_options::option_description(
name.data(), boost::program_options::value<std::string>()->composing()->notifier(on_program_option), field.getDescription())));
}
template <typename T>
void addProgramOptions(T &cmd_settings, boost::program_options::options_description & options)
{
const auto & settings_to_aliases = T::Traits::settingsToAliases();
for (const auto & field : cmd_settings.all())
{
std::string_view name = field.getName();
addProgramOption(cmd_settings, options, name, field);
if (auto it = settings_to_aliases.find(name); it != settings_to_aliases.end())
for (const auto alias : it->second)
addProgramOption(cmd_settings, options, alias, field);
}
}
}

16
src/Core/LogsLevel.h Normal file
View File

@ -0,0 +1,16 @@
#pragma once
namespace DB
{
enum class LogsLevel
{
none = 0, /// Disable
fatal,
error,
warning,
information,
debug,
trace,
test,
};
}

4
src/Core/Settings.h
View File

@ -114,6 +114,7 @@ class IColumn;
M(Bool, enable_s3_requests_logging, false, "Enable very explicit logging of S3 requests. Makes sense for debug only.", 0) \
M(String, s3queue_default_zookeeper_path, "/clickhouse/s3queue/", "Default zookeeper path prefix for S3Queue engine", 0) \
M(Bool, s3queue_enable_logging_to_s3queue_log, false, "Enable writing to system.s3queue_log. The value can be overwritten per table with table settings", 0) \
M(Bool, s3queue_allow_experimental_sharded_mode, false, "Enable experimental sharded mode of S3Queue table engine. It is experimental because it will be rewritten", 0) \
M(UInt64, hdfs_replication, 0, "The actual number of replications can be specified when the hdfs file is created.", 0) \
M(Bool, hdfs_truncate_on_insert, false, "Enables or disables truncate before insert in s3 engine tables", 0) \
M(Bool, hdfs_create_new_file_on_insert, false, "Enables or disables creating a new file on each insert in hdfs engine tables", 0) \
@ -372,6 +373,7 @@ class IColumn;
M(Bool, opentelemetry_trace_processors, false, "Collect OpenTelemetry spans for processors.", 0) \
M(Bool, prefer_column_name_to_alias, false, "Prefer using column names instead of aliases if possible.", 0) \
M(Bool, allow_experimental_analyzer, false, "Allow experimental analyzer", 0) \
M(Bool, analyzer_compatibility_join_using_top_level_identifier, false, "Force to resolve identifier in JOIN USING from projection (for example, in `SELECT a + 1 AS b FROM t1 JOIN t2 USING (b)` join will be performed by `t1.a + 1 = t2.b`, rather then `t1.b = t2.b`).", 0) \
M(Bool, prefer_global_in_and_join, false, "If enabled, all IN/JOIN operators will be rewritten as GLOBAL IN/JOIN. It's useful when the to-be-joined tables are only available on the initiator and we need to always scatter their data on-the-fly during distributed processing with the GLOBAL keyword. It's also useful to reduce the need to access the external sources joining external tables.", 0) \
M(Bool, enable_vertical_final, true, "If enable, remove duplicated rows during FINAL by marking rows as deleted and filtering them later instead of merging rows", 0) \
\
@ -450,7 +452,7 @@ class IColumn;
\
M(Bool, compatibility_ignore_collation_in_create_table, true, "Compatibility ignore collation in create table", 0) \
\
M(String, temporary_files_codec, "LZ4", "Set compression codec for temporary files (sort and join on disk). I.e. LZ4, NONE.", 0) \
M(String, temporary_files_codec, "LZ4", "Set compression codec for temporary files produced by (JOINs, external GROUP BY, external ORDER BY). I.e. LZ4, NONE.", 0) \
\
M(UInt64, max_rows_to_transfer, 0, "Maximum size (in rows) of the transmitted external table obtained when the GLOBAL IN/JOIN section is executed.", 0) \
M(UInt64, max_bytes_to_transfer, 0, "Maximum size (in uncompressed bytes) of the transmitted external table obtained when the GLOBAL IN/JOIN section is executed.", 0) \

4
src/Core/SettingsChangesHistory.h
View File

@ -97,9 +97,11 @@ static std::map<ClickHouseVersion, SettingsChangesHistory::SettingsChanges> sett
{"function_locate_has_mysql_compatible_argument_order", false, true, "Increase compatibility with MySQL's locate function."},
{"filesystem_cache_reserve_space_wait_lock_timeout_milliseconds", 1000, 1000, "Wait time to lock cache for sapce reservation in filesystem cache"},
{"max_parser_backtracks", 0, 1000000, "Limiting the complexity of parsing"},
{"analyzer_compatibility_join_using_top_level_identifier", false, false, "Force to resolve identifier in JOIN USING from projection"},
{"keeper_max_retries", 10, 10, "Max retries for general keeper operations"},
{"keeper_retry_initial_backoff_ms", 100, 100, "Initial backoff timeout for general keeper operations"},
{"keeper_retry_max_backoff_ms", 5000, 5000, "Max backoff timeout for general keeper operations"},
{"s3queue_allow_experimental_sharded_mode", false, false, "Enable experimental sharded mode of S3Queue table engine. It is experimental because it will be rewritten"},
}},
{"24.2", {{"allow_suspicious_variant_types", true, false, "Don't allow creating Variant type with suspicious variants by default"},
{"validate_experimental_and_suspicious_types_inside_nested_types", false, true, "Validate usage of experimental and suspicious types inside nested types"},
@ -109,7 +111,7 @@ static std::map<ClickHouseVersion, SettingsChangesHistory::SettingsChanges> sett
{"query_plan_optimize_prewhere", true, true, "Allow to push down filter to PREWHERE expression for supported storages"},
{"async_insert_max_data_size", 1000000, 10485760, "The previous value appeared to be too small."},
{"async_insert_poll_timeout_ms", 10, 10, "Timeout in milliseconds for polling data from asynchronous insert queue"},
{"async_insert_use_adaptive_busy_timeout", true, true, "Use adaptive asynchronous insert timeout"},
{"async_insert_use_adaptive_busy_timeout", false, true, "Use adaptive asynchronous insert timeout"},
{"async_insert_busy_timeout_min_ms", 50, 50, "The minimum value of the asynchronous insert timeout in milliseconds; it also serves as the initial value, which may be increased later by the adaptive algorithm"},
{"async_insert_busy_timeout_max_ms", 200, 200, "The minimum value of the asynchronous insert timeout in milliseconds; async_insert_busy_timeout_ms is aliased to async_insert_busy_timeout_max_ms"},
{"async_insert_busy_timeout_increase_rate", 0.2, 0.2, "The exponential growth rate at which the adaptive asynchronous insert timeout increases"},

19
src/Core/SettingsEnums.h
View File

@ -1,12 +1,13 @@
#pragma once
#include <Core/SettingsFields.h>
#include <Core/Joins.h>
#include <QueryPipeline/SizeLimits.h>
#include <Core/LogsLevel.h>
#include <Core/SettingsFields.h>
#include <Formats/FormatSettings.h>
#include <IO/ReadSettings.h>
#include <Common/ShellCommandSettings.h>
#include <Parsers/ASTSQLSecurity.h>
#include <QueryPipeline/SizeLimits.h>
#include <Common/ShellCommandSettings.h>
namespace DB
@ -90,18 +91,6 @@ DECLARE_SETTING_ENUM_WITH_RENAME(IntervalOutputFormat, FormatSettings::IntervalO
DECLARE_SETTING_ENUM_WITH_RENAME(ParquetVersion, FormatSettings::ParquetVersion)
enum class LogsLevel
{
none = 0, /// Disable
fatal,
error,
warning,
information,
debug,
trace,
test,
};
DECLARE_SETTING_ENUM(LogsLevel)

