diff --git a/.github/workflows/cancel.yml b/.github/workflows/cancel.yml index cb06d853219..3c2be767ad2 100644 --- a/.github/workflows/cancel.yml +++ b/.github/workflows/cancel.yml @@ -6,7 +6,7 @@ env: on: # yamllint disable-line rule:truthy workflow_run: - workflows: ["PullRequestCI", "ReleaseCI", "DocsCheck", "BackportPR"] + workflows: ["PullRequestCI", "ReleaseBranchCI", "DocsCheck", "BackportPR"] types: - requested jobs: diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml index 801f7eda94a..612bb1f8f9b 100644 --- a/.github/workflows/nightly.yml +++ b/.github/workflows/nightly.yml @@ -122,3 +122,58 @@ jobs: docker ps --quiet | xargs --no-run-if-empty docker kill ||: docker ps --all --quiet | xargs --no-run-if-empty docker rm -f ||: sudo rm -fr "$TEMP_PATH" "$CACHES_PATH" + SonarCloud: + runs-on: [self-hosted, builder] + env: + SONAR_SCANNER_VERSION: 4.7.0.2747 + SONAR_SERVER_URL: "https://sonarcloud.io" + BUILD_WRAPPER_OUT_DIR: build_wrapper_output_directory # Directory where build-wrapper output will be placed + CC: clang-15 + CXX: clang++-15 + steps: + - uses: actions/checkout@v2 + with: + fetch-depth: 0 # Shallow clones should be disabled for a better relevancy of analysis + submodules: true + - name: Set up JDK 11 + uses: actions/setup-java@v1 + with: + java-version: 11 + - name: Download and set up sonar-scanner + env: + SONAR_SCANNER_DOWNLOAD_URL: https://binaries.sonarsource.com/Distribution/sonar-scanner-cli/sonar-scanner-cli-${{ env.SONAR_SCANNER_VERSION }}-linux.zip + run: | + mkdir -p "$HOME/.sonar" + curl -sSLo "$HOME/.sonar/sonar-scanner.zip" "${{ env.SONAR_SCANNER_DOWNLOAD_URL }}" + unzip -o "$HOME/.sonar/sonar-scanner.zip" -d "$HOME/.sonar/" + echo "$HOME/.sonar/sonar-scanner-${{ env.SONAR_SCANNER_VERSION }}-linux/bin" >> "$GITHUB_PATH" + - name: Download and set up build-wrapper + env: + BUILD_WRAPPER_DOWNLOAD_URL: ${{ env.SONAR_SERVER_URL }}/static/cpp/build-wrapper-linux-x86.zip + run: | + curl -sSLo "$HOME/.sonar/build-wrapper-linux-x86.zip" "${{ env.BUILD_WRAPPER_DOWNLOAD_URL }}" + unzip -o "$HOME/.sonar/build-wrapper-linux-x86.zip" -d "$HOME/.sonar/" + echo "$HOME/.sonar/build-wrapper-linux-x86" >> "$GITHUB_PATH" + - name: Set Up Build Tools + run: | + sudo apt-get update + sudo apt-get install -yq git cmake ccache python3 ninja-build + sudo bash -c "$(wget -O - https://apt.llvm.org/llvm.sh)" + - name: Run build-wrapper + run: | + mkdir build + cd build + cmake .. + cd .. + build-wrapper-linux-x86-64 --out-dir ${{ env.BUILD_WRAPPER_OUT_DIR }} cmake --build build/ + - name: Run sonar-scanner + env: + GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} + SONAR_TOKEN: ${{ secrets.SONAR_TOKEN }} + run: | + sonar-scanner \ + --define sonar.host.url="${{ env.SONAR_SERVER_URL }}" \ + --define sonar.cfamily.build-wrapper-output="${{ env.BUILD_WRAPPER_OUT_DIR }}" \ + --define sonar.projectKey="ClickHouse_ClickHouse" \ + --define sonar.organization="clickhouse-java" \ + --define sonar.exclusions="**/*.java,**/*.ts,**/*.js,**/*.css,**/*.sql" diff --git a/.github/workflows/pull_request.yml b/.github/workflows/pull_request.yml index 3951f99b16b..23245c16374 100644 --- a/.github/workflows/pull_request.yml +++ b/.github/workflows/pull_request.yml @@ -2023,6 +2023,7 @@ jobs: docker ps --all --quiet | xargs --no-run-if-empty docker rm -f ||: sudo rm -fr "$TEMP_PATH" TestsBugfixCheck: + needs: [CheckLabels, StyleCheck] runs-on: [self-hosted, stress-tester] steps: - name: Set envs diff --git a/.snyk b/.snyk new file mode 100644 index 00000000000..7acc6b9fbf5 --- /dev/null +++ b/.snyk @@ -0,0 +1,4 @@ +# Snyk (https://snyk.io) policy file +exclude: + global: + - tests/** diff --git a/CHANGELOG.md b/CHANGELOG.md index 22f6afc4901..68767612892 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -31,6 +31,8 @@ * Add OpenTelemetry support to ON CLUSTER DDL (require `distributed_ddl_entry_format_version` to be set to 4). [#41484](https://github.com/ClickHouse/ClickHouse/pull/41484) ([Frank Chen](https://github.com/FrankChen021)). * Added system table `asynchronous_insert_log`. It contains information about asynchronous inserts (including results of queries in fire-and-forget mode (with `wait_for_async_insert=0`)) for better introspection. [#42040](https://github.com/ClickHouse/ClickHouse/pull/42040) ([Anton Popov](https://github.com/CurtizJ)). * Add support for methods `lz4`, `bz2`, `snappy` in HTTP's `Accept-Encoding` which is a non-standard extension to HTTP protocol. [#42071](https://github.com/ClickHouse/ClickHouse/pull/42071) ([Nikolay Degterinsky](https://github.com/evillique)). +* Adds Morton Coding (ZCurve) encode/decode functions. [#41753](https://github.com/ClickHouse/ClickHouse/pull/41753) ([Constantine Peresypkin](https://github.com/pkit)). +* Add support for `SET setting_name = DEFAULT`. [#42187](https://github.com/ClickHouse/ClickHouse/pull/42187) ([Filatenkov Artur](https://github.com/FArthur-cmd)). #### Experimental Feature * Added new infrastructure for query analysis and planning under the `allow_experimental_analyzer` setting. [#31796](https://github.com/ClickHouse/ClickHouse/pull/31796) ([Maksim Kita](https://github.com/kitaisreal)). @@ -66,8 +68,7 @@ * Allow readable size values (like `1TB`) in cache config. [#41688](https://github.com/ClickHouse/ClickHouse/pull/41688) ([Kseniia Sumarokova](https://github.com/kssenii)). * ClickHouse could cache stale DNS entries for some period of time (15 seconds by default) until the cache won't be updated asynchronously. During these periods ClickHouse can nevertheless try to establish a connection and produce errors. This behavior is fixed. [#41707](https://github.com/ClickHouse/ClickHouse/pull/41707) ([Nikita Mikhaylov](https://github.com/nikitamikhaylov)). * Add interactive history search with fzf-like utility (fzf/sk) for `clickhouse-client`/`clickhouse-local` (note you can use `FZF_DEFAULT_OPTS`/`SKIM_DEFAULT_OPTIONS` to additionally configure the behavior). [#41730](https://github.com/ClickHouse/ClickHouse/pull/41730) ([Azat Khuzhin](https://github.com/azat)). -* -Only allow clients connecting to a secure server with an invalid certificate only to proceed with the '--accept-certificate' flag. [#41743](https://github.com/ClickHouse/ClickHouse/pull/41743) ([Yakov Olkhovskiy](https://github.com/yakov-olkhovskiy)). +* Only allow clients connecting to a secure server with an invalid certificate only to proceed with the '--accept-certificate' flag. [#41743](https://github.com/ClickHouse/ClickHouse/pull/41743) ([Yakov Olkhovskiy](https://github.com/yakov-olkhovskiy)). * Add function `tryBase58Decode`, similar to the existing function `tryBase64Decode`. [#41824](https://github.com/ClickHouse/ClickHouse/pull/41824) ([Robert Schulze](https://github.com/rschu1ze)). * Improve feedback when replacing partition with different primary key. Fixes [#34798](https://github.com/ClickHouse/ClickHouse/issues/34798). [#41838](https://github.com/ClickHouse/ClickHouse/pull/41838) ([Salvatore](https://github.com/tbsal)). * Fix parallel parsing: segmentator now checks `max_block_size`. This fixed memory overallocation in case of parallel parsing and small LIMIT. [#41852](https://github.com/ClickHouse/ClickHouse/pull/41852) ([Vitaly Baranov](https://github.com/vitlibar)). @@ -86,6 +87,8 @@ Only allow clients connecting to a secure server with an invalid certificate onl * Fix rarely invalid cast of aggregate state types with complex types such as Decimal. This fixes [#42408](https://github.com/ClickHouse/ClickHouse/issues/42408). [#42417](https://github.com/ClickHouse/ClickHouse/pull/42417) ([Amos Bird](https://github.com/amosbird)). * Allow to use `Date32` arguments for `dateName` function. [#42554](https://github.com/ClickHouse/ClickHouse/pull/42554) ([Roman Vasin](https://github.com/rvasin)). * Now filters with NULL literals will be used during index analysis. [#34063](https://github.com/ClickHouse/ClickHouse/issues/34063). [#41842](https://github.com/ClickHouse/ClickHouse/pull/41842) ([Amos Bird](https://github.com/amosbird)). +* Merge parts if every part in the range is older than a certain threshold. The threshold can be set by using `min_age_to_force_merge_seconds`. This closes [#35836](https://github.com/ClickHouse/ClickHouse/issues/35836). [#42423](https://github.com/ClickHouse/ClickHouse/pull/42423) ([Antonio Andelic](https://github.com/antonio2368)). This is continuation of [#39550i](https://github.com/ClickHouse/ClickHouse/pull/39550) by [@fastio](https://github.com/fastio) who implemented most of the logic. +* Improve the time to recover lost keeper connections. [#42541](https://github.com/ClickHouse/ClickHouse/pull/42541) ([Raúl Marín](https://github.com/Algunenano)). #### Build/Testing/Packaging Improvement * Add fuzzer for table definitions [#40096](https://github.com/ClickHouse/ClickHouse/pull/40096) ([Anton Popov](https://github.com/CurtizJ)). This represents the biggest advancement for ClickHouse testing in this year so far. diff --git a/README.md b/README.md index 003b78a3cbb..f90df9686c2 100644 --- a/README.md +++ b/README.md @@ -5,7 +5,7 @@ ClickHouse® is an open-source column-oriented database management system that a ## Useful Links * [Official website](https://clickhouse.com/) has a quick high-level overview of ClickHouse on the main page. -* [ClickHouse Cloud](https://clickhouse.com/cloud) ClickHouse as a service, built by the creators and maintainers. +* [ClickHouse Cloud](https://clickhouse.cloud) ClickHouse as a service, built by the creators and maintainers. * [Tutorial](https://clickhouse.com/docs/en/getting_started/tutorial/) shows how to set up and query a small ClickHouse cluster. * [Documentation](https://clickhouse.com/docs/en/) provides more in-depth information. * [YouTube channel](https://www.youtube.com/c/ClickHouseDB) has a lot of content about ClickHouse in video format. @@ -16,5 +16,6 @@ ClickHouse® is an open-source column-oriented database management system that a * [Contacts](https://clickhouse.com/company/contact) can help to get your questions answered if there are any. ## Upcoming events -* [**v22.10 Release Webinar**](https://clickhouse.com/company/events/v22-10-release-webinar) Original creator, co-founder, and CTO of ClickHouse Alexey Milovidov will walk us through the highlights of the release, provide live demos, and share vision into what is coming in the roadmap. -* [**Introducing ClickHouse Cloud**](https://clickhouse.com/company/events/cloud-beta) Introducing ClickHouse as a service, built by creators and maintainers of the fastest OLAP database on earth. Join Tanya Bragin for a detailed walkthrough of ClickHouse Cloud capabilities, as well as a peek behind the curtain to understand the unique architecture that makes our service tick. +* [**v22.11 Release Webinar**](https://clickhouse.com/company/events/v22-11-release-webinar) Original creator, co-founder, and CTO of ClickHouse Alexey Milovidov will walk us through the highlights of the release, provide live demos, and share vision into what is coming in the roadmap. +* [**ClickHouse Meetup at the Deutsche Bank office in Berlin**](https://www.meetup.com/clickhouse-berlin-user-group/events/289311596/) Hear from Deutsche Bank on why they chose ClickHouse for big sensitive data in a regulated environment. The ClickHouse team will then present how ClickHouse is used for real time financial data analytics, including tick data, trade analytics and risk management. +* [**AWS re:Invent**](https://clickhouse.com/company/events/aws-reinvent) Core members of the ClickHouse team -- including 2 of our founders -- will be at re:Invent from November 29 to December 3. We are available on the show floor, but are also determining interest in holding an event during the time there. diff --git a/cmake/clang_tidy.cmake b/cmake/clang_tidy.cmake index 200282234ca..57295682487 100644 --- a/cmake/clang_tidy.cmake +++ b/cmake/clang_tidy.cmake @@ -3,10 +3,20 @@ option (ENABLE_CLANG_TIDY "Use clang-tidy static analyzer" OFF) if (ENABLE_CLANG_TIDY) - find_program (CLANG_TIDY_PATH NAMES "clang-tidy" "clang-tidy-15" "clang-tidy-14" "clang-tidy-13" "clang-tidy-12") + find_program (CLANG_TIDY_CACHE_PATH NAMES "clang-tidy-cache") + if (CLANG_TIDY_CACHE_PATH) + find_program (_CLANG_TIDY_PATH NAMES "clang-tidy" "clang-tidy-15" "clang-tidy-14" "clang-tidy-13" "clang-tidy-12") + + # Why do we use ';' here? + # It's a cmake black magic: https://cmake.org/cmake/help/latest/prop_tgt/LANG_CLANG_TIDY.html#prop_tgt:%3CLANG%3E_CLANG_TIDY + # The CLANG_TIDY_PATH is passed to CMAKE_CXX_CLANG_TIDY, which follows CXX_CLANG_TIDY syntax. + set (CLANG_TIDY_PATH "${CLANG_TIDY_CACHE_PATH};${_CLANG_TIDY_PATH}" CACHE STRING "A combined command to run clang-tidy with caching wrapper") + else () + find_program (CLANG_TIDY_PATH NAMES "clang-tidy" "clang-tidy-15" "clang-tidy-14" "clang-tidy-13" "clang-tidy-12") + endif () if (CLANG_TIDY_PATH) - message(STATUS + message (STATUS "Using clang-tidy: ${CLANG_TIDY_PATH}. The checks will be run during build process. See the .clang-tidy file at the root directory to configure the checks.") @@ -15,11 +25,15 @@ if (ENABLE_CLANG_TIDY) # clang-tidy requires assertions to guide the analysis # Note that NDEBUG is set implicitly by CMake for non-debug builds - set(COMPILER_FLAGS "${COMPILER_FLAGS} -UNDEBUG") + set (COMPILER_FLAGS "${COMPILER_FLAGS} -UNDEBUG") - # The variable CMAKE_CXX_CLANG_TIDY will be set inside src and base directories with non third-party code. + # The variable CMAKE_CXX_CLANG_TIDY will be set inside the following directories with non third-party code. + # - base + # - programs + # - src + # - utils # set (CMAKE_CXX_CLANG_TIDY "${CLANG_TIDY_PATH}") else () - message(${RECONFIGURE_MESSAGE_LEVEL} "clang-tidy is not found") + message (${RECONFIGURE_MESSAGE_LEVEL} "clang-tidy is not found") endif () endif () diff --git a/contrib/cctz b/contrib/cctz index 7a454c25c7d..5c8528fb35e 160000 --- a/contrib/cctz +++ b/contrib/cctz @@ -1 +1 @@ -Subproject commit 7a454c25c7d16053bcd327cdd16329212a08fa4a +Subproject commit 5c8528fb35e89ee0b3a7157490423fba0d4dd7b5 diff --git a/contrib/libcxx b/contrib/libcxx index 172b2ae074f..4db7f838afd 160000 --- a/contrib/libcxx +++ b/contrib/libcxx @@ -1 +1 @@ -Subproject commit 172b2ae074f6755145b91c53a95c8540c1468239 +Subproject commit 4db7f838afd3139eb3761694b04d31275df45d2d diff --git a/contrib/libcxx-cmake/CMakeLists.txt b/contrib/libcxx-cmake/CMakeLists.txt index 6f42a479588..53c6ff58f83 100644 --- a/contrib/libcxx-cmake/CMakeLists.txt +++ b/contrib/libcxx-cmake/CMakeLists.txt @@ -25,6 +25,7 @@ set(SRCS "${LIBCXX_SOURCE_DIR}/src/ios.cpp" "${LIBCXX_SOURCE_DIR}/src/ios.instantiations.cpp" "${LIBCXX_SOURCE_DIR}/src/iostream.cpp" +"${LIBCXX_SOURCE_DIR}/src/legacy_debug_handler.cpp" "${LIBCXX_SOURCE_DIR}/src/legacy_pointer_safety.cpp" "${LIBCXX_SOURCE_DIR}/src/locale.cpp" "${LIBCXX_SOURCE_DIR}/src/memory.cpp" @@ -49,6 +50,7 @@ set(SRCS "${LIBCXX_SOURCE_DIR}/src/valarray.cpp" "${LIBCXX_SOURCE_DIR}/src/variant.cpp" "${LIBCXX_SOURCE_DIR}/src/vector.cpp" +"${LIBCXX_SOURCE_DIR}/src/verbose_abort.cpp" ) add_library(cxx ${SRCS}) diff --git a/contrib/libcxxabi b/contrib/libcxxabi index 6eb7cc7a7bd..a736a6b3c6a 160000 --- a/contrib/libcxxabi +++ b/contrib/libcxxabi @@ -1 +1 @@ -Subproject commit 6eb7cc7a7bdd779e6734d1b9fb451df2274462d7 +Subproject commit a736a6b3c6a7b8aae2ebad629ca21b2c55b4820e diff --git a/contrib/libcxxabi-cmake/CMakeLists.txt b/contrib/libcxxabi-cmake/CMakeLists.txt index bf1ede8a60e..a59452eee9a 100644 --- a/contrib/libcxxabi-cmake/CMakeLists.txt +++ b/contrib/libcxxabi-cmake/CMakeLists.txt @@ -9,6 +9,7 @@ set(SRCS "${LIBCXXABI_SOURCE_DIR}/src/cxa_exception_storage.cpp" "${LIBCXXABI_SOURCE_DIR}/src/cxa_guard.cpp" "${LIBCXXABI_SOURCE_DIR}/src/cxa_handlers.cpp" +# "${LIBCXXABI_SOURCE_DIR}/src/cxa_noexception.cpp" "${LIBCXXABI_SOURCE_DIR}/src/cxa_personality.cpp" "${LIBCXXABI_SOURCE_DIR}/src/cxa_thread_atexit.cpp" "${LIBCXXABI_SOURCE_DIR}/src/cxa_vector.cpp" diff --git a/contrib/llvm-project-cmake/CMakeLists.txt b/contrib/llvm-project-cmake/CMakeLists.txt index 6a73ae0f0c6..7af4a23bc9d 100644 --- a/contrib/llvm-project-cmake/CMakeLists.txt +++ b/contrib/llvm-project-cmake/CMakeLists.txt @@ -21,6 +21,9 @@ set (LLVM_INCLUDE_DIRS "${ClickHouse_BINARY_DIR}/contrib/llvm-project/llvm/include" ) set (LLVM_LIBRARY_DIRS "${ClickHouse_BINARY_DIR}/contrib/llvm-project/llvm") +# NOTE: You should not remove this line since otherwise it will use default 20, +# and llvm cannot be compiled with bundled libcxx and 20 standard. +set (CMAKE_CXX_STANDARD 14) # This list was generated by listing all LLVM libraries, compiling the binary and removing all libraries while it still compiles. set (REQUIRED_LLVM_LIBRARIES diff --git a/contrib/rocksdb b/contrib/rocksdb index e7c2b2f7bcf..2c8998e26c6 160000 --- a/contrib/rocksdb +++ b/contrib/rocksdb @@ -1 +1 @@ -Subproject commit e7c2b2f7bcf3b4b33892a1a6d25c32a93edfbdb9 +Subproject commit 2c8998e26c6d46b27c710d7829c3a15e34959f70 diff --git a/contrib/zlib-ng b/contrib/zlib-ng index bffad6f6fe7..50f0eae1a41 160000 --- a/contrib/zlib-ng +++ b/contrib/zlib-ng @@ -1 +1 @@ -Subproject commit bffad6f6fe74d6a2f92e2668390664a926c68733 +Subproject commit 50f0eae1a411764cd6d1e85b3ce471438acd3c1c diff --git a/docker/packager/binary/Dockerfile b/docker/packager/binary/Dockerfile index 77afc3e924b..06c3c0d80f0 100644 --- a/docker/packager/binary/Dockerfile +++ b/docker/packager/binary/Dockerfile @@ -91,6 +91,9 @@ ENV PATH="$PATH:/usr/local/go/bin" ENV GOPATH=/workdir/go ENV GOCACHE=/workdir/ +RUN curl https://raw.githubusercontent.com/matus-chochlik/ctcache/7fd516e91c17779cbc6fc18bd119313d9532dd90/clang-tidy-cache -Lo /usr/bin/clang-tidy-cache \ + && chmod +x /usr/bin/clang-tidy-cache + RUN mkdir /workdir && chmod 777 /workdir WORKDIR /workdir diff --git a/docker/packager/packager b/docker/packager/packager index 83629dc7408..7f6bd8818fb 100755 --- a/docker/packager/packager +++ b/docker/packager/packager @@ -258,6 +258,10 @@ def parse_env_variables( if clang_tidy: # 15G is not enough for tidy build cache_maxsize = "25G" + + # `CTCACHE_DIR` has the same purpose as the `CCACHE_DIR` above. + # It's there to have the clang-tidy cache embedded into our standard `CCACHE_DIR` + result.append("CTCACHE_DIR=/ccache/clang-tidy-cache") result.append(f"CCACHE_MAXSIZE={cache_maxsize}") if distcc_hosts: @@ -282,9 +286,7 @@ def parse_env_variables( cmake_flags.append("-DENABLE_TESTS=1") if shared_libraries: - cmake_flags.append( - "-DUSE_STATIC_LIBRARIES=0 -DSPLIT_SHARED_LIBRARIES=1" - ) + cmake_flags.append("-DUSE_STATIC_LIBRARIES=0 -DSPLIT_SHARED_LIBRARIES=1") # We can't always build utils because it requires too much space, but # we have to build them at least in some way in CI. The shared library # build is probably the least heavy disk-wise. diff --git a/docker/server/Dockerfile.alpine b/docker/server/Dockerfile.alpine index cf4eb3fe645..8f1cf6ee98b 100644 --- a/docker/server/Dockerfile.alpine +++ b/docker/server/Dockerfile.alpine @@ -33,7 +33,7 @@ RUN arch=${TARGETARCH:-amd64} \ # lts / testing / prestable / etc ARG REPO_CHANNEL="stable" ARG REPOSITORY="https://packages.clickhouse.com/tgz/${REPO_CHANNEL}" -ARG VERSION="22.10.1.1877" +ARG VERSION="22.10.2.11" ARG PACKAGES="clickhouse-client clickhouse-server clickhouse-common-static" # user/group precreated explicitly with fixed uid/gid on purpose. diff --git a/docker/server/Dockerfile.ubuntu b/docker/server/Dockerfile.ubuntu index d26657a7979..d5fc5d8e0d3 100644 --- a/docker/server/Dockerfile.ubuntu +++ b/docker/server/Dockerfile.ubuntu @@ -21,7 +21,7 @@ RUN sed -i "s|http://archive.ubuntu.com|${apt_archive}|g" /etc/apt/sources.list ARG REPO_CHANNEL="stable" ARG REPOSITORY="deb https://packages.clickhouse.com/deb ${REPO_CHANNEL} main" -ARG VERSION="22.10.1.1877" +ARG VERSION="22.10.2.11" ARG PACKAGES="clickhouse-client clickhouse-server clickhouse-common-static" # set non-empty deb_location_url url to create a docker image diff --git a/docker/test/integration/base/Dockerfile b/docker/test/integration/base/Dockerfile index 9b6318a5426..a2d86187a23 100644 --- a/docker/test/integration/base/Dockerfile +++ b/docker/test/integration/base/Dockerfile @@ -27,9 +27,14 @@ RUN apt-get update \ tar \ tzdata \ unixodbc \ + python3-pip \ + libcurl4-openssl-dev \ + libssl-dev \ && apt-get clean \ && rm -rf /var/lib/apt/lists/* /var/cache/debconf /tmp/* +RUN pip3 install pycurl + # Architecture of the image when BuildKit/buildx is used ARG TARGETARCH diff --git a/docs/README.md b/docs/README.md index fa8b6bed85c..3ca87dc03c3 100644 --- a/docs/README.md +++ b/docs/README.md @@ -212,4 +212,4 @@ Templates: ## How to Build Documentation -You can build your documentation manually by following the instructions in [docs/tools/README.md](../docs/tools/README.md). Also, our CI runs the documentation build after the `documentation` label is added to PR. You can see the results of a build in the GitHub interface. If you have no permissions to add labels, a reviewer of your PR will add it. +You can build your documentation manually by following the instructions in the docs repo [contrib-writing-guide](https://github.com/ClickHouse/clickhouse-docs/blob/main/contrib-writing-guide.md). Also, our CI runs the documentation build after the `documentation` label is added to PR. You can see the results of a build in the GitHub interface. If you have no permissions to add labels, a reviewer of your PR will add it. diff --git a/docs/changelogs/v22.10.2.11-stable.md b/docs/changelogs/v22.10.2.11-stable.md new file mode 100644 index 00000000000..e4507f4e745 --- /dev/null +++ b/docs/changelogs/v22.10.2.11-stable.md @@ -0,0 +1,18 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.10.2.11-stable (d2bfcaba002) FIXME as compared to v22.10.1.1877-stable (98ab5a3c189) + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42750](https://github.com/ClickHouse/ClickHouse/issues/42750): A segmentation fault related to DNS & c-ares has been reported. The below error ocurred in multiple threads: ``` 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008088 [ 356 ] {} BaseDaemon: ######################################## 2022-09-28 15:41:19.008,"2022.09.28 15:41:19.008147 [ 356 ] {} BaseDaemon: (version 22.8.5.29 (official build), build id: 92504ACA0B8E2267) (from thread 353) (no query) Received signal Segmentation fault (11)" 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008196 [ 356 ] {} BaseDaemon: Address: 0xf Access: write. Address not mapped to object. 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008216 [ 356 ] {} BaseDaemon: Stack trace: 0x188f8212 0x1626851b 0x1626a69e 0x16269b3f 0x16267eab 0x13cf8284 0x13d24afc 0x13c5217e 0x14ec2495 0x15ba440f 0x15b9d13b 0x15bb2699 0x1891ccb3 0x1891e00d 0x18ae0769 0x18ade022 0x7f76aa985609 0x7f76aa8aa133 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008274 [ 356 ] {} BaseDaemon: 2. Poco::Net::IPAddress::family() const @ 0x188f8212 in /usr/bin/clickhouse 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008297 [ 356 ] {} BaseDaemon: 3. ? @ 0x1626851b in /usr/bin/clickhouse 2022-09-28 15:41:19.008,2022.09.28 15:41:19.008309 [ 356 ] {} BaseDaemon: 4. ? @ 0x1626a69e in /usr/bin/clickhouse ```. [#42234](https://github.com/ClickHouse/ClickHouse/pull/42234) ([Arthur Passos](https://github.com/arthurpassos)). +* Backported in [#42793](https://github.com/ClickHouse/ClickHouse/issues/42793): Fix a bug in ParserFunction that could have led to a segmentation fault. [#42724](https://github.com/ClickHouse/ClickHouse/pull/42724) ([Nikolay Degterinsky](https://github.com/evillique)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Always run `BuilderReport` and `BuilderSpecialReport` in all CI types [#42684](https://github.com/ClickHouse/ClickHouse/pull/42684) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). + diff --git a/docs/changelogs/v22.3.14.18-lts.md b/docs/changelogs/v22.3.14.18-lts.md new file mode 100644 index 00000000000..d0c67a2b241 --- /dev/null +++ b/docs/changelogs/v22.3.14.18-lts.md @@ -0,0 +1,26 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.3.14.18-lts (642946f61b2) FIXME as compared to v22.3.13.80-lts (e2708b01fba) + +#### Bug Fix +* Backported in [#42432](https://github.com/ClickHouse/ClickHouse/issues/42432): - Choose correct aggregation method for LowCardinality with BigInt. [#42342](https://github.com/ClickHouse/ClickHouse/pull/42342) ([Duc Canh Le](https://github.com/canhld94)). + +#### Build/Testing/Packaging Improvement +* Backported in [#42328](https://github.com/ClickHouse/ClickHouse/issues/42328): Update cctz to the latest master, update tzdb to 2020e. [#42273](https://github.com/ClickHouse/ClickHouse/pull/42273) ([Dom Del Nano](https://github.com/ddelnano)). +* Backported in [#42358](https://github.com/ClickHouse/ClickHouse/issues/42358): Update tzdata to 2022e to support the new timezone changes. Palestine transitions are now Saturdays at 02:00. Simplify three Ukraine zones into one. Jordan and Syria switch from +02/+03 with DST to year-round +03. (https://data.iana.org/time-zones/tzdb/NEWS). This closes [#42252](https://github.com/ClickHouse/ClickHouse/issues/42252). [#42327](https://github.com/ClickHouse/ClickHouse/pull/42327) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42298](https://github.com/ClickHouse/ClickHouse/issues/42298): Fix a bug with projections and the `aggregate_functions_null_for_empty` setting. This bug is very rare and appears only if you enable the `aggregate_functions_null_for_empty` setting in the server's config. This closes [#41647](https://github.com/ClickHouse/ClickHouse/issues/41647). [#42198](https://github.com/ClickHouse/ClickHouse/pull/42198) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42592](https://github.com/ClickHouse/ClickHouse/issues/42592): This closes [#42453](https://github.com/ClickHouse/ClickHouse/issues/42453). [#42573](https://github.com/ClickHouse/ClickHouse/pull/42573) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Add a warning message to release.py script, require release type [#41975](https://github.com/ClickHouse/ClickHouse/pull/41975) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). +* Revert [#27787](https://github.com/ClickHouse/ClickHouse/issues/27787) [#42136](https://github.com/ClickHouse/ClickHouse/pull/42136) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). + diff --git a/docs/changelogs/v22.3.14.23-lts.md b/docs/changelogs/v22.3.14.23-lts.md new file mode 100644 index 00000000000..663d8b43f6f --- /dev/null +++ b/docs/changelogs/v22.3.14.23-lts.md @@ -0,0 +1,29 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.3.14.23-lts (74956bfee4d) FIXME as compared to v22.3.13.80-lts (e2708b01fba) + +#### Improvement +* Backported in [#42527](https://github.com/ClickHouse/ClickHouse/issues/42527): Fix issue with passing MySQL timeouts for MySQL database engine and MySQL table function. Closes [#34168](https://github.com/ClickHouse/ClickHouse/issues/34168)?notification_referrer_id=NT_kwDOAzsV57MzMDMxNjAzNTY5OjU0MjAzODc5. [#40751](https://github.com/ClickHouse/ClickHouse/pull/40751) ([Kseniia Sumarokova](https://github.com/kssenii)). + +#### Bug Fix +* Backported in [#42432](https://github.com/ClickHouse/ClickHouse/issues/42432): - Choose correct aggregation method for LowCardinality with BigInt. [#42342](https://github.com/ClickHouse/ClickHouse/pull/42342) ([Duc Canh Le](https://github.com/canhld94)). + +#### Build/Testing/Packaging Improvement +* Backported in [#42328](https://github.com/ClickHouse/ClickHouse/issues/42328): Update cctz to the latest master, update tzdb to 2020e. [#42273](https://github.com/ClickHouse/ClickHouse/pull/42273) ([Dom Del Nano](https://github.com/ddelnano)). +* Backported in [#42358](https://github.com/ClickHouse/ClickHouse/issues/42358): Update tzdata to 2022e to support the new timezone changes. Palestine transitions are now Saturdays at 02:00. Simplify three Ukraine zones into one. Jordan and Syria switch from +02/+03 with DST to year-round +03. (https://data.iana.org/time-zones/tzdb/NEWS). This closes [#42252](https://github.com/ClickHouse/ClickHouse/issues/42252). [#42327](https://github.com/ClickHouse/ClickHouse/pull/42327) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42298](https://github.com/ClickHouse/ClickHouse/issues/42298): Fix a bug with projections and the `aggregate_functions_null_for_empty` setting. This bug is very rare and appears only if you enable the `aggregate_functions_null_for_empty` setting in the server's config. This closes [#41647](https://github.com/ClickHouse/ClickHouse/issues/41647). [#42198](https://github.com/ClickHouse/ClickHouse/pull/42198) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42592](https://github.com/ClickHouse/ClickHouse/issues/42592): This closes [#42453](https://github.com/ClickHouse/ClickHouse/issues/42453). [#42573](https://github.com/ClickHouse/ClickHouse/pull/42573) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Add a warning message to release.py script, require release type [#41975](https://github.com/ClickHouse/ClickHouse/pull/41975) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). +* Revert [#27787](https://github.com/ClickHouse/ClickHouse/issues/27787) [#42136](https://github.com/ClickHouse/ClickHouse/pull/42136) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). + diff --git a/docs/changelogs/v22.7.7.24-stable.md b/docs/changelogs/v22.7.7.24-stable.md new file mode 100644 index 00000000000..d7b83775502 --- /dev/null +++ b/docs/changelogs/v22.7.7.24-stable.md @@ -0,0 +1,29 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.7.7.24-stable (02ad1f979a8) FIXME as compared to v22.7.6.74-stable (c00ffb3c11a) + +#### Bug Fix +* Backported in [#42433](https://github.com/ClickHouse/ClickHouse/issues/42433): - Choose correct aggregation method for LowCardinality with BigInt. [#42342](https://github.com/ClickHouse/ClickHouse/pull/42342) ([Duc Canh Le](https://github.com/canhld94)). + +#### Build/Testing/Packaging Improvement +* Backported in [#42329](https://github.com/ClickHouse/ClickHouse/issues/42329): Update cctz to the latest master, update tzdb to 2020e. [#42273](https://github.com/ClickHouse/ClickHouse/pull/42273) ([Dom Del Nano](https://github.com/ddelnano)). +* Backported in [#42359](https://github.com/ClickHouse/ClickHouse/issues/42359): Update tzdata to 2022e to support the new timezone changes. Palestine transitions are now Saturdays at 02:00. Simplify three Ukraine zones into one. Jordan and Syria switch from +02/+03 with DST to year-round +03. (https://data.iana.org/time-zones/tzdb/NEWS). This closes [#42252](https://github.com/ClickHouse/ClickHouse/issues/42252). [#42327](https://github.com/ClickHouse/ClickHouse/pull/42327) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42268](https://github.com/ClickHouse/ClickHouse/issues/42268): Fix reusing of files > 4GB from base backup. [#42146](https://github.com/ClickHouse/ClickHouse/pull/42146) ([Azat Khuzhin](https://github.com/azat)). +* Backported in [#42299](https://github.com/ClickHouse/ClickHouse/issues/42299): Fix a bug with projections and the `aggregate_functions_null_for_empty` setting. This bug is very rare and appears only if you enable the `aggregate_functions_null_for_empty` setting in the server's config. This closes [#41647](https://github.com/ClickHouse/ClickHouse/issues/41647). [#42198](https://github.com/ClickHouse/ClickHouse/pull/42198) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42386](https://github.com/ClickHouse/ClickHouse/issues/42386): `ALTER UPDATE` of attached part (with columns different from table schema) could create an invalid `columns.txt` metadata on disk. Reading from such part could fail with errors or return invalid data. Fixes [#42161](https://github.com/ClickHouse/ClickHouse/issues/42161). [#42319](https://github.com/ClickHouse/ClickHouse/pull/42319) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42498](https://github.com/ClickHouse/ClickHouse/issues/42498): Setting `additional_table_filters` were not applied to `Distributed` storage. Fixes [#41692](https://github.com/ClickHouse/ClickHouse/issues/41692). [#42322](https://github.com/ClickHouse/ClickHouse/pull/42322) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42593](https://github.com/ClickHouse/ClickHouse/issues/42593): This closes [#42453](https://github.com/ClickHouse/ClickHouse/issues/42453). [#42573](https://github.com/ClickHouse/ClickHouse/pull/42573) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Add a warning message to release.py script, require release type [#41975](https://github.com/ClickHouse/ClickHouse/pull/41975) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). +* Revert [#27787](https://github.com/ClickHouse/ClickHouse/issues/27787) [#42136](https://github.com/ClickHouse/ClickHouse/pull/42136) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). + diff --git a/docs/changelogs/v22.8.7.34-lts.md b/docs/changelogs/v22.8.7.34-lts.md new file mode 100644 index 00000000000..0dc899f4717 --- /dev/null +++ b/docs/changelogs/v22.8.7.34-lts.md @@ -0,0 +1,37 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.8.7.34-lts (3c38e5e8ab9) FIXME as compared to v22.8.6.71-lts (7bf38a43e30) + +#### Improvement +* Backported in [#42096](https://github.com/ClickHouse/ClickHouse/issues/42096): Replace back `clickhouse su` command with `sudo -u` in start in order to respect limits in `/etc/security/limits.conf`. [#41847](https://github.com/ClickHouse/ClickHouse/pull/41847) ([Eugene Konkov](https://github.com/ekonkov)). + +#### Bug Fix +* Backported in [#42434](https://github.com/ClickHouse/ClickHouse/issues/42434): - Choose correct aggregation method for LowCardinality with BigInt. [#42342](https://github.com/ClickHouse/ClickHouse/pull/42342) ([Duc Canh Le](https://github.com/canhld94)). + +#### Build/Testing/Packaging Improvement +* Backported in [#42296](https://github.com/ClickHouse/ClickHouse/issues/42296): Update cctz to the latest master, update tzdb to 2020e. [#42273](https://github.com/ClickHouse/ClickHouse/pull/42273) ([Dom Del Nano](https://github.com/ddelnano)). +* Backported in [#42360](https://github.com/ClickHouse/ClickHouse/issues/42360): Update tzdata to 2022e to support the new timezone changes. Palestine transitions are now Saturdays at 02:00. Simplify three Ukraine zones into one. Jordan and Syria switch from +02/+03 with DST to year-round +03. (https://data.iana.org/time-zones/tzdb/NEWS). This closes [#42252](https://github.com/ClickHouse/ClickHouse/issues/42252). [#42327](https://github.com/ClickHouse/ClickHouse/pull/42327) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42489](https://github.com/ClickHouse/ClickHouse/issues/42489): Removed skipping of mutations in unaffected partitions of `MergeTree` tables, because this feature never worked correctly and might cause resurrection of finished mutations. [#40589](https://github.com/ClickHouse/ClickHouse/pull/40589) ([Alexander Tokmakov](https://github.com/tavplubix)). +* Backported in [#42121](https://github.com/ClickHouse/ClickHouse/issues/42121): Fixed "Part ... intersects part ..." error that might happen in extremely rare cases if replica was restarted just after detaching some part as broken. [#41741](https://github.com/ClickHouse/ClickHouse/pull/41741) ([Alexander Tokmakov](https://github.com/tavplubix)). +* - Prevent crash when passing wrong aggregation states to groupBitmap*. [#41972](https://github.com/ClickHouse/ClickHouse/pull/41972) ([Raúl Marín](https://github.com/Algunenano)). +* - Fix read bytes/rows in X-ClickHouse-Summary with materialized views. [#41973](https://github.com/ClickHouse/ClickHouse/pull/41973) ([Raúl Marín](https://github.com/Algunenano)). +* Backported in [#42269](https://github.com/ClickHouse/ClickHouse/issues/42269): Fix reusing of files > 4GB from base backup. [#42146](https://github.com/ClickHouse/ClickHouse/pull/42146) ([Azat Khuzhin](https://github.com/azat)). +* Backported in [#42300](https://github.com/ClickHouse/ClickHouse/issues/42300): Fix a bug with projections and the `aggregate_functions_null_for_empty` setting. This bug is very rare and appears only if you enable the `aggregate_functions_null_for_empty` setting in the server's config. This closes [#41647](https://github.com/ClickHouse/ClickHouse/issues/41647). [#42198](https://github.com/ClickHouse/ClickHouse/pull/42198) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42387](https://github.com/ClickHouse/ClickHouse/issues/42387): `ALTER UPDATE` of attached part (with columns different from table schema) could create an invalid `columns.txt` metadata on disk. Reading from such part could fail with errors or return invalid data. Fixes [#42161](https://github.com/ClickHouse/ClickHouse/issues/42161). [#42319](https://github.com/ClickHouse/ClickHouse/pull/42319) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42499](https://github.com/ClickHouse/ClickHouse/issues/42499): Setting `additional_table_filters` were not applied to `Distributed` storage. Fixes [#41692](https://github.com/ClickHouse/ClickHouse/issues/41692). [#42322](https://github.com/ClickHouse/ClickHouse/pull/42322) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42571](https://github.com/ClickHouse/ClickHouse/issues/42571): Fix buffer overflow in the processing of Decimal data types. This closes [#42451](https://github.com/ClickHouse/ClickHouse/issues/42451). [#42465](https://github.com/ClickHouse/ClickHouse/pull/42465) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42594](https://github.com/ClickHouse/ClickHouse/issues/42594): This closes [#42453](https://github.com/ClickHouse/ClickHouse/issues/42453). [#42573](https://github.com/ClickHouse/ClickHouse/pull/42573) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Add a warning message to release.py script, require release type [#41975](https://github.com/ClickHouse/ClickHouse/pull/41975) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). +* Revert [#27787](https://github.com/ClickHouse/ClickHouse/issues/27787) [#42136](https://github.com/ClickHouse/ClickHouse/pull/42136) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). + diff --git a/docs/changelogs/v22.8.8.3-lts.md b/docs/changelogs/v22.8.8.3-lts.md new file mode 100644 index 00000000000..deaab51fce9 --- /dev/null +++ b/docs/changelogs/v22.8.8.3-lts.md @@ -0,0 +1,13 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.8.8.3-lts (ac5a6cababc) FIXME as compared to v22.8.7.34-lts (3c38e5e8ab9) + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42677](https://github.com/ClickHouse/ClickHouse/issues/42677): keeper-fix: fix race in accessing logs while snapshot is being installed. [#40627](https://github.com/ClickHouse/ClickHouse/pull/40627) ([Antonio Andelic](https://github.com/antonio2368)). + diff --git a/docs/changelogs/v22.9.4.32-stable.md b/docs/changelogs/v22.9.4.32-stable.md new file mode 100644 index 00000000000..d6c3f4ba498 --- /dev/null +++ b/docs/changelogs/v22.9.4.32-stable.md @@ -0,0 +1,33 @@ +--- +sidebar_position: 1 +sidebar_label: 2022 +--- + +# 2022 Changelog + +### ClickHouse release v22.9.4.32-stable (3db8bcf1a70) FIXME as compared to v22.9.3.18-stable (0cb4b15d2fa) + +#### Bug Fix +* Backported in [#42435](https://github.com/ClickHouse/ClickHouse/issues/42435): - Choose correct aggregation method for LowCardinality with BigInt. [#42342](https://github.com/ClickHouse/ClickHouse/pull/42342) ([Duc Canh Le](https://github.com/canhld94)). + +#### Build/Testing/Packaging Improvement +* Backported in [#42297](https://github.com/ClickHouse/ClickHouse/issues/42297): Update cctz to the latest master, update tzdb to 2020e. [#42273](https://github.com/ClickHouse/ClickHouse/pull/42273) ([Dom Del Nano](https://github.com/ddelnano)). +* Backported in [#42361](https://github.com/ClickHouse/ClickHouse/issues/42361): Update tzdata to 2022e to support the new timezone changes. Palestine transitions are now Saturdays at 02:00. Simplify three Ukraine zones into one. Jordan and Syria switch from +02/+03 with DST to year-round +03. (https://data.iana.org/time-zones/tzdb/NEWS). This closes [#42252](https://github.com/ClickHouse/ClickHouse/issues/42252). [#42327](https://github.com/ClickHouse/ClickHouse/pull/42327) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### Bug Fix (user-visible misbehavior in official stable or prestable release) + +* Backported in [#42122](https://github.com/ClickHouse/ClickHouse/issues/42122): Fixed "Part ... intersects part ..." error that might happen in extremely rare cases if replica was restarted just after detaching some part as broken. [#41741](https://github.com/ClickHouse/ClickHouse/pull/41741) ([Alexander Tokmakov](https://github.com/tavplubix)). +* Backported in [#41938](https://github.com/ClickHouse/ClickHouse/issues/41938): Don't allow to create or alter merge tree tables with virtual column name _row_exists, which is reserved for lightweight delete. Fixed [#41716](https://github.com/ClickHouse/ClickHouse/issues/41716). [#41763](https://github.com/ClickHouse/ClickHouse/pull/41763) ([Jianmei Zhang](https://github.com/zhangjmruc)). +* Backported in [#42179](https://github.com/ClickHouse/ClickHouse/issues/42179): Fix reusing of files > 4GB from base backup. [#42146](https://github.com/ClickHouse/ClickHouse/pull/42146) ([Azat Khuzhin](https://github.com/azat)). +* Backported in [#42301](https://github.com/ClickHouse/ClickHouse/issues/42301): Fix a bug with projections and the `aggregate_functions_null_for_empty` setting. This bug is very rare and appears only if you enable the `aggregate_functions_null_for_empty` setting in the server's config. This closes [#41647](https://github.com/ClickHouse/ClickHouse/issues/41647). [#42198](https://github.com/ClickHouse/ClickHouse/pull/42198) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42388](https://github.com/ClickHouse/ClickHouse/issues/42388): `ALTER UPDATE` of attached part (with columns different from table schema) could create an invalid `columns.txt` metadata on disk. Reading from such part could fail with errors or return invalid data. Fixes [#42161](https://github.com/ClickHouse/ClickHouse/issues/42161). [#42319](https://github.com/ClickHouse/ClickHouse/pull/42319) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42500](https://github.com/ClickHouse/ClickHouse/issues/42500): Setting `additional_table_filters` were not applied to `Distributed` storage. Fixes [#41692](https://github.com/ClickHouse/ClickHouse/issues/41692). [#42322](https://github.com/ClickHouse/ClickHouse/pull/42322) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). +* Backported in [#42581](https://github.com/ClickHouse/ClickHouse/issues/42581): This reverts [#40217](https://github.com/ClickHouse/ClickHouse/issues/40217) which introduced a regression in date/time functions. [#42367](https://github.com/ClickHouse/ClickHouse/pull/42367) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42572](https://github.com/ClickHouse/ClickHouse/issues/42572): Fix buffer overflow in the processing of Decimal data types. This closes [#42451](https://github.com/ClickHouse/ClickHouse/issues/42451). [#42465](https://github.com/ClickHouse/ClickHouse/pull/42465) ([Alexey Milovidov](https://github.com/alexey-milovidov)). +* Backported in [#42595](https://github.com/ClickHouse/ClickHouse/issues/42595): This closes [#42453](https://github.com/ClickHouse/ClickHouse/issues/42453). [#42573](https://github.com/ClickHouse/ClickHouse/pull/42573) ([Alexey Milovidov](https://github.com/alexey-milovidov)). + +#### NOT FOR CHANGELOG / INSIGNIFICANT + +* Add a warning message to release.py script, require release type [#41975](https://github.com/ClickHouse/ClickHouse/pull/41975) ([Mikhail f. Shiryaev](https://github.com/Felixoid)). +* Revert [#27787](https://github.com/ClickHouse/ClickHouse/issues/27787) [#42136](https://github.com/ClickHouse/ClickHouse/pull/42136) ([Nikolai Kochetov](https://github.com/KochetovNicolai)). + diff --git a/docs/en/development/build.md b/docs/en/development/build.md index f397dc0d037..8982a3bc0a4 100644 --- a/docs/en/development/build.md +++ b/docs/en/development/build.md @@ -105,7 +105,7 @@ ninja Example for Fedora Rawhide: ``` bash sudo yum update -yum --nogpg install git cmake make clang-c++ python3 +sudo yum --nogpg install git cmake make clang python3 ccache git clone --recursive https://github.com/ClickHouse/ClickHouse.git mkdir build && cd build cmake ../ClickHouse diff --git a/docs/en/engines/table-engines/integrations/s3.md b/docs/en/engines/table-engines/integrations/s3.md index 986a29b8307..db983ab9c68 100644 --- a/docs/en/engines/table-engines/integrations/s3.md +++ b/docs/en/engines/table-engines/integrations/s3.md @@ -139,7 +139,7 @@ The following settings can be specified in configuration file for given endpoint - `use_environment_credentials` — If set to `true`, S3 client will try to obtain credentials from environment variables and [Amazon EC2](https://en.wikipedia.org/wiki/Amazon_Elastic_Compute_Cloud) metadata for given endpoint. Optional, default value is `false`. - `region` — Specifies S3 region name. Optional. - `use_insecure_imds_request` — If set to `true`, S3 client will use insecure IMDS request while obtaining credentials from Amazon EC2 metadata. Optional, default value is `false`. -- `header` — Adds specified HTTP header to a request to given endpoint. Optional, can be speficied multiple times. +- `header` — Adds specified HTTP header to a request to given endpoint. Optional, can be specified multiple times. - `server_side_encryption_customer_key_base64` — If specified, required headers for accessing S3 objects with SSE-C encryption will be set. Optional. - `max_single_read_retries` — The maximum number of attempts during single read. Default value is `4`. Optional. diff --git a/docs/en/getting-started/example-datasets/nyc-taxi.md b/docs/en/getting-started/example-datasets/nyc-taxi.md index e24fb4b01a7..69098f63037 100644 --- a/docs/en/getting-started/example-datasets/nyc-taxi.md +++ b/docs/en/getting-started/example-datasets/nyc-taxi.md @@ -33,7 +33,7 @@ CREATE TABLE trips ( tip_amount Float32, tolls_amount Float32, total_amount Float32, - payment_type Enum('CSH' = 1, 'CRE' = 2, 'NOC' = 3, 'DIS' = 4), + payment_type Enum('CSH' = 1, 'CRE' = 2, 'NOC' = 3, 'DIS' = 4, 'UNK' = 5), pickup_ntaname LowCardinality(String), dropoff_ntaname LowCardinality(String) ) @@ -63,7 +63,7 @@ SELECT payment_type, pickup_ntaname, dropoff_ntaname -FROM url( +FROM s3( 'https://datasets-documentation.s3.eu-west-3.amazonaws.com/nyc-taxi/trips_{0..2}.gz', 'TabSeparatedWithNames' ) diff --git a/docs/en/getting-started/install.md b/docs/en/getting-started/install.md index 61303eddab9..e88e9e06a68 100644 --- a/docs/en/getting-started/install.md +++ b/docs/en/getting-started/install.md @@ -128,6 +128,24 @@ clickhouse-client # or "clickhouse-client --password" if you set up a password. +
+Migration Method for installing the deb-packages + +```bash +sudo apt-key del E0C56BD4 +sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 8919F6BD2B48D754 +echo "deb https://packages.clickhouse.com/deb stable main" | sudo tee \ + /etc/apt/sources.list.d/clickhouse.list +sudo apt-get update + +sudo apt-get install -y clickhouse-server clickhouse-client + +sudo service clickhouse-server start +clickhouse-client # or "clickhouse-client --password" if you set up a password. +``` + +
+ You can replace `stable` with `lts` to use different [release kinds](/docs/en/faq/operations/production.md) based on your needs. You can also download and install packages manually from [here](https://packages.clickhouse.com/deb/pool/main/c/). diff --git a/docs/en/operations/backup.md b/docs/en/operations/_backup.md similarity index 67% rename from docs/en/operations/backup.md rename to docs/en/operations/_backup.md index a755e3ef9a6..d694c51cee6 100644 --- a/docs/en/operations/backup.md +++ b/docs/en/operations/_backup.md @@ -1,9 +1,12 @@ ---- -slug: /en/operations/backup -sidebar_position: 49 -sidebar_label: Data backup and restore -title: Data backup and restore ---- + +[//]: # (This file is included in Manage > Backups) + +- [Backup to a local disk](#backup-to-a-local-disk) +- [Configuring backup/restore to use an S3 endpoint](#configuring-backuprestore-to-use-an-s3-endpoint) +- [Backup/restore using an S3 disk](#backuprestore-using-an-s3-disk) +- [Alternatives](#alternatives) + +## Background While [replication](../engines/table-engines/mergetree-family/replication.md) provides protection from hardware failures, it does not protect against human errors: accidental deletion of data, deletion of the wrong table or a table on the wrong cluster, and software bugs that result in incorrect data processing or data corruption. In many cases mistakes like these will affect all replicas. ClickHouse has built-in safeguards to prevent some types of mistakes — for example, by default [you can’t just drop tables with a MergeTree-like engine containing more than 50 Gb of data](server-configuration-parameters/settings.md#max-table-size-to-drop). However, these safeguards do not cover all possible cases and can be circumvented. @@ -15,7 +18,9 @@ Each company has different resources available and business requirements, so the Keep in mind that if you backed something up and never tried to restore it, chances are that restore will not work properly when you actually need it (or at least it will take longer than business can tolerate). So whatever backup approach you choose, make sure to automate the restore process as well, and practice it on a spare ClickHouse cluster regularly. ::: -## Configure a backup destination +## Backup to a local disk + +### Configure a backup destination In the examples below you will see the backup destination specified like `Disk('backups', '1.zip')`. To prepare the destination add a file to `/etc/clickhouse-server/config.d/backup_disk.xml` specifying the backup destination. For example, this file defines disk named `backups` and then adds that disk to the **backups > allowed_disk** list: @@ -39,7 +44,7 @@ In the examples below you will see the backup destination specified like `Disk(' ``` -## Parameters +### Parameters Backups can be either full or incremental, and can include tables (including materialized views, projections, and dictionaries), and databases. Backups can be synchronous (default) or asynchronous. They can be compressed. Backups can be password protected. @@ -52,7 +57,7 @@ The BACKUP and RESTORE statements take a list of DATABASE and TABLE names, a des - `password` for the file on disk - `base_backup`: the destination of the previous backup of this source. For example, `Disk('backups', '1.zip')` -## Usage examples +### Usage examples Backup and then restore a table: ``` @@ -81,7 +86,7 @@ RESTORE TABLE test.table AS test.table2 FROM Disk('backups', '1.zip') BACKUP TABLE test.table3 AS test.table4 TO Disk('backups', '2.zip') ``` -## Incremental backups +### Incremental backups Incremental backups can be taken by specifying the `base_backup`. :::note @@ -100,7 +105,7 @@ RESTORE TABLE test.table AS test.table2 FROM Disk('backups', 'incremental-a.zip'); ``` -## Assign a password to the backup +### Assign a password to the backup Backups written to disk can have a password applied to the file: ``` @@ -116,7 +121,7 @@ RESTORE TABLE test.table SETTINGS password='qwerty' ``` -## Compression settings +### Compression settings If you would like to specify the compression method or level: ``` @@ -125,14 +130,14 @@ BACKUP TABLE test.table SETTINGS compression_method='lzma', compression_level=3 ``` -## Restore specific partitions +### Restore specific partitions If specific partitions associated with a table need to be restored these can be specified. To restore partitions 1 and 4 from backup: ``` RESTORE TABLE test.table PARTITIONS '2', '3' FROM Disk('backups', 'filename.zip') ``` -## Check the status of backups +### Check the status of backups The backup command returns an `id` and `status`, and that `id` can be used to get the status of the backup. This is very useful to check the progress of long ASYNC backups. The example below shows a failure that happened when trying to overwrite an existing backup file: ```sql @@ -171,13 +176,118 @@ end_time: 2022-08-30 09:21:46 1 row in set. Elapsed: 0.002 sec. ``` -## Backup to S3 +## Configuring BACKUP/RESTORE to use an S3 Endpoint -It is possible to `BACKUP`/`RESTORE` to S3, but this disk should be configured -in a proper way, since by default you will need to backup metadata from local -disk to make backup full. +To write backups to an S3 bucket you need three pieces of information: +- S3 endpoint, + for example `https://mars-doc-test.s3.amazonaws.com/backup-S3/` +- Access key ID, + for example `ABC123` +- Secret access key, + for example `Abc+123` -First of all, you need to configure S3 disk in a special way: +:::note +Creating an S3 bucket is covered in [Use S3 Object Storage as a ClickHouse disk](/docs/en/integrations/data-ingestion/s3/configuring-s3-for-clickhouse-use.md), just come back to this doc after saving the policy, there is no need to configure ClickHouse to use the S3 bucket. +::: + +The destination for a backup will be specified like this: +``` +S3('/', '', ') +``` + +```sql +CREATE TABLE data +( + `key` Int, + `value` String, + `array` Array(String) +) +ENGINE = MergeTree +ORDER BY tuple() +``` + +```sql +INSERT INTO data SELECT * +FROM generateRandom('key Int, value String, array Array(String)') +LIMIT 1000 +``` + +### Create a base (initial) backup + +Incremental backups require a _base_ backup to start from, this example will be used +later as the base backup. The first parameter of the S3 destination is the S3 endpoint followed by the directory within the bucket to use for this backup. In this example the directory is named `my_backup`. + +```sql +BACKUP TABLE data TO S3('https://mars-doc-test.s3.amazonaws.com/backup-S3/my_backup', 'ABC123', 'Abc+123') +``` + +```response +┌─id───────────────────────────────────┬─status─────────┐ +│ de442b75-a66c-4a3c-a193-f76f278c70f3 │ BACKUP_CREATED │ +└──────────────────────────────────────┴────────────────┘ +``` + +### Add more data + +Incremental backups are populated with the difference between the base backup and the current content of the table being backed up. Add more data before taking the incremental backup: + +```sql +INSERT INTO data SELECT * +FROM generateRandom('key Int, value String, array Array(String)') +LIMIT 100 +``` +### Take an incremental backup + +This backup command is similar to the base backup, but adds `SETTINGS base_backup` and the location of the base backup. Note that the destination for the incremental backup is not the same directory as the base, it is the same endpoint with a different target directory within the bucket. The base backup is in `my_backup`, and the incremental will be written to `my_incremental`: +```sql +BACKUP TABLE data TO S3('https://mars-doc-test.s3.amazonaws.com/backup-S3/my_incremental', 'ABC123', 'Abc+123') SETTINGS base_backup = S3('https://mars-doc-test.s3.amazonaws.com/backup-S3/my_backup', 'ABC123', 'Abc+123') +``` + +```response +┌─id───────────────────────────────────┬─status─────────┐ +│ f6cd3900-850f-41c9-94f1-0c4df33ea528 │ BACKUP_CREATED │ +└──────────────────────────────────────┴────────────────┘ +``` +### Restore from the incremental backup + +This command restores the incremental backup into a new table, `data3`. Note that when an incremental backup is restored, the base backup is also included. Specify only the incremental backup when restoring: +```sql +RESTORE TABLE data AS data3 FROM S3('https://mars-doc-test.s3.amazonaws.com/backup-S3/my_incremental', 'ABC123', 'Abc+123') +``` + +```response +┌─id───────────────────────────────────┬─status───┐ +│ ff0c8c39-7dff-4324-a241-000796de11ca │ RESTORED │ +└──────────────────────────────────────┴──────────┘ +``` + +### Verify the count + +There were two inserts into the original table `data`, one with 1,000 rows and one with 100 rows, for a total of 1,100. Verify that the restored table has 1,100 rows: +```sql +SELECT count() +FROM data3 +``` +```response +┌─count()─┐ +│ 1100 │ +└─────────┘ +``` + +### Verify the content +This compares the content of the original table, `data` with the restored table `data3`: +```sql +SELECT throwIf(( + SELECT groupArray(tuple(*)) + FROM data + ) != ( + SELECT groupArray(tuple(*)) + FROM data3 + ), 'Data does not match after BACKUP/RESTORE') +``` +## BACKUP/RESTORE Using an S3 Disk + +It is also possible to `BACKUP`/`RESTORE` to S3 by configuring an S3 disk in the ClickHouse storage configuration. Configure the disk like this by adding a file to `/etc/clickhouse-server/config.d`: ```xml diff --git a/docs/en/operations/update.md b/docs/en/operations/_update.md similarity index 88% rename from docs/en/operations/update.md rename to docs/en/operations/_update.md index 24f7efecc7b..86981da2be6 100644 --- a/docs/en/operations/update.md +++ b/docs/en/operations/_update.md @@ -1,10 +1,7 @@ ---- -slug: /en/operations/update -sidebar_position: 47 -sidebar_label: ClickHouse Upgrade ---- -# ClickHouse Upgrade +[//]: # (This file is included in Manage > Updates) + +## Self-managed ClickHouse Upgrade If ClickHouse was installed from `deb` packages, execute the following commands on the server: diff --git a/docs/en/operations/clickhouse-keeper.md b/docs/en/operations/clickhouse-keeper.md index 82fa5c114ea..aad20da0010 100644 --- a/docs/en/operations/clickhouse-keeper.md +++ b/docs/en/operations/clickhouse-keeper.md @@ -309,7 +309,7 @@ Sessions with Ephemerals (1): /clickhouse/task_queue/ddl ``` -## [experimental] Migration from ZooKeeper {#migration-from-zookeeper} +## Migration from ZooKeeper {#migration-from-zookeeper} Seamlessly migration from ZooKeeper to ClickHouse Keeper is impossible you have to stop your ZooKeeper cluster, convert data and start ClickHouse Keeper. `clickhouse-keeper-converter` tool allows converting ZooKeeper logs and snapshots to ClickHouse Keeper snapshot. It works only with ZooKeeper > 3.4. Steps for migration: diff --git a/docs/en/operations/system-tables/information_schema.md b/docs/en/operations/system-tables/information_schema.md index a573491282a..a8e516f02a3 100644 --- a/docs/en/operations/system-tables/information_schema.md +++ b/docs/en/operations/system-tables/information_schema.md @@ -178,7 +178,7 @@ Columns: - `view_definition` ([String](../../sql-reference/data-types/string.md)) — `SELECT` query for view. - `check_option` ([String](../../sql-reference/data-types/string.md)) — `NONE`, no checking. - `is_updatable` ([Enum8](../../sql-reference/data-types/enum.md)) — `NO`, the view is not updated. -- `is_insertable_into` ([Enum8](../../sql-reference/data-types/enum.md)) — Shows whether the created view is [materialized](../../sql-reference/statements/create/view/#materialized). Possible values: +- `is_insertable_into` ([Enum8](../../sql-reference/data-types/enum.md)) — Shows whether the created view is [materialized](../../sql-reference/statements/create/view.md/#materialized-view). Possible values: - `NO` — The created view is not materialized. - `YES` — The created view is materialized. - `is_trigger_updatable` ([Enum8](../../sql-reference/data-types/enum.md)) — `NO`, the trigger is not updated. diff --git a/docs/en/operations/system-tables/replicated_fetches.md b/docs/en/operations/system-tables/replicated_fetches.md index 3536bbaff4d..74888fd2f13 100644 --- a/docs/en/operations/system-tables/replicated_fetches.md +++ b/docs/en/operations/system-tables/replicated_fetches.md @@ -68,6 +68,5 @@ thread_id: 54 **See Also** -- [Managing ReplicatedMergeTree Tables](../../sql-reference/statements/system/#query-language-system-replicated) +- [Managing ReplicatedMergeTree Tables](../../sql-reference/statements/system.md/#managing-replicatedmergetree-tables) -[Original article](https://clickhouse.com/docs/en/operations/system_tables/replicated_fetches) diff --git a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md index 198ff12f1d6..02a4ad57a3b 100644 --- a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md +++ b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-layout.md @@ -303,17 +303,25 @@ or CREATE DICTIONARY somedict ( id UInt64, first Date, - last Date + last Date, + advertiser_id UInt64 ) PRIMARY KEY id +SOURCE(CLICKHOUSE(TABLE 'date_table')) +LIFETIME(MIN 1 MAX 1000) LAYOUT(RANGE_HASHED()) RANGE(MIN first MAX last) ``` -To work with these dictionaries, you need to pass an additional argument to the `dictGetT` function, for which a range is selected: +To work with these dictionaries, you need to pass an additional argument to the `dictGet` function, for which a range is selected: ``` sql -dictGetT('dict_name', 'attr_name', id, date) +dictGet('dict_name', 'attr_name', id, date) +``` +Query example: + +``` sql +SELECT dictGet('somedict', 'advertiser_id', 1, '2022-10-20 23:20:10.000'::DateTime64::UInt64); ``` This function returns the value for the specified `id`s and the date range that includes the passed date. diff --git a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-polygon.md b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-polygon.md index 912af5b5bce..e5ee48c9166 100644 --- a/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-polygon.md +++ b/docs/en/sql-reference/dictionaries/external-dictionaries/external-dicts-dict-polygon.md @@ -14,8 +14,10 @@ Example of a polygon dictionary configuration: - key - Array(Array(Array(Array(Float64)))) + + key + Array(Array(Array(Array(Float64)))) + diff --git a/docs/en/sql-reference/functions/date-time-functions.md b/docs/en/sql-reference/functions/date-time-functions.md index 15fc9ef0c89..f7ea2690b21 100644 --- a/docs/en/sql-reference/functions/date-time-functions.md +++ b/docs/en/sql-reference/functions/date-time-functions.md @@ -1068,7 +1068,7 @@ Example: SELECT timeSlots(toDateTime('2012-01-01 12:20:00'), toUInt32(600)); SELECT timeSlots(toDateTime('1980-12-12 21:01:02', 'UTC'), toUInt32(600), 299); SELECT timeSlots(toDateTime64('1980-12-12 21:01:02.1234', 4, 'UTC'), toDecimal64(600.1, 1), toDecimal64(299, 0)); -``` +``` ``` text ┌─timeSlots(toDateTime('2012-01-01 12:20:00'), toUInt32(600))─┐ │ ['2012-01-01 12:00:00','2012-01-01 12:30:00'] │ @@ -1244,7 +1244,7 @@ Result: └──────────────────────────┘ ``` -When there are two arguments: first is an [Integer](../../sql-reference/data-types/int-uint.md) or [DateTime](../../sql-reference/data-types/datetime.md), second is a constant format string — it acts in the same way as [formatDateTime](#formatdatetime) and return [String](../../sql-reference/data-types/string.md#string) type. +When there are two or three arguments, the first an [Integer](../../sql-reference/data-types/int-uint.md), [Date](../../sql-reference/data-types/date.md), [Date32](../../sql-reference/data-types/date32.md), [DateTime](../../sql-reference/data-types/datetime.md) or [DateTime64](../../sql-reference/data-types/datetime64.md), the second a constant format string and the third an optional constant time zone string — it acts in the same way as [formatDateTime](#formatdatetime) and return [String](../../sql-reference/data-types/string.md#string) type. For example: diff --git a/docs/en/sql-reference/functions/string-functions.md b/docs/en/sql-reference/functions/string-functions.md index a8ba4843279..e0418a81f14 100644 --- a/docs/en/sql-reference/functions/string-functions.md +++ b/docs/en/sql-reference/functions/string-functions.md @@ -571,13 +571,13 @@ Similar to base58Decode, but returns an empty string in case of error. ## base64Encode(s) -Encodes ‘s’ string into base64 +Encodes ‘s’ FixedString or String into base64. Alias: `TO_BASE64`. ## base64Decode(s) -Decode base64-encoded string ‘s’ into original string. In case of failure raises an exception. +Decode base64-encoded FixedString or String ‘s’ into original string. In case of failure raises an exception. Alias: `FROM_BASE64`. diff --git a/docs/en/sql-reference/statements/alter/column.md b/docs/en/sql-reference/statements/alter/column.md index 067a350dca7..cc278465437 100644 --- a/docs/en/sql-reference/statements/alter/column.md +++ b/docs/en/sql-reference/statements/alter/column.md @@ -107,7 +107,7 @@ ALTER TABLE visits RENAME COLUMN webBrowser TO browser CLEAR COLUMN [IF EXISTS] name IN PARTITION partition_name ``` -Resets all data in a column for a specified partition. Read more about setting the partition name in the section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Resets all data in a column for a specified partition. Read more about setting the partition name in the section [How to set the partition expression](partition.md#how-to-set-partition-expression). If the `IF EXISTS` clause is specified, the query won’t return an error if the column does not exist. @@ -204,8 +204,9 @@ It is used if it is necessary to add or update a column with a complicated expre Syntax: ```sql -ALTER TABLE table MATERIALIZE COLUMN col; +ALTER TABLE [db.]table [ON CLUSTER cluster] MATERIALIZE COLUMN col [IN PARTITION partition | IN PARTITION ID 'partition_id']; ``` +- If you specify a PARTITION, a column will be materialized with only the specified partition. **Example** diff --git a/docs/en/sql-reference/statements/alter/partition.md b/docs/en/sql-reference/statements/alter/partition.md index a216de85cfc..2d89c1d5d18 100644 --- a/docs/en/sql-reference/statements/alter/partition.md +++ b/docs/en/sql-reference/statements/alter/partition.md @@ -39,7 +39,7 @@ ALTER TABLE mt DETACH PARTITION '2020-11-21'; ALTER TABLE mt DETACH PART 'all_2_2_0'; ``` -Read about setting the partition expression in a section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Read about setting the partition expression in a section [How to set the partition expression](#how-to-set-partition-expression). After the query is executed, you can do whatever you want with the data in the `detached` directory — delete it from the file system, or just leave it. @@ -53,7 +53,7 @@ ALTER TABLE table_name [ON CLUSTER cluster] DROP PARTITION|PART partition_expr Deletes the specified partition from the table. This query tags the partition as inactive and deletes data completely, approximately in 10 minutes. -Read about setting the partition expression in a section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Read about setting the partition expression in a section [How to set the partition expression](#how-to-set-partition-expression). The query is replicated – it deletes data on all replicas. @@ -71,7 +71,7 @@ ALTER TABLE table_name [ON CLUSTER cluster] DROP DETACHED PARTITION|PART partiti ``` Removes the specified part or all parts of the specified partition from `detached`. -Read more about setting the partition expression in a section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Read more about setting the partition expression in a section [How to set the partition expression](#how-to-set-partition-expression). ## ATTACH PARTITION\|PART @@ -86,7 +86,7 @@ ALTER TABLE visits ATTACH PARTITION 201901; ALTER TABLE visits ATTACH PART 201901_2_2_0; ``` -Read more about setting the partition expression in a section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Read more about setting the partition expression in a section [How to set the partition expression](#how-to-set-partition-expression). This query is replicated. The replica-initiator checks whether there is data in the `detached` directory. If data exists, the query checks its integrity. If everything is correct, the query adds the data to the table. @@ -166,7 +166,7 @@ This query creates a local backup of a specified partition. If the `PARTITION` c The entire backup process is performed without stopping the server. ::: -Note that for old-styled tables you can specify the prefix of the partition name (for example, `2019`) - then the query creates the backup for all the corresponding partitions. Read about setting the partition expression in a section [How to specify the partition expression](#alter-how-to-specify-part-expr). +Note that for old-styled tables you can specify the prefix of the partition name (for example, `2019`) - then the query creates the backup for all the corresponding partitions. Read about setting the partition expression in a section [How to set the partition expression](#how-to-set-partition-expression). At the time of execution, for a data snapshot, the query creates hardlinks to a table data. Hardlinks are placed in the directory `/var/lib/clickhouse/shadow/N/...`, where: @@ -194,7 +194,7 @@ To restore data from a backup, do the following: Restoring from a backup does not require stopping the server. -For more information about backups and restoring data, see the [Data Backup](../../../operations/backup.md) section. +For more information about backups and restoring data, see the [Data Backup](/docs/en/manage/backups.mdx) section. ## UNFREEZE PARTITION diff --git a/docs/en/sql-reference/statements/alter/projection.md b/docs/en/sql-reference/statements/alter/projection.md index a1981b4a0bb..ff8ecf3a77f 100644 --- a/docs/en/sql-reference/statements/alter/projection.md +++ b/docs/en/sql-reference/statements/alter/projection.md @@ -7,18 +7,26 @@ title: "Manipulating Projections" The following operations with [projections](../../../engines/table-engines/mergetree-family/mergetree.md#projections) are available: -- `ALTER TABLE [db].name ADD PROJECTION name ( SELECT [GROUP BY] [ORDER BY] )` - Adds projection description to tables metadata. +## ADD PROJECTION -- `ALTER TABLE [db].name DROP PROJECTION name` - Removes projection description from tables metadata and deletes projection files from disk. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). +`ALTER TABLE [db].name ADD PROJECTION name ( SELECT [GROUP BY] [ORDER BY] )` - Adds projection description to tables metadata. -- `ALTER TABLE [db.]table MATERIALIZE PROJECTION name IN PARTITION partition_name` - The query rebuilds the projection `name` in the partition `partition_name`. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). +## DROP PROJECTION -- `ALTER TABLE [db.]table CLEAR PROJECTION name IN PARTITION partition_name` - Deletes projection files from disk without removing description. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). +`ALTER TABLE [db].name DROP PROJECTION name` - Removes projection description from tables metadata and deletes projection files from disk. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). + +## MATERIALIZE PROJECTION + +`ALTER TABLE [db.]table MATERIALIZE PROJECTION name IN PARTITION partition_name` - The query rebuilds the projection `name` in the partition `partition_name`. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). + +## CLEAR PROJECTION + +`ALTER TABLE [db.]table CLEAR PROJECTION name IN PARTITION partition_name` - Deletes projection files from disk without removing description. Implemented as a [mutation](../../../sql-reference/statements/alter/index.md#mutations). The commands `ADD`, `DROP` and `CLEAR` are lightweight in a sense that they only change metadata or remove files. -Also, they are replicated, syncing projections metadata via ZooKeeper. +Also, they are replicated, syncing projections metadata via ClickHouse Keeper or ZooKeeper. :::note Projection manipulation is supported only for tables with [`*MergeTree`](../../../engines/table-engines/mergetree-family/mergetree.md) engine (including [replicated](../../../engines/table-engines/mergetree-family/replication.md) variants). diff --git a/docs/en/sql-reference/statements/create/database.md b/docs/en/sql-reference/statements/create/database.md index 432f5975cc8..7954d1362f1 100644 --- a/docs/en/sql-reference/statements/create/database.md +++ b/docs/en/sql-reference/statements/create/database.md @@ -31,7 +31,7 @@ By default, ClickHouse uses its own [Atomic](../../../engines/database-engines/a ### COMMENT -You can add a comment to the database when you creating it. +You can add a comment to the database when you are creating it. The comment is supported for all database engines. diff --git a/docs/en/sql-reference/statements/create/function.md b/docs/en/sql-reference/statements/create/function.md index 63c006b1e3e..90be007bf43 100644 --- a/docs/en/sql-reference/statements/create/function.md +++ b/docs/en/sql-reference/statements/create/function.md @@ -4,7 +4,7 @@ sidebar_position: 38 sidebar_label: FUNCTION --- -# CREATE FUNCTION +# CREATE FUNCTION — user defined function (UDF) Creates a user defined function from a lambda expression. The expression must consist of function parameters, constants, operators, or other function calls. diff --git a/docs/en/sql-reference/statements/optimize.md b/docs/en/sql-reference/statements/optimize.md index 680ff773992..036d3f0599a 100644 --- a/docs/en/sql-reference/statements/optimize.md +++ b/docs/en/sql-reference/statements/optimize.md @@ -22,7 +22,7 @@ The `OPTIMIZE` query is supported for [MergeTree](../../engines/table-engines/me When `OPTIMIZE` is used with the [ReplicatedMergeTree](../../engines/table-engines/mergetree-family/replication.md) family of table engines, ClickHouse creates a task for merging and waits for execution on all replicas (if the [replication_alter_partitions_sync](../../operations/settings/settings.md#replication-alter-partitions-sync) setting is set to `2`) or on current replica (if the [replication_alter_partitions_sync](../../operations/settings/settings.md#replication-alter-partitions-sync) setting is set to `1`). - If `OPTIMIZE` does not perform a merge for any reason, it does not notify the client. To enable notifications, use the [optimize_throw_if_noop](../../operations/settings/settings.md#setting-optimize_throw_if_noop) setting. -- If you specify a `PARTITION`, only the specified partition is optimized. [How to set partition expression](../../sql-reference/statements/alter/index.md#alter-how-to-specify-part-expr). +- If you specify a `PARTITION`, only the specified partition is optimized. [How to set partition expression](alter/partition.md#how-to-set-partition-expression). - If you specify `FINAL`, optimization is performed even when all the data is already in one part. Also merge is forced even if concurrent merges are performed. - If you specify `DEDUPLICATE`, then completely identical rows (unless by-clause is specified) will be deduplicated (all columns are compared), it makes sense only for the MergeTree engine. diff --git a/docs/en/sql-reference/statements/select/intersect.md b/docs/en/sql-reference/statements/select/intersect.md index d3b2b51b6be..f1eb4738543 100644 --- a/docs/en/sql-reference/statements/select/intersect.md +++ b/docs/en/sql-reference/statements/select/intersect.md @@ -7,7 +7,7 @@ sidebar_label: INTERSECT The `INTERSECT` clause returns only those rows that result from both the first and the second queries. The queries must match the number of columns, order, and type. The result of `INTERSECT` can contain duplicate rows. -Multiple `INTERSECT` statements are executes left to right if parenthesis are not specified. The `INTERSECT` operator has a higher priority than the `UNION` and `EXCEPT` clause. +Multiple `INTERSECT` statements are executed left to right if parentheses are not specified. The `INTERSECT` operator has a higher priority than the `UNION` and `EXCEPT` clauses. ``` sql diff --git a/docs/ru/engines/table-engines/integrations/kafka.md b/docs/ru/engines/table-engines/integrations/kafka.md index 37fc902e777..a5f091e1b23 100644 --- a/docs/ru/engines/table-engines/integrations/kafka.md +++ b/docs/ru/engines/table-engines/integrations/kafka.md @@ -87,14 +87,15 @@ SETTINGS Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот метод в новых проектах. По возможности переключите старые проекты на метод, описанный выше. +:::note "Attention" +Не используйте этот метод в новых проектах. По возможности переключите старые проекты на метод, описанный выше. +::: ``` sql Kafka(kafka_broker_list, kafka_topic_list, kafka_group_name, kafka_format [, kafka_row_delimiter, kafka_schema, kafka_num_consumers, kafka_skip_broken_messages]) ``` - ::: + ## Описание {#opisanie} diff --git a/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md index aa16113192e..86a275767a0 100644 --- a/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/aggregatingmergetree.md @@ -39,9 +39,10 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. - ::: +:::note "Attention" +Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. +::: + ``` sql CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] ( diff --git a/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md index ecaaa6b8417..72b4725c6ed 100644 --- a/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/collapsingmergetree.md @@ -43,9 +43,10 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. - ::: +:::note "Attention" +Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. +::: + ``` sql CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] ( @@ -59,7 +60,7 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] - `sign` — Имя столбца с типом строки: `1` — строка состояния, `-1` — строка отмены состояния. - Тип данных столбца — `Int8`. + Тип данных столбца — `Int8`. diff --git a/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md b/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md index 818f85f7e37..324a3fd1633 100644 --- a/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/graphitemergetree.md @@ -55,9 +55,10 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. - ::: +:::note "Attention" +Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. +::: + ``` sql CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] ( diff --git a/docs/ru/engines/table-engines/mergetree-family/mergetree.md b/docs/ru/engines/table-engines/mergetree-family/mergetree.md index e01e0006b87..f024d5f1985 100644 --- a/docs/ru/engines/table-engines/mergetree-family/mergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/mergetree.md @@ -115,9 +115,10 @@ ENGINE MergeTree() PARTITION BY toYYYYMM(EventDate) ORDER BY (CounterID, EventDa Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ, описанный выше. - ::: +:::note "Attention" +Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ, описанный выше. +::: + ``` sql CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] ( diff --git a/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md b/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md index 0d9d268fa46..7b69927e161 100644 --- a/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md +++ b/docs/ru/engines/table-engines/mergetree-family/summingmergetree.md @@ -42,9 +42,10 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] Устаревший способ создания таблицы - :::note "Attention" - Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. - ::: +:::note "Attention" +Не используйте этот способ в новых проектах и по возможности переведите старые проекты на способ описанный выше. +::: + ``` sql CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] ( diff --git a/docs/ru/sql-reference/functions/date-time-functions.md b/docs/ru/sql-reference/functions/date-time-functions.md index 80e2561a8d7..f18c2ea258a 100644 --- a/docs/ru/sql-reference/functions/date-time-functions.md +++ b/docs/ru/sql-reference/functions/date-time-functions.md @@ -316,9 +316,9 @@ SELECT toStartOfISOYear(toDate('2017-01-01')) AS ISOYear20170101; Возвращается дата. :::note "Attention" - Возвращаемое значение для некорректных дат зависит от реализации. ClickHouse может вернуть нулевую дату, выбросить исключение, или выполнить «естественное» перетекание дат между месяцами. +Возвращаемое значение для некорректных дат зависит от реализации. ClickHouse может вернуть нулевую дату, выбросить исключение, или выполнить «естественное» перетекание дат между месяцами. ::: - + ## toMonday {#tomonday} Округляет дату или дату-с-временем вниз до ближайшего понедельника. @@ -1126,8 +1126,7 @@ SELECT FROM_UNIXTIME(423543535); └──────────────────────────┘ ``` -В случае, когда есть два аргумента: первый типа [Integer](../../sql-reference/data-types/int-uint.md) или [DateTime](../../sql-reference/data-types/datetime.md), а второй является строкой постоянного формата — функция работает также, как [formatDateTime](#formatdatetime), и возвращает значение типа [String](../../sql-reference/data-types/string.md#string). - +В случае, когда есть два или три аргумента: первый типа [Integer](../../sql-reference/data-types/int-uint.md), [Date](../../sql-reference/data-types/date.md), [Date32](../../sql-reference/data-types/date32.md), [DateTime](../../sql-reference/data-types/datetime.md) или [DateTime64](../../sql-reference/data-types/datetime64.md), а второй является строкой постоянного формата и третий является строкой постоянной временной зоны — функция работает также, как [formatDateTime](#formatdatetime), и возвращает значение типа [String](../../sql-reference/data-types/string.md#string). Запрос: diff --git a/docs/ru/sql-reference/operators/in.md b/docs/ru/sql-reference/operators/in.md index 2b3d87a877f..fa679b890a7 100644 --- a/docs/ru/sql-reference/operators/in.md +++ b/docs/ru/sql-reference/operators/in.md @@ -122,9 +122,9 @@ FROM t_null Существует два варианта IN-ов с подзапросами (аналогично для JOIN-ов): обычный `IN` / `JOIN` и `GLOBAL IN` / `GLOBAL JOIN`. Они отличаются способом выполнения при распределённой обработке запроса. - :::note "Attention" - Помните, что алгоритмы, описанные ниже, могут работать иначе в зависимости от [настройки](../../operations/settings/settings.md) `distributed_product_mode`. - ::: +:::note "Attention" +Помните, что алгоритмы, описанные ниже, могут работать иначе в зависимости от [настройки](../../operations/settings/settings.md) `distributed_product_mode`. +::: При использовании обычного IN-а, запрос отправляется на удалённые серверы, и на каждом из них выполняются подзапросы в секциях `IN` / `JOIN`. При использовании `GLOBAL IN` / `GLOBAL JOIN-а`, сначала выполняются все подзапросы для `GLOBAL IN` / `GLOBAL JOIN-ов`, и результаты складываются во временные таблицы. Затем эти временные таблицы передаются на каждый удалённый сервер, и на них выполняются запросы, с использованием этих переданных временных данных. diff --git a/programs/copier/Aliases.h b/programs/copier/Aliases.h index c4d9c40d9f1..02be3441acd 100644 --- a/programs/copier/Aliases.h +++ b/programs/copier/Aliases.h @@ -1,6 +1,10 @@ #pragma once -#include +#include + +#include + +#include namespace DB { @@ -8,21 +12,4 @@ namespace DB using DatabaseAndTableName = std::pair; using ListOfDatabasesAndTableNames = std::vector; - - /// Hierarchical description of the tasks - struct ShardPartitionPiece; - struct ShardPartition; - struct TaskShard; - struct TaskTable; - struct TaskCluster; - struct ClusterPartition; - - using PartitionPieces = std::vector; - using TasksPartition = std::map>; - using ShardInfo = Cluster::ShardInfo; - using TaskShardPtr = std::shared_ptr; - using TasksShard = std::vector; - using TasksTable = std::list; - using ClusterPartitions = std::map>; } - diff --git a/programs/copier/CMakeLists.txt b/programs/copier/CMakeLists.txt index 57e0996ed78..2c17e70bc5e 100644 --- a/programs/copier/CMakeLists.txt +++ b/programs/copier/CMakeLists.txt @@ -1,7 +1,13 @@ set(CLICKHOUSE_COPIER_SOURCES "${CMAKE_CURRENT_SOURCE_DIR}/ClusterCopierApp.cpp" "${CMAKE_CURRENT_SOURCE_DIR}/ClusterCopier.cpp" - "${CMAKE_CURRENT_SOURCE_DIR}/Internals.cpp") + "${CMAKE_CURRENT_SOURCE_DIR}/Internals.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/ShardPartition.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/ShardPartitionPiece.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/StatusAccumulator.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/TaskCluster.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/TaskShard.cpp" + "${CMAKE_CURRENT_SOURCE_DIR}/TaskTable.cpp") set (CLICKHOUSE_COPIER_LINK PRIVATE diff --git a/programs/copier/ClusterCopier.h b/programs/copier/ClusterCopier.h index b354fc59eee..063b13e9078 100644 --- a/programs/copier/ClusterCopier.h +++ b/programs/copier/ClusterCopier.h @@ -3,7 +3,8 @@ #include "Aliases.h" #include "Internals.h" #include "TaskCluster.h" -#include "TaskTableAndShard.h" +#include "TaskShard.h" +#include "TaskTable.h" #include "ShardPartition.h" #include "ShardPartitionPiece.h" #include "ZooKeeperStaff.h" diff --git a/programs/copier/ClusterPartition.h b/programs/copier/ClusterPartition.h index ed69bfa8c26..22063989e22 100644 --- a/programs/copier/ClusterPartition.h +++ b/programs/copier/ClusterPartition.h @@ -1,17 +1,22 @@ #pragma once -#include "Aliases.h" +#include +#include namespace DB { - /// Contains info about all shards that contain a partition - struct ClusterPartition - { - double elapsed_time_seconds = 0; - UInt64 bytes_copied = 0; - UInt64 rows_copied = 0; - UInt64 blocks_copied = 0; - UInt64 total_tries = 0; - }; +/// Contains info about all shards that contain a partition +struct ClusterPartition +{ + double elapsed_time_seconds = 0; + UInt64 bytes_copied = 0; + UInt64 rows_copied = 0; + UInt64 blocks_copied = 0; + + UInt64 total_tries = 0; +}; + +using ClusterPartitions = std::map>; + } diff --git a/programs/copier/ShardPartition.cpp b/programs/copier/ShardPartition.cpp new file mode 100644 index 00000000000..4c962fc807d --- /dev/null +++ b/programs/copier/ShardPartition.cpp @@ -0,0 +1,70 @@ +#include "ShardPartition.h" + +#include "TaskShard.h" +#include "TaskTable.h" + +namespace DB +{ + +ShardPartition::ShardPartition(TaskShard & parent, String name_quoted_, size_t number_of_splits) + : task_shard(parent) + , name(std::move(name_quoted_)) +{ + pieces.reserve(number_of_splits); +} + +String ShardPartition::getPartitionCleanStartPath() const +{ + return getPartitionPath() + "/clean_start"; +} + +String ShardPartition::getPartitionPieceCleanStartPath(size_t current_piece_number) const +{ + assert(current_piece_number < task_shard.task_table.number_of_splits); + return getPartitionPiecePath(current_piece_number) + "/clean_start"; +} + +String ShardPartition::getPartitionPath() const +{ + return task_shard.task_table.getPartitionPath(name); +} + +String ShardPartition::getPartitionPiecePath(size_t current_piece_number) const +{ + assert(current_piece_number < task_shard.task_table.number_of_splits); + return task_shard.task_table.getPartitionPiecePath(name, current_piece_number); +} + +String ShardPartition::getShardStatusPath() const +{ + // schema: //tables///shards/ + // e.g. /root/table_test.hits/201701/shards/1 + return getPartitionShardsPath() + "/" + toString(task_shard.numberInCluster()); +} + +String ShardPartition::getPartitionShardsPath() const +{ + return getPartitionPath() + "/shards"; +} + +String ShardPartition::getPartitionActiveWorkersPath() const +{ + return getPartitionPath() + "/partition_active_workers"; +} + +String ShardPartition::getActiveWorkerPath() const +{ + return getPartitionActiveWorkersPath() + "/" + toString(task_shard.numberInCluster()); +} + +String ShardPartition::getCommonPartitionIsDirtyPath() const +{ + return getPartitionPath() + "/is_dirty"; +} + +String ShardPartition::getCommonPartitionIsCleanedPath() const +{ + return getCommonPartitionIsDirtyPath() + "/cleaned"; +} + +} diff --git a/programs/copier/ShardPartition.h b/programs/copier/ShardPartition.h index 7de381977f9..2457213733c 100644 --- a/programs/copier/ShardPartition.h +++ b/programs/copier/ShardPartition.h @@ -1,19 +1,23 @@ #pragma once -#include "Aliases.h" -#include "TaskTableAndShard.h" +#include "ShardPartitionPiece.h" + +#include + +#include namespace DB { +struct TaskShard; + /// Just destination partition of a shard /// I don't know what this comment means. /// In short, when we discovered what shards contain currently processing partition, /// This class describes a partition (name) that is stored on the shard (parent). struct ShardPartition { - ShardPartition(TaskShard &parent, String name_quoted_, size_t number_of_splits = 10) - : task_shard(parent), name(std::move(name_quoted_)) { pieces.reserve(number_of_splits); } + ShardPartition(TaskShard &parent, String name_quoted_, size_t number_of_splits = 10); String getPartitionPath() const; @@ -45,58 +49,6 @@ struct ShardPartition String name; }; -inline String ShardPartition::getPartitionCleanStartPath() const -{ - return getPartitionPath() + "/clean_start"; -} - -inline String ShardPartition::getPartitionPieceCleanStartPath(size_t current_piece_number) const -{ - assert(current_piece_number < task_shard.task_table.number_of_splits); - return getPartitionPiecePath(current_piece_number) + "/clean_start"; -} - -inline String ShardPartition::getPartitionPath() const -{ - return task_shard.task_table.getPartitionPath(name); -} - -inline String ShardPartition::getPartitionPiecePath(size_t current_piece_number) const -{ - assert(current_piece_number < task_shard.task_table.number_of_splits); - return task_shard.task_table.getPartitionPiecePath(name, current_piece_number); -} - -inline String ShardPartition::getShardStatusPath() const -{ - // schema: //tables/
//shards/ - // e.g. /root/table_test.hits/201701/shards/1 - return getPartitionShardsPath() + "/" + toString(task_shard.numberInCluster()); -} - -inline String ShardPartition::getPartitionShardsPath() const -{ - return getPartitionPath() + "/shards"; -} - -inline String ShardPartition::getPartitionActiveWorkersPath() const -{ - return getPartitionPath() + "/partition_active_workers"; -} - -inline String ShardPartition::getActiveWorkerPath() const -{ - return getPartitionActiveWorkersPath() + "/" + toString(task_shard.numberInCluster()); -} - -inline String ShardPartition::getCommonPartitionIsDirtyPath() const -{ - return getPartitionPath() + "/is_dirty"; -} - -inline String ShardPartition::getCommonPartitionIsCleanedPath() const -{ - return getCommonPartitionIsDirtyPath() + "/cleaned"; -} +using TasksPartition = std::map>; } diff --git a/programs/copier/ShardPartitionPiece.cpp b/programs/copier/ShardPartitionPiece.cpp new file mode 100644 index 00000000000..36d1621e012 --- /dev/null +++ b/programs/copier/ShardPartitionPiece.cpp @@ -0,0 +1,64 @@ +#include "ShardPartitionPiece.h" + +#include "ShardPartition.h" +#include "TaskShard.h" + +#include + +namespace DB +{ + +ShardPartitionPiece::ShardPartitionPiece(ShardPartition & parent, size_t current_piece_number_, bool is_present_piece_) + : is_absent_piece(!is_present_piece_) + , current_piece_number(current_piece_number_) + , shard_partition(parent) +{ +} + +String ShardPartitionPiece::getPartitionPiecePath() const +{ + return shard_partition.getPartitionPath() + "/piece_" + toString(current_piece_number); +} + +String ShardPartitionPiece::getPartitionPieceCleanStartPath() const +{ + return getPartitionPiecePath() + "/clean_start"; +} + +String ShardPartitionPiece::getPartitionPieceIsDirtyPath() const +{ + return getPartitionPiecePath() + "/is_dirty"; +} + +String ShardPartitionPiece::getPartitionPieceIsCleanedPath() const +{ + return getPartitionPieceIsDirtyPath() + "/cleaned"; +} + +String ShardPartitionPiece::getPartitionPieceActiveWorkersPath() const +{ + return getPartitionPiecePath() + "/partition_piece_active_workers"; +} + +String ShardPartitionPiece::getActiveWorkerPath() const +{ + return getPartitionPieceActiveWorkersPath() + "/" + toString(shard_partition.task_shard.numberInCluster()); +} + +/// On what shards do we have current partition. +String ShardPartitionPiece::getPartitionPieceShardsPath() const +{ + return getPartitionPiecePath() + "/shards"; +} + +String ShardPartitionPiece::getShardStatusPath() const +{ + return getPartitionPieceShardsPath() + "/" + toString(shard_partition.task_shard.numberInCluster()); +} + +String ShardPartitionPiece::getPartitionPieceCleanerPath() const +{ + return getPartitionPieceIsDirtyPath() + "/cleaner"; +} + +} diff --git a/programs/copier/ShardPartitionPiece.h b/programs/copier/ShardPartitionPiece.h index a21fd531da4..aba378d466d 100644 --- a/programs/copier/ShardPartitionPiece.h +++ b/programs/copier/ShardPartitionPiece.h @@ -1,16 +1,15 @@ #pragma once -#include "Internals.h" +#include namespace DB { +struct ShardPartition; + struct ShardPartitionPiece { - - ShardPartitionPiece(ShardPartition &parent, size_t current_piece_number_, bool is_present_piece_) - : is_absent_piece(!is_present_piece_), current_piece_number(current_piece_number_), - shard_partition(parent) {} + ShardPartitionPiece(ShardPartition & parent, size_t current_piece_number_, bool is_present_piece_); String getPartitionPiecePath() const; @@ -37,52 +36,6 @@ struct ShardPartitionPiece ShardPartition & shard_partition; }; - -inline String ShardPartitionPiece::getPartitionPiecePath() const -{ - return shard_partition.getPartitionPath() + "/piece_" + toString(current_piece_number); -} - -inline String ShardPartitionPiece::getPartitionPieceCleanStartPath() const -{ - return getPartitionPiecePath() + "/clean_start"; -} - -inline String ShardPartitionPiece::getPartitionPieceIsDirtyPath() const -{ - return getPartitionPiecePath() + "/is_dirty"; -} - -inline String ShardPartitionPiece::getPartitionPieceIsCleanedPath() const -{ - return getPartitionPieceIsDirtyPath() + "/cleaned"; -} - -inline String ShardPartitionPiece::getPartitionPieceActiveWorkersPath() const -{ - return getPartitionPiecePath() + "/partition_piece_active_workers"; -} - -inline String ShardPartitionPiece::getActiveWorkerPath() const -{ - return getPartitionPieceActiveWorkersPath() + "/" + toString(shard_partition.task_shard.numberInCluster()); -} - -/// On what shards do we have current partition. -inline String ShardPartitionPiece::getPartitionPieceShardsPath() const -{ - return getPartitionPiecePath() + "/shards"; -} - -inline String ShardPartitionPiece::getShardStatusPath() const -{ - return getPartitionPieceShardsPath() + "/" + toString(shard_partition.task_shard.numberInCluster()); -} - -inline String ShardPartitionPiece::getPartitionPieceCleanerPath() const -{ - return getPartitionPieceIsDirtyPath() + "/cleaner"; -} - +using PartitionPieces = std::vector; } diff --git a/programs/copier/StatusAccumulator.cpp b/programs/copier/StatusAccumulator.cpp new file mode 100644 index 00000000000..77adeac708c --- /dev/null +++ b/programs/copier/StatusAccumulator.cpp @@ -0,0 +1,48 @@ +#include "StatusAccumulator.h" + +#include +#include +#include +#include + +#include + +namespace DB +{ + +StatusAccumulator::MapPtr StatusAccumulator::fromJSON(String state_json) +{ + Poco::JSON::Parser parser; + auto state = parser.parse(state_json).extract(); + MapPtr result_ptr = std::make_shared(); + for (const auto & table_name : state->getNames()) + { + auto table_status_json = state->getValue(table_name); + auto table_status = parser.parse(table_status_json).extract(); + /// Map entry will be created if it is absent + auto & map_table_status = (*result_ptr)[table_name]; + map_table_status.all_partitions_count += table_status->getValue("all_partitions_count"); + map_table_status.processed_partitions_count += table_status->getValue("processed_partitions_count"); + } + return result_ptr; +} + +String StatusAccumulator::serializeToJSON(MapPtr statuses) +{ + Poco::JSON::Object result_json; + for (const auto & [table_name, table_status] : *statuses) + { + Poco::JSON::Object status_json; + status_json.set("all_partitions_count", table_status.all_partitions_count); + status_json.set("processed_partitions_count", table_status.processed_partitions_count); + + result_json.set(table_name, status_json); + } + std::ostringstream oss; // STYLE_CHECK_ALLOW_STD_STRING_STREAM + oss.exceptions(std::ios::failbit); + Poco::JSON::Stringifier::stringify(result_json, oss); + auto result = oss.str(); + return result; +} + +} diff --git a/programs/copier/StatusAccumulator.h b/programs/copier/StatusAccumulator.h index 6e20e3dc95d..d420b611602 100644 --- a/programs/copier/StatusAccumulator.h +++ b/programs/copier/StatusAccumulator.h @@ -1,65 +1,27 @@ #pragma once +#include -#include -#include -#include -#include - -#include #include -#include -#include +#include namespace DB { class StatusAccumulator { - public: - struct TableStatus - { - size_t all_partitions_count; - size_t processed_partitions_count; - }; +public: + struct TableStatus + { + size_t all_partitions_count; + size_t processed_partitions_count; + }; - using Map = std::unordered_map; - using MapPtr = std::shared_ptr; + using Map = std::unordered_map; + using MapPtr = std::shared_ptr; - static MapPtr fromJSON(std::string state_json) - { - Poco::JSON::Parser parser; - auto state = parser.parse(state_json).extract(); - MapPtr result_ptr = std::make_shared(); - for (const auto & table_name : state->getNames()) - { - auto table_status_json = state->getValue(table_name); - auto table_status = parser.parse(table_status_json).extract(); - /// Map entry will be created if it is absent - auto & map_table_status = (*result_ptr)[table_name]; - map_table_status.all_partitions_count += table_status->getValue("all_partitions_count"); - map_table_status.processed_partitions_count += table_status->getValue("processed_partitions_count"); - } - return result_ptr; - } - - static std::string serializeToJSON(MapPtr statuses) - { - Poco::JSON::Object result_json; - for (const auto & [table_name, table_status] : *statuses) - { - Poco::JSON::Object status_json; - status_json.set("all_partitions_count", table_status.all_partitions_count); - status_json.set("processed_partitions_count", table_status.processed_partitions_count); - - result_json.set(table_name, status_json); - } - std::ostringstream oss; // STYLE_CHECK_ALLOW_STD_STRING_STREAM - oss.exceptions(std::ios::failbit); - Poco::JSON::Stringifier::stringify(result_json, oss); - auto result = oss.str(); - return result; - } + static MapPtr fromJSON(String state_json); + static String serializeToJSON(MapPtr statuses); }; } diff --git a/programs/copier/TaskCluster.cpp b/programs/copier/TaskCluster.cpp new file mode 100644 index 00000000000..957c7d2120d --- /dev/null +++ b/programs/copier/TaskCluster.cpp @@ -0,0 +1,74 @@ +#include "TaskCluster.h" + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int BAD_ARGUMENTS; +} + +TaskCluster::TaskCluster(const String & task_zookeeper_path_, const String & default_local_database_) + : task_zookeeper_path(task_zookeeper_path_) + , default_local_database(default_local_database_) +{} + +void DB::TaskCluster::loadTasks(const Poco::Util::AbstractConfiguration & config, const String & base_key) +{ + String prefix = base_key.empty() ? "" : base_key + "."; + + clusters_prefix = prefix + "remote_servers"; + if (!config.has(clusters_prefix)) + throw Exception("You should specify list of clusters in " + clusters_prefix, ErrorCodes::BAD_ARGUMENTS); + + Poco::Util::AbstractConfiguration::Keys tables_keys; + config.keys(prefix + "tables", tables_keys); + + for (const auto & table_key : tables_keys) + { + table_tasks.emplace_back(*this, config, prefix + "tables", table_key); + } +} + +void DB::TaskCluster::reloadSettings(const Poco::Util::AbstractConfiguration & config, const String & base_key) +{ + String prefix = base_key.empty() ? "" : base_key + "."; + + max_workers = config.getUInt64(prefix + "max_workers"); + + settings_common = Settings(); + if (config.has(prefix + "settings")) + settings_common.loadSettingsFromConfig(prefix + "settings", config); + + settings_common.prefer_localhost_replica = false; + + settings_pull = settings_common; + if (config.has(prefix + "settings_pull")) + settings_pull.loadSettingsFromConfig(prefix + "settings_pull", config); + + settings_push = settings_common; + if (config.has(prefix + "settings_push")) + settings_push.loadSettingsFromConfig(prefix + "settings_push", config); + + auto set_default_value = [] (auto && setting, auto && default_value) + { + setting = setting.changed ? setting.value : default_value; + }; + + /// Override important settings + settings_pull.readonly = 1; + settings_pull.prefer_localhost_replica = false; + settings_push.insert_distributed_sync = true; + settings_push.prefer_localhost_replica = false; + + set_default_value(settings_pull.load_balancing, LoadBalancing::NEAREST_HOSTNAME); + set_default_value(settings_pull.max_threads, 1); + set_default_value(settings_pull.max_block_size, 8192UL); + set_default_value(settings_pull.preferred_block_size_bytes, 0); + + set_default_value(settings_push.insert_distributed_timeout, 0); + set_default_value(settings_push.replication_alter_partitions_sync, 2); +} + +} + diff --git a/programs/copier/TaskCluster.h b/programs/copier/TaskCluster.h index 7d8f01ba15f..fc1c8a663ec 100644 --- a/programs/copier/TaskCluster.h +++ b/programs/copier/TaskCluster.h @@ -1,21 +1,20 @@ #pragma once -#include "Aliases.h" +#include "TaskTable.h" + +#include +#include + #include +#include + namespace DB { -namespace ErrorCodes -{ - extern const int BAD_ARGUMENTS; -} struct TaskCluster { - TaskCluster(const String & task_zookeeper_path_, const String & default_local_database_) - : task_zookeeper_path(task_zookeeper_path_) - , default_local_database(default_local_database_) - {} + TaskCluster(const String & task_zookeeper_path_, const String & default_local_database_); void loadTasks(const Poco::Util::AbstractConfiguration & config, const String & base_key = ""); @@ -50,61 +49,4 @@ struct TaskCluster pcg64 random_engine; }; -inline void DB::TaskCluster::loadTasks(const Poco::Util::AbstractConfiguration & config, const String & base_key) -{ - String prefix = base_key.empty() ? "" : base_key + "."; - - clusters_prefix = prefix + "remote_servers"; - if (!config.has(clusters_prefix)) - throw Exception("You should specify list of clusters in " + clusters_prefix, ErrorCodes::BAD_ARGUMENTS); - - Poco::Util::AbstractConfiguration::Keys tables_keys; - config.keys(prefix + "tables", tables_keys); - - for (const auto & table_key : tables_keys) - { - table_tasks.emplace_back(*this, config, prefix + "tables", table_key); - } -} - -inline void DB::TaskCluster::reloadSettings(const Poco::Util::AbstractConfiguration & config, const String & base_key) -{ - String prefix = base_key.empty() ? "" : base_key + "."; - - max_workers = config.getUInt64(prefix + "max_workers"); - - settings_common = Settings(); - if (config.has(prefix + "settings")) - settings_common.loadSettingsFromConfig(prefix + "settings", config); - - settings_common.prefer_localhost_replica = 0; - - settings_pull = settings_common; - if (config.has(prefix + "settings_pull")) - settings_pull.loadSettingsFromConfig(prefix + "settings_pull", config); - - settings_push = settings_common; - if (config.has(prefix + "settings_push")) - settings_push.loadSettingsFromConfig(prefix + "settings_push", config); - - auto set_default_value = [] (auto && setting, auto && default_value) - { - setting = setting.changed ? setting.value : default_value; - }; - - /// Override important settings - settings_pull.readonly = 1; - settings_pull.prefer_localhost_replica = false; - settings_push.insert_distributed_sync = true; - settings_push.prefer_localhost_replica = false; - - set_default_value(settings_pull.load_balancing, LoadBalancing::NEAREST_HOSTNAME); - set_default_value(settings_pull.max_threads, 1); - set_default_value(settings_pull.max_block_size, 8192UL); - set_default_value(settings_pull.preferred_block_size_bytes, 0); - - set_default_value(settings_push.insert_distributed_timeout, 0); - set_default_value(settings_push.replication_alter_partitions_sync, 2); -} - } diff --git a/programs/copier/TaskShard.cpp b/programs/copier/TaskShard.cpp new file mode 100644 index 00000000000..d156f451a84 --- /dev/null +++ b/programs/copier/TaskShard.cpp @@ -0,0 +1,37 @@ +#include "TaskShard.h" + +#include "TaskTable.h" + +namespace DB +{ + +TaskShard::TaskShard(TaskTable & parent, const Cluster::ShardInfo & info_) + : task_table(parent) + , info(info_) +{ + list_of_split_tables_on_shard.assign(task_table.number_of_splits, DatabaseAndTableName()); +} + +UInt32 TaskShard::numberInCluster() const +{ + return info.shard_num; +} + +UInt32 TaskShard::indexInCluster() const +{ + return info.shard_num - 1; +} + +String DB::TaskShard::getDescription() const +{ + return fmt::format("N{} (having a replica {}, pull table {} of cluster {}", + numberInCluster(), getHostNameExample(), getQuotedTable(task_table.table_pull), task_table.cluster_pull_name); +} + +String DB::TaskShard::getHostNameExample() const +{ + const auto & replicas = task_table.cluster_pull->getShardsAddresses().at(indexInCluster()); + return replicas.at(0).readableString(); +} + +} diff --git a/programs/copier/TaskShard.h b/programs/copier/TaskShard.h new file mode 100644 index 00000000000..05d652077ea --- /dev/null +++ b/programs/copier/TaskShard.h @@ -0,0 +1,56 @@ +#pragma once + +#include "Aliases.h" +#include "Internals.h" +#include "ClusterPartition.h" +#include "ShardPartition.h" + + +namespace DB +{ + +struct TaskTable; + +struct TaskShard +{ + TaskShard(TaskTable & parent, const Cluster::ShardInfo & info_); + + TaskTable & task_table; + + Cluster::ShardInfo info; + + UInt32 numberInCluster() const; + + UInt32 indexInCluster() const; + + String getDescription() const; + + String getHostNameExample() const; + + /// Used to sort clusters by their proximity + ShardPriority priority; + + /// Column with unique destination partitions (computed from engine_push_partition_key expr.) in the shard + ColumnWithTypeAndName partition_key_column; + + /// There is a task for each destination partition + TasksPartition partition_tasks; + + /// Which partitions have been checked for existence + /// If some partition from this lists is exists, it is in partition_tasks + std::set checked_partitions; + + /// Last CREATE TABLE query of the table of the shard + ASTPtr current_pull_table_create_query; + ASTPtr current_push_table_create_query; + + /// Internal distributed tables + DatabaseAndTableName table_read_shard; + DatabaseAndTableName main_table_split_shard; + ListOfDatabasesAndTableNames list_of_split_tables_on_shard; +}; + +using TaskShardPtr = std::shared_ptr; +using TasksShard = std::vector; + +} diff --git a/programs/copier/TaskTable.cpp b/programs/copier/TaskTable.cpp new file mode 100644 index 00000000000..5b09a9c99a7 --- /dev/null +++ b/programs/copier/TaskTable.cpp @@ -0,0 +1,221 @@ +#include "TaskTable.h" + +#include "ClusterPartition.h" +#include "TaskCluster.h" + +#include + +#include + +namespace DB +{ +namespace ErrorCodes +{ + extern const int UNKNOWN_ELEMENT_IN_CONFIG; + extern const int LOGICAL_ERROR; +} + +TaskTable::TaskTable(TaskCluster & parent, const Poco::Util::AbstractConfiguration & config, + const String & prefix_, const String & table_key) + : task_cluster(parent) +{ + String table_prefix = prefix_ + "." + table_key + "."; + + name_in_config = table_key; + + number_of_splits = config.getUInt64(table_prefix + "number_of_splits", 3); + + allow_to_copy_alias_and_materialized_columns = config.getBool(table_prefix + "allow_to_copy_alias_and_materialized_columns", false); + allow_to_drop_target_partitions = config.getBool(table_prefix + "allow_to_drop_target_partitions", false); + + cluster_pull_name = config.getString(table_prefix + "cluster_pull"); + cluster_push_name = config.getString(table_prefix + "cluster_push"); + + table_pull.first = config.getString(table_prefix + "database_pull"); + table_pull.second = config.getString(table_prefix + "table_pull"); + + table_push.first = config.getString(table_prefix + "database_push"); + table_push.second = config.getString(table_prefix + "table_push"); + + /// Used as node name in ZooKeeper + table_id = escapeForFileName(cluster_push_name) + + "." + escapeForFileName(table_push.first) + + "." + escapeForFileName(table_push.second); + + engine_push_str = config.getString(table_prefix + "engine", "rand()"); + + { + ParserStorage parser_storage; + engine_push_ast = parseQuery(parser_storage, engine_push_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); + engine_push_partition_key_ast = extractPartitionKey(engine_push_ast); + primary_key_comma_separated = boost::algorithm::join(extractPrimaryKeyColumnNames(engine_push_ast), ", "); + is_replicated_table = isReplicatedTableEngine(engine_push_ast); + } + + sharding_key_str = config.getString(table_prefix + "sharding_key"); + + auxiliary_engine_split_asts.reserve(number_of_splits); + { + ParserExpressionWithOptionalAlias parser_expression(false); + sharding_key_ast = parseQuery(parser_expression, sharding_key_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); + main_engine_split_ast = createASTStorageDistributed(cluster_push_name, table_push.first, table_push.second, + sharding_key_ast); + + for (const auto piece_number : collections::range(0, number_of_splits)) + { + auxiliary_engine_split_asts.emplace_back + ( + createASTStorageDistributed(cluster_push_name, table_push.first, + table_push.second + "_piece_" + toString(piece_number), sharding_key_ast) + ); + } + } + + where_condition_str = config.getString(table_prefix + "where_condition", ""); + if (!where_condition_str.empty()) + { + ParserExpressionWithOptionalAlias parser_expression(false); + where_condition_ast = parseQuery(parser_expression, where_condition_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); + + // Will use canonical expression form + where_condition_str = queryToString(where_condition_ast); + } + + String enabled_partitions_prefix = table_prefix + "enabled_partitions"; + has_enabled_partitions = config.has(enabled_partitions_prefix); + + if (has_enabled_partitions) + { + Strings keys; + config.keys(enabled_partitions_prefix, keys); + + if (keys.empty()) + { + /// Parse list of partition from space-separated string + String partitions_str = config.getString(table_prefix + "enabled_partitions"); + boost::trim_if(partitions_str, isWhitespaceASCII); + boost::split(enabled_partitions, partitions_str, isWhitespaceASCII, boost::token_compress_on); + } + else + { + /// Parse sequence of ... + for (const String &key : keys) + { + if (!startsWith(key, "partition")) + throw Exception("Unknown key " + key + " in " + enabled_partitions_prefix, ErrorCodes::UNKNOWN_ELEMENT_IN_CONFIG); + + enabled_partitions.emplace_back(config.getString(enabled_partitions_prefix + "." + key)); + } + } + + std::copy(enabled_partitions.begin(), enabled_partitions.end(), std::inserter(enabled_partitions_set, enabled_partitions_set.begin())); + } +} + + +String TaskTable::getPartitionPath(const String & partition_name) const +{ + return task_cluster.task_zookeeper_path // root + + "/tables/" + table_id // tables/dst_cluster.merge.hits + + "/" + escapeForFileName(partition_name); // 201701 +} + +String TaskTable::getPartitionAttachIsActivePath(const String & partition_name) const +{ + return getPartitionPath(partition_name) + "/attach_active"; +} + +String TaskTable::getPartitionAttachIsDonePath(const String & partition_name) const +{ + return getPartitionPath(partition_name) + "/attach_is_done"; +} + +String TaskTable::getPartitionPiecePath(const String & partition_name, size_t piece_number) const +{ + assert(piece_number < number_of_splits); + return getPartitionPath(partition_name) + "/piece_" + toString(piece_number); // 1...number_of_splits +} + +String TaskTable::getCertainPartitionIsDirtyPath(const String &partition_name) const +{ + return getPartitionPath(partition_name) + "/is_dirty"; +} + +String TaskTable::getCertainPartitionPieceIsDirtyPath(const String & partition_name, const size_t piece_number) const +{ + return getPartitionPiecePath(partition_name, piece_number) + "/is_dirty"; +} + +String TaskTable::getCertainPartitionIsCleanedPath(const String & partition_name) const +{ + return getCertainPartitionIsDirtyPath(partition_name) + "/cleaned"; +} + +String TaskTable::getCertainPartitionPieceIsCleanedPath(const String & partition_name, const size_t piece_number) const +{ + return getCertainPartitionPieceIsDirtyPath(partition_name, piece_number) + "/cleaned"; +} + +String TaskTable::getCertainPartitionTaskStatusPath(const String & partition_name) const +{ + return getPartitionPath(partition_name) + "/shards"; +} + +String TaskTable::getCertainPartitionPieceTaskStatusPath(const String & partition_name, const size_t piece_number) const +{ + return getPartitionPiecePath(partition_name, piece_number) + "/shards"; +} + +bool TaskTable::isReplicatedTable() const +{ + return is_replicated_table; +} + +String TaskTable::getStatusAllPartitionCount() const +{ + return task_cluster.task_zookeeper_path + "/status/all_partitions_count"; +} + +String TaskTable::getStatusProcessedPartitionsCount() const +{ + return task_cluster.task_zookeeper_path + "/status/processed_partitions_count"; +} + +ASTPtr TaskTable::rewriteReplicatedCreateQueryToPlain() const +{ + ASTPtr prev_engine_push_ast = engine_push_ast->clone(); + + auto & new_storage_ast = prev_engine_push_ast->as(); + auto & new_engine_ast = new_storage_ast.engine->as(); + + /// Remove "Replicated" from name + new_engine_ast.name = new_engine_ast.name.substr(10); + + if (new_engine_ast.arguments) + { + auto & replicated_table_arguments = new_engine_ast.arguments->children; + + + /// In some cases of Atomic database engine usage ReplicatedMergeTree tables + /// could be created without arguments. + if (!replicated_table_arguments.empty()) + { + /// Delete first two arguments of Replicated...MergeTree() table. + replicated_table_arguments.erase(replicated_table_arguments.begin()); + replicated_table_arguments.erase(replicated_table_arguments.begin()); + } + } + + return new_storage_ast.clone(); +} + +ClusterPartition & TaskTable::getClusterPartition(const String & partition_name) +{ + auto it = cluster_partitions.find(partition_name); + if (it == cluster_partitions.end()) + throw Exception("There are no cluster partition " + partition_name + " in " + table_id, + ErrorCodes::LOGICAL_ERROR); + return it->second; +} + +} diff --git a/programs/copier/TaskTable.h b/programs/copier/TaskTable.h new file mode 100644 index 00000000000..2bb7f078bc6 --- /dev/null +++ b/programs/copier/TaskTable.h @@ -0,0 +1,173 @@ +#pragma once + +#include "Aliases.h" +#include "TaskShard.h" + + +namespace DB +{ + +struct ClusterPartition; +struct TaskCluster; + +struct TaskTable +{ + TaskTable(TaskCluster & parent, const Poco::Util::AbstractConfiguration & config, const String & prefix, const String & table_key); + + TaskCluster & task_cluster; + + /// These functions used in checkPartitionIsDone() or checkPartitionPieceIsDone() + /// They are implemented here not to call task_table.tasks_shard[partition_name].second.pieces[current_piece_number] etc. + + String getPartitionPath(const String & partition_name) const; + + String getPartitionAttachIsActivePath(const String & partition_name) const; + + String getPartitionAttachIsDonePath(const String & partition_name) const; + + String getPartitionPiecePath(const String & partition_name, size_t piece_number) const; + + String getCertainPartitionIsDirtyPath(const String & partition_name) const; + + String getCertainPartitionPieceIsDirtyPath(const String & partition_name, size_t piece_number) const; + + String getCertainPartitionIsCleanedPath(const String & partition_name) const; + + String getCertainPartitionPieceIsCleanedPath(const String & partition_name, size_t piece_number) const; + + String getCertainPartitionTaskStatusPath(const String & partition_name) const; + + String getCertainPartitionPieceTaskStatusPath(const String & partition_name, size_t piece_number) const; + + bool isReplicatedTable() const; + + /// These nodes are used for check-status option + String getStatusAllPartitionCount() const; + String getStatusProcessedPartitionsCount() const; + + /// Partitions will be split into number-of-splits pieces. + /// Each piece will be copied independently. (10 by default) + size_t number_of_splits; + + bool allow_to_copy_alias_and_materialized_columns{false}; + bool allow_to_drop_target_partitions{false}; + + String name_in_config; + + /// Used as task ID + String table_id; + + /// Column names in primary key + String primary_key_comma_separated; + + /// Source cluster and table + String cluster_pull_name; + DatabaseAndTableName table_pull; + + /// Destination cluster and table + String cluster_push_name; + DatabaseAndTableName table_push; + + /// Storage of destination table + /// (tables that are stored on each shard of target cluster) + String engine_push_str; + ASTPtr engine_push_ast; + ASTPtr engine_push_partition_key_ast; + + /// First argument of Replicated...MergeTree() + String engine_push_zk_path; + bool is_replicated_table; + + ASTPtr rewriteReplicatedCreateQueryToPlain() const; + + /* + * A Distributed table definition used to split data + * Distributed table will be created on each shard of default + * cluster to perform data copying and resharding + * */ + String sharding_key_str; + ASTPtr sharding_key_ast; + ASTPtr main_engine_split_ast; + + /* + * To copy partition piece form one cluster to another we have to use Distributed table. + * In case of usage separate table (engine_push) for each partition piece, + * we have to use many Distributed tables. + * */ + ASTs auxiliary_engine_split_asts; + + /// Additional WHERE expression to filter input data + String where_condition_str; + ASTPtr where_condition_ast; + + /// Resolved clusters + ClusterPtr cluster_pull; + ClusterPtr cluster_push; + + /// Filter partitions that should be copied + bool has_enabled_partitions = false; + Strings enabled_partitions; + NameSet enabled_partitions_set; + + /** + * Prioritized list of shards + * all_shards contains information about all shards in the table. + * So we have to check whether particular shard have current partition or not while processing. + */ + TasksShard all_shards; + TasksShard local_shards; + + /// All partitions of the current table. + ClusterPartitions cluster_partitions; + NameSet finished_cluster_partitions; + + /// Partition names to process in user-specified order + Strings ordered_partition_names; + + ClusterPartition & getClusterPartition(const String & partition_name); + + Stopwatch watch; + UInt64 bytes_copied = 0; + UInt64 rows_copied = 0; + + template + void initShards(RandomEngine &&random_engine); +}; + +using TasksTable = std::list; + + +template +inline void TaskTable::initShards(RandomEngine && random_engine) +{ + const String & fqdn_name = getFQDNOrHostName(); + std::uniform_int_distribution get_urand(0, std::numeric_limits::max()); + + // Compute the priority + for (const auto & shard_info : cluster_pull->getShardsInfo()) + { + TaskShardPtr task_shard = std::make_shared(*this, shard_info); + const auto & replicas = cluster_pull->getShardsAddresses().at(task_shard->indexInCluster()); + task_shard->priority = getReplicasPriority(replicas, fqdn_name, get_urand(random_engine)); + + all_shards.emplace_back(task_shard); + } + + // Sort by priority + std::sort(all_shards.begin(), all_shards.end(), + [](const TaskShardPtr & lhs, const TaskShardPtr & rhs) + { + return ShardPriority::greaterPriority(lhs->priority, rhs->priority); + }); + + // Cut local shards + auto it_first_remote = std::lower_bound(all_shards.begin(), all_shards.end(), 1, + [](const TaskShardPtr & lhs, UInt8 is_remote) + { + return lhs->priority.is_remote < is_remote; + }); + + local_shards.assign(all_shards.begin(), it_first_remote); +} + +} diff --git a/programs/copier/TaskTableAndShard.h b/programs/copier/TaskTableAndShard.h deleted file mode 100644 index cef9b669971..00000000000 --- a/programs/copier/TaskTableAndShard.h +++ /dev/null @@ -1,434 +0,0 @@ -#pragma once - -#include "Aliases.h" -#include "Internals.h" -#include "ClusterPartition.h" - -#include -#include - -#include -#include - - -namespace DB -{ -namespace ErrorCodes -{ - extern const int UNKNOWN_ELEMENT_IN_CONFIG; - extern const int LOGICAL_ERROR; -} - -struct TaskShard; - -struct TaskTable -{ - TaskTable(TaskCluster & parent, const Poco::Util::AbstractConfiguration & config, const String & prefix, - const String & table_key); - - TaskCluster & task_cluster; - - /// These functions used in checkPartitionIsDone() or checkPartitionPieceIsDone() - /// They are implemented here not to call task_table.tasks_shard[partition_name].second.pieces[current_piece_number] etc. - - String getPartitionPath(const String & partition_name) const; - - String getPartitionAttachIsActivePath(const String & partition_name) const; - - String getPartitionAttachIsDonePath(const String & partition_name) const; - - String getPartitionPiecePath(const String & partition_name, size_t piece_number) const; - - String getCertainPartitionIsDirtyPath(const String & partition_name) const; - - String getCertainPartitionPieceIsDirtyPath(const String & partition_name, size_t piece_number) const; - - String getCertainPartitionIsCleanedPath(const String & partition_name) const; - - String getCertainPartitionPieceIsCleanedPath(const String & partition_name, size_t piece_number) const; - - String getCertainPartitionTaskStatusPath(const String & partition_name) const; - - String getCertainPartitionPieceTaskStatusPath(const String & partition_name, size_t piece_number) const; - - bool isReplicatedTable() const { return is_replicated_table; } - - /// These nodes are used for check-status option - String getStatusAllPartitionCount() const; - String getStatusProcessedPartitionsCount() const; - - /// Partitions will be split into number-of-splits pieces. - /// Each piece will be copied independently. (10 by default) - size_t number_of_splits; - - bool allow_to_copy_alias_and_materialized_columns{false}; - bool allow_to_drop_target_partitions{false}; - - String name_in_config; - - /// Used as task ID - String table_id; - - /// Column names in primary key - String primary_key_comma_separated; - - /// Source cluster and table - String cluster_pull_name; - DatabaseAndTableName table_pull; - - /// Destination cluster and table - String cluster_push_name; - DatabaseAndTableName table_push; - - /// Storage of destination table - /// (tables that are stored on each shard of target cluster) - String engine_push_str; - ASTPtr engine_push_ast; - ASTPtr engine_push_partition_key_ast; - - /// First argument of Replicated...MergeTree() - String engine_push_zk_path; - bool is_replicated_table; - - ASTPtr rewriteReplicatedCreateQueryToPlain() const; - - /* - * A Distributed table definition used to split data - * Distributed table will be created on each shard of default - * cluster to perform data copying and resharding - * */ - String sharding_key_str; - ASTPtr sharding_key_ast; - ASTPtr main_engine_split_ast; - - /* - * To copy partition piece form one cluster to another we have to use Distributed table. - * In case of usage separate table (engine_push) for each partition piece, - * we have to use many Distributed tables. - * */ - ASTs auxiliary_engine_split_asts; - - /// Additional WHERE expression to filter input data - String where_condition_str; - ASTPtr where_condition_ast; - - /// Resolved clusters - ClusterPtr cluster_pull; - ClusterPtr cluster_push; - - /// Filter partitions that should be copied - bool has_enabled_partitions = false; - Strings enabled_partitions; - NameSet enabled_partitions_set; - - /** - * Prioritized list of shards - * all_shards contains information about all shards in the table. - * So we have to check whether particular shard have current partition or not while processing. - */ - TasksShard all_shards; - TasksShard local_shards; - - /// All partitions of the current table. - ClusterPartitions cluster_partitions; - NameSet finished_cluster_partitions; - - /// Partition names to process in user-specified order - Strings ordered_partition_names; - - ClusterPartition & getClusterPartition(const String & partition_name) - { - auto it = cluster_partitions.find(partition_name); - if (it == cluster_partitions.end()) - throw Exception("There are no cluster partition " + partition_name + " in " + table_id, - ErrorCodes::LOGICAL_ERROR); - return it->second; - } - - Stopwatch watch; - UInt64 bytes_copied = 0; - UInt64 rows_copied = 0; - - template - void initShards(RandomEngine &&random_engine); -}; - - -struct TaskShard -{ - TaskShard(TaskTable & parent, const ShardInfo & info_) : task_table(parent), info(info_) - { - list_of_split_tables_on_shard.assign(task_table.number_of_splits, DatabaseAndTableName()); - } - - TaskTable & task_table; - - ShardInfo info; - - UInt32 numberInCluster() const { return info.shard_num; } - - UInt32 indexInCluster() const { return info.shard_num - 1; } - - String getDescription() const; - - String getHostNameExample() const; - - /// Used to sort clusters by their proximity - ShardPriority priority; - - /// Column with unique destination partitions (computed from engine_push_partition_key expr.) in the shard - ColumnWithTypeAndName partition_key_column; - - /// There is a task for each destination partition - TasksPartition partition_tasks; - - /// Which partitions have been checked for existence - /// If some partition from this lists is exists, it is in partition_tasks - std::set checked_partitions; - - /// Last CREATE TABLE query of the table of the shard - ASTPtr current_pull_table_create_query; - ASTPtr current_push_table_create_query; - - /// Internal distributed tables - DatabaseAndTableName table_read_shard; - DatabaseAndTableName main_table_split_shard; - ListOfDatabasesAndTableNames list_of_split_tables_on_shard; -}; - - -inline String TaskTable::getPartitionPath(const String & partition_name) const -{ - return task_cluster.task_zookeeper_path // root - + "/tables/" + table_id // tables/dst_cluster.merge.hits - + "/" + escapeForFileName(partition_name); // 201701 -} - -inline String TaskTable::getPartitionAttachIsActivePath(const String & partition_name) const -{ - return getPartitionPath(partition_name) + "/attach_active"; -} - -inline String TaskTable::getPartitionAttachIsDonePath(const String & partition_name) const -{ - return getPartitionPath(partition_name) + "/attach_is_done"; -} - -inline String TaskTable::getPartitionPiecePath(const String & partition_name, size_t piece_number) const -{ - assert(piece_number < number_of_splits); - return getPartitionPath(partition_name) + "/piece_" + toString(piece_number); // 1...number_of_splits -} - -inline String TaskTable::getCertainPartitionIsDirtyPath(const String &partition_name) const -{ - return getPartitionPath(partition_name) + "/is_dirty"; -} - -inline String TaskTable::getCertainPartitionPieceIsDirtyPath(const String & partition_name, const size_t piece_number) const -{ - return getPartitionPiecePath(partition_name, piece_number) + "/is_dirty"; -} - -inline String TaskTable::getCertainPartitionIsCleanedPath(const String & partition_name) const -{ - return getCertainPartitionIsDirtyPath(partition_name) + "/cleaned"; -} - -inline String TaskTable::getCertainPartitionPieceIsCleanedPath(const String & partition_name, const size_t piece_number) const -{ - return getCertainPartitionPieceIsDirtyPath(partition_name, piece_number) + "/cleaned"; -} - -inline String TaskTable::getCertainPartitionTaskStatusPath(const String & partition_name) const -{ - return getPartitionPath(partition_name) + "/shards"; -} - -inline String TaskTable::getCertainPartitionPieceTaskStatusPath(const String & partition_name, const size_t piece_number) const -{ - return getPartitionPiecePath(partition_name, piece_number) + "/shards"; -} - -inline String TaskTable::getStatusAllPartitionCount() const -{ - return task_cluster.task_zookeeper_path + "/status/all_partitions_count"; -} - -inline String TaskTable::getStatusProcessedPartitionsCount() const -{ - return task_cluster.task_zookeeper_path + "/status/processed_partitions_count"; -} - -inline TaskTable::TaskTable(TaskCluster & parent, const Poco::Util::AbstractConfiguration & config, - const String & prefix_, const String & table_key) - : task_cluster(parent) -{ - String table_prefix = prefix_ + "." + table_key + "."; - - name_in_config = table_key; - - number_of_splits = config.getUInt64(table_prefix + "number_of_splits", 3); - - allow_to_copy_alias_and_materialized_columns = config.getBool(table_prefix + "allow_to_copy_alias_and_materialized_columns", false); - allow_to_drop_target_partitions = config.getBool(table_prefix + "allow_to_drop_target_partitions", false); - - cluster_pull_name = config.getString(table_prefix + "cluster_pull"); - cluster_push_name = config.getString(table_prefix + "cluster_push"); - - table_pull.first = config.getString(table_prefix + "database_pull"); - table_pull.second = config.getString(table_prefix + "table_pull"); - - table_push.first = config.getString(table_prefix + "database_push"); - table_push.second = config.getString(table_prefix + "table_push"); - - /// Used as node name in ZooKeeper - table_id = escapeForFileName(cluster_push_name) - + "." + escapeForFileName(table_push.first) - + "." + escapeForFileName(table_push.second); - - engine_push_str = config.getString(table_prefix + "engine", "rand()"); - - { - ParserStorage parser_storage; - engine_push_ast = parseQuery(parser_storage, engine_push_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); - engine_push_partition_key_ast = extractPartitionKey(engine_push_ast); - primary_key_comma_separated = boost::algorithm::join(extractPrimaryKeyColumnNames(engine_push_ast), ", "); - is_replicated_table = isReplicatedTableEngine(engine_push_ast); - } - - sharding_key_str = config.getString(table_prefix + "sharding_key"); - - auxiliary_engine_split_asts.reserve(number_of_splits); - { - ParserExpressionWithOptionalAlias parser_expression(false); - sharding_key_ast = parseQuery(parser_expression, sharding_key_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); - main_engine_split_ast = createASTStorageDistributed(cluster_push_name, table_push.first, table_push.second, - sharding_key_ast); - - for (const auto piece_number : collections::range(0, number_of_splits)) - { - auxiliary_engine_split_asts.emplace_back - ( - createASTStorageDistributed(cluster_push_name, table_push.first, - table_push.second + "_piece_" + toString(piece_number), sharding_key_ast) - ); - } - } - - where_condition_str = config.getString(table_prefix + "where_condition", ""); - if (!where_condition_str.empty()) - { - ParserExpressionWithOptionalAlias parser_expression(false); - where_condition_ast = parseQuery(parser_expression, where_condition_str, 0, DBMS_DEFAULT_MAX_PARSER_DEPTH); - - // Will use canonical expression form - where_condition_str = queryToString(where_condition_ast); - } - - String enabled_partitions_prefix = table_prefix + "enabled_partitions"; - has_enabled_partitions = config.has(enabled_partitions_prefix); - - if (has_enabled_partitions) - { - Strings keys; - config.keys(enabled_partitions_prefix, keys); - - if (keys.empty()) - { - /// Parse list of partition from space-separated string - String partitions_str = config.getString(table_prefix + "enabled_partitions"); - boost::trim_if(partitions_str, isWhitespaceASCII); - boost::split(enabled_partitions, partitions_str, isWhitespaceASCII, boost::token_compress_on); - } - else - { - /// Parse sequence of ... - for (const String &key : keys) - { - if (!startsWith(key, "partition")) - throw Exception("Unknown key " + key + " in " + enabled_partitions_prefix, ErrorCodes::UNKNOWN_ELEMENT_IN_CONFIG); - - enabled_partitions.emplace_back(config.getString(enabled_partitions_prefix + "." + key)); - } - } - - std::copy(enabled_partitions.begin(), enabled_partitions.end(), std::inserter(enabled_partitions_set, enabled_partitions_set.begin())); - } -} - -template -inline void TaskTable::initShards(RandomEngine && random_engine) -{ - const String & fqdn_name = getFQDNOrHostName(); - std::uniform_int_distribution get_urand(0, std::numeric_limits::max()); - - // Compute the priority - for (const auto & shard_info : cluster_pull->getShardsInfo()) - { - TaskShardPtr task_shard = std::make_shared(*this, shard_info); - const auto & replicas = cluster_pull->getShardsAddresses().at(task_shard->indexInCluster()); - task_shard->priority = getReplicasPriority(replicas, fqdn_name, get_urand(random_engine)); - - all_shards.emplace_back(task_shard); - } - - // Sort by priority - std::sort(all_shards.begin(), all_shards.end(), - [](const TaskShardPtr & lhs, const TaskShardPtr & rhs) - { - return ShardPriority::greaterPriority(lhs->priority, rhs->priority); - }); - - // Cut local shards - auto it_first_remote = std::lower_bound(all_shards.begin(), all_shards.end(), 1, - [](const TaskShardPtr & lhs, UInt8 is_remote) - { - return lhs->priority.is_remote < is_remote; - }); - - local_shards.assign(all_shards.begin(), it_first_remote); -} - -inline ASTPtr TaskTable::rewriteReplicatedCreateQueryToPlain() const -{ - ASTPtr prev_engine_push_ast = engine_push_ast->clone(); - - auto & new_storage_ast = prev_engine_push_ast->as(); - auto & new_engine_ast = new_storage_ast.engine->as(); - - /// Remove "Replicated" from name - new_engine_ast.name = new_engine_ast.name.substr(10); - - if (new_engine_ast.arguments) - { - auto & replicated_table_arguments = new_engine_ast.arguments->children; - - - /// In some cases of Atomic database engine usage ReplicatedMergeTree tables - /// could be created without arguments. - if (!replicated_table_arguments.empty()) - { - /// Delete first two arguments of Replicated...MergeTree() table. - replicated_table_arguments.erase(replicated_table_arguments.begin()); - replicated_table_arguments.erase(replicated_table_arguments.begin()); - } - } - - return new_storage_ast.clone(); -} - - -inline String DB::TaskShard::getDescription() const -{ - return fmt::format("N{} (having a replica {}, pull table {} of cluster {}", - numberInCluster(), getHostNameExample(), getQuotedTable(task_table.table_pull), task_table.cluster_pull_name); -} - -inline String DB::TaskShard::getHostNameExample() const -{ - const auto & replicas = task_table.cluster_pull->getShardsAddresses().at(indexInCluster()); - return replicas.at(0).readableString(); -} - -} diff --git a/src/Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.cpp b/src/Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.cpp index 0232d8958ff..e4d6633b6e6 100644 --- a/src/Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.cpp +++ b/src/Analyzer/Passes/OrderByLimitByDuplicateEliminationPass.cpp @@ -10,23 +10,34 @@ namespace DB namespace { -struct QueryTreeNodeHash +struct QueryTreeNodeWithHash { - size_t operator()(const IQueryTreeNode * node) const + explicit QueryTreeNodeWithHash(const IQueryTreeNode * node_) + : node(node_) + , hash(node->getTreeHash().first) + {} + + const IQueryTreeNode * node = nullptr; + size_t hash = 0; +}; + +struct QueryTreeNodeWithHashHash +{ + size_t operator()(const QueryTreeNodeWithHash & node_with_hash) const { - return node->getTreeHash().first; + return node_with_hash.hash; } }; -struct QueryTreeNodeEqualTo +struct QueryTreeNodeWithHashEqualTo { - size_t operator()(const IQueryTreeNode * lhs_node, const IQueryTreeNode * rhs_node) const + bool operator()(const QueryTreeNodeWithHash & lhs_node, const QueryTreeNodeWithHash & rhs_node) const { - return lhs_node->isEqual(*rhs_node); + return lhs_node.hash == rhs_node.hash && lhs_node.node->isEqual(*rhs_node.node); } }; -using QueryTreeNodeSet = std::unordered_set; +using QueryTreeNodeWithHashSet = std::unordered_set; class OrderByLimitByDuplicateEliminationVisitor : public InDepthQueryTreeVisitor { @@ -82,7 +93,7 @@ public: } private: - QueryTreeNodeSet unique_expressions_nodes_set; + QueryTreeNodeWithHashSet unique_expressions_nodes_set; }; } diff --git a/src/Analyzer/Passes/QueryAnalysisPass.cpp b/src/Analyzer/Passes/QueryAnalysisPass.cpp index 6b91d0f8053..9db2d66d99d 100644 --- a/src/Analyzer/Passes/QueryAnalysisPass.cpp +++ b/src/Analyzer/Passes/QueryAnalysisPass.cpp @@ -1621,34 +1621,7 @@ void QueryAnalyzer::validateTableExpressionModifiers(const QueryTreeNodePtr & ta table_expression_node->formatASTForErrorMessage(), scope.scope_node->formatASTForErrorMessage()); - if (query_node || union_node) - { - auto table_expression_modifiers = query_node ? query_node->getTableExpressionModifiers() : union_node->getTableExpressionModifiers(); - - if (table_expression_modifiers.has_value()) - { - String table_expression_modifiers_error_message; - - if (table_expression_modifiers->hasFinal()) - { - table_expression_modifiers_error_message += "FINAL"; - - if (table_expression_modifiers->hasSampleSizeRatio()) - table_expression_modifiers_error_message += ", SAMPLE"; - } - else if (table_expression_modifiers->hasSampleSizeRatio()) - { - table_expression_modifiers_error_message += "SAMPLE"; - } - - throw Exception(ErrorCodes::UNSUPPORTED_METHOD, - "Table expression modifiers {} are not supported for subquery {}. In scope {}", - table_expression_modifiers_error_message, - table_expression_node->formatASTForErrorMessage(), - scope.scope_node->formatASTForErrorMessage()); - } - } - else if (table_node || table_function_node) + if (table_node || table_function_node) { auto table_expression_modifiers = table_node ? table_node->getTableExpressionModifiers() : table_function_node->getTableExpressionModifiers(); @@ -4661,17 +4634,23 @@ void QueryAnalyzer::initializeQueryJoinTreeNode(QueryTreeNodePtr & join_tree_nod auto table_expression_modifiers = from_table_identifier.getTableExpressionModifiers(); - if (auto * resolved_identifier_query_node = resolved_identifier->as()) + auto * resolved_identifier_query_node = resolved_identifier->as(); + auto * resolved_identifier_union_node = resolved_identifier->as(); + + if (resolved_identifier_query_node || resolved_identifier_union_node) { - resolved_identifier_query_node->setIsCTE(false); + if (resolved_identifier_query_node) + resolved_identifier_query_node->setIsCTE(false); + else + resolved_identifier_union_node->setIsCTE(false); + if (table_expression_modifiers.has_value()) - resolved_identifier_query_node->setTableExpressionModifiers(*table_expression_modifiers); - } - else if (auto * resolved_identifier_union_node = resolved_identifier->as()) - { - resolved_identifier_union_node->setIsCTE(false); - if (table_expression_modifiers.has_value()) - resolved_identifier_union_node->setTableExpressionModifiers(*table_expression_modifiers); + { + throw Exception(ErrorCodes::UNSUPPORTED_METHOD, + "Table expression modifiers {} are not supported for subquery {}", + table_expression_modifiers->formatForErrorMessage(), + resolved_identifier->formatASTForErrorMessage()); + } } else if (auto * resolved_identifier_table_node = resolved_identifier->as()) { diff --git a/src/Analyzer/QueryNode.cpp b/src/Analyzer/QueryNode.cpp index 557baad2654..c5bbc193544 100644 --- a/src/Analyzer/QueryNode.cpp +++ b/src/Analyzer/QueryNode.cpp @@ -74,12 +74,6 @@ void QueryNode::dumpTreeImpl(WriteBuffer & buffer, FormatState & format_state, s buffer << ", constant_value_type: " << constant_value->getType()->getName(); } - if (table_expression_modifiers) - { - buffer << ", "; - table_expression_modifiers->dump(buffer); - } - if (hasWith()) { buffer << '\n' << std::string(indent + 2, ' ') << "WITH\n"; @@ -195,13 +189,6 @@ bool QueryNode::isEqualImpl(const IQueryTreeNode & rhs) const else if (!constant_value && rhs_typed.constant_value) return false; - if (table_expression_modifiers && rhs_typed.table_expression_modifiers && table_expression_modifiers != rhs_typed.table_expression_modifiers) - return false; - else if (table_expression_modifiers && !rhs_typed.table_expression_modifiers) - return false; - else if (!table_expression_modifiers && rhs_typed.table_expression_modifiers) - return false; - return is_subquery == rhs_typed.is_subquery && is_cte == rhs_typed.is_cte && cte_name == rhs_typed.cte_name && @@ -250,9 +237,6 @@ void QueryNode::updateTreeHashImpl(HashState & state) const state.update(constant_value_type_name.size()); state.update(constant_value_type_name); } - - if (table_expression_modifiers) - table_expression_modifiers->updateTreeHash(state); } QueryTreeNodePtr QueryNode::cloneImpl() const @@ -270,7 +254,6 @@ QueryTreeNodePtr QueryNode::cloneImpl() const result_query_node->cte_name = cte_name; result_query_node->projection_columns = projection_columns; result_query_node->constant_value = constant_value; - result_query_node->table_expression_modifiers = table_expression_modifiers; return result_query_node; } diff --git a/src/Analyzer/QueryNode.h b/src/Analyzer/QueryNode.h index 6bb6613fc2b..1bb381c95c9 100644 --- a/src/Analyzer/QueryNode.h +++ b/src/Analyzer/QueryNode.h @@ -176,24 +176,6 @@ public: is_group_by_with_grouping_sets = is_group_by_with_grouping_sets_value; } - /// Return true if query node has table expression modifiers, false otherwise - bool hasTableExpressionModifiers() const - { - return table_expression_modifiers.has_value(); - } - - /// Get table expression modifiers - const std::optional & getTableExpressionModifiers() const - { - return table_expression_modifiers; - } - - /// Set table expression modifiers - void setTableExpressionModifiers(TableExpressionModifiers table_expression_modifiers_value) - { - table_expression_modifiers = std::move(table_expression_modifiers_value); - } - /// Returns true if query node WITH section is not empty, false otherwise bool hasWith() const { @@ -602,7 +584,6 @@ private: std::string cte_name; NamesAndTypes projection_columns; ConstantValuePtr constant_value; - std::optional table_expression_modifiers; SettingsChanges settings_changes; static constexpr size_t with_child_index = 0; diff --git a/src/Analyzer/QueryTreeBuilder.cpp b/src/Analyzer/QueryTreeBuilder.cpp index 890aa2b01a2..51745d820e7 100644 --- a/src/Analyzer/QueryTreeBuilder.cpp +++ b/src/Analyzer/QueryTreeBuilder.cpp @@ -145,12 +145,10 @@ QueryTreeNodePtr QueryTreeBuilder::buildSelectWithUnionExpression(const ASTPtr & if (select_lists.children.size() == 1) return buildSelectOrUnionExpression(select_lists.children[0], is_subquery, cte_name); - auto union_node = std::make_shared(); + auto union_node = std::make_shared(select_with_union_query_typed.union_mode); union_node->setIsSubquery(is_subquery); union_node->setIsCTE(!cte_name.empty()); union_node->setCTEName(cte_name); - union_node->setUnionMode(select_with_union_query_typed.union_mode); - union_node->setUnionModes(select_with_union_query_typed.list_of_modes); union_node->setOriginalAST(select_with_union_query); size_t select_lists_children_size = select_lists.children.size(); @@ -173,23 +171,22 @@ QueryTreeNodePtr QueryTreeBuilder::buildSelectIntersectExceptQuery(const ASTPtr if (select_lists.size() == 1) return buildSelectExpression(select_lists[0], is_subquery, cte_name); - auto union_node = std::make_shared(); - union_node->setIsSubquery(is_subquery); - union_node->setIsCTE(!cte_name.empty()); - union_node->setCTEName(cte_name); - + SelectUnionMode union_mode; if (select_intersect_except_query_typed.final_operator == ASTSelectIntersectExceptQuery::Operator::INTERSECT_ALL) - union_node->setUnionMode(SelectUnionMode::INTERSECT_ALL); + union_mode = SelectUnionMode::INTERSECT_ALL; else if (select_intersect_except_query_typed.final_operator == ASTSelectIntersectExceptQuery::Operator::INTERSECT_DISTINCT) - union_node->setUnionMode(SelectUnionMode::INTERSECT_DISTINCT); + union_mode = SelectUnionMode::INTERSECT_DISTINCT; else if (select_intersect_except_query_typed.final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_ALL) - union_node->setUnionMode(SelectUnionMode::EXCEPT_ALL); + union_mode = SelectUnionMode::EXCEPT_ALL; else if (select_intersect_except_query_typed.final_operator == ASTSelectIntersectExceptQuery::Operator::EXCEPT_DISTINCT) - union_node->setUnionMode(SelectUnionMode::EXCEPT_DISTINCT); + union_mode = SelectUnionMode::EXCEPT_DISTINCT; else throw Exception(ErrorCodes::LOGICAL_ERROR, "UNION type is not initialized"); - union_node->setUnionModes(SelectUnionModes(select_lists.size() - 1, union_node->getUnionMode())); + auto union_node = std::make_shared(union_mode); + union_node->setIsSubquery(is_subquery); + union_node->setIsCTE(!cte_name.empty()); + union_node->setCTEName(cte_name); union_node->setOriginalAST(select_intersect_except_query); size_t select_lists_size = select_lists.size(); @@ -676,14 +673,10 @@ QueryTreeNodePtr QueryTreeBuilder::buildJoinTree(const ASTPtr & tables_in_select if (table_expression_modifiers) { - if (auto * query_node = node->as()) - query_node->setTableExpressionModifiers(*table_expression_modifiers); - else if (auto * union_node = node->as()) - union_node->setTableExpressionModifiers(*table_expression_modifiers); - else - throw Exception(ErrorCodes::LOGICAL_ERROR, - "Unexpected table expression subquery node. Expected union or query. Actual {}", - node->formatASTForErrorMessage()); + throw Exception(ErrorCodes::UNSUPPORTED_METHOD, + "Table expression modifiers {} are not supported for subquery {}", + table_expression_modifiers->formatForErrorMessage(), + node->formatASTForErrorMessage()); } table_expressions.push_back(std::move(node)); diff --git a/src/Analyzer/TableExpressionModifiers.cpp b/src/Analyzer/TableExpressionModifiers.cpp index 79b5a8dba41..c8002f44c97 100644 --- a/src/Analyzer/TableExpressionModifiers.cpp +++ b/src/Analyzer/TableExpressionModifiers.cpp @@ -5,6 +5,7 @@ #include #include #include +#include namespace DB { @@ -39,4 +40,27 @@ void TableExpressionModifiers::updateTreeHash(SipHash & hash_state) const } } +String TableExpressionModifiers::formatForErrorMessage() const +{ + WriteBufferFromOwnString buffer; + if (has_final) + buffer << "FINAL"; + + if (sample_size_ratio) + { + if (has_final) + buffer << ' '; + buffer << "SAMPLE " << ASTSampleRatio::toString(*sample_size_ratio); + } + + if (sample_offset_ratio) + { + if (has_final || sample_size_ratio) + buffer << ' '; + buffer << "OFFSET " << ASTSampleRatio::toString(*sample_offset_ratio); + } + + return buffer.str(); +} + } diff --git a/src/Analyzer/TableExpressionModifiers.h b/src/Analyzer/TableExpressionModifiers.h index cc5ac3948bf..f61c2a61610 100644 --- a/src/Analyzer/TableExpressionModifiers.h +++ b/src/Analyzer/TableExpressionModifiers.h @@ -58,6 +58,9 @@ public: /// Update tree hash void updateTreeHash(SipHash & hash_state) const; + /// Format for error message + String formatForErrorMessage() const; + private: bool has_final = false; std::optional sample_size_ratio; diff --git a/src/Analyzer/UnionNode.cpp b/src/Analyzer/UnionNode.cpp index b8ed46c645e..67860438335 100644 --- a/src/Analyzer/UnionNode.cpp +++ b/src/Analyzer/UnionNode.cpp @@ -30,11 +30,18 @@ namespace DB namespace ErrorCodes { extern const int TYPE_MISMATCH; + extern const int BAD_ARGUMENTS; } -UnionNode::UnionNode() +UnionNode::UnionNode(SelectUnionMode union_mode_) : IQueryTreeNode(children_size) + , union_mode(union_mode_) { + if (union_mode == SelectUnionMode::UNION_DEFAULT || + union_mode == SelectUnionMode::EXCEPT_DEFAULT || + union_mode == SelectUnionMode::INTERSECT_DEFAULT) + throw Exception(ErrorCodes::BAD_ARGUMENTS, "UNION mode {} must be normalized", toString(union_mode)); + children[queries_child_index] = std::make_shared(); } @@ -101,28 +108,8 @@ void UnionNode::dumpTreeImpl(WriteBuffer & buffer, FormatState & format_state, s buffer << ", constant_value_type: " << constant_value->getType()->getName(); } - if (table_expression_modifiers) - { - buffer << ", "; - table_expression_modifiers->dump(buffer); - } - buffer << ", union_mode: " << toString(union_mode); - size_t union_modes_size = union_modes.size(); - buffer << '\n' << std::string(indent + 2, ' ') << "UNION MODES " << union_modes_size << '\n'; - - for (size_t i = 0; i < union_modes_size; ++i) - { - buffer << std::string(indent + 4, ' '); - - auto query_union_mode = union_modes[i]; - buffer << toString(query_union_mode); - - if (i + 1 != union_modes_size) - buffer << '\n'; - } - buffer << '\n' << std::string(indent + 2, ' ') << "QUERIES\n"; getQueriesNode()->dumpTreeImpl(buffer, format_state, indent + 4); } @@ -137,15 +124,8 @@ bool UnionNode::isEqualImpl(const IQueryTreeNode & rhs) const else if (!constant_value && rhs_typed.constant_value) return false; - if (table_expression_modifiers && rhs_typed.table_expression_modifiers && table_expression_modifiers != rhs_typed.table_expression_modifiers) - return false; - else if (table_expression_modifiers && !rhs_typed.table_expression_modifiers) - return false; - else if (!table_expression_modifiers && rhs_typed.table_expression_modifiers) - return false; - return is_subquery == rhs_typed.is_subquery && is_cte == rhs_typed.is_cte && cte_name == rhs_typed.cte_name && - union_mode == rhs_typed.union_mode && union_modes == rhs_typed.union_modes; + union_mode == rhs_typed.union_mode; } void UnionNode::updateTreeHashImpl(HashState & state) const @@ -158,10 +138,6 @@ void UnionNode::updateTreeHashImpl(HashState & state) const state.update(static_cast(union_mode)); - state.update(union_modes.size()); - for (const auto & query_union_mode : union_modes) - state.update(static_cast(query_union_mode)); - if (constant_value) { auto constant_dump = applyVisitor(FieldVisitorToString(), constant_value->getValue()); @@ -172,23 +148,16 @@ void UnionNode::updateTreeHashImpl(HashState & state) const state.update(constant_value_type_name.size()); state.update(constant_value_type_name); } - - if (table_expression_modifiers) - table_expression_modifiers->updateTreeHash(state); } QueryTreeNodePtr UnionNode::cloneImpl() const { - auto result_union_node = std::make_shared(); + auto result_union_node = std::make_shared(union_mode); result_union_node->is_subquery = is_subquery; result_union_node->is_cte = is_cte; result_union_node->cte_name = cte_name; - result_union_node->union_mode = union_mode; - result_union_node->union_modes = union_modes; - result_union_node->union_modes_set = union_modes_set; result_union_node->constant_value = constant_value; - result_union_node->table_expression_modifiers = table_expression_modifiers; return result_union_node; } @@ -197,14 +166,7 @@ ASTPtr UnionNode::toASTImpl() const { auto select_with_union_query = std::make_shared(); select_with_union_query->union_mode = union_mode; - - if (union_mode != SelectUnionMode::UNION_DEFAULT && - union_mode != SelectUnionMode::EXCEPT_DEFAULT && - union_mode != SelectUnionMode::INTERSECT_DEFAULT) - select_with_union_query->is_normalized = true; - - select_with_union_query->list_of_modes = union_modes; - select_with_union_query->set_of_modes = union_modes_set; + select_with_union_query->is_normalized = true; select_with_union_query->children.push_back(getQueriesNode()->toAST()); select_with_union_query->list_of_selects = select_with_union_query->children.back(); diff --git a/src/Analyzer/UnionNode.h b/src/Analyzer/UnionNode.h index 05e70b87a27..9ef76591597 100644 --- a/src/Analyzer/UnionNode.h +++ b/src/Analyzer/UnionNode.h @@ -19,6 +19,7 @@ namespace ErrorCodes } /** Union node represents union of queries in query tree. + * Union node must be initialized with normalized union mode. * * Example: (SELECT id FROM test_table) UNION ALL (SELECT id FROM test_table_2); * Example: (SELECT id FROM test_table) UNION DISTINCT (SELECT id FROM test_table_2); @@ -41,7 +42,8 @@ using UnionNodePtr = std::shared_ptr; class UnionNode final : public IQueryTreeNode { public: - explicit UnionNode(); + /// Construct union node with normalized union mode + explicit UnionNode(SelectUnionMode union_mode_); /// Returns true if union node is subquery, false otherwise bool isSubquery() const @@ -85,25 +87,6 @@ public: return union_mode; } - /// Set union mode value - void setUnionMode(SelectUnionMode union_mode_value) - { - union_mode = union_mode_value; - } - - /// Get union modes - const SelectUnionModes & getUnionModes() const - { - return union_modes; - } - - /// Set union modes value - void setUnionModes(const SelectUnionModes & union_modes_value) - { - union_modes = union_modes_value; - union_modes_set = SelectUnionModesSet(union_modes.begin(), union_modes.end()); - } - /// Get union node queries const ListNode & getQueries() const { @@ -128,24 +111,6 @@ public: return children[queries_child_index]; } - /// Return true if union node has table expression modifiers, false otherwise - bool hasTableExpressionModifiers() const - { - return table_expression_modifiers.has_value(); - } - - /// Get table expression modifiers - const std::optional & getTableExpressionModifiers() const - { - return table_expression_modifiers; - } - - /// Set table expression modifiers - void setTableExpressionModifiers(TableExpressionModifiers table_expression_modifiers_value) - { - table_expression_modifiers = std::move(table_expression_modifiers_value); - } - /// Compute union node projection columns NamesAndTypes computeProjectionColumns() const; @@ -189,10 +154,7 @@ private: bool is_cte = false; std::string cte_name; SelectUnionMode union_mode; - SelectUnionModes union_modes; - SelectUnionModesSet union_modes_set; ConstantValuePtr constant_value; - std::optional table_expression_modifiers; static constexpr size_t queries_child_index = 0; static constexpr size_t children_size = queries_child_index + 1; diff --git a/src/Analyzer/Utils.cpp b/src/Analyzer/Utils.cpp index 5f0d682865f..b504a5b5787 100644 --- a/src/Analyzer/Utils.cpp +++ b/src/Analyzer/Utils.cpp @@ -98,11 +98,6 @@ static ASTPtr convertIntoTableExpressionAST(const QueryTreeNodePtr & table_expre if (node_type == QueryTreeNodeType::QUERY || node_type == QueryTreeNodeType::UNION) { - if (auto * query_node = table_expression_node->as()) - table_expression_modifiers = query_node->getTableExpressionModifiers(); - else if (auto * union_node = table_expression_node->as()) - table_expression_modifiers = union_node->getTableExpressionModifiers(); - result_table_expression->subquery = result_table_expression->children.back(); } else if (node_type == QueryTreeNodeType::TABLE || node_type == QueryTreeNodeType::IDENTIFIER) diff --git a/src/Columns/ColumnArray.h b/src/Columns/ColumnArray.h index 4e951ec28b8..5970802f598 100644 --- a/src/Columns/ColumnArray.h +++ b/src/Columns/ColumnArray.h @@ -176,6 +176,9 @@ public: void getIndicesOfNonDefaultRows(Offsets & indices, size_t from, size_t limit) const override; + void finalize() override { data->finalize(); } + bool isFinalized() const override { return data->isFinalized(); } + bool isCollationSupported() const override { return getData().isCollationSupported(); } size_t getNumberOfDimensions() const; diff --git a/src/Columns/ColumnMap.h b/src/Columns/ColumnMap.h index a3e171008ff..1e03633ced7 100644 --- a/src/Columns/ColumnMap.h +++ b/src/Columns/ColumnMap.h @@ -93,6 +93,8 @@ public: bool structureEquals(const IColumn & rhs) const override; double getRatioOfDefaultRows(double sample_ratio) const override; void getIndicesOfNonDefaultRows(Offsets & indices, size_t from, size_t limit) const override; + void finalize() override { nested->finalize(); } + bool isFinalized() const override { return nested->isFinalized(); } const ColumnArray & getNestedColumn() const { return assert_cast(*nested); } ColumnArray & getNestedColumn() { return assert_cast(*nested); } diff --git a/src/Columns/ColumnObject.cpp b/src/Columns/ColumnObject.cpp index 86586559ff7..bf4630137d5 100644 --- a/src/Columns/ColumnObject.cpp +++ b/src/Columns/ColumnObject.cpp @@ -732,8 +732,8 @@ void ColumnObject::get(size_t n, Field & res) const { assert(n < size()); res = Object(); - auto & object = res.get(); + for (const auto & entry : subcolumns) { auto it = object.try_emplace(entry->path.getPath()).first; @@ -744,7 +744,6 @@ void ColumnObject::get(size_t n, Field & res) const void ColumnObject::insertFrom(const IColumn & src, size_t n) { insert(src[n]); - finalize(); } void ColumnObject::insertRangeFrom(const IColumn & src, size_t start, size_t length) @@ -792,9 +791,8 @@ MutableColumnPtr ColumnObject::applyForSubcolumns(Func && func) const { if (!isFinalized()) { - auto finalized = IColumn::mutate(getPtr()); + auto finalized = cloneFinalized(); auto & finalized_object = assert_cast(*finalized); - finalized_object.finalize(); return finalized_object.applyForSubcolumns(std::forward(func)); } diff --git a/src/Columns/ColumnObject.h b/src/Columns/ColumnObject.h index f32356fed6e..8fcf3d41fba 100644 --- a/src/Columns/ColumnObject.h +++ b/src/Columns/ColumnObject.h @@ -198,10 +198,6 @@ public: Subcolumns & getSubcolumns() { return subcolumns; } PathsInData getKeys() const; - /// Finalizes all subcolumns. - void finalize(); - bool isFinalized() const; - /// Part of interface const char * getFamilyName() const override { return "Object"; } @@ -219,12 +215,17 @@ public: void popBack(size_t length) override; Field operator[](size_t n) const override; void get(size_t n, Field & res) const override; + ColumnPtr permute(const Permutation & perm, size_t limit) const override; ColumnPtr filter(const Filter & filter, ssize_t result_size_hint) const override; ColumnPtr index(const IColumn & indexes, size_t limit) const override; ColumnPtr replicate(const Offsets & offsets) const override; MutableColumnPtr cloneResized(size_t new_size) const override; + /// Finalizes all subcolumns. + void finalize() override; + bool isFinalized() const override; + /// Order of rows in ColumnObject is undefined. void getPermutation(PermutationSortDirection, PermutationSortStability, size_t, int, Permutation & res) const override; void compareColumn(const IColumn & rhs, size_t rhs_row_num, @@ -264,9 +265,7 @@ private: template MutableColumnPtr applyForSubcolumns(Func && func) const; - /// For given subcolumn return subcolumn from the same Nested type. /// It's used to get shared sized of Nested to insert correct default values. const Subcolumns::Node * getLeafOfTheSameNested(const Subcolumns::NodePtr & entry) const; }; - } diff --git a/src/Columns/ColumnTuple.cpp b/src/Columns/ColumnTuple.cpp index 3577b6dee28..d8a43bf510d 100644 --- a/src/Columns/ColumnTuple.cpp +++ b/src/Columns/ColumnTuple.cpp @@ -570,4 +570,15 @@ void ColumnTuple::getIndicesOfNonDefaultRows(Offsets & indices, size_t from, siz return getIndicesOfNonDefaultRowsImpl(indices, from, limit); } +void ColumnTuple::finalize() +{ + for (auto & column : columns) + column->finalize(); +} + +bool ColumnTuple::isFinalized() const +{ + return std::all_of(columns.begin(), columns.end(), [](const auto & column) { return column->isFinalized(); }); +} + } diff --git a/src/Columns/ColumnTuple.h b/src/Columns/ColumnTuple.h index 385de7db1e7..96395d4edfb 100644 --- a/src/Columns/ColumnTuple.h +++ b/src/Columns/ColumnTuple.h @@ -103,6 +103,8 @@ public: ColumnPtr compress() const override; double getRatioOfDefaultRows(double sample_ratio) const override; void getIndicesOfNonDefaultRows(Offsets & indices, size_t from, size_t limit) const override; + void finalize() override; + bool isFinalized() const override; size_t tupleSize() const { return columns.size(); } diff --git a/src/Columns/IColumn.h b/src/Columns/IColumn.h index 19f3dea4f82..461e41e3eec 100644 --- a/src/Columns/IColumn.h +++ b/src/Columns/IColumn.h @@ -85,8 +85,8 @@ public: [[nodiscard]] virtual MutablePtr cloneEmpty() const { return cloneResized(0); } /// Creates column with the same type and specified size. - /// If size is less current size, then data is cut. - /// If size is greater, than default values are appended. + /// If size is less than current size, then data is cut. + /// If size is greater, then default values are appended. [[nodiscard]] virtual MutablePtr cloneResized(size_t /*size*/) const { throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Cannot cloneResized() column {}", getName()); } /// Returns number of values in column. @@ -453,6 +453,16 @@ public: return getPtr(); } + /// Some columns may require finalization before using of other operations. + virtual void finalize() {} + virtual bool isFinalized() const { return true; } + + MutablePtr cloneFinalized() const + { + auto finalized = IColumn::mutate(getPtr()); + finalized->finalize(); + return finalized; + } [[nodiscard]] static MutablePtr mutate(Ptr ptr) { diff --git a/src/Common/CaresPTRResolver.cpp b/src/Common/CaresPTRResolver.cpp index a02909309b6..99b4c34dfbd 100644 --- a/src/Common/CaresPTRResolver.cpp +++ b/src/Common/CaresPTRResolver.cpp @@ -2,6 +2,7 @@ #include #include #include +#include #include "ares.h" #include "netdb.h" @@ -40,6 +41,8 @@ namespace DB } } + std::mutex CaresPTRResolver::mutex; + CaresPTRResolver::CaresPTRResolver(CaresPTRResolver::provider_token) : channel(nullptr) { /* @@ -73,6 +76,8 @@ namespace DB std::unordered_set CaresPTRResolver::resolve(const std::string & ip) { + std::lock_guard guard(mutex); + std::unordered_set ptr_records; resolve(ip, ptr_records); @@ -83,6 +88,8 @@ namespace DB std::unordered_set CaresPTRResolver::resolve_v6(const std::string & ip) { + std::lock_guard guard(mutex); + std::unordered_set ptr_records; resolve_v6(ip, ptr_records); @@ -110,23 +117,83 @@ namespace DB void CaresPTRResolver::wait() { - timeval * tvp, tv; - fd_set read_fds; - fd_set write_fds; - int nfds; + int sockets[ARES_GETSOCK_MAXNUM]; + pollfd pollfd[ARES_GETSOCK_MAXNUM]; - for (;;) + while (true) { - FD_ZERO(&read_fds); - FD_ZERO(&write_fds); - nfds = ares_fds(channel, &read_fds,&write_fds); - if (nfds == 0) + auto readable_sockets = get_readable_sockets(sockets, pollfd); + auto timeout = calculate_timeout(); + + int number_of_fds_ready = 0; + if (!readable_sockets.empty()) + { + number_of_fds_ready = poll(readable_sockets.data(), static_cast(readable_sockets.size()), static_cast(timeout)); + } + + if (number_of_fds_ready > 0) + { + process_readable_sockets(readable_sockets); + } + else + { + process_possible_timeout(); + break; + } + } + } + + std::span CaresPTRResolver::get_readable_sockets(int * sockets, pollfd * pollfd) + { + int sockets_bitmask = ares_getsock(channel, sockets, ARES_GETSOCK_MAXNUM); + + int number_of_sockets_to_poll = 0; + + for (int i = 0; i < ARES_GETSOCK_MAXNUM; i++, number_of_sockets_to_poll++) + { + pollfd[i].events = 0; + pollfd[i].revents = 0; + + if (ARES_GETSOCK_READABLE(sockets_bitmask, i)) + { + pollfd[i].fd = sockets[i]; + pollfd[i].events = POLLIN; + } + else { break; } - tvp = ares_timeout(channel, nullptr, &tv); - select(nfds, &read_fds, &write_fds, nullptr, tvp); - ares_process(channel, &read_fds, &write_fds); + } + + return std::span(pollfd, number_of_sockets_to_poll); + } + + int64_t CaresPTRResolver::calculate_timeout() + { + timeval tv; + if (auto * tvp = ares_timeout(channel, nullptr, &tv)) + { + auto timeout = tvp->tv_sec * 1000 + tvp->tv_usec / 1000; + + return timeout; + } + + return 0; + } + + void CaresPTRResolver::process_possible_timeout() + { + /* Call ares_process() unconditonally here, even if we simply timed out + above, as otherwise the ares name resolve won't timeout! */ + ares_process_fd(channel, ARES_SOCKET_BAD, ARES_SOCKET_BAD); + } + + void CaresPTRResolver::process_readable_sockets(std::span readable_sockets) + { + for (auto readable_socket : readable_sockets) + { + auto fd = readable_socket.revents & POLLIN ? readable_socket.fd : ARES_SOCKET_BAD; + ares_process_fd(channel, fd, ARES_SOCKET_BAD); } } } diff --git a/src/Common/CaresPTRResolver.h b/src/Common/CaresPTRResolver.h index e5182d34682..9df6d7aeb72 100644 --- a/src/Common/CaresPTRResolver.h +++ b/src/Common/CaresPTRResolver.h @@ -1,5 +1,8 @@ #pragma once +#include +#include +#include #include "DNSPTRResolver.h" using ares_channel = struct ares_channeldata *; @@ -20,7 +23,6 @@ namespace DB * Allow only DNSPTRProvider to instantiate this class * */ struct provider_token {}; - public: explicit CaresPTRResolver(provider_token); ~CaresPTRResolver() override; @@ -36,7 +38,17 @@ namespace DB void resolve_v6(const std::string & ip, std::unordered_set & response); + std::span get_readable_sockets(int * sockets, pollfd * pollfd); + + int64_t calculate_timeout(); + + void process_possible_timeout(); + + void process_readable_sockets(std::span readable_sockets); + ares_channel channel; + + static std::mutex mutex; }; } diff --git a/src/Common/DNSPTRResolverProvider.cpp b/src/Common/DNSPTRResolverProvider.cpp index 97d601a3a78..91ce4dbb938 100644 --- a/src/Common/DNSPTRResolverProvider.cpp +++ b/src/Common/DNSPTRResolverProvider.cpp @@ -5,8 +5,10 @@ namespace DB { std::shared_ptr DNSPTRResolverProvider::get() { - return std::make_shared( + static auto resolver = std::make_shared( CaresPTRResolver::provider_token {} ); + + return resolver; } } diff --git a/src/Common/JSONParsers/DummyJSONParser.h b/src/Common/JSONParsers/DummyJSONParser.h index 3cedd59decd..50c112affe2 100644 --- a/src/Common/JSONParsers/DummyJSONParser.h +++ b/src/Common/JSONParsers/DummyJSONParser.h @@ -3,6 +3,7 @@ #include #include #include +#include "ElementTypes.h" namespace DB @@ -25,6 +26,7 @@ struct DummyJSONParser { public: Element() = default; + static ElementType type() { return ElementType::NULL_VALUE; } static bool isInt64() { return false; } static bool isUInt64() { return false; } static bool isDouble() { return false; } diff --git a/src/Common/JSONParsers/ElementTypes.h b/src/Common/JSONParsers/ElementTypes.h new file mode 100644 index 00000000000..44e4c850a2f --- /dev/null +++ b/src/Common/JSONParsers/ElementTypes.h @@ -0,0 +1,17 @@ +#pragma once + +namespace DB +{ +// Enum values match simdjson's for fast conversion +enum class ElementType +{ + ARRAY = '[', + OBJECT = '{', + INT64 = 'l', + UINT64 = 'u', + DOUBLE = 'd', + STRING = '"', + BOOL = 't', + NULL_VALUE = 'n' +}; +} diff --git a/src/Common/JSONParsers/RapidJSONParser.h b/src/Common/JSONParsers/RapidJSONParser.h index 01730bc0692..6c5ea938bfe 100644 --- a/src/Common/JSONParsers/RapidJSONParser.h +++ b/src/Common/JSONParsers/RapidJSONParser.h @@ -6,7 +6,7 @@ # include # include # include - +# include "ElementTypes.h" namespace DB { @@ -26,6 +26,20 @@ struct RapidJSONParser ALWAYS_INLINE Element() = default; ALWAYS_INLINE Element(const rapidjson::Value & value_) : ptr(&value_) {} /// NOLINT + ALWAYS_INLINE ElementType type() const + { + switch (ptr->GetType()) + { + case rapidjson::kNumberType: return ptr->IsDouble() ? ElementType::DOUBLE : (ptr->IsUint64() ? ElementType::UINT64 : ElementType::INT64); + case rapidjson::kStringType: return ElementType::STRING; + case rapidjson::kArrayType: return ElementType::ARRAY; + case rapidjson::kObjectType: return ElementType::OBJECT; + case rapidjson::kTrueType: return ElementType::BOOL; + case rapidjson::kFalseType: return ElementType::BOOL; + case rapidjson::kNullType: return ElementType::NULL_VALUE; + } + } + ALWAYS_INLINE bool isInt64() const { return ptr->IsInt64(); } ALWAYS_INLINE bool isUInt64() const { return ptr->IsUint64(); } ALWAYS_INLINE bool isDouble() const { return ptr->IsDouble(); } diff --git a/src/Common/JSONParsers/SimdJSONParser.h b/src/Common/JSONParsers/SimdJSONParser.h index 14eb3cd6d78..f0f8f91109f 100644 --- a/src/Common/JSONParsers/SimdJSONParser.h +++ b/src/Common/JSONParsers/SimdJSONParser.h @@ -7,7 +7,7 @@ # include # include # include - +# include "ElementTypes.h" namespace DB { @@ -31,6 +31,21 @@ struct SimdJSONParser ALWAYS_INLINE Element() {} /// NOLINT ALWAYS_INLINE Element(const simdjson::dom::element & element_) : element(element_) {} /// NOLINT + ALWAYS_INLINE ElementType type() const + { + switch (element.type()) + { + case simdjson::dom::element_type::INT64: return ElementType::INT64; + case simdjson::dom::element_type::UINT64: return ElementType::UINT64; + case simdjson::dom::element_type::DOUBLE: return ElementType::DOUBLE; + case simdjson::dom::element_type::STRING: return ElementType::STRING; + case simdjson::dom::element_type::ARRAY: return ElementType::ARRAY; + case simdjson::dom::element_type::OBJECT: return ElementType::OBJECT; + case simdjson::dom::element_type::BOOL: return ElementType::BOOL; + case simdjson::dom::element_type::NULL_VALUE: return ElementType::NULL_VALUE; + } + } + ALWAYS_INLINE bool isInt64() const { return element.type() == simdjson::dom::element_type::INT64; } ALWAYS_INLINE bool isUInt64() const { return element.type() == simdjson::dom::element_type::UINT64; } ALWAYS_INLINE bool isDouble() const { return element.type() == simdjson::dom::element_type::DOUBLE; } diff --git a/src/Common/MemoryTracker.cpp b/src/Common/MemoryTracker.cpp index 8bd31681706..b530410ec63 100644 --- a/src/Common/MemoryTracker.cpp +++ b/src/Common/MemoryTracker.cpp @@ -10,6 +10,7 @@ #include #include #include +#include #include #include "config.h" @@ -86,6 +87,8 @@ inline std::string_view toDescription(OvercommitResult result) namespace ProfileEvents { extern const Event QueryMemoryLimitExceeded; + extern const Event MemoryAllocatorPurge; + extern const Event MemoryAllocatorPurgeTimeMicroseconds; } using namespace std::chrono_literals; @@ -229,7 +232,10 @@ void MemoryTracker::allocImpl(Int64 size, bool throw_if_memory_exceeded, MemoryT { if (free_memory_in_allocator_arenas.exchange(-current_free_memory_in_allocator_arenas) > 0) { + Stopwatch watch; mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".purge", nullptr, nullptr, nullptr, 0); + ProfileEvents::increment(ProfileEvents::MemoryAllocatorPurge); + ProfileEvents::increment(ProfileEvents::MemoryAllocatorPurgeTimeMicroseconds, watch.elapsedMicroseconds()); } } @@ -432,7 +438,7 @@ void MemoryTracker::reset() void MemoryTracker::setRSS(Int64 rss_, Int64 free_memory_in_allocator_arenas_) { - Int64 new_amount = rss_; // - free_memory_in_allocator_arenas_; + Int64 new_amount = rss_; total_memory_tracker.amount.store(new_amount, std::memory_order_relaxed); free_memory_in_allocator_arenas.store(free_memory_in_allocator_arenas_, std::memory_order_relaxed); diff --git a/src/Common/ProfileEvents.cpp b/src/Common/ProfileEvents.cpp index 46bec669626..2f801e496fa 100644 --- a/src/Common/ProfileEvents.cpp +++ b/src/Common/ProfileEvents.cpp @@ -229,6 +229,8 @@ The server successfully detected this situation and will download merged part fr M(UserTimeMicroseconds, "Total time spent in processing (queries and other tasks) threads executing CPU instructions in user space. This include time CPU pipeline was stalled due to cache misses, branch mispredictions, hyper-threading, etc.") \ M(SystemTimeMicroseconds, "Total time spent in processing (queries and other tasks) threads executing CPU instructions in OS kernel space. This include time CPU pipeline was stalled due to cache misses, branch mispredictions, hyper-threading, etc.") \ M(MemoryOvercommitWaitTimeMicroseconds, "Total time spent in waiting for memory to be freed in OvercommitTracker.") \ + M(MemoryAllocatorPurge, "Total number of times memory allocator purge was requested") \ + M(MemoryAllocatorPurgeTimeMicroseconds, "Total number of times memory allocator purge was requested") \ M(SoftPageFaults, "The number of soft page faults in query execution threads. Soft page fault usually means a miss in the memory allocator cache which required a new memory mapping from the OS and subsequent allocation of a page of physical memory.") \ M(HardPageFaults, "The number of hard page faults in query execution threads. High values indicate either that you forgot to turn off swap on your server, or eviction of memory pages of the ClickHouse binary during very high memory pressure, or successful usage of the 'mmap' read method for the tables data.") \ \ diff --git a/src/Common/Throttler.cpp b/src/Common/Throttler.cpp index 2c9279e21e1..b38777efc03 100644 --- a/src/Common/Throttler.cpp +++ b/src/Common/Throttler.cpp @@ -3,7 +3,6 @@ #include #include #include -#include namespace ProfileEvents { @@ -21,63 +20,56 @@ namespace ErrorCodes /// Just 10^9. static constexpr auto NS = 1000000000UL; -/// Tracking window. Actually the size is not really important. We just want to avoid -/// throttles when there are no actions for a long period time. -static const double window_ns = 1ULL * NS; +static const size_t default_burst_seconds = 1; + +Throttler::Throttler(size_t max_speed_, const std::shared_ptr & parent_) + : max_speed(max_speed_) + , max_burst(max_speed_ * default_burst_seconds) + , limit_exceeded_exception_message("") + , tokens(max_burst) + , parent(parent_) +{} + +Throttler::Throttler(size_t max_speed_, size_t limit_, const char * limit_exceeded_exception_message_, + const std::shared_ptr & parent_) + : max_speed(max_speed_) + , max_burst(max_speed_ * default_burst_seconds) + , limit(limit_) + , limit_exceeded_exception_message(limit_exceeded_exception_message_) + , tokens(max_burst) + , parent(parent_) +{} void Throttler::add(size_t amount) { - size_t new_count; - /// This outer variable is always equal to smoothed_speed. - /// We use to avoid race condition. - double current_speed = 0; - + // Values obtained under lock to be checked after release + size_t count_value; + double tokens_value; { std::lock_guard lock(mutex); - auto now = clock_gettime_ns_adjusted(prev_ns); - /// If prev_ns is equal to zero (first `add` call) we known nothing about speed - /// and don't track anything. - if (max_speed && prev_ns != 0) + if (max_speed) { - /// Time spent to process the amount of bytes - double time_spent = now - prev_ns; - - /// The speed in bytes per second is equal to amount / time_spent in seconds - auto new_speed = amount / (time_spent / NS); - - /// We want to make old values of speed less important for our smoothed value - /// so we decay it's value with coef. - auto decay_coeff = std::pow(0.5, time_spent / window_ns); - - /// Weighted average between previous and new speed - smoothed_speed = smoothed_speed * decay_coeff + (1 - decay_coeff) * new_speed; - current_speed = smoothed_speed; + double delta_seconds = prev_ns ? static_cast(now - prev_ns) / NS : 0; + tokens = std::min(tokens + max_speed * delta_seconds - amount, max_burst); } - count += amount; - new_count = count; + count_value = count; + tokens_value = tokens; prev_ns = now; } - if (limit && new_count > limit) + if (limit && count_value > limit) throw Exception(limit_exceeded_exception_message + std::string(" Maximum: ") + toString(limit), ErrorCodes::LIMIT_EXCEEDED); - if (max_speed && current_speed > max_speed) + /// Wait unless there is positive amount of tokens - throttling + if (max_speed && tokens_value < 0) { - /// If we was too fast then we have to sleep until our smoothed speed became <= max_speed - int64_t sleep_time = static_cast(-window_ns * std::log2(max_speed / current_speed)); - - if (sleep_time > 0) - { - accumulated_sleep += sleep_time; - - sleepForNanoseconds(sleep_time); - - accumulated_sleep -= sleep_time; - - ProfileEvents::increment(ProfileEvents::ThrottlerSleepMicroseconds, sleep_time / 1000UL); - } + int64_t sleep_time = static_cast(-tokens_value / max_speed * NS); + accumulated_sleep += sleep_time; + sleepForNanoseconds(sleep_time); + accumulated_sleep -= sleep_time; + ProfileEvents::increment(ProfileEvents::ThrottlerSleepMicroseconds, sleep_time / 1000UL); } if (parent) @@ -89,9 +81,9 @@ void Throttler::reset() std::lock_guard lock(mutex); count = 0; - accumulated_sleep = 0; - smoothed_speed = 0; + tokens = max_burst; prev_ns = 0; + // NOTE: do not zero `accumulated_sleep` to avoid races } bool Throttler::isThrottling() const diff --git a/src/Common/Throttler.h b/src/Common/Throttler.h index 6d44ad6ca5f..9b6eff13506 100644 --- a/src/Common/Throttler.h +++ b/src/Common/Throttler.h @@ -10,25 +10,26 @@ namespace DB { -/** Allows you to limit the speed of something (in entities per second) using sleep. - * Specifics of work: - * Tracks exponentially (pow of 1/2) smoothed speed with hardcoded window. - * See more comments in .cpp file. - * - * Also allows you to set a limit on the maximum number of entities. If exceeded, an exception will be thrown. +/** Allows you to limit the speed of something (in tokens per second) using sleep. + * Implemented using Token Bucket Throttling algorithm. + * Also allows you to set a limit on the maximum number of tokens. If exceeded, an exception will be thrown. */ class Throttler { public: - explicit Throttler(size_t max_speed_, const std::shared_ptr & parent_ = nullptr) - : max_speed(max_speed_), limit_exceeded_exception_message(""), parent(parent_) {} + Throttler(size_t max_speed_, size_t max_burst_, const std::shared_ptr & parent_ = nullptr) + : max_speed(max_speed_), max_burst(max_burst_), limit_exceeded_exception_message(""), tokens(max_burst), parent(parent_) {} + + explicit Throttler(size_t max_speed_, const std::shared_ptr & parent_ = nullptr); + + Throttler(size_t max_speed_, size_t max_burst_, size_t limit_, const char * limit_exceeded_exception_message_, + const std::shared_ptr & parent_ = nullptr) + : max_speed(max_speed_), max_burst(max_burst_), limit(limit_), limit_exceeded_exception_message(limit_exceeded_exception_message_), tokens(max_burst), parent(parent_) {} Throttler(size_t max_speed_, size_t limit_, const char * limit_exceeded_exception_message_, - const std::shared_ptr & parent_ = nullptr) - : max_speed(max_speed_), limit(limit_), limit_exceeded_exception_message(limit_exceeded_exception_message_), parent(parent_) {} + const std::shared_ptr & parent_ = nullptr); - /// Calculates the smoothed speed, sleeps if required and throws exception on - /// limit overflow. + /// Use `amount` tokens, sleeps if required or throws exception on limit overflow. void add(size_t amount); /// Not thread safe @@ -45,15 +46,14 @@ public: private: size_t count{0}; - const size_t max_speed{0}; - const uint64_t limit{0}; /// 0 - not limited. + const size_t max_speed{0}; /// in tokens per second. + const size_t max_burst{0}; /// in tokens. + const uint64_t limit{0}; /// 0 - not limited. const char * limit_exceeded_exception_message = nullptr; std::mutex mutex; - std::atomic accumulated_sleep{0}; - /// Smoothed value of current speed. Updated in `add` method. - double smoothed_speed{0}; - /// previous `add` call time (in nanoseconds) - uint64_t prev_ns{0}; + std::atomic accumulated_sleep{0}; // Accumulated sleep time over all waiting threads + double tokens{0}; /// Amount of tokens available in token bucket. Updated in `add` method. + uint64_t prev_ns{0}; /// Previous `add` call time (in nanoseconds). /// Used to implement a hierarchy of throttlers std::shared_ptr parent; diff --git a/src/Common/formatIPv6.h b/src/Common/formatIPv6.h index 83b9d6e9fb1..31d5e83760a 100644 --- a/src/Common/formatIPv6.h +++ b/src/Common/formatIPv6.h @@ -5,6 +5,7 @@ #include #include #include +#include #include #include @@ -55,8 +56,11 @@ inline bool parseIPv4(const char * src, unsigned char * dst) } if (*(src - 1) != '\0') return false; - +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + reverseMemcpy(dst, &result, sizeof(result)); +#else memcpy(dst, &result, sizeof(result)); +#endif return true; } diff --git a/src/Compression/getCompressionCodecForFile.cpp b/src/Compression/getCompressionCodecForFile.cpp index f9365862c5b..027ee0ac57a 100644 --- a/src/Compression/getCompressionCodecForFile.cpp +++ b/src/Compression/getCompressionCodecForFile.cpp @@ -13,9 +13,9 @@ namespace DB using Checksum = CityHash_v1_0_2::uint128; -CompressionCodecPtr getCompressionCodecForFile(const DataPartStoragePtr & data_part_storage, const String & relative_path) +CompressionCodecPtr getCompressionCodecForFile(const IDataPartStorage & data_part_storage, const String & relative_path) { - auto read_buffer = data_part_storage->readFile(relative_path, {}, std::nullopt, std::nullopt); + auto read_buffer = data_part_storage.readFile(relative_path, {}, std::nullopt, std::nullopt); read_buffer->ignore(sizeof(Checksum)); UInt8 header_size = ICompressionCodec::getHeaderSize(); diff --git a/src/Compression/getCompressionCodecForFile.h b/src/Compression/getCompressionCodecForFile.h index ad855684128..b6f22750e4d 100644 --- a/src/Compression/getCompressionCodecForFile.h +++ b/src/Compression/getCompressionCodecForFile.h @@ -11,6 +11,6 @@ namespace DB /// clickhouse fashion (with checksums, headers for each block, etc). This /// method should be used as fallback when we cannot deduce compression codec /// from metadata. -CompressionCodecPtr getCompressionCodecForFile(const DataPartStoragePtr & data_part_storage, const String & relative_path); +CompressionCodecPtr getCompressionCodecForFile(const IDataPartStorage & data_part_storage, const String & relative_path); } diff --git a/src/Compression/tests/gtest_compressionCodec.cpp b/src/Compression/tests/gtest_compressionCodec.cpp index c203b3e9d5c..9b44c60cd81 100644 --- a/src/Compression/tests/gtest_compressionCodec.cpp +++ b/src/Compression/tests/gtest_compressionCodec.cpp @@ -1049,7 +1049,7 @@ INSTANTIATE_TEST_SUITE_P(RandomInt, ::testing::Combine( DefaultCodecsToTest, ::testing::Values( - generateSeq(G(RandomGenerator(0))), + generateSeq(G(RandomGenerator(0))), generateSeq(G(RandomGenerator(0))), generateSeq(G(RandomGenerator(0, 0, 1000'000'000))), generateSeq(G(RandomGenerator(0, 0, 1000'000'000))) diff --git a/src/DataTypes/DataTypeArray.h b/src/DataTypes/DataTypeArray.h index 122ac8e03a3..033a657c845 100644 --- a/src/DataTypes/DataTypeArray.h +++ b/src/DataTypes/DataTypeArray.h @@ -48,6 +48,7 @@ public: bool textCanContainOnlyValidUTF8() const override { return nested->textCanContainOnlyValidUTF8(); } bool isComparable() const override { return nested->isComparable(); } bool canBeComparedWithCollation() const override { return nested->canBeComparedWithCollation(); } + bool hasDynamicSubcolumns() const override { return nested->hasDynamicSubcolumns(); } bool isValueUnambiguouslyRepresentedInContiguousMemoryRegion() const override { diff --git a/src/DataTypes/DataTypeMap.cpp b/src/DataTypes/DataTypeMap.cpp index 42ec739c33b..d49c205fc59 100644 --- a/src/DataTypes/DataTypeMap.cpp +++ b/src/DataTypes/DataTypeMap.cpp @@ -22,6 +22,27 @@ namespace ErrorCodes extern const int BAD_ARGUMENTS; } +DataTypeMap::DataTypeMap(const DataTypePtr & nested_) + : nested(nested_) +{ + const auto * type_array = typeid_cast(nested.get()); + if (!type_array) + throw Exception(ErrorCodes::BAD_ARGUMENTS, + "Expected Array(Tuple(key, value)) type, got {}", nested->getName()); + + const auto * type_tuple = typeid_cast(type_array->getNestedType().get()); + if (!type_tuple) + throw Exception(ErrorCodes::BAD_ARGUMENTS, + "Expected Array(Tuple(key, value)) type, got {}", nested->getName()); + + if (type_tuple->getElements().size() != 2) + throw Exception(ErrorCodes::BAD_ARGUMENTS, + "Expected Array(Tuple(key, value)) type, got {}", nested->getName()); + + key_type = type_tuple->getElement(0); + value_type = type_tuple->getElement(1); + assertKeyType(); +} DataTypeMap::DataTypeMap(const DataTypes & elems_) { diff --git a/src/DataTypes/DataTypeMap.h b/src/DataTypes/DataTypeMap.h index 479008031fe..2ab5c602a25 100644 --- a/src/DataTypes/DataTypeMap.h +++ b/src/DataTypes/DataTypeMap.h @@ -23,6 +23,7 @@ private: public: static constexpr bool is_parametric = true; + explicit DataTypeMap(const DataTypePtr & nested_); explicit DataTypeMap(const DataTypes & elems); DataTypeMap(const DataTypePtr & key_type_, const DataTypePtr & value_type_); @@ -40,6 +41,7 @@ public: bool isComparable() const override { return key_type->isComparable() && value_type->isComparable(); } bool isParametric() const override { return true; } bool haveSubtypes() const override { return true; } + bool hasDynamicSubcolumns() const override { return nested->hasDynamicSubcolumns(); } const DataTypePtr & getKeyType() const { return key_type; } const DataTypePtr & getValueType() const { return value_type; } diff --git a/src/DataTypes/DataTypeObject.h b/src/DataTypes/DataTypeObject.h index 503947c3738..2f6ad961512 100644 --- a/src/DataTypes/DataTypeObject.h +++ b/src/DataTypes/DataTypeObject.h @@ -36,6 +36,7 @@ public: bool haveSubtypes() const override { return false; } bool equals(const IDataType & rhs) const override; bool isParametric() const override { return true; } + bool hasDynamicSubcolumns() const override { return true; } SerializationPtr doGetDefaultSerialization() const override; diff --git a/src/DataTypes/DataTypeTuple.cpp b/src/DataTypes/DataTypeTuple.cpp index ef05cd440c0..87cbac4cfb2 100644 --- a/src/DataTypes/DataTypeTuple.cpp +++ b/src/DataTypes/DataTypeTuple.cpp @@ -247,6 +247,11 @@ bool DataTypeTuple::haveMaximumSizeOfValue() const return std::all_of(elems.begin(), elems.end(), [](auto && elem) { return elem->haveMaximumSizeOfValue(); }); } +bool DataTypeTuple::hasDynamicSubcolumns() const +{ + return std::any_of(elems.begin(), elems.end(), [](auto && elem) { return elem->hasDynamicSubcolumns(); }); +} + bool DataTypeTuple::isComparable() const { return std::all_of(elems.begin(), elems.end(), [](auto && elem) { return elem->isComparable(); }); diff --git a/src/DataTypes/DataTypeTuple.h b/src/DataTypes/DataTypeTuple.h index 5abec18bd3b..152f21015f5 100644 --- a/src/DataTypes/DataTypeTuple.h +++ b/src/DataTypes/DataTypeTuple.h @@ -50,6 +50,7 @@ public: bool isComparable() const override; bool textCanContainOnlyValidUTF8() const override; bool haveMaximumSizeOfValue() const override; + bool hasDynamicSubcolumns() const override; size_t getMaximumSizeOfValueInMemory() const override; size_t getSizeOfValueInMemory() const override; diff --git a/src/DataTypes/IDataType.h b/src/DataTypes/IDataType.h index c93128ced95..4db944c8c3f 100644 --- a/src/DataTypes/IDataType.h +++ b/src/DataTypes/IDataType.h @@ -291,6 +291,9 @@ public: /// Strings, Numbers, Date, DateTime, Nullable virtual bool canBeInsideLowCardinality() const { return false; } + /// Object, Array(Object), Tuple(..., Object, ...) + virtual bool hasDynamicSubcolumns() const { return false; } + /// Updates avg_value_size_hint for newly read column. Uses to optimize deserialization. Zero expected for first column. static void updateAvgValueSizeHint(const IColumn & column, double & avg_value_size_hint); diff --git a/src/DataTypes/ObjectUtils.cpp b/src/DataTypes/ObjectUtils.cpp index e5d8d05acb5..e711b34ffa9 100644 --- a/src/DataTypes/ObjectUtils.cpp +++ b/src/DataTypes/ObjectUtils.cpp @@ -1,17 +1,19 @@ -#include #include #include #include #include +#include #include #include #include #include #include #include +#include #include #include #include +#include #include #include #include @@ -105,10 +107,11 @@ Array createEmptyArrayField(size_t num_dimensions) DataTypePtr getDataTypeByColumn(const IColumn & column) { auto idx = column.getDataType(); - if (WhichDataType(idx).isSimple()) + WhichDataType which(idx); + if (which.isSimple()) return DataTypeFactory::instance().get(String(magic_enum::enum_name(idx))); - if (WhichDataType(idx).isNothing()) + if (which.isNothing()) return std::make_shared(); if (const auto * column_array = checkAndGetColumn(&column)) @@ -132,41 +135,124 @@ static auto extractVector(const std::vector & vec) return res; } -void convertObjectsToTuples(Block & block, const NamesAndTypesList & extended_storage_columns) +static DataTypePtr recreateTupleWithElements(const DataTypeTuple & type_tuple, const DataTypes & elements) { - std::unordered_map storage_columns_map; - for (const auto & [name, type] : extended_storage_columns) - storage_columns_map[name] = type; - - for (auto & column : block) - { - if (!isObject(column.type)) - continue; - - const auto & column_object = assert_cast(*column.column); - if (!column_object.isFinalized()) - throw Exception(ErrorCodes::LOGICAL_ERROR, - "Cannot convert to tuple column '{}' from type {}. Column should be finalized first", - column.name, column.type->getName()); - - std::tie(column.column, column.type) = unflattenObjectToTuple(column_object); - - auto it = storage_columns_map.find(column.name); - if (it == storage_columns_map.end()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Column '{}' not found in storage", column.name); - - /// Check that constructed Tuple type and type in storage are compatible. - getLeastCommonTypeForObject({column.type, it->second}, true); - } + return type_tuple.haveExplicitNames() + ? std::make_shared(elements, type_tuple.getElementNames()) + : std::make_shared(elements); } -void deduceTypesOfObjectColumns(const StorageSnapshotPtr & storage_snapshot, Block & block) +static std::pair convertObjectColumnToTuple( + const ColumnObject & column_object, const DataTypeObject & type_object) { - if (!storage_snapshot->object_columns.empty()) + if (!column_object.isFinalized()) { - auto options = GetColumnsOptions(GetColumnsOptions::AllPhysical).withExtendedObjects(); - auto storage_columns = storage_snapshot->getColumns(options); - convertObjectsToTuples(block, storage_columns); + auto finalized = column_object.cloneFinalized(); + const auto & finalized_object = assert_cast(*finalized); + return convertObjectColumnToTuple(finalized_object, type_object); + } + + const auto & subcolumns = column_object.getSubcolumns(); + + PathsInData tuple_paths; + DataTypes tuple_types; + Columns tuple_columns; + + for (const auto & entry : subcolumns) + { + tuple_paths.emplace_back(entry->path); + tuple_types.emplace_back(entry->data.getLeastCommonType()); + tuple_columns.emplace_back(entry->data.getFinalizedColumnPtr()); + } + + return unflattenTuple(tuple_paths, tuple_types, tuple_columns); +} + +static std::pair recursivlyConvertDynamicColumnToTuple( + const ColumnPtr & column, const DataTypePtr & type) +{ + if (!type->hasDynamicSubcolumns()) + return {column, type}; + + if (const auto * type_object = typeid_cast(type.get())) + { + const auto & column_object = assert_cast(*column); + return convertObjectColumnToTuple(column_object, *type_object); + } + + if (const auto * type_array = typeid_cast(type.get())) + { + const auto & column_array = assert_cast(*column); + auto [new_column, new_type] = recursivlyConvertDynamicColumnToTuple( + column_array.getDataPtr(), type_array->getNestedType()); + + return + { + ColumnArray::create(new_column, column_array.getOffsetsPtr()), + std::make_shared(std::move(new_type)), + }; + } + + if (const auto * type_map = typeid_cast(type.get())) + { + const auto & column_map = assert_cast(*column); + auto [new_column, new_type] = recursivlyConvertDynamicColumnToTuple( + column_map.getNestedColumnPtr(), type_map->getNestedType()); + + return + { + ColumnMap::create(new_column), + std::make_shared(std::move(new_type)), + }; + } + + if (const auto * type_tuple = typeid_cast(type.get())) + { + const auto & tuple_columns = assert_cast(*column).getColumns(); + const auto & tuple_types = type_tuple->getElements(); + + assert(tuple_columns.size() == tuple_types.size()); + const size_t tuple_size = tuple_types.size(); + + Columns new_tuple_columns(tuple_size); + DataTypes new_tuple_types(tuple_size); + + for (size_t i = 0; i < tuple_size; ++i) + { + std::tie(new_tuple_columns[i], new_tuple_types[i]) + = recursivlyConvertDynamicColumnToTuple(tuple_columns[i], tuple_types[i]); + } + + return + { + ColumnTuple::create(new_tuple_columns), + recreateTupleWithElements(*type_tuple, new_tuple_types) + }; + } + + throw Exception(ErrorCodes::LOGICAL_ERROR, "Type {} unexpectedly has dynamic columns", type->getName()); +} + +void convertDynamicColumnsToTuples(Block & block, const StorageSnapshotPtr & storage_snapshot) +{ + for (auto & column : block) + { + if (!column.type->hasDynamicSubcolumns()) + continue; + + std::tie(column.column, column.type) + = recursivlyConvertDynamicColumnToTuple(column.column, column.type); + + GetColumnsOptions options(GetColumnsOptions::AllPhysical); + auto storage_column = storage_snapshot->tryGetColumn(options, column.name); + if (!storage_column) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Column '{}' not found in storage", column.name); + + auto storage_column_concrete = storage_snapshot->getColumn(options.withExtendedObjects(), column.name); + + /// Check that constructed Tuple type and type in storage are compatible. + getLeastCommonTypeForDynamicColumns( + storage_column->type, {column.type, storage_column_concrete.type}, true); } } @@ -217,24 +303,8 @@ void checkObjectHasNoAmbiguosPaths(const PathsInData & paths) } } -DataTypePtr getLeastCommonTypeForObject(const DataTypes & types, bool check_ambiguos_paths) +static DataTypePtr getLeastCommonTypeForObject(const DataTypes & types, bool check_ambiguos_paths) { - if (types.empty()) - return nullptr; - - bool all_equal = true; - for (size_t i = 1; i < types.size(); ++i) - { - if (!types[i]->equals(*types[0])) - { - all_equal = false; - break; - } - } - - if (all_equal) - return types[0]; - /// Types of subcolumns by path from all tuples. std::unordered_map subcolumns_types; @@ -287,19 +357,139 @@ DataTypePtr getLeastCommonTypeForObject(const DataTypes & types, bool check_ambi return unflattenTuple(tuple_paths, tuple_types); } -NameSet getNamesOfObjectColumns(const NamesAndTypesList & columns_list) -{ - NameSet res; - for (const auto & [name, type] : columns_list) - if (isObject(type)) - res.insert(name); +static DataTypePtr getLeastCommonTypeForDynamicColumnsImpl( + const DataTypePtr & type_in_storage, const DataTypes & concrete_types, bool check_ambiguos_paths); - return res; +template +static DataTypePtr getLeastCommonTypeForColumnWithNestedType( + const Type & type, const DataTypes & concrete_types, bool check_ambiguos_paths) +{ + DataTypes nested_types; + nested_types.reserve(concrete_types.size()); + + for (const auto & concrete_type : concrete_types) + { + const auto * type_with_nested_conctete = typeid_cast(concrete_type.get()); + if (!type_with_nested_conctete) + throw Exception(ErrorCodes::TYPE_MISMATCH, "Expected {} type, got {}", demangle(typeid(Type).name()), concrete_type->getName()); + + nested_types.push_back(type_with_nested_conctete->getNestedType()); + } + + return std::make_shared( + getLeastCommonTypeForDynamicColumnsImpl( + type.getNestedType(), nested_types, check_ambiguos_paths)); } -bool hasObjectColumns(const ColumnsDescription & columns) +static DataTypePtr getLeastCommonTypeForTuple( + const DataTypeTuple & type, const DataTypes & concrete_types, bool check_ambiguos_paths) { - return std::any_of(columns.begin(), columns.end(), [](const auto & column) { return isObject(column.type); }); + const auto & element_types = type.getElements(); + DataTypes new_element_types(element_types.size()); + + for (size_t i = 0; i < element_types.size(); ++i) + { + DataTypes concrete_element_types; + concrete_element_types.reserve(concrete_types.size()); + + for (const auto & type_concrete : concrete_types) + { + const auto * type_tuple_conctete = typeid_cast(type_concrete.get()); + if (!type_tuple_conctete) + throw Exception(ErrorCodes::TYPE_MISMATCH, "Expected Tuple type, got {}", type_concrete->getName()); + + concrete_element_types.push_back(type_tuple_conctete->getElement(i)); + } + + new_element_types[i] = getLeastCommonTypeForDynamicColumnsImpl( + element_types[i], concrete_element_types, check_ambiguos_paths); + } + + return recreateTupleWithElements(type, new_element_types); +} + +static DataTypePtr getLeastCommonTypeForDynamicColumnsImpl( + const DataTypePtr & type_in_storage, const DataTypes & concrete_types, bool check_ambiguos_paths) +{ + if (!type_in_storage->hasDynamicSubcolumns()) + return type_in_storage; + + if (isObject(type_in_storage)) + return getLeastCommonTypeForObject(concrete_types, check_ambiguos_paths); + + if (const auto * type_array = typeid_cast(type_in_storage.get())) + return getLeastCommonTypeForColumnWithNestedType(*type_array, concrete_types, check_ambiguos_paths); + + if (const auto * type_map = typeid_cast(type_in_storage.get())) + return getLeastCommonTypeForColumnWithNestedType(*type_map, concrete_types, check_ambiguos_paths); + + if (const auto * type_tuple = typeid_cast(type_in_storage.get())) + return getLeastCommonTypeForTuple(*type_tuple, concrete_types, check_ambiguos_paths); + + throw Exception(ErrorCodes::LOGICAL_ERROR, "Type {} unexpectedly has dynamic columns", type_in_storage->getName()); +} + +DataTypePtr getLeastCommonTypeForDynamicColumns( + const DataTypePtr & type_in_storage, const DataTypes & concrete_types, bool check_ambiguos_paths) +{ + if (concrete_types.empty()) + return nullptr; + + bool all_equal = true; + for (size_t i = 1; i < concrete_types.size(); ++i) + { + if (!concrete_types[i]->equals(*concrete_types[0])) + { + all_equal = false; + break; + } + } + + if (all_equal) + return concrete_types[0]; + + return getLeastCommonTypeForDynamicColumnsImpl(type_in_storage, concrete_types, check_ambiguos_paths); +} + +DataTypePtr createConcreteEmptyDynamicColumn(const DataTypePtr & type_in_storage) +{ + if (!type_in_storage->hasDynamicSubcolumns()) + return type_in_storage; + + if (isObject(type_in_storage)) + return std::make_shared( + DataTypes{std::make_shared()}, Names{ColumnObject::COLUMN_NAME_DUMMY}); + + if (const auto * type_array = typeid_cast(type_in_storage.get())) + return std::make_shared( + createConcreteEmptyDynamicColumn(type_array->getNestedType())); + + if (const auto * type_map = typeid_cast(type_in_storage.get())) + return std::make_shared( + createConcreteEmptyDynamicColumn(type_map->getNestedType())); + + if (const auto * type_tuple = typeid_cast(type_in_storage.get())) + { + const auto & elements = type_tuple->getElements(); + DataTypes new_elements; + new_elements.reserve(elements.size()); + + for (const auto & element : elements) + new_elements.push_back(createConcreteEmptyDynamicColumn(element)); + + return recreateTupleWithElements(*type_tuple, new_elements); + } + + throw Exception(ErrorCodes::LOGICAL_ERROR, "Type {} unexpectedly has dynamic columns", type_in_storage->getName()); +} + +bool hasDynamicSubcolumns(const ColumnsDescription & columns) +{ + return std::any_of(columns.begin(), columns.end(), + [](const auto & column) + { + return column.type->hasDynamicSubcolumns(); + }); } void extendObjectColumns(NamesAndTypesList & columns_list, const ColumnsDescription & object_columns, bool with_subcolumns) @@ -320,16 +510,20 @@ void extendObjectColumns(NamesAndTypesList & columns_list, const ColumnsDescript columns_list.splice(columns_list.end(), std::move(subcolumns_list)); } -void updateObjectColumns(ColumnsDescription & object_columns, const NamesAndTypesList & new_columns) +void updateObjectColumns( + ColumnsDescription & object_columns, + const ColumnsDescription & storage_columns, + const NamesAndTypesList & new_columns) { for (const auto & new_column : new_columns) { auto object_column = object_columns.tryGetColumn(GetColumnsOptions::All, new_column.name); if (object_column && !object_column->type->equals(*new_column.type)) { + auto storage_column = storage_columns.getColumn(GetColumnsOptions::All, new_column.name); object_columns.modify(new_column.name, [&](auto & column) { - column.type = getLeastCommonTypeForObject({object_column->type, new_column.type}); + column.type = getLeastCommonTypeForDynamicColumns(storage_column.type, {object_column->type, new_column.type}); }); } } @@ -745,13 +939,6 @@ void replaceMissedSubcolumnsByConstants( addConstantToWithClause(query, name, type); } -void finalizeObjectColumns(const MutableColumns & columns) -{ - for (const auto & column : columns) - if (auto * column_object = typeid_cast(column.get())) - column_object->finalize(); -} - Field FieldVisitorReplaceScalars::operator()(const Array & x) const { if (num_dimensions_to_keep == 0) @@ -768,11 +955,13 @@ size_t FieldVisitorToNumberOfDimensions::operator()(const Array & x) { const size_t size = x.size(); size_t dimensions = 0; + for (size_t i = 0; i < size; ++i) { size_t element_dimensions = applyVisitor(*this, x[i]); if (i > 0 && element_dimensions != dimensions) need_fold_dimension = true; + dimensions = std::max(dimensions, element_dimensions); } @@ -783,12 +972,13 @@ Field FieldVisitorFoldDimension::operator()(const Array & x) const { if (num_dimensions_to_fold == 0) return x; + const size_t size = x.size(); Array res(size); for (size_t i = 0; i < size; ++i) - { res[i] = applyVisitor(FieldVisitorFoldDimension(num_dimensions_to_fold - 1), x[i]); - } + return res; } + } diff --git a/src/DataTypes/ObjectUtils.h b/src/DataTypes/ObjectUtils.h index c60d5bec208..bd15edfe851 100644 --- a/src/DataTypes/ObjectUtils.h +++ b/src/DataTypes/ObjectUtils.h @@ -39,27 +39,31 @@ Array createEmptyArrayField(size_t num_dimensions); DataTypePtr getDataTypeByColumn(const IColumn & column); /// Converts Object types and columns to Tuples in @columns_list and @block -/// and checks that types are consistent with types in @extended_storage_columns. -void convertObjectsToTuples(Block & block, const NamesAndTypesList & extended_storage_columns); -void deduceTypesOfObjectColumns(const StorageSnapshotPtr & storage_snapshot, Block & block); +/// and checks that types are consistent with types in @storage_snapshot. +void convertDynamicColumnsToTuples(Block & block, const StorageSnapshotPtr & storage_snapshot); /// Checks that each path is not the prefix of any other path. void checkObjectHasNoAmbiguosPaths(const PathsInData & paths); /// Receives several Tuple types and deduces the least common type among them. -DataTypePtr getLeastCommonTypeForObject(const DataTypes & types, bool check_ambiguos_paths = false); +DataTypePtr getLeastCommonTypeForDynamicColumns( + const DataTypePtr & type_in_storage, const DataTypes & types, bool check_ambiguos_paths = false); + +DataTypePtr createConcreteEmptyDynamicColumn(const DataTypePtr & type_in_storage); /// Converts types of object columns to tuples in @columns_list /// according to @object_columns and adds all tuple's subcolumns if needed. void extendObjectColumns(NamesAndTypesList & columns_list, const ColumnsDescription & object_columns, bool with_subcolumns); -NameSet getNamesOfObjectColumns(const NamesAndTypesList & columns_list); -bool hasObjectColumns(const ColumnsDescription & columns); -void finalizeObjectColumns(const MutableColumns & columns); +/// Checks whether @columns contain any column with dynamic subcolumns. +bool hasDynamicSubcolumns(const ColumnsDescription & columns); /// Updates types of objects in @object_columns inplace /// according to types in new_columns. -void updateObjectColumns(ColumnsDescription & object_columns, const NamesAndTypesList & new_columns); +void updateObjectColumns( + ColumnsDescription & object_columns, + const ColumnsDescription & storage_columns, + const NamesAndTypesList & new_columns); using DataTypeTuplePtr = std::shared_ptr; @@ -142,13 +146,15 @@ public: { if (num_dimensions_to_fold == 0) return x; - Array res(1,x); + + Array res(1, x); for (size_t i = 1; i < num_dimensions_to_fold; ++i) { Array new_res; new_res.push_back(std::move(res)); res = std::move(new_res); } + return res; } @@ -163,7 +169,7 @@ private: /// columns-like objects from entry to which Iterator points. /// columns-like object should have fields "name" and "type". template -ColumnsDescription getObjectColumns( +ColumnsDescription getConcreteObjectColumns( Iterator begin, Iterator end, const ColumnsDescription & storage_columns, EntryColumnsGetter && entry_columns_getter) @@ -176,14 +182,8 @@ ColumnsDescription getObjectColumns( /// dummy column will be removed. for (const auto & column : storage_columns) { - if (isObject(column.type)) - { - auto tuple_type = std::make_shared( - DataTypes{std::make_shared()}, - Names{ColumnObject::COLUMN_NAME_DUMMY}); - - types_in_entries[column.name].push_back(std::move(tuple_type)); - } + if (column.type->hasDynamicSubcolumns()) + types_in_entries[column.name].push_back(createConcreteEmptyDynamicColumn(column.type)); } for (auto it = begin; it != end; ++it) @@ -192,14 +192,17 @@ ColumnsDescription getObjectColumns( for (const auto & column : entry_columns) { auto storage_column = storage_columns.tryGetPhysical(column.name); - if (storage_column && isObject(storage_column->type)) + if (storage_column && storage_column->type->hasDynamicSubcolumns()) types_in_entries[column.name].push_back(column.type); } } ColumnsDescription res; for (const auto & [name, types] : types_in_entries) - res.add({name, getLeastCommonTypeForObject(types)}); + { + auto storage_column = storage_columns.getPhysical(name); + res.add({name, getLeastCommonTypeForDynamicColumns(storage_column.type, types)}); + } return res; } diff --git a/src/DataTypes/Serializations/ISerialization.h b/src/DataTypes/Serializations/ISerialization.h index 1193c15b939..d64b41253f5 100644 --- a/src/DataTypes/Serializations/ISerialization.h +++ b/src/DataTypes/Serializations/ISerialization.h @@ -249,7 +249,9 @@ public: }; /// Call before serializeBinaryBulkWithMultipleStreams chain to write something before first mark. + /// Column may be used only to retrieve the structure. virtual void serializeBinaryBulkStatePrefix( + const IColumn & /*column*/, SerializeBinaryBulkSettings & /*settings*/, SerializeBinaryBulkStatePtr & /*state*/) const {} diff --git a/src/DataTypes/Serializations/SerializationArray.cpp b/src/DataTypes/Serializations/SerializationArray.cpp index eb93b5049a0..143a3264381 100644 --- a/src/DataTypes/Serializations/SerializationArray.cpp +++ b/src/DataTypes/Serializations/SerializationArray.cpp @@ -246,11 +246,13 @@ void SerializationArray::enumerateStreams( } void SerializationArray::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { settings.path.push_back(Substream::ArrayElements); - nested->serializeBinaryBulkStatePrefix(settings, state); + const auto & column_array = assert_cast(column); + nested->serializeBinaryBulkStatePrefix(column_array.getData(), settings, state); settings.path.pop_back(); } diff --git a/src/DataTypes/Serializations/SerializationArray.h b/src/DataTypes/Serializations/SerializationArray.h index 84e37acbaad..860461d667f 100644 --- a/src/DataTypes/Serializations/SerializationArray.h +++ b/src/DataTypes/Serializations/SerializationArray.h @@ -41,6 +41,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationLowCardinality.cpp b/src/DataTypes/Serializations/SerializationLowCardinality.cpp index 761adf3b765..c70bb1e1465 100644 --- a/src/DataTypes/Serializations/SerializationLowCardinality.cpp +++ b/src/DataTypes/Serializations/SerializationLowCardinality.cpp @@ -221,6 +221,7 @@ struct DeserializeStateLowCardinality : public ISerialization::DeserializeBinary }; void SerializationLowCardinality::serializeBinaryBulkStatePrefix( + const IColumn & /*column*/, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { diff --git a/src/DataTypes/Serializations/SerializationLowCardinality.h b/src/DataTypes/Serializations/SerializationLowCardinality.h index cc090f2044e..1d0c3226faf 100644 --- a/src/DataTypes/Serializations/SerializationLowCardinality.h +++ b/src/DataTypes/Serializations/SerializationLowCardinality.h @@ -23,6 +23,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationMap.cpp b/src/DataTypes/Serializations/SerializationMap.cpp index 958e33fbaf4..cd0a99c0c68 100644 --- a/src/DataTypes/Serializations/SerializationMap.cpp +++ b/src/DataTypes/Serializations/SerializationMap.cpp @@ -270,10 +270,11 @@ void SerializationMap::enumerateStreams( } void SerializationMap::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { - nested->serializeBinaryBulkStatePrefix(settings, state); + nested->serializeBinaryBulkStatePrefix(extractNestedColumn(column), settings, state); } void SerializationMap::serializeBinaryBulkStateSuffix( diff --git a/src/DataTypes/Serializations/SerializationMap.h b/src/DataTypes/Serializations/SerializationMap.h index 42f99ca7991..864ac1f3a99 100644 --- a/src/DataTypes/Serializations/SerializationMap.h +++ b/src/DataTypes/Serializations/SerializationMap.h @@ -37,6 +37,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationNamed.cpp b/src/DataTypes/Serializations/SerializationNamed.cpp index 4dac4b3a922..ca60948ce68 100644 --- a/src/DataTypes/Serializations/SerializationNamed.cpp +++ b/src/DataTypes/Serializations/SerializationNamed.cpp @@ -17,11 +17,12 @@ void SerializationNamed::enumerateStreams( } void SerializationNamed::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { addToPath(settings.path); - nested_serialization->serializeBinaryBulkStatePrefix(settings, state); + nested_serialization->serializeBinaryBulkStatePrefix(column, settings, state); settings.path.pop_back(); } diff --git a/src/DataTypes/Serializations/SerializationNamed.h b/src/DataTypes/Serializations/SerializationNamed.h index 2a2c7c0dfc7..52bbb039442 100644 --- a/src/DataTypes/Serializations/SerializationNamed.h +++ b/src/DataTypes/Serializations/SerializationNamed.h @@ -31,6 +31,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationNullable.cpp b/src/DataTypes/Serializations/SerializationNullable.cpp index 560b73bc827..c46fde27ddb 100644 --- a/src/DataTypes/Serializations/SerializationNullable.cpp +++ b/src/DataTypes/Serializations/SerializationNullable.cpp @@ -70,11 +70,13 @@ void SerializationNullable::enumerateStreams( } void SerializationNullable::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { settings.path.push_back(Substream::NullableElements); - nested->serializeBinaryBulkStatePrefix(settings, state); + const auto & column_nullable = assert_cast(column); + nested->serializeBinaryBulkStatePrefix(column_nullable.getNestedColumn(), settings, state); settings.path.pop_back(); } diff --git a/src/DataTypes/Serializations/SerializationNullable.h b/src/DataTypes/Serializations/SerializationNullable.h index ea3958065e7..9aabbe299cc 100644 --- a/src/DataTypes/Serializations/SerializationNullable.h +++ b/src/DataTypes/Serializations/SerializationNullable.h @@ -19,6 +19,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationObject.cpp b/src/DataTypes/Serializations/SerializationObject.cpp index b893407e7a5..98a94886f67 100644 --- a/src/DataTypes/Serializations/SerializationObject.cpp +++ b/src/DataTypes/Serializations/SerializationObject.cpp @@ -13,8 +13,6 @@ #include #include -#include - #include #include #include @@ -30,6 +28,7 @@ namespace ErrorCodes extern const int NOT_IMPLEMENTED; extern const int INCORRECT_DATA; extern const int CANNOT_READ_ALL_DATA; + extern const int ARGUMENT_OUT_OF_BOUND; extern const int LOGICAL_ERROR; } @@ -141,7 +140,6 @@ void SerializationObject::checkSerializationIsSupported(const TSettings template struct SerializationObject::SerializeStateObject : public ISerialization::SerializeBinaryBulkState { - bool is_first = true; DataTypePtr nested_type; SerializationPtr nested_serialization; SerializeBinaryBulkStatePtr nested_state; @@ -158,6 +156,7 @@ struct SerializationObject::DeserializeStateObject : public ISerializati template void SerializationObject::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { @@ -166,15 +165,34 @@ void SerializationObject::serializeBinaryBulkStatePrefix( throw Exception(ErrorCodes::NOT_IMPLEMENTED, "DataTypeObject doesn't support serialization with non-trivial state"); + const auto & column_object = assert_cast(column); + if (!column_object.isFinalized()) + { + auto finalized = column_object.cloneFinalized(); + serializeBinaryBulkStatePrefix(*finalized, settings, state); + return; + } + settings.path.push_back(Substream::ObjectStructure); auto * stream = settings.getter(settings.path); - settings.path.pop_back(); if (!stream) throw Exception(ErrorCodes::LOGICAL_ERROR, "Missing stream for kind of binary serialization"); + auto [tuple_column, tuple_type] = unflattenObjectToTuple(column_object); + writeIntBinary(static_cast(BinarySerializationKind::TUPLE), *stream); - state = std::make_shared(); + writeStringBinary(tuple_type->getName(), *stream); + + auto state_object = std::make_shared(); + state_object->nested_type = tuple_type; + state_object->nested_serialization = tuple_type->getDefaultSerialization(); + + settings.path.back() = Substream::ObjectData; + state_object->nested_serialization->serializeBinaryBulkStatePrefix(*tuple_column, settings, state_object->nested_state); + + state = std::move(state_object); + settings.path.pop_back(); } template @@ -261,33 +279,14 @@ void SerializationObject::serializeBinaryBulkWithMultipleStreams( if (!column_object.isFinalized()) { - auto finalized_object = column_object.clone(); - assert_cast(*finalized_object).finalize(); - serializeBinaryBulkWithMultipleStreams(*finalized_object, offset, limit, settings, state); + auto finalized = column_object.cloneFinalized(); + serializeBinaryBulkWithMultipleStreams(*finalized, offset, limit, settings, state); return; } auto [tuple_column, tuple_type] = unflattenObjectToTuple(column_object); - if (state_object->is_first) - { - /// Actually it's a part of serializeBinaryBulkStatePrefix, - /// but it cannot be done there, because we have to know the - /// structure of column. - - settings.path.push_back(Substream::ObjectStructure); - if (auto * stream = settings.getter(settings.path)) - writeStringBinary(tuple_type->getName(), *stream); - - state_object->nested_type = tuple_type; - state_object->nested_serialization = tuple_type->getDefaultSerialization(); - state_object->is_first = false; - - settings.path.back() = Substream::ObjectData; - state_object->nested_serialization->serializeBinaryBulkStatePrefix(settings, state_object->nested_state); - settings.path.pop_back(); - } - else if (!state_object->nested_type->equals(*tuple_type)) + if (!state_object->nested_type->equals(*tuple_type)) { throw Exception(ErrorCodes::LOGICAL_ERROR, "Types of internal column of Object mismatched. Expected: {}, Got: {}", @@ -411,18 +410,63 @@ void SerializationObject::serializeTextImpl(const IColumn & column, size writeChar('{', ostr); for (auto it = subcolumns.begin(); it != subcolumns.end(); ++it) { + const auto & entry = *it; if (it != subcolumns.begin()) writeCString(",", ostr); - writeDoubleQuoted((*it)->path.getPath(), ostr); + writeDoubleQuoted(entry->path.getPath(), ostr); writeChar(':', ostr); - - auto serialization = (*it)->data.getLeastCommonType()->getDefaultSerialization(); - serialization->serializeTextJSON((*it)->data.getFinalizedColumn(), row_num, ostr, settings); + serializeTextFromSubcolumn(entry->data, row_num, ostr, settings); } writeChar('}', ostr); } +template +void SerializationObject::serializeTextFromSubcolumn( + const ColumnObject::Subcolumn & subcolumn, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const +{ + const auto & least_common_type = subcolumn.getLeastCommonType(); + + if (subcolumn.isFinalized()) + { + const auto & finalized_column = subcolumn.getFinalizedColumn(); + auto info = least_common_type->getSerializationInfo(finalized_column); + auto serialization = least_common_type->getSerialization(*info); + serialization->serializeTextJSON(finalized_column, row_num, ostr, settings); + return; + } + + size_t ind = row_num; + if (ind < subcolumn.getNumberOfDefaultsInPrefix()) + { + /// Suboptimal, but it should happen rarely. + auto tmp_column = subcolumn.getLeastCommonType()->createColumn(); + tmp_column->insertDefault(); + + auto info = least_common_type->getSerializationInfo(*tmp_column); + auto serialization = least_common_type->getSerialization(*info); + serialization->serializeTextJSON(*tmp_column, 0, ostr, settings); + return; + } + + ind -= subcolumn.getNumberOfDefaultsInPrefix(); + for (const auto & part : subcolumn.getData()) + { + if (ind < part->size()) + { + auto part_type = getDataTypeByColumn(*part); + auto info = part_type->getSerializationInfo(*part); + auto serialization = part_type->getSerialization(*info); + serialization->serializeTextJSON(*part, ind, ostr, settings); + return; + } + + ind -= part->size(); + } + + throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Index ({}) for text serialization is out of range", row_num); +} + template void SerializationObject::serializeText(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const { diff --git a/src/DataTypes/Serializations/SerializationObject.h b/src/DataTypes/Serializations/SerializationObject.h index ff72c84faaa..47a7127cd1c 100644 --- a/src/DataTypes/Serializations/SerializationObject.h +++ b/src/DataTypes/Serializations/SerializationObject.h @@ -8,7 +8,7 @@ namespace DB { /** Serialization for data type Object. - * Supported only test serialization/deserialization. + * Supported only text serialization/deserialization. * and binary bulk serialization/deserialization without position independent * encoding, i.e. serialization/deserialization into Native format. */ @@ -31,6 +31,7 @@ public: */ void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; @@ -104,6 +105,7 @@ private: void deserializeTextImpl(IColumn & column, Reader && reader) const; void serializeTextImpl(const IColumn & column, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const; + void serializeTextFromSubcolumn(const ColumnObject::Subcolumn & subcolumn, size_t row_num, WriteBuffer & ostr, const FormatSettings & settings) const; /// Pool of parser objects to make SerializationObject thread safe. mutable SimpleObjectPool parsers_pool; diff --git a/src/DataTypes/Serializations/SerializationSparse.cpp b/src/DataTypes/Serializations/SerializationSparse.cpp index 855bdfa1b3e..cd09cd7be5a 100644 --- a/src/DataTypes/Serializations/SerializationSparse.cpp +++ b/src/DataTypes/Serializations/SerializationSparse.cpp @@ -178,11 +178,16 @@ void SerializationSparse::enumerateStreams( } void SerializationSparse::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { settings.path.push_back(Substream::SparseElements); - nested->serializeBinaryBulkStatePrefix(settings, state); + if (const auto * column_sparse = typeid_cast(&column)) + nested->serializeBinaryBulkStatePrefix(column_sparse->getValuesColumn(), settings, state); + else + nested->serializeBinaryBulkStatePrefix(column, settings, state); + settings.path.pop_back(); } diff --git a/src/DataTypes/Serializations/SerializationSparse.h b/src/DataTypes/Serializations/SerializationSparse.h index dc2f63c5a05..c157fe7ce98 100644 --- a/src/DataTypes/Serializations/SerializationSparse.h +++ b/src/DataTypes/Serializations/SerializationSparse.h @@ -33,6 +33,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationTuple.cpp b/src/DataTypes/Serializations/SerializationTuple.cpp index c2c40cbb507..8ffb1fe86bc 100644 --- a/src/DataTypes/Serializations/SerializationTuple.cpp +++ b/src/DataTypes/Serializations/SerializationTuple.cpp @@ -314,6 +314,7 @@ struct DeserializeBinaryBulkStateTuple : public ISerialization::DeserializeBinar void SerializationTuple::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { @@ -321,7 +322,7 @@ void SerializationTuple::serializeBinaryBulkStatePrefix( tuple_state->states.resize(elems.size()); for (size_t i = 0; i < elems.size(); ++i) - elems[i]->serializeBinaryBulkStatePrefix(settings, tuple_state->states[i]); + elems[i]->serializeBinaryBulkStatePrefix(extractElementColumn(column, i), settings, tuple_state->states[i]); state = std::move(tuple_state); } diff --git a/src/DataTypes/Serializations/SerializationTuple.h b/src/DataTypes/Serializations/SerializationTuple.h index d1caeb73dad..db0339bc996 100644 --- a/src/DataTypes/Serializations/SerializationTuple.h +++ b/src/DataTypes/Serializations/SerializationTuple.h @@ -39,6 +39,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/SerializationWrapper.cpp b/src/DataTypes/Serializations/SerializationWrapper.cpp index 7c50c1c6e26..c83de614751 100644 --- a/src/DataTypes/Serializations/SerializationWrapper.cpp +++ b/src/DataTypes/Serializations/SerializationWrapper.cpp @@ -13,10 +13,11 @@ void SerializationWrapper::enumerateStreams( } void SerializationWrapper::serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const { - nested_serialization->serializeBinaryBulkStatePrefix(settings, state); + nested_serialization->serializeBinaryBulkStatePrefix(column, settings, state); } void SerializationWrapper::serializeBinaryBulkStateSuffix( diff --git a/src/DataTypes/Serializations/SerializationWrapper.h b/src/DataTypes/Serializations/SerializationWrapper.h index d010c6b5314..46941f150e1 100644 --- a/src/DataTypes/Serializations/SerializationWrapper.h +++ b/src/DataTypes/Serializations/SerializationWrapper.h @@ -26,6 +26,7 @@ public: const SubstreamData & data) const override; void serializeBinaryBulkStatePrefix( + const IColumn & column, SerializeBinaryBulkSettings & settings, SerializeBinaryBulkStatePtr & state) const override; diff --git a/src/DataTypes/Serializations/tests/gtest_json_parser.cpp b/src/DataTypes/Serializations/tests/gtest_json_parser.cpp index 4dddb3cd03d..9b0c8e44d02 100644 --- a/src/DataTypes/Serializations/tests/gtest_json_parser.cpp +++ b/src/DataTypes/Serializations/tests/gtest_json_parser.cpp @@ -69,7 +69,7 @@ static std::ostream & operator<<(std::ostream & ostr, const JSONPathAndValue & p bool first = true; for (const auto & part : path_and_value.path.getParts()) { - ostr << (first ? "{" : ", {") << part.key << ", " << part.is_nested << ", " << part.anonymous_array_level << "}"; + ostr << (first ? "{" : ", {") << part.key << ", " << part.is_nested << ", " << static_cast(part.anonymous_array_level) << "}"; first = false; } diff --git a/src/DataTypes/Serializations/tests/gtest_object_serialization.cpp b/src/DataTypes/Serializations/tests/gtest_object_serialization.cpp index f1fbbe115e2..fc7432d5bf6 100644 --- a/src/DataTypes/Serializations/tests/gtest_object_serialization.cpp +++ b/src/DataTypes/Serializations/tests/gtest_object_serialization.cpp @@ -31,7 +31,7 @@ TEST(SerializationObject, FromString) settings.getter = [&out](const auto &) { return &out; }; writeIntBinary(static_cast(1), out); - serialization->serializeBinaryBulkStatePrefix(settings, state); + serialization->serializeBinaryBulkStatePrefix(*column_string, settings, state); serialization->serializeBinaryBulkWithMultipleStreams(*column_string, 0, column_string->size(), settings, state); serialization->serializeBinaryBulkStateSuffix(settings, state); } diff --git a/src/Disks/IStoragePolicy.cpp b/src/Disks/IStoragePolicy.cpp index 2ba6df4be8f..c843ee11563 100644 --- a/src/Disks/IStoragePolicy.cpp +++ b/src/Disks/IStoragePolicy.cpp @@ -31,4 +31,33 @@ VolumePtr IStoragePolicy::getVolumeByName(const String & volume_name) const return volume; } +size_t IStoragePolicy::getVolumeIndexByDiskName(const String & disk_name) const +{ + auto index = tryGetVolumeIndexByDiskName(disk_name); + if (!index) + throw Exception(ErrorCodes::UNKNOWN_DISK, + "No disk {} in policy {}", backQuote(disk_name), backQuote(getName())); + + return *index; +} + +VolumePtr IStoragePolicy::tryGetVolumeByDiskName(const String & disk_name) const +{ + auto index = tryGetVolumeIndexByDiskName(disk_name); + if (!index) + return nullptr; + + return getVolume(*index); +} + +VolumePtr IStoragePolicy::getVolumeByDiskName(const String & disk_name) const +{ + auto volume = tryGetVolumeByDiskName(disk_name); + if (!volume) + throw Exception(ErrorCodes::UNKNOWN_DISK, + "No disk {} in policy {}", backQuote(disk_name), backQuote(getName())); + + return volume; +} + } diff --git a/src/Disks/IStoragePolicy.h b/src/Disks/IStoragePolicy.h index 8d14a26691b..a6a5fe5f692 100644 --- a/src/Disks/IStoragePolicy.h +++ b/src/Disks/IStoragePolicy.h @@ -4,6 +4,7 @@ #include #include +#include #include namespace DB @@ -55,12 +56,15 @@ public: /// Get volume by index. virtual VolumePtr getVolume(size_t index) const = 0; virtual VolumePtr tryGetVolumeByName(const String & volume_name) const = 0; - virtual VolumePtr tryGetVolumeByDisk(const DiskPtr & disk_ptr) const = 0; VolumePtr getVolumeByName(const String & volume_name) const; /// Checks if storage policy can be replaced by another one. virtual void checkCompatibleWith(const StoragePolicyPtr & new_storage_policy) const = 0; - /// Find volume index, which contains disk - virtual size_t getVolumeIndexByDisk(const DiskPtr & disk_ptr) const = 0; + /// Finds a volume index, which contains disk + virtual std::optional tryGetVolumeIndexByDiskName(const String & disk_name) const = 0; + size_t getVolumeIndexByDiskName(const String & disk_name) const; + /// Finds a volume which contains a specified disk. + VolumePtr tryGetVolumeByDiskName(const String & disk_name) const; + VolumePtr getVolumeByDiskName(const String & disk_name) const; /// Check if we have any volume with stopped merges virtual bool hasAnyVolumeWithDisabledMerges() const = 0; virtual bool containsVolume(const String & volume_name) const = 0; diff --git a/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp b/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp index b3dcfdafa9e..c3549701ec1 100644 --- a/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp +++ b/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.cpp @@ -141,7 +141,7 @@ std::unique_ptr AzureObjectStorage::writeObject( /// NO return std::make_unique(std::move(buffer), std::move(finalize_callback), object.absolute_path); } -void AzureObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const +void AzureObjectStorage::findAllFiles(const std::string & path, RelativePathsWithSize & children) const { auto client_ptr = client.get(); diff --git a/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h b/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h index 47ac0d6badd..6fd41dae2ec 100644 --- a/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h +++ b/src/Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h @@ -84,7 +84,7 @@ public: size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE, const WriteSettings & write_settings = {}) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; + void findAllFiles(const std::string & path, RelativePathsWithSize & children) const override; /// Remove file. Throws exception if file doesn't exists or it's a directory. void removeObject(const StoredObject & object) override; diff --git a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp index 7e829847846..f3d3f049dc1 100644 --- a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp +++ b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.cpp @@ -282,9 +282,9 @@ std::unique_ptr CachedObjectStorage::cloneObjectStorage( return object_storage->cloneObjectStorage(new_namespace, config, config_prefix, context); } -void CachedObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const +void CachedObjectStorage::findAllFiles(const std::string & path, RelativePathsWithSize & children) const { - object_storage->listPrefix(path, children); + object_storage->findAllFiles(path, children); } ObjectMetadata CachedObjectStorage::getObjectMetadata(const std::string & path) const diff --git a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h index 68ded61a9f1..64e6eed45bb 100644 --- a/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h +++ b/src/Disks/ObjectStorages/Cached/CachedObjectStorage.h @@ -72,7 +72,7 @@ public: const std::string & config_prefix, ContextPtr context) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; + void findAllFiles(const std::string & path, RelativePathsWithSize & children) const override; ObjectMetadata getObjectMetadata(const std::string & path) const override; diff --git a/src/Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.cpp b/src/Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.cpp index 65f0d24035a..4ea42616ba2 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.cpp @@ -390,7 +390,7 @@ void DiskObjectStorageRemoteMetadataRestoreHelper::restoreFiles(IObjectStorage * }; RelativePathsWithSize children; - source_object_storage->listPrefix(restore_information.source_path, children); + source_object_storage->findAllFiles(restore_information.source_path, children); restore_files(children); @@ -540,7 +540,7 @@ void DiskObjectStorageRemoteMetadataRestoreHelper::restoreFileOperations(IObject }; RelativePathsWithSize children; - source_object_storage->listPrefix(restore_information.source_path + "operations/", children); + source_object_storage->findAllFiles(restore_information.source_path + "operations/", children); restore_file_operations(children); if (restore_information.detached) diff --git a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp index 2a75668dd76..b55fb2c4fa5 100644 --- a/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp +++ b/src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp @@ -599,7 +599,7 @@ std::unique_ptr DiskObjectStorageTransaction::writeFile auto write_operation = std::make_unique(object_storage, metadata_storage, object); std::function create_metadata_callback; - if (autocommit) + if (autocommit) { create_metadata_callback = [tx = shared_from_this(), mode, path, blob_name] (size_t count) { diff --git a/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.cpp b/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.cpp index 2f82458ecd8..80c4bb2bc64 100644 --- a/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.cpp +++ b/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.cpp @@ -101,18 +101,6 @@ std::unique_ptr HDFSObjectStorage::writeObject( /// NOL } -void HDFSObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const -{ - const size_t begin_of_path = path.find('/', path.find("//") + 2); - int32_t num_entries; - auto * files_list = hdfsListDirectory(hdfs_fs.get(), path.substr(begin_of_path).c_str(), &num_entries); - if (num_entries == -1) - throw Exception(ErrorCodes::HDFS_ERROR, "HDFSDelete failed with path: " + path); - - for (int32_t i = 0; i < num_entries; ++i) - children.emplace_back(files_list[i].mName, files_list[i].mSize); -} - /// Remove file. Throws exception if file doesn't exists or it's a directory. void HDFSObjectStorage::removeObject(const StoredObject & object) { diff --git a/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h b/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h index 82cddfb9122..4064a5c5b7f 100644 --- a/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h +++ b/src/Disks/ObjectStorages/HDFS/HDFSObjectStorage.h @@ -85,8 +85,6 @@ public: size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE, const WriteSettings & write_settings = {}) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; - /// Remove file. Throws exception if file doesn't exists or it's a directory. void removeObject(const StoredObject & object) override; diff --git a/src/Disks/ObjectStorages/IMetadataStorage.h b/src/Disks/ObjectStorages/IMetadataStorage.h index 3d6c772157d..597d7744c78 100644 --- a/src/Disks/ObjectStorages/IMetadataStorage.h +++ b/src/Disks/ObjectStorages/IMetadataStorage.h @@ -11,10 +11,16 @@ #include #include #include +#include namespace DB { +namespace ErrorCodes +{ + extern const int NOT_IMPLEMENTED; +} + class IMetadataStorage; /// Tries to provide some "transactions" interface, which allow @@ -33,32 +39,71 @@ public: /// General purpose methods /// Write metadata string to file - virtual void writeStringToFile(const std::string & path, const std::string & data) = 0; + virtual void writeStringToFile(const std::string & /* path */, const std::string & /* data */) + { + throwNotImplemented(); + } - virtual void setLastModified(const std::string & path, const Poco::Timestamp & timestamp) = 0; + virtual void setLastModified(const std::string & /* path */, const Poco::Timestamp & /* timestamp */) + { + throwNotImplemented(); + } virtual bool supportsChmod() const = 0; - virtual void chmod(const String & path, mode_t mode) = 0; + virtual void chmod(const String & /* path */, mode_t /* mode */) + { + throwNotImplemented(); + } - virtual void setReadOnly(const std::string & path) = 0; + virtual void setReadOnly(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void unlinkFile(const std::string & path) = 0; + virtual void unlinkFile(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void createDirectory(const std::string & path) = 0; + virtual void createDirectory(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void createDirectoryRecursive(const std::string & path) = 0; + virtual void createDirectoryRecursive(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void removeDirectory(const std::string & path) = 0; + virtual void removeDirectory(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void removeRecursive(const std::string & path) = 0; + virtual void removeRecursive(const std::string & /* path */) + { + throwNotImplemented(); + } - virtual void createHardLink(const std::string & path_from, const std::string & path_to) = 0; + virtual void createHardLink(const std::string & /* path_from */, const std::string & /* path_to */) + { + throwNotImplemented(); + } - virtual void moveFile(const std::string & path_from, const std::string & path_to) = 0; + virtual void moveFile(const std::string & /* path_from */, const std::string & /* path_to */) + { + throwNotImplemented(); + } - virtual void moveDirectory(const std::string & path_from, const std::string & path_to) = 0; + virtual void moveDirectory(const std::string & /* path_from */, const std::string & /* path_to */) + { + throwNotImplemented(); + } - virtual void replaceFile(const std::string & path_from, const std::string & path_to) = 0; + virtual void replaceFile(const std::string & /* path_from */, const std::string & /* path_to */) + { + throwNotImplemented(); + } /// Metadata related methods @@ -69,7 +114,10 @@ public: virtual void createMetadataFile(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) = 0; /// Add to new blob to metadata file (way to implement appends) - virtual void addBlobToMetadata(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) = 0; + virtual void addBlobToMetadata(const std::string & /* path */, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) + { + throwNotImplemented(); + } /// Unlink metadata file and do something special if required /// By default just remove file (unlink file). @@ -79,6 +127,12 @@ public: } virtual ~IMetadataTransaction() = default; + +private: + [[noreturn]] static void throwNotImplemented() + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Operation is not implemented"); + } }; using MetadataTransactionPtr = std::shared_ptr; @@ -106,12 +160,18 @@ public: virtual Poco::Timestamp getLastModified(const std::string & path) const = 0; - virtual time_t getLastChanged(const std::string & path) const = 0; + virtual time_t getLastChanged(const std::string & /* path */) const + { + throwNotImplemented(); + } virtual bool supportsChmod() const = 0; virtual bool supportsStat() const = 0; - virtual struct stat stat(const String & path) const = 0; + virtual struct stat stat(const String & /* path */) const + { + throwNotImplemented(); + } virtual std::vector listDirectory(const std::string & path) const = 0; @@ -120,20 +180,32 @@ public: virtual uint32_t getHardlinkCount(const std::string & path) const = 0; /// Read metadata file to string from path - virtual std::string readFileToString(const std::string & path) const = 0; + virtual std::string readFileToString(const std::string & /* path */) const + { + throwNotImplemented(); + } virtual ~IMetadataStorage() = default; /// ==== More specific methods. Previous were almost general purpose. ==== /// Read multiple metadata files into strings and return mapping from file_path -> metadata - virtual std::unordered_map getSerializedMetadata(const std::vector & file_paths) const = 0; + virtual std::unordered_map getSerializedMetadata(const std::vector & /* file_paths */) const + { + throwNotImplemented(); + } /// Return object information (absolute_path, bytes_size, ...) for metadata path. /// object_storage_path is absolute. virtual StoredObjects getStorageObjects(const std::string & path) const = 0; virtual std::string getObjectStorageRootPath() const = 0; + +private: + [[noreturn]] static void throwNotImplemented() + { + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Operation is not implemented"); + } }; using MetadataStoragePtr = std::shared_ptr; diff --git a/src/Disks/ObjectStorages/IObjectStorage.cpp b/src/Disks/ObjectStorages/IObjectStorage.cpp index 9d6610ee326..3f8ac566603 100644 --- a/src/Disks/ObjectStorages/IObjectStorage.cpp +++ b/src/Disks/ObjectStorages/IObjectStorage.cpp @@ -14,6 +14,17 @@ namespace ErrorCodes extern const int LOGICAL_ERROR; } +void IObjectStorage::findAllFiles(const std::string &, RelativePathsWithSize &) const +{ + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "findAllFiles() is not supported"); +} +void IObjectStorage::getDirectoryContents(const std::string &, + RelativePathsWithSize &, + std::vector &) const +{ + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getDirectoryContents() is not supported"); +} + IAsynchronousReader & IObjectStorage::getThreadPoolReader() { auto context = Context::getGlobalContextInstance(); diff --git a/src/Disks/ObjectStorages/IObjectStorage.h b/src/Disks/ObjectStorages/IObjectStorage.h index 52e1a2cb270..9451ae31b07 100644 --- a/src/Disks/ObjectStorages/IObjectStorage.h +++ b/src/Disks/ObjectStorages/IObjectStorage.h @@ -65,8 +65,32 @@ public: /// Object exists or not virtual bool exists(const StoredObject & object) const = 0; - /// List on prefix, return children (relative paths) with their sizes. - virtual void listPrefix(const std::string & path, RelativePathsWithSize & children) const = 0; + /// List all objects with specific prefix. + /// + /// For example if you do this over filesystem, you should skip folders and + /// return files only, so something like on local filesystem: + /// + /// find . -type f + /// + /// @param children - out files (relative paths) with their sizes. + /// + /// NOTE: It makes sense only for real object storages (S3, Azure), since + /// it is used only for one of the following: + /// - send_metadata (to restore metadata) + /// - see DiskObjectStorage::restoreMetadataIfNeeded() + /// - MetadataStorageFromPlainObjectStorage - only for s3_plain disk + virtual void findAllFiles(const std::string & path, RelativePathsWithSize & children) const; + + /// Analog of directory content for object storage (object storage does not + /// have "directory" definition, but it can be emulated with usage of + /// "delimiter"), so this is analog of: + /// + /// find . -maxdepth 1 $path + /// + /// Return files in @files and directories in @directories + virtual void getDirectoryContents(const std::string & path, + RelativePathsWithSize & files, + std::vector & directories) const; /// Get object metadata if supported. It should be possible to receive /// at least size of object diff --git a/src/Disks/ObjectStorages/LocalObjectStorage.cpp b/src/Disks/ObjectStorages/LocalObjectStorage.cpp index dbb3a7c2aba..67e2cc2d74b 100644 --- a/src/Disks/ObjectStorages/LocalObjectStorage.cpp +++ b/src/Disks/ObjectStorages/LocalObjectStorage.cpp @@ -104,13 +104,6 @@ std::unique_ptr LocalObjectStorage::writeObject( /// NO return std::make_unique(path, buf_size, flags); } -void LocalObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const -{ - fs::directory_iterator end_it; - for (auto it = fs::directory_iterator(path); it != end_it; ++it) - children.emplace_back(it->path().filename(), it->file_size()); -} - void LocalObjectStorage::removeObject(const StoredObject & object) { /// For local object storage files are actually removed when "metadata" is removed. diff --git a/src/Disks/ObjectStorages/LocalObjectStorage.h b/src/Disks/ObjectStorages/LocalObjectStorage.h index 0e4c71b4a47..b04e3fa6285 100644 --- a/src/Disks/ObjectStorages/LocalObjectStorage.h +++ b/src/Disks/ObjectStorages/LocalObjectStorage.h @@ -45,8 +45,6 @@ public: size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE, const WriteSettings & write_settings = {}) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; - void removeObject(const StoredObject & object) override; void removeObjects(const StoredObjects & objects) override; diff --git a/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp b/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp index 35cd3be15d2..259f6e01fd7 100644 --- a/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp +++ b/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp @@ -12,7 +12,6 @@ namespace DB namespace ErrorCodes { - extern const int NOT_IMPLEMENTED; extern const int LOGICAL_ERROR; } @@ -33,194 +32,102 @@ const std::string & MetadataStorageFromPlainObjectStorage::getPath() const { return object_storage_root_path; } +std::filesystem::path MetadataStorageFromPlainObjectStorage::getAbsolutePath(const std::string & path) const +{ + return fs::path(object_storage_root_path) / path; +} bool MetadataStorageFromPlainObjectStorage::exists(const std::string & path) const { - auto object = StoredObject::create(*object_storage, fs::path(object_storage_root_path) / path); + auto object = StoredObject::create(*object_storage, getAbsolutePath(path)); return object_storage->exists(object); } bool MetadataStorageFromPlainObjectStorage::isFile(const std::string & path) const { /// NOTE: This check is inaccurate and has excessive API calls - return !isDirectory(path) && exists(path); + return exists(path) && !isDirectory(path); } bool MetadataStorageFromPlainObjectStorage::isDirectory(const std::string & path) const { - std::string directory = path; + std::string directory = getAbsolutePath(path); trimRight(directory); directory += "/"; /// NOTE: This check is far from ideal, since it work only if the directory /// really has files, and has excessive API calls - RelativePathsWithSize children; - object_storage->listPrefix(directory, children); - return !children.empty(); -} - -Poco::Timestamp MetadataStorageFromPlainObjectStorage::getLastModified(const std::string &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getLastModified is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -struct stat MetadataStorageFromPlainObjectStorage::stat(const std::string &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "stat is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -time_t MetadataStorageFromPlainObjectStorage::getLastChanged(const std::string &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getLastChanged is not implemented for MetadataStorageFromPlainObjectStorage"); + RelativePathsWithSize files; + std::vector directories; + object_storage->getDirectoryContents(directory, files, directories); + return !files.empty() || !directories.empty(); } uint64_t MetadataStorageFromPlainObjectStorage::getFileSize(const String & path) const { RelativePathsWithSize children; - object_storage->listPrefix(path, children); + object_storage->findAllFiles(getAbsolutePath(path), children); if (children.empty()) return 0; if (children.size() != 1) - throw Exception(ErrorCodes::LOGICAL_ERROR, "listPrefix() return multiple paths ({}) for {}", children.size(), path); + throw Exception(ErrorCodes::LOGICAL_ERROR, "findAllFiles() return multiple paths ({}) for {}", children.size(), path); return children.front().bytes_size; } std::vector MetadataStorageFromPlainObjectStorage::listDirectory(const std::string & path) const { - RelativePathsWithSize children; - object_storage->listPrefix(path, children); + RelativePathsWithSize files; + std::vector directories; + object_storage->getDirectoryContents(getAbsolutePath(path), files, directories); std::vector result; - for (const auto & path_size : children) - { + for (const auto & path_size : files) result.push_back(path_size.relative_path); - } + for (const auto & directory : directories) + result.push_back(directory); return result; } DirectoryIteratorPtr MetadataStorageFromPlainObjectStorage::iterateDirectory(const std::string & path) const { - /// NOTE: this is not required for BACKUP/RESTORE, but this is a first step - /// towards MergeTree on plain S3. + /// Required for MergeTree auto paths = listDirectory(path); std::vector fs_paths(paths.begin(), paths.end()); return std::make_unique(std::move(fs_paths)); } -std::string MetadataStorageFromPlainObjectStorage::readFileToString(const std::string &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "readFileToString is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -std::unordered_map MetadataStorageFromPlainObjectStorage::getSerializedMetadata(const std::vector &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getSerializedMetadata is not implemented for MetadataStorageFromPlainObjectStorage"); -} - StoredObjects MetadataStorageFromPlainObjectStorage::getStorageObjects(const std::string & path) const { std::string blob_name = object_storage->generateBlobNameForPath(path); - - std::string object_path = fs::path(object_storage_root_path) / blob_name; - size_t object_size = getFileSize(object_path); - - auto object = StoredObject::create(*object_storage, object_path, object_size, /* exists */true); + size_t object_size = getFileSize(blob_name); + auto object = StoredObject::create(*object_storage, getAbsolutePath(blob_name), object_size, /* exists */true); return {std::move(object)}; } -uint32_t MetadataStorageFromPlainObjectStorage::getHardlinkCount(const std::string &) const -{ - return 1; -} - const IMetadataStorage & MetadataStorageFromPlainObjectStorageTransaction::getStorageForNonTransactionalReads() const { return metadata_storage; } -void MetadataStorageFromPlainObjectStorageTransaction::writeStringToFile(const std::string &, const std::string & /* data */) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "writeStringToFile is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::setLastModified(const std::string &, const Poco::Timestamp &) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "setLastModified is not implemented for MetadataStorageFromPlainObjectStorage"); -} - void MetadataStorageFromPlainObjectStorageTransaction::unlinkFile(const std::string & path) { - auto object = StoredObject::create(*metadata_storage.object_storage, fs::path(metadata_storage.object_storage_root_path) / path); + auto object = StoredObject::create(*metadata_storage.object_storage, metadata_storage.getAbsolutePath(path)); metadata_storage.object_storage->removeObject(object); } -void MetadataStorageFromPlainObjectStorageTransaction::removeRecursive(const std::string &) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "removeRecursive is not implemented for MetadataStorageFromPlainObjectStorage"); -} - void MetadataStorageFromPlainObjectStorageTransaction::createDirectory(const std::string &) { /// Noop. It is an Object Storage not a filesystem. } - void MetadataStorageFromPlainObjectStorageTransaction::createDirectoryRecursive(const std::string &) { /// Noop. It is an Object Storage not a filesystem. } - -void MetadataStorageFromPlainObjectStorageTransaction::removeDirectory(const std::string &) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "removeDirectory is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::moveFile(const std::string & /* path_from */, const std::string & /* path_to */) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "moveFile is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::moveDirectory(const std::string & /* path_from */, const std::string & /* path_to */) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "moveDirectory is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::replaceFile(const std::string & /* path_from */, const std::string & /* path_to */) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "replaceFile is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::chmod(const String &, mode_t) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "chmod is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::setReadOnly(const std::string &) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "setReadOnly is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::createHardLink(const std::string & /* path_from */, const std::string & /* path_to */) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "createHardLink is not implemented for MetadataStorageFromPlainObjectStorage"); -} - -void MetadataStorageFromPlainObjectStorageTransaction::createEmptyMetadataFile(const std::string &) -{ - /// Noop, no separate metadata. -} - -void MetadataStorageFromPlainObjectStorageTransaction::createMetadataFile( - const std::string &, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) -{ - /// Noop, no separate metadata. -} - void MetadataStorageFromPlainObjectStorageTransaction::addBlobToMetadata( const std::string &, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) { /// Noop, local metadata files is only one file, it is the metadata file itself. } - void MetadataStorageFromPlainObjectStorageTransaction::unlinkMetadata(const std::string &) { /// Noop, no separate metadata. diff --git a/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.h b/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.h index bd993918413..99cc960b9e4 100644 --- a/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.h +++ b/src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.h @@ -45,31 +45,32 @@ public: uint64_t getFileSize(const String & path) const override; - Poco::Timestamp getLastModified(const std::string & path) const override; - - time_t getLastChanged(const std::string & path) const override; - - bool supportsChmod() const override { return false; } - - bool supportsStat() const override { return false; } - - struct stat stat(const String & path) const override; - std::vector listDirectory(const std::string & path) const override; DirectoryIteratorPtr iterateDirectory(const std::string & path) const override; - std::string readFileToString(const std::string & path) const override; - - std::unordered_map getSerializedMetadata(const std::vector & file_paths) const override; - - uint32_t getHardlinkCount(const std::string & path) const override; - DiskPtr getDisk() const { return {}; } StoredObjects getStorageObjects(const std::string & path) const override; std::string getObjectStorageRootPath() const override { return object_storage_root_path; } + + Poco::Timestamp getLastModified(const std::string & /* path */) const override + { + /// Required by MergeTree + return {}; + } + + uint32_t getHardlinkCount(const std::string & /* path */) const override + { + return 1; + } + + bool supportsChmod() const override { return false; } + bool supportsStat() const override { return false; } + +private: + std::filesystem::path getAbsolutePath(const std::string & path) const; }; class MetadataStorageFromPlainObjectStorageTransaction final : public IMetadataTransaction @@ -83,47 +84,34 @@ public: : metadata_storage(metadata_storage_) {} - ~MetadataStorageFromPlainObjectStorageTransaction() override = default; - - const IMetadataStorage & getStorageForNonTransactionalReads() const final; - - void commit() final {} - - void writeStringToFile(const std::string & path, const std::string & data) override; - - void createEmptyMetadataFile(const std::string & path) override; - - void createMetadataFile(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) override; + const IMetadataStorage & getStorageForNonTransactionalReads() const override; void addBlobToMetadata(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) override; - void setLastModified(const std::string & path, const Poco::Timestamp & timestamp) override; + void createEmptyMetadataFile(const std::string & /* path */) override + { + /// No metadata, no need to create anything. + } - bool supportsChmod() const override { return false; } - - void chmod(const String & path, mode_t mode) override; - - void setReadOnly(const std::string & path) override; - - void unlinkFile(const std::string & path) override; + void createMetadataFile(const std::string & /* path */, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) override + { + /// Noop + } void createDirectory(const std::string & path) override; void createDirectoryRecursive(const std::string & path) override; - void removeDirectory(const std::string & path) override; - - void removeRecursive(const std::string & path) override; - - void createHardLink(const std::string & path_from, const std::string & path_to) override; - - void moveFile(const std::string & path_from, const std::string & path_to) override; - - void moveDirectory(const std::string & path_from, const std::string & path_to) override; - - void replaceFile(const std::string & path_from, const std::string & path_to) override; + void unlinkFile(const std::string & path) override; void unlinkMetadata(const std::string & path) override; + + void commit() override + { + /// Nothing to commit. + } + + bool supportsChmod() const override { return false; } }; } diff --git a/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp b/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp index 83908d02f48..0c421ee03d7 100644 --- a/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp +++ b/src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp @@ -28,7 +28,7 @@ #include #include - +#include #include #include @@ -248,7 +248,7 @@ std::unique_ptr S3ObjectStorage::writeObject( /// NOLIN std::move(s3_buffer), std::move(finalize_callback), object.absolute_path); } -void S3ObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const +void S3ObjectStorage::findAllFiles(const std::string & path, RelativePathsWithSize & children) const { auto settings_ptr = s3_settings.get(); auto client_ptr = client.get(); @@ -279,6 +279,49 @@ void S3ObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize } while (outcome.GetResult().GetIsTruncated()); } +void S3ObjectStorage::getDirectoryContents(const std::string & path, + RelativePathsWithSize & files, + std::vector & directories) const +{ + auto settings_ptr = s3_settings.get(); + auto client_ptr = client.get(); + + Aws::S3::Model::ListObjectsV2Request request; + request.SetBucket(bucket); + request.SetPrefix(path); + request.SetMaxKeys(settings_ptr->list_object_keys_size); + request.SetDelimiter("/"); + + Aws::S3::Model::ListObjectsV2Outcome outcome; + do + { + ProfileEvents::increment(ProfileEvents::S3ListObjects); + ProfileEvents::increment(ProfileEvents::DiskS3ListObjects); + outcome = client_ptr->ListObjectsV2(request); + throwIfError(outcome); + + auto result = outcome.GetResult(); + auto result_objects = result.GetContents(); + auto result_common_prefixes = result.GetCommonPrefixes(); + + if (result_objects.empty() && result_common_prefixes.empty()) + break; + + for (const auto & object : result_objects) + files.emplace_back(object.GetKey(), object.GetSize()); + + for (const auto & common_prefix : result_common_prefixes) + { + std::string directory = common_prefix.GetPrefix(); + /// Make it compatible with std::filesystem::path::filename() + trimRight(directory, '/'); + directories.emplace_back(directory); + } + + request.SetContinuationToken(outcome.GetResult().GetNextContinuationToken()); + } while (outcome.GetResult().GetIsTruncated()); +} + void S3ObjectStorage::removeObjectImpl(const StoredObject & object, bool if_exists) { auto client_ptr = client.get(); diff --git a/src/Disks/ObjectStorages/S3/S3ObjectStorage.h b/src/Disks/ObjectStorages/S3/S3ObjectStorage.h index a193653db9a..6b1e8289b15 100644 --- a/src/Disks/ObjectStorages/S3/S3ObjectStorage.h +++ b/src/Disks/ObjectStorages/S3/S3ObjectStorage.h @@ -105,7 +105,10 @@ public: size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE, const WriteSettings & write_settings = {}) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; + void findAllFiles(const std::string & path, RelativePathsWithSize & children) const override; + void getDirectoryContents(const std::string & path, + RelativePathsWithSize & files, + std::vector & directories) const override; /// Uses `DeleteObjectRequest`. void removeObject(const StoredObject & object) override; diff --git a/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.cpp b/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.cpp index 06e36a2ddd8..12c2cd16a9f 100644 --- a/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.cpp +++ b/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.cpp @@ -12,7 +12,6 @@ namespace DB namespace ErrorCodes { - extern const int NOT_IMPLEMENTED; extern const int FILE_DOESNT_EXIST; extern const int NETWORK_ERROR; } @@ -55,6 +54,9 @@ bool MetadataStorageFromStaticFilesWebServer::exists(const std::string & path) c path, [](const auto & file, const std::string & path_) { return file.first < path_; } ); + if (it == object_storage.files.end()) + return false; + if (startsWith(it->first, path) || (it != object_storage.files.begin() && startsWith(std::prev(it)->first, path))) return true; @@ -165,91 +167,11 @@ DirectoryIteratorPtr MetadataStorageFromStaticFilesWebServer::iterateDirectory(c return std::make_unique(std::move(dir_file_paths)); } -std::string MetadataStorageFromStaticFilesWebServer::readFileToString(const std::string &) const -{ - WebObjectStorage::throwNotAllowed(); -} - -Poco::Timestamp MetadataStorageFromStaticFilesWebServer::getLastModified(const std::string &) const -{ - return {}; -} - -time_t MetadataStorageFromStaticFilesWebServer::getLastChanged(const std::string &) const -{ - return {}; -} - -uint32_t MetadataStorageFromStaticFilesWebServer::getHardlinkCount(const std::string &) const -{ - return 1; -} - const IMetadataStorage & MetadataStorageFromStaticFilesWebServerTransaction::getStorageForNonTransactionalReads() const { return metadata_storage; } -void MetadataStorageFromStaticFilesWebServerTransaction::writeStringToFile(const std::string &, const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::setLastModified(const std::string &, const Poco::Timestamp &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::unlinkFile(const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::removeRecursive(const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::removeDirectory(const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::moveFile(const std::string &, const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::moveDirectory(const std::string &, const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::replaceFile(const std::string &, const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::setReadOnly(const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::createHardLink(const std::string &, const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::addBlobToMetadata(const std::string &, const std::string &, uint64_t) -{ - WebObjectStorage::throwNotAllowed(); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::unlinkMetadata(const std::string &) -{ - WebObjectStorage::throwNotAllowed(); -} - void MetadataStorageFromStaticFilesWebServerTransaction::createDirectory(const std::string &) { /// Noop. @@ -260,30 +182,4 @@ void MetadataStorageFromStaticFilesWebServerTransaction::createDirectoryRecursiv /// Noop. } -void MetadataStorageFromStaticFilesWebServerTransaction::createEmptyMetadataFile(const std::string & /* path */) -{ - /// Noop. -} - -void MetadataStorageFromStaticFilesWebServerTransaction::createMetadataFile( - const std::string & /* path */, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) -{ - /// Noop. -} - -void MetadataStorageFromStaticFilesWebServerTransaction::commit() -{ - /// Noop. -} - -std::unordered_map MetadataStorageFromStaticFilesWebServer::getSerializedMetadata(const std::vector &) const -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getSerializedMetadata is not implemented for MetadataStorageFromStaticFilesWebServer"); -} - -void MetadataStorageFromStaticFilesWebServerTransaction::chmod(const String &, mode_t) -{ - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "chmod is not implemented for MetadataStorageFromStaticFilesWebServer"); -} - } diff --git a/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.h b/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.h index 27a1ae8b8fa..338a2690b8f 100644 --- a/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.h +++ b/src/Disks/ObjectStorages/Web/MetadataStorageFromStaticFilesWebServer.h @@ -36,29 +36,28 @@ public: uint64_t getFileSize(const String & path) const override; - Poco::Timestamp getLastModified(const std::string & path) const override; - - time_t getLastChanged(const std::string & path) const override; - std::vector listDirectory(const std::string & path) const override; DirectoryIteratorPtr iterateDirectory(const std::string & path) const override; - std::string readFileToString(const std::string & path) const override; - - std::unordered_map getSerializedMetadata(const std::vector & file_paths) const override; - - uint32_t getHardlinkCount(const std::string & path) const override; - StoredObjects getStorageObjects(const std::string & path) const override; std::string getObjectStorageRootPath() const override { return ""; } + struct stat stat(const String & /* path */) const override { return {}; } + + Poco::Timestamp getLastModified(const std::string & /* path */) const override + { + /// Required by MergeTree + return {}; + } + uint32_t getHardlinkCount(const std::string & /* path */) const override + { + return 1; + } + bool supportsChmod() const override { return false; } - bool supportsStat() const override { return false; } - - struct stat stat(const String &) const override { return {}; } }; class MetadataStorageFromStaticFilesWebServerTransaction final : public IMetadataTransaction @@ -73,47 +72,28 @@ public: : metadata_storage(metadata_storage_) {} - ~MetadataStorageFromStaticFilesWebServerTransaction() override = default; - const IMetadataStorage & getStorageForNonTransactionalReads() const override; - void commit() override; + void createEmptyMetadataFile(const std::string & /* path */) override + { + /// No metadata, no need to create anything. + } - void writeStringToFile(const std::string & path, const std::string & data) override; - - void createEmptyMetadataFile(const std::string & path) override; - - void createMetadataFile(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) override; - - void addBlobToMetadata(const std::string & path, const std::string & blob_name, uint64_t size_in_bytes) override; - - void setLastModified(const std::string & path, const Poco::Timestamp & timestamp) override; - - void setReadOnly(const std::string & path) override; - - void unlinkFile(const std::string & path) override; + void createMetadataFile(const std::string & /* path */, const std::string & /* blob_name */, uint64_t /* size_in_bytes */) override + { + /// Noop + } void createDirectory(const std::string & path) override; void createDirectoryRecursive(const std::string & path) override; - void removeDirectory(const std::string & path) override; - - void removeRecursive(const std::string & path) override; - - void createHardLink(const std::string & path_from, const std::string & path_to) override; - - void moveFile(const std::string & path_from, const std::string & path_to) override; - - void moveDirectory(const std::string & path_from, const std::string & path_to) override; - - void replaceFile(const std::string & path_from, const std::string & path_to) override; - - void unlinkMetadata(const std::string & path) override; + void commit() override + { + /// Nothing to commit. + } bool supportsChmod() const override { return false; } - - void chmod(const String &, mode_t) override; }; } diff --git a/src/Disks/ObjectStorages/Web/WebObjectStorage.cpp b/src/Disks/ObjectStorages/Web/WebObjectStorage.cpp index 71bde110fa6..f97409cfc6c 100644 --- a/src/Disks/ObjectStorages/Web/WebObjectStorage.cpp +++ b/src/Disks/ObjectStorages/Web/WebObjectStorage.cpp @@ -178,17 +178,6 @@ std::unique_ptr WebObjectStorage::readObject( /// NOLINT } } -void WebObjectStorage::listPrefix(const std::string & path, RelativePathsWithSize & children) const -{ - for (const auto & [file_path, file_info] : files) - { - if (file_info.type == FileType::File && file_path.starts_with(path)) - { - children.emplace_back(file_path, file_info.size); - } - } -} - void WebObjectStorage::throwNotAllowed() { throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Only read-only operations are supported"); diff --git a/src/Disks/ObjectStorages/Web/WebObjectStorage.h b/src/Disks/ObjectStorages/Web/WebObjectStorage.h index 2fda5e576aa..2dab8fdb62d 100644 --- a/src/Disks/ObjectStorages/Web/WebObjectStorage.h +++ b/src/Disks/ObjectStorages/Web/WebObjectStorage.h @@ -55,8 +55,6 @@ public: size_t buf_size = DBMS_DEFAULT_BUFFER_SIZE, const WriteSettings & write_settings = {}) override; - void listPrefix(const std::string & path, RelativePathsWithSize & children) const override; - void removeObject(const StoredObject & object) override; void removeObjects(const StoredObjects & objects) override; diff --git a/src/Disks/StoragePolicy.cpp b/src/Disks/StoragePolicy.cpp index 10513c6beae..10524ffcc0f 100644 --- a/src/Disks/StoragePolicy.cpp +++ b/src/Disks/StoragePolicy.cpp @@ -26,7 +26,6 @@ namespace ErrorCodes extern const int BAD_ARGUMENTS; extern const int EXCESSIVE_ELEMENT_IN_CONFIG; extern const int NO_ELEMENTS_IN_CONFIG; - extern const int UNKNOWN_DISK; extern const int UNKNOWN_POLICY; extern const int UNKNOWN_VOLUME; extern const int LOGICAL_ERROR; @@ -311,22 +310,12 @@ void StoragePolicy::checkCompatibleWith(const StoragePolicyPtr & new_storage_pol } -size_t StoragePolicy::getVolumeIndexByDisk(const DiskPtr & disk_ptr) const +std::optional StoragePolicy::tryGetVolumeIndexByDiskName(const String & disk_name) const { - auto it = volume_index_by_disk_name.find(disk_ptr->getName()); + auto it = volume_index_by_disk_name.find(disk_name); if (it != volume_index_by_disk_name.end()) return it->second; - else - throw Exception("No disk " + backQuote(disk_ptr->getName()) + " in policy " + backQuote(name), ErrorCodes::UNKNOWN_DISK); -} - - -VolumePtr StoragePolicy::tryGetVolumeByDisk(const DiskPtr & disk_ptr) const -{ - auto it = volume_index_by_disk_name.find(disk_ptr->getName()); - if (it == volume_index_by_disk_name.end()) - return nullptr; - return getVolume(it->second); + return {}; } diff --git a/src/Disks/StoragePolicy.h b/src/Disks/StoragePolicy.h index fd0169a6ebe..9631f1c2e52 100644 --- a/src/Disks/StoragePolicy.h +++ b/src/Disks/StoragePolicy.h @@ -68,7 +68,7 @@ public: ReservationPtr reserve(UInt64 bytes, size_t min_volume_index) const override; /// Find volume index, which contains disk - size_t getVolumeIndexByDisk(const DiskPtr & disk_ptr) const override; + std::optional tryGetVolumeIndexByDiskName(const String & disk_name) const override; /// Reserves 0 bytes on disk with max available space /// Do not use this function when it is possible to predict size. @@ -85,9 +85,6 @@ public: VolumePtr tryGetVolumeByName(const String & volume_name) const override; - /// Finds a volume which contains a specified disk. - VolumePtr tryGetVolumeByDisk(const DiskPtr & disk_ptr) const override; - /// Checks if storage policy can be replaced by another one. void checkCompatibleWith(const StoragePolicyPtr & new_storage_policy) const override; diff --git a/src/Formats/EscapingRuleUtils.cpp b/src/Formats/EscapingRuleUtils.cpp index e47525d089a..e80ab50968d 100644 --- a/src/Formats/EscapingRuleUtils.cpp +++ b/src/Formats/EscapingRuleUtils.cpp @@ -859,7 +859,7 @@ String getAdditionalFormatInfoByEscapingRule(const FormatSettings & settings, Fo result += fmt::format( ", use_best_effort_in_schema_inference={}, bool_true_representation={}, bool_false_representation={}," " null_representation={}, delimiter={}, tuple_delimiter={}", - settings.tsv.use_best_effort_in_schema_inference, + settings.csv.use_best_effort_in_schema_inference, settings.bool_true_representation, settings.bool_false_representation, settings.csv.null_representation, diff --git a/src/Formats/NativeWriter.cpp b/src/Formats/NativeWriter.cpp index 0cae2a2e789..c4dea371afd 100644 --- a/src/Formats/NativeWriter.cpp +++ b/src/Formats/NativeWriter.cpp @@ -58,7 +58,7 @@ static void writeData(const ISerialization & serialization, const ColumnPtr & co settings.low_cardinality_max_dictionary_size = 0; //-V1048 ISerialization::SerializeBinaryBulkStatePtr state; - serialization.serializeBinaryBulkStatePrefix(settings, state); + serialization.serializeBinaryBulkStatePrefix(*full_column, settings, state); serialization.serializeBinaryBulkWithMultipleStreams(*full_column, offset, limit, settings, state); serialization.serializeBinaryBulkStateSuffix(settings, state); } diff --git a/src/Formats/newLineSegmentationEngine.cpp b/src/Formats/newLineSegmentationEngine.cpp new file mode 100644 index 00000000000..a605bba7e5b --- /dev/null +++ b/src/Formats/newLineSegmentationEngine.cpp @@ -0,0 +1,50 @@ +#include +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +std::pair newLineFileSegmentationEngine(ReadBuffer & in, DB::Memory<> & memory, size_t min_bytes, size_t max_rows) +{ + char * pos = in.position(); + bool need_more_data = true; + size_t number_of_rows = 0; + + while (loadAtPosition(in, memory, pos) && need_more_data) + { + pos = find_first_symbols<'\r', '\n'>(pos, in.buffer().end()); + if (pos > in.buffer().end()) + throw Exception("Position in buffer is out of bounds. There must be a bug.", ErrorCodes::LOGICAL_ERROR); + else if (pos == in.buffer().end()) + continue; + + ++number_of_rows; + if ((memory.size() + static_cast(pos - in.position()) >= min_bytes) || (number_of_rows == max_rows)) + need_more_data = false; + + if (*pos == '\n') + { + ++pos; + if (loadAtPosition(in, memory, pos) && *pos == '\r') + ++pos; + } + else if (*pos == '\r') + { + ++pos; + if (loadAtPosition(in, memory, pos) && *pos == '\n') + ++pos; + } + } + + saveUpToPosition(in, memory, pos); + + return {loadAtPosition(in, memory, pos), number_of_rows}; +} + +} diff --git a/src/Formats/newLineSegmentationEngine.h b/src/Formats/newLineSegmentationEngine.h new file mode 100644 index 00000000000..598f808b798 --- /dev/null +++ b/src/Formats/newLineSegmentationEngine.h @@ -0,0 +1,9 @@ +#pragma once + +#include +#include + +namespace DB +{ + std::pair newLineFileSegmentationEngine(ReadBuffer & in, DB::Memory<> & memory, size_t min_bytes, size_t max_rows); +} diff --git a/src/Formats/registerFormats.cpp b/src/Formats/registerFormats.cpp index 593e4568be1..ba40fe442ab 100644 --- a/src/Formats/registerFormats.cpp +++ b/src/Formats/registerFormats.cpp @@ -18,6 +18,7 @@ void registerFileSegmentationEngineJSONCompactEachRow(FormatFactory & factory); #if USE_HIVE void registerFileSegmentationEngineHiveText(FormatFactory & factory); #endif +void registerFileSegmentationEngineLineAsString(FormatFactory & factory); /// Formats for both input/output. @@ -153,6 +154,7 @@ void registerFormats() #if USE_HIVE registerFileSegmentationEngineHiveText(factory); #endif + registerFileSegmentationEngineLineAsString(factory); registerInputFormatNative(factory); diff --git a/src/Functions/FunctionBase64Conversion.h b/src/Functions/FunctionBase64Conversion.h index d7e4ea3486d..ef340a33149 100644 --- a/src/Functions/FunctionBase64Conversion.h +++ b/src/Functions/FunctionBase64Conversion.h @@ -2,21 +2,19 @@ #include "config.h" #if USE_BASE64 -# include -# include +# include # include # include -# include # include -# include -# include +# include +# include # include +# include +# include namespace DB { -using namespace GatherUtils; - namespace ErrorCodes { extern const int BAD_ARGUMENTS; @@ -25,33 +23,86 @@ namespace ErrorCodes extern const int INCORRECT_DATA; } +namespace Detail +{ + inline size_t base64Decode(const std::span src, UInt8 * dst) + { +# if defined(__aarch64__) + return tb64sdec(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); +# else + return _tb64d(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); +# endif + } +} + struct Base64Encode { static constexpr auto name = "base64Encode"; - static size_t getBufferSize(size_t string_length, size_t string_count) + + static size_t getBufferSize(const size_t string_length, const size_t string_count) { return ((string_length - string_count) / 3 + string_count) * 4 + string_count; } + + static size_t performCoding(const std::span src, UInt8 * dst) + { + /* + * Some bug in sse arm64 implementation? + * `base64Encode(repeat('a', 46))` returns wrong padding character + */ +# if defined(__aarch64__) + return tb64senc(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); +# else + return _tb64e(reinterpret_cast(src.data()), src.size(), reinterpret_cast(dst)); +# endif + } }; struct Base64Decode { static constexpr auto name = "base64Decode"; - static size_t getBufferSize(size_t string_length, size_t string_count) + static size_t getBufferSize(const size_t string_length, const size_t string_count) { return ((string_length - string_count) / 4 + string_count) * 3 + string_count; } + + static size_t performCoding(const std::span src, UInt8 * dst) + { + const auto outlen = Detail::base64Decode(src, dst); + if (src.size() > 0 && !outlen) + throw Exception( + ErrorCodes::INCORRECT_DATA, + "Failed to {} input '{}'", + name, + String(reinterpret_cast(src.data()), src.size())); + + return outlen; + } }; struct TryBase64Decode { static constexpr auto name = "tryBase64Decode"; - static size_t getBufferSize(size_t string_length, size_t string_count) + static size_t getBufferSize(const size_t string_length, const size_t string_count) { return Base64Decode::getBufferSize(string_length, string_count); } + + static size_t performCoding(const std::span src, UInt8 * dst) + { + if (src.empty()) + return 0; + + const auto outlen = Detail::base64Decode(src, dst); + // during decoding character array can be partially polluted + // if fail, revert back and clean + if (!outlen) + *dst = 0; + + return outlen; + } }; template @@ -71,99 +122,60 @@ public: if (arguments.size() != 1) throw Exception(ErrorCodes::BAD_ARGUMENTS, "Wrong number of arguments for function {}: 1 expected.", getName()); - if (!WhichDataType(arguments[0].type).isString()) + if (!WhichDataType(arguments[0].type).isStringOrFixedString()) throw Exception( ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, - "Illegal type {} of 1st argument of function {}. Must be String.", - arguments[0].type->getName(), getName()); + "Illegal type {} of 1st argument of function {}. Must be FixedString or String.", + arguments[0].type->getName(), + getName()); return std::make_shared(); } - ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, size_t input_rows_count) const override + ColumnPtr executeImpl(const ColumnsWithTypeAndName & arguments, const DataTypePtr &, const size_t input_rows_count) const override { - const ColumnPtr column_string = arguments[0].column; - const ColumnString * input = checkAndGetColumn(column_string.get()); + const auto & input_column = arguments[0].column; + if (const auto * src_column_as_fixed_string = checkAndGetColumn(*input_column)) + return execute(*src_column_as_fixed_string, input_rows_count); + else if (const auto * src_column_as_string = checkAndGetColumn(*input_column)) + return execute(*src_column_as_string, input_rows_count); - if (!input) - throw Exception( - ErrorCodes::ILLEGAL_COLUMN, - "Illegal column {} of first argument of function {}, must be of type String", - arguments[0].column->getName(), getName()); + throw Exception( + ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of first argument of function {}, must be of type FixedString or String.", + input_column->getName(), + getName()); + } +private: + static ColumnPtr execute(const ColumnString & src_column, const size_t src_row_count) + { auto dst_column = ColumnString::create(); - auto & dst_data = dst_column->getChars(); + auto & dst_chars = dst_column->getChars(); auto & dst_offsets = dst_column->getOffsets(); - size_t reserve = Func::getBufferSize(input->getChars().size(), input->size()); - dst_data.resize(reserve); - dst_offsets.resize(input_rows_count); + const auto reserve = Func::getBufferSize(src_column.byteSize(), src_column.size()); + dst_chars.resize(reserve); + dst_offsets.resize(src_row_count); - const ColumnString::Offsets & src_offsets = input->getOffsets(); + const auto & src_chars = src_column.getChars(); + const auto & src_offsets = src_column.getOffsets(); - const auto * source = input->getChars().data(); - auto * dst = dst_data.data(); + auto * dst = dst_chars.data(); auto * dst_pos = dst; + const auto * src = src_chars.data(); size_t src_offset_prev = 0; - - for (size_t row = 0; row < input_rows_count; ++row) + for (size_t row = 0; row < src_row_count; ++row) { - size_t srclen = src_offsets[row] - src_offset_prev - 1; - size_t outlen = 0; - - if constexpr (std::is_same_v) - { - /* - * Some bug in sse arm64 implementation? - * `base64Encode(repeat('a', 46))` returns wrong padding character - */ -#if defined(__aarch64__) - outlen = tb64senc(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); -#else - outlen = _tb64e(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); -#endif - } - else if constexpr (std::is_same_v) - { - if (srclen > 0) - { -#if defined(__aarch64__) - outlen = tb64sdec(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); -#else - outlen = _tb64d(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); -#endif - - if (!outlen) - throw Exception( - ErrorCodes::INCORRECT_DATA, - "Failed to {} input '{}'", - getName(), String(reinterpret_cast(source), srclen)); - } - } - else - { - if (srclen > 0) - { - // during decoding character array can be partially polluted - // if fail, revert back and clean - auto * savepoint = dst_pos; - outlen = _tb64d(reinterpret_cast(source), srclen, reinterpret_cast(dst_pos)); - if (!outlen) - { - outlen = 0; - dst_pos = savepoint; //-V1048 - // clean the symbol - dst_pos[0] = 0; - } - } - } + const size_t src_length = src_offsets[row] - src_offset_prev - 1; + const auto outlen = Func::performCoding({src, src_length}, dst_pos); /// Base64 library is using AVX-512 with some shuffle operations. /// Memory sanitizer don't understand if there was uninitialized memory in SIMD register but it was not used in the result of shuffle. __msan_unpoison(dst_pos, outlen); - source += srclen + 1; + src += src_length + 1; dst_pos += outlen; *dst_pos = '\0'; dst_pos += 1; @@ -172,8 +184,44 @@ public: src_offset_prev = src_offsets[row]; } - dst_data.resize(dst_pos - dst); + dst_chars.resize(dst_pos - dst); + return dst_column; + } + static ColumnPtr execute(const ColumnFixedString & src_column, const size_t src_row_count) + { + auto dst_column = ColumnString::create(); + auto & dst_chars = dst_column->getChars(); + auto & dst_offsets = dst_column->getOffsets(); + + const auto reserve = Func::getBufferSize(src_column.byteSize(), src_column.size()); + dst_chars.resize(reserve); + dst_offsets.resize(src_row_count); + + const auto & src_chars = src_column.getChars(); + const auto & src_n = src_column.getN(); + + auto * dst = dst_chars.data(); + auto * dst_pos = dst; + const auto * src = src_chars.data(); + + for (size_t row = 0; row < src_row_count; ++row) + { + const auto outlen = Func::performCoding({src, src_n}, dst_pos); + + /// Base64 library is using AVX-512 with some shuffle operations. + /// Memory sanitizer don't understand if there was uninitialized memory in SIMD register but it was not used in the result of shuffle. + __msan_unpoison(dst_pos, outlen); + + src += src_n; + dst_pos += outlen; + *dst_pos = '\0'; + dst_pos += 1; + + dst_offsets[row] = dst_pos - dst; + } + + dst_chars.resize(dst_pos - dst); return dst_column; } }; diff --git a/src/Functions/FunctionsConversion.h b/src/Functions/FunctionsConversion.h index 3a69ba43642..dd494d821bf 100644 --- a/src/Functions/FunctionsConversion.h +++ b/src/Functions/FunctionsConversion.h @@ -3360,9 +3360,8 @@ private: { return [] (ColumnsWithTypeAndName & arguments, const DataTypePtr & result_type, const ColumnNullable * nullable_source, size_t input_rows_count) { - auto res = ConvertImplGenericFromString::execute(arguments, result_type, nullable_source, input_rows_count); - auto & res_object = assert_cast(res->assumeMutableRef()); - res_object.finalize(); + auto res = ConvertImplGenericFromString::execute(arguments, result_type, nullable_source, input_rows_count)->assumeMutable(); + res->finalize(); return res; }; } diff --git a/src/Functions/FunctionsJSON.cpp b/src/Functions/FunctionsJSON.cpp index 493fc36ca3c..c856419c9e8 100644 --- a/src/Functions/FunctionsJSON.cpp +++ b/src/Functions/FunctionsJSON.cpp @@ -25,7 +25,6 @@ #include #include #include -#include #include #include #include @@ -40,6 +39,7 @@ #include #include +#include #include @@ -191,7 +191,7 @@ private: for (const auto i : collections::range(first_index_argument, first_index_argument + num_index_arguments)) { const auto & column = columns[i]; - if (!isString(column.type) && !isInteger(column.type)) + if (!isString(column.type) && !isNativeInteger(column.type)) throw Exception{"The argument " + std::to_string(i + 1) + " of function " + String(function_name) + " should be a string specifying key or an integer specifying index, illegal type: " + column.type->getName(), ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT}; @@ -623,24 +623,32 @@ public: static bool insertResultToColumn(IColumn & dest, const Element & element, std::string_view) { UInt8 type; - if (element.isInt64()) - type = 'i'; - else if (element.isUInt64()) - type = 'u'; - else if (element.isDouble()) - type = 'd'; - else if (element.isBool()) - type = 'b'; - else if (element.isString()) - type = '"'; - else if (element.isArray()) - type = '['; - else if (element.isObject()) - type = '{'; - else if (element.isNull()) - type = 0; - else - return false; + switch (element.type()) + { + case ElementType::INT64: + type = 'i'; + break; + case ElementType::UINT64: + type = 'u'; + break; + case ElementType::DOUBLE: + type = 'd'; + break; + case ElementType::STRING: + type = '"'; + break; + case ElementType::ARRAY: + type = '['; + break; + case ElementType::OBJECT: + type = '{'; + break; + case ElementType::NULL_VALUE: + type = 0; + break; + default: + return false; + } ColumnVector & col_vec = assert_cast &>(dest); col_vec.insertValue(type); @@ -666,34 +674,51 @@ public: { NumberType value; - if (element.isInt64()) + switch (element.type()) { - if (!accurate::convertNumeric(element.getInt64(), value)) + case ElementType::DOUBLE: + if constexpr (std::is_floating_point_v) + { + /// We permit inaccurate conversion of double to float. + /// Example: double 0.1 from JSON is not representable in float. + /// But it will be more convenient for user to perform conversion. + value = static_cast(element.getDouble()); + } + else if (!accurate::convertNumeric(element.getDouble(), value)) + return false; + break; + case ElementType::UINT64: + if (!accurate::convertNumeric(element.getUInt64(), value)) + return false; + break; + case ElementType::INT64: + if (!accurate::convertNumeric(element.getInt64(), value)) + return false; + break; + case ElementType::BOOL: + if constexpr (is_integer && convert_bool_to_integer) + { + value = static_cast(element.getBool()); + break; + } return false; - } - else if (element.isUInt64()) - { - if (!accurate::convertNumeric(element.getUInt64(), value)) - return false; - } - else if (element.isDouble()) - { - if constexpr (std::is_floating_point_v) - { - /// We permit inaccurate conversion of double to float. - /// Example: double 0.1 from JSON is not representable in float. - /// But it will be more convenient for user to perform conversion. - value = static_cast(element.getDouble()); + case ElementType::STRING: { + auto rb = ReadBufferFromMemory{element.getString()}; + if constexpr (std::is_floating_point_v) + { + if (!tryReadFloatText(value, rb) || !rb.eof()) + return false; + } + else + { + if (!tryReadIntText(value, rb) || !rb.eof()) + return false; + } + break; } - else if (!accurate::convertNumeric(element.getDouble(), value)) + default: return false; } - else if (element.isBool() && is_integer && convert_bool_to_integer) - { - value = static_cast(element.getBool()); - } - else - return false; auto & col_vec = assert_cast &>(dest); col_vec.insertValue(value); @@ -719,9 +744,25 @@ using JSONExtractInt64Impl = JSONExtractNumericImpl; template using JSONExtractUInt64Impl = JSONExtractNumericImpl; template +using JSONExtractInt128Impl = JSONExtractNumericImpl; +template +using JSONExtractUInt128Impl = JSONExtractNumericImpl; +template +using JSONExtractInt256Impl = JSONExtractNumericImpl; +template +using JSONExtractUInt256Impl = JSONExtractNumericImpl; +template using JSONExtractFloat32Impl = JSONExtractNumericImpl; template using JSONExtractFloat64Impl = JSONExtractNumericImpl; +template +using JSONExtractDecimal32Impl = JSONExtractNumericImpl; +template +using JSONExtractDecimal64Impl = JSONExtractNumericImpl; +template +using JSONExtractDecimal128Impl = JSONExtractNumericImpl; +template +using JSONExtractDecimal256Impl = JSONExtractNumericImpl; template @@ -739,11 +780,22 @@ public: static bool insertResultToColumn(IColumn & dest, const Element & element, std::string_view) { - if (!element.isBool()) - return false; + bool value; + switch (element.type()) + { + case ElementType::BOOL: + value = element.getBool(); + break; + case ElementType::INT64: + case ElementType::UINT64: + value = element.getUInt64() != 0; + break; + default: + return false; + } auto & col_vec = assert_cast &>(dest); - col_vec.insertValue(static_cast(element.getBool())); + col_vec.insertValue(static_cast(value)); return true; } }; @@ -845,12 +897,35 @@ struct JSONExtractTree explicit DecimalNode(DataTypePtr data_type_) : data_type(data_type_) {} bool insertResultToColumn(IColumn & dest, const Element & element) override { - if (!element.isDouble()) - return false; - const auto * type = assert_cast *>(data_type.get()); - auto result = convertToDecimal, DataTypeDecimal>(element.getDouble(), type->getScale()); - assert_cast &>(dest).insert(result); + + DecimalType value{}; + + switch (element.type()) + { + case ElementType::DOUBLE: + value = convertToDecimal, DataTypeDecimal>( + element.getDouble(), type->getScale()); + break; + case ElementType::UINT64: + value = convertToDecimal, DataTypeDecimal>( + element.getUInt64(), type->getScale()); + break; + case ElementType::INT64: + value = convertToDecimal, DataTypeDecimal>( + element.getInt64(), type->getScale()); + break; + case ElementType::STRING: { + auto rb = ReadBufferFromMemory{element.getString()}; + if (!SerializationDecimal::tryReadText(value, rb, DecimalUtils::max_precision, type->getScale())) + return false; + break; + } + default: + return false; + } + + assert_cast &>(dest).insert(value); return true; } private: @@ -1088,10 +1163,14 @@ struct JSONExtractTree case TypeIndex::UInt16: return std::make_unique>(); case TypeIndex::UInt32: return std::make_unique>(); case TypeIndex::UInt64: return std::make_unique>(); + case TypeIndex::UInt128: return std::make_unique>(); + case TypeIndex::UInt256: return std::make_unique>(); case TypeIndex::Int8: return std::make_unique>(); case TypeIndex::Int16: return std::make_unique>(); case TypeIndex::Int32: return std::make_unique>(); case TypeIndex::Int64: return std::make_unique>(); + case TypeIndex::Int128: return std::make_unique>(); + case TypeIndex::Int256: return std::make_unique>(); case TypeIndex::Float32: return std::make_unique>(); case TypeIndex::Float64: return std::make_unique>(); case TypeIndex::String: return std::make_unique(); diff --git a/src/Functions/FunctionsLogical.cpp b/src/Functions/FunctionsLogical.cpp index 2ac7688737f..7e52c55e5b0 100644 --- a/src/Functions/FunctionsLogical.cpp +++ b/src/Functions/FunctionsLogical.cpp @@ -168,10 +168,7 @@ public: inline ResultValueType apply(const size_t i) const { const auto a = !!vec[i]; - if constexpr (Op::isSaturable()) - return Op::isSaturatedValue(a) ? a : Op::apply(a, next.apply(i)); - else - return Op::apply(a, next.apply(i)); + return Op::apply(a, next.apply(i)); } private: diff --git a/src/Functions/LowerUpperUTF8Impl.h b/src/Functions/LowerUpperUTF8Impl.h index 3a050e2bd6f..b8fd20d56de 100644 --- a/src/Functions/LowerUpperUTF8Impl.h +++ b/src/Functions/LowerUpperUTF8Impl.h @@ -104,7 +104,7 @@ struct LowerUpperUTF8Impl /** Converts a single code point starting at `src` to desired case, storing result starting at `dst`. * `src` and `dst` are incremented by corresponding sequence lengths. */ - static void toCase(const UInt8 *& src, const UInt8 * src_end, UInt8 *& dst) + static bool toCase(const UInt8 *& src, const UInt8 * src_end, UInt8 *& dst, bool partial) { if (src[0] <= ascii_upper_bound) { @@ -136,6 +136,11 @@ struct LowerUpperUTF8Impl static const Poco::UTF8Encoding utf8; size_t src_sequence_length = UTF8::seqLength(*src); + /// In case partial buffer was passed (due to SSE optimization) + /// we cannot convert it with current src_end, but we may have more + /// bytes to convert and eventually got correct symbol. + if (partial && src_sequence_length > static_cast(src_end-src)) + return false; auto src_code_point = UTF8::convertUTF8ToCodePoint(src, src_end - src); if (src_code_point) @@ -152,7 +157,7 @@ struct LowerUpperUTF8Impl { src += dst_sequence_length; dst += dst_sequence_length; - return; + return true; } } } @@ -161,6 +166,8 @@ struct LowerUpperUTF8Impl ++dst; ++src; } + + return true; } private: @@ -229,16 +236,13 @@ private: const UInt8 * expected_end = std::min(src + bytes_sse, row_end); while (src < expected_end) - toCase(src, expected_end, dst); - - /// adjust src_end_sse by pushing it forward or backward - const auto diff = src - expected_end; - if (diff != 0) { - if (src_end_sse + diff < src_end) - src_end_sse += diff; - else - src_end_sse -= bytes_sse - diff; + if (!toCase(src, expected_end, dst, /* partial= */ true)) + { + /// Fallback to handling byte by byte. + src_end_sse = src; + break; + } } } } @@ -255,7 +259,7 @@ private: chassert(row_end >= src); while (src < row_end) - toCase(src, row_end, dst); + toCase(src, row_end, dst, /* partial= */ false); ++offset_it; } } diff --git a/src/Functions/URL/cutToFirstSignificantSubdomain.cpp b/src/Functions/URL/cutToFirstSignificantSubdomain.cpp index 10c41b6a4c3..7bf09d1eb00 100644 --- a/src/Functions/URL/cutToFirstSignificantSubdomain.cpp +++ b/src/Functions/URL/cutToFirstSignificantSubdomain.cpp @@ -43,10 +43,34 @@ using FunctionCutToFirstSignificantSubdomainWithWWWRFC = FunctionStringToString< REGISTER_FUNCTION(CutToFirstSignificantSubdomain) { - factory.registerFunction(); - factory.registerFunction(); - factory.registerFunction(); - factory.registerFunction(); + factory.registerFunction( + { + R"(Returns the part of the domain that includes top-level subdomains up to the "first significant subdomain" (see documentation of the `firstSignificantSubdomain`).)", + Documentation::Examples{ + {"cutToFirstSignificantSubdomain1", "SELECT cutToFirstSignificantSubdomain('https://news.clickhouse.com.tr/')"}, + {"cutToFirstSignificantSubdomain2", "SELECT cutToFirstSignificantSubdomain('www.tr')"}, + {"cutToFirstSignificantSubdomain3", "SELECT cutToFirstSignificantSubdomain('tr')"}, + }, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Returns the part of the domain that includes top-level subdomains up to the "first significant subdomain", without stripping "www".)