1
src/Daemon/SentryWriter.cpp
View File

@ -12,6 +12,7 @@
#include <Common/StackTrace.h>
#include <Common/getNumberOfPhysicalCPUCores.h>
#include <Core/ServerUUID.h>
#include <IO/WriteHelpers.h>
#include "config.h"
#include <Common/config_version.h>

5
src/DataTypes/DataTypeDomainBool.cpp
View File

@ -1,6 +1,7 @@
#include <DataTypes/Serializations/SerializationBool.h>
#include <DataTypes/DataTypeFactory.h>
#include <DataTypes/DataTypeCustom.h>
#include <DataTypes/DataTypeFactory.h>
#include <DataTypes/IDataType.h>
#include <DataTypes/Serializations/SerializationBool.h>
namespace DB
{

1
src/DataTypes/DataTypeFactory.h
View File

@ -1,6 +1,5 @@
#pragma once
#include <DataTypes/IDataType.h>
#include <Parsers/IAST_fwd.h>
#include <Common/IFactoryWithAliases.h>
#include <DataTypes/DataTypeCustom.h>

2
src/DataTypes/DataTypeTuple.cpp
View File

@ -346,7 +346,7 @@ SerializationPtr DataTypeTuple::getSerialization(const SerializationInfo & info)
return std::make_shared<SerializationTuple>(std::move(serializations), have_explicit_names);
}
MutableSerializationInfoPtr DataTypeTuple::createSerializationInfo(const SerializationInfo::Settings & settings) const
MutableSerializationInfoPtr DataTypeTuple::createSerializationInfo(const SerializationInfoSettings & settings) const
{
MutableSerializationInfos infos;
infos.reserve(elems.size());

2
src/DataTypes/DataTypeTuple.h
View File

@ -58,7 +58,7 @@ public:
SerializationPtr doGetDefaultSerialization() const override;
SerializationPtr getSerialization(const SerializationInfo & info) const override;
MutableSerializationInfoPtr createSerializationInfo(const SerializationInfo::Settings & settings) const override;
MutableSerializationInfoPtr createSerializationInfo(const SerializationInfoSettings & settings) const override;
SerializationInfoPtr getSerializationInfo(const IColumn & column) const override;
const DataTypePtr & getElement(size_t i) const { return elems[i]; }

2
src/DataTypes/IDataType.cpp
View File

@ -202,7 +202,7 @@ void IDataType::setCustomization(DataTypeCustomDescPtr custom_desc_) const
custom_serialization = std::move(custom_desc_->serialization);
}
MutableSerializationInfoPtr IDataType::createSerializationInfo(const SerializationInfo::Settings & settings) const
MutableSerializationInfoPtr IDataType::createSerializationInfo(const SerializationInfoSettings & settings) const
{
return std::make_shared<SerializationInfo>(ISerialization::Kind::DEFAULT, settings);
}

9
src/DataTypes/IDataType.h
View File

@ -7,8 +7,6 @@
#include <Common/COW.h>
#include <DataTypes/DataTypeCustom.h>
#include <DataTypes/Serializations/ISerialization.h>
#include <DataTypes/Serializations/SerializationInfo.h>
namespace DB
{
@ -38,6 +36,11 @@ struct DataTypeWithConstInfo
using DataTypesWithConstInfo = std::vector<DataTypeWithConstInfo>;
class SerializationInfo;
using SerializationInfoPtr = std::shared_ptr<const SerializationInfo>;
using MutableSerializationInfoPtr = std::shared_ptr<SerializationInfo>;
struct SerializationInfoSettings;
/** Properties of data type.
*
* Contains methods for getting serialization instances.
@ -117,7 +120,7 @@ public:
Names getSubcolumnNames() const;
virtual MutableSerializationInfoPtr createSerializationInfo(const SerializationInfo::Settings & settings) const;
virtual MutableSerializationInfoPtr createSerializationInfo(const SerializationInfoSettings & settings) const;
virtual SerializationInfoPtr getSerializationInfo(const IColumn & column) const;
/// TODO: support more types.

1
src/DataTypes/ObjectUtils.h
View File

@ -13,6 +13,7 @@ namespace DB
struct StorageSnapshot;
using StorageSnapshotPtr = std::shared_ptr<StorageSnapshot>;
class ColumnsDescription;
/// Returns number of dimensions in Array type. 0 if type is not array.
size_t getNumberOfDimensions(const IDataType & type);

26
src/DataTypes/Serializations/SerializationInfo.h
View File

@ -2,6 +2,8 @@
#include <Core/Types_fwd.h>
#include <DataTypes/Serializations/ISerialization.h>
#include <DataTypes/Serializations/SerializationInfoSettings.h>
#include <Poco/JSON/Object.h>
@ -28,6 +30,8 @@ constexpr auto SERIALIZATION_INFO_VERSION = 0;
class SerializationInfo
{
public:
using Settings = SerializationInfoSettings;
struct Data
{
size_t num_rows = 0;
@ -38,16 +42,8 @@ public:
void addDefaults(size_t length);
};
struct Settings
{
const double ratio_of_defaults_for_sparse = 1.0;
const bool choose_kind = false;
bool isAlwaysDefault() const { return ratio_of_defaults_for_sparse >= 1.0; }
};
SerializationInfo(ISerialization::Kind kind_, const Settings & settings_);
SerializationInfo(ISerialization::Kind kind_, const Settings & settings_, const Data & data_);
SerializationInfo(ISerialization::Kind kind_, const SerializationInfoSettings & settings_);
SerializationInfo(ISerialization::Kind kind_, const SerializationInfoSettings & settings_, const Data & data_);
virtual ~SerializationInfo() = default;
@ -64,7 +60,7 @@ public:
virtual std::shared_ptr<SerializationInfo> createWithType(
const IDataType & old_type,
const IDataType & new_type,
const Settings & new_settings) const;
const SerializationInfoSettings & new_settings) const;
virtual void serialializeKindBinary(WriteBuffer & out) const;
virtual void deserializeFromKindsBinary(ReadBuffer & in);
@ -73,14 +69,14 @@ public:
virtual void fromJSON(const Poco::JSON::Object & object);
void setKind(ISerialization::Kind kind_) { kind = kind_; }
const Settings & getSettings() const { return settings; }
const SerializationInfoSettings & getSettings() const { return settings; }
const Data & getData() const { return data; }
ISerialization::Kind getKind() const { return kind; }
static ISerialization::Kind chooseKind(const Data & data, const Settings & settings);
static ISerialization::Kind chooseKind(const Data & data, const SerializationInfoSettings & settings);
protected:
const Settings settings;
const SerializationInfoSettings settings;
ISerialization::Kind kind;
Data data;
@ -96,7 +92,7 @@ using MutableSerializationInfos = std::vector<MutableSerializationInfoPtr>;
class SerializationInfoByName : public std::map<String, MutableSerializationInfoPtr>
{
public:
using Settings = SerializationInfo::Settings;
using Settings = SerializationInfoSettings;
SerializationInfoByName() = default;
SerializationInfoByName(const NamesAndTypesList & columns, const Settings & settings);

14
src/DataTypes/Serializations/SerializationInfoSettings.h Normal file
View File

@ -0,0 +1,14 @@
#pragma once
namespace DB
{
struct SerializationInfoSettings
{
const double ratio_of_defaults_for_sparse = 1.0;
const bool choose_kind = false;
bool isAlwaysDefault() const { return ratio_of_defaults_for_sparse >= 1.0; }
};
}

1
src/DataTypes/tests/gtest_DataType_deserializeAsText.cpp
View File

@ -1,6 +1,7 @@
#include <Columns/IColumn.h>
#include <Core/Field.h>
#include <DataTypes/DataTypeFactory.h>
#include <DataTypes/IDataType.h>
#include <Formats/FormatSettings.h>
#include <IO/ReadBuffer.h>

1
src/Databases/DatabaseAtomic.cpp
View File

@ -11,6 +11,7 @@
#include <Common/filesystemHelpers.h>
#include <Storages/StorageMaterializedView.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/ExternalDictionariesLoader.h>
#include <filesystem>
#include <Interpreters/DDLTask.h>

1
src/Databases/DatabaseMemory.cpp
View File

@ -6,6 +6,7 @@
#include <Databases/DDLDependencyVisitor.h>
#include <Databases/DDLLoadingDependencyVisitor.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Parsers/ASTCreateQuery.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/formatAST.h>

3
src/Databases/DatabaseOnDisk.cpp
View File

@ -5,8 +5,9 @@
#include <IO/WriteBufferFromFile.h>
#include <IO/WriteHelpers.h>
#include <Interpreters/Context.h>
#include <Interpreters/InterpreterCreateQuery.h>
#include <Interpreters/ApplyWithSubqueryVisitor.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/InterpreterCreateQuery.h>
#include <Parsers/ASTCreateQuery.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/ParserCreateQuery.h>

1
src/Databases/DatabaseReplicated.cpp
View File

@ -20,6 +20,7 @@
#include <Databases/TablesDependencyGraph.h>
#include <Interpreters/Cluster.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/DDLTask.h>
#include <Interpreters/evaluateConstantExpression.h>
#include <Interpreters/executeDDLQueryOnCluster.h>

1
src/Databases/DatabasesCommon.cpp
View File

@ -1,6 +1,7 @@
#include <Databases/DatabasesCommon.h>
#include <Interpreters/InterpreterCreateQuery.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Parsers/ASTCreateQuery.h>
#include <Parsers/ASTSelectWithUnionQuery.h>
#include <Parsers/ParserCreateQuery.h>

9
src/Databases/IDatabase.cpp
View File

@ -1,11 +1,12 @@
#include <memory>
#include <Databases/IDatabase.h>
#include <Storages/IStorage.h>
#include <Parsers/ASTCreateQuery.h>
#include <Common/quoteString.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Common/NamePrompter.h>
#include <Interpreters/TableNameHints.h>
#include <Parsers/ASTCreateQuery.h>
#include <Storages/IStorage.h>
#include <Common/CurrentMetrics.h>
#include <Common/NamePrompter.h>
#include <Common/quoteString.h>
namespace CurrentMetrics

1
src/Disks/ObjectStorages/DiskObjectStorage.cpp
View File

@ -9,6 +9,7 @@
#include <Common/logger_useful.h>
#include <Common/filesystemHelpers.h>
#include <Common/CurrentMetrics.h>
#include <Common/Scheduler/IResourceManager.h>
#include <Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.h>
#include <Disks/ObjectStorages/DiskObjectStorageTransaction.h>
#include <Disks/FakeDiskTransaction.h>

1
src/Formats/FormatFactory.h
View File

@ -23,6 +23,7 @@ namespace DB
class Block;
struct Settings;
struct FormatFactorySettings;
struct ReadSettings;
class ReadBuffer;
class WriteBuffer;

1
src/Formats/FormatSchemaInfo.h
View File

@ -8,6 +8,7 @@
namespace DB
{
class Context;
class Block;
/// Extracts information about where the format schema file is from passed context and keep it.
class FormatSchemaInfo

1
src/Formats/ReadSchemaUtils.cpp
View File

@ -3,6 +3,7 @@
#include <IO/PeekableReadBuffer.h>
#include <IO/WithFileSize.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Processors/Formats/ISchemaReader.h>
#include <Storages/IStorage.h>
#include <Common/assert_cast.h>

6
src/Formats/SchemaInferenceUtils.h
View File

@ -3,9 +3,15 @@
#include <DataTypes/IDataType.h>
#include <IO/ReadBuffer.h>
#include <vector>
namespace DB
{
class Block;
class NamesAndTypesList;
using NamesAndTypesLists = std::vector<NamesAndTypesList>;
/// Struct with some additional information about inferred types for JSON formats.
struct JSONInferenceInfo
{

1
src/Functions/CMakeLists.txt
View File

@ -37,6 +37,7 @@ list (APPEND PUBLIC_LIBS
clickhouse_dictionaries_embedded
clickhouse_parsers
ch_contrib::consistent_hashing
common
dbms
ch_contrib::metrohash
ch_contrib::murmurhash

1
src/Functions/DateTimeTransforms.h
View File

@ -11,6 +11,7 @@
#include <Columns/ColumnsNumber.h>
#include <Columns/ColumnVector.h>
#include <Columns/ColumnDecimal.h>
#include <Formats/FormatSettings.h>
#include <Functions/FunctionHelpers.h>
#include <Functions/IFunction.h>
#include <Functions/extractTimeZoneFromFunctionArguments.h>

7
src/Functions/FunctionJoinGet.cpp
View File

@ -1,12 +1,13 @@
#include <Columns/ColumnString.h>
#include <Core/Block.h>
#include <Functions/FunctionFactory.h>
#include <Functions/FunctionHelpers.h>
#include <Interpreters/Context.h>
#include <Interpreters/HashJoin.h>
#include <Functions/IFunction.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/HashJoin.h>
#include <Storages/StorageJoin.h>
#include <Storages/TableLockHolder.h>
#include <Core/Block.h>
namespace DB

2
src/Functions/FunctionsStringHashFixedString.cpp
View File

@ -274,7 +274,7 @@ public:
const typename ColumnIPv6::Container & data = col_from_ip->getData();
const auto size = col_from_ip->size();
auto & chars_to = col_to->getChars();
const auto length = IPV6_BINARY_LENGTH;
const auto length = sizeof(IPv6::UnderlyingType);
chars_to.resize(size * Impl::length);
for (size_t i = 0; i < size; ++i)
{

3
src/Functions/appendTrailingCharIfAbsent.cpp
View File

@ -1,9 +1,10 @@
#include <Columns/ColumnString.h>
#include <Common/assert_cast.h>
#include <DataTypes/DataTypeString.h>
#include <Functions/FunctionFactory.h>
#include <Functions/FunctionHelpers.h>
#include <Functions/IFunction.h>
#include <base/range.h>
#include <Common/assert_cast.h>
namespace DB