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `cutToFirstSignificantSubdomain` but follows stricter rules to be compatible with RFC 3986 and less performant.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `cutToFirstSignificantSubdomainWithWWW` but follows stricter rules to be compatible with RFC 3986 and less performant.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/cutToFirstSignificantSubdomainCustom.cpp b/src/Functions/URL/cutToFirstSignificantSubdomainCustom.cpp index 521216c84a7..e81921d69ff 100644 --- a/src/Functions/URL/cutToFirstSignificantSubdomainCustom.cpp +++ b/src/Functions/URL/cutToFirstSignificantSubdomainCustom.cpp @@ -42,10 +42,41 @@ using FunctionCutToFirstSignificantSubdomainCustomWithWWWRFC = FunctionCutToFirs REGISTER_FUNCTION(CutToFirstSignificantSubdomainCustom) { - factory.registerFunction(); - factory.registerFunction(); - factory.registerFunction(); - factory.registerFunction(); + factory.registerFunction( + { + R"( +Returns the part of the domain that includes top-level subdomains up to the first significant subdomain. Accepts custom TLD list name. + +Can be useful if you need fresh TLD list or you have custom. + )", + Documentation::Examples{ + {"cutToFirstSignificantSubdomainCustom", "SELECT cutToFirstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list');"}, + }, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"( +Returns the part of the domain that includes top-level subdomains up to the first significant subdomain without stripping `www`. +Accepts custom TLD list name from config. + +Can be useful if you need fresh TLD list or you have custom. + )", + Documentation::Examples{{"cutToFirstSignificantSubdomainCustomWithWWW", "SELECT cutToFirstSignificantSubdomainCustomWithWWW('www.foo', 'public_suffix_list')"}}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `cutToFirstSignificantSubdomainCustom` but follows stricter rules according to RFC 3986.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `cutToFirstSignificantSubdomainCustomWithWWW` but follows stricter rules according to RFC 3986.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/domain.cpp b/src/Functions/URL/domain.cpp index e7fead24dc9..fce7cea4693 100644 --- a/src/Functions/URL/domain.cpp +++ b/src/Functions/URL/domain.cpp @@ -14,8 +14,24 @@ using FunctionDomainRFC = FunctionStringToString(); - factory.registerFunction(); + factory.registerFunction( + { + R"( +Extracts the hostname from a URL. + +The URL can be specified with or without a scheme. +If the argument can't be parsed as URL, the function returns an empty string. + )", + Documentation::Examples{{"domain", "SELECT domain('svn+ssh://some.svn-hosting.com:80/repo/trunk')"}}, + Documentation::Categories{"URL"} + }); + + factory.registerFunction( + { + R"(Similar to `domain` but follows stricter rules to be compatible with RFC 3986 and less performant.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/domainWithoutWWW.cpp b/src/Functions/URL/domainWithoutWWW.cpp index 2fa9159d7af..48401e5e6e5 100644 --- a/src/Functions/URL/domainWithoutWWW.cpp +++ b/src/Functions/URL/domainWithoutWWW.cpp @@ -14,8 +14,23 @@ using FunctionDomainWithoutWWWRFC = FunctionStringToString(); - factory.registerFunction(); + factory.registerFunction( + { + R"( +Extracts the hostname from a URL, removing the leading "www." if present. + +The URL can be specified with or without a scheme. +If the argument can't be parsed as URL, the function returns an empty string. + )", + Documentation::Examples{{"domainWithoutWWW", "SELECT domainWithoutWWW('https://www.clickhouse.com')"}}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `domainWithoutWWW` but follows stricter rules to be compatible with RFC 3986 and less performant.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/firstSignificantSubdomain.cpp b/src/Functions/URL/firstSignificantSubdomain.cpp index 902a4f43fba..62307ef816c 100644 --- a/src/Functions/URL/firstSignificantSubdomain.cpp +++ b/src/Functions/URL/firstSignificantSubdomain.cpp @@ -14,8 +14,28 @@ using FunctionFirstSignificantSubdomainRFC = FunctionStringToString(); - factory.registerFunction(); + factory.registerFunction( + { + R"( +Returns the "first significant subdomain". + +The first significant subdomain is a second-level domain if it is 'com', 'net', 'org', or 'co'. +Otherwise, it is a third-level domain. + +For example, firstSignificantSubdomain('https://news.clickhouse.com/') = 'clickhouse', firstSignificantSubdomain ('https://news.clickhouse.com.tr/') = 'clickhouse'. + +The list of "insignificant" second-level domains and other implementation details may change in the future. + )", + Documentation::Examples{{"firstSignificantSubdomain", "SELECT firstSignificantSubdomain('https://news.clickhouse.com/')"}}, + Documentation::Categories{"URL"} + }); + + factory.registerFunction( + { + R"(Returns the "first significant subdomain" according to RFC 1034.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/port.cpp b/src/Functions/URL/port.cpp index f716f3e454b..52fa4077c18 100644 --- a/src/Functions/URL/port.cpp +++ b/src/Functions/URL/port.cpp @@ -139,8 +139,18 @@ struct FunctionPortRFC : public FunctionPortImpl REGISTER_FUNCTION(Port) { - factory.registerFunction(); - factory.registerFunction(); + factory.registerFunction( + { + R"(Returns the port or `default_port` if there is no port in the URL (or in case of validation error).)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); + factory.registerFunction( + { + R"(Similar to `port`, but conforms to RFC 3986.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/URL/topLevelDomain.cpp b/src/Functions/URL/topLevelDomain.cpp index f5610ed93b7..ed9b40d4b73 100644 --- a/src/Functions/URL/topLevelDomain.cpp +++ b/src/Functions/URL/topLevelDomain.cpp @@ -53,8 +53,23 @@ using FunctionTopLevelDomainRFC = FunctionStringToString(); - factory.registerFunction(); + factory.registerFunction( + { + R"( +Extracts the the top-level domain from a URL. + +Returns an empty string if the argument cannot be parsed as a URL or does not contain a top-level domain. + )", + Documentation::Examples{{"topLevelDomain", "SELECT topLevelDomain('svn+ssh://www.some.svn-hosting.com:80/repo/trunk')"}}, + Documentation::Categories{"URL"} + }); + + factory.registerFunction( + { + R"(Similar to topLevelDomain, but conforms to RFC 3986.)", + Documentation::Examples{}, + Documentation::Categories{"URL"} + }); } } diff --git a/src/Functions/base64Decode.cpp b/src/Functions/base64Decode.cpp index f6943233d44..4060aafe1a3 100644 --- a/src/Functions/base64Decode.cpp +++ b/src/Functions/base64Decode.cpp @@ -1,8 +1,7 @@ #include + #if USE_BASE64 #include -#include - namespace DB { @@ -15,4 +14,5 @@ REGISTER_FUNCTION(Base64Decode) factory.registerAlias("FROM_BASE64", "base64Decode", FunctionFactory::CaseInsensitive); } } + #endif diff --git a/src/Functions/base64Encode.cpp b/src/Functions/base64Encode.cpp index fc06935e0a1..773db7e09d9 100644 --- a/src/Functions/base64Encode.cpp +++ b/src/Functions/base64Encode.cpp @@ -1,10 +1,7 @@ -#include #include -#include "config.h" - #if USE_BASE64 -# include +#include namespace DB { @@ -17,4 +14,5 @@ REGISTER_FUNCTION(Base64Encode) factory.registerAlias("TO_BASE64", "base64Encode", FunctionFactory::CaseInsensitive); } } + #endif diff --git a/src/Functions/blockSerializedSize.cpp b/src/Functions/blockSerializedSize.cpp index d406984c51c..35be65f3fed 100644 --- a/src/Functions/blockSerializedSize.cpp +++ b/src/Functions/blockSerializedSize.cpp @@ -54,7 +54,7 @@ public: auto serialization = elem.type->getDefaultSerialization(); - serialization->serializeBinaryBulkStatePrefix(settings, state); + serialization->serializeBinaryBulkStatePrefix(*full_column, settings, state); serialization->serializeBinaryBulkWithMultipleStreams(*full_column, 0 /** offset */, 0 /** limit */, settings, state); diff --git a/src/Functions/formatDateTime.cpp b/src/Functions/formatDateTime.cpp index a10c059b342..4db04d61d84 100644 --- a/src/Functions/formatDateTime.cpp +++ b/src/Functions/formatDateTime.cpp @@ -1,5 +1,6 @@ #include #include +#include #include #include #include @@ -45,6 +46,7 @@ template <> struct ActionValueTypeMap { using ActionValueTyp template <> struct ActionValueTypeMap { using ActionValueType = UInt32; }; template <> struct ActionValueTypeMap { using ActionValueType = UInt32; }; template <> struct ActionValueTypeMap { using ActionValueType = UInt16; }; +template <> struct ActionValueTypeMap { using ActionValueType = Int32; }; template <> struct ActionValueTypeMap { using ActionValueType = UInt32; }; // TODO(vnemkov): to add sub-second format instruction, make that DateTime64 and do some math in Action. template <> struct ActionValueTypeMap { using ActionValueType = Int64; }; @@ -315,44 +317,39 @@ public: if constexpr (support_integer) { if (arguments.size() != 1 && arguments.size() != 2 && arguments.size() != 3) - throw Exception( - "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) - + ", should be 1, 2 or 3", - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); + throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, + "Number of arguments for function {} doesn't match: passed {}, should be 1, 2 or 3", + getName(), arguments.size()); if (arguments.size() == 1 && !isInteger(arguments[0].type)) - throw Exception( - "Illegal type " + arguments[0].type->getName() + " of 1 argument of function " + getName() - + " when arguments size is 1. Should be integer", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); - if (arguments.size() > 1 && !(isInteger(arguments[0].type) || isDate(arguments[0].type) || isDateTime(arguments[0].type) || isDateTime64(arguments[0].type))) - throw Exception( - "Illegal type " + arguments[0].type->getName() + " of 1 argument of function " + getName() - + " when arguments size is 2 or 3. Should be a integer or a date with time", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of first argument of function {} when arguments size is 1. Should be integer", + arguments[0].type->getName(), getName()); + if (arguments.size() > 1 && !(isInteger(arguments[0].type) || isDate(arguments[0].type) || isDateTime(arguments[0].type) || isDate32(arguments[0].type) || isDateTime64(arguments[0].type))) + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of first argument of function {} when arguments size is 2 or 3. Should be a integer or a date with time", + arguments[0].type->getName(), getName()); } else { if (arguments.size() != 2 && arguments.size() != 3) - throw Exception( - "Number of arguments for function " + getName() + " doesn't match: passed " + toString(arguments.size()) - + ", should be 2 or 3", - ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH); - if (!isDate(arguments[0].type) && !isDateTime(arguments[0].type) && !isDateTime64(arguments[0].type)) - throw Exception( - "Illegal type " + arguments[0].type->getName() + " of 1 argument of function " + getName() - + ". Should be a date or a date with time", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH, + "Number of arguments for function {} doesn't match: passed {}, should be 2 or 3", + getName(), arguments.size()); + if (!isDate(arguments[0].type) && !isDateTime(arguments[0].type) && !isDate32(arguments[0].type) && !isDateTime64(arguments[0].type)) + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of first argument of function {}. Should be a date or a date with time", + arguments[0].type->getName(), getName()); } if (arguments.size() == 2 && !WhichDataType(arguments[1].type).isString()) - throw Exception( - "Illegal type " + arguments[1].type->getName() + " of 2 argument of function " + getName() + ". Must be String.", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of second argument of function {}. Must be String.", + arguments[1].type->getName(), getName()); if (arguments.size() == 3 && !WhichDataType(arguments[2].type).isString()) - throw Exception( - "Illegal type " + arguments[2].type->getName() + " of 3 argument of function " + getName() + ". Must be String.", - ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); + throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT, + "Illegal type {} of third argument of function {}. Must be String.", + arguments[2].type->getName(), getName()); if (arguments.size() == 1) return std::make_shared(); @@ -373,10 +370,9 @@ public: return true; })) { - throw Exception( - "Illegal column " + arguments[0].column->getName() + " of function " + getName() - + ", must be Integer or DateTime when arguments size is 1.", - ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of function {}, must be Integer, Date, Date32, DateTime or DateTime64 when arguments size is 1.", + arguments[0].column->getName(), getName()); } } else @@ -385,32 +381,31 @@ public: { using FromDataType = std::decay_t; if (!(res = executeType(arguments, result_type))) - throw Exception( - "Illegal column " + arguments[0].column->getName() + " of function " + getName() - + ", must be Integer or DateTime.", - ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of function {}, must be Integer, Date, Date32, DateTime or DateTime64.", + arguments[0].column->getName(), getName()); return true; })) { if (!((res = executeType(arguments, result_type)) + || (res = executeType(arguments, result_type)) || (res = executeType(arguments, result_type)) || (res = executeType(arguments, result_type)))) - throw Exception( - "Illegal column " + arguments[0].column->getName() + " of function " + getName() - + ", must be Integer or DateTime.", - ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of function {}, must be Integer or DateTime.", + arguments[0].column->getName(), getName()); } } } else { if (!((res = executeType(arguments, result_type)) + || (res = executeType(arguments, result_type)) || (res = executeType(arguments, result_type)) || (res = executeType(arguments, result_type)))) - throw Exception( - "Illegal column " + arguments[0].column->getName() + " of function " + getName() - + ", must be Date or DateTime.", - ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of function {}, must be Date or DateTime.", + arguments[0].column->getName(), getName()); } return res; @@ -425,10 +420,9 @@ public: const ColumnConst * pattern_column = checkAndGetColumnConst(arguments[1].column.get()); if (!pattern_column) - throw Exception("Illegal column " + arguments[1].column->getName() - + " of second ('format') argument of function " + getName() - + ". Must be constant string.", - ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Illegal column {} of second ('format') argument of function {}. Must be constant string.", + arguments[1].column->getName(), getName()); String pattern = pattern_column->getValue(); @@ -712,12 +706,14 @@ public: // Unimplemented case 'U': [[fallthrough]]; case 'W': - throw Exception("Wrong pattern '" + pattern + "', symbol '" + *pos + " is not implemented ' for function " + getName(), - ErrorCodes::NOT_IMPLEMENTED); + throw Exception(ErrorCodes::NOT_IMPLEMENTED, + "Wrong pattern '{}', symbol '{}' is not implemented for function {}", + pattern, *pos, getName()); default: - throw Exception( - "Wrong pattern '" + pattern + "', unexpected symbol '" + *pos + "' for function " + getName(), ErrorCodes::ILLEGAL_COLUMN); + throw Exception(ErrorCodes::ILLEGAL_COLUMN, + "Wrong pattern '{}', unexpected symbol '{}' for function {}", + pattern, *pos, getName()); } ++pos; diff --git a/src/Functions/tryBase64Decode.cpp b/src/Functions/tryBase64Decode.cpp index 1102c7a3418..bd452b8357b 100644 --- a/src/Functions/tryBase64Decode.cpp +++ b/src/Functions/tryBase64Decode.cpp @@ -1,7 +1,7 @@ #include + #if USE_BASE64 #include -#include namespace DB { @@ -10,4 +10,5 @@ REGISTER_FUNCTION(TryBase64Decode) factory.registerFunction>(); } } + #endif diff --git a/src/IO/ReadBufferFromMemory.h b/src/IO/ReadBufferFromMemory.h index dc5c464604b..ad96e4bfa28 100644 --- a/src/IO/ReadBufferFromMemory.h +++ b/src/IO/ReadBufferFromMemory.h @@ -16,6 +16,8 @@ public: requires (sizeof(CharT) == 1) ReadBufferFromMemory(const CharT * buf, size_t size) : SeekableReadBuffer(const_cast(reinterpret_cast(buf)), size, 0) {} + explicit ReadBufferFromMemory(const std::string_view&& str) + : SeekableReadBuffer(const_cast(str.data()), str.size(), 0) {} off_t seek(off_t off, int whence) override; diff --git a/src/IO/readDecimalText.h b/src/IO/readDecimalText.h index 64374a20574..9d7f8137136 100644 --- a/src/IO/readDecimalText.h +++ b/src/IO/readDecimalText.h @@ -147,23 +147,32 @@ inline bool readDigits(ReadBuffer & buf, T & x, uint32_t & digits, int32_t & exp return true; } -template -inline void readDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale, bool digits_only = false) +template +inline ReturnType readDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale, bool digits_only = false) { + static constexpr bool throw_exception = std::is_same_v; + uint32_t digits = precision; int32_t exponent; - readDigits(buf, x, digits, exponent, digits_only); + auto ok = readDigits(buf, x, digits, exponent, digits_only); + + if (!throw_exception && !ok) + return ReturnType(false); if (static_cast(digits) + exponent > static_cast(precision - scale)) { - static constexpr const char * pattern = - "Decimal value is too big: {} digits were read: {}e{}." - " Expected to read decimal with scale {} and precision {}"; + if constexpr (throw_exception) + { + static constexpr const char * pattern = "Decimal value is too big: {} digits were read: {}e{}." + " Expected to read decimal with scale {} and precision {}"; - if constexpr (is_big_int_v) - throw Exception(fmt::format(pattern, digits, x.value, exponent, scale, precision), ErrorCodes::ARGUMENT_OUT_OF_BOUND); + if constexpr (is_big_int_v) + throw Exception(fmt::format(pattern, digits, x.value, exponent, scale, precision), ErrorCodes::ARGUMENT_OUT_OF_BOUND); + else + throw Exception(fmt::format(pattern, digits, x, exponent, scale, precision), ErrorCodes::ARGUMENT_OUT_OF_BOUND); + } else - throw Exception(fmt::format(pattern, digits, x, exponent, scale, precision), ErrorCodes::ARGUMENT_OUT_OF_BOUND); + return ReturnType(false); } if (static_cast(scale) + exponent < 0) @@ -175,7 +184,7 @@ inline void readDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_ /// Too big negative exponent x.value = 0; scale = 0; - return; + return ReturnType(true); } else { @@ -184,26 +193,18 @@ inline void readDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_ assert(divisor > 0); /// This is for Clang Static Analyzer. It is not smart enough to infer it automatically. x.value /= divisor; scale = 0; - return; + return ReturnType(true); } } scale += exponent; + return ReturnType(true); } template inline bool tryReadDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale) { - uint32_t digits = precision; - int32_t exponent; - - if (!readDigits(buf, x, digits, exponent, true) || - static_cast(digits) + exponent > static_cast(precision - scale) || - static_cast(scale) + exponent < 0) - return false; - - scale += exponent; - return true; + return readDecimalText(buf, x, precision, scale, true); } template diff --git a/src/Interpreters/AsynchronousMetrics.cpp b/src/Interpreters/AsynchronousMetrics.cpp index 338ae1bbbfd..488ac77e956 100644 --- a/src/Interpreters/AsynchronousMetrics.cpp +++ b/src/Interpreters/AsynchronousMetrics.cpp @@ -703,19 +703,26 @@ void AsynchronousMetrics::update(TimePoint update_time) Int64 free_memory_in_allocator_arenas = 0; #if USE_JEMALLOC - /// This is a memory which is kept by allocator. - /// Will subsract it from RSS to decrease memory drift. + /// According to jemalloc man, pdirty is: + /// + /// Number of pages within unused extents that are potentially + /// dirty, and for which madvise() or similar has not been called. + /// + /// So they will be subtracted from RSS to make accounting more + /// accurate, since those pages are not really RSS but a memory + /// that can be used at anytime via jemalloc. free_memory_in_allocator_arenas = je_malloc_pdirty * getPageSize(); #endif - Int64 difference = rss - free_memory_in_allocator_arenas - amount; + Int64 difference = rss - amount; /// Log only if difference is high. This is for convenience. The threshold is arbitrary. if (difference >= 1048576 || difference <= -1048576) LOG_TRACE(log, - "MemoryTracking: was {}, peak {}, will set to {} (RSS), difference: {}", + "MemoryTracking: was {}, peak {}, free memory in arenas {}, will set to {} (RSS), difference: {}", ReadableSize(amount), ReadableSize(peak), + ReadableSize(free_memory_in_allocator_arenas), ReadableSize(rss), ReadableSize(difference)); diff --git a/src/Interpreters/HashJoin.cpp b/src/Interpreters/HashJoin.cpp index 26b9b843567..41c7c28a6fa 100644 --- a/src/Interpreters/HashJoin.cpp +++ b/src/Interpreters/HashJoin.cpp @@ -658,7 +658,9 @@ void HashJoin::initRightBlockStructure(Block & saved_block_sample) /// Save non key columns for (auto & column : sample_block_with_columns_to_add) { - if (!saved_block_sample.findByName(column.name)) + if (auto * col = saved_block_sample.findByName(column.name)) + *col = column; + else saved_block_sample.insert(column); } } diff --git a/src/Interpreters/InterpreterCreateQuery.cpp b/src/Interpreters/InterpreterCreateQuery.cpp index f41ae31b239..4a6ce63eb84 100644 --- a/src/Interpreters/InterpreterCreateQuery.cpp +++ b/src/Interpreters/InterpreterCreateQuery.cpp @@ -829,7 +829,7 @@ void InterpreterCreateQuery::validateTableStructure(const ASTCreateQuery & creat { for (const auto & [name, type] : properties.columns.getAllPhysical()) { - if (isObject(type)) + if (type->hasDynamicSubcolumns()) { throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot create table with column '{}' which type is '{}' " @@ -1398,7 +1398,7 @@ bool InterpreterCreateQuery::doCreateTable(ASTCreateQuery & create, /// we can safely destroy the object without a call to "shutdown", because there is guarantee /// that no background threads/similar resources remain after exception from "startup". - if (!res->supportsDynamicSubcolumns() && hasObjectColumns(res->getInMemoryMetadataPtr()->getColumns())) + if (!res->supportsDynamicSubcolumns() && hasDynamicSubcolumns(res->getInMemoryMetadataPtr()->getColumns())) { throw Exception(ErrorCodes::ILLEGAL_COLUMN, "Cannot create table with column of type Object, " diff --git a/src/Interpreters/InterpreterDropQuery.cpp b/src/Interpreters/InterpreterDropQuery.cpp index 8be3dce7bf1..28f8e43ee9b 100644 --- a/src/Interpreters/InterpreterDropQuery.cpp +++ b/src/Interpreters/InterpreterDropQuery.cpp @@ -213,7 +213,7 @@ BlockIO InterpreterDropQuery::executeToTableImpl(ContextPtr context_, ASTDropQue { /// And for simple MergeTree we can stop merges before acquiring the lock auto merges_blocker = table->getActionLock(ActionLocks::PartsMerge); - auto table_lock = table->lockExclusively(context_->getCurrentQueryId(), context_->getSettingsRef().lock_acquire_timeout); + table_lock = table->lockExclusively(context_->getCurrentQueryId(), context_->getSettingsRef().lock_acquire_timeout); } auto metadata_snapshot = table->getInMemoryMetadataPtr(); diff --git a/src/Interpreters/PartLog.cpp b/src/Interpreters/PartLog.cpp index 75e6d02d6e1..b35ee50b98e 100644 --- a/src/Interpreters/PartLog.cpp +++ b/src/Interpreters/PartLog.cpp @@ -207,8 +207,8 @@ bool PartLog::addNewParts( elem.table_name = table_id.table_name; elem.partition_id = part->info.partition_id; elem.part_name = part->name; - elem.disk_name = part->data_part_storage->getDiskName(); - elem.path_on_disk = part->data_part_storage->getFullPath(); + elem.disk_name = part->getDataPartStorage().getDiskName(); + elem.path_on_disk = part->getDataPartStorage().getFullPath(); elem.part_type = part->getType(); elem.bytes_compressed_on_disk = part->getBytesOnDisk(); diff --git a/src/Interpreters/convertFieldToType.cpp b/src/Interpreters/convertFieldToType.cpp index 9ba171d2665..e57016d969a 100644 --- a/src/Interpreters/convertFieldToType.cpp +++ b/src/Interpreters/convertFieldToType.cpp @@ -236,10 +236,11 @@ Field convertFieldToTypeImpl(const Field & src, const IDataType & type, const ID } if (which_type.isDateTime64() - && (which_from_type.isNativeInt() || which_from_type.isNativeUInt() || which_from_type.isDate() || which_from_type.isDate32() || which_from_type.isDateTime() || which_from_type.isDateTime64())) + && (src.getType() == Field::Types::UInt64 || src.getType() == Field::Types::Int64 || src.getType() == Field::Types::Decimal64)) { const auto scale = static_cast(type).getScale(); - const auto decimal_value = DecimalUtils::decimalFromComponents(applyVisitor(FieldVisitorConvertToNumber(), src), 0, scale); + const auto decimal_value + = DecimalUtils::decimalFromComponents(applyVisitor(FieldVisitorConvertToNumber(), src), 0, scale); return Field(DecimalField(decimal_value, scale)); } } @@ -386,6 +387,9 @@ Field convertFieldToTypeImpl(const Field & src, const IDataType & type, const ID } else if (isObject(type)) { + if (src.getType() == Field::Types::Object) + return src; /// Already in needed type. + const auto * from_type_tuple = typeid_cast(from_type_hint); if (src.getType() == Field::Types::Tuple && from_type_tuple && from_type_tuple->haveExplicitNames()) { diff --git a/src/Parsers/Access/ParserCreateUserQuery.cpp b/src/Parsers/Access/ParserCreateUserQuery.cpp index 9e32b3c4618..ed6ecb62667 100644 --- a/src/Parsers/Access/ParserCreateUserQuery.cpp +++ b/src/Parsers/Access/ParserCreateUserQuery.cpp @@ -295,11 +295,11 @@ namespace } - bool parseHosts(IParserBase::Pos & pos, Expected & expected, const String & prefix, AllowedClientHosts & hosts) + bool parseHosts(IParserBase::Pos & pos, Expected & expected, std::string_view prefix, AllowedClientHosts & hosts) { return IParserBase::wrapParseImpl(pos, [&] { - if (!prefix.empty() && !ParserKeyword{prefix.c_str()}.ignore(pos, expected)) + if (!prefix.empty() && !ParserKeyword{prefix}.ignore(pos, expected)) return false; if (!ParserKeyword{"HOST"}.ignore(pos, expected)) @@ -492,7 +492,6 @@ bool ParserCreateUserQuery::parseImpl(Pos & pos, ASTPtr & node, Expected & expec if (alter) { - String maybe_new_name; if (!new_name && (names->size() == 1) && parseRenameTo(pos, expected, new_name)) continue; diff --git a/src/Parsers/ExpressionListParsers.cpp b/src/Parsers/ExpressionListParsers.cpp index 4af4dabb12e..c362340d013 100644 --- a/src/Parsers/ExpressionListParsers.cpp +++ b/src/Parsers/ExpressionListParsers.cpp @@ -561,13 +561,10 @@ public: virtual bool getResult(ASTPtr & node) { - if (elements.size() == 1) - { - node = std::move(elements[0]); - return true; - } + if (!finished) + return false; - return false; + return getResultImpl(node); } virtual bool parse(IParser::Pos & /*pos*/, Expected & /*expected*/, Action & /*action*/) = 0; @@ -746,6 +743,17 @@ public: Checkpoint current_checkpoint = Checkpoint::None; protected: + virtual bool getResultImpl(ASTPtr & node) + { + if (elements.size() == 1) + { + node = std::move(elements[0]); + return true; + } + + return false; + } + std::vector operators; ASTs operands; ASTs elements; @@ -766,17 +774,12 @@ public: bool getResult(ASTPtr & node) override { /// We can exit the main cycle outside the parse() function, - /// so we need to merge the element here + /// so we need to merge the element here. + /// Because of this 'finished' flag can also not be set. if (!mergeElement()) return false; - if (elements.size() == 1) - { - node = std::move(elements[0]); - return true; - } - - return false; + return Layer::getResultImpl(node); } bool parse(IParser::Pos & pos, Expected & /*expected*/, Action & /*action*/) override @@ -1029,17 +1032,6 @@ private: class RoundBracketsLayer : public Layer { public: - bool getResult(ASTPtr & node) override - { - // Round brackets can mean priority operator as well as function tuple() - if (!is_tuple && elements.size() == 1) - node = std::move(elements[0]); - else - node = makeASTFunction("tuple", std::move(elements)); - - return true; - } - bool parse(IParser::Pos & pos, Expected & expected, Action & action) override { if (ParserToken(TokenType::Comma).ignore(pos, expected)) @@ -1069,6 +1061,19 @@ public: return true; } + +protected: + bool getResultImpl(ASTPtr & node) override + { + // Round brackets can mean priority operator as well as function tuple() + if (!is_tuple && elements.size() == 1) + node = std::move(elements[0]); + else + node = makeASTFunction("tuple", std::move(elements)); + + return true; + } + private: bool is_tuple = false; }; @@ -1077,16 +1082,17 @@ private: class ArrayLayer : public LayerWithSeparator { public: - bool getResult(ASTPtr & node) override - { - node = makeASTFunction("array", std::move(elements)); - return true; - } - bool parse(IParser::Pos & pos, Expected & expected, Action & action) override { return LayerWithSeparator::parse(pos, expected, action); } + +protected: + bool getResultImpl(ASTPtr & node) override + { + node = makeASTFunction("array", std::move(elements)); + return true; + } }; /// Layer for arrayElement square brackets operator @@ -1206,23 +1212,6 @@ class ExtractLayer : public LayerWithSeparator(interval_kind.toDateDiffUnit()), elements[0], elements[1]); - else if (elements.size() == 3) - node = makeASTFunction("dateDiff", std::make_shared(interval_kind.toDateDiffUnit()), elements[0], elements[1], elements[2]); - else - return false; - } - else - { - node = makeASTFunction("dateDiff", std::move(elements)); - } - return true; - } - bool parse(IParser::Pos & pos, Expected & expected, Action & action) override { /// 0. Try to parse interval_kind (-> 1) @@ -1699,6 +1693,25 @@ public: return true; } +protected: + bool getResultImpl(ASTPtr & node) override + { + if (parsed_interval_kind) + { + if (elements.size() == 2) + node = makeASTFunction("dateDiff", std::make_shared(interval_kind.toDateDiffUnit()), elements[0], elements[1]); + else if (elements.size() == 3) + node = makeASTFunction("dateDiff", std::make_shared(interval_kind.toDateDiffUnit()), elements[0], elements[1], elements[2]); + else + return false; + } + else + { + node = makeASTFunction("dateDiff", std::move(elements)); + } + return true; + } + private: IntervalKind interval_kind; bool parsed_interval_kind = false; @@ -1882,16 +1895,6 @@ class ViewLayer : public Layer public: explicit ViewLayer(bool if_permitted_) : if_permitted(if_permitted_) {} - bool getResult(ASTPtr & node) override - { - if (if_permitted) - node = makeASTFunction("viewIfPermitted", std::move(elements)); - else - node = makeASTFunction("view", std::move(elements)); - - return true; - } - bool parse(IParser::Pos & pos, Expected & expected, Action & /*action*/) override { /// view(SELECT ...) @@ -1948,6 +1951,17 @@ public: return true; } +protected: + bool getResultImpl(ASTPtr & node) override + { + if (if_permitted) + node = makeASTFunction("viewIfPermitted", std::move(elements)); + else + node = makeASTFunction("view", std::move(elements)); + + return true; + } + private: bool if_permitted; }; diff --git a/src/Processors/Executors/ExecutingGraph.cpp b/src/Processors/Executors/ExecutingGraph.cpp index 9d69abc5e87..4ab2c5b3802 100644 --- a/src/Processors/Executors/ExecutingGraph.cpp +++ b/src/Processors/Executors/ExecutingGraph.cpp @@ -71,7 +71,7 @@ bool ExecutingGraph::addEdges(uint64_t node) } } - /// Add direct edges form output ports. + /// Add direct edges from output ports. auto & outputs = from->getOutputs(); auto from_output = nodes[node]->direct_edges.size(); diff --git a/src/Processors/Formats/IRowInputFormat.cpp b/src/Processors/Formats/IRowInputFormat.cpp index abd91ddcf35..6f153019df5 100644 --- a/src/Processors/Formats/IRowInputFormat.cpp +++ b/src/Processors/Formats/IRowInputFormat.cpp @@ -232,7 +232,9 @@ Chunk IRowInputFormat::generate() return {}; } - finalizeObjectColumns(columns); + for (const auto & column : columns) + column->finalize(); + Chunk chunk(std::move(columns), num_rows); return chunk; } diff --git a/src/Processors/Formats/Impl/LineAsStringRowInputFormat.cpp b/src/Processors/Formats/Impl/LineAsStringRowInputFormat.cpp index 30084804d92..677f8bb28ec 100644 --- a/src/Processors/Formats/Impl/LineAsStringRowInputFormat.cpp +++ b/src/Processors/Formats/Impl/LineAsStringRowInputFormat.cpp @@ -1,5 +1,5 @@ #include -#include +#include #include #include #include @@ -63,6 +63,12 @@ void registerInputFormatLineAsString(FormatFactory & factory) }); } +void registerFileSegmentationEngineLineAsString(FormatFactory & factory) +{ + factory.registerFileSegmentationEngine("LineAsString", &newLineFileSegmentationEngine); +} + + void registerLineAsStringSchemaReader(FormatFactory & factory) { factory.registerExternalSchemaReader("LineAsString", []( diff --git a/src/Processors/Formats/Impl/RegexpRowInputFormat.cpp b/src/Processors/Formats/Impl/RegexpRowInputFormat.cpp index 6eacfe621e1..2ad2ad6f7a3 100644 --- a/src/Processors/Formats/Impl/RegexpRowInputFormat.cpp +++ b/src/Processors/Formats/Impl/RegexpRowInputFormat.cpp @@ -3,6 +3,7 @@ #include #include #include +#include #include namespace DB @@ -178,46 +179,9 @@ void registerInputFormatRegexp(FormatFactory & factory) }); } -static std::pair fileSegmentationEngineRegexpImpl(ReadBuffer & in, DB::Memory<> & memory, size_t min_bytes, size_t max_rows) -{ - char * pos = in.position(); - bool need_more_data = true; - size_t number_of_rows = 0; - - while (loadAtPosition(in, memory, pos) && need_more_data) - { - pos = find_first_symbols<'\r', '\n'>(pos, in.buffer().end()); - if (pos > in.buffer().end()) - throw Exception("Position in buffer is out of bounds. There must be a bug.", ErrorCodes::LOGICAL_ERROR); - else if (pos == in.buffer().end()) - continue; - - ++number_of_rows; - if ((memory.size() + static_cast(pos - in.position()) >= min_bytes) || (number_of_rows == max_rows)) - need_more_data = false; - - if (*pos == '\n') - { - ++pos; - if (loadAtPosition(in, memory, pos) && *pos == '\r') - ++pos; - } - else if (*pos == '\r') - { - ++pos; - if (loadAtPosition(in, memory, pos) && *pos == '\n') - ++pos; - } - } - - saveUpToPosition(in, memory, pos); - - return {loadAtPosition(in, memory, pos), number_of_rows}; -} - void registerFileSegmentationEngineRegexp(FormatFactory & factory) { - factory.registerFileSegmentationEngine("Regexp", &fileSegmentationEngineRegexpImpl); + factory.registerFileSegmentationEngine("Regexp", &newLineFileSegmentationEngine); } void registerRegexpSchemaReader(FormatFactory & factory) diff --git a/src/Processors/Formats/Impl/ValuesBlockInputFormat.cpp b/src/Processors/Formats/Impl/ValuesBlockInputFormat.cpp index 72e89ea0013..108b4203e3e 100644 --- a/src/Processors/Formats/Impl/ValuesBlockInputFormat.cpp +++ b/src/Processors/Formats/Impl/ValuesBlockInputFormat.cpp @@ -101,7 +101,9 @@ Chunk ValuesBlockInputFormat::generate() return {}; } - finalizeObjectColumns(columns); + for (const auto & column : columns) + column->finalize(); + size_t rows_in_block = columns[0]->size(); return Chunk{std::move(columns), rows_in_block}; } diff --git a/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.cpp b/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.cpp index ebc1b37074b..db08f3ffbd3 100644 --- a/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.cpp +++ b/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.cpp @@ -1,7 +1,5 @@ #include -#include -#include #include #include #include @@ -18,70 +16,6 @@ AggregatingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition() = default; AggregatingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition(ColumnsDefinition &&) noexcept = default; AggregatingSortedAlgorithm::ColumnsDefinition::~ColumnsDefinition() = default; -/// Stores information for aggregation of AggregateFunction columns -struct AggregatingSortedAlgorithm::AggregateDescription -{ - ColumnAggregateFunction * column = nullptr; - const size_t column_number = 0; /// Position in header. - - AggregateDescription() = default; - explicit AggregateDescription(size_t col_number) : column_number(col_number) {} -}; - -/// Stores information for aggregation of SimpleAggregateFunction columns -struct AggregatingSortedAlgorithm::SimpleAggregateDescription -{ - /// An aggregate function 'anyLast', 'sum'... - AggregateFunctionPtr function; - IAggregateFunction::AddFunc add_function = nullptr; - - size_t column_number = 0; - IColumn * column = nullptr; - - /// For LowCardinality, convert is converted to nested type. nested_type is nullptr if no conversion needed. - const DataTypePtr nested_type; /// Nested type for LowCardinality, if it is. - const DataTypePtr real_type; /// Type in header. - - AlignedBuffer state; - bool created = false; - - SimpleAggregateDescription( - AggregateFunctionPtr function_, const size_t column_number_, - DataTypePtr nested_type_, DataTypePtr real_type_) - : function(std::move(function_)), column_number(column_number_) - , nested_type(std::move(nested_type_)), real_type(std::move(real_type_)) - { - add_function = function->getAddressOfAddFunction(); - state.reset(function->sizeOfData(), function->alignOfData()); - } - - void createState() - { - if (created) - return; - function->create(state.data()); - created = true; - } - - void destroyState() - { - if (!created) - return; - function->destroy(state.data()); - created = false; - } - - /// Explicitly destroy aggregation state if the stream is terminated - ~SimpleAggregateDescription() - { - destroyState(); - } - - SimpleAggregateDescription() = default; - SimpleAggregateDescription(SimpleAggregateDescription &&) = default; - SimpleAggregateDescription(const SimpleAggregateDescription &) = delete; -}; - static AggregatingSortedAlgorithm::ColumnsDefinition defineColumns( const Block & header, const SortDescription & description) { @@ -191,6 +125,39 @@ static void postprocessChunk(Chunk & chunk, const AggregatingSortedAlgorithm::Co } +AggregatingSortedAlgorithm::SimpleAggregateDescription::SimpleAggregateDescription( + AggregateFunctionPtr function_, const size_t column_number_, + DataTypePtr nested_type_, DataTypePtr real_type_) + : function(std::move(function_)), column_number(column_number_) + , nested_type(std::move(nested_type_)), real_type(std::move(real_type_)) +{ + add_function = function->getAddressOfAddFunction(); + state.reset(function->sizeOfData(), function->alignOfData()); +} + +void AggregatingSortedAlgorithm::SimpleAggregateDescription::createState() +{ + if (created) + return; + function->create(state.data()); + created = true; +} + +void AggregatingSortedAlgorithm::SimpleAggregateDescription::destroyState() +{ + if (!created) + return; + function->destroy(state.data()); + created = false; +} + +/// Explicitly destroy aggregation state if the stream is terminated +AggregatingSortedAlgorithm::SimpleAggregateDescription::~SimpleAggregateDescription() +{ + destroyState(); +} + + AggregatingSortedAlgorithm::AggregatingMergedData::AggregatingMergedData( MutableColumns columns_, UInt64 max_block_size_, ColumnsDefinition & def_) : MergedData(std::move(columns_), false, max_block_size_), def(def_) diff --git a/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.h b/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.h index e572ed7d526..d670242ed81 100644 --- a/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.h +++ b/src/Processors/Merges/Algorithms/AggregatingSortedAlgorithm.h @@ -1,5 +1,7 @@ #pragma once +#include +#include #include #include @@ -23,8 +25,48 @@ public: void consume(Input & input, size_t source_num) override; Status merge() override; - struct SimpleAggregateDescription; - struct AggregateDescription; + /// Stores information for aggregation of SimpleAggregateFunction columns + struct SimpleAggregateDescription + { + /// An aggregate function 'anyLast', 'sum'... + AggregateFunctionPtr function; + IAggregateFunction::AddFunc add_function = nullptr; + + size_t column_number = 0; + IColumn * column = nullptr; + + /// For LowCardinality, convert is converted to nested type. nested_type is nullptr if no conversion needed. + const DataTypePtr nested_type; /// Nested type for LowCardinality, if it is. + const DataTypePtr real_type; /// Type in header. + + AlignedBuffer state; + bool created = false; + + SimpleAggregateDescription( + AggregateFunctionPtr function_, const size_t column_number_, + DataTypePtr nested_type_, DataTypePtr real_type_); + + void createState(); + + void destroyState(); + + /// Explicitly destroy aggregation state if the stream is terminated + ~SimpleAggregateDescription(); + + SimpleAggregateDescription() = default; + SimpleAggregateDescription(SimpleAggregateDescription &&) = default; + SimpleAggregateDescription(const SimpleAggregateDescription &) = delete; + }; + + /// Stores information for aggregation of AggregateFunction columns + struct AggregateDescription + { + ColumnAggregateFunction * column = nullptr; + const size_t column_number = 0; /// Position in header. + + AggregateDescription() = default; + explicit AggregateDescription(size_t col_number) : column_number(col_number) {} + }; /// This structure define columns into one of three types: /// * columns which are not aggregate functions and not needed to be aggregated diff --git a/src/Processors/Merges/Algorithms/SummingSortedAlgorithm.cpp b/src/Processors/Merges/Algorithms/SummingSortedAlgorithm.cpp index 8636813132d..c79c667a988 100644 --- a/src/Processors/Merges/Algorithms/SummingSortedAlgorithm.cpp +++ b/src/Processors/Merges/Algorithms/SummingSortedAlgorithm.cpp @@ -23,10 +23,6 @@ namespace ErrorCodes extern const int CORRUPTED_DATA; } -SummingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition() = default; -SummingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition(ColumnsDefinition &&) noexcept = default; -SummingSortedAlgorithm::ColumnsDefinition::~ColumnsDefinition() = default; - /// Stores numbers of key-columns and value-columns. struct SummingSortedAlgorithm::MapDescription { @@ -777,4 +773,8 @@ IMergingAlgorithm::Status SummingSortedAlgorithm::merge() return Status(merged_data.pull(), true); } +SummingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition() = default; +SummingSortedAlgorithm::ColumnsDefinition::ColumnsDefinition(ColumnsDefinition &&) noexcept = default; +SummingSortedAlgorithm::ColumnsDefinition::~ColumnsDefinition() = default; + } diff --git a/src/Processors/QueryPlan/Optimizations/optimizePrimaryKeyCondition.cpp b/src/Processors/QueryPlan/Optimizations/optimizePrimaryKeyCondition.cpp index 7d682c408e5..984c76701ba 100644 --- a/src/Processors/QueryPlan/Optimizations/optimizePrimaryKeyCondition.cpp +++ b/src/Processors/QueryPlan/Optimizations/optimizePrimaryKeyCondition.cpp @@ -17,7 +17,7 @@ void optimizePrimaryKeyCondition(QueryPlan::Node & root) size_t next_child = 0; }; - std::deque stack; + std::vector stack; stack.push_back({.node = &root}); while (!stack.empty()) @@ -27,29 +27,29 @@ void optimizePrimaryKeyCondition(QueryPlan::Node & root) /// Traverse all children first. if (frame.next_child < frame.node->children.size()) { - stack.push_back({.node = frame.node->children[frame.next_child]}); - + auto next_frame = Frame{.node = frame.node->children[frame.next_child]}; ++frame.next_child; + stack.push_back(next_frame); continue; } - auto add_filter = [&](auto & storage) + auto add_read_from_storage_filter = [&](auto & storage) { - for (auto iter=stack.rbegin() + 1; iter!=stack.rend(); ++iter) + for (auto iter = stack.rbegin() + 1; iter != stack.rend(); ++iter) { if (auto * filter_step = typeid_cast(iter->node->step.get())) storage.addFilter(filter_step->getExpression(), filter_step->getFilterColumnName()); else if (typeid_cast(iter->node->step.get())) - ; + continue; else break; } }; if (auto * read_from_merge_tree = typeid_cast(frame.node->step.get())) - add_filter(*read_from_merge_tree); + add_read_from_storage_filter(*read_from_merge_tree); else if (auto * read_from_merge = typeid_cast(frame.node->step.get())) - add_filter(*read_from_merge); + add_read_from_storage_filter(*read_from_merge); stack.pop_back(); } diff --git a/src/Processors/QueryPlan/ReadFromMergeTree.cpp b/src/Processors/QueryPlan/ReadFromMergeTree.cpp index c88a244b963..b268e7deff0 100644 --- a/src/Processors/QueryPlan/ReadFromMergeTree.cpp +++ b/src/Processors/QueryPlan/ReadFromMergeTree.cpp @@ -925,8 +925,15 @@ MergeTreeDataSelectAnalysisResultPtr ReadFromMergeTree::selectRangesToRead( for (const auto & node : added_filter_nodes.nodes) nodes.nodes.push_back(node); - key_condition.emplace( - std::move(nodes), query_info.syntax_analyzer_result, query_info.prepared_sets, context, primary_key_columns, primary_key.expression); + NameSet array_join_name_set; + if (query_info.syntax_analyzer_result) + array_join_name_set = query_info.syntax_analyzer_result->getArrayJoinSourceNameSet(); + + key_condition.emplace(std::move(nodes), + context, + primary_key_columns, + primary_key.expression, + array_join_name_set); } else { diff --git a/src/Storages/HDFS/StorageHDFSCluster.cpp b/src/Storages/HDFS/StorageHDFSCluster.cpp index 64fdf77dbc2..5f9d5ea3d6d 100644 --- a/src/Storages/HDFS/StorageHDFSCluster.cpp +++ b/src/Storages/HDFS/StorageHDFSCluster.cpp @@ -11,7 +11,6 @@ #include #include #include -#include #include #include #include @@ -25,6 +24,8 @@ #include #include #include +#include +#include #include @@ -56,6 +57,7 @@ StorageHDFSCluster::StorageHDFSCluster( { auto columns = StorageHDFS::getTableStructureFromData(format_name, uri_, compression_method, context_); storage_metadata.setColumns(columns); + add_columns_structure_to_query = true; } else storage_metadata.setColumns(columns_); @@ -92,6 +94,11 @@ Pipe StorageHDFSCluster::read( const bool add_agg_info = processed_stage == QueryProcessingStage::WithMergeableState; + auto query_to_send = query_info.original_query->clone(); + if (add_columns_structure_to_query) + addColumnsStructureToQueryWithClusterEngine( + query_to_send, StorageDictionary::generateNamesAndTypesDescription(storage_snapshot->metadata->getColumns().getAll()), 3, getName()); + for (const auto & replicas : cluster->getShardsAddresses()) { /// There will be only one replica, because we consider each replica as a shard @@ -110,7 +117,7 @@ Pipe StorageHDFSCluster::read( /// So, task_identifier is passed as constructor argument. It is more obvious. auto remote_query_executor = std::make_shared( connection, - queryToString(query_info.original_query), + queryToString(query_to_send), header, context, /*throttler=*/nullptr, diff --git a/src/Storages/HDFS/StorageHDFSCluster.h b/src/Storages/HDFS/StorageHDFSCluster.h index 0d6f2bbe1b8..adcc3f5db6e 100644 --- a/src/Storages/HDFS/StorageHDFSCluster.h +++ b/src/Storages/HDFS/StorageHDFSCluster.h @@ -44,6 +44,7 @@ private: String uri; String format_name; String compression_method; + bool add_columns_structure_to_query = false; }; diff --git a/src/Storages/MergeTree/DataPartStorageOnDisk.cpp b/src/Storages/MergeTree/DataPartStorageOnDisk.cpp index efc7710f640..7b36a9873e4 100644 --- a/src/Storages/MergeTree/DataPartStorageOnDisk.cpp +++ b/src/Storages/MergeTree/DataPartStorageOnDisk.cpp @@ -6,12 +6,12 @@ #include #include #include -#include #include #include #include #include #include +#include namespace DB { @@ -29,6 +29,16 @@ DataPartStorageOnDisk::DataPartStorageOnDisk(VolumePtr volume_, std::string root { } +DataPartStorageOnDisk::DataPartStorageOnDisk( + VolumePtr volume_, std::string root_path_, std::string part_dir_, DiskTransactionPtr transaction_) + : volume(std::move(volume_)) + , root_path(std::move(root_path_)) + , part_dir(std::move(part_dir_)) + , transaction(std::move(transaction_)) + , has_shared_transaction(transaction != nullptr) +{ +} + std::string DataPartStorageOnDisk::getFullPath() const { return fs::path(volume->getDisk()->getPath()) / root_path / part_dir / ""; @@ -49,6 +59,11 @@ std::string DataPartStorageOnDisk::getFullRootPath() const return fs::path(volume->getDisk()->getPath()) / root_path / ""; } +MutableDataPartStoragePtr DataPartStorageOnDisk::getProjection(const std::string & name) +{ + return std::shared_ptr(new DataPartStorageOnDisk(volume, std::string(fs::path(root_path) / part_dir), name, transaction)); +} + DataPartStoragePtr DataPartStorageOnDisk::getProjection(const std::string & name) const { return std::make_shared(volume, std::string(fs::path(root_path) / part_dir), name); @@ -113,6 +128,7 @@ static UInt64 calculateTotalSizeOnDiskImpl(const DiskPtr & disk, const String & { if (disk->isFile(from)) return disk->getFileSize(from); + std::vector files; disk->listFiles(from, files); UInt64 res = 0; @@ -135,75 +151,11 @@ std::unique_ptr DataPartStorageOnDisk::readFile( return volume->getDisk()->readFile(fs::path(root_path) / part_dir / name, settings, read_hint, file_size); } -static std::unique_ptr openForReading(const DiskPtr & disk, const String & path) -{ - size_t file_size = disk->getFileSize(path); - return disk->readFile(path, ReadSettings().adjustBufferSize(file_size), file_size); -} - -void DataPartStorageOnDisk::loadVersionMetadata(VersionMetadata & version, Poco::Logger * log) const -{ - std::string version_file_name = fs::path(root_path) / part_dir / "txn_version.txt"; - String tmp_version_file_name = version_file_name + ".tmp"; - DiskPtr disk = volume->getDisk(); - - auto remove_tmp_file = [&]() - { - auto last_modified = disk->getLastModified(tmp_version_file_name); - auto buf = openForReading(disk, tmp_version_file_name); - String content; - readStringUntilEOF(content, *buf); - LOG_WARNING(log, "Found file {} that was last modified on {}, has size {} and the following content: {}", - tmp_version_file_name, last_modified.epochTime(), content.size(), content); - disk->removeFile(tmp_version_file_name); - }; - - if (disk->exists(version_file_name)) - { - auto buf = openForReading(disk, version_file_name); - version.read(*buf); - if (disk->exists(tmp_version_file_name)) - remove_tmp_file(); - return; - } - - /// Four (?) cases are possible: - /// 1. Part was created without transactions. - /// 2. Version metadata file was not renamed from *.tmp on part creation. - /// 3. Version metadata were written to *.tmp file, but hard restart happened before fsync. - /// 4. Fsyncs in storeVersionMetadata() work incorrectly. - - if (!disk->exists(tmp_version_file_name)) - { - /// Case 1. - /// We do not have version metadata and transactions history for old parts, - /// so let's consider that such parts were created by some ancient transaction - /// and were committed with some prehistoric CSN. - /// NOTE It might be Case 3, but version metadata file is written on part creation before other files, - /// so it's not Case 3 if part is not broken. - version.setCreationTID(Tx::PrehistoricTID, nullptr); - version.creation_csn = Tx::PrehistoricCSN; - return; - } - - /// Case 2. - /// Content of *.tmp file may be broken, just use fake TID. - /// Transaction was not committed if *.tmp file was not renamed, so we should complete rollback by removing part. - version.setCreationTID(Tx::DummyTID, nullptr); - version.creation_csn = Tx::RolledBackCSN; - remove_tmp_file(); -} - void DataPartStorageOnDisk::checkConsistency(const MergeTreeDataPartChecksums & checksums) const { checksums.checkSizes(volume->getDisk(), getRelativePath()); } -DataPartStorageBuilderPtr DataPartStorageOnDisk::getBuilder() const -{ - return std::make_shared(volume, root_path, part_dir); -} - void DataPartStorageOnDisk::remove( CanRemoveCallback && can_remove_callback, const MergeTreeDataPartChecksums & checksums, @@ -273,7 +225,7 @@ void DataPartStorageOnDisk::remove( try { disk->moveDirectory(from, to); - onRename(root_path, part_dir_without_slash); + part_dir = part_dir_without_slash; } catch (const Exception & e) { @@ -488,11 +440,6 @@ bool DataPartStorageOnDisk::looksLikeBrokenDetachedPartHasTheSameContent(const S return original_files_list == detached_files_list; } -void DataPartStorageBuilderOnDisk::setRelativePath(const std::string & path) -{ - part_dir = path; -} - std::string DataPartStorageOnDisk::getDiskName() const { return volume->getDisk()->getName(); @@ -523,7 +470,7 @@ bool DataPartStorageOnDisk::isBroken() const return volume->getDisk()->isBroken(); } -void DataPartStorageOnDisk::syncRevision(UInt64 revision) +void DataPartStorageOnDisk::syncRevision(UInt64 revision) const { volume->getDisk()->syncRevision(revision); } @@ -543,11 +490,6 @@ std::string DataPartStorageOnDisk::getDiskPath() const return volume->getDisk()->getPath(); } -DataPartStorageOnDisk::DisksSet::const_iterator DataPartStorageOnDisk::isStoredOnDisk(const DisksSet & disks) const -{ - return disks.find(volume->getDisk()); -} - ReservationPtr DataPartStorageOnDisk::reserve(UInt64 bytes) const { auto res = volume->reserve(bytes); @@ -562,159 +504,6 @@ ReservationPtr DataPartStorageOnDisk::tryReserve(UInt64 bytes) const return volume->reserve(bytes); } -size_t DataPartStorageOnDisk::getVolumeIndex(const IStoragePolicy & storage_policy) const -{ - return storage_policy.getVolumeIndexByDisk(volume->getDisk()); -} - -void DataPartStorageOnDisk::writeChecksums(const MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const -{ - std::string path = fs::path(root_path) / part_dir / "checksums.txt"; - - try - { - { - auto out = volume->getDisk()->writeFile(path + ".tmp", 4096, WriteMode::Rewrite, settings); - checksums.write(*out); - } - - volume->getDisk()->moveFile(path + ".tmp", path); - } - catch (...) - { - try - { - if (volume->getDisk()->exists(path + ".tmp")) - volume->getDisk()->removeFile(path + ".tmp"); - } - catch (...) - { - tryLogCurrentException("DataPartStorageOnDisk"); - } - - throw; - } -} - -void DataPartStorageOnDisk::writeColumns(const NamesAndTypesList & columns, const WriteSettings & settings) const -{ - std::string path = fs::path(root_path) / part_dir / "columns.txt"; - - try - { - auto buf = volume->getDisk()->writeFile(path + ".tmp", 4096, WriteMode::Rewrite, settings); - columns.writeText(*buf); - buf->finalize(); - - volume->getDisk()->moveFile(path + ".tmp", path); - } - catch (...) - { - try - { - if (volume->getDisk()->exists(path + ".tmp")) - volume->getDisk()->removeFile(path + ".tmp"); - } - catch (...) - { - tryLogCurrentException("DataPartStorageOnDisk"); - } - - throw; - } -} - -void DataPartStorageOnDisk::writeVersionMetadata(const VersionMetadata & version, bool fsync_part_dir) const -{ - std::string path = fs::path(root_path) / part_dir / "txn_version.txt"; - try - { - { - /// TODO IDisk interface does not allow to open file with O_EXCL flag (for DiskLocal), - /// so we create empty file at first (expecting that createFile throws if file already exists) - /// and then overwrite it. - volume->getDisk()->createFile(path + ".tmp"); - auto buf = volume->getDisk()->writeFile(path + ".tmp", 256); - version.write(*buf); - buf->finalize(); - buf->sync(); - } - - SyncGuardPtr sync_guard; - if (fsync_part_dir) - sync_guard = volume->getDisk()->getDirectorySyncGuard(getRelativePath()); - volume->getDisk()->replaceFile(path + ".tmp", path); - - } - catch (...) - { - try - { - if (volume->getDisk()->exists(path + ".tmp")) - volume->getDisk()->removeFile(path + ".tmp"); - } - catch (...) - { - tryLogCurrentException("DataPartStorageOnDisk"); - } - - throw; - } -} - -void DataPartStorageOnDisk::appendCSNToVersionMetadata(const VersionMetadata & version, VersionMetadata::WhichCSN which_csn) const -{ - /// Small enough appends to file are usually atomic, - /// so we append new metadata instead of rewriting file to reduce number of fsyncs. - /// We don't need to do fsync when writing CSN, because in case of hard restart - /// we will be able to restore CSN from transaction log in Keeper. - - std::string version_file_name = fs::path(root_path) / part_dir / "txn_version.txt"; - DiskPtr disk = volume->getDisk(); - auto out = disk->writeFile(version_file_name, 256, WriteMode::Append); - version.writeCSN(*out, which_csn); - out->finalize(); -} - -void DataPartStorageOnDisk::appendRemovalTIDToVersionMetadata(const VersionMetadata & version, bool clear) const -{ - String version_file_name = fs::path(root_path) / part_dir / "txn_version.txt"; - DiskPtr disk = volume->getDisk(); - auto out = disk->writeFile(version_file_name, 256, WriteMode::Append); - version.writeRemovalTID(*out, clear); - out->finalize(); - - /// fsync is not required when we clearing removal TID, because after hard restart we will fix metadata - if (!clear) - out->sync(); -} - -void DataPartStorageOnDisk::writeDeleteOnDestroyMarker(Poco::Logger * log) const -{ - String marker_path = fs::path(root_path) / part_dir / "delete-on-destroy.txt"; - auto disk = volume->getDisk(); - try - { - volume->getDisk()->createFile(marker_path); - } - catch (Poco::Exception & e) - { - LOG_ERROR(log, "{} (while creating DeleteOnDestroy marker: {})", e.what(), backQuote(fullPath(disk, marker_path))); - } -} - -void DataPartStorageOnDisk::removeDeleteOnDestroyMarker() const -{ - std::string delete_on_destroy_file_name = fs::path(root_path) / part_dir / "delete-on-destroy.txt"; - volume->getDisk()->removeFileIfExists(delete_on_destroy_file_name); -} - -void DataPartStorageOnDisk::removeVersionMetadata() const -{ - std::string version_file_name = fs::path(root_path) / part_dir / "txn_version.txt"; - volume->getDisk()->removeFileIfExists(version_file_name); -} - String DataPartStorageOnDisk::getUniqueId() const { auto disk = volume->getDisk(); @@ -724,16 +513,6 @@ String DataPartStorageOnDisk::getUniqueId() const return disk->getUniqueId(fs::path(getRelativePath()) / "checksums.txt"); } -bool DataPartStorageOnDisk::shallParticipateInMerges(const IStoragePolicy & storage_policy) const -{ - /// `IMergeTreeDataPart::volume` describes space where current part belongs, and holds - /// `SingleDiskVolume` object which does not contain up-to-date settings of corresponding volume. - /// Therefore we shall obtain volume from storage policy. - auto volume_ptr = storage_policy.getVolume(storage_policy.getVolumeIndexByDisk(volume->getDisk())); - - return !volume_ptr->areMergesAvoided(); -} - void DataPartStorageOnDisk::backup( const MergeTreeDataPartChecksums & checksums, const NameSet & files_without_checksums, @@ -798,7 +577,7 @@ void DataPartStorageOnDisk::backup( } } -DataPartStoragePtr DataPartStorageOnDisk::freeze( +MutableDataPartStoragePtr DataPartStorageOnDisk::freeze( const std::string & to, const std::string & dir_path, bool make_source_readonly, @@ -822,7 +601,7 @@ DataPartStoragePtr DataPartStorageOnDisk::freeze( return std::make_shared(single_disk_volume, to, dir_path); } -DataPartStoragePtr DataPartStorageOnDisk::clone( +MutableDataPartStoragePtr DataPartStorageOnDisk::clonePart( const std::string & to, const std::string & dir_path, const DiskPtr & disk, @@ -835,6 +614,7 @@ DataPartStoragePtr DataPartStorageOnDisk::clone( LOG_WARNING(log, "Path {} already exists. Will remove it and clone again.", fullPath(disk, path_to_clone)); disk->removeRecursive(path_to_clone); } + disk->createDirectories(to); volume->getDisk()->copy(getRelativePath(), disk, to); volume->getDisk()->removeFileIfExists(fs::path(path_to_clone) / "delete-on-destroy.txt"); @@ -843,13 +623,7 @@ DataPartStoragePtr DataPartStorageOnDisk::clone( return std::make_shared(single_disk_volume, to, dir_path); } -void DataPartStorageOnDisk::onRename(const std::string & new_root_path, const std::string & new_part_dir) -{ - part_dir = new_part_dir; - root_path = new_root_path; -} - -void DataPartStorageBuilderOnDisk::rename( +void DataPartStorageOnDisk::rename( const std::string & new_root_path, const std::string & new_part_dir, Poco::Logger * log, @@ -870,7 +644,7 @@ void DataPartStorageBuilderOnDisk::rename( "Part directory {} already exists and contains {} files. Removing it.", fullPath(volume->getDisk(), to), files.size()); - transaction->removeRecursive(to); + executeOperation([&](auto & disk) { disk.removeRecursive(to); }); } else { @@ -884,8 +658,12 @@ void DataPartStorageBuilderOnDisk::rename( String from = getRelativePath(); /// Why? - transaction->setLastModified(from, Poco::Timestamp::fromEpochTime(time(nullptr))); - transaction->moveDirectory(from, to); + executeOperation([&](auto & disk) + { + disk.setLastModified(from, Poco::Timestamp::fromEpochTime(time(nullptr))); + disk.moveDirectory(from, to); + }); + part_dir = new_part_dir; root_path = new_root_path; @@ -907,7 +685,7 @@ void DataPartStorageOnDisk::changeRootPath(const std::string & from_root, const --prefix_size; if (prefix_size > root_path.size() - || std::string_view(from_root).substr(0, prefix_size) != std::string_view(root_path).substr(0, prefix_size)) + || std::string_view(from_root).substr(0, prefix_size) != std::string_view(root_path).substr(0, prefix_size)) throw Exception( ErrorCodes::LOGICAL_ERROR, "Cannot change part root to {} because it is not a prefix of current root {}", @@ -920,51 +698,80 @@ void DataPartStorageOnDisk::changeRootPath(const std::string & from_root, const root_path = to_root.substr(0, dst_size) + root_path.substr(prefix_size); } -DataPartStorageBuilderOnDisk::DataPartStorageBuilderOnDisk( - VolumePtr volume_, - std::string root_path_, - std::string part_dir_) - : volume(std::move(volume_)) - , root_path(std::move(root_path_)) - , part_dir(std::move(part_dir_)) - , transaction(volume->getDisk()->createTransaction()) -{ -} - -std::unique_ptr DataPartStorageBuilderOnDisk::writeFile( - const String & name, - size_t buf_size, - const WriteSettings & settings) -{ - return transaction->writeFile(fs::path(root_path) / part_dir / name, buf_size, WriteMode::Rewrite, settings, /* autocommit = */ false); -} - -void DataPartStorageBuilderOnDisk::removeFile(const String & name) -{ - transaction->removeFile(fs::path(root_path) / part_dir / name); -} - -void DataPartStorageBuilderOnDisk::removeFileIfExists(const String & name) -{ - transaction->removeFileIfExists(fs::path(root_path) / part_dir / name); -} - -void DataPartStorageBuilderOnDisk::removeRecursive() -{ - transaction->removeRecursive(fs::path(root_path) / part_dir); -} - -void DataPartStorageBuilderOnDisk::removeSharedRecursive(bool keep_in_remote_fs) -{ - transaction->removeSharedRecursive(fs::path(root_path) / part_dir, keep_in_remote_fs, {}); -} - -SyncGuardPtr DataPartStorageBuilderOnDisk::getDirectorySyncGuard() const +SyncGuardPtr DataPartStorageOnDisk::getDirectorySyncGuard() const { return volume->getDisk()->getDirectorySyncGuard(fs::path(root_path) / part_dir); } -void DataPartStorageBuilderOnDisk::createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) const +template +void DataPartStorageOnDisk::executeOperation(Op && op) +{ + if (transaction) + op(*transaction); + else + op(*volume->getDisk()); +} + +std::unique_ptr DataPartStorageOnDisk::writeFile( + const String & name, + size_t buf_size, + const WriteSettings & settings) +{ + if (transaction) + return transaction->writeFile(fs::path(root_path) / part_dir / name, buf_size, WriteMode::Rewrite, settings, /* autocommit = */ false); + + return volume->getDisk()->writeFile(fs::path(root_path) / part_dir / name, buf_size, WriteMode::Rewrite, settings); +} + +std::unique_ptr DataPartStorageOnDisk::writeTransactionFile(WriteMode mode) const +{ + return volume->getDisk()->writeFile(fs::path(root_path) / part_dir / "txn_version.txt", 256, mode); +} + +void DataPartStorageOnDisk::createFile(const String & name) +{ + executeOperation([&](auto & disk) { disk.createFile(fs::path(root_path) / part_dir / name); }); +} + +void DataPartStorageOnDisk::moveFile(const String & from_name, const String & to_name) +{ + executeOperation([&](auto & disk) + { + auto relative_path = fs::path(root_path) / part_dir; + disk.moveFile(relative_path / from_name, relative_path / to_name); + }); +} + +void DataPartStorageOnDisk::replaceFile(const String & from_name, const String & to_name) +{ + executeOperation([&](auto & disk) + { + auto relative_path = fs::path(root_path) / part_dir; + disk.replaceFile(relative_path / from_name, relative_path / to_name); + }); +} + +void DataPartStorageOnDisk::removeFile(const String & name) +{ + executeOperation([&](auto & disk) { disk.removeFile(fs::path(root_path) / part_dir / name); }); +} + +void DataPartStorageOnDisk::removeFileIfExists(const String & name) +{ + executeOperation([&](auto & disk) { disk.removeFileIfExists(fs::path(root_path) / part_dir / name); }); +} + +void DataPartStorageOnDisk::removeRecursive() +{ + executeOperation([&](auto & disk) { disk.removeRecursive(fs::path(root_path) / part_dir); }); +} + +void DataPartStorageOnDisk::removeSharedRecursive(bool keep_in_remote_fs) +{ + executeOperation([&](auto & disk) { disk.removeSharedRecursive(fs::path(root_path) / part_dir, keep_in_remote_fs, {}); }); +} + +void DataPartStorageOnDisk::createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) { const auto * source_on_disk = typeid_cast(&source); if (!source_on_disk) @@ -973,58 +780,43 @@ void DataPartStorageBuilderOnDisk::createHardLinkFrom(const IDataPartStorage & s "Cannot create hardlink from different storage. Expected DataPartStorageOnDisk, got {}", typeid(source).name()); - transaction->createHardLink( - fs::path(source_on_disk->getRelativePath()) / from, - fs::path(root_path) / part_dir / to); + executeOperation([&](auto & disk) + { + disk.createHardLink( + fs::path(source_on_disk->getRelativePath()) / from, + fs::path(root_path) / part_dir / to); + }); } -bool DataPartStorageBuilderOnDisk::exists() const +void DataPartStorageOnDisk::createDirectories() { - return volume->getDisk()->exists(fs::path(root_path) / part_dir); + executeOperation([&](auto & disk) { disk.createDirectories(fs::path(root_path) / part_dir); }); } -std::string DataPartStorageBuilderOnDisk::getFullPath() const +void DataPartStorageOnDisk::createProjection(const std::string & name) { - return fs::path(volume->getDisk()->getPath()) / root_path / part_dir; + executeOperation([&](auto & disk) { disk.createDirectory(fs::path(root_path) / part_dir / name); }); } -std::string DataPartStorageBuilderOnDisk::getRelativePath() const +void DataPartStorageOnDisk::beginTransaction() { - return fs::path(root_path) / part_dir; + if (transaction) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "Uncommitted {}transaction already exists", has_shared_transaction ? "shared " : ""); + + transaction = volume->getDisk()->createTransaction(); } -void DataPartStorageBuilderOnDisk::createDirectories() +void DataPartStorageOnDisk::commitTransaction() { - transaction->createDirectories(fs::path(root_path) / part_dir); -} + if (!transaction) + throw Exception(ErrorCodes::LOGICAL_ERROR, "There is no uncommitted transaction"); -void DataPartStorageBuilderOnDisk::createProjection(const std::string & name) -{ - transaction->createDirectory(fs::path(root_path) / part_dir / name); -} + if (has_shared_transaction) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot commit shared transaction"); -ReservationPtr DataPartStorageBuilderOnDisk::reserve(UInt64 bytes) -{ - auto res = volume->reserve(bytes); - if (!res) - throw Exception(ErrorCodes::NOT_ENOUGH_SPACE, "Cannot reserve {}, not enough space", ReadableSize(bytes)); - - return res; -} - -DataPartStorageBuilderPtr DataPartStorageBuilderOnDisk::getProjection(const std::string & name) const -{ - return std::make_shared(volume, std::string(fs::path(root_path) / part_dir), name); -} - -DataPartStoragePtr DataPartStorageBuilderOnDisk::getStorage() const -{ - return std::make_shared(volume, root_path, part_dir); -} - -void DataPartStorageBuilderOnDisk::commit() -{ transaction->commit(); + transaction.reset(); } } diff --git a/src/Storages/MergeTree/DataPartStorageOnDisk.h b/src/Storages/MergeTree/DataPartStorageOnDisk.h index d325049f056..bea1596e1f7 100644 --- a/src/Storages/MergeTree/DataPartStorageOnDisk.h +++ b/src/Storages/MergeTree/DataPartStorageOnDisk.h @@ -21,6 +21,7 @@ public: std::string getPartDirectory() const override { return part_dir; } std::string getFullRootPath() const override; + MutableDataPartStoragePtr getProjection(const std::string & name) override; DataPartStoragePtr getProjection(const std::string & name) const override; bool exists() const override; @@ -41,7 +42,6 @@ public: std::optional read_hint, std::optional file_size) const override; - void loadVersionMetadata(VersionMetadata & version, Poco::Logger * log) const override; void checkConsistency(const MergeTreeDataPartChecksums & checksums) const override; void remove( @@ -60,7 +60,6 @@ public: std::optional & original_files_list) const; void setRelativePath(const std::string & path) override; - void onRename(const std::string & new_root_path, const std::string & new_part_dir) override; std::string getDiskName() const override; std::string getDiskType() const override; @@ -68,30 +67,14 @@ public: bool supportZeroCopyReplication() const override; bool supportParallelWrite() const override; bool isBroken() const override; - void syncRevision(UInt64 revision) override; + void syncRevision(UInt64 revision) const override; UInt64 getRevision() const override; std::unordered_map getSerializedMetadata(const std::vector & paths) const override; std::string getDiskPath() const override; - - DisksSet::const_iterator isStoredOnDisk(const DisksSet & disks) const override; - ReservationPtr reserve(UInt64 bytes) const override; ReservationPtr tryReserve(UInt64 bytes) const override; - size_t getVolumeIndex(const IStoragePolicy &) const override; - - void writeChecksums(const MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const override; - void writeColumns(const NamesAndTypesList & columns, const WriteSettings & settings) const override; - void writeVersionMetadata(const VersionMetadata & version, bool fsync_part_dir) const override; - void appendCSNToVersionMetadata(const VersionMetadata & version, VersionMetadata::WhichCSN which_csn) const override; - void appendRemovalTIDToVersionMetadata(const VersionMetadata & version, bool clear) const override; - void writeDeleteOnDestroyMarker(Poco::Logger * log) const override; - void removeDeleteOnDestroyMarker() const override; - void removeVersionMetadata() const override; - String getUniqueId() const override; - bool shallParticipateInMerges(const IStoragePolicy &) const override; - void backup( const MergeTreeDataPartChecksums & checksums, const NameSet & files_without_checksums, @@ -100,7 +83,7 @@ public: bool make_temporary_hard_links, TemporaryFilesOnDisks * temp_dirs) const override; - DataPartStoragePtr freeze( + MutableDataPartStoragePtr freeze( const std::string & to, const std::string & dir_path, bool make_source_readonly, @@ -108,7 +91,7 @@ public: bool copy_instead_of_hardlink, const NameSet & files_to_copy_instead_of_hardlinks) const override; - DataPartStoragePtr clone( + MutableDataPartStoragePtr clonePart( const std::string & to, const std::string & dir_path, const DiskPtr & disk, @@ -116,11 +99,51 @@ public: void changeRootPath(const std::string & from_root, const std::string & to_root) override; - DataPartStorageBuilderPtr getBuilder() const override; + void createDirectories() override; + void createProjection(const std::string & name) override; + + std::unique_ptr writeFile( + const String & name, + size_t buf_size, + const WriteSettings & settings) override; + + std::unique_ptr writeTransactionFile(WriteMode mode) const override; + + void createFile(const String & name) override; + void moveFile(const String & from_name, const String & to_name) override; + void replaceFile(const String & from_name, const String & to_name) override; + + void removeFile(const String & name) override; + void removeFileIfExists(const String & name) override; + void removeRecursive() override; + void removeSharedRecursive(bool keep_in_remote_fs) override; + + SyncGuardPtr getDirectorySyncGuard() const override; + + void createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) override; + + void rename( + const std::string & new_root_path, + const std::string & new_part_dir, + Poco::Logger * log, + bool remove_new_dir_if_exists, + bool fsync_part_dir) override; + + void beginTransaction() override; + void commitTransaction() override; + bool hasActiveTransaction() const override { return transaction != nullptr; } + private: VolumePtr volume; std::string root_path; std::string part_dir; + DiskTransactionPtr transaction; + bool has_shared_transaction = false; + + DataPartStorageOnDisk(VolumePtr volume_, std::string root_path_, std::string part_dir_, DiskTransactionPtr transaction_); + + template + void executeOperation(Op && op); void clearDirectory( const std::string & dir, @@ -134,56 +157,4 @@ private: bool is_projection) const; }; -class DataPartStorageBuilderOnDisk final : public IDataPartStorageBuilder -{ -public: - DataPartStorageBuilderOnDisk(VolumePtr volume_, std::string root_path_, std::string part_dir_); - - void setRelativePath(const std::string & path) override; - - bool exists() const override; - - void createDirectories() override; - void createProjection(const std::string & name) override; - - std::string getPartDirectory() const override { return part_dir; } - std::string getFullPath() const override; - std::string getRelativePath() const override; - - std::unique_ptr writeFile( - const String & name, - size_t buf_size, - const WriteSettings & settings) override; - - void removeFile(const String & name) override; - void removeFileIfExists(const String & name) override; - void removeRecursive() override; - void removeSharedRecursive(bool keep_in_remote_fs) override; - - SyncGuardPtr getDirectorySyncGuard() const override; - - void createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) const override; - - ReservationPtr reserve(UInt64 bytes) override; - - DataPartStorageBuilderPtr getProjection(const std::string & name) const override; - - DataPartStoragePtr getStorage() const override; - - void rename( - const std::string & new_root_path, - const std::string & new_part_dir, - Poco::Logger * log, - bool remove_new_dir_if_exists, - bool fsync_part_dir) override; - - void commit() override; - -private: - VolumePtr volume; - std::string root_path; - std::string part_dir; - DiskTransactionPtr transaction; -}; - } diff --git a/src/Storages/MergeTree/DataPartsExchange.cpp b/src/Storages/MergeTree/DataPartsExchange.cpp index 475461aa0d6..4f9c9ffd596 100644 --- a/src/Storages/MergeTree/DataPartsExchange.cpp +++ b/src/Storages/MergeTree/DataPartsExchange.cpp @@ -13,9 +13,9 @@ #include #include #include +#include #include #include -#include #include #include #include @@ -147,12 +147,13 @@ void Service::processQuery(const HTMLForm & params, ReadBuffer & /*body*/, Write CurrentMetrics::Increment metric_increment{CurrentMetrics::ReplicatedSend}; - if (part->data_part_storage->isStoredOnRemoteDisk()) + if (part->getDataPartStorage().isStoredOnRemoteDisk()) { UInt64 revision = parse(params.get("disk_revision", "0")); if (revision) - part->data_part_storage->syncRevision(revision); - revision = part->data_part_storage->getRevision(); + part->getDataPartStorage().syncRevision(revision); + + revision = part->getDataPartStorage().getRevision(); if (revision) response.addCookie({"disk_revision", toString(revision)}); } @@ -179,43 +180,32 @@ void Service::processQuery(const HTMLForm & params, ReadBuffer & /*body*/, Write std::sregex_token_iterator(remote_fs_metadata.begin(), remote_fs_metadata.end(), re, -1), std::sregex_token_iterator()); + bool send_projections = client_protocol_version >= REPLICATION_PROTOCOL_VERSION_WITH_PARTS_PROJECTION; + if (send_projections) + { + const auto & projections = part->getProjectionParts(); + writeBinary(projections.size(), out); + } + if (data_settings->allow_remote_fs_zero_copy_replication && /// In memory data part does not have metadata yet. !isInMemoryPart(part) && client_protocol_version >= REPLICATION_PROTOCOL_VERSION_WITH_PARTS_ZERO_COPY) { - auto disk_type = part->data_part_storage->getDiskType(); - if (part->data_part_storage->supportZeroCopyReplication() && std::find(capability.begin(), capability.end(), disk_type) != capability.end()) + auto disk_type = part->getDataPartStorage().getDiskType(); + if (part->getDataPartStorage().supportZeroCopyReplication() && std::find(capability.begin(), capability.end(), disk_type) != capability.end()) { /// Send metadata if the receiver's capability covers the source disk type. response.addCookie({"remote_fs_metadata", disk_type}); - if (client_protocol_version >= REPLICATION_PROTOCOL_VERSION_WITH_PARTS_PROJECTION) - { - const auto & projections = part->getProjectionParts(); - writeBinary(projections.size(), out); - } - - sendPartFromDiskRemoteMeta(part, out, true, part->getProjectionParts()); + sendPartFromDiskRemoteMeta(part, out, true, send_projections); return; } } - if (client_protocol_version >= REPLICATION_PROTOCOL_VERSION_WITH_PARTS_PROJECTION) - { - const auto & projections = part->getProjectionParts(); - writeBinary(projections.size(), out); - if (isInMemoryPart(part)) - sendPartFromMemory(part, out, projections); - else - sendPartFromDisk(part, out, client_protocol_version, projections); - } + if (isInMemoryPart(part)) + sendPartFromMemory(part, out, send_projections); else - { - if (isInMemoryPart(part)) - sendPartFromMemory(part, out); - else - sendPartFromDisk(part, out, client_protocol_version); - } + sendPartFromDisk(part, out, client_protocol_version, send_projections); } catch (const NetException &) { @@ -237,20 +227,23 @@ void Service::processQuery(const HTMLForm & params, ReadBuffer & /*body*/, Write } void Service::sendPartFromMemory( - const MergeTreeData::DataPartPtr & part, WriteBuffer & out, const std::map> & projections) + const MergeTreeData::DataPartPtr & part, WriteBuffer & out, bool send_projections) { auto metadata_snapshot = data.getInMemoryMetadataPtr(); - for (const auto & [name, projection] : projections) + if (send_projections) { - auto projection_sample_block = metadata_snapshot->projections.get(name).sample_block; - auto part_in_memory = asInMemoryPart(projection); - if (!part_in_memory) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Projection {} of part {} is not stored in memory", name, part->name); + for (const auto & [name, projection] : part->getProjectionParts()) + { + auto projection_sample_block = metadata_snapshot->projections.get(name).sample_block; + auto part_in_memory = asInMemoryPart(projection); + if (!part_in_memory) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Projection {} of part {} is not stored in memory", name, part->name); - writeStringBinary(name, out); - projection->checksums.write(out); - NativeWriter block_out(out, 0, projection_sample_block); - block_out.write(part_in_memory->block); + writeStringBinary(name, out); + projection->checksums.write(out); + NativeWriter block_out(out, 0, projection_sample_block); + block_out.write(part_in_memory->block); + } } auto part_in_memory = asInMemoryPart(part); @@ -268,7 +261,7 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( const MergeTreeData::DataPartPtr & part, WriteBuffer & out, int client_protocol_version, - const std::map> & projections) + bool send_projections) { /// We'll take a list of files from the list of checksums. MergeTreeData::DataPart::Checksums checksums = part->checksums; @@ -276,7 +269,8 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( auto file_names_without_checksums = part->getFileNamesWithoutChecksums(); for (const auto & file_name : file_names_without_checksums) { - if (client_protocol_version < REPLICATION_PROTOCOL_VERSION_WITH_PARTS_DEFAULT_COMPRESSION && file_name == IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME) + if (client_protocol_version < REPLICATION_PROTOCOL_VERSION_WITH_PARTS_DEFAULT_COMPRESSION + && file_name == IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME) continue; checksums.files[file_name] = {}; @@ -287,11 +281,10 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( { // Get rid of projection files checksums.files.erase(name + ".proj"); - auto it = projections.find(name); - if (it != projections.end()) + if (send_projections) { writeStringBinary(name, out); - MergeTreeData::DataPart::Checksums projection_checksum = sendPartFromDisk(it->second, out, client_protocol_version); + MergeTreeData::DataPart::Checksums projection_checksum = sendPartFromDisk(projection, out, client_protocol_version, false); data_checksums.addFile(name + ".proj", projection_checksum.getTotalSizeOnDisk(), projection_checksum.getTotalChecksumUInt128()); } else if (part->checksums.has(name + ".proj")) @@ -307,12 +300,12 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( { String file_name = it.first; - UInt64 size = part->data_part_storage->getFileSize(file_name); + UInt64 size = part->getDataPartStorage().getFileSize(file_name); writeStringBinary(it.first, out); writeBinary(size, out); - auto file_in = part->data_part_storage->readFile(file_name, {}, std::nullopt, std::nullopt); + auto file_in = part->getDataPartStorage().readFile(file_name, {}, std::nullopt, std::nullopt); HashingWriteBuffer hashing_out(out); copyDataWithThrottler(*file_in, hashing_out, blocker.getCounter(), data.getSendsThrottler()); @@ -323,7 +316,7 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Unexpected size of file {}, expected {} got {}", - std::string(fs::path(part->data_part_storage->getRelativePath()) / file_name), + std::string(fs::path(part->getDataPartStorage().getRelativePath()) / file_name), hashing_out.count(), size); writePODBinary(hashing_out.getHash(), out); @@ -336,18 +329,15 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDisk( return data_checksums; } -MergeTreeData::DataPart::Checksums Service::sendPartFromDiskRemoteMeta( +void Service::sendPartFromDiskRemoteMeta( const MergeTreeData::DataPartPtr & part, WriteBuffer & out, bool send_part_id, - const std::map> & projections) + bool send_projections) { - const auto * data_part_storage_on_disk = dynamic_cast(part->data_part_storage.get()); - if (!data_part_storage_on_disk) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Storage '{}' doesn't support zero-copy replication", part->data_part_storage->getDiskName()); - - if (!data_part_storage_on_disk->supportZeroCopyReplication()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Disk '{}' doesn't support zero-copy replication", data_part_storage_on_disk->getDiskName()); + auto data_part_storage = part->getDataPartStoragePtr(); + if (!data_part_storage->supportZeroCopyReplication()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Disk '{}' doesn't support zero-copy replication", data_part_storage->getDiskName()); /// We'll take a list of files from the list of checksums. MergeTreeData::DataPart::Checksums checksums = part->checksums; @@ -365,33 +355,23 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDiskRemoteMeta( std::vector paths; paths.reserve(checksums.files.size()); for (const auto & it : checksums.files) - paths.push_back(fs::path(part->data_part_storage->getRelativePath()) / it.first); + paths.push_back(fs::path(part->getDataPartStorage().getRelativePath()) / it.first); /// Serialized metadatadatas with zero ref counts. - auto metadatas = data_part_storage_on_disk->getSerializedMetadata(paths); + auto metadatas = data_part_storage->getSerializedMetadata(paths); if (send_part_id) { - String part_id = data_part_storage_on_disk->getUniqueId(); + String part_id = data_part_storage->getUniqueId(); writeStringBinary(part_id, out); } - MergeTreeData::DataPart::Checksums data_checksums; - for (const auto & [name, projection] : part->getProjectionParts()) + if (send_projections) { - auto it = projections.find(name); - if (it != projections.end()) + for (const auto & [name, projection] : part->getProjectionParts()) { - writeStringBinary(name, out); - MergeTreeData::DataPart::Checksums projection_checksum = sendPartFromDiskRemoteMeta(it->second, out, false); - data_checksums.addFile(name + ".proj", projection_checksum.getTotalSizeOnDisk(), projection_checksum.getTotalChecksumUInt128()); - } - else if (part->checksums.has(name + ".proj")) - { - // We don't send this projection, just add out checksum to bypass the following check - const auto & our_checksum = part->checksums.files.find(name + ".proj")->second; - data_checksums.addFile(name + ".proj", our_checksum.file_size, our_checksum.file_hash); + sendPartFromDiskRemoteMeta(projection, out, false, false); } } @@ -399,10 +379,10 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDiskRemoteMeta( for (const auto & it : checksums.files) { const String & file_name = it.first; - String file_path_prefix = fs::path(part->data_part_storage->getRelativePath()) / file_name; + String file_path_prefix = fs::path(part->getDataPartStorage().getRelativePath()) / file_name; /// Just some additional checks - String metadata_file_path = fs::path(data_part_storage_on_disk->getDiskPath()) / file_path_prefix; + String metadata_file_path = fs::path(data_part_storage->getDiskPath()) / file_path_prefix; fs::path metadata(metadata_file_path); if (!fs::exists(metadata)) throw Exception(ErrorCodes::CORRUPTED_DATA, "Remote metadata '{}' is not exists", file_name); @@ -426,12 +406,7 @@ MergeTreeData::DataPart::Checksums Service::sendPartFromDiskRemoteMeta( throw Exception(ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Unexpected size of file {}", metadata_file_path); writePODBinary(hashing_out.getHash(), out); - - if (!file_names_without_checksums.contains(file_name)) - data_checksums.addFile(file_name, hashing_out.count(), hashing_out.getHash()); } - - return data_checksums; } MergeTreeData::DataPartPtr Service::findPart(const String & name) @@ -706,74 +681,54 @@ MergeTreeData::MutableDataPartPtr Fetcher::fetchSelectedPart( in->setNextCallback(ReplicatedFetchReadCallback(*entry)); - return part_type == "InMemory" - ? downloadPartToMemory(part_name, part_uuid, metadata_snapshot, context, disk, *in, projections, throttler) - : downloadPartToDisk(part_name, replica_path, to_detached, tmp_prefix, sync, disk, *in, projections, checksums, throttler); + if (part_type == "InMemory") + { + auto volume = std::make_shared("volume_" + part_name, disk, 0); + + auto data_part_storage = std::make_shared( + volume, + data.getRelativeDataPath(), + part_name); + + return downloadPartToMemory( + data_part_storage, part_name, + MergeTreePartInfo::fromPartName(part_name, data.format_version), + part_uuid, metadata_snapshot, context, *in, + projections, false, throttler); + } + + return downloadPartToDisk( + part_name, replica_path, to_detached, tmp_prefix, + sync, disk, *in, projections, checksums, throttler); } MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToMemory( + MutableDataPartStoragePtr data_part_storage, const String & part_name, + const MergeTreePartInfo & part_info, const UUID & part_uuid, const StorageMetadataPtr & metadata_snapshot, ContextPtr context, - DiskPtr disk, PooledReadWriteBufferFromHTTP & in, size_t projections, + bool is_projection, ThrottlerPtr throttler) { - auto volume = std::make_shared("volume_" + part_name, disk, 0); + auto new_data_part = std::make_shared(data, part_name, part_info, data_part_storage); - auto data_part_storage = std::make_shared( - volume, - data.getRelativeDataPath(), - part_name); - - auto data_part_storage_builder = std::make_shared( - volume, - data.getRelativeDataPath(), - part_name); - - MergeTreeData::MutableDataPartPtr new_data_part = - std::make_shared(data, part_name, data_part_storage); - new_data_part->version.setCreationTID(Tx::PrehistoricTID, nullptr); - - for (auto i = 0ul; i < projections; ++i) + for (size_t i = 0; i < projections; ++i) { String projection_name; readStringBinary(projection_name, in); - MergeTreeData::DataPart::Checksums checksums; - if (!checksums.read(in)) - throw Exception("Cannot deserialize checksums", ErrorCodes::CORRUPTED_DATA); - - NativeReader block_in(in, 0); - auto block = block_in.read(); - throttler->add(block.bytes()); - - auto projection_part_storage = data_part_storage->getProjection(projection_name + ".proj"); - auto projection_part_storage_builder = data_part_storage_builder->getProjection(projection_name + ".proj"); MergeTreePartInfo new_part_info("all", 0, 0, 0); - MergeTreeData::MutableDataPartPtr new_projection_part = - std::make_shared(data, projection_name, new_part_info, projection_part_storage, new_data_part.get()); + auto projection_part_storage = data_part_storage->getProjection(projection_name + ".proj"); - new_projection_part->is_temp = false; - new_projection_part->setColumns(block.getNamesAndTypesList(), {}); - MergeTreePartition partition{}; - new_projection_part->partition = std::move(partition); - new_projection_part->minmax_idx = std::make_shared(); + auto new_projection_part = downloadPartToMemory( + projection_part_storage, projection_name, + new_part_info, part_uuid, metadata_snapshot, + context, in, 0, true, throttler); - MergedBlockOutputStream part_out( - new_projection_part, - projection_part_storage_builder, - metadata_snapshot->projections.get(projection_name).metadata, - block.getNamesAndTypesList(), - {}, - CompressionCodecFactory::instance().get("NONE", {}), - NO_TRANSACTION_PTR); - - part_out.write(block); - part_out.finalizePart(new_projection_part, false); - new_projection_part->checksums.checkEqual(checksums, /* have_uncompressed = */ true); new_data_part->addProjectionPart(projection_name, std::move(new_projection_part)); } @@ -785,14 +740,19 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToMemory( auto block = block_in.read(); throttler->add(block.bytes()); - new_data_part->uuid = part_uuid; - new_data_part->is_temp = true; new_data_part->setColumns(block.getNamesAndTypesList(), {}); - new_data_part->minmax_idx->update(block, data.getMinMaxColumnsNames(metadata_snapshot->getPartitionKey())); - new_data_part->partition.create(metadata_snapshot, block, 0, context); + + if (!is_projection) + { + new_data_part->version.setCreationTID(Tx::PrehistoricTID, nullptr); + new_data_part->uuid = part_uuid; + new_data_part->is_temp = true; + new_data_part->minmax_idx->update(block, data.getMinMaxColumnsNames(metadata_snapshot->getPartitionKey())); + new_data_part->partition.create(metadata_snapshot, block, 0, context); + } MergedBlockOutputStream part_out( - new_data_part, data_part_storage_builder, metadata_snapshot, block.getNamesAndTypesList(), {}, + new_data_part, metadata_snapshot, block.getNamesAndTypesList(), {}, CompressionCodecFactory::instance().get("NONE", {}), NO_TRANSACTION_PTR); part_out.write(block); @@ -804,7 +764,7 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToMemory( void Fetcher::downloadBasePartOrProjectionPartToDiskRemoteMeta( const String & replica_path, - DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, PooledReadWriteBufferFromHTTP & in, MergeTreeData::DataPart::Checksums & checksums, ThrottlerPtr throttler) const @@ -820,7 +780,7 @@ void Fetcher::downloadBasePartOrProjectionPartToDiskRemoteMeta( readStringBinary(file_name, in); readBinary(file_size, in); - String metadata_file = fs::path(data_part_storage_builder->getFullPath()) / file_name; + String metadata_file = fs::path(data_part_storage->getFullPath()) / file_name; { auto file_out = std::make_unique(metadata_file, DBMS_DEFAULT_BUFFER_SIZE, -1, 0666, nullptr, 0); @@ -834,8 +794,8 @@ void Fetcher::downloadBasePartOrProjectionPartToDiskRemoteMeta( /// NOTE The is_cancelled flag also makes sense to check every time you read over the network, /// performing a poll with a not very large timeout. /// And now we check it only between read chunks (in the `copyData` function). - data_part_storage_builder->removeSharedRecursive(true); - data_part_storage_builder->commit(); + data_part_storage->removeSharedRecursive(true); + data_part_storage->commitTransaction(); throw Exception("Fetching of part was cancelled", ErrorCodes::ABORTED); } @@ -855,13 +815,12 @@ void Fetcher::downloadBasePartOrProjectionPartToDiskRemoteMeta( checksums.addFile(file_name, file_size, expected_hash); } } - } void Fetcher::downloadBaseOrProjectionPartToDisk( const String & replica_path, - DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, bool sync, PooledReadWriteBufferFromHTTP & in, MergeTreeData::DataPart::Checksums & checksums, @@ -880,14 +839,14 @@ void Fetcher::downloadBaseOrProjectionPartToDisk( /// File must be inside "absolute_part_path" directory. /// Otherwise malicious ClickHouse replica may force us to write to arbitrary path. - String absolute_file_path = fs::weakly_canonical(fs::path(data_part_storage_builder->getRelativePath()) / file_name); - if (!startsWith(absolute_file_path, fs::weakly_canonical(data_part_storage_builder->getRelativePath()).string())) + String absolute_file_path = fs::weakly_canonical(fs::path(data_part_storage->getRelativePath()) / file_name); + if (!startsWith(absolute_file_path, fs::weakly_canonical(data_part_storage->getRelativePath()).string())) throw Exception(ErrorCodes::INSECURE_PATH, "File path ({}) doesn't appear to be inside part path ({}). " "This may happen if we are trying to download part from malicious replica or logical error.", - absolute_file_path, data_part_storage_builder->getRelativePath()); + absolute_file_path, data_part_storage->getRelativePath()); - auto file_out = data_part_storage_builder->writeFile(file_name, std::min(file_size, DBMS_DEFAULT_BUFFER_SIZE), {}); + auto file_out = data_part_storage->writeFile(file_name, std::min(file_size, DBMS_DEFAULT_BUFFER_SIZE), {}); HashingWriteBuffer hashing_out(*file_out); copyDataWithThrottler(in, hashing_out, file_size, blocker.getCounter(), throttler); @@ -896,7 +855,7 @@ void Fetcher::downloadBaseOrProjectionPartToDisk( /// NOTE The is_cancelled flag also makes sense to check every time you read over the network, /// performing a poll with a not very large timeout. /// And now we check it only between read chunks (in the `copyData` function). - data_part_storage_builder->removeRecursive(); + data_part_storage->removeRecursive(); throw Exception("Fetching of part was cancelled", ErrorCodes::ABORTED); } @@ -906,7 +865,7 @@ void Fetcher::downloadBaseOrProjectionPartToDisk( if (expected_hash != hashing_out.getHash()) throw Exception(ErrorCodes::CHECKSUM_DOESNT_MATCH, "Checksum mismatch for file {} transferred from {}", - (fs::path(data_part_storage_builder->getFullPath()) / file_name).string(), + (fs::path(data_part_storage->getFullPath()) / file_name).string(), replica_path); if (file_name != "checksums.txt" && @@ -951,15 +910,12 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToDisk( part_relative_path, part_dir); - DataPartStorageBuilderPtr data_part_storage_builder = std::make_shared( - volume, - part_relative_path, - part_dir); + data_part_storage->beginTransaction(); - if (data_part_storage_builder->exists()) + if (data_part_storage->exists()) { LOG_WARNING(log, "Directory {} already exists, probably result of a failed fetch. Will remove it before fetching part.", - data_part_storage_builder->getFullPath()); + data_part_storage->getFullPath()); /// Even if it's a temporary part it could be downloaded with zero copy replication and this function /// is executed as a callback. @@ -967,37 +923,36 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToDisk( /// We don't control the amount of refs for temporary parts so we cannot decide can we remove blobs /// or not. So we are not doing it bool keep_shared = disk->supportZeroCopyReplication() && data_settings->allow_remote_fs_zero_copy_replication; - data_part_storage_builder->removeSharedRecursive(keep_shared); + data_part_storage->removeSharedRecursive(keep_shared); } - data_part_storage_builder->createDirectories(); + data_part_storage->createDirectories(); SyncGuardPtr sync_guard; if (data.getSettings()->fsync_part_directory) - sync_guard = disk->getDirectorySyncGuard(data_part_storage->getRelativePath()); + sync_guard = data_part_storage->getDirectorySyncGuard(); CurrentMetrics::Increment metric_increment{CurrentMetrics::ReplicatedFetch}; - for (auto i = 0ul; i < projections; ++i) + for (size_t i = 0; i < projections; ++i) { String projection_name; readStringBinary(projection_name, in); MergeTreeData::DataPart::Checksums projection_checksum; auto projection_part_storage = data_part_storage->getProjection(projection_name + ".proj"); - auto projection_part_storage_builder = data_part_storage_builder->getProjection(projection_name + ".proj"); - - projection_part_storage_builder->createDirectories(); + projection_part_storage->createDirectories(); downloadBaseOrProjectionPartToDisk( - replica_path, projection_part_storage_builder, sync, in, projection_checksum, throttler); + replica_path, projection_part_storage, sync, in, projection_checksum, throttler); checksums.addFile( projection_name + ".proj", projection_checksum.getTotalSizeOnDisk(), projection_checksum.getTotalChecksumUInt128()); } // Download the base part - downloadBaseOrProjectionPartToDisk(replica_path, data_part_storage_builder, sync, in, checksums, throttler); + downloadBaseOrProjectionPartToDisk(replica_path, data_part_storage, sync, in, checksums, throttler); assertEOF(in); + data_part_storage->commitTransaction(); MergeTreeData::MutableDataPartPtr new_data_part = data.createPart(part_name, data_part_storage); new_data_part->version.setCreationTID(Tx::PrehistoricTID, nullptr); new_data_part->is_temp = true; @@ -1043,49 +998,43 @@ MergeTreeData::MutableDataPartPtr Fetcher::downloadPartToDiskRemoteMeta( part_relative_path, part_dir); - DataPartStorageBuilderPtr data_part_storage_builder = std::make_shared( - volume, - part_relative_path, - part_dir); + data_part_storage->beginTransaction(); if (data_part_storage->exists()) throw Exception(ErrorCodes::DIRECTORY_ALREADY_EXISTS, "Directory {} already exists.", data_part_storage->getFullPath()); CurrentMetrics::Increment metric_increment{CurrentMetrics::ReplicatedFetch}; - volume->getDisk()->createDirectories(data_part_storage->getFullPath()); + data_part_storage->createDirectories(); - for (auto i = 0ul; i < projections; ++i) + for (size_t i = 0; i < projections; ++i) { String projection_name; readStringBinary(projection_name, in); MergeTreeData::DataPart::Checksums projection_checksum; auto projection_part_storage = data_part_storage->getProjection(projection_name + ".proj"); - auto projection_part_storage_builder = data_part_storage_builder->getProjection(projection_name + ".proj"); - - projection_part_storage_builder->createDirectories(); + projection_part_storage->createDirectories(); downloadBasePartOrProjectionPartToDiskRemoteMeta( - replica_path, projection_part_storage_builder, in, projection_checksum, throttler); + replica_path, projection_part_storage, in, projection_checksum, throttler); checksums.addFile( projection_name + ".proj", projection_checksum.getTotalSizeOnDisk(), projection_checksum.getTotalChecksumUInt128()); } downloadBasePartOrProjectionPartToDiskRemoteMeta( - replica_path, data_part_storage_builder, in, checksums, throttler); + replica_path, data_part_storage, in, checksums, throttler); assertEOF(in); MergeTreeData::MutableDataPartPtr new_data_part; try { - data_part_storage_builder->commit(); + data_part_storage->commitTransaction(); new_data_part = data.createPart(part_name, data_part_storage); new_data_part->version.setCreationTID(Tx::PrehistoricTID, nullptr); new_data_part->is_temp = true; new_data_part->modification_time = time(nullptr); - new_data_part->loadColumnsChecksumsIndexes(true, false); } #if USE_AWS_S3 diff --git a/src/Storages/MergeTree/DataPartsExchange.h b/src/Storages/MergeTree/DataPartsExchange.h index 9e453ffb422..6c92fad4092 100644 --- a/src/Storages/MergeTree/DataPartsExchange.h +++ b/src/Storages/MergeTree/DataPartsExchange.h @@ -1,5 +1,6 @@ #pragma once +#include "Storages/MergeTree/MergeTreePartInfo.h" #include #include #include @@ -42,19 +43,19 @@ private: void sendPartFromMemory( const MergeTreeData::DataPartPtr & part, WriteBuffer & out, - const std::map> & projections = {}); + bool send_projections); MergeTreeData::DataPart::Checksums sendPartFromDisk( const MergeTreeData::DataPartPtr & part, WriteBuffer & out, int client_protocol_version, - const std::map> & projections = {}); + bool send_projections); - MergeTreeData::DataPart::Checksums sendPartFromDiskRemoteMeta( + void sendPartFromDiskRemoteMeta( const MergeTreeData::DataPartPtr & part, WriteBuffer & out, bool send_part_id, - const std::map> & projections = {}); + bool send_projections); /// StorageReplicatedMergeTree::shutdown() waits for all parts exchange handlers to finish, /// so Service will never access dangling reference to storage @@ -94,7 +95,7 @@ public: private: void downloadBaseOrProjectionPartToDisk( const String & replica_path, - DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, bool sync, PooledReadWriteBufferFromHTTP & in, MergeTreeData::DataPart::Checksums & checksums, @@ -102,12 +103,11 @@ private: void downloadBasePartOrProjectionPartToDiskRemoteMeta( const String & replica_path, - DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, PooledReadWriteBufferFromHTTP & in, MergeTreeData::DataPart::Checksums & checksums, ThrottlerPtr throttler) const; - MergeTreeData::MutableDataPartPtr downloadPartToDisk( const String & part_name, const String & replica_path, @@ -121,13 +121,15 @@ private: ThrottlerPtr throttler); MergeTreeData::MutableDataPartPtr downloadPartToMemory( + MutableDataPartStoragePtr data_part_storage, const String & part_name, + const MergeTreePartInfo & part_info, const UUID & part_uuid, const StorageMetadataPtr & metadata_snapshot, ContextPtr context, - DiskPtr disk, PooledReadWriteBufferFromHTTP & in, size_t projections, + bool is_projection, ThrottlerPtr throttler); MergeTreeData::MutableDataPartPtr downloadPartToDiskRemoteMeta( diff --git a/src/Storages/MergeTree/IDataPartStorage.h b/src/Storages/MergeTree/IDataPartStorage.h index 03627938348..c6669908db4 100644 --- a/src/Storages/MergeTree/IDataPartStorage.h +++ b/src/Storages/MergeTree/IDataPartStorage.h @@ -4,6 +4,9 @@ #include #include #include +#include +#include +#include #include namespace DB @@ -18,6 +21,7 @@ struct CanRemoveDescription NameSet files_not_to_remove; }; + using CanRemoveCallback = std::function; class IDataPartStorageIterator @@ -61,13 +65,10 @@ struct WriteSettings; class TemporaryFileOnDisk; -class IDataPartStorageBuilder; -using DataPartStorageBuilderPtr = std::shared_ptr; - /// This is an abstraction of storage for data part files. /// Ideally, it is assumed to contains read-only methods from IDisk. /// It is not fulfilled now, but let's try our best. -class IDataPartStorage +class IDataPartStorage : public boost::noncopyable { public: virtual ~IDataPartStorage() = default; @@ -81,16 +82,19 @@ public: /// virtual std::string getRelativeRootPath() const = 0; /// Get a storage for projection. - virtual std::shared_ptr getProjection(const std::string & name) const = 0; + virtual std::shared_ptr getProjection(const std::string & name) = 0; + virtual std::shared_ptr getProjection(const std::string & name) const = 0; /// Part directory exists. virtual bool exists() const = 0; + /// File inside part directory exists. Specified path is relative to the part path. virtual bool exists(const std::string & name) const = 0; virtual bool isDirectory(const std::string & name) const = 0; /// Modification time for part directory. virtual Poco::Timestamp getLastModified() const = 0; + /// Iterate part directory. Iteration in subdirectory is not needed yet. virtual DataPartStorageIteratorPtr iterate() const = 0; @@ -107,7 +111,6 @@ public: std::optional read_hint, std::optional file_size) const = 0; - virtual void loadVersionMetadata(VersionMetadata & version, Poco::Logger * log) const = 0; virtual void checkConsistency(const MergeTreeDataPartChecksums & checksums) const = 0; struct ProjectionChecksums @@ -129,12 +132,12 @@ public: /// Get a name like 'prefix_partdir_tryN' which does not exist in a root dir. /// TODO: remove it. - virtual std::optional getRelativePathForPrefix(Poco::Logger * log, const String & prefix, bool detached, bool broken) const = 0; + virtual std::optional getRelativePathForPrefix( + Poco::Logger * log, const String & prefix, bool detached, bool broken) const = 0; - /// Reset part directory, used for im-memory parts. + /// Reset part directory, used for in-memory parts. /// TODO: remove it. virtual void setRelativePath(const std::string & path) = 0; - virtual void onRename(const std::string & new_root_path, const std::string & new_part_dir) = 0; /// Some methods from IDisk. Needed to avoid getting internal IDisk interface. virtual std::string getDiskName() const = 0; @@ -143,41 +146,26 @@ public: virtual bool supportZeroCopyReplication() const { return false; } virtual bool supportParallelWrite() const = 0; virtual bool isBroken() const = 0; - virtual void syncRevision(UInt64 revision) = 0; + + /// TODO: remove or at least remove const. + virtual void syncRevision(UInt64 revision) const = 0; virtual UInt64 getRevision() const = 0; + virtual std::unordered_map getSerializedMetadata(const std::vector & paths) const = 0; /// Get a path for internal disk if relevant. It is used mainly for logging. virtual std::string getDiskPath() const = 0; - /// Check if data part is stored on one of the specified disk in set. - using DisksSet = std::unordered_set; - virtual DisksSet::const_iterator isStoredOnDisk(const DisksSet & disks) const { return disks.end(); } - /// Reserve space on the same disk. /// Probably we should try to remove it later. - virtual ReservationPtr reserve(UInt64 /*bytes*/) const { return nullptr; } - virtual ReservationPtr tryReserve(UInt64 /*bytes*/) const { return nullptr; } - virtual size_t getVolumeIndex(const IStoragePolicy &) const { return 0; } - - /// Some methods which change data part internals possibly after creation. - /// Probably we should try to remove it later. - virtual void writeChecksums(const MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const = 0; - virtual void writeColumns(const NamesAndTypesList & columns, const WriteSettings & settings) const = 0; - virtual void writeVersionMetadata(const VersionMetadata & version, bool fsync_part_dir) const = 0; - virtual void appendCSNToVersionMetadata(const VersionMetadata & version, VersionMetadata::WhichCSN which_csn) const = 0; - virtual void appendRemovalTIDToVersionMetadata(const VersionMetadata & version, bool clear) const = 0; - virtual void writeDeleteOnDestroyMarker(Poco::Logger * log) const = 0; - virtual void removeDeleteOnDestroyMarker() const = 0; - virtual void removeVersionMetadata() const = 0; + /// TODO: remove constness + virtual ReservationPtr reserve(UInt64 /*bytes*/) const { return nullptr; } + virtual ReservationPtr tryReserve(UInt64 /*bytes*/) const { return nullptr; } /// A leak of abstraction. /// Return some uniq string for file. /// Required for distinguish different copies of the same part on remote FS. virtual String getUniqueId() const = 0; - /// A leak of abstraction - virtual bool shallParticipateInMerges(const IStoragePolicy &) const { return true; } - /// Create a backup of a data part. /// This method adds a new entry to backup_entries. /// Also creates a new tmp_dir for internal disk (if disk is mentioned the first time). @@ -205,7 +193,7 @@ public: const NameSet & files_to_copy_instead_of_hardlinks) const = 0; /// Make a full copy of a data part into 'to/dir_path' (possibly to a different disk). - virtual std::shared_ptr clone( + virtual std::shared_ptr clonePart( const std::string & to, const std::string & dir_path, const DiskPtr & disk, @@ -215,33 +203,22 @@ public: /// Right now, this is needed for rename table query. virtual void changeRootPath(const std::string & from_root, const std::string & to_root) = 0; - /// Leak of abstraction as well. We should use builder as one-time object which allow - /// us to build parts, while storage should be read-only method to access part properties - /// related to disk. However our code is really tricky and sometimes we need ad-hoc builders. - virtual DataPartStorageBuilderPtr getBuilder() const = 0; -}; - -using DataPartStoragePtr = std::shared_ptr; - -/// This interface is needed to write data part. -class IDataPartStorageBuilder -{ -public: - virtual ~IDataPartStorageBuilder() = default; - - /// Reset part directory, used for im-memory parts - virtual void setRelativePath(const std::string & path) = 0; - - virtual std::string getPartDirectory() const = 0; - virtual std::string getFullPath() const = 0; - virtual std::string getRelativePath() const = 0; - - virtual bool exists() const = 0; - virtual void createDirectories() = 0; virtual void createProjection(const std::string & name) = 0; - virtual std::unique_ptr writeFile(const String & name, size_t buf_size, const WriteSettings & settings) = 0; + virtual std::unique_ptr writeFile( + const String & name, + size_t buf_size, + const WriteSettings & settings) = 0; + + /// A special const method to write transaction file. + /// It's const, because file with transaction metadata + /// can be modified after part creation. + virtual std::unique_ptr writeTransactionFile(WriteMode mode) const = 0; + + virtual void createFile(const String & name) = 0; + virtual void moveFile(const String & from_name, const String & to_name) = 0; + virtual void replaceFile(const String & from_name, const String & to_name) = 0; virtual void removeFile(const String & name) = 0; virtual void removeFileIfExists(const String & name) = 0; @@ -250,20 +227,12 @@ public: virtual SyncGuardPtr getDirectorySyncGuard() const { return nullptr; } - virtual void createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) const = 0; - - virtual ReservationPtr reserve(UInt64 /*bytes*/) { return nullptr; } - - virtual std::shared_ptr getProjection(const std::string & name) const = 0; - - virtual DataPartStoragePtr getStorage() const = 0; + virtual void createHardLinkFrom(const IDataPartStorage & source, const std::string & from, const std::string & to) = 0; /// Rename part. /// Ideally, new_root_path should be the same as current root (but it is not true). /// Examples are: 'all_1_2_1' -> 'detached/all_1_2_1' /// 'moving/tmp_all_1_2_1' -> 'all_1_2_1' - /// - /// To notify storage also call onRename for it with first two args virtual void rename( const std::string & new_root_path, const std::string & new_part_dir, @@ -271,7 +240,35 @@ public: bool remove_new_dir_if_exists, bool fsync_part_dir) = 0; - virtual void commit() = 0; + /// Starts a transaction of mutable operations. + virtual void beginTransaction() = 0; + /// Commits a transaction of mutable operations. + virtual void commitTransaction() = 0; + virtual bool hasActiveTransaction() const = 0; +}; + +using DataPartStoragePtr = std::shared_ptr; +using MutableDataPartStoragePtr = std::shared_ptr; + +/// A holder that encapsulates data part storage and +/// gives access to const storage from const methods +/// and to mutable storage from non-const methods. +class DataPartStorageHolder : public boost::noncopyable +{ +public: + explicit DataPartStorageHolder(MutableDataPartStoragePtr storage_) + : storage(std::move(storage_)) + { + } + + IDataPartStorage & getDataPartStorage() { return *storage; } + const IDataPartStorage & getDataPartStorage() const { return *storage; } + + MutableDataPartStoragePtr getDataPartStoragePtr() { return storage; } + DataPartStoragePtr getDataPartStoragePtr() const { return storage; } + +private: + MutableDataPartStoragePtr storage; }; } diff --git a/src/Storages/MergeTree/IMergeTreeDataPart.cpp b/src/Storages/MergeTree/IMergeTreeDataPart.cpp index cc9a14162f8..368af55aa15 100644 --- a/src/Storages/MergeTree/IMergeTreeDataPart.cpp +++ b/src/Storages/MergeTree/IMergeTreeDataPart.cpp @@ -1,4 +1,5 @@ #include "IMergeTreeDataPart.h" +#include "Storages/MergeTree/IDataPartStorage.h" #include #include @@ -101,7 +102,7 @@ void IMergeTreeDataPart::MinMaxIndex::load(const MergeTreeData & data, const Par } IMergeTreeDataPart::MinMaxIndex::WrittenFiles IMergeTreeDataPart::MinMaxIndex::store( - const MergeTreeData & data, const DataPartStorageBuilderPtr & data_part_storage_builder, Checksums & out_checksums) const + const MergeTreeData & data, IDataPartStorage & part_storage, Checksums & out_checksums) const { auto metadata_snapshot = data.getInMemoryMetadataPtr(); const auto & partition_key = metadata_snapshot->getPartitionKey(); @@ -109,20 +110,20 @@ IMergeTreeDataPart::MinMaxIndex::WrittenFiles IMergeTreeDataPart::MinMaxIndex::s auto minmax_column_names = data.getMinMaxColumnsNames(partition_key); auto minmax_column_types = data.getMinMaxColumnsTypes(partition_key); - return store(minmax_column_names, minmax_column_types, data_part_storage_builder, out_checksums); + return store(minmax_column_names, minmax_column_types, part_storage, out_checksums); } IMergeTreeDataPart::MinMaxIndex::WrittenFiles IMergeTreeDataPart::MinMaxIndex::store( const Names & column_names, const DataTypes & data_types, - const DataPartStorageBuilderPtr & data_part_storage_builder, + IDataPartStorage & part_storage, Checksums & out_checksums) const { if (!initialized) throw Exception( ErrorCodes::LOGICAL_ERROR, "Attempt to store uninitialized MinMax index for part {}. This is a bug", - data_part_storage_builder->getFullPath()); + part_storage.getFullPath()); WrittenFiles written_files; @@ -131,7 +132,7 @@ IMergeTreeDataPart::MinMaxIndex::WrittenFiles IMergeTreeDataPart::MinMaxIndex::s String file_name = "minmax_" + escapeForFileName(column_names[i]) + ".idx"; auto serialization = data_types.at(i)->getDefaultSerialization(); - auto out = data_part_storage_builder->writeFile(file_name, DBMS_DEFAULT_BUFFER_SIZE, {}); + auto out = part_storage.writeFile(file_name, DBMS_DEFAULT_BUFFER_SIZE, {}); HashingWriteBuffer out_hashing(*out); serialization->serializeBinary(hyperrectangle[i].left, out_hashing); serialization->serializeBinary(hyperrectangle[i].right, out_hashing); @@ -301,13 +302,13 @@ static void decrementTypeMetric(MergeTreeDataPartType type) IMergeTreeDataPart::IMergeTreeDataPart( const MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, Type part_type_, const IMergeTreeDataPart * parent_part_) - : storage(storage_) + : DataPartStorageHolder(data_part_storage_) + , storage(storage_) , name(name_) , info(MergeTreePartInfo::fromPartName(name_, storage.format_version)) - , data_part_storage(parent_part_ ? parent_part_->data_part_storage : data_part_storage_) , index_granularity_info(storage_, part_type_) , part_type(part_type_) , parent_part(parent_part_) @@ -315,6 +316,7 @@ IMergeTreeDataPart::IMergeTreeDataPart( { if (parent_part) state = MergeTreeDataPartState::Active; + incrementStateMetric(state); incrementTypeMetric(part_type); @@ -328,13 +330,13 @@ IMergeTreeDataPart::IMergeTreeDataPart( const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, Type part_type_, const IMergeTreeDataPart * parent_part_) - : storage(storage_) + : DataPartStorageHolder(data_part_storage_) + , storage(storage_) , name(name_) , info(info_) - , data_part_storage(data_part_storage_) , index_granularity_info(storage_, part_type_) , part_type(part_type_) , parent_part(parent_part_) @@ -342,6 +344,7 @@ IMergeTreeDataPart::IMergeTreeDataPart( { if (parent_part) state = MergeTreeDataPartState::Active; + incrementStateMetric(state); incrementTypeMetric(part_type); @@ -505,17 +508,17 @@ void IMergeTreeDataPart::removeIfNeeded() std::string path; try { - path = data_part_storage->getRelativePath(); + path = getDataPartStorage().getRelativePath(); - if (!data_part_storage->exists()) // path + if (!getDataPartStorage().exists()) // path return; if (is_temp) { - String file_name = fileName(data_part_storage->getPartDirectory()); + String file_name = fileName(getDataPartStorage().getPartDirectory()); if (file_name.empty()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "relative_path {} of part {} is invalid or not set", data_part_storage->getPartDirectory(), name); + throw Exception(ErrorCodes::LOGICAL_ERROR, "relative_path {} of part {} is invalid or not set", getDataPartStorage().getPartDirectory(), name); if (!startsWith(file_name, "tmp") && !endsWith(file_name, ".tmp_proj")) { @@ -620,7 +623,7 @@ String IMergeTreeDataPart::getColumnNameWithMinimumCompressedSize(bool with_subc } if (!minimum_size_column) - throw Exception("Could not find a column of minimum size in MergeTree, part " + data_part_storage->getFullPath(), ErrorCodes::LOGICAL_ERROR); + throw Exception("Could not find a column of minimum size in MergeTree, part " + getDataPartStorage().getFullPath(), ErrorCodes::LOGICAL_ERROR); return *minimum_size_column; } @@ -698,9 +701,9 @@ void IMergeTreeDataPart::loadProjections(bool require_columns_checksums, bool ch for (const auto & projection : metadata_snapshot->projections) { String path = /*getRelativePath() + */ projection.name + ".proj"; - if (data_part_storage->exists(path)) + if (getDataPartStorage().exists(path)) { - auto projection_part_storage = data_part_storage->getProjection(projection.name + ".proj"); + auto projection_part_storage = getDataPartStorage().getProjection(projection.name + ".proj"); auto part = storage.createPart(projection.name, {"all", 0, 0, 0}, projection_part_storage, this); part->loadColumnsChecksumsIndexes(require_columns_checksums, check_consistency); projection_parts.emplace(projection.name, std::move(part)); @@ -741,8 +744,8 @@ void IMergeTreeDataPart::loadIndex() loaded_index[i]->reserve(index_granularity.getMarksCount()); } - String index_name = "primary" + getIndexExtensionFromFilesystem(data_part_storage).value(); - String index_path = fs::path(data_part_storage->getRelativePath()) / index_name; + String index_name = "primary" + getIndexExtensionFromFilesystem(getDataPartStorage()).value(); + String index_path = fs::path(getDataPartStorage().getRelativePath()) / index_name; auto index_file = metadata_manager->read(index_name); size_t marks_count = index_granularity.getMarksCount(); @@ -781,7 +784,7 @@ void IMergeTreeDataPart::appendFilesOfIndex(Strings & files) const if (metadata_snapshot->hasPrimaryKey()) { - String index_name = "primary" + getIndexExtensionFromFilesystem(data_part_storage).value(); + String index_name = "primary" + getIndexExtensionFromFilesystem(getDataPartStorage()).value(); files.push_back(index_name); } } @@ -793,10 +796,10 @@ NameSet IMergeTreeDataPart::getFileNamesWithoutChecksums() const NameSet result = {"checksums.txt", "columns.txt"}; - if (data_part_storage->exists(DEFAULT_COMPRESSION_CODEC_FILE_NAME)) + if (getDataPartStorage().exists(DEFAULT_COMPRESSION_CODEC_FILE_NAME)) result.emplace(DEFAULT_COMPRESSION_CODEC_FILE_NAME); - if (data_part_storage->exists(TXN_VERSION_METADATA_FILE_NAME)) + if (getDataPartStorage().exists(TXN_VERSION_METADATA_FILE_NAME)) result.emplace(TXN_VERSION_METADATA_FILE_NAME); return result; @@ -811,7 +814,7 @@ void IMergeTreeDataPart::loadDefaultCompressionCodec() return; } - String path = fs::path(data_part_storage->getRelativePath()) / DEFAULT_COMPRESSION_CODEC_FILE_NAME; + String path = fs::path(getDataPartStorage().getRelativePath()) / DEFAULT_COMPRESSION_CODEC_FILE_NAME; bool exists = metadata_manager->exists(DEFAULT_COMPRESSION_CODEC_FILE_NAME); if (!exists) { @@ -851,6 +854,120 @@ void IMergeTreeDataPart::loadDefaultCompressionCodec() } } +template +void IMergeTreeDataPart::writeMetadata(const String & filename, const WriteSettings & settings, Writer && writer) +{ + auto & data_part_storage = getDataPartStorage(); + auto tmp_filename = filename + ".tmp"; + + try + { + { + auto out = data_part_storage.writeFile(tmp_filename, 4096, settings); + writer(*out); + out->finalize(); + } + + data_part_storage.moveFile(tmp_filename, filename); + } + catch (...) + { + try + { + if (data_part_storage.exists(tmp_filename)) + data_part_storage.removeFile(tmp_filename); + } + catch (...) + { + tryLogCurrentException("DataPartStorageOnDisk"); + } + + throw; + } +} + +void IMergeTreeDataPart::writeChecksums(const MergeTreeDataPartChecksums & checksums_, const WriteSettings & settings) +{ + writeMetadata("checksums.txt", settings, [&checksums_](auto & buffer) + { + checksums_.write(buffer); + }); +} + +void IMergeTreeDataPart::writeColumns(const NamesAndTypesList & columns_, const WriteSettings & settings) +{ + writeMetadata("columns.txt", settings, [&columns_](auto & buffer) + { + columns_.writeText(buffer); + }); +} + +void IMergeTreeDataPart::writeVersionMetadata(const VersionMetadata & version_, bool fsync_part_dir) const +{ + static constexpr auto filename = "txn_version.txt"; + static constexpr auto tmp_filename = "txn_version.txt.tmp"; + auto & data_part_storage = const_cast(getDataPartStorage()); + + try + { + { + /// TODO IDisk interface does not allow to open file with O_EXCL flag (for DiskLocal), + /// so we create empty file at first (expecting that createFile throws if file already exists) + /// and then overwrite it. + data_part_storage.createFile(tmp_filename); + auto write_settings = storage.getContext()->getWriteSettings(); + auto buf = data_part_storage.writeFile(tmp_filename, 256, write_settings); + version_.write(*buf); + buf->finalize(); + buf->sync(); + } + + SyncGuardPtr sync_guard; + if (fsync_part_dir) + sync_guard = data_part_storage.getDirectorySyncGuard(); + data_part_storage.replaceFile(tmp_filename, filename); + } + catch (...) + { + try + { + if (data_part_storage.exists(tmp_filename)) + data_part_storage.removeFile(tmp_filename); + } + catch (...) + { + tryLogCurrentException("DataPartStorageOnDisk"); + } + + throw; + } +} + +void IMergeTreeDataPart::writeDeleteOnDestroyMarker() +{ + static constexpr auto marker_path = "delete-on-destroy.txt"; + + try + { + getDataPartStorage().createFile(marker_path); + } + catch (Poco::Exception & e) + { + LOG_ERROR(storage.log, "{} (while creating DeleteOnDestroy marker: {})", + e.what(), (fs::path(getDataPartStorage().getFullPath()) / marker_path).string()); + } +} + +void IMergeTreeDataPart::removeDeleteOnDestroyMarker() +{ + getDataPartStorage().removeFileIfExists("delete-on-destroy.txt"); +} + +void IMergeTreeDataPart::removeVersionMetadata() +{ + getDataPartStorage().removeFileIfExists("txn_version.txt"); +} + void IMergeTreeDataPart::appendFilesOfDefaultCompressionCodec(Strings & files) { files.push_back(DEFAULT_COMPRESSION_CODEC_FILE_NAME); @@ -880,7 +997,7 @@ CompressionCodecPtr IMergeTreeDataPart::detectDefaultCompressionCodec() const String candidate_path = /*fs::path(getRelativePath()) */ (ISerialization::getFileNameForStream(part_column, substream_path) + ".bin"); /// We can have existing, but empty .bin files. Example: LowCardinality(Nullable(...)) columns and column_name.dict.null.bin file. - if (data_part_storage->exists(candidate_path) && data_part_storage->getFileSize(candidate_path) != 0) + if (getDataPartStorage().exists(candidate_path) && getDataPartStorage().getFileSize(candidate_path) != 0) path_to_data_file = candidate_path; } }); @@ -891,7 +1008,7 @@ CompressionCodecPtr IMergeTreeDataPart::detectDefaultCompressionCodec() const continue; } - result = getCompressionCodecForFile(data_part_storage, path_to_data_file); + result = getCompressionCodecForFile(getDataPartStorage(), path_to_data_file); break; } } @@ -936,7 +1053,7 @@ void IMergeTreeDataPart::loadPartitionAndMinMaxIndex() String calculated_partition_id = partition.getID(metadata_snapshot->getPartitionKey().sample_block); if (calculated_partition_id != info.partition_id) throw Exception( - "While loading part " + data_part_storage->getFullPath() + ": calculated partition ID: " + calculated_partition_id + "While loading part " + getDataPartStorage().getFullPath() + ": calculated partition ID: " + calculated_partition_id + " differs from partition ID in part name: " + info.partition_id, ErrorCodes::CORRUPTED_DATA); } @@ -965,7 +1082,7 @@ void IMergeTreeDataPart::loadChecksums(bool require) bytes_on_disk = checksums.getTotalSizeOnDisk(); } else - bytes_on_disk = data_part_storage->calculateTotalSizeOnDisk(); + bytes_on_disk = getDataPartStorage().calculateTotalSizeOnDisk(); } else { @@ -977,7 +1094,7 @@ void IMergeTreeDataPart::loadChecksums(bool require) LOG_WARNING(storage.log, "Checksums for part {} not found. Will calculate them from data on disk.", name); checksums = checkDataPart(shared_from_this(), false); - data_part_storage->writeChecksums(checksums, {}); + writeChecksums(checksums, {}); bytes_on_disk = checksums.getTotalSizeOnDisk(); } @@ -990,8 +1107,6 @@ void IMergeTreeDataPart::appendFilesOfChecksums(Strings & files) void IMergeTreeDataPart::loadRowsCount() { - //String path = fs::path(getRelativePath()) / "count.txt"; - auto read_rows_count = [&]() { auto buf = metadata_manager->read("count.txt"); @@ -1062,7 +1177,7 @@ void IMergeTreeDataPart::loadRowsCount() } else { - if (data_part_storage->exists("count.txt")) + if (getDataPartStorage().exists("count.txt")) { read_rows_count(); return; @@ -1161,7 +1276,7 @@ void IMergeTreeDataPart::appendFilesOfUUID(Strings & files) void IMergeTreeDataPart::loadColumns(bool require) { - String path = fs::path(data_part_storage->getRelativePath()) / "columns.txt"; + String path = fs::path(getDataPartStorage().getRelativePath()) / "columns.txt"; auto metadata_snapshot = storage.getInMemoryMetadataPtr(); if (parent_part) metadata_snapshot = metadata_snapshot->projections.get(name).metadata; @@ -1172,18 +1287,18 @@ void IMergeTreeDataPart::loadColumns(bool require) { /// We can get list of columns only from columns.txt in compact parts. if (require || part_type == Type::Compact) - throw Exception("No columns.txt in part " + name + ", expected path " + path + " on drive " + data_part_storage->getDiskName(), + throw Exception("No columns.txt in part " + name + ", expected path " + path + " on drive " + getDataPartStorage().getDiskName(), ErrorCodes::NO_FILE_IN_DATA_PART); /// If there is no file with a list of columns, write it down. for (const NameAndTypePair & column : metadata_snapshot->getColumns().getAllPhysical()) - if (data_part_storage->exists(getFileNameForColumn(column) + ".bin")) + if (getDataPartStorage().exists(getFileNameForColumn(column) + ".bin")) loaded_columns.push_back(column); if (columns.empty()) throw Exception("No columns in part " + name, ErrorCodes::NO_FILE_IN_DATA_PART); - data_part_storage->writeColumns(loaded_columns, {}); + writeColumns(loaded_columns, {}); } else { @@ -1227,7 +1342,7 @@ void IMergeTreeDataPart::assertHasVersionMetadata(MergeTreeTransaction * txn) co name, storage.getStorageID().getNameForLogs(), version.creation_tid, txn ? txn->dumpDescription() : ""); assert(!txn || storage.supportsTransactions()); - assert(!txn || data_part_storage->exists(TXN_VERSION_METADATA_FILE_NAME)); + assert(!txn || getDataPartStorage().exists(TXN_VERSION_METADATA_FILE_NAME)); } void IMergeTreeDataPart::storeVersionMetadata(bool force) const @@ -1242,7 +1357,7 @@ void IMergeTreeDataPart::storeVersionMetadata(bool force) const throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Transactions are not supported for in-memory parts (table: {}, part: {})", storage.getStorageID().getNameForLogs(), name); - data_part_storage->writeVersionMetadata(version, storage.getSettings()->fsync_part_directory); + writeVersionMetadata(version, storage.getSettings()->fsync_part_directory); } void IMergeTreeDataPart::appendCSNToVersionMetadata(VersionMetadata::WhichCSN which_csn) const @@ -1254,7 +1369,14 @@ void IMergeTreeDataPart::appendCSNToVersionMetadata(VersionMetadata::WhichCSN wh chassert(!