100
src/Functions/array/arrayDistance.cpp
View File

@ -18,11 +18,11 @@ namespace DB
{
namespace ErrorCodes
{
extern const int ARGUMENT_OUT_OF_BOUND;
extern const int ILLEGAL_COLUMN;
extern const int ILLEGAL_TYPE_OF_ARGUMENT;
extern const int LOGICAL_ERROR;
extern const int SIZES_OF_ARRAYS_DONT_MATCH;
extern const int ARGUMENT_OUT_OF_BOUND;
}
struct L1Distance
@ -357,7 +357,7 @@ public:
throw Exception(
ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
"Arguments of function {} has nested type {}. "
"Support: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
"Supported types: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
getName(),
common_type->getName());
}
@ -379,17 +379,17 @@ public:
}
#define SUPPORTED_TYPES(action) \
action(UInt8) \
action(UInt16) \
action(UInt32) \
action(UInt64) \
action(Int8) \
action(Int16) \
action(Int32) \
action(Int64) \
action(Float32) \
action(Float64)
#define SUPPORTED_TYPES(ACTION) \
ACTION(UInt8) \
ACTION(UInt16) \
ACTION(UInt32) \
ACTION(UInt64) \
ACTION(Int8) \
ACTION(Int16) \
ACTION(Int32) \
ACTION(Int64) \
ACTION(Float32) \
ACTION(Float64)
private:
@ -398,12 +398,11 @@ private:
{
DataTypePtr type_x = typeid_cast<const DataTypeArray *>(arguments[0].type.get())->getNestedType();
/// Dynamic disaptch based on the 1st argument type
switch (type_x->getTypeId())
{
#define ON_TYPE(type) \
case TypeIndex::type: \
return executeWithFirstType<ResultType, type>(arguments, input_rows_count); \
return executeWithResultTypeAndLeftType<ResultType, type>(arguments, input_rows_count); \
break;
SUPPORTED_TYPES(ON_TYPE)
@ -413,23 +412,22 @@ private:
throw Exception(
ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
"Arguments of function {} has nested type {}. "
"Support: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
"Supported types: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
getName(),
type_x->getName());
}
}
template <typename ResultType, typename FirstArgType>
ColumnPtr executeWithFirstType(const ColumnsWithTypeAndName & arguments, size_t input_rows_count) const
template <typename ResultType, typename LeftType>
ColumnPtr executeWithResultTypeAndLeftType(const ColumnsWithTypeAndName & arguments, size_t input_rows_count) const
{
DataTypePtr type_y = typeid_cast<const DataTypeArray *>(arguments[1].type.get())->getNestedType();
/// Dynamic disaptch based on the 2nd argument type
switch (type_y->getTypeId())
{
#define ON_TYPE(type) \
case TypeIndex::type: \
return executeWithTypes<ResultType, FirstArgType, type>(arguments[0].column, arguments[1].column, input_rows_count, arguments); \
return executeWithResultTypeAndLeftTypeAndRightType<ResultType, LeftType, type>(arguments[0].column, arguments[1].column, input_rows_count, arguments); \
break;
SUPPORTED_TYPES(ON_TYPE)
@ -439,59 +437,43 @@ private:
throw Exception(
ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
"Arguments of function {} has nested type {}. "
"Support: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
"Supported types: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
getName(),
type_y->getName());
}
}
template <typename ResultType, typename FirstArgType, typename SecondArgType>
ColumnPtr executeWithTypes(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count, const ColumnsWithTypeAndName & arguments) const
template <typename ResultType, typename LeftType, typename RightType>
ColumnPtr executeWithResultTypeAndLeftTypeAndRightType(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count, const ColumnsWithTypeAndName & arguments) const
{
if (typeid_cast<const ColumnConst *>(col_x.get()))
{
return executeWithTypesFirstArgConst<ResultType, FirstArgType, SecondArgType>(col_x, col_y, input_rows_count, arguments);
return executeWithLeftArgConst<ResultType, LeftType, RightType>(col_x, col_y, input_rows_count, arguments);
}
else if (typeid_cast<const ColumnConst *>(col_y.get()))
{
return executeWithTypesFirstArgConst<ResultType, SecondArgType, FirstArgType>(col_y, col_x, input_rows_count, arguments);
return executeWithLeftArgConst<ResultType, RightType, LeftType>(col_y, col_x, input_rows_count, arguments);
}
col_x = col_x->convertToFullColumnIfConst();
col_y = col_y->convertToFullColumnIfConst();
const auto & array_x = *assert_cast<const ColumnArray *>(col_x.get());
const auto & array_y = *assert_cast<const ColumnArray *>(col_y.get());
const auto & data_x = typeid_cast<const ColumnVector<FirstArgType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<SecondArgType> &>(array_y.getData()).getData();
const auto & data_x = typeid_cast<const ColumnVector<LeftType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<RightType> &>(array_y.getData()).getData();
const auto & offsets_x = array_x.getOffsets();
const auto & offsets_y = array_y.getOffsets();
/// Check that arrays in both columns are the sames size
for (size_t row = 0; row < offsets_x.size(); ++row)
{
if (offsets_x[row] != offsets_y[row]) [[unlikely]]
{
ColumnArray::Offset prev_offset = row > 0 ? offsets_x[row] : 0;
throw Exception(
ErrorCodes::SIZES_OF_ARRAYS_DONT_MATCH,
"Arguments of function {} have different array sizes: {} and {}",
getName(),
offsets_x[row] - prev_offset,
offsets_y[row] - prev_offset);
}
}
if (!array_x.hasEqualOffsets(array_y))
throw Exception(ErrorCodes::SIZES_OF_ARRAYS_DONT_MATCH, "Array arguments for function {} must have equal sizes", getName());
const typename Kernel::ConstParams kernel_params = initConstParams(arguments);
auto result = ColumnVector<ResultType>::create(input_rows_count);
auto & result_data = result->getData();
auto col_res = ColumnVector<ResultType>::create(input_rows_count);
auto & result_data = col_res->getData();
/// Do the actual computation
ColumnArray::Offset prev = 0;
size_t row = 0;
for (auto off : offsets_x)
{
/// Process chunks in vectorized manner
@ -517,12 +499,12 @@ private:
result_data[row] = Kernel::finalize(state, kernel_params);
row++;
}
return result;
return col_res;
}
/// Special case when the 1st parameter is Const
template <typename ResultType, typename FirstArgType, typename SecondArgType>
ColumnPtr executeWithTypesFirstArgConst(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count, const ColumnsWithTypeAndName & arguments) const
template <typename ResultType, typename LeftType, typename RightType>
ColumnPtr executeWithLeftArgConst(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count, const ColumnsWithTypeAndName & arguments) const
{
col_x = assert_cast<const ColumnConst *>(col_x.get())->getDataColumnPtr();
col_y = col_y->convertToFullColumnIfConst();
@ -530,26 +512,25 @@ private:
const auto & array_x = *assert_cast<const ColumnArray *>(col_x.get());
const auto & array_y = *assert_cast<const ColumnArray *>(col_y.get());
const auto & data_x = typeid_cast<const ColumnVector<FirstArgType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<SecondArgType> &>(array_y.getData()).getData();
const auto & data_x = typeid_cast<const ColumnVector<LeftType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<RightType> &>(array_y.getData()).getData();
const auto & offsets_x = array_x.getOffsets();
const auto & offsets_y = array_y.getOffsets();
/// Check that arrays in both columns are the sames size
ColumnArray::Offset prev_offset = 0;
for (size_t row : collections::range(0, offsets_y.size()))
for (auto offset_y : offsets_y)
{
if (offsets_x[0] != offsets_y[row] - prev_offset) [[unlikely]]
if (offsets_x[0] != offset_y - prev_offset) [[unlikely]]
{
throw Exception(
ErrorCodes::SIZES_OF_ARRAYS_DONT_MATCH,
"Arguments of function {} have different array sizes: {} and {}",
getName(),
offsets_x[0],
offsets_y[row] - prev_offset);
offset_y - prev_offset);
}
prev_offset = offsets_y[row];
prev_offset = offset_y;
}
const typename Kernel::ConstParams kernel_params = initConstParams(arguments);
@ -557,7 +538,6 @@ private:
auto result = ColumnVector<ResultType>::create(input_rows_count);
auto & result_data = result->getData();
/// Do the actual computation
size_t prev = 0;
size_t row = 0;
@ -574,7 +554,7 @@ private:
/// - the two most common metrics L2 and cosine distance,
/// - the most powerful SIMD instruction set (AVX-512F).
#if USE_MULTITARGET_CODE
if constexpr (std::is_same_v<ResultType, FirstArgType> && std::is_same_v<ResultType, SecondArgType>) /// ResultType is Float32 or Float64
if constexpr (std::is_same_v<ResultType, LeftType> && std::is_same_v<ResultType, RightType>) /// ResultType is Float32 or Float64
{
if constexpr (std::is_same_v<Kernel, L2Distance>
|| std::is_same_v<Kernel, CosineDistance>)

225
src/Functions/array/arrayDotProduct.cpp
View File

@ -18,10 +18,9 @@ namespace DB
namespace ErrorCodes
{
extern const int BAD_ARGUMENTS;
extern const int ILLEGAL_COLUMN;
extern const int ILLEGAL_TYPE_OF_ARGUMENT;
extern const int LOGICAL_ERROR;
extern const int SIZES_OF_ARRAYS_DONT_MATCH;
}
@ -141,6 +140,7 @@ public:
static FunctionPtr create(ContextPtr) { return std::make_shared<FunctionArrayScalarProduct>(); }
size_t getNumberOfArguments() const override { return 2; }
bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
bool useDefaultImplementationForConstants() const override { return true; }
DataTypePtr getReturnTypeImpl(const DataTypes & arguments) const override
{
@ -163,26 +163,29 @@ public:
return Kernel::getReturnType(nested_types[0], nested_types[1]);
}
ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t /* input_rows_count */) const override
#define SUPPORTED_TYPES(ACTION) \
ACTION(UInt8) \
ACTION(UInt16) \
ACTION(UInt32) \
ACTION(UInt64) \
ACTION(Int8) \
ACTION(Int16) \
ACTION(Int32) \
ACTION(Int64) \
ACTION(Float32) \
ACTION(Float64)
ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override
{
switch (result_type->getTypeId())
{
#define SUPPORTED_TYPE(type) \
#define ON_TYPE(type) \
case TypeIndex::type: \
return executeWithResultType<type>(arguments); \
return executeWithResultType<type>(arguments, input_rows_count); \
break;
SUPPORTED_TYPE(UInt8)
SUPPORTED_TYPE(UInt16)
SUPPORTED_TYPE(UInt32)
SUPPORTED_TYPE(UInt64)
SUPPORTED_TYPE(Int8)
SUPPORTED_TYPE(Int16)
SUPPORTED_TYPE(Int32)
SUPPORTED_TYPE(Int64)
SUPPORTED_TYPE(Float32)
SUPPORTED_TYPE(Float64)
#undef SUPPORTED_TYPE
SUPPORTED_TYPES(ON_TYPE)
#undef ON_TYPE
default:
throw Exception(ErrorCodes::LOGICAL_ERROR, "Unexpected result type {}", result_type->getName());
@ -191,90 +194,150 @@ public:
private:
template <typename ResultType>
ColumnPtr executeWithResultType(const ColumnsWithTypeAndName & arguments) const
ColumnPtr executeWithResultType(const ColumnsWithTypeAndName & arguments, size_t input_rows_count) const
{
ColumnPtr res;
if (!((res = executeWithResultTypeAndLeft<ResultType, UInt8>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, UInt16>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, UInt32>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, UInt64>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Int8>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Int16>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Int32>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Int64>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Float32>(arguments))
|| (res = executeWithResultTypeAndLeft<ResultType, Float64>(arguments))))
throw Exception(ErrorCodes::ILLEGAL_COLUMN,
"Illegal column {} of first argument of function {}", arguments[0].column->getName(), getName());
DataTypePtr type_x = typeid_cast<const DataTypeArray *>(arguments[0].type.get())->getNestedType();
return res;
switch (type_x->getTypeId())
{
#define ON_TYPE(type) \
case TypeIndex::type: \
return executeWithResultTypeAndLeftType<ResultType, type>(arguments, input_rows_count); \
break;
SUPPORTED_TYPES(ON_TYPE)
#undef ON_TYPE
default:
throw Exception(
ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
"Arguments of function {} has nested type {}. "
"Supported types: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
getName(),
type_x->getName());
}
}
template <typename ResultType, typename LeftType>
ColumnPtr executeWithResultTypeAndLeft(const ColumnsWithTypeAndName & arguments) const
ColumnPtr executeWithResultTypeAndLeftType(const ColumnsWithTypeAndName & arguments, size_t input_rows_count) const
{
ColumnPtr res;
if ( (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, UInt8>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, UInt16>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, UInt32>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, UInt64>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Int8>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Int16>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Int32>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Int64>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Float32>(arguments))
|| (res = executeWithResultTypeAndLeftAndRight<ResultType, LeftType, Float64>(arguments)))
return res;
DataTypePtr type_y = typeid_cast<const DataTypeArray *>(arguments[1].type.get())->getNestedType();
return nullptr;
switch (type_y->getTypeId())
{
#define ON_TYPE(type) \
case TypeIndex::type: \
return executeWithResultTypeAndLeftTypeAndRightType<ResultType, LeftType, type>(arguments[0].column, arguments[1].column, input_rows_count); \
break;
SUPPORTED_TYPES(ON_TYPE)
#undef ON_TYPE
default:
throw Exception(
ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
"Arguments of function {} has nested type {}. "
"Supported types: UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64, Float32, Float64.",
getName(),
type_y->getName());
}
}
template <typename ResultType, typename LeftType, typename RightType>
ColumnPtr executeWithResultTypeAndLeftAndRight(const ColumnsWithTypeAndName & arguments) const
ColumnPtr executeWithResultTypeAndLeftTypeAndRightType(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count) const
{
ColumnPtr col_left = arguments[0].column->convertToFullColumnIfConst();
ColumnPtr col_right = arguments[1].column->convertToFullColumnIfConst();
if (!col_left || !col_right)
return nullptr;
if (typeid_cast<const ColumnConst *>(col_x.get()))
{
return executeWithLeftArgConst<ResultType, LeftType, RightType>(col_x, col_y, input_rows_count);
}
else if (typeid_cast<const ColumnConst *>(col_y.get()))
{
return executeWithLeftArgConst<ResultType, RightType, LeftType>(col_y, col_x, input_rows_count);
}
const ColumnArray * col_arr_left = checkAndGetColumn<ColumnArray>(col_left.get());
const ColumnArray * cokl_arr_right = checkAndGetColumn<ColumnArray>(col_right.get());
if (!col_arr_left || !cokl_arr_right)
return nullptr;
const auto & array_x = *assert_cast<const ColumnArray *>(col_x.get());
const auto & array_y = *assert_cast<const ColumnArray *>(col_y.get());
const ColumnVector<LeftType> * col_arr_nested_left = checkAndGetColumn<ColumnVector<LeftType>>(col_arr_left->getData());
const ColumnVector<RightType> * col_arr_nested_right = checkAndGetColumn<ColumnVector<RightType>>(cokl_arr_right->getData());
if (!col_arr_nested_left || !col_arr_nested_right)
return nullptr;
const auto & data_x = typeid_cast<const ColumnVector<LeftType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<RightType> &>(array_y.getData()).getData();
if (!col_arr_left->hasEqualOffsets(*cokl_arr_right))
throw Exception(ErrorCodes::BAD_ARGUMENTS, "Array arguments for function {} must have equal sizes", getName());
const auto & offsets_x = array_x.getOffsets();
auto col_res = ColumnVector<ResultType>::create();
if (!array_x.hasEqualOffsets(array_y))
throw Exception(ErrorCodes::SIZES_OF_ARRAYS_DONT_MATCH, "Array arguments for function {} must have equal sizes", getName());
vector(
col_arr_nested_left->getData(),
col_arr_nested_right->getData(),
col_arr_left->getOffsets(),
col_res->getData());
auto col_res = ColumnVector<ResultType>::create(input_rows_count);
auto & result_data = col_res->getData();
ColumnArray::Offset current_offset = 0;
for (size_t row = 0; row < input_rows_count; ++row)
{
const size_t array_size = offsets_x[row] - current_offset;
size_t i = 0;
/// Process chunks in vectorized manner
static constexpr size_t VEC_SIZE = 4;
typename Kernel::template State<ResultType> states[VEC_SIZE];
for (; i + VEC_SIZE < array_size; i += VEC_SIZE)
{
for (size_t j = 0; j < VEC_SIZE; ++j)
Kernel::template accumulate<ResultType>(states[j], static_cast<ResultType>(data_x[current_offset + i + j]), static_cast<ResultType>(data_y[current_offset + i + j]));
}
typename Kernel::template State<ResultType> state;
for (const auto & other_state : states)
Kernel::template combine<ResultType>(state, other_state);
/// Process the tail
for (; i < array_size; ++i)
Kernel::template accumulate<ResultType>(state, static_cast<ResultType>(data_x[current_offset + i]), static_cast<ResultType>(data_y[current_offset + i]));
result_data[row] = Kernel::template finalize<ResultType>(state);
current_offset = offsets_x[row];
}
return col_res;
}
template <typename ResultType, typename LeftType, typename RightType>
static void vector(
const PaddedPODArray<LeftType> & left,
const PaddedPODArray<RightType> & right,
const ColumnArray::Offsets & offsets,
PaddedPODArray<ResultType> & result)
ColumnPtr executeWithLeftArgConst(ColumnPtr col_x, ColumnPtr col_y, size_t input_rows_count) const
{
size_t size = offsets.size();
result.resize(size);
col_x = assert_cast<const ColumnConst *>(col_x.get())->getDataColumnPtr();
col_y = col_y->convertToFullColumnIfConst();
const auto & array_x = *assert_cast<const ColumnArray *>(col_x.get());
const auto & array_y = *assert_cast<const ColumnArray *>(col_y.get());
const auto & data_x = typeid_cast<const ColumnVector<LeftType> &>(array_x.getData()).getData();
const auto & data_y = typeid_cast<const ColumnVector<RightType> &>(array_y.getData()).getData();
const auto & offsets_x = array_x.getOffsets();
const auto & offsets_y = array_y.getOffsets();
ColumnArray::Offset prev_offset = 0;
for (auto offset_y : offsets_y)
{
if (offsets_x[0] != offset_y - prev_offset) [[unlikely]]
{
throw Exception(
ErrorCodes::SIZES_OF_ARRAYS_DONT_MATCH,
"Arguments of function {} have different array sizes: {} and {}",
getName(),
offsets_x[0],
offset_y - prev_offset);
}
prev_offset = offset_y;
}
auto col_res = ColumnVector<ResultType>::create(input_rows_count);
auto & result = col_res->getData();
ColumnArray::Offset current_offset = 0;
for (size_t row = 0; row < size; ++row)
for (size_t row = 0; row < input_rows_count; ++row)
{
size_t array_size = offsets[row] - current_offset;
const size_t array_size = offsets_x[0];
typename Kernel::template State<ResultType> state;
size_t i = 0;
@ -283,13 +346,14 @@ private:
/// To avoid combinatorial explosion of SIMD kernels, focus on
/// - the two most common input/output types (Float32 x Float32) --> Float32 and (Float64 x Float64) --> Float64 instead of 10 x
/// 10 input types x 8 output types,
/// - const/non-const inputs instead of non-const/non-const inputs
/// - the most powerful SIMD instruction set (AVX-512F).
#if USE_MULTITARGET_CODE
if constexpr ((std::is_same_v<ResultType, Float32> || std::is_same_v<ResultType, Float64>)
&& std::is_same_v<ResultType, LeftType> && std::is_same_v<LeftType, RightType>)
{
if (isArchSupported(TargetArch::AVX512F))
Kernel::template accumulateCombine<ResultType>(&left[current_offset], &right[current_offset], array_size, i, state);
Kernel::template accumulateCombine<ResultType>(&data_x[0], &data_y[current_offset], array_size, i, state);
}
#else
/// Process chunks in vectorized manner
@ -298,7 +362,7 @@ private:
for (; i + VEC_SIZE < array_size; i += VEC_SIZE)
{
for (size_t j = 0; j < VEC_SIZE; ++j)
Kernel::template accumulate<ResultType>(states[j], static_cast<ResultType>(left[i + j]), static_cast<ResultType>(right[i + j]));
Kernel::template accumulate<ResultType>(states[j], static_cast<ResultType>(data_x[i + j]), static_cast<ResultType>(data_y[current_offset + i + j]));
}
for (const auto & other_state : states)
@ -307,13 +371,14 @@ private:
/// Process the tail
for (; i < array_size; ++i)
Kernel::template accumulate<ResultType>(state, static_cast<ResultType>(left[i]), static_cast<ResultType>(right[i]));
Kernel::template accumulate<ResultType>(state, static_cast<ResultType>(data_x[i]), static_cast<ResultType>(data_y[current_offset + i]));
/// ResultType res = Kernel::template finalize<ResultType>(state);
result[row] = Kernel::template finalize<ResultType>(state);
current_offset = offsets[row];
current_offset = offsets_y[row];
}
return col_res;
}
};

1
src/Functions/array/arrayIntersect.cpp
View File

@ -20,6 +20,7 @@
#include <Columns/ColumnTuple.h>
#include <Common/HashTable/ClearableHashMap.h>
#include <Common/assert_cast.h>
#include <base/range.h>
#include <base/TypeLists.h>
#include <Interpreters/castColumn.h>

3
src/Functions/array/arrayNorm.cpp
View File

@ -175,8 +175,7 @@ public:
}
}
ColumnPtr
executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override
ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, size_t input_rows_count) const override
{
DataTypePtr type = typeid_cast<const DataTypeArray *>(arguments[0].type.get())->getNestedType();
ColumnPtr column = arguments[0].column->convertToFullColumnIfConst();

5
src/Functions/array/arraySort.cpp
View File

@ -46,7 +46,10 @@ ColumnPtr ArraySortImpl<positive, is_partial>::execute(
ErrorCodes::LOGICAL_ERROR,
"Expected fixed arguments to get the limit for partial array sort"
);
return fixed_arguments[0].column.get()->getUInt(0);
/// During dryRun the input column might be empty
if (!fixed_arguments[0].column->empty())
return fixed_arguments[0].column->getUInt(0);
}
return 0;
}();

34
src/Functions/countDigits.cpp
View File

@ -20,6 +20,40 @@ namespace ErrorCodes
namespace
{
template <typename T>
int digits10(T x)
{
if (x < 10ULL)
return 1;
if (x < 100ULL)
return 2;
if (x < 1000ULL)
return 3;
if (x < 1000000000000ULL)
{
if (x < 100000000ULL)
{
if (x < 1000000ULL)
{
if (x < 10000ULL)
return 4;
else
return 5 + (x >= 100000ULL);
}
return 7 + (x >= 10000000ULL);
}
if (x < 10000000000ULL)
return 9 + (x >= 1000000000ULL);
return 11 + (x >= 100000000000ULL);
}
return 12 + digits10(x / 1000000000000ULL);
}
/// Returns number of decimal digits you need to represent the value.
/// For Decimal values takes in account their scales: calculates result over underlying int type which is (value * scale).
/// countDigits(42) = 2, countDigits(42.000) = 5, countDigits(0.04200) = 4.

2
src/Functions/dateDiff.cpp
View File

@ -6,7 +6,7 @@
#include <Columns/ColumnsDateTime.h>
#include <Columns/ColumnsNumber.h>
#include <Columns/ColumnDecimal.h>
#include <Formats/FormatSettings.h>
#include <Functions/IFunction.h>
#include <Functions/FunctionHelpers.h>
#include <Functions/FunctionFactory.h>

3
src/Functions/date_trunc.cpp
View File

@ -1,9 +1,10 @@
#include <Columns/ColumnConst.h>
#include <Columns/ColumnsNumber.h>
#include <Columns/ColumnString.h>
#include <Columns/ColumnsNumber.h>
#include <DataTypes/DataTypeDate.h>
#include <DataTypes/DataTypeDateTime.h>
#include <DataTypes/DataTypeInterval.h>
#include <Formats/FormatSettings.h>
#include <Functions/DateTimeTransforms.h>
#include <Functions/FunctionFactory.h>

1
src/Functions/hasColumnInTable.cpp
View File

@ -7,6 +7,7 @@
#include <Storages/IStorage.h>
#include <Interpreters/Cluster.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Storages/getStructureOfRemoteTable.h>

1
src/Functions/keyvaluepair/ArgumentExtractor.h
View File

@ -4,6 +4,7 @@
#include <Columns/ColumnsNumber.h>
#include <Core/ColumnsWithTypeAndName.h>
#include <list>
#include <optional>
namespace DB

2
src/Functions/tupleConcat.cpp
View File

@ -4,6 +4,8 @@
#include <Functions/FunctionHelpers.h>
#include <Functions/IFunction.h>
#include <base/range.h>
namespace DB
{
namespace ErrorCodes

3
src/IO/CachedInMemoryReadBufferFromFile.h
View File

@ -1,7 +1,8 @@
#pragma once
#include <Common/PageCache.h>
#include <IO/ReadBufferFromFileBase.h>
#include <IO/ReadSettings.h>
#include <Common/PageCache.h>
namespace DB
{

9
src/IO/ReadHelpers.h
View File

@ -154,9 +154,12 @@ inline void readIPv6Binary(IPv6 & ip, ReadBuffer & buf)
size_t size = 0;
readVarUInt(size, buf);
if (size != IPV6_BINARY_LENGTH)
throw Exception(ErrorCodes::SIZE_OF_FIXED_STRING_DOESNT_MATCH,
"Size of the string {} doesn't match size of binary IPv6 {}", size, IPV6_BINARY_LENGTH);
if (size != sizeof(IPv6::UnderlyingType))
throw Exception(
ErrorCodes::SIZE_OF_FIXED_STRING_DOESNT_MATCH,
"Size of the string {} doesn't match size of binary IPv6 {}",
size,
sizeof(IPv6::UnderlyingType));
buf.readStrict(reinterpret_cast<char*>(&ip.toUnderType()), size);
}

1
src/IO/S3/copyS3File.h
View File

@ -14,6 +14,7 @@
namespace DB
{
struct ReadSettings;
class SeekableReadBuffer;
using CreateReadBuffer = std::function<std::unique_ptr<SeekableReadBuffer>()>;

2
src/IO/examples/read_buffer_from_hdfs.cpp
View File

@ -6,6 +6,8 @@
#include <base/types.h>
#include <Common/Config/ConfigProcessor.h>
#include <Poco/Util/MapConfiguration.h>
using namespace DB;
int main()

1
src/Interpreters/ActionLocksManager.cpp
View File

@ -1,5 +1,6 @@
#include "ActionLocksManager.h"
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Databases/IDatabase.h>
#include <Storages/IStorage.h>

1
src/Interpreters/ActionsVisitor.cpp
View File

@ -44,6 +44,7 @@
#include <Interpreters/ExpressionActions.h>
#include <Interpreters/misc.h>
#include <Interpreters/ActionsVisitor.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/Set.h>
#include <Interpreters/evaluateConstantExpression.h>
#include <Interpreters/convertFieldToType.h>

5
src/Interpreters/ActionsVisitor.h
View File

@ -1,13 +1,14 @@
#pragma once
#include <string_view>
#include <Core/ColumnNumbers.h>
#include <Core/ColumnWithTypeAndName.h>
#include <Core/NamesAndTypes.h>
#include <Interpreters/Context_fwd.h>
#include <Interpreters/InDepthNodeVisitor.h>
#include <Interpreters/PreparedSets.h>
#include <Parsers/IAST.h>
#include <Core/ColumnNumbers.h>
#include <Core/ColumnWithTypeAndName.h>
#include <QueryPipeline/SizeLimits.h>
namespace DB
{

1
src/Interpreters/AsynchronousInsertQueue.cpp
View File

@ -11,6 +11,7 @@
#include <IO/copyData.h>
#include <Interpreters/AsynchronousInsertLog.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/InterpreterInsertQuery.h>
#include <Interpreters/ProcessList.h>
#include <Interpreters/executeQuery.h>

7
src/Interpreters/CollectJoinOnKeysVisitor.h
View File

@ -1,11 +1,12 @@
#pragma once
#include <Core/Joins.h>
#include <Core/Names.h>
#include <Interpreters/Aliases.h>
#include <Interpreters/DatabaseAndTableWithAlias.h>
#include <Interpreters/InDepthNodeVisitor.h>
#include <Parsers/ASTFunction.h>
#include <Parsers/queryToString.h>
#include <Interpreters/InDepthNodeVisitor.h>
#include <Interpreters/DatabaseAndTableWithAlias.h>
#include <Interpreters/Aliases.h>
namespace DB

15
src/Interpreters/Context.cpp
View File

@ -1073,7 +1073,9 @@ void Context::setTemporaryStoragePath(const String & path, size_t max_size)
setupTmpPath(shared->log, disk->getPath());
}
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(volume, max_size);
TemporaryDataOnDiskSettings temporary_data_on_disk_settings;
temporary_data_on_disk_settings.max_size_on_disk = max_size;
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(std::move(volume), std::move(temporary_data_on_disk_settings));
}
void Context::setTemporaryStoragePolicy(const String & policy_name, size_t max_size)
@ -1093,7 +1095,7 @@ void Context::setTemporaryStoragePolicy(const String & policy_name, size_t max_s
VolumePtr volume = tmp_policy->getVolume(0);
if (volume->getDisks().empty())
throw Exception(ErrorCodes::NO_ELEMENTS_IN_CONFIG, "No disks volume for temporary files");
throw Exception(ErrorCodes::NO_ELEMENTS_IN_CONFIG, "No disks volume for temporary files");
for (const auto & disk : volume->getDisks())
{
@ -1119,7 +1121,9 @@ void Context::setTemporaryStoragePolicy(const String & policy_name, size_t max_s
if (shared->root_temp_data_on_disk)
throw Exception(ErrorCodes::LOGICAL_ERROR, "Temporary storage is already set");
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(volume, max_size);
TemporaryDataOnDiskSettings temporary_data_on_disk_settings;
temporary_data_on_disk_settings.max_size_on_disk = max_size;
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(std::move(volume), std::move(temporary_data_on_disk_settings));
}
void Context::setTemporaryStorageInCache(const String & cache_disk_name, size_t max_size)
@ -1140,7 +1144,10 @@ void Context::setTemporaryStorageInCache(const String & cache_disk_name, size_t
shared->tmp_path = file_cache->getBasePath();
VolumePtr volume = createLocalSingleDiskVolume(shared->tmp_path, shared->getConfigRefWithLock(lock));
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(volume, file_cache.get(), max_size);
TemporaryDataOnDiskSettings temporary_data_on_disk_settings;
temporary_data_on_disk_settings.max_size_on_disk = max_size;
shared->root_temp_data_on_disk = std::make_shared<TemporaryDataOnDiskScope>(std::move(volume), file_cache.get(), std::move(temporary_data_on_disk_settings));
}
void Context::setFlagsPath(const String & path)

15
src/Interpreters/Context.h
View File

@ -19,9 +19,8 @@
#include <Disks/IO/getThreadPoolReader.h>
#include <Interpreters/ClientInfo.h>
#include <Interpreters/Context_fwd.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/StorageID.h>
#include <Interpreters/MergeTreeTransactionHolder.h>
#include <Common/Scheduler/IResourceManager.h>
#include <Parsers/IAST_fwd.h>
#include <Server/HTTP/HTTPContext.h>
#include <Storages/ColumnsDescription.h>
@ -149,6 +148,18 @@ template <class Queue>
class MergeTreeBackgroundExecutor;
class AsyncLoader;
struct TemporaryTableHolder;
using TemporaryTablesMapping = std::map<String, std::shared_ptr<TemporaryTableHolder>>;
class LoadTask;
using LoadTaskPtr = std::shared_ptr<LoadTask>;
using LoadTaskPtrs = std::vector<LoadTaskPtr>;
class IClassifier;
using ClassifierPtr = std::shared_ptr<IClassifier>;
class IResourceManager;
using ResourceManagerPtr = std::shared_ptr<IResourceManager>;
/// Scheduling policy can be changed using `background_merges_mutations_scheduling_policy` config option.
/// By default concurrent merges are scheduled using "round_robin" to ensure fair and starvation-free operation.
/// Previously in heavily overloaded shards big merges could possibly be starved by smaller

3
src/Interpreters/DDLTask.cpp
View File

@ -2,6 +2,8 @@
#include <base/sort.h>
#include <Common/DNSResolver.h>
#include <Common/isLocalAddress.h>
#include <Databases/DatabaseReplicated.h>
#include <Interpreters/DatabaseCatalog.h>
#include <IO/WriteHelpers.h>
#include <IO/ReadHelpers.h>
#include <IO/Operators.h>
@ -14,7 +16,6 @@
#include <Parsers/parseQuery.h>
#include <Parsers/queryToString.h>
#include <Parsers/ASTQueryWithTableAndOutput.h>
#include <Databases/DatabaseReplicated.h>
namespace DB

44
src/Interpreters/DatabaseCatalog.cpp
View File

@ -2,6 +2,7 @@
#include <mutex>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/Context.h>
#include <Interpreters/TableNameHints.h>
#include <Interpreters/loadMetadata.h>
#include <Interpreters/executeQuery.h>
#include <Interpreters/InterpreterCreateQuery.h>
@ -1142,7 +1143,7 @@ void DatabaseCatalog::dequeueDroppedTableCleanup(StorageID table_id)
TableMarkedAsDropped dropped_table;
{
std::lock_guard lock(tables_marked_dropped_mutex);
auto latest_drop_time = std::numeric_limits<time_t>::min();
time_t latest_drop_time = std::numeric_limits<time_t>::min();
auto it_dropped_table = tables_marked_dropped.end();
for (auto it = tables_marked_dropped.begin(); it != tables_marked_dropped.end(); ++it)
{
@ -1167,7 +1168,7 @@ void DatabaseCatalog::dequeueDroppedTableCleanup(StorageID table_id)
}
if (it_dropped_table == tables_marked_dropped.end())
throw Exception(ErrorCodes::UNKNOWN_TABLE,
"Table {} is being dropped, has been dropped, or the database engine does not support UNDROP",
"The drop task of table {} is in progress, has been dropped or the database engine doesn't support it",
table_id.getNameForLogs());
latest_metadata_dropped_path = it_dropped_table->metadata_path;
String table_metadata_path = getPathForMetadata(it_dropped_table->table_id);
@ -1705,4 +1706,43 @@ DDLGuard::~DDLGuard()
releaseTableLock();
}
std::pair<String, String> TableNameHints::getHintForTable(const String & table_name) const
{
auto results = this->getHints(table_name, getAllRegisteredNames());
if (results.empty())
return getExtendedHintForTable(table_name);
return std::make_pair(database->getDatabaseName(), results[0]);
}
std::pair<String, String> TableNameHints::getExtendedHintForTable(const String & table_name) const
{
/// load all available databases from the DatabaseCatalog instance
auto & database_catalog = DatabaseCatalog::instance();
auto all_databases = database_catalog.getDatabases();
for (const auto & [db_name, db] : all_databases)
{
/// this case should be covered already by getHintForTable
if (db_name == database->getDatabaseName())
continue;
TableNameHints hints(db, context);
auto results = hints.getHints(table_name);
/// if the results are not empty, return the first instance of the table_name
/// and the corresponding database_name that was found.
if (!results.empty())
return std::make_pair(db_name, results[0]);
}
return {};
}
Names TableNameHints::getAllRegisteredNames() const
{
Names result;
if (database)
for (auto table_it = database->getTablesIterator(context); table_it->isValid(); table_it->next())
result.emplace_back(table_it->name());
return result;
}
}

64
src/Interpreters/DatabaseCatalog.h
View File

@ -1,13 +1,11 @@
#pragma once
#include <Core/UUID.h>
#include <Databases/IDatabase.h>
#include <Databases/TablesDependencyGraph.h>
#include <Interpreters/Context_fwd.h>
#include <Interpreters/StorageID.h>
#include <Parsers/IAST_fwd.h>
#include <Storages/IStorage_fwd.h>
#include <Common/NamePrompter.h>
#include <Common/SharedMutex.h>
#include <boost/noncopyable.hpp>
@ -365,68 +363,6 @@ private:
static constexpr time_t DBMS_DEFAULT_DISK_RELOAD_PERIOD_SEC = 5;
};
class TableNameHints : public IHints<>
{
public:
TableNameHints(ConstDatabasePtr database_, ContextPtr context_)
: context(context_),
database(database_)
{
}
/// getHintForTable tries to get a hint for the provided table_name in the provided
/// database. If the results are empty, it goes for extended hints for the table
/// with getExtendedHintForTable which looks for the table name in every database that's
/// available in the database catalog. It finally returns a single hint which is the database
/// name and table_name pair which is similar to the table_name provided. Perhaps something to
/// consider is should we return more than one pair of hint?
std::pair<String, String> getHintForTable(const String & table_name) const
{
auto results = this->getHints(table_name, getAllRegisteredNames());
if (results.empty())
return getExtendedHintForTable(table_name);
return std::make_pair(database->getDatabaseName(), results[0]);
}
/// getExtendedHintsForTable tries to get hint for the given table_name across all
/// the databases that are available in the database catalog.
std::pair<String, String> getExtendedHintForTable(const String & table_name) const
{
/// load all available databases from the DatabaseCatalog instance
auto & database_catalog = DatabaseCatalog::instance();
auto all_databases = database_catalog.getDatabases();
for (const auto & [db_name, db] : all_databases)
{
/// this case should be covered already by getHintForTable
if (db_name == database->getDatabaseName())
continue;
TableNameHints hints(db, context);
auto results = hints.getHints(table_name);
/// if the results are not empty, return the first instance of the table_name
/// and the corresponding database_name that was found.
if (!results.empty())
return std::make_pair(db_name, results[0]);
}
return {};
}
Names getAllRegisteredNames() const override
{
Names result;
if (database)
for (auto table_it = database->getTablesIterator(context); table_it->isValid(); table_it->next())
result.emplace_back(table_it->name());
return result;
}
private:
ContextPtr context;
ConstDatabasePtr database;
};
/// This class is useful when creating a table or database.
/// Usually we create IStorage/IDatabase object first and then add it to IDatabase/DatabaseCatalog.

1
src/Interpreters/IKeyValueEntity.h
View File

@ -1,5 +1,6 @@
#pragma once
#include <Core/Block.h>
#include <Core/Names.h>
#include <Processors/Chunk.h>

1
src/Interpreters/InJoinSubqueriesPreprocessor.cpp
View File

@ -1,6 +1,7 @@
#include <Interpreters/InJoinSubqueriesPreprocessor.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseAndTableWithAlias.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/IdentifierSemantic.h>
#include <Interpreters/InDepthNodeVisitor.h>
#include <Storages/StorageDistributed.h>

1
src/Interpreters/InterpreterCheckQuery.cpp
View File

@ -16,6 +16,7 @@
#include <DataTypes/DataTypeString.h>
#include <Interpreters/Context.h>
#include <Interpreters/DatabaseCatalog.h>
#include <Interpreters/ProcessList.h>
#include <Parsers/ASTCheckQuery.h>

Some files were not shown because too many files have changed in this diff Show More