(which_csn == VersionMetadata::WhichCSN::REMOVAL && version.removal_csn == 0)); chassert(isStoredOnDisk()); - data_part_storage->appendCSNToVersionMetadata(version, which_csn); + /// Small enough appends to file are usually atomic, + /// so we append new metadata instead of rewriting file to reduce number of fsyncs. + /// We don't need to do fsync when writing CSN, because in case of hard restart + /// we will be able to restore CSN from transaction log in Keeper. + + auto out = getDataPartStorage().writeTransactionFile(WriteMode::Append); + version.writeCSN(*out, which_csn); + out->finalize(); } void IMergeTreeDataPart::appendRemovalTIDToVersionMetadata(bool clear) const @@ -1277,13 +1399,74 @@ void IMergeTreeDataPart::appendRemovalTIDToVersionMetadata(bool clear) const else LOG_TEST(storage.log, "Appending removal TID for {} (creation: {}, removal {})", name, version.creation_tid, version.removal_tid); - data_part_storage->appendRemovalTIDToVersionMetadata(version, clear); + auto out = getDataPartStorage().writeTransactionFile(WriteMode::Append); + version.writeRemovalTID(*out, clear); + out->finalize(); + + /// fsync is not required when we clearing removal TID, because after hard restart we will fix metadata + if (!clear) + out->sync(); +} + +static std::unique_ptr openForReading(const IDataPartStorage & part_storage, const String & filename) +{ + size_t file_size = part_storage.getFileSize(filename); + return part_storage.readFile(filename, ReadSettings().adjustBufferSize(file_size), file_size, file_size); } void IMergeTreeDataPart::loadVersionMetadata() const try { - data_part_storage->loadVersionMetadata(version, storage.log); + static constexpr auto version_file_name = "txn_version.txt"; + static constexpr auto tmp_version_file_name = "txn_version.txt.tmp"; + auto & data_part_storage = const_cast(getDataPartStorage()); + + auto remove_tmp_file = [&]() + { + auto last_modified = data_part_storage.getLastModified(); + auto buf = openForReading(data_part_storage, tmp_version_file_name); + + String content; + readStringUntilEOF(content, *buf); + LOG_WARNING(storage.log, "Found file {} that was last modified on {}, has size {} and the following content: {}", + tmp_version_file_name, last_modified.epochTime(), content.size(), content); + data_part_storage.removeFile(tmp_version_file_name); + }; + + if (data_part_storage.exists(version_file_name)) + { + auto buf = openForReading(data_part_storage, version_file_name); + version.read(*buf); + if (data_part_storage.exists(tmp_version_file_name)) + remove_tmp_file(); + return; + } + + /// Four (?) cases are possible: + /// 1. Part was created without transactions. + /// 2. Version metadata file was not renamed from *.tmp on part creation. + /// 3. Version metadata were written to *.tmp file, but hard restart happened before fsync. + /// 4. Fsyncs in storeVersionMetadata() work incorrectly. + + if (!data_part_storage.exists(tmp_version_file_name)) + { + /// Case 1. + /// We do not have version metadata and transactions history for old parts, + /// so let's consider that such parts were created by some ancient transaction + /// and were committed with some prehistoric CSN. + /// NOTE It might be Case 3, but version metadata file is written on part creation before other files, + /// so it's not Case 3 if part is not broken. + version.setCreationTID(Tx::PrehistoricTID, nullptr); + version.creation_csn = Tx::PrehistoricCSN; + return; + } + + /// Case 2. + /// Content of *.tmp file may be broken, just use fake TID. + /// Transaction was not committed if *.tmp file was not renamed, so we should complete rollback by removing part. + version.setCreationTID(Tx::DummyTID, nullptr); + version.creation_csn = Tx::RolledBackCSN; + remove_tmp_file(); } catch (Exception & e) { @@ -1320,15 +1503,15 @@ bool IMergeTreeDataPart::assertHasValidVersionMetadata() const if (state == MergeTreeDataPartState::Temporary) return true; - if (!data_part_storage->exists()) + if (!getDataPartStorage().exists()) return true; String content; String version_file_name = TXN_VERSION_METADATA_FILE_NAME; try { - size_t file_size = data_part_storage->getFileSize(TXN_VERSION_METADATA_FILE_NAME); - auto buf = data_part_storage->readFile(TXN_VERSION_METADATA_FILE_NAME, ReadSettings().adjustBufferSize(file_size), file_size, std::nullopt); + size_t file_size = getDataPartStorage().getFileSize(TXN_VERSION_METADATA_FILE_NAME); + auto buf = getDataPartStorage().readFile(TXN_VERSION_METADATA_FILE_NAME, ReadSettings().adjustBufferSize(file_size), file_size, std::nullopt); readStringUntilEOF(content, *buf); ReadBufferFromString str_buf{content}; @@ -1362,10 +1545,11 @@ void IMergeTreeDataPart::appendFilesOfColumns(Strings & files) bool IMergeTreeDataPart::shallParticipateInMerges(const StoragePolicyPtr & storage_policy) const { - return data_part_storage->shallParticipateInMerges(*storage_policy); + auto disk_name = getDataPartStorage().getDiskName(); + return !storage_policy->getVolumeByDiskName(disk_name)->areMergesAvoided(); } -void IMergeTreeDataPart::renameTo(const String & new_relative_path, bool remove_new_dir_if_exists, DataPartStorageBuilderPtr builder) const +void IMergeTreeDataPart::renameTo(const String & new_relative_path, bool remove_new_dir_if_exists) try { assertOnDisk(); @@ -1376,22 +1560,21 @@ try if (parent_part) { /// For projections, move is only possible inside parent part dir. - relative_path = parent_part->data_part_storage->getRelativePath(); + relative_path = parent_part->getDataPartStorage().getRelativePath(); } - String from = data_part_storage->getRelativePath(); + auto old_projection_root_path = getDataPartStorage().getRelativePath(); auto to = fs::path(relative_path) / new_relative_path; metadata_manager->deleteAll(true); metadata_manager->assertAllDeleted(true); - builder->rename(to.parent_path(), to.filename(), storage.log, remove_new_dir_if_exists, fsync_dir); - data_part_storage->onRename(to.parent_path(), to.filename()); + getDataPartStorage().rename(to.parent_path(), to.filename(), storage.log, remove_new_dir_if_exists, fsync_dir); metadata_manager->updateAll(true); - for (const auto & [p_name, part] : projection_parts) - { - part->data_part_storage = data_part_storage->getProjection(p_name + ".proj"); - } + auto new_projection_root_path = to.string(); + + for (const auto & [_, part] : projection_parts) + part->getDataPartStorage().changeRootPath(old_projection_root_path, new_projection_root_path); } catch (...) { @@ -1432,14 +1615,14 @@ void IMergeTreeDataPart::initializePartMetadataManager() void IMergeTreeDataPart::initializeIndexGranularityInfo() { - auto mrk_ext = MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(data_part_storage); + auto mrk_ext = MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(getDataPartStorage()); if (mrk_ext) index_granularity_info = MergeTreeIndexGranularityInfo(storage, MarkType{*mrk_ext}); else index_granularity_info = MergeTreeIndexGranularityInfo(storage, part_type); } -void IMergeTreeDataPart::remove() const +void IMergeTreeDataPart::remove() { assert(assertHasValidVersionMetadata()); part_is_probably_removed_from_disk = true; @@ -1456,7 +1639,6 @@ void IMergeTreeDataPart::remove() const return CanRemoveDescription{.can_remove_anything = can_remove, .files_not_to_remove = files_not_to_remove }; }; - if (!isStoredOnDisk()) return; @@ -1475,7 +1657,7 @@ void IMergeTreeDataPart::remove() const projection_checksums.emplace_back(IDataPartStorage::ProjectionChecksums{.name = p_name, .checksums = projection_part->checksums}); } - data_part_storage->remove(std::move(can_remove_callback), checksums, projection_checksums, is_temp, getState(), storage.log); + getDataPartStorage().remove(std::move(can_remove_callback), checksums, projection_checksums, is_temp, getState(), storage.log); } std::optional IMergeTreeDataPart::getRelativePathForPrefix(const String & prefix, bool detached, bool broken) const @@ -1492,7 +1674,7 @@ std::optional IMergeTreeDataPart::getRelativePathForPrefix(const String if (detached && parent_part) throw Exception(ErrorCodes::LOGICAL_ERROR, "Cannot detach projection"); - return data_part_storage->getRelativePathForPrefix(storage.log, prefix, detached, broken); + return getDataPartStorage().getRelativePathForPrefix(storage.log, prefix, detached, broken); } std::optional IMergeTreeDataPart::getRelativePathForDetachedPart(const String & prefix, bool broken) const @@ -1507,11 +1689,11 @@ std::optional IMergeTreeDataPart::getRelativePathForDetachedPart(const S return {}; } -void IMergeTreeDataPart::renameToDetached(const String & prefix, DataPartStorageBuilderPtr builder) const +void IMergeTreeDataPart::renameToDetached(const String & prefix) { auto path_to_detach = getRelativePathForDetachedPart(prefix, /* broken */ false); assert(path_to_detach); - renameTo(path_to_detach.value(), true, builder); + renameTo(path_to_detach.value(), true); part_is_probably_removed_from_disk = true; } @@ -1530,7 +1712,7 @@ void IMergeTreeDataPart::makeCloneInDetached(const String & prefix, const Storag if (!maybe_path_in_detached) return; - data_part_storage->freeze( + getDataPartStorage().freeze( storage.relative_data_path, *maybe_path_in_detached, /*make_source_readonly*/ true, @@ -1539,17 +1721,17 @@ void IMergeTreeDataPart::makeCloneInDetached(const String & prefix, const Storag {}); } -DataPartStoragePtr IMergeTreeDataPart::makeCloneOnDisk(const DiskPtr & disk, const String & directory_name) const +MutableDataPartStoragePtr IMergeTreeDataPart::makeCloneOnDisk(const DiskPtr & disk, const String & directory_name) const { assertOnDisk(); - if (disk->getName() == data_part_storage->getDiskName()) - throw Exception("Can not clone data part " + name + " to same disk " + data_part_storage->getDiskName(), ErrorCodes::LOGICAL_ERROR); + if (disk->getName() == getDataPartStorage().getDiskName()) + throw Exception("Can not clone data part " + name + " to same disk " + getDataPartStorage().getDiskName(), ErrorCodes::LOGICAL_ERROR); if (directory_name.empty()) throw Exception("Can not clone data part " + name + " to empty directory.", ErrorCodes::LOGICAL_ERROR); String path_to_clone = fs::path(storage.relative_data_path) / directory_name / ""; - return data_part_storage->clone(path_to_clone, data_part_storage->getPartDirectory(), disk, storage.log); + return getDataPartStorage().clonePart(path_to_clone, getDataPartStorage().getPartDirectory(), disk, storage.log); } void IMergeTreeDataPart::checkConsistencyBase() const @@ -1590,26 +1772,26 @@ void IMergeTreeDataPart::checkConsistencyBase() const } } - data_part_storage->checkConsistency(checksums); + getDataPartStorage().checkConsistency(checksums); } else { auto check_file_not_empty = [this](const String & file_path) { UInt64 file_size; - if (!data_part_storage->exists(file_path) || (file_size = data_part_storage->getFileSize(file_path)) == 0) + if (!getDataPartStorage().exists(file_path) || (file_size = getDataPartStorage().getFileSize(file_path)) == 0) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Part {} is broken: {} is empty", - data_part_storage->getFullPath(), - std::string(fs::path(data_part_storage->getFullPath()) / file_path)); + getDataPartStorage().getFullPath(), + std::string(fs::path(getDataPartStorage().getFullPath()) / file_path)); return file_size; }; /// Check that the primary key index is not empty. if (!pk.column_names.empty()) { - String index_name = "primary" + getIndexExtensionFromFilesystem(data_part_storage).value(); + String index_name = "primary" + getIndexExtensionFromFilesystem(getDataPartStorage()).value(); check_file_not_empty(index_name); } @@ -1753,7 +1935,7 @@ bool IMergeTreeDataPart::checkAllTTLCalculated(const StorageMetadataPtr & metada String IMergeTreeDataPart::getUniqueId() const { - return data_part_storage->getUniqueId(); + return getDataPartStorage().getUniqueId(); } String IMergeTreeDataPart::getZeroLevelPartBlockID(std::string_view token) const @@ -1792,11 +1974,11 @@ IMergeTreeDataPart::uint128 IMergeTreeDataPart::getActualChecksumByFile(const St return it->second.file_hash; } - if (!data_part_storage->exists(file_name)) + if (!getDataPartStorage().exists(file_name)) { return {}; } - std::unique_ptr in_file = data_part_storage->readFile(file_name, {}, std::nullopt, std::nullopt); + std::unique_ptr in_file = getDataPartStorage().readFile(file_name, {}, std::nullopt, std::nullopt); HashingReadBuffer in_hash(*in_file); String value; @@ -1824,11 +2006,11 @@ bool isInMemoryPart(const MergeTreeDataPartPtr & data_part) return (data_part && data_part->getType() == MergeTreeDataPartType::InMemory); } -std::optional getIndexExtensionFromFilesystem(const DataPartStoragePtr & data_part_storage) +std::optional getIndexExtensionFromFilesystem(const IDataPartStorage & data_part_storage) { - if (data_part_storage->exists()) + if (data_part_storage.exists()) { - for (auto it = data_part_storage->iterate(); it->isValid(); it->next()) + for (auto it = data_part_storage.iterate(); it->isValid(); it->next()) { const auto & extension = fs::path(it->name()).extension(); if (extension == getIndexExtension(false) diff --git a/src/Storages/MergeTree/IMergeTreeDataPart.h b/src/Storages/MergeTree/IMergeTreeDataPart.h index 6f034574fb4..6515eb1a65c 100644 --- a/src/Storages/MergeTree/IMergeTreeDataPart.h +++ b/src/Storages/MergeTree/IMergeTreeDataPart.h @@ -1,5 +1,6 @@ #pragma once +#include "IO/WriteSettings.h" #include #include #include @@ -46,7 +47,7 @@ class UncompressedCache; class MergeTreeTransaction; /// Description of the data part. -class IMergeTreeDataPart : public std::enable_shared_from_this +class IMergeTreeDataPart : public std::enable_shared_from_this, public DataPartStorageHolder { public: static constexpr auto DATA_FILE_EXTENSION = ".bin"; @@ -67,19 +68,18 @@ public: using uint128 = IPartMetadataManager::uint128; - IMergeTreeDataPart( const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, Type part_type_, const IMergeTreeDataPart * parent_part_); IMergeTreeDataPart( const MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, Type part_type_, const IMergeTreeDataPart * parent_part_); @@ -94,13 +94,12 @@ public: const ReadBufferFromFileBase::ProfileCallback & profile_callback_) const = 0; virtual MergeTreeWriterPtr getWriter( - DataPartStorageBuilderPtr data_part_storage_builder, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const = 0; + const MergeTreeIndexGranularity & computed_index_granularity) = 0; virtual bool isStoredOnDisk() const = 0; @@ -152,7 +151,7 @@ public: /// Throws an exception if part is not stored in on-disk format. void assertOnDisk() const; - void remove() const; + void remove(); /// Initialize columns (from columns.txt if exists, or create from column files if not). /// Load checksums from checksums.txt if exists. Load index if required. @@ -200,10 +199,6 @@ public: /// processed by multiple shards. UUID uuid = UUIDHelpers::Nil; - /// This is an object which encapsulates all the operations with disk. - /// Contains a path to stored data. - DataPartStoragePtr data_part_storage; - MergeTreeIndexGranularityInfo index_granularity_info; size_t rows_count = 0; @@ -289,8 +284,8 @@ public: using WrittenFiles = std::vector>; - [[nodiscard]] WrittenFiles store(const MergeTreeData & data, const DataPartStorageBuilderPtr & data_part_storage_builder, Checksums & checksums) const; - [[nodiscard]] WrittenFiles store(const Names & column_names, const DataTypes & data_types, const DataPartStorageBuilderPtr & data_part_storage_builder, Checksums & checksums) const; + [[nodiscard]] WrittenFiles store(const MergeTreeData & data, IDataPartStorage & part_storage, Checksums & checksums) const; + [[nodiscard]] WrittenFiles store(const Names & column_names, const DataTypes & data_types, IDataPartStorage & part_storage, Checksums & checksums) const; void update(const Block & block, const Names & column_names); void merge(const MinMaxIndex & other); @@ -321,17 +316,17 @@ public: size_t getFileSizeOrZero(const String & file_name) const; /// Moves a part to detached/ directory and adds prefix to its name - void renameToDetached(const String & prefix, DataPartStorageBuilderPtr builder) const; + void renameToDetached(const String & prefix); /// Makes checks and move part to new directory /// Changes only relative_dir_name, you need to update other metadata (name, is_temp) explicitly - virtual void renameTo(const String & new_relative_path, bool remove_new_dir_if_exists, DataPartStorageBuilderPtr builder) const; + virtual void renameTo(const String & new_relative_path, bool remove_new_dir_if_exists); /// Makes clone of a part in detached/ directory via hard links virtual void makeCloneInDetached(const String & prefix, const StorageMetadataPtr & metadata_snapshot) const; /// Makes full clone of part in specified subdirectory (relative to storage data directory, e.g. "detached") on another disk - DataPartStoragePtr makeCloneOnDisk(const DiskPtr & disk, const String & directory_name) const; + MutableDataPartStoragePtr makeCloneOnDisk(const DiskPtr & disk, const String & directory_name) const; /// Checks that .bin and .mrk files exist. /// @@ -445,6 +440,12 @@ public: /// True if here is lightweight deleted mask file in part. bool hasLightweightDelete() const { return columns.contains(LightweightDeleteDescription::FILTER_COLUMN.name); } + void writeChecksums(const MergeTreeDataPartChecksums & checksums_, const WriteSettings & settings); + + void writeDeleteOnDestroyMarker(); + void removeDeleteOnDestroyMarker(); + void removeVersionMetadata(); + protected: /// Total size of all columns, calculated once in calcuateColumnSizesOnDisk @@ -566,6 +567,12 @@ private: /// any specifial compression. void loadDefaultCompressionCodec(); + void writeColumns(const NamesAndTypesList & columns_, const WriteSettings & settings); + void writeVersionMetadata(const VersionMetadata & version_, bool fsync_part_dir) const; + + template + void writeMetadata(const String & filename, const WriteSettings & settings, Writer && writer); + static void appendFilesOfDefaultCompressionCodec(Strings & files); /// Found column without specific compression and return codec @@ -585,7 +592,7 @@ bool isCompactPart(const MergeTreeDataPartPtr & data_part); bool isWidePart(const MergeTreeDataPartPtr & data_part); bool isInMemoryPart(const MergeTreeDataPartPtr & data_part); inline String getIndexExtension(bool is_compressed_primary_key) { return is_compressed_primary_key ? ".cidx" : ".idx"; } -std::optional getIndexExtensionFromFilesystem(const DataPartStoragePtr & data_part_storage); +std::optional getIndexExtensionFromFilesystem(const IDataPartStorage & data_part_storage); bool isCompressedFromIndexExtension(const String & index_extension); } diff --git a/src/Storages/MergeTree/IMergeTreeDataPartInfoForReader.h b/src/Storages/MergeTree/IMergeTreeDataPartInfoForReader.h index 28f834d661d..2e4972c2788 100644 --- a/src/Storages/MergeTree/IMergeTreeDataPartInfoForReader.h +++ b/src/Storages/MergeTree/IMergeTreeDataPartInfoForReader.h @@ -7,7 +7,8 @@ namespace DB { class IDataPartStorage; -using DataPartStoragePtr = std::shared_ptr; +using DataPartStoragePtr = std::shared_ptr; + class MergeTreeIndexGranularity; struct MergeTreeDataPartChecksums; struct MergeTreeIndexGranularityInfo; @@ -36,7 +37,7 @@ public: virtual bool isProjectionPart() const = 0; - virtual const DataPartStoragePtr & getDataPartStorage() const = 0; + virtual DataPartStoragePtr getDataPartStorage() const = 0; virtual const NamesAndTypesList & getColumns() const = 0; diff --git a/src/Storages/MergeTree/IMergeTreeDataPartWriter.cpp b/src/Storages/MergeTree/IMergeTreeDataPartWriter.cpp index 84d0b50ae2f..2488c63e309 100644 --- a/src/Storages/MergeTree/IMergeTreeDataPartWriter.cpp +++ b/src/Storages/MergeTree/IMergeTreeDataPartWriter.cpp @@ -38,14 +38,12 @@ Block permuteBlockIfNeeded(const Block & block, const IColumn::Permutation * per } IMergeTreeDataPartWriter::IMergeTreeDataPartWriter( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const MergeTreeWriterSettings & settings_, const MergeTreeIndexGranularity & index_granularity_) : data_part(data_part_) - , data_part_storage_builder(std::move(data_part_storage_builder_)) , storage(data_part_->storage) , metadata_snapshot(metadata_snapshot_) , columns_list(columns_list_) diff --git a/src/Storages/MergeTree/IMergeTreeDataPartWriter.h b/src/Storages/MergeTree/IMergeTreeDataPartWriter.h index 417e2713180..fa3c675f7da 100644 --- a/src/Storages/MergeTree/IMergeTreeDataPartWriter.h +++ b/src/Storages/MergeTree/IMergeTreeDataPartWriter.h @@ -22,8 +22,7 @@ class IMergeTreeDataPartWriter : private boost::noncopyable { public: IMergeTreeDataPartWriter( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const MergeTreeWriterSettings & settings_, @@ -42,8 +41,7 @@ public: protected: - const MergeTreeData::DataPartPtr data_part; - DataPartStorageBuilderPtr data_part_storage_builder; + const MergeTreeMutableDataPartPtr data_part; const MergeTreeData & storage; const StorageMetadataPtr metadata_snapshot; const NamesAndTypesList columns_list; diff --git a/src/Storages/MergeTree/IMergedBlockOutputStream.cpp b/src/Storages/MergeTree/IMergedBlockOutputStream.cpp index 5af9bbd3ed8..37da6014d1b 100644 --- a/src/Storages/MergeTree/IMergedBlockOutputStream.cpp +++ b/src/Storages/MergeTree/IMergedBlockOutputStream.cpp @@ -6,14 +6,13 @@ namespace DB { IMergedBlockOutputStream::IMergedBlockOutputStream( - DataPartStorageBuilderPtr data_part_storage_builder_, - const MergeTreeDataPartPtr & data_part, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const NamesAndTypesList & columns_list, bool reset_columns_) : storage(data_part->storage) , metadata_snapshot(metadata_snapshot_) - , data_part_storage_builder(std::move(data_part_storage_builder_)) + , data_part_storage(data_part->getDataPartStoragePtr()) , reset_columns(reset_columns_) { if (reset_columns) diff --git a/src/Storages/MergeTree/IMergedBlockOutputStream.h b/src/Storages/MergeTree/IMergedBlockOutputStream.h index dbcca1443b5..ca4e3899b29 100644 --- a/src/Storages/MergeTree/IMergedBlockOutputStream.h +++ b/src/Storages/MergeTree/IMergedBlockOutputStream.h @@ -1,5 +1,6 @@ #pragma once +#include "Storages/MergeTree/IDataPartStorage.h" #include #include #include @@ -12,8 +13,7 @@ class IMergedBlockOutputStream { public: IMergedBlockOutputStream( - DataPartStorageBuilderPtr data_part_storage_builder_, - const MergeTreeDataPartPtr & data_part, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const NamesAndTypesList & columns_list, bool reset_columns_); @@ -42,7 +42,7 @@ protected: const MergeTreeData & storage; StorageMetadataPtr metadata_snapshot; - DataPartStorageBuilderPtr data_part_storage_builder; + MutableDataPartStoragePtr data_part_storage; IMergeTreeDataPart::MergeTreeWriterPtr writer; bool reset_columns = false; diff --git a/src/Storages/MergeTree/KeyCondition.cpp b/src/Storages/MergeTree/KeyCondition.cpp index ac2fcdb45d4..99c14ede3e2 100644 --- a/src/Storages/MergeTree/KeyCondition.cpp +++ b/src/Storages/MergeTree/KeyCondition.cpp @@ -27,6 +27,7 @@ #include #include #include +#include #include #include @@ -86,6 +87,88 @@ String extractFixedPrefixFromLikePattern(const String & like_pattern) return fixed_prefix; } +/// for "^prefix..." string it returns "prefix" +static String extractFixedPrefixFromRegularExpression(const String & regexp) +{ + if (regexp.size() <= 1 || regexp[0] != '^') + return {}; + + String fixed_prefix; + const char * begin = regexp.data() + 1; + const char * pos = begin; + const char * end = regexp.data() + regexp.size(); + + while (pos != end) + { + switch (*pos) + { + case '\0': + pos = end; + break; + + case '\\': + { + ++pos; + if (pos == end) + break; + + switch (*pos) + { + case '|': + case '(': + case ')': + case '^': + case '$': + case '.': + case '[': + case '?': + case '*': + case '+': + case '{': + fixed_prefix += *pos; + break; + default: + /// all other escape sequences are not supported + pos = end; + break; + } + + ++pos; + break; + } + + /// non-trivial cases + case '|': + fixed_prefix.clear(); + [[fallthrough]]; + case '(': + case '[': + case '^': + case '$': + case '.': + case '+': + pos = end; + break; + + /// Quantifiers that allow a zero number of occurrences. + case '{': + case '?': + case '*': + if (!fixed_prefix.empty()) + fixed_prefix.pop_back(); + + pos = end; + break; + default: + fixed_prefix += *pos; + pos++; + break; + } + } + + return fixed_prefix; +} + /** For a given string, get a minimum string that is strictly greater than all strings with this prefix, * or return an empty string if there are no such strings. @@ -112,289 +195,6 @@ static String firstStringThatIsGreaterThanAllStringsWithPrefix(const String & pr return res; } -static void appendColumnNameWithoutAlias(const ActionsDAG::Node & node, WriteBuffer & out, bool legacy = false) -{ - switch (node.type) - { - case (ActionsDAG::ActionType::INPUT): - writeString(node.result_name, out); - break; - case (ActionsDAG::ActionType::COLUMN): - { - /// If it was created from ASTLiteral, then result_name can be an alias. - /// We need to convert value back to string here. - if (const auto * column_const = typeid_cast(node.column.get())) - writeString(applyVisitor(FieldVisitorToString(), column_const->getField()), out); - /// It may be possible that column is ColumnSet - else - writeString(node.result_name, out); - break; - } - case (ActionsDAG::ActionType::ALIAS): - appendColumnNameWithoutAlias(*node.children.front(), out, legacy); - break; - case (ActionsDAG::ActionType::ARRAY_JOIN): - writeCString("arrayJoin(", out); - appendColumnNameWithoutAlias(*node.children.front(), out, legacy); - writeChar(')', out); - break; - case (ActionsDAG::ActionType::FUNCTION): - { - auto name = node.function_base->getName(); - if (legacy && name == "modulo") - writeCString("moduleLegacy", out); - else - writeString(name, out); - - writeChar('(', out); - bool first = true; - for (const auto * arg : node.children) - { - if (!first) - writeCString(", ", out); - first = false; - - appendColumnNameWithoutAlias(*arg, out, legacy); - } - writeChar(')', out); - } - } -} - -static std::string getColumnNameWithoutAlias(const ActionsDAG::Node & node, bool legacy = false) -{ - WriteBufferFromOwnString out; - appendColumnNameWithoutAlias(node, out, legacy); - return std::move(out.str()); -} - -class KeyCondition::Tree -{ -public: - explicit Tree(const IAST * ast_) : ast(ast_) { assert(ast); } - explicit Tree(const ActionsDAG::Node * dag_) : dag(dag_) { assert(dag); } - - std::string getColumnName() const - { - if (ast) - return ast->getColumnNameWithoutAlias(); - else - return getColumnNameWithoutAlias(*dag); - } - - std::string getColumnNameLegacy() const - { - if (ast) - { - auto adjusted_ast = ast->clone(); - KeyDescription::moduloToModuloLegacyRecursive(adjusted_ast); - return adjusted_ast->getColumnNameWithoutAlias(); - } - else - return getColumnNameWithoutAlias(*dag, true); - } - - bool isFunction() const - { - if (ast) - return typeid_cast(ast); - else - return dag->type == ActionsDAG::ActionType::FUNCTION; - } - - bool isConstant() const - { - if (ast) - return typeid_cast(ast); - else - return dag->column && isColumnConst(*dag->column); - } - - ColumnWithTypeAndName getConstant() const - { - if (!isConstant()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "KeyCondition::Tree node is not a constant"); - - ColumnWithTypeAndName res; - - if (ast) - { - const auto * literal = assert_cast(ast); - res.type = applyVisitor(FieldToDataType(), literal->value); - res.column = res.type->createColumnConst(0, literal->value); - - } - else - { - res.type = dag->result_type; - res.column = dag->column; - } - - return res; - } - - bool tryGetConstant(const Block & block_with_constants, Field & out_value, DataTypePtr & out_type) const - { - if (ast) - { - // Constant expr should use alias names if any - String column_name = ast->getColumnName(); - - if (const auto * lit = ast->as()) - { - /// By default block_with_constants has only one column named "_dummy". - /// If block contains only constants it's may not be preprocessed by - // ExpressionAnalyzer, so try to look up in the default column. - if (!block_with_constants.has(column_name)) - column_name = "_dummy"; - - /// Simple literal - out_value = lit->value; - out_type = block_with_constants.getByName(column_name).type; - - /// If constant is not Null, we can assume it's type is not Nullable as well. - if (!out_value.isNull()) - out_type = removeNullable(out_type); - - return true; - } - else if (block_with_constants.has(column_name) && isColumnConst(*block_with_constants.getByName(column_name).column)) - { - /// An expression which is dependent on constants only - const auto & expr_info = block_with_constants.getByName(column_name); - out_value = (*expr_info.column)[0]; - out_type = expr_info.type; - - if (!out_value.isNull()) - out_type = removeNullable(out_type); - - return true; - } - } - else - { - if (dag->column && isColumnConst(*dag->column)) - { - out_value = (*dag->column)[0]; - out_type = dag->result_type; - - if (!out_value.isNull()) - out_type = removeNullable(out_type); - - return true; - } - } - - return false; - } - - ConstSetPtr tryGetPreparedSet( - const PreparedSetsPtr & sets, - const std::vector & indexes_mapping, - const DataTypes & data_types) const - { - if (sets && ast) - { - if (ast->as() || ast->as()) - return sets->get(PreparedSetKey::forSubquery(*ast)); - - /// We have `PreparedSetKey::forLiteral` but it is useless here as we don't have enough information - /// about types in left argument of the IN operator. Instead, we manually iterate through all the sets - /// and find the one for the right arg based on the AST structure (getTreeHash), after that we check - /// that the types it was prepared with are compatible with the types of the primary key. - auto types_match = [&indexes_mapping, &data_types](const SetPtr & candidate_set) - { - assert(indexes_mapping.size() == data_types.size()); - - for (size_t i = 0; i < indexes_mapping.size(); ++i) - { - if (!candidate_set->areTypesEqual(indexes_mapping[i].tuple_index, data_types[i])) - return false; - } - - return true; - }; - - for (const auto & set : sets->getByTreeHash(ast->getTreeHash())) - { - if (types_match(set)) - return set; - } - } - else if (dag->column) - { - const IColumn * col = dag->column.get(); - if (const auto * col_const = typeid_cast(col)) - col = &col_const->getDataColumn(); - - if (const auto * col_set = typeid_cast(col)) - { - auto set = col_set->getData(); - if (set->isCreated()) - return set; - } - } - - return nullptr; - } - - FunctionTree asFunction() const; - -protected: - const IAST * ast = nullptr; - const ActionsDAG::Node * dag = nullptr; -}; - -class KeyCondition::FunctionTree : public KeyCondition::Tree -{ -public: - std::string getFunctionName() const - { - if (ast) - return assert_cast(ast)->name; - else - return dag->function_base->getName(); - } - - size_t numArguments() const - { - if (ast) - { - const auto * func = assert_cast(ast); - return func->arguments ? func->arguments->children.size() : 0; - } - else - return dag->children.size(); - } - - Tree getArgumentAt(size_t idx) const - { - if (ast) - return Tree(assert_cast(ast)->arguments->children[idx].get()); - else - return Tree(dag->children[idx]); - } - -private: - using Tree::Tree; - - friend class Tree; -}; - - -KeyCondition::FunctionTree KeyCondition::Tree::asFunction() const -{ - if (!isFunction()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "KeyCondition::Tree node is not a function"); - - if (ast) - return KeyCondition::FunctionTree(ast); - else - return KeyCondition::FunctionTree(dag); -} - - -/// A dictionary containing actions to the corresponding functions to turn them into `RPNElement` const KeyCondition::AtomMap KeyCondition::atom_map { { @@ -581,6 +381,27 @@ const KeyCondition::AtomMap KeyCondition::atom_map return true; } }, + { + "match", + [] (RPNElement & out, const Field & value) + { + if (value.getType() != Field::Types::String) + return false; + + String prefix = extractFixedPrefixFromRegularExpression(value.get()); + if (prefix.empty()) + return false; + + String right_bound = firstStringThatIsGreaterThanAllStringsWithPrefix(prefix); + + out.function = RPNElement::FUNCTION_IN_RANGE; + out.range = !right_bound.empty() + ? Range(prefix, true, right_bound, false) + : Range::createLeftBounded(prefix, true); + + return true; + } + }, { "isNotNull", [] (RPNElement & out, const Field &) @@ -869,16 +690,17 @@ static NameSet getAllSubexpressionNames(const ExpressionActions & key_expr) KeyCondition::KeyCondition( const ASTPtr & query, const ASTs & additional_filter_asts, - TreeRewriterResultPtr syntax_analyzer_result, - PreparedSetsPtr prepared_sets_, + Block block_with_constants, + PreparedSetsPtr prepared_sets, ContextPtr context, const Names & key_column_names, const ExpressionActionsPtr & key_expr_, + NameSet array_joined_column_names_, bool single_point_, bool strict_) : key_expr(key_expr_) , key_subexpr_names(getAllSubexpressionNames(*key_expr)) - , prepared_sets(prepared_sets_) + , array_joined_column_names(std::move(array_joined_column_names_)) , single_point(single_point_) , strict(strict_) { @@ -889,82 +711,64 @@ KeyCondition::KeyCondition( key_columns[name] = i; } - if (!syntax_analyzer_result) + auto filter_node = buildFilterNode(query, additional_filter_asts); + + if (!filter_node) { rpn.emplace_back(RPNElement::FUNCTION_UNKNOWN); return; } - /** Evaluation of expressions that depend only on constants. - * For the index to be used, if it is written, for example `WHERE Date = toDate(now())`. + /** When non-strictly monotonic functions are employed in functional index (e.g. ORDER BY toStartOfHour(dateTime)), + * the use of NOT operator in predicate will result in the indexing algorithm leave out some data. + * This is caused by rewriting in KeyCondition::tryParseAtomFromAST of relational operators to less strict + * when parsing the AST into internal RPN representation. + * To overcome the problem, before parsing the AST we transform it to its semantically equivalent form where all NOT's + * are pushed down and applied (when possible) to leaf nodes. */ - Block block_with_constants = getBlockWithConstants(query, syntax_analyzer_result, context); + auto inverted_filter_node = cloneASTWithInversionPushDown(filter_node); - if (syntax_analyzer_result) - { - for (const auto & [name, _] : syntax_analyzer_result->array_join_result_to_source) - array_joined_columns.insert(name); - } + RPNBuilder builder( + inverted_filter_node, + std::move(context), + std::move(block_with_constants), + std::move(prepared_sets), + [&](const RPNBuilderTreeNode & node, RPNElement & out) { return extractAtomFromTree(node, out); }); + rpn = std::move(builder).extractRPN(); +} - const ASTSelectQuery & select = query->as(); - - ASTs filters; - if (select.where()) - filters.push_back(select.where()); - - if (select.prewhere()) - filters.push_back(select.prewhere()); - - for (const auto & filter_ast : additional_filter_asts) - filters.push_back(filter_ast); - - if (!filters.empty()) - { - ASTPtr filter_query; - if (filters.size() == 1) - { - filter_query = filters.front(); - } - else - { - auto function = std::make_shared(); - - function->name = "and"; - function->arguments = std::make_shared(); - function->children.push_back(function->arguments); - function->arguments->children = std::move(filters); - - filter_query = function; - } - - /** When non-strictly monotonic functions are employed in functional index (e.g. ORDER BY toStartOfHour(dateTime)), - * the use of NOT operator in predicate will result in the indexing algorithm leave out some data. - * This is caused by rewriting in KeyCondition::tryParseAtomFromAST of relational operators to less strict - * when parsing the AST into internal RPN representation. - * To overcome the problem, before parsing the AST we transform it to its semantically equivalent form where all NOT's - * are pushed down and applied (when possible) to leaf nodes. - */ - auto ast = cloneASTWithInversionPushDown(filter_query); - traverseAST(Tree(ast.get()), context, block_with_constants); - } - else - { - rpn.emplace_back(RPNElement::FUNCTION_UNKNOWN); - } +KeyCondition::KeyCondition( + const SelectQueryInfo & query_info, + ContextPtr context, + const Names & key_column_names, + const ExpressionActionsPtr & key_expr_, + bool single_point_, + bool strict_) + : KeyCondition( + query_info.query, + query_info.filter_asts, + KeyCondition::getBlockWithConstants(query_info.query, query_info.syntax_analyzer_result, context), + query_info.prepared_sets, + context, + key_column_names, + key_expr_, + query_info.syntax_analyzer_result->getArrayJoinSourceNameSet(), + single_point_, + strict_) +{ } KeyCondition::KeyCondition( ActionDAGNodes dag_nodes, - TreeRewriterResultPtr syntax_analyzer_result, - PreparedSetsPtr prepared_sets_, ContextPtr context, const Names & key_column_names, const ExpressionActionsPtr & key_expr_, + NameSet array_joined_column_names_, bool single_point_, bool strict_) : key_expr(key_expr_) , key_subexpr_names(getAllSubexpressionNames(*key_expr)) - , prepared_sets(prepared_sets_) + , array_joined_column_names(std::move(array_joined_column_names_)) , single_point(single_point_) , strict(strict_) { @@ -975,29 +779,23 @@ KeyCondition::KeyCondition( key_columns[name] = i; } - if (!syntax_analyzer_result) + if (dag_nodes.nodes.empty()) { rpn.emplace_back(RPNElement::FUNCTION_UNKNOWN); return; } - for (const auto & [name, _] : syntax_analyzer_result->array_join_result_to_source) - array_joined_columns.insert(name); + auto inverted_dag = cloneASTWithInversionPushDown(std::move(dag_nodes.nodes), context); + assert(inverted_dag->getOutputs().size() == 1); - if (!dag_nodes.nodes.empty()) + const auto * inverted_dag_filter_node = inverted_dag->getOutputs()[0]; + + RPNBuilder builder(inverted_dag_filter_node, context, [&](const RPNBuilderTreeNode & node, RPNElement & out) { - auto inverted_dag = cloneASTWithInversionPushDown(std::move(dag_nodes.nodes), context); + return extractAtomFromTree(node, out); + }); - // std::cerr << "========== inverted dag: " << inverted_dag->dumpDAG() << std::endl; - - Block empty; - for (const auto * node : inverted_dag->getOutputs()) - traverseAST(Tree(node), context, empty); - } - else - { - rpn.emplace_back(RPNElement::FUNCTION_UNKNOWN); - } + rpn = std::move(builder).extractRPN(); } bool KeyCondition::addCondition(const String & column, const Range & range) @@ -1009,12 +807,12 @@ bool KeyCondition::addCondition(const String & column, const Range & range) return true; } -/** Computes value of constant expression and its data type. - * Returns false, if expression isn't constant. - */ bool KeyCondition::getConstant(const ASTPtr & expr, Block & block_with_constants, Field & out_value, DataTypePtr & out_type) { - return Tree(expr.get()).tryGetConstant(block_with_constants, out_value, out_type); + RPNBuilderTreeContext tree_context(nullptr, block_with_constants, nullptr); + RPNBuilderTreeNode node(expr.get(), tree_context); + + return node.tryGetConstant(out_value, out_type); } @@ -1098,39 +896,6 @@ static FieldRef applyFunction(const FunctionBasePtr & func, const DataTypePtr & return {field.columns, field.row_idx, result_idx}; } -void KeyCondition::traverseAST(const Tree & node, ContextPtr context, Block & block_with_constants) -{ - RPNElement element; - - if (node.isFunction()) - { - auto func = node.asFunction(); - if (tryParseLogicalOperatorFromAST(func, element)) - { - size_t num_args = func.numArguments(); - for (size_t i = 0; i < num_args; ++i) - { - traverseAST(func.getArgumentAt(i), context, block_with_constants); - - /** The first part of the condition is for the correct support of `and` and `or` functions of arbitrary arity - * - in this case `n - 1` elements are added (where `n` is the number of arguments). - */ - if (i != 0 || element.function == RPNElement::FUNCTION_NOT) - rpn.emplace_back(element); - } - - return; - } - } - - if (!tryParseAtomFromAST(node, context, block_with_constants, element)) - { - element.function = RPNElement::FUNCTION_UNKNOWN; - } - - rpn.emplace_back(std::move(element)); -} - /** The key functional expression constraint may be inferred from a plain column in the expression. * For example, if the key contains `toStartOfHour(Timestamp)` and query contains `WHERE Timestamp >= now()`, * it can be assumed that if `toStartOfHour()` is monotonic on [now(), inf), the `toStartOfHour(Timestamp) >= toStartOfHour(now())` @@ -1197,7 +962,8 @@ bool KeyCondition::transformConstantWithValidFunctions( if (is_valid_chain) { - auto const_type = cur_node->result_type; + out_type = removeLowCardinality(out_type); + auto const_type = removeLowCardinality(cur_node->result_type); auto const_column = out_type->createColumnConst(1, out_value); auto const_value = (*castColumnAccurateOrNull({const_column, out_type, ""}, const_type))[0]; @@ -1251,7 +1017,7 @@ bool KeyCondition::transformConstantWithValidFunctions( } bool KeyCondition::canConstantBeWrappedByMonotonicFunctions( - const Tree & node, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, Field & out_value, @@ -1259,7 +1025,7 @@ bool KeyCondition::canConstantBeWrappedByMonotonicFunctions( { String expr_name = node.getColumnName(); - if (array_joined_columns.contains(expr_name)) + if (array_joined_column_names.contains(expr_name)) return false; if (!key_subexpr_names.contains(expr_name)) @@ -1286,11 +1052,15 @@ bool KeyCondition::canConstantBeWrappedByMonotonicFunctions( /// Looking for possible transformation of `column = constant` into `partition_expr = function(constant)` bool KeyCondition::canConstantBeWrappedByFunctions( - const Tree & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, Field & out_value, DataTypePtr & out_type) + const RPNBuilderTreeNode & node, + size_t & out_key_column_num, + DataTypePtr & out_key_column_type, + Field & out_value, + DataTypePtr & out_type) { String expr_name = node.getColumnName(); - if (array_joined_columns.contains(expr_name)) + if (array_joined_column_names.contains(expr_name)) return false; if (!key_subexpr_names.contains(expr_name)) @@ -1304,7 +1074,7 @@ bool KeyCondition::canConstantBeWrappedByFunctions( /// The case `f(modulo(...))` for totally monotonic `f ` is considered to be rare. /// /// Note: for negative values, we can filter more partitions then needed. - expr_name = node.getColumnNameLegacy(); + expr_name = node.getColumnNameWithModuloLegacy(); if (!key_subexpr_names.contains(expr_name)) return false; @@ -1321,8 +1091,7 @@ bool KeyCondition::canConstantBeWrappedByFunctions( } bool KeyCondition::tryPrepareSetIndex( - const FunctionTree & func, - ContextPtr context, + const RPNBuilderFunctionTreeNode & func, RPNElement & out, size_t & out_key_column_num) { @@ -1332,13 +1101,12 @@ bool KeyCondition::tryPrepareSetIndex( std::vector indexes_mapping; DataTypes data_types; - auto get_key_tuple_position_mapping = [&](const Tree & node, size_t tuple_index) + auto get_key_tuple_position_mapping = [&](const RPNBuilderTreeNode & node, size_t tuple_index) { MergeTreeSetIndex::KeyTuplePositionMapping index_mapping; index_mapping.tuple_index = tuple_index; DataTypePtr data_type; - if (isKeyPossiblyWrappedByMonotonicFunctions( - node, context, index_mapping.key_index, data_type, index_mapping.functions)) + if (isKeyPossiblyWrappedByMonotonicFunctions(node, index_mapping.key_index, data_type, index_mapping.functions)) { indexes_mapping.push_back(index_mapping); data_types.push_back(data_type); @@ -1352,25 +1120,29 @@ bool KeyCondition::tryPrepareSetIndex( { /// Note: in case of ActionsDAG, tuple may be a constant. /// In this case, there is no keys in tuple. So, we don't have to check it. - auto left_arg_tuple = left_arg.asFunction(); + auto left_arg_tuple = left_arg.toFunctionNode(); if (left_arg_tuple.getFunctionName() == "tuple") { - left_args_count = left_arg_tuple.numArguments(); + left_args_count = left_arg_tuple.getArgumentsSize(); for (size_t i = 0; i < left_args_count; ++i) get_key_tuple_position_mapping(left_arg_tuple.getArgumentAt(i), i); } else + { get_key_tuple_position_mapping(left_arg, 0); + } } else + { get_key_tuple_position_mapping(left_arg, 0); + } if (indexes_mapping.empty()) return false; const auto right_arg = func.getArgumentAt(1); - auto prepared_set = right_arg.tryGetPreparedSet(prepared_sets, indexes_mapping, data_types); + auto prepared_set = right_arg.tryGetPreparedSet(indexes_mapping, data_types); if (!prepared_set) return false; @@ -1464,13 +1236,12 @@ private: bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctions( - const Tree & node, - ContextPtr context, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_res_column_type, MonotonicFunctionsChain & out_functions_chain) { - std::vector chain_not_tested_for_monotonicity; + std::vector chain_not_tested_for_monotonicity; DataTypePtr key_column_type; if (!isKeyPossiblyWrappedByMonotonicFunctionsImpl(node, out_key_column_num, key_column_type, chain_not_tested_for_monotonicity)) @@ -1479,17 +1250,17 @@ bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctions( for (auto it = chain_not_tested_for_monotonicity.rbegin(); it != chain_not_tested_for_monotonicity.rend(); ++it) { auto function = *it; - auto func_builder = FunctionFactory::instance().tryGet(function.getFunctionName(), context); + auto func_builder = FunctionFactory::instance().tryGet(function.getFunctionName(), node.getTreeContext().getQueryContext()); if (!func_builder) return false; ColumnsWithTypeAndName arguments; ColumnWithTypeAndName const_arg; FunctionWithOptionalConstArg::Kind kind = FunctionWithOptionalConstArg::Kind::NO_CONST; - if (function.numArguments() == 2) + if (function.getArgumentsSize() == 2) { if (function.getArgumentAt(0).isConstant()) { - const_arg = function.getArgumentAt(0).getConstant(); + const_arg = function.getArgumentAt(0).getConstantColumn(); arguments.push_back(const_arg); arguments.push_back({ nullptr, key_column_type, "" }); kind = FunctionWithOptionalConstArg::Kind::LEFT_CONST; @@ -1497,7 +1268,7 @@ bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctions( else if (function.getArgumentAt(1).isConstant()) { arguments.push_back({ nullptr, key_column_type, "" }); - const_arg = function.getArgumentAt(1).getConstant(); + const_arg = function.getArgumentAt(1).getConstantColumn(); arguments.push_back(const_arg); kind = FunctionWithOptionalConstArg::Kind::RIGHT_CONST; } @@ -1523,10 +1294,10 @@ bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctions( } bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctionsImpl( - const Tree & node, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, - std::vector & out_functions_chain) + std::vector & out_functions_chain) { /** By itself, the key column can be a functional expression. for example, `intHash32(UserID)`. * Therefore, use the full name of the expression for search. @@ -1536,7 +1307,7 @@ bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctionsImpl( // Key columns should use canonical names for index analysis String name = node.getColumnName(); - if (array_joined_columns.contains(name)) + if (array_joined_column_names.contains(name)) return false; auto it = key_columns.find(name); @@ -1549,37 +1320,39 @@ bool KeyCondition::isKeyPossiblyWrappedByMonotonicFunctionsImpl( if (node.isFunction()) { - auto func = node.asFunction(); + auto function_node = node.toFunctionNode(); - size_t num_args = func.numArguments(); - if (num_args > 2 || num_args == 0) + size_t arguments_size = function_node.getArgumentsSize(); + if (arguments_size > 2 || arguments_size == 0) return false; - out_functions_chain.push_back(func); - bool ret = false; - if (num_args == 2) + out_functions_chain.push_back(function_node); + + bool result = false; + if (arguments_size == 2) { - if (func.getArgumentAt(0).isConstant()) + if (function_node.getArgumentAt(0).isConstant()) { - ret = isKeyPossiblyWrappedByMonotonicFunctionsImpl(func.getArgumentAt(1), out_key_column_num, out_key_column_type, out_functions_chain); + result = isKeyPossiblyWrappedByMonotonicFunctionsImpl(function_node.getArgumentAt(1), out_key_column_num, out_key_column_type, out_functions_chain); } - else if (func.getArgumentAt(1).isConstant()) + else if (function_node.getArgumentAt(1).isConstant()) { - ret = isKeyPossiblyWrappedByMonotonicFunctionsImpl(func.getArgumentAt(0), out_key_column_num, out_key_column_type, out_functions_chain); + result = isKeyPossiblyWrappedByMonotonicFunctionsImpl(function_node.getArgumentAt(0), out_key_column_num, out_key_column_type, out_functions_chain); } } else { - ret = isKeyPossiblyWrappedByMonotonicFunctionsImpl(func.getArgumentAt(0), out_key_column_num, out_key_column_type, out_functions_chain); + result = isKeyPossiblyWrappedByMonotonicFunctionsImpl(function_node.getArgumentAt(0), out_key_column_num, out_key_column_type, out_functions_chain); } - return ret; + + return result; } return false; } -static void castValueToType(const DataTypePtr & desired_type, Field & src_value, const DataTypePtr & src_type, const KeyCondition::Tree & node) +static void castValueToType(const DataTypePtr & desired_type, Field & src_value, const DataTypePtr & src_type, const String & node_column_name) { try { @@ -1589,13 +1362,13 @@ static void castValueToType(const DataTypePtr & desired_type, Field & src_value, { throw Exception("Key expression contains comparison between inconvertible types: " + desired_type->getName() + " and " + src_type->getName() + - " inside " + node.getColumnName(), + " inside " + node_column_name, ErrorCodes::BAD_TYPE_OF_FIELD); } } -bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Block & block_with_constants, RPNElement & out) +bool KeyCondition::extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out) { /** Functions < > = != <= >= in `notIn` isNull isNotNull, where one argument is a constant, and the other is one of columns of key, * or itself, wrapped in a chain of possibly-monotonic functions, @@ -1605,8 +1378,8 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl DataTypePtr const_type; if (node.isFunction()) { - auto func = node.asFunction(); - size_t num_args = func.numArguments(); + auto func = node.toFunctionNode(); + size_t num_args = func.getArgumentsSize(); DataTypePtr key_expr_type; /// Type of expression containing key column size_t key_column_num = -1; /// Number of a key column (inside key_column_names array) @@ -1618,7 +1391,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl if (num_args == 1) { - if (!(isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(0), context, key_column_num, key_expr_type, chain))) + if (!(isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(0), key_column_num, key_expr_type, chain))) return false; if (key_column_num == static_cast(-1)) @@ -1649,7 +1422,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl if (functionIsInOrGlobalInOperator(func_name)) { - if (tryPrepareSetIndex(func, context, out, key_column_num)) + if (tryPrepareSetIndex(func, out, key_column_num)) { key_arg_pos = 0; is_set_const = true; @@ -1657,7 +1430,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl else return false; } - else if (func.getArgumentAt(1).tryGetConstant(block_with_constants, const_value, const_type)) + else if (func.getArgumentAt(1).tryGetConstant(const_value, const_type)) { /// If the const operand is null, the atom will be always false if (const_value.isNull()) @@ -1666,7 +1439,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl return true; } - if (isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(0), context, key_column_num, key_expr_type, chain)) + if (isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(0), key_column_num, key_expr_type, chain)) { key_arg_pos = 0; } @@ -1687,7 +1460,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl else return false; } - else if (func.getArgumentAt(0).tryGetConstant(block_with_constants, const_value, const_type)) + else if (func.getArgumentAt(0).tryGetConstant(const_value, const_type)) { /// If the const operand is null, the atom will be always false if (const_value.isNull()) @@ -1696,7 +1469,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl return true; } - if (isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(1), context, key_column_num, key_expr_type, chain)) + if (isKeyPossiblyWrappedByMonotonicFunctions(func.getArgumentAt(1), key_column_num, key_expr_type, chain)) { key_arg_pos = 1; } @@ -1737,7 +1510,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl else if (func_name == "in" || func_name == "notIn" || func_name == "like" || func_name == "notLike" || func_name == "ilike" || func_name == "notIlike" || - func_name == "startsWith") + func_name == "startsWith" || func_name == "match") { /// "const IN data_column" doesn't make sense (unlike "data_column IN const") return false; @@ -1776,7 +1549,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl if (!const_type->equals(*common_type)) { - castValueToType(common_type, const_value, const_type, node); + castValueToType(common_type, const_value, const_type, node.getColumnName()); // Need to set is_constant_transformed unless we're doing exact conversion if (!key_expr_type_not_null->equals(*common_type)) @@ -1821,7 +1594,7 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl return atom_it->second(out, const_value); } - else if (node.tryGetConstant(block_with_constants, const_value, const_type)) + else if (node.tryGetConstant(const_value, const_type)) { /// For cases where it says, for example, `WHERE 0 AND something` @@ -1844,32 +1617,6 @@ bool KeyCondition::tryParseAtomFromAST(const Tree & node, ContextPtr context, Bl return false; } -bool KeyCondition::tryParseLogicalOperatorFromAST(const FunctionTree & func, RPNElement & out) -{ - /// Functions AND, OR, NOT. - /// Also a special function `indexHint` - works as if instead of calling a function there are just parentheses - /// (or, the same thing - calling the function `and` from one argument). - - if (func.getFunctionName() == "not") - { - if (func.numArguments() != 1) - return false; - - out.function = RPNElement::FUNCTION_NOT; - } - else - { - if (func.getFunctionName() == "and" || func.getFunctionName() == "indexHint") - out.function = RPNElement::FUNCTION_AND; - else if (func.getFunctionName() == "or") - out.function = RPNElement::FUNCTION_OR; - else - return false; - } - - return true; -} - String KeyCondition::toString() const { String res; diff --git a/src/Storages/MergeTree/KeyCondition.h b/src/Storages/MergeTree/KeyCondition.h index d00a25a1077..fe1bffa9305 100644 --- a/src/Storages/MergeTree/KeyCondition.h +++ b/src/Storages/MergeTree/KeyCondition.h @@ -2,11 +2,16 @@ #include -#include #include -#include -#include +#include + +#include +#include +#include + +#include +#include namespace DB { @@ -205,45 +210,37 @@ public: class KeyCondition { public: - /// Does not take into account the SAMPLE section. all_columns - the set of all columns of the table. + /// Construct key condition from AST SELECT query WHERE, PREWHERE and additional filters KeyCondition( const ASTPtr & query, const ASTs & additional_filter_asts, - TreeRewriterResultPtr syntax_analyzer_result, + Block block_with_constants, PreparedSetsPtr prepared_sets_, ContextPtr context, const Names & key_column_names, const ExpressionActionsPtr & key_expr, + NameSet array_joined_column_names, bool single_point_ = false, bool strict_ = false); + /** Construct key condition from AST SELECT query WHERE, PREWHERE and additional filters. + * Select query, additional filters, prepared sets are initialized using query info. + */ KeyCondition( const SelectQueryInfo & query_info, ContextPtr context, const Names & key_column_names, const ExpressionActionsPtr & key_expr_, bool single_point_ = false, - bool strict_ = false) - : KeyCondition( - query_info.query, - query_info.filter_asts, - query_info.syntax_analyzer_result, - query_info.prepared_sets, - context, - key_column_names, - key_expr_, - single_point_, - strict_) - { - } + bool strict_ = false); + /// Construct key condition from ActionsDAG nodes KeyCondition( ActionDAGNodes dag_nodes, - TreeRewriterResultPtr syntax_analyzer_result, - PreparedSetsPtr prepared_sets_, ContextPtr context, const Names & key_column_names, const ExpressionActionsPtr & key_expr, + NameSet array_joined_column_names, bool single_point_ = false, bool strict_ = false); @@ -275,6 +272,7 @@ public: /// Checks that the index can not be used /// FUNCTION_UNKNOWN will be AND'ed (if any). bool alwaysUnknownOrTrue() const; + /// Checks that the index can not be used /// Does not allow any FUNCTION_UNKNOWN (will instantly return true). bool anyUnknownOrAlwaysTrue() const; @@ -313,10 +311,18 @@ public: * Returns false, if expression isn't constant. */ static bool getConstant( - const ASTPtr & expr, Block & block_with_constants, Field & out_value, DataTypePtr & out_type); + const ASTPtr & expr, + Block & block_with_constants, + Field & out_value, + DataTypePtr & out_type); + /** Calculate expressions, that depend only on constants. + * For index to work when something like "WHERE Date = toDate(now())" is written. + */ static Block getBlockWithConstants( - const ASTPtr & query, const TreeRewriterResultPtr & syntax_analyzer_result, ContextPtr context); + const ASTPtr & query, + const TreeRewriterResultPtr & syntax_analyzer_result, + ContextPtr context); static std::optional applyMonotonicFunctionsChainToRange( Range key_range, @@ -373,14 +379,11 @@ private: using RPN = std::vector; using ColumnIndices = std::map; - using AtomMap = std::unordered_map; public: + using AtomMap = std::unordered_map; static const AtomMap atom_map; - class Tree; - class FunctionTree; - private: BoolMask checkInRange( size_t used_key_size, @@ -390,9 +393,7 @@ private: bool right_bounded, BoolMask initial_mask) const; - void traverseAST(const Tree & node, ContextPtr context, Block & block_with_constants); - bool tryParseAtomFromAST(const Tree & node, ContextPtr context, Block & block_with_constants, RPNElement & out); - static bool tryParseLogicalOperatorFromAST(const FunctionTree & func, RPNElement & out); + bool extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out); /** Is node the key column * or expression in which column of key is wrapped by chain of functions, @@ -401,17 +402,16 @@ private: * and fills chain of possibly-monotonic functions. */ bool isKeyPossiblyWrappedByMonotonicFunctions( - const Tree & node, - ContextPtr context, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_res_column_type, MonotonicFunctionsChain & out_functions_chain); bool isKeyPossiblyWrappedByMonotonicFunctionsImpl( - const Tree & node, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, - std::vector & out_functions_chain); + std::vector & out_functions_chain); bool transformConstantWithValidFunctions( const String & expr_name, @@ -422,21 +422,24 @@ private: std::function always_monotonic) const; bool canConstantBeWrappedByMonotonicFunctions( - const Tree & node, + const RPNBuilderTreeNode & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, Field & out_value, DataTypePtr & out_type); bool canConstantBeWrappedByFunctions( - const Tree & node, size_t & out_key_column_num, DataTypePtr & out_key_column_type, Field & out_value, DataTypePtr & out_type); + const RPNBuilderTreeNode & node, + size_t & out_key_column_num, + DataTypePtr & out_key_column_type, + Field & out_value, + DataTypePtr & out_type); /// If it's possible to make an RPNElement /// that will filter values (possibly tuples) by the content of 'prepared_set', /// do it and return true. bool tryPrepareSetIndex( - const FunctionTree & func, - ContextPtr context, + const RPNBuilderFunctionTreeNode & func, RPNElement & out, size_t & out_key_column_num); @@ -472,11 +475,12 @@ private: /// All intermediate columns are used to calculate key_expr. const NameSet key_subexpr_names; - NameSet array_joined_columns; - PreparedSetsPtr prepared_sets; + /// Array joined column names + NameSet array_joined_column_names; // If true, always allow key_expr to be wrapped by function bool single_point; + // If true, do not use always_monotonic information to transform constants bool strict; }; diff --git a/src/Storages/MergeTree/LoadedMergeTreeDataPartInfoForReader.h b/src/Storages/MergeTree/LoadedMergeTreeDataPartInfoForReader.h index a16aaa728ae..bc786ec0428 100644 --- a/src/Storages/MergeTree/LoadedMergeTreeDataPartInfoForReader.h +++ b/src/Storages/MergeTree/LoadedMergeTreeDataPartInfoForReader.h @@ -12,7 +12,8 @@ public: explicit LoadedMergeTreeDataPartInfoForReader(MergeTreeData::DataPartPtr data_part_) : IMergeTreeDataPartInfoForReader(data_part_->storage.getContext()) , data_part(data_part_) - {} + { + } bool isCompactPart() const override { return DB::isCompactPart(data_part); } @@ -22,7 +23,7 @@ public: bool isProjectionPart() const override { return data_part->isProjectionPart(); } - const DataPartStoragePtr & getDataPartStorage() const override { return data_part->data_part_storage; } + DataPartStoragePtr getDataPartStorage() const override { return data_part->getDataPartStoragePtr(); } const NamesAndTypesList & getColumns() const override { return data_part->getColumns(); } diff --git a/src/Storages/MergeTree/MergeFromLogEntryTask.cpp b/src/Storages/MergeTree/MergeFromLogEntryTask.cpp index 18982c3bbf4..9a9b8a4a6bb 100644 --- a/src/Storages/MergeTree/MergeFromLogEntryTask.cpp +++ b/src/Storages/MergeTree/MergeFromLogEntryTask.cpp @@ -160,7 +160,9 @@ ReplicatedMergeMutateTaskBase::PrepareResult MergeFromLogEntryTask::prepare() for (auto & part_ptr : parts) { ttl_infos.update(part_ptr->ttl_infos); - max_volume_index = std::max(max_volume_index, part_ptr->data_part_storage->getVolumeIndex(*storage.getStoragePolicy())); + auto disk_name = part_ptr->getDataPartStorage().getDiskName(); + size_t volume_index = storage.getStoragePolicy()->getVolumeIndexByDiskName(disk_name); + max_volume_index = std::max(max_volume_index, volume_index); } /// It will live until the whole task is being destroyed @@ -294,12 +296,10 @@ ReplicatedMergeMutateTaskBase::PrepareResult MergeFromLogEntryTask::prepare() bool MergeFromLogEntryTask::finalize(ReplicatedMergeMutateTaskBase::PartLogWriter write_part_log) { part = merge_task->getFuture().get(); - auto builder = merge_task->getBuilder(); /// Task is not needed merge_task.reset(); - - storage.merger_mutator.renameMergedTemporaryPart(part, parts, NO_TRANSACTION_PTR, *transaction_ptr, builder); + storage.merger_mutator.renameMergedTemporaryPart(part, parts, NO_TRANSACTION_PTR, *transaction_ptr); try { diff --git a/src/Storages/MergeTree/MergeList.cpp b/src/Storages/MergeTree/MergeList.cpp index ebe826531d2..02e61a70eb6 100644 --- a/src/Storages/MergeTree/MergeList.cpp +++ b/src/Storages/MergeTree/MergeList.cpp @@ -65,7 +65,7 @@ MergeListElement::MergeListElement( for (const auto & source_part : future_part->parts) { source_part_names.emplace_back(source_part->name); - source_part_paths.emplace_back(source_part->data_part_storage->getFullPath()); + source_part_paths.emplace_back(source_part->getDataPartStorage().getFullPath()); total_size_bytes_compressed += source_part->getBytesOnDisk(); total_size_marks += source_part->getMarksCount(); diff --git a/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp b/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp index 0dcdd927e7b..cc5e87956a1 100644 --- a/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp +++ b/src/Storages/MergeTree/MergePlainMergeTreeTask.cpp @@ -115,10 +115,9 @@ void MergePlainMergeTreeTask::prepare() void MergePlainMergeTreeTask::finish() { new_part = merge_task->getFuture().get(); - auto builder = merge_task->getBuilder(); MergeTreeData::Transaction transaction(storage, txn.get()); - storage.merger_mutator.renameMergedTemporaryPart(new_part, future_part->parts, txn, transaction, builder); + storage.merger_mutator.renameMergedTemporaryPart(new_part, future_part->parts, txn, transaction); transaction.commit(); write_part_log({}); diff --git a/src/Storages/MergeTree/MergeTask.cpp b/src/Storages/MergeTree/MergeTask.cpp index c247d2d2476..0b6fe23e961 100644 --- a/src/Storages/MergeTree/MergeTask.cpp +++ b/src/Storages/MergeTree/MergeTask.cpp @@ -1,3 +1,4 @@ +#include "Storages/MergeTree/IDataPartStorage.h" #include #include @@ -125,23 +126,26 @@ bool MergeTask::ExecuteAndFinalizeHorizontalPart::prepare() ctx->disk = global_ctx->space_reservation->getDisk(); String local_tmp_part_basename = local_tmp_prefix + global_ctx->future_part->name + local_tmp_suffix; + MutableDataPartStoragePtr data_part_storage; - if (global_ctx->parent_path_storage_builder) + if (global_ctx->parent_part) { - global_ctx->data_part_storage_builder = global_ctx->parent_path_storage_builder->getProjection(local_tmp_part_basename); + data_part_storage = global_ctx->parent_part->getDataPartStorage().getProjection(local_tmp_part_basename); } else { auto local_single_disk_volume = std::make_shared("volume_" + global_ctx->future_part->name, ctx->disk, 0); - global_ctx->data_part_storage_builder = std::make_shared( + data_part_storage = std::make_shared( local_single_disk_volume, global_ctx->data->relative_data_path, local_tmp_part_basename); + + data_part_storage->beginTransaction(); } - if (global_ctx->data_part_storage_builder->exists()) - throw Exception("Directory " + global_ctx->data_part_storage_builder->getFullPath() + " already exists", ErrorCodes::DIRECTORY_ALREADY_EXISTS); + if (data_part_storage->exists()) + throw Exception("Directory " + data_part_storage->getFullPath() + " already exists", ErrorCodes::DIRECTORY_ALREADY_EXISTS); if (!global_ctx->parent_part) global_ctx->temporary_directory_lock = global_ctx->data->getTemporaryPartDirectoryHolder(local_tmp_part_basename); @@ -149,7 +153,7 @@ bool MergeTask::ExecuteAndFinalizeHorizontalPart::prepare() global_ctx->all_column_names = global_ctx->metadata_snapshot->getColumns().getNamesOfPhysical(); global_ctx->storage_columns = global_ctx->metadata_snapshot->getColumns().getAllPhysical(); - auto object_columns = MergeTreeData::getObjectColumns(global_ctx->future_part->parts, global_ctx->metadata_snapshot->getColumns()); + auto object_columns = MergeTreeData::getConcreteObjectColumns(global_ctx->future_part->parts, global_ctx->metadata_snapshot->getColumns()); global_ctx->storage_snapshot = std::make_shared(*global_ctx->data, global_ctx->metadata_snapshot, object_columns); extendObjectColumns(global_ctx->storage_columns, object_columns, false); @@ -163,8 +167,6 @@ bool MergeTask::ExecuteAndFinalizeHorizontalPart::prepare() global_ctx->merging_columns, global_ctx->merging_column_names); - auto data_part_storage = global_ctx->data_part_storage_builder->getStorage(); - global_ctx->new_data_part = global_ctx->data->createPart( global_ctx->future_part->name, global_ctx->future_part->type, @@ -302,7 +304,6 @@ bool MergeTask::ExecuteAndFinalizeHorizontalPart::prepare() global_ctx->to = std::make_shared( global_ctx->new_data_part, - global_ctx->data_part_storage_builder, global_ctx->metadata_snapshot, global_ctx->merging_columns, MergeTreeIndexFactory::instance().getMany(global_ctx->metadata_snapshot->getSecondaryIndices()), @@ -501,7 +502,6 @@ void MergeTask::VerticalMergeStage::prepareVerticalMergeForOneColumn() const ctx->executor = std::make_unique(ctx->column_parts_pipeline); ctx->column_to = std::make_unique( - global_ctx->data_part_storage_builder, global_ctx->new_data_part, global_ctx->metadata_snapshot, ctx->executor->getHeader(), @@ -654,7 +654,6 @@ bool MergeTask::MergeProjectionsStage::mergeMinMaxIndexAndPrepareProjections() c global_ctx->deduplicate_by_columns, projection_merging_params, global_ctx->new_data_part.get(), - global_ctx->data_part_storage_builder.get(), ".proj", NO_TRANSACTION_PTR, global_ctx->data, diff --git a/src/Storages/MergeTree/MergeTask.h b/src/Storages/MergeTree/MergeTask.h index 43aba602052..6a29cdbb5ca 100644 --- a/src/Storages/MergeTree/MergeTask.h +++ b/src/Storages/MergeTree/MergeTask.h @@ -59,8 +59,7 @@ public: bool deduplicate_, Names deduplicate_by_columns_, MergeTreeData::MergingParams merging_params_, - const IMergeTreeDataPart * parent_part_, - const IDataPartStorageBuilder * parent_path_storage_builder_, + IMergeTreeDataPart * parent_part_, String suffix_, MergeTreeTransactionPtr txn, MergeTreeData * data_, @@ -82,7 +81,6 @@ public: global_ctx->deduplicate = std::move(deduplicate_); global_ctx->deduplicate_by_columns = std::move(deduplicate_by_columns_); global_ctx->parent_part = std::move(parent_part_); - global_ctx->parent_path_storage_builder = std::move(parent_path_storage_builder_); global_ctx->data = std::move(data_); global_ctx->mutator = std::move(mutator_); global_ctx->merges_blocker = std::move(merges_blocker_); @@ -102,11 +100,6 @@ public: return global_ctx->promise.get_future(); } - DataPartStorageBuilderPtr getBuilder() - { - return global_ctx->data_part_storage_builder; - } - bool execute(); private: @@ -141,8 +134,7 @@ private: StorageMetadataPtr metadata_snapshot{nullptr}; FutureMergedMutatedPartPtr future_part{nullptr}; /// This will be either nullptr or new_data_part, so raw pointer is ok. - const IMergeTreeDataPart * parent_part{nullptr}; - const IDataPartStorageBuilder * parent_path_storage_builder{nullptr}; + IMergeTreeDataPart * parent_part{nullptr}; ContextPtr context{nullptr}; time_t time_of_merge{0}; ReservationSharedPtr space_reservation{nullptr}; @@ -168,7 +160,6 @@ private: std::unique_ptr merging_executor; MergeTreeData::MutableDataPartPtr new_data_part{nullptr}; - DataPartStorageBuilderPtr data_part_storage_builder; /// If lightweight delete mask is present then some input rows are filtered out right after reading. std::shared_ptr> input_rows_filtered{std::make_shared>(0)}; diff --git a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp b/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp index 475407a402b..b63e08b733d 100644 --- a/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp +++ b/src/Storages/MergeTree/MergeTreeBaseSelectProcessor.cpp @@ -607,7 +607,7 @@ Block MergeTreeBaseSelectProcessor::transformHeader( if (!row_level_column.type->canBeUsedInBooleanContext()) { throw Exception("Invalid type for filter in PREWHERE: " + row_level_column.type->getName(), - ErrorCodes::LOGICAL_ERROR); + ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER); } block.erase(prewhere_info->row_level_column_name); @@ -620,7 +620,7 @@ Block MergeTreeBaseSelectProcessor::transformHeader( if (!prewhere_column.type->canBeUsedInBooleanContext()) { throw Exception("Invalid type for filter in PREWHERE: " + prewhere_column.type->getName(), - ErrorCodes::LOGICAL_ERROR); + ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER); } if (prewhere_info->remove_prewhere_column) @@ -628,13 +628,13 @@ Block MergeTreeBaseSelectProcessor::transformHeader( else { WhichDataType which(removeNullable(recursiveRemoveLowCardinality(prewhere_column.type))); - if (which.isInt() || which.isUInt()) + if (which.isNativeInt() || which.isNativeUInt()) prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1u)->convertToFullColumnIfConst(); else if (which.isFloat()) prewhere_column.column = prewhere_column.type->createColumnConst(block.rows(), 1.0f)->convertToFullColumnIfConst(); else - throw Exception("Illegal type " + prewhere_column.type->getName() + " of column for filter.", - ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER); + throw Exception( + ErrorCodes::ILLEGAL_TYPE_OF_COLUMN_FOR_FILTER, "Illegal type {} of column for filter", prewhere_column.type->getName()); } } diff --git a/src/Storages/MergeTree/MergeTreeData.cpp b/src/Storages/MergeTree/MergeTreeData.cpp index 510e8d0da84..83e87a0e462 100644 --- a/src/Storages/MergeTree/MergeTreeData.cpp +++ b/src/Storages/MergeTree/MergeTreeData.cpp @@ -943,8 +943,8 @@ Int64 MergeTreeData::getMaxBlockNumber() const } void MergeTreeData::loadDataPartsFromDisk( - DataPartsVector & broken_parts_to_detach, - DataPartsVector & duplicate_parts_to_remove, + MutableDataPartsVector & broken_parts_to_detach, + MutableDataPartsVector & duplicate_parts_to_remove, ThreadPool & pool, size_t num_parts, std::queue>> & parts_queue, @@ -1082,7 +1082,6 @@ void MergeTreeData::loadDataPartsFromDisk( if (size_of_part.has_value()) part_size_str = formatReadableSizeWithBinarySuffix(*size_of_part); - LOG_ERROR(log, "Detaching broken part {}{} (size: {}). " "If it happened after update, it is likely because of backward incompatibility. " @@ -1200,8 +1199,7 @@ void MergeTreeData::loadDataPartsFromDisk( void MergeTreeData::loadDataPartsFromWAL( - DataPartsVector & /* broken_parts_to_detach */, - DataPartsVector & duplicate_parts_to_remove, + MutableDataPartsVector & duplicate_parts_to_remove, MutableDataPartsVector & parts_from_wal) { for (auto & part : parts_from_wal) @@ -1215,7 +1213,7 @@ void MergeTreeData::loadDataPartsFromWAL( { if ((*it)->checksums.getTotalChecksumHex() == part->checksums.getTotalChecksumHex()) { - LOG_ERROR(log, "Remove duplicate part {}", part->data_part_storage->getFullPath()); + LOG_ERROR(log, "Remove duplicate part {}", part->getDataPartStorage().getFullPath()); duplicate_parts_to_remove.push_back(part); } else @@ -1329,8 +1327,8 @@ void MergeTreeData::loadDataParts(bool skip_sanity_checks) auto part_lock = lockParts(); data_parts_indexes.clear(); - DataPartsVector broken_parts_to_detach; - DataPartsVector duplicate_parts_to_remove; + MutableDataPartsVector broken_parts_to_detach; + MutableDataPartsVector duplicate_parts_to_remove; if (num_parts > 0) loadDataPartsFromDisk( @@ -1384,7 +1382,7 @@ void MergeTreeData::loadDataParts(bool skip_sanity_checks) parts_from_wal.insert( parts_from_wal.end(), std::make_move_iterator(disk_wal_parts.begin()), std::make_move_iterator(disk_wal_parts.end())); - loadDataPartsFromWAL(broken_parts_to_detach, duplicate_parts_to_remove, parts_from_wal); + loadDataPartsFromWAL(duplicate_parts_to_remove, parts_from_wal); num_parts += parts_from_wal.size(); } @@ -1397,11 +1395,7 @@ void MergeTreeData::loadDataParts(bool skip_sanity_checks) } for (auto & part : broken_parts_to_detach) - { - auto builder = part->data_part_storage->getBuilder(); - part->renameToDetached("broken-on-start", builder); /// detached parts must not have '_' in prefixes - builder->commit(); - } + part->renameToDetached("broken-on-start"); /// detached parts must not have '_' in prefixes for (auto & part : duplicate_parts_to_remove) part->remove(); @@ -1689,6 +1683,15 @@ scope_guard MergeTreeData::getTemporaryPartDirectoryHolder(const String & part_d return [this, part_dir_name]() { temporary_parts.remove(part_dir_name); }; } +MergeTreeData::MutableDataPartPtr MergeTreeData::preparePartForRemoval(const DataPartPtr & part) +{ + auto state = part->getState(); + if (state != DataPartState::Deleting && state != DataPartState::DeleteOnDestroy) + throw Exception(ErrorCodes::LOGICAL_ERROR, + "Cannot remove part {}, because it has state: {}", part->name, magic_enum::enum_name(part->getState())); + + return std::const_pointer_cast(part); +} MergeTreeData::DataPartsVector MergeTreeData::grabOldParts(bool force) { @@ -1864,7 +1867,7 @@ void MergeTreeData::flushAllInMemoryPartsIfNeeded() { if (auto part_in_memory = asInMemoryPart(part)) { - part_in_memory->flushToDisk(part_in_memory->data_part_storage->getPartDirectory(), metadata_snapshot); + part_in_memory->flushToDisk(part_in_memory->getDataPartStorage().getPartDirectory(), metadata_snapshot); } } } @@ -1948,7 +1951,7 @@ void MergeTreeData::clearPartsFromFilesystemImpl(const DataPartsVector & parts_t if (thread_group) CurrentThread::attachToIfDetached(thread_group); - part->remove(); + preparePartForRemoval(part)->remove(); if (part_names_succeed) { std::lock_guard lock(part_names_mutex); @@ -1964,7 +1967,7 @@ void MergeTreeData::clearPartsFromFilesystemImpl(const DataPartsVector & parts_t LOG_DEBUG(log, "Removing {} parts from filesystem: {}", parts_to_remove.size(), fmt::join(parts_to_remove, ", ")); for (const DataPartPtr & part : parts_to_remove) { - part->remove(); + preparePartForRemoval(part)->remove(); if (part_names_succeed) part_names_succeed->insert(part->name); } @@ -2144,11 +2147,14 @@ void MergeTreeData::rename(const String & new_table_path, const StorageID & new_ if (!getStorageID().hasUUID()) getContext()->dropCaches(); + /// TODO: remove const_cast for (const auto & part : data_parts_by_info) - part->data_part_storage->changeRootPath(relative_data_path, new_table_path); + { + auto & part_mutable = const_cast(*part); + part_mutable.getDataPartStorage().changeRootPath(relative_data_path, new_table_path); + } relative_data_path = new_table_path; - renameInMemory(new_table_id); } @@ -2166,7 +2172,12 @@ void MergeTreeData::dropAllData() auto lock = lockParts(); - DataPartsVector all_parts(data_parts_by_info.begin(), data_parts_by_info.end()); + DataPartsVector all_parts; + for (auto it = data_parts_by_info.begin(); it != data_parts_by_info.end(); ++it) + { + modifyPartState(it, DataPartState::Deleting); + all_parts.push_back(*it); + } { std::lock_guard wal_lock(write_ahead_log_mutex); @@ -2179,7 +2190,6 @@ void MergeTreeData::dropAllData() if (!getStorageID().hasUUID()) getContext()->dropCaches(); - /// Removing of each data part before recursive removal of directory is to speed-up removal, because there will be less number of syscalls. NameSet part_names_failed; try @@ -2189,6 +2199,7 @@ void MergeTreeData::dropAllData() LOG_TRACE(log, "dropAllData: removing all data parts from memory."); data_parts_indexes.clear(); + all_data_dropped = true; } catch (...) { @@ -2726,7 +2737,7 @@ MergeTreeDataPartType MergeTreeData::choosePartTypeOnDisk(size_t bytes_uncompres MergeTreeData::MutableDataPartPtr MergeTreeData::createPart(const String & name, MergeTreeDataPartType type, const MergeTreePartInfo & part_info, - const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const + const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const { if (type == MergeTreeDataPartType::Compact) return std::make_shared(*this, name, part_info, data_part_storage, parent_part); @@ -2739,17 +2750,17 @@ MergeTreeData::MutableDataPartPtr MergeTreeData::createPart(const String & name, } MergeTreeData::MutableDataPartPtr MergeTreeData::createPart( - const String & name, const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const + const String & name, const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const { return createPart(name, MergeTreePartInfo::fromPartName(name, format_version), data_part_storage, parent_part); } MergeTreeData::MutableDataPartPtr MergeTreeData::createPart( const String & name, const MergeTreePartInfo & part_info, - const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const + const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part) const { MergeTreeDataPartType type; - auto mrk_ext = MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(data_part_storage); + auto mrk_ext = MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(*data_part_storage); if (mrk_ext) { @@ -2943,12 +2954,11 @@ MergeTreeData::DataPartsVector MergeTreeData::getActivePartsToReplace( bool MergeTreeData::renameTempPartAndAdd( MutableDataPartPtr & part, Transaction & out_transaction, - DataPartStorageBuilderPtr builder, DataPartsLock & lock) { DataPartsVector covered_parts; - if (!renameTempPartAndReplaceImpl(part, out_transaction, lock, builder, &covered_parts)) + if (!renameTempPartAndReplaceImpl(part, out_transaction, lock, &covered_parts)) return false; if (!covered_parts.empty()) @@ -2982,32 +2992,31 @@ void MergeTreeData::checkPartCanBeAddedToTable(MutableDataPartPtr & part, DataPa } } -void MergeTreeData::preparePartForCommit(MutableDataPartPtr & part, Transaction & out_transaction, DataPartStorageBuilderPtr builder) +void MergeTreeData::preparePartForCommit(MutableDataPartPtr & part, Transaction & out_transaction) { part->is_temp = false; part->setState(DataPartState::PreActive); assert([&]() { - String dir_name = fs::path(part->data_part_storage->getRelativePath()).filename(); + String dir_name = fs::path(part->getDataPartStorage().getRelativePath()).filename(); bool may_be_cleaned_up = dir_name.starts_with("tmp_") || dir_name.starts_with("tmp-fetch_"); return !may_be_cleaned_up || temporary_parts.contains(dir_name); }()); - part->renameTo(part->name, true, builder); + part->renameTo(part->name, true); data_parts_indexes.insert(part); - out_transaction.addPart(part, builder); + out_transaction.addPart(part); } bool MergeTreeData::renameTempPartAndReplaceImpl( MutableDataPartPtr & part, Transaction & out_transaction, DataPartsLock & lock, - DataPartStorageBuilderPtr builder, DataPartsVector * out_covered_parts) { - LOG_TRACE(log, "Renaming temporary part {} to {}.", part->data_part_storage->getPartDirectory(), part->name); + LOG_TRACE(log, "Renaming temporary part {} to {}.", part->getDataPartStorage().getPartDirectory(), part->name); if (&out_transaction.data != this) throw Exception("MergeTreeData::Transaction for one table cannot be used with another. It is a bug.", @@ -3029,7 +3038,7 @@ bool MergeTreeData::renameTempPartAndReplaceImpl( /// All checks are passed. Now we can rename the part on disk. /// So, we maintain invariant: if a non-temporary part in filesystem then it is in data_parts - preparePartForCommit(part, out_transaction, builder); + preparePartForCommit(part, out_transaction); if (out_covered_parts) { @@ -3045,21 +3054,19 @@ bool MergeTreeData::renameTempPartAndReplaceImpl( MergeTreeData::DataPartsVector MergeTreeData::renameTempPartAndReplaceUnlocked( MutableDataPartPtr & part, Transaction & out_transaction, - DataPartStorageBuilderPtr builder, DataPartsLock & lock) { DataPartsVector covered_parts; - renameTempPartAndReplaceImpl(part, out_transaction, lock, builder, &covered_parts); + renameTempPartAndReplaceImpl(part, out_transaction, lock, &covered_parts); return covered_parts; } MergeTreeData::DataPartsVector MergeTreeData::renameTempPartAndReplace( MutableDataPartPtr & part, - Transaction & out_transaction, - DataPartStorageBuilderPtr builder) + Transaction & out_transaction) { auto part_lock = lockParts(); - return renameTempPartAndReplaceUnlocked(part, out_transaction, builder, part_lock); + return renameTempPartAndReplaceUnlocked(part, out_transaction, part_lock); } void MergeTreeData::removePartsFromWorkingSet(MergeTreeTransaction * txn, const MergeTreeData::DataPartsVector & remove, bool clear_without_timeout, DataPartsLock & acquired_lock) @@ -3136,7 +3143,7 @@ void MergeTreeData::removePartsInRangeFromWorkingSet(MergeTreeTransaction * txn, removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper(txn, drop_range, lock); } -MergeTreeData::DataPartsVector MergeTreeData::removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper( +MergeTreeData::PartsToRemoveFromZooKeeper MergeTreeData::removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper( MergeTreeTransaction * txn, const MergeTreePartInfo & drop_range, DataPartsLock & lock) { DataPartsVector parts_to_remove; @@ -3214,15 +3221,20 @@ MergeTreeData::DataPartsVector MergeTreeData::removePartsInRangeFromWorkingSetAn /// FIXME refactor removePartsFromWorkingSet(...), do not remove parts twice removePartsFromWorkingSet(txn, parts_to_remove, clear_without_timeout, lock); + /// Since we can return parts in Deleting state, we have to use a wrapper that restricts access to such parts. + PartsToRemoveFromZooKeeper parts_to_remove_from_zookeeper; + for (auto & part : parts_to_remove) + parts_to_remove_from_zookeeper.emplace_back(std::move(part)); + for (auto & part : inactive_parts_to_remove_immediately) { if (!drop_range.contains(part->info)) continue; part->remove_time.store(0, std::memory_order_relaxed); - parts_to_remove.push_back(std::move(part)); + parts_to_remove_from_zookeeper.emplace_back(std::move(part), /* was_active */ false); } - return parts_to_remove; + return parts_to_remove_from_zookeeper; } void MergeTreeData::restoreAndActivatePart(const DataPartPtr & part, DataPartsLock * acquired_lock) @@ -3240,9 +3252,9 @@ void MergeTreeData::outdateBrokenPartAndCloneToDetached(const DataPartPtr & part { auto metadata_snapshot = getInMemoryMetadataPtr(); if (prefix.empty()) - LOG_INFO(log, "Cloning part {} to {} and making it obsolete.", part_to_detach->data_part_storage->getPartDirectory(), part_to_detach->name); + LOG_INFO(log, "Cloning part {} to {} and making it obsolete.", part_to_detach->getDataPartStorage().getPartDirectory(), part_to_detach->name); else - LOG_INFO(log, "Cloning part {} to {}_{} and making it obsolete.", part_to_detach->data_part_storage->getPartDirectory(), prefix, part_to_detach->name); + LOG_INFO(log, "Cloning part {} to {}_{} and making it obsolete.", part_to_detach->getDataPartStorage().getPartDirectory(), prefix, part_to_detach->name); part_to_detach->makeCloneInDetached(prefix, metadata_snapshot); @@ -3254,9 +3266,9 @@ void MergeTreeData::outdateBrokenPartAndCloneToDetached(const DataPartPtr & part void MergeTreeData::forcefullyMovePartToDetachedAndRemoveFromMemory(const MergeTreeData::DataPartPtr & part_to_detach, const String & prefix, bool restore_covered) { if (prefix.empty()) - LOG_INFO(log, "Renaming {} to {} and forgetting it.", part_to_detach->data_part_storage->getPartDirectory(), part_to_detach->name); + LOG_INFO(log, "Renaming {} to {} and forgetting it.", part_to_detach->getDataPartStorage().getPartDirectory(), part_to_detach->name); else - LOG_INFO(log, "Renaming {} to {}_{} and forgetting it.", part_to_detach->data_part_storage->getPartDirectory(), prefix, part_to_detach->name); + LOG_INFO(log, "Renaming {} to {}_{} and forgetting it.", part_to_detach->getDataPartStorage().getPartDirectory(), prefix, part_to_detach->name); auto lock = lockParts(); bool removed_active_part = false; @@ -3279,11 +3291,7 @@ void MergeTreeData::forcefullyMovePartToDetachedAndRemoveFromMemory(const MergeT } modifyPartState(it_part, DataPartState::Deleting); - - auto builder = part->data_part_storage->getBuilder(); - part->renameToDetached(prefix, builder); - builder->commit(); - + preparePartForRemoval(part)->renameToDetached(prefix); data_parts_indexes.erase(it_part); if (restore_covered && part->info.level == 0) @@ -3437,7 +3445,7 @@ void MergeTreeData::tryRemovePartImmediately(DataPartPtr && part) try { - part_to_delete->remove(); + preparePartForRemoval(part_to_delete)->remove(); } catch (...) { @@ -3647,9 +3655,9 @@ void MergeTreeData::swapActivePart(MergeTreeData::DataPartPtr part_copy) /// when allow_remote_fs_zero_copy_replication turned on and off again original_active_part->force_keep_shared_data = false; - if (original_active_part->data_part_storage->supportZeroCopyReplication() && - part_copy->data_part_storage->supportZeroCopyReplication() && - original_active_part->data_part_storage->getUniqueId() == part_copy->data_part_storage->getUniqueId()) + if (original_active_part->getDataPartStorage().supportZeroCopyReplication() && + part_copy->getDataPartStorage().supportZeroCopyReplication() && + original_active_part->getDataPartStorage().getUniqueId() == part_copy->getDataPartStorage().getUniqueId()) { /// May be when several volumes use the same S3/HDFS storage original_active_part->force_keep_shared_data = true; @@ -3669,7 +3677,7 @@ void MergeTreeData::swapActivePart(MergeTreeData::DataPartPtr part_copy) /// All other locks are taken in StorageReplicatedMergeTree lockSharedData(*part_copy); - original_active_part->data_part_storage->writeDeleteOnDestroyMarker(log); + preparePartForRemoval(original_active_part)->writeDeleteOnDestroyMarker(); return; } } @@ -3803,9 +3811,9 @@ MergeTreeData::DataPartPtr MergeTreeData::getPartIfExists(const String & part_na static void loadPartAndFixMetadataImpl(MergeTreeData::MutableDataPartPtr part) { part->loadColumnsChecksumsIndexes(false, true); - part->modification_time = part->data_part_storage->getLastModified().epochTime(); - part->data_part_storage->removeDeleteOnDestroyMarker(); - part->data_part_storage->removeVersionMetadata(); + part->modification_time = part->getDataPartStorage().getLastModified().epochTime(); + part->removeDeleteOnDestroyMarker(); + part->removeVersionMetadata(); } void MergeTreeData::calculateColumnAndSecondaryIndexSizesImpl() @@ -3965,7 +3973,7 @@ void MergeTreeData::movePartitionToDisk(const ASTPtr & partition, const String & auto disk = getStoragePolicy()->getDiskByName(name); std::erase_if(parts, [&](auto part_ptr) { - return part_ptr->data_part_storage->getDiskName() == disk->getName(); + return part_ptr->getDataPartStorage().getDiskName() == disk->getName(); }); if (parts.empty()) @@ -4015,7 +4023,7 @@ void MergeTreeData::movePartitionToVolume(const ASTPtr & partition, const String { for (const auto & disk : volume->getDisks()) { - if (part_ptr->data_part_storage->getDiskName() == disk->getName()) + if (part_ptr->getDataPartStorage().getDiskName() == disk->getName()) { return true; } @@ -4212,7 +4220,7 @@ BackupEntries MergeTreeData::backupParts(const DataPartsVector & data_parts, con make_temporary_hard_links = false; hold_storage_and_part_ptrs = true; } - else if (supportsReplication() && part->data_part_storage->supportZeroCopyReplication() && getSettings()->allow_remote_fs_zero_copy_replication) + else if (supportsReplication() && part->getDataPartStorage().supportZeroCopyReplication() && getSettings()->allow_remote_fs_zero_copy_replication) { /// Hard links don't work correctly with zero copy replication. make_temporary_hard_links = false; @@ -4224,7 +4232,7 @@ BackupEntries MergeTreeData::backupParts(const DataPartsVector & data_parts, con table_lock = lockForShare(local_context->getCurrentQueryId(), local_context->getSettingsRef().lock_acquire_timeout); BackupEntries backup_entries_from_part; - part->data_part_storage->backup( + part->getDataPartStorage().backup( part->checksums, part->getFileNamesWithoutChecksums(), data_path_in_backup, @@ -4235,7 +4243,7 @@ BackupEntries MergeTreeData::backupParts(const DataPartsVector & data_parts, con auto projection_parts = part->getProjectionParts(); for (const auto & [projection_name, projection_part] : projection_parts) { - projection_part->data_part_storage->backup( + projection_part->getDataPartStorage().backup( projection_part->checksums, projection_part->getFileNamesWithoutChecksums(), fs::path{data_path_in_backup} / part->name, @@ -4911,22 +4919,16 @@ ReservationPtr MergeTreeData::reserveSpace(UInt64 expected_size, SpacePtr space) return checkAndReturnReservation(expected_size, std::move(reservation)); } -ReservationPtr MergeTreeData::reserveSpace(UInt64 expected_size, const DataPartStoragePtr & data_part_storage) +ReservationPtr MergeTreeData::reserveSpace(UInt64 expected_size, const IDataPartStorage & data_part_storage) { expected_size = std::max(RESERVATION_MIN_ESTIMATION_SIZE, expected_size); - return data_part_storage->reserve(expected_size); + return data_part_storage.reserve(expected_size); } -ReservationPtr MergeTreeData::reserveSpace(UInt64 expected_size, const DataPartStorageBuilderPtr & data_part_storage_builder) +ReservationPtr MergeTreeData::tryReserveSpace(UInt64 expected_size, const IDataPartStorage & data_part_storage) { expected_size = std::max(RESERVATION_MIN_ESTIMATION_SIZE, expected_size); - return data_part_storage_builder->reserve(expected_size); -} - -ReservationPtr MergeTreeData::tryReserveSpace(UInt64 expected_size, const DataPartStoragePtr & data_part_storage) -{ - expected_size = std::max(RESERVATION_MIN_ESTIMATION_SIZE, expected_size); - return data_part_storage->tryReserve(expected_size); + return data_part_storage.tryReserve(expected_size); } ReservationPtr MergeTreeData::tryReserveSpace(UInt64 expected_size, SpacePtr space) @@ -5063,7 +5065,7 @@ bool MergeTreeData::shouldPerformTTLMoveOnInsert(const SpacePtr & move_destinati if (move_destination->isDisk()) { auto disk = std::static_pointer_cast(move_destination); - if (auto volume = getStoragePolicy()->tryGetVolumeByDisk(disk)) + if (auto volume = getStoragePolicy()->tryGetVolumeByDiskName(disk->getName())) return volume->perform_ttl_move_on_insert; } return false; @@ -5075,11 +5077,11 @@ bool MergeTreeData::isPartInTTLDestination(const TTLDescription & ttl, const IMe if (ttl.destination_type == DataDestinationType::VOLUME) { for (const auto & disk : policy->getVolumeByName(ttl.destination_name)->getDisks()) - if (disk->getName() == part.data_part_storage->getDiskName()) + if (disk->getName() == part.getDataPartStorage().getDiskName()) return true; } else if (ttl.destination_type == DataDestinationType::DISK) - return policy->getDiskByName(ttl.destination_name)->getName() == part.data_part_storage->getDiskName(); + return policy->getDiskByName(ttl.destination_name)->getName() == part.getDataPartStorage().getDiskName(); return false; } @@ -5151,7 +5153,7 @@ void MergeTreeData::Transaction::rollbackPartsToTemporaryState() WriteBufferFromOwnString buf; buf << " Rollbacking parts state to temporary and removing from working set:"; for (const auto & part : precommitted_parts) - buf << " " << part->data_part_storage->getPartDirectory(); + buf << " " << part->getDataPartStorage().getPartDirectory(); buf << "."; LOG_DEBUG(data.log, "Undoing transaction.{}", buf.str()); @@ -5162,12 +5164,11 @@ void MergeTreeData::Transaction::rollbackPartsToTemporaryState() clear(); } -void MergeTreeData::Transaction::addPart(MutableDataPartPtr & part, DataPartStorageBuilderPtr builder) +void MergeTreeData::Transaction::addPart(MutableDataPartPtr & part) { precommitted_parts.insert(part); if (asInMemoryPart(part)) has_in_memory_parts = true; - part_builders.push_back(builder); } void MergeTreeData::Transaction::rollback() @@ -5177,13 +5178,31 @@ void MergeTreeData::Transaction::rollback() WriteBufferFromOwnString buf; buf << " Removing parts:"; for (const auto & part : precommitted_parts) - buf << " " << part->data_part_storage->getPartDirectory(); + buf << " " << part->getDataPartStorage().getPartDirectory(); buf << "."; LOG_DEBUG(data.log, "Undoing transaction.{}", buf.str()); - data.removePartsFromWorkingSet(txn, - DataPartsVector(precommitted_parts.begin(), precommitted_parts.end()), - /* clear_without_timeout = */ true); + auto lock = data.lockParts(); + + if (data.data_parts_indexes.empty()) + { + /// Table was dropped concurrently and all parts (including PreActive parts) were cleared, so there's nothing to rollback + if (!data.all_data_dropped) + { + Strings part_names; + for (const auto & part : precommitted_parts) + part_names.emplace_back(part->name); + throw Exception(ErrorCodes::LOGICAL_ERROR, "There are some PreActive parts ({}) to rollback, " + "but data parts set is empty and table {} was not dropped. It's a bug", + fmt::join(part_names, ", "), data.getStorageID().getNameForLogs()); + } + } + else + { + data.removePartsFromWorkingSet(txn, + DataPartsVector(precommitted_parts.begin(), precommitted_parts.end()), + /* clear_without_timeout = */ true, &lock); + } } clear(); @@ -5205,8 +5224,9 @@ MergeTreeData::DataPartsVector MergeTreeData::Transaction::commit(MergeTreeData: auto parts_lock = acquired_parts_lock ? MergeTreeData::DataPartsLock() : data.lockParts(); auto * owing_parts_lock = acquired_parts_lock ? acquired_parts_lock : &parts_lock; - for (auto & builder : part_builders) - builder->commit(); + for (const auto & part : precommitted_parts) + if (part->getDataPartStorage().hasActiveTransaction()) + part->getDataPartStorage().commitTransaction(); bool commit_to_wal = has_in_memory_parts && settings->in_memory_parts_enable_wal; if (txn || commit_to_wal) @@ -5215,7 +5235,7 @@ MergeTreeData::DataPartsVector MergeTreeData::Transaction::commit(MergeTreeData: if (commit_to_wal) wal = data.getWriteAheadLog(); - for (const DataPartPtr & part : precommitted_parts) + for (const auto & part : precommitted_parts) { if (txn) { @@ -5240,7 +5260,7 @@ MergeTreeData::DataPartsVector MergeTreeData::Transaction::commit(MergeTreeData: size_t reduce_rows = 0; size_t reduce_parts = 0; - for (const DataPartPtr & part : precommitted_parts) + for (const auto & part : precommitted_parts) { DataPartPtr covering_part; DataPartsVector covered_parts = data.getActivePartsToReplace(part->info, part->name, covering_part, *owing_parts_lock); @@ -6232,7 +6252,7 @@ std::pair MergeTreeData::cloneAn bool does_storage_policy_allow_same_disk = false; for (const DiskPtr & disk : getStoragePolicy()->getDisks()) { - if (disk->getName() == src_part->data_part_storage->getDiskName()) + if (disk->getName() == src_part->getDataPartStorage().getDiskName()) { does_storage_policy_allow_same_disk = true; break; @@ -6242,7 +6262,7 @@ std::pair MergeTreeData::cloneAn throw Exception( ErrorCodes::BAD_ARGUMENTS, "Could not clone and load part {} because disk does not belong to storage policy", - quoteString(src_part->data_part_storage->getFullPath())); + quoteString(src_part->getDataPartStorage().getFullPath())); String dst_part_name = src_part->getNewName(dst_part_info); assert(!tmp_part_prefix.empty()); @@ -6250,9 +6270,8 @@ std::pair MergeTreeData::cloneAn auto temporary_directory_lock = getTemporaryPartDirectoryHolder(tmp_dst_part_name); /// Why it is needed if we only hardlink files? - auto reservation = src_part->data_part_storage->reserve(src_part->getBytesOnDisk()); - - auto src_part_storage = src_part->data_part_storage; + auto reservation = src_part->getDataPartStorage().reserve(src_part->getBytesOnDisk()); + auto src_part_storage = src_part->getDataPartStoragePtr(); /// If source part is in memory, flush it to disk and clone it already in on-disk format if (auto src_part_in_memory = asInMemoryPart(src_part)) @@ -6279,7 +6298,7 @@ std::pair MergeTreeData::cloneAn hardlinked_files->source_part_name = src_part->name; hardlinked_files->source_table_shared_id = src_part->storage.getTableSharedID(); - for (auto it = src_part->data_part_storage->iterate(); it->isValid(); it->next()) + for (auto it = src_part->getDataPartStorage().iterate(); it->isValid(); it->next()) { if (!files_to_copy_instead_of_hardlinks.contains(it->name()) && it->name() != IMergeTreeDataPart::DELETE_ON_DESTROY_MARKER_FILE_NAME @@ -6338,14 +6357,14 @@ Strings MergeTreeData::getDataPaths() const void MergeTreeData::reportBrokenPart(MergeTreeData::DataPartPtr & data_part) const { - if (data_part->data_part_storage && data_part->data_part_storage->isBroken()) + if (data_part->getDataPartStorage().isBroken()) { auto parts = getDataPartsForInternalUsage(); - LOG_WARNING(log, "Scanning parts to recover on broken disk {}@{}.", data_part->data_part_storage->getDiskName(), data_part->data_part_storage->getDiskPath()); + LOG_WARNING(log, "Scanning parts to recover on broken disk {}@{}.", data_part->getDataPartStorage().getDiskName(), data_part->getDataPartStorage().getDiskPath()); for (const auto & part : parts) { - if (part->data_part_storage && part->data_part_storage->getDiskName() == data_part->data_part_storage->getDiskName()) + if (part->getDataPartStorage().getDiskName() == data_part->getDataPartStorage().getDiskName()) broken_part_callback(part->name); } } @@ -6436,7 +6455,7 @@ PartitionCommandsResultInfo MergeTreeData::freezePartitionsByMatcher( LOG_DEBUG(log, "Freezing part {} snapshot will be placed at {}", part->name, backup_path); - auto data_part_storage = part->data_part_storage; + auto data_part_storage = part->getDataPartStoragePtr(); String src_part_path = data_part_storage->getRelativePath(); String backup_part_path = fs::path(backup_path) / relative_data_path; if (auto part_in_memory = asInMemoryPart(part)) @@ -6450,12 +6469,12 @@ PartitionCommandsResultInfo MergeTreeData::freezePartitionsByMatcher( // Store metadata for replicated table. // Do nothing for non-replicated. - createAndStoreFreezeMetadata(disk, part, fs::path(backup_part_path) / part->data_part_storage->getPartDirectory()); + createAndStoreFreezeMetadata(disk, part, fs::path(backup_part_path) / part->getDataPartStorage().getPartDirectory()); }; auto new_storage = data_part_storage->freeze( backup_part_path, - part->data_part_storage->getPartDirectory(), + part->getDataPartStorage().getPartDirectory(), /*make_source_readonly*/ true, callback, /*copy_instead_of_hardlink*/ false, @@ -6577,8 +6596,8 @@ try if (result_part) { - part_log_elem.disk_name = result_part->data_part_storage->getDiskName(); - part_log_elem.path_on_disk = result_part->data_part_storage->getFullPath(); + part_log_elem.disk_name = result_part->getDataPartStorage().getDiskName(); + part_log_elem.path_on_disk = result_part->getDataPartStorage().getFullPath(); part_log_elem.bytes_compressed_on_disk = result_part->getBytesOnDisk(); part_log_elem.rows = result_part->rows_count; part_log_elem.part_type = result_part->getType(); @@ -6734,7 +6753,7 @@ bool MergeTreeData::moveParts(const CurrentlyMovingPartsTaggerPtr & moving_tagge for (const auto & moving_part : moving_tagger->parts_to_move) { Stopwatch stopwatch; - DataPartPtr cloned_part; + MutableDataPartPtr cloned_part; auto write_part_log = [&](const ExecutionStatus & execution_status) { @@ -6997,7 +7016,7 @@ ReservationPtr MergeTreeData::balancedReservation( if (part->isStoredOnDisk() && part->getBytesOnDisk() >= min_bytes_to_rebalance_partition_over_jbod && part_info.partition_id == part->info.partition_id) { - auto name = part->data_part_storage->getDiskName(); + auto name = part->getDataPartStorage().getDiskName(); auto it = disk_occupation.find(name); if (it != disk_occupation.end()) { @@ -7105,18 +7124,18 @@ ReservationPtr MergeTreeData::balancedReservation( return reserved_space; } -ColumnsDescription MergeTreeData::getObjectColumns( +ColumnsDescription MergeTreeData::getConcreteObjectColumns( const DataPartsVector & parts, const ColumnsDescription & storage_columns) { - return DB::getObjectColumns( + return DB::getConcreteObjectColumns( parts.begin(), parts.end(), storage_columns, [](const auto & part) -> const auto & { return part->getColumns(); }); } -ColumnsDescription MergeTreeData::getObjectColumns( +ColumnsDescription MergeTreeData::getConcreteObjectColumns( boost::iterator_range range, const ColumnsDescription & storage_columns) { - return DB::getObjectColumns( + return DB::getConcreteObjectColumns( range.begin(), range.end(), storage_columns, [](const auto & part) -> const auto & { return part->getColumns(); }); } @@ -7125,21 +7144,21 @@ void MergeTreeData::resetObjectColumnsFromActiveParts(const DataPartsLock & /*lo { auto metadata_snapshot = getInMemoryMetadataPtr(); const auto & columns = metadata_snapshot->getColumns(); - if (!hasObjectColumns(columns)) + if (!hasDynamicSubcolumns(columns)) return; auto range = getDataPartsStateRange(DataPartState::Active); - object_columns = getObjectColumns(range, columns); + object_columns = getConcreteObjectColumns(range, columns); } void MergeTreeData::updateObjectColumns(const DataPartPtr & part, const DataPartsLock & /*lock*/) { auto metadata_snapshot = getInMemoryMetadataPtr(); const auto & columns = metadata_snapshot->getColumns(); - if (!hasObjectColumns(columns)) + if (!hasDynamicSubcolumns(columns)) return; - DB::updateObjectColumns(object_columns, part->getColumns()); + DB::updateObjectColumns(object_columns, columns, part->getColumns()); } StorageSnapshotPtr MergeTreeData::getStorageSnapshot(const StorageMetadataPtr & metadata_snapshot, ContextPtr query_context) const diff --git a/src/Storages/MergeTree/MergeTreeData.h b/src/Storages/MergeTree/MergeTreeData.h index c4a5d66ccbe..8bd0fc1f280 100644 --- a/src/Storages/MergeTree/MergeTreeData.h +++ b/src/Storages/MergeTree/MergeTreeData.h @@ -214,6 +214,7 @@ public: }; using DataParts = std::set; + using MutableDataParts = std::set; using DataPartsVector = std::vector; using DataPartsLock = std::unique_lock; @@ -225,15 +226,15 @@ public: /// After this method setColumns must be called MutableDataPartPtr createPart(const String & name, MergeTreeDataPartType type, const MergeTreePartInfo & part_info, - const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; + const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; /// Create part, that already exists on filesystem. /// After this methods 'loadColumnsChecksumsIndexes' must be called. MutableDataPartPtr createPart(const String & name, - const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; + const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; MutableDataPartPtr createPart(const String & name, const MergeTreePartInfo & part_info, - const DataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; + const MutableDataPartStoragePtr & data_part_storage, const IMergeTreeDataPart * parent_part = nullptr) const; /// Auxiliary object to add a set of parts into the working set in two steps: /// * First, as PreActive parts (the parts are ready, but not yet in the active set). @@ -247,7 +248,7 @@ public: DataPartsVector commit(MergeTreeData::DataPartsLock * acquired_parts_lock = nullptr); - void addPart(MutableDataPartPtr & part, DataPartStorageBuilderPtr builder); + void addPart(MutableDataPartPtr & part); void rollback(); @@ -275,9 +276,8 @@ public: MergeTreeData & data; MergeTreeTransaction * txn; - DataParts precommitted_parts; - std::vector part_builders; - DataParts locked_parts; + MutableDataParts precommitted_parts; + MutableDataParts locked_parts; bool has_in_memory_parts = false; void clear(); @@ -414,9 +414,8 @@ public: SelectQueryInfo & info) const override; ReservationPtr reserveSpace(UInt64 expected_size, VolumePtr & volume) const; - static ReservationPtr tryReserveSpace(UInt64 expected_size, const DataPartStoragePtr & data_part_storage); - static ReservationPtr reserveSpace(UInt64 expected_size, const DataPartStoragePtr & data_part_storage); - static ReservationPtr reserveSpace(UInt64 expected_size, const DataPartStorageBuilderPtr & data_part_storage_builder); + static ReservationPtr tryReserveSpace(UInt64 expected_size, const IDataPartStorage & data_part_storage); + static ReservationPtr reserveSpace(UInt64 expected_size, const IDataPartStorage & data_part_storage); static bool partsContainSameProjections(const DataPartPtr & left, const DataPartPtr & right); @@ -555,21 +554,18 @@ public: bool renameTempPartAndAdd( MutableDataPartPtr & part, Transaction & transaction, - DataPartStorageBuilderPtr builder, DataPartsLock & lock); /// The same as renameTempPartAndAdd but the block range of the part can contain existing parts. /// Returns all parts covered by the added part (in ascending order). DataPartsVector renameTempPartAndReplace( MutableDataPartPtr & part, - Transaction & out_transaction, - DataPartStorageBuilderPtr builder); + Transaction & out_transaction); /// Unlocked version of previous one. Useful when added multiple parts with a single lock. DataPartsVector renameTempPartAndReplaceUnlocked( MutableDataPartPtr & part, Transaction & out_transaction, - DataPartStorageBuilderPtr builder, DataPartsLock & lock); /// Remove parts from working set immediately (without wait for background @@ -588,10 +584,33 @@ public: /// Used in REPLACE PARTITION command. void removePartsInRangeFromWorkingSet(MergeTreeTransaction * txn, const MergeTreePartInfo & drop_range, DataPartsLock & lock); + /// This wrapper is required to restrict access to parts in Deleting state + class PartToRemoveFromZooKeeper + { + DataPartPtr part; + bool was_active; + + public: + explicit PartToRemoveFromZooKeeper(DataPartPtr && part_, bool was_active_ = true) + : part(std::move(part_)), was_active(was_active_) + { + } + + /// It's safe to get name of any part + const String & getPartName() const { return part->name; } + + DataPartPtr getPartIfItWasActive() const + { + return was_active ? part : nullptr; + } + }; + + using PartsToRemoveFromZooKeeper = std::vector; + /// Same as above, but also returns list of parts to remove from ZooKeeper. /// It includes parts that have been just removed by these method /// and Outdated parts covered by drop_range that were removed earlier for any reason. - DataPartsVector removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper( + PartsToRemoveFromZooKeeper removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper( MergeTreeTransaction * txn, const MergeTreePartInfo & drop_range, DataPartsLock & lock); /// Restores Outdated part and adds it to working set @@ -644,6 +663,9 @@ public: /// Deletes the data directory and flushes the uncompressed blocks cache and the marks cache. void dropAllData(); + /// This flag is for hardening and assertions. + bool all_data_dropped = false; + /// Drop data directories if they are empty. It is safe to call this method if table creation was unsuccessful. void dropIfEmpty(); @@ -757,10 +779,10 @@ public: return column_sizes; } - const ColumnsDescription & getObjectColumns() const { return object_columns; } + const ColumnsDescription & getConcreteObjectColumns() const { return object_columns; } /// Creates description of columns of data type Object from the range of data parts. - static ColumnsDescription getObjectColumns( + static ColumnsDescription getConcreteObjectColumns( const DataPartsVector & parts, const ColumnsDescription & storage_columns); IndexSizeByName getSecondaryIndexSizes() const override @@ -979,7 +1001,7 @@ public: /// Fetch part only if some replica has it on shared storage like S3 /// Overridden in StorageReplicatedMergeTree - virtual DataPartStoragePtr tryToFetchIfShared(const IMergeTreeDataPart &, const DiskPtr &, const String &) { return nullptr; } + virtual MutableDataPartStoragePtr tryToFetchIfShared(const IMergeTreeDataPart &, const DiskPtr &, const String &) { return nullptr; } /// Check shared data usage on other replicas for detached/freezed part /// Remove local files and remote files if needed @@ -1129,7 +1151,7 @@ protected: } /// Creates description of columns of data type Object from the range of data parts. - static ColumnsDescription getObjectColumns( + static ColumnsDescription getConcreteObjectColumns( boost::iterator_range range, const ColumnsDescription & storage_columns); std::optional totalRowsByPartitionPredicateImpl( @@ -1264,13 +1286,12 @@ protected: static void incrementMergedPartsProfileEvent(MergeTreeDataPartType type); private: - /// Checking that candidate part doesn't break invariants: correct partition and doesn't exist already void checkPartCanBeAddedToTable(MutableDataPartPtr & part, DataPartsLock & lock) const; /// Preparing itself to be committed in memory: fill some fields inside part, add it to data_parts_indexes /// in precommitted state and to transaction - void preparePartForCommit(MutableDataPartPtr & part, Transaction & out_transaction, DataPartStorageBuilderPtr builder); + void preparePartForCommit(MutableDataPartPtr & part, Transaction & out_transaction); /// Low-level method for preparing parts for commit (in-memory). /// FIXME Merge MergeTreeTransaction and Transaction @@ -1278,7 +1299,6 @@ private: MutableDataPartPtr & part, Transaction & out_transaction, DataPartsLock & lock, - DataPartStorageBuilderPtr builder, DataPartsVector * out_covered_parts); /// RAII Wrapper for atomic work with currently moving parts @@ -1334,8 +1354,8 @@ private: virtual std::unique_ptr getDefaultSettings() const = 0; void loadDataPartsFromDisk( - DataPartsVector & broken_parts_to_detach, - DataPartsVector & duplicate_parts_to_remove, + MutableDataPartsVector & broken_parts_to_detach, + MutableDataPartsVector & duplicate_parts_to_remove, ThreadPool & pool, size_t num_parts, std::queue>> & parts_queue, @@ -1343,8 +1363,7 @@ private: const MergeTreeSettingsPtr & settings); void loadDataPartsFromWAL( - DataPartsVector & broken_parts_to_detach, - DataPartsVector & duplicate_parts_to_remove, + MutableDataPartsVector & duplicate_parts_to_remove, MutableDataPartsVector & parts_from_wal); /// Create zero-copy exclusive lock for part and disk. Useful for coordination of @@ -1356,6 +1375,8 @@ private: /// Otherwise, in non-parallel case will break and return. void clearPartsFromFilesystemImpl(const DataPartsVector & parts, NameSet * part_names_succeed); + static MutableDataPartPtr preparePartForRemoval(const DataPartPtr & part); + TemporaryParts temporary_parts; }; diff --git a/src/Storages/MergeTree/MergeTreeDataMergerMutator.cpp b/src/Storages/MergeTree/MergeTreeDataMergerMutator.cpp index b0ef1522685..0b5c5285d15 100644 --- a/src/Storages/MergeTree/MergeTreeDataMergerMutator.cpp +++ b/src/Storages/MergeTree/MergeTreeDataMergerMutator.cpp @@ -483,8 +483,7 @@ MergeTaskPtr MergeTreeDataMergerMutator::mergePartsToTemporaryPart( const Names & deduplicate_by_columns, const MergeTreeData::MergingParams & merging_params, const MergeTreeTransactionPtr & txn, - const IMergeTreeDataPart * parent_part, - const IDataPartStorageBuilder * parent_path_storage_builder, + IMergeTreeDataPart * parent_part, const String & suffix) { return std::make_shared( @@ -499,7 +498,6 @@ MergeTaskPtr MergeTreeDataMergerMutator::mergePartsToTemporaryPart( deduplicate_by_columns, merging_params, parent_part, - parent_path_storage_builder, suffix, txn, &data, @@ -541,8 +539,7 @@ MergeTreeData::DataPartPtr MergeTreeDataMergerMutator::renameMergedTemporaryPart MergeTreeData::MutableDataPartPtr & new_data_part, const MergeTreeData::DataPartsVector & parts, const MergeTreeTransactionPtr & txn, - MergeTreeData::Transaction & out_transaction, - DataPartStorageBuilderPtr builder) + MergeTreeData::Transaction & out_transaction) { /// Some of source parts was possibly created in transaction, so non-transactional merge may break isolation. if (data.transactions_enabled.load(std::memory_order_relaxed) && !txn) @@ -550,7 +547,7 @@ MergeTreeData::DataPartPtr MergeTreeDataMergerMutator::renameMergedTemporaryPart "but transactions were enabled for this table"); /// Rename new part, add to the set and remove original parts. - auto replaced_parts = data.renameTempPartAndReplace(new_data_part, out_transaction, builder); + auto replaced_parts = data.renameTempPartAndReplace(new_data_part, out_transaction); /// Let's check that all original parts have been deleted and only them. if (replaced_parts.size() != parts.size()) diff --git a/src/Storages/MergeTree/MergeTreeDataMergerMutator.h b/src/Storages/MergeTree/MergeTreeDataMergerMutator.h index 14eb82c641c..5d98f526325 100644 --- a/src/Storages/MergeTree/MergeTreeDataMergerMutator.h +++ b/src/Storages/MergeTree/MergeTreeDataMergerMutator.h @@ -113,8 +113,7 @@ public: const Names & deduplicate_by_columns, const MergeTreeData::MergingParams & merging_params, const MergeTreeTransactionPtr & txn, - const IMergeTreeDataPart * parent_part = nullptr, - const IDataPartStorageBuilder * parent_path_storage_builder = nullptr, + IMergeTreeDataPart * parent_part = nullptr, const String & suffix = ""); /// Mutate a single data part with the specified commands. Will create and return a temporary part. @@ -133,8 +132,7 @@ public: MergeTreeData::MutableDataPartPtr & new_data_part, const MergeTreeData::DataPartsVector & parts, const MergeTreeTransactionPtr & txn, - MergeTreeData::Transaction & out_transaction, - DataPartStorageBuilderPtr builder); + MergeTreeData::Transaction & out_transaction); /// The approximate amount of disk space needed for merge or mutation. With a surplus. diff --git a/src/Storages/MergeTree/MergeTreeDataPartCompact.cpp b/src/Storages/MergeTree/MergeTreeDataPartCompact.cpp index 9298e841072..a537b44d9ea 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartCompact.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartCompact.cpp @@ -22,7 +22,7 @@ namespace ErrorCodes MergeTreeDataPartCompact::MergeTreeDataPartCompact( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, data_part_storage_, Type::Compact, parent_part_) { @@ -32,7 +32,7 @@ MergeTreeDataPartCompact::MergeTreeDataPartCompact( const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, info_, data_part_storage_, Type::Compact, parent_part_) { @@ -58,13 +58,12 @@ IMergeTreeDataPart::MergeTreeReaderPtr MergeTreeDataPartCompact::getReader( } IMergeTreeDataPart::MergeTreeWriterPtr MergeTreeDataPartCompact::getWriter( - DataPartStorageBuilderPtr data_part_storage_builder, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const + const MergeTreeIndexGranularity & computed_index_granularity) { NamesAndTypesList ordered_columns_list; std::copy_if(columns_list.begin(), columns_list.end(), std::back_inserter(ordered_columns_list), @@ -75,7 +74,7 @@ IMergeTreeDataPart::MergeTreeWriterPtr MergeTreeDataPartCompact::getWriter( { return *getColumnPosition(lhs.name) < *getColumnPosition(rhs.name); }); return std::make_unique( - shared_from_this(), std::move(data_part_storage_builder), ordered_columns_list, metadata_snapshot, + shared_from_this(), ordered_columns_list, metadata_snapshot, indices_to_recalc, getMarksFileExtension(), default_codec_, writer_settings, computed_index_granularity); } @@ -97,21 +96,21 @@ void MergeTreeDataPartCompact::calculateEachColumnSizes(ColumnSizeByName & /*eac void MergeTreeDataPartCompact::loadIndexGranularityImpl( MergeTreeIndexGranularity & index_granularity_, const MergeTreeIndexGranularityInfo & index_granularity_info_, - size_t columns_count, const DataPartStoragePtr & data_part_storage_) + size_t columns_count, const IDataPartStorage & data_part_storage_) { if (!index_granularity_info_.mark_type.adaptive) throw Exception("MergeTreeDataPartCompact cannot be created with non-adaptive granulary.", ErrorCodes::NOT_IMPLEMENTED); auto marks_file_path = index_granularity_info_.getMarksFilePath("data"); - if (!data_part_storage_->exists(marks_file_path)) + if (!data_part_storage_.exists(marks_file_path)) throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "Marks file '{}' doesn't exist", - std::string(fs::path(data_part_storage_->getFullPath()) / marks_file_path)); + std::string(fs::path(data_part_storage_.getFullPath()) / marks_file_path)); - size_t marks_file_size = data_part_storage_->getFileSize(marks_file_path); + size_t marks_file_size = data_part_storage_.getFileSize(marks_file_path); - std::unique_ptr buffer = data_part_storage_->readFile( + std::unique_ptr buffer = data_part_storage_.readFile( marks_file_path, ReadSettings().adjustBufferSize(marks_file_size), marks_file_size, std::nullopt); std::unique_ptr marks_reader; @@ -140,7 +139,7 @@ void MergeTreeDataPartCompact::loadIndexGranularity() if (columns.empty()) throw Exception("No columns in part " + name, ErrorCodes::NO_FILE_IN_DATA_PART); - loadIndexGranularityImpl(index_granularity, index_granularity_info, columns.size(), data_part_storage); + loadIndexGranularityImpl(index_granularity, index_granularity_info, columns.size(), getDataPartStorage()); } bool MergeTreeDataPartCompact::hasColumnFiles(const NameAndTypePair & column) const @@ -171,12 +170,12 @@ void MergeTreeDataPartCompact::checkConsistency(bool require_part_metadata) cons throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "No marks file checksum for column in part {}", - data_part_storage->getFullPath()); + getDataPartStorage().getFullPath()); if (!checksums.files.contains(DATA_FILE_NAME_WITH_EXTENSION)) throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "No data file checksum for in part {}", - data_part_storage->getFullPath()); + getDataPartStorage().getFullPath()); } } else @@ -184,33 +183,33 @@ void MergeTreeDataPartCompact::checkConsistency(bool require_part_metadata) cons { /// count.txt should be present even in non custom-partitioned parts std::string file_path = "count.txt"; - if (!data_part_storage->exists(file_path) || data_part_storage->getFileSize(file_path) == 0) + if (!getDataPartStorage().exists(file_path) || getDataPartStorage().getFileSize(file_path) == 0) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Part {} is broken: {} is empty", - data_part_storage->getRelativePath(), - std::string(fs::path(data_part_storage->getFullPath()) / file_path)); + getDataPartStorage().getRelativePath(), + std::string(fs::path(getDataPartStorage().getFullPath()) / file_path)); } /// Check that marks are nonempty and have the consistent size with columns number. - if (data_part_storage->exists(mrk_file_name)) + if (getDataPartStorage().exists(mrk_file_name)) { - UInt64 file_size = data_part_storage->getFileSize(mrk_file_name); + UInt64 file_size = getDataPartStorage().getFileSize(mrk_file_name); if (!file_size) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Part {} is broken: {} is empty.", - data_part_storage->getRelativePath(), - std::string(fs::path(data_part_storage->getFullPath()) / mrk_file_name)); + getDataPartStorage().getRelativePath(), + std::string(fs::path(getDataPartStorage().getFullPath()) / mrk_file_name)); UInt64 expected_file_size = index_granularity_info.getMarkSizeInBytes(columns.size()) * index_granularity.getMarksCount(); if (expected_file_size != file_size) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Part {} is broken: bad size of marks file '{}': {}, must be: {}", - data_part_storage->getRelativePath(), - std::string(fs::path(data_part_storage->getFullPath()) / mrk_file_name), + getDataPartStorage().getRelativePath(), + std::string(fs::path(getDataPartStorage().getFullPath()) / mrk_file_name), std::to_string(file_size), std::to_string(expected_file_size)); } } @@ -218,12 +217,12 @@ void MergeTreeDataPartCompact::checkConsistency(bool require_part_metadata) cons bool MergeTreeDataPartCompact::isStoredOnRemoteDisk() const { - return data_part_storage->isStoredOnRemoteDisk(); + return getDataPartStorage().isStoredOnRemoteDisk(); } bool MergeTreeDataPartCompact::isStoredOnRemoteDiskWithZeroCopySupport() const { - return data_part_storage->supportZeroCopyReplication(); + return getDataPartStorage().supportZeroCopyReplication(); } MergeTreeDataPartCompact::~MergeTreeDataPartCompact() diff --git a/src/Storages/MergeTree/MergeTreeDataPartCompact.h b/src/Storages/MergeTree/MergeTreeDataPartCompact.h index d3ac71cb02a..e275c586cb9 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartCompact.h +++ b/src/Storages/MergeTree/MergeTreeDataPartCompact.h @@ -25,13 +25,13 @@ public: const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeDataPartCompact( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeReaderPtr getReader( @@ -45,13 +45,12 @@ public: const ReadBufferFromFileBase::ProfileCallback & profile_callback) const override; MergeTreeWriterPtr getWriter( - DataPartStorageBuilderPtr data_part_storage_builder, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const override; + const MergeTreeIndexGranularity & computed_index_granularity) override; bool isStoredOnDisk() const override { return true; } @@ -68,7 +67,7 @@ public: protected: static void loadIndexGranularityImpl( MergeTreeIndexGranularity & index_granularity_, const MergeTreeIndexGranularityInfo & index_granularity_info_, - size_t columns_count, const DataPartStoragePtr & data_part_storage_); + size_t columns_count, const IDataPartStorage & data_part_storage_); private: void checkConsistency(bool require_part_metadata) const override; diff --git a/src/Storages/MergeTree/MergeTreeDataPartInMemory.cpp b/src/Storages/MergeTree/MergeTreeDataPartInMemory.cpp index 7a3c5f11c81..48b1b6bab60 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartInMemory.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartInMemory.cpp @@ -1,10 +1,12 @@ -#include "MergeTreeDataPartInMemory.h" +#include #include #include #include #include #include +#include #include +#include #include #include #include @@ -21,7 +23,7 @@ namespace ErrorCodes MergeTreeDataPartInMemory::MergeTreeDataPartInMemory( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, data_part_storage_, Type::InMemory, parent_part_) { @@ -32,7 +34,7 @@ MergeTreeDataPartInMemory::MergeTreeDataPartInMemory( const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, info_, data_part_storage_, Type::InMemory, parent_part_) { @@ -56,27 +58,33 @@ IMergeTreeDataPart::MergeTreeReaderPtr MergeTreeDataPartInMemory::getReader( } IMergeTreeDataPart::MergeTreeWriterPtr MergeTreeDataPartInMemory::getWriter( - DataPartStorageBuilderPtr data_part_storage_builder_, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & /* indices_to_recalc */, const CompressionCodecPtr & /* default_codec */, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & /* computed_index_granularity */) const + const MergeTreeIndexGranularity & /* computed_index_granularity */) { - data_part_storage_builder = data_part_storage_builder_; - auto ptr = std::static_pointer_cast(shared_from_this()); + auto ptr = std::static_pointer_cast(shared_from_this()); return std::make_unique( ptr, columns_list, metadata_snapshot, writer_settings); } -DataPartStoragePtr MergeTreeDataPartInMemory::flushToDisk(const String & new_relative_path, const StorageMetadataPtr & metadata_snapshot) const +MutableDataPartStoragePtr MergeTreeDataPartInMemory::flushToDisk(const String & new_relative_path, const StorageMetadataPtr & metadata_snapshot) const { - auto current_full_path = data_part_storage_builder->getFullPath(); - data_part_storage_builder->setRelativePath(new_relative_path); + auto reservation = storage.reserveSpace(block.bytes(), getDataPartStorage()); + VolumePtr volume = storage.getStoragePolicy()->getVolume(0); + VolumePtr data_part_volume = createVolumeFromReservation(reservation, volume); + auto new_data_part_storage = std::make_shared( + data_part_volume, + storage.getRelativeDataPath(), + new_relative_path); + + new_data_part_storage->beginTransaction(); + + auto current_full_path = getDataPartStorage().getFullPath(); auto new_type = storage.choosePartTypeOnDisk(block.bytes(), rows_count); - auto new_data_part_storage = data_part_storage_builder->getStorage(); auto new_data_part = storage.createPart(name, new_type, info, new_data_part_storage); new_data_part->uuid = uuid; @@ -84,50 +92,50 @@ DataPartStoragePtr MergeTreeDataPartInMemory::flushToDisk(const String & new_rel new_data_part->partition.value = partition.value; new_data_part->minmax_idx = minmax_idx; - if (data_part_storage_builder->exists()) + if (new_data_part_storage->exists()) { throw Exception( ErrorCodes::DIRECTORY_ALREADY_EXISTS, "Could not flush part {}. Part in {} already exists", quoteString(current_full_path), - data_part_storage_builder->getFullPath()); + new_data_part_storage->getFullPath()); } - data_part_storage_builder->createDirectories(); + new_data_part_storage->createDirectories(); auto compression_codec = storage.getContext()->chooseCompressionCodec(0, 0); auto indices = MergeTreeIndexFactory::instance().getMany(metadata_snapshot->getSecondaryIndices()); - MergedBlockOutputStream out(new_data_part, data_part_storage_builder, metadata_snapshot, columns, indices, compression_codec, NO_TRANSACTION_PTR); + MergedBlockOutputStream out(new_data_part, metadata_snapshot, columns, indices, compression_codec, NO_TRANSACTION_PTR); out.write(block); const auto & projections = metadata_snapshot->getProjections(); for (const auto & [projection_name, projection] : projection_parts) { if (projections.has(projection_name)) { - auto projection_part_storage_builder = data_part_storage_builder->getProjection(projection_name + ".proj"); - if (projection_part_storage_builder->exists()) + auto projection_part_storage = new_data_part_storage->getProjection(projection_name + ".proj"); + if (projection_part_storage->exists()) { throw Exception( ErrorCodes::DIRECTORY_ALREADY_EXISTS, "Could not flush projection part {}. Projection part in {} already exists", projection_name, - projection_part_storage_builder->getFullPath()); + projection_part_storage->getFullPath()); } auto projection_part = asInMemoryPart(projection); auto projection_type = storage.choosePartTypeOnDisk(projection_part->block.bytes(), rows_count); MergeTreePartInfo projection_info("all", 0, 0, 0); auto projection_data_part - = storage.createPart(projection_name, projection_type, projection_info, projection_part_storage_builder->getStorage(), parent_part); + = storage.createPart(projection_name, projection_type, projection_info, projection_part_storage, parent_part); projection_data_part->is_temp = false; // clean up will be done on parent part projection_data_part->setColumns(projection->getColumns(), {}); - projection_part_storage_builder->createDirectories(); + projection_part_storage->createDirectories(); const auto & desc = projections.get(name); auto projection_compression_codec = storage.getContext()->chooseCompressionCodec(0, 0); auto projection_indices = MergeTreeIndexFactory::instance().getMany(desc.metadata->getSecondaryIndices()); MergedBlockOutputStream projection_out( - projection_data_part, projection_part_storage_builder, desc.metadata, projection_part->columns, projection_indices, + projection_data_part, desc.metadata, projection_part->columns, projection_indices, projection_compression_codec, NO_TRANSACTION_PTR); projection_out.write(projection_part->block); @@ -137,6 +145,7 @@ DataPartStoragePtr MergeTreeDataPartInMemory::flushToDisk(const String & new_rel } out.finalizePart(new_data_part, false); + new_data_part_storage->commitTransaction(); return new_data_part_storage; } @@ -146,12 +155,9 @@ void MergeTreeDataPartInMemory::makeCloneInDetached(const String & prefix, const flushToDisk(detached_path, metadata_snapshot); } -void MergeTreeDataPartInMemory::renameTo(const String & new_relative_path, bool /* remove_new_dir_if_exists */, DataPartStorageBuilderPtr) const +void MergeTreeDataPartInMemory::renameTo(const String & new_relative_path, bool /* remove_new_dir_if_exists */) { - data_part_storage->setRelativePath(new_relative_path); - - if (data_part_storage_builder) - data_part_storage_builder->setRelativePath(new_relative_path); + getDataPartStorage().setRelativePath(new_relative_path); } void MergeTreeDataPartInMemory::calculateEachColumnSizes(ColumnSizeByName & each_columns_size, ColumnSize & total_size) const diff --git a/src/Storages/MergeTree/MergeTreeDataPartInMemory.h b/src/Storages/MergeTree/MergeTreeDataPartInMemory.h index d985c7f055e..e58701b04a1 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartInMemory.h +++ b/src/Storages/MergeTree/MergeTreeDataPartInMemory.h @@ -14,13 +14,13 @@ public: const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeDataPartInMemory( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeReaderPtr getReader( @@ -34,29 +34,27 @@ public: const ReadBufferFromFileBase::ProfileCallback & profile_callback) const override; MergeTreeWriterPtr getWriter( - DataPartStorageBuilderPtr data_part_storage_builder_, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const override; + const MergeTreeIndexGranularity & computed_index_granularity) override; bool isStoredOnDisk() const override { return false; } bool isStoredOnRemoteDisk() const override { return false; } bool isStoredOnRemoteDiskWithZeroCopySupport() const override { return false; } bool hasColumnFiles(const NameAndTypePair & column) const override { return !!getColumnPosition(column.getNameInStorage()); } String getFileNameForColumn(const NameAndTypePair & /* column */) const override { return ""; } - void renameTo(const String & new_relative_path, bool remove_new_dir_if_exists, DataPartStorageBuilderPtr) const override; + void renameTo(const String & new_relative_path, bool remove_new_dir_if_exists) override; void makeCloneInDetached(const String & prefix, const StorageMetadataPtr & metadata_snapshot) const override; - DataPartStoragePtr flushToDisk(const String & new_relative_path, const StorageMetadataPtr & metadata_snapshot) const; + MutableDataPartStoragePtr flushToDisk(const String & new_relative_path, const StorageMetadataPtr & metadata_snapshot) const; /// Returns hash of parts's block Checksum calculateBlockChecksum() const; mutable Block block; - mutable DataPartStorageBuilderPtr data_part_storage_builder; private: mutable std::condition_variable is_merged; @@ -66,6 +64,8 @@ private: }; using DataPartInMemoryPtr = std::shared_ptr; +using MutableDataPartInMemoryPtr = std::shared_ptr; + DataPartInMemoryPtr asInMemoryPart(const MergeTreeDataPartPtr & part); } diff --git a/src/Storages/MergeTree/MergeTreeDataPartWide.cpp b/src/Storages/MergeTree/MergeTreeDataPartWide.cpp index 170d1b1d703..2418960f992 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWide.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartWide.cpp @@ -21,7 +21,7 @@ namespace ErrorCodes MergeTreeDataPartWide::MergeTreeDataPartWide( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, data_part_storage_, Type::Wide, parent_part_) { @@ -31,7 +31,7 @@ MergeTreeDataPartWide::MergeTreeDataPartWide( const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_) : IMergeTreeDataPart(storage_, name_, info_, data_part_storage_, Type::Wide, parent_part_) { @@ -56,17 +56,16 @@ IMergeTreeDataPart::MergeTreeReaderPtr MergeTreeDataPartWide::getReader( } IMergeTreeDataPart::MergeTreeWriterPtr MergeTreeDataPartWide::getWriter( - DataPartStorageBuilderPtr data_part_storage_builder, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const + const MergeTreeIndexGranularity & computed_index_granularity) { return std::make_unique( - shared_from_this(), data_part_storage_builder, - columns_list, metadata_snapshot, indices_to_recalc, + shared_from_this(), columns_list, + metadata_snapshot, indices_to_recalc, getMarksFileExtension(), default_codec_, writer_settings, computed_index_granularity); } @@ -105,18 +104,18 @@ ColumnSize MergeTreeDataPartWide::getColumnSizeImpl( void MergeTreeDataPartWide::loadIndexGranularityImpl( MergeTreeIndexGranularity & index_granularity_, MergeTreeIndexGranularityInfo & index_granularity_info_, - const DataPartStoragePtr & data_part_storage_, const std::string & any_column_file_name) + const IDataPartStorage & data_part_storage_, const std::string & any_column_file_name) { index_granularity_info_.changeGranularityIfRequired(data_part_storage_); /// We can use any column, it doesn't matter std::string marks_file_path = index_granularity_info_.getMarksFilePath(any_column_file_name); - if (!data_part_storage_->exists(marks_file_path)) + if (!data_part_storage_.exists(marks_file_path)) throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "Marks file '{}' doesn't exist", - std::string(fs::path(data_part_storage_->getFullPath()) / marks_file_path)); + std::string(fs::path(data_part_storage_.getFullPath()) / marks_file_path)); - size_t marks_file_size = data_part_storage_->getFileSize(marks_file_path); + size_t marks_file_size = data_part_storage_.getFileSize(marks_file_path); if (!index_granularity_info_.mark_type.adaptive && !index_granularity_info_.mark_type.compressed) { @@ -126,7 +125,7 @@ void MergeTreeDataPartWide::loadIndexGranularityImpl( } else { - auto marks_file = data_part_storage_->readFile(marks_file_path, ReadSettings().adjustBufferSize(marks_file_size), marks_file_size, std::nullopt); + auto marks_file = data_part_storage_.readFile(marks_file_path, ReadSettings().adjustBufferSize(marks_file_size), marks_file_size, std::nullopt); std::unique_ptr marks_reader; if (!index_granularity_info_.mark_type.compressed) @@ -163,18 +162,18 @@ void MergeTreeDataPartWide::loadIndexGranularity() if (columns.empty()) throw Exception("No columns in part " + name, ErrorCodes::NO_FILE_IN_DATA_PART); - loadIndexGranularityImpl(index_granularity, index_granularity_info, data_part_storage, getFileNameForColumn(columns.front())); + loadIndexGranularityImpl(index_granularity, index_granularity_info, getDataPartStorage(), getFileNameForColumn(columns.front())); } bool MergeTreeDataPartWide::isStoredOnRemoteDisk() const { - return data_part_storage->isStoredOnRemoteDisk(); + return getDataPartStorage().isStoredOnRemoteDisk(); } bool MergeTreeDataPartWide::isStoredOnRemoteDiskWithZeroCopySupport() const { - return data_part_storage->supportZeroCopyReplication(); + return getDataPartStorage().supportZeroCopyReplication(); } MergeTreeDataPartWide::~MergeTreeDataPartWide() @@ -203,13 +202,13 @@ void MergeTreeDataPartWide::checkConsistency(bool require_part_metadata) const throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "No {} file checksum for column {} in part {} ", - mrk_file_name, name_type.name, data_part_storage->getFullPath()); + mrk_file_name, name_type.name, getDataPartStorage().getFullPath()); if (!checksums.files.contains(bin_file_name)) throw Exception( ErrorCodes::NO_FILE_IN_DATA_PART, "No {} file checksum for column {} in part ", - bin_file_name, name_type.name, data_part_storage->getFullPath()); + bin_file_name, name_type.name, getDataPartStorage().getFullPath()); }); } } @@ -225,23 +224,23 @@ void MergeTreeDataPartWide::checkConsistency(bool require_part_metadata) const auto file_path = ISerialization::getFileNameForStream(name_type, substream_path) + marks_file_extension; /// Missing file is Ok for case when new column was added. - if (data_part_storage->exists(file_path)) + if (getDataPartStorage().exists(file_path)) { - UInt64 file_size = data_part_storage->getFileSize(file_path); + UInt64 file_size = getDataPartStorage().getFileSize(file_path); if (!file_size) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, "Part {} is broken: {} is empty.", - data_part_storage->getFullPath(), - std::string(fs::path(data_part_storage->getFullPath()) / file_path)); + getDataPartStorage().getFullPath(), + std::string(fs::path(getDataPartStorage().getFullPath()) / file_path)); if (!marks_size) marks_size = file_size; else if (file_size != *marks_size) throw Exception( ErrorCodes::BAD_SIZE_OF_FILE_IN_DATA_PART, - "Part {} is broken: marks have different sizes.", data_part_storage->getFullPath()); + "Part {} is broken: marks have different sizes.", getDataPartStorage().getFullPath()); } }); } diff --git a/src/Storages/MergeTree/MergeTreeDataPartWide.h b/src/Storages/MergeTree/MergeTreeDataPartWide.h index 52afa9e82d4..601bdff51a1 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWide.h +++ b/src/Storages/MergeTree/MergeTreeDataPartWide.h @@ -1,5 +1,6 @@ #pragma once +#include "Storages/MergeTree/IDataPartStorage.h" #include namespace DB @@ -19,13 +20,13 @@ public: const MergeTreeData & storage_, const String & name_, const MergeTreePartInfo & info_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeDataPartWide( MergeTreeData & storage_, const String & name_, - const DataPartStoragePtr & data_part_storage_, + const MutableDataPartStoragePtr & data_part_storage_, const IMergeTreeDataPart * parent_part_ = nullptr); MergeTreeReaderPtr getReader( @@ -39,13 +40,12 @@ public: const ReadBufferFromFileBase::ProfileCallback & profile_callback) const override; MergeTreeWriterPtr getWriter( - DataPartStorageBuilderPtr data_part_storage_builder, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & writer_settings, - const MergeTreeIndexGranularity & computed_index_granularity) const override; + const MergeTreeIndexGranularity & computed_index_granularity) override; bool isStoredOnDisk() const override { return true; } @@ -64,7 +64,7 @@ public: protected: static void loadIndexGranularityImpl( MergeTreeIndexGranularity & index_granularity_, MergeTreeIndexGranularityInfo & index_granularity_info_, - const DataPartStoragePtr & data_part_storage_, const std::string & any_column_file_name); + const IDataPartStorage & data_part_storage_, const std::string & any_column_file_name); private: void checkConsistency(bool require_part_metadata) const override; diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.cpp b/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.cpp index 457aad55023..020121e59d7 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.cpp @@ -10,8 +10,7 @@ namespace ErrorCodes } MergeTreeDataPartWriterCompact::MergeTreeDataPartWriterCompact( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const std::vector & indices_to_recalc_, @@ -19,16 +18,16 @@ MergeTreeDataPartWriterCompact::MergeTreeDataPartWriterCompact( const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & settings_, const MergeTreeIndexGranularity & index_granularity_) - : MergeTreeDataPartWriterOnDisk(data_part_, std::move(data_part_storage_builder_), columns_list_, metadata_snapshot_, + : MergeTreeDataPartWriterOnDisk(data_part_, columns_list_, metadata_snapshot_, indices_to_recalc_, marks_file_extension_, default_codec_, settings_, index_granularity_) - , plain_file(data_part_storage_builder->writeFile( + , plain_file(data_part_->getDataPartStorage().writeFile( MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION, settings.max_compress_block_size, settings_.query_write_settings)) , plain_hashing(*plain_file) { - marks_file = data_part_storage_builder->writeFile( + marks_file = data_part_->getDataPartStorage().writeFile( MergeTreeDataPartCompact::DATA_FILE_NAME + marks_file_extension_, 4096, settings_.query_write_settings); @@ -132,7 +131,7 @@ void writeColumnSingleGranule( serialize_settings.position_independent_encoding = true; //-V1048 serialize_settings.low_cardinality_max_dictionary_size = 0; //-V1048 - serialization->serializeBinaryBulkStatePrefix(serialize_settings, state); + serialization->serializeBinaryBulkStatePrefix(*column.column, serialize_settings, state); serialization->serializeBinaryBulkWithMultipleStreams(*column.column, from_row, number_of_rows, serialize_settings, state); serialization->serializeBinaryBulkStateSuffix(serialize_settings, state); } diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.h b/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.h index 7b68f61925f..06f8122393f 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.h +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterCompact.h @@ -11,8 +11,7 @@ class MergeTreeDataPartWriterCompact : public MergeTreeDataPartWriterOnDisk { public: MergeTreeDataPartWriterCompact( - const MergeTreeData::DataPartPtr & data_part, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot_, const std::vector & indices_to_recalc, diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.cpp b/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.cpp index e1145868ce2..8066a097499 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.cpp @@ -11,11 +11,11 @@ namespace ErrorCodes } MergeTreeDataPartWriterInMemory::MergeTreeDataPartWriterInMemory( - const DataPartInMemoryPtr & part_, + const MutableDataPartInMemoryPtr & part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const MergeTreeWriterSettings & settings_) - : IMergeTreeDataPartWriter(part_, nullptr, columns_list_, metadata_snapshot_, settings_) + : IMergeTreeDataPartWriter(part_, columns_list_, metadata_snapshot_, settings_) , part_in_memory(part_) {} void MergeTreeDataPartWriterInMemory::write( diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.h b/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.h index 233ca81a697..9e1e868beac 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.h +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterInMemory.h @@ -10,7 +10,7 @@ class MergeTreeDataPartWriterInMemory : public IMergeTreeDataPartWriter { public: MergeTreeDataPartWriterInMemory( - const DataPartInMemoryPtr & part_, + const MutableDataPartInMemoryPtr & part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot, const MergeTreeWriterSettings & settings_); @@ -24,7 +24,7 @@ public: private: void calculateAndSerializePrimaryIndex(const Block & primary_index_block); - DataPartInMemoryPtr part_in_memory; + MutableDataPartInMemoryPtr part_in_memory; }; } diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.cpp b/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.cpp index 1d2b095330e..d085bb29b20 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.cpp @@ -48,7 +48,7 @@ void MergeTreeDataPartWriterOnDisk::Stream::sync() const MergeTreeDataPartWriterOnDisk::Stream::Stream( const String & escaped_column_name_, - const DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, const String & data_path_, const std::string & data_file_extension_, const std::string & marks_path_, @@ -61,11 +61,11 @@ MergeTreeDataPartWriterOnDisk::Stream::Stream( escaped_column_name(escaped_column_name_), data_file_extension{data_file_extension_}, marks_file_extension{marks_file_extension_}, - plain_file(data_part_storage_builder->writeFile(data_path_ + data_file_extension, max_compress_block_size_, query_write_settings)), + plain_file(data_part_storage->writeFile(data_path_ + data_file_extension, max_compress_block_size_, query_write_settings)), plain_hashing(*plain_file), compressor(plain_hashing, compression_codec_, max_compress_block_size_), compressed_hashing(compressor), - marks_file(data_part_storage_builder->writeFile(marks_path_ + marks_file_extension, 4096, query_write_settings)), + marks_file(data_part_storage->writeFile(marks_path_ + marks_file_extension, 4096, query_write_settings)), marks_hashing(*marks_file), marks_compressor(marks_hashing, marks_compression_codec_, marks_compress_block_size_), marks_compressed_hashing(marks_compressor), @@ -96,8 +96,7 @@ void MergeTreeDataPartWriterOnDisk::Stream::addToChecksums(MergeTreeData::DataPa MergeTreeDataPartWriterOnDisk::MergeTreeDataPartWriterOnDisk( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const MergeTreeIndices & indices_to_recalc_, @@ -105,8 +104,7 @@ MergeTreeDataPartWriterOnDisk::MergeTreeDataPartWriterOnDisk( const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & settings_, const MergeTreeIndexGranularity & index_granularity_) - : IMergeTreeDataPartWriter(data_part_, std::move(data_part_storage_builder_), - columns_list_, metadata_snapshot_, settings_, index_granularity_) + : IMergeTreeDataPartWriter(data_part_, columns_list_, metadata_snapshot_, settings_, index_granularity_) , skip_indices(indices_to_recalc_) , marks_file_extension(marks_file_extension_) , default_codec(default_codec_) @@ -116,8 +114,8 @@ MergeTreeDataPartWriterOnDisk::MergeTreeDataPartWriterOnDisk( if (settings.blocks_are_granules_size && !index_granularity.empty()) throw Exception("Can't take information about index granularity from blocks, when non empty index_granularity array specified", ErrorCodes::LOGICAL_ERROR); - if (!data_part_storage_builder->exists()) - data_part_storage_builder->createDirectories(); + if (!data_part->getDataPartStorage().exists()) + data_part->getDataPartStorage().createDirectories(); if (settings.rewrite_primary_key) initPrimaryIndex(); @@ -178,7 +176,7 @@ void MergeTreeDataPartWriterOnDisk::initPrimaryIndex() if (metadata_snapshot->hasPrimaryKey()) { String index_name = "primary" + getIndexExtension(compress_primary_key); - index_file_stream = data_part_storage_builder->writeFile(index_name, DBMS_DEFAULT_BUFFER_SIZE, settings.query_write_settings); + index_file_stream = data_part->getDataPartStorage().writeFile(index_name, DBMS_DEFAULT_BUFFER_SIZE, settings.query_write_settings); index_file_hashing_stream = std::make_unique(*index_file_stream); if (compress_primary_key) @@ -204,7 +202,7 @@ void MergeTreeDataPartWriterOnDisk::initSkipIndices() skip_indices_streams.emplace_back( std::make_unique( stream_name, - data_part_storage_builder, + data_part->getDataPartStoragePtr(), stream_name, index_helper->getSerializedFileExtension(), stream_name, marks_file_extension, default_codec, settings.max_compress_block_size, diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.h b/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.h index 4b58224de78..ab1adfe7f59 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.h +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterOnDisk.h @@ -50,7 +50,7 @@ public: { Stream( const String & escaped_column_name_, - const DataPartStorageBuilderPtr & data_part_storage_builder, + const MutableDataPartStoragePtr & data_part_storage, const String & data_path_, const std::string & data_file_extension_, const std::string & marks_path_, @@ -92,8 +92,7 @@ public: using StreamPtr = std::unique_ptr; MergeTreeDataPartWriterOnDisk( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot_, const std::vector & indices_to_recalc, diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterWide.cpp b/src/Storages/MergeTree/MergeTreeDataPartWriterWide.cpp index 70654f521a1..62917bcb084 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterWide.cpp +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterWide.cpp @@ -71,8 +71,7 @@ Granules getGranulesToWrite(const MergeTreeIndexGranularity & index_granularity, } MergeTreeDataPartWriterWide::MergeTreeDataPartWriterWide( - const MergeTreeData::DataPartPtr & data_part_, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part_, const NamesAndTypesList & columns_list_, const StorageMetadataPtr & metadata_snapshot_, const std::vector & indices_to_recalc_, @@ -80,7 +79,7 @@ MergeTreeDataPartWriterWide::MergeTreeDataPartWriterWide( const CompressionCodecPtr & default_codec_, const MergeTreeWriterSettings & settings_, const MergeTreeIndexGranularity & index_granularity_) - : MergeTreeDataPartWriterOnDisk(data_part_, std::move(data_part_storage_builder_), columns_list_, metadata_snapshot_, + : MergeTreeDataPartWriterOnDisk(data_part_, columns_list_, metadata_snapshot_, indices_to_recalc_, marks_file_extension_, default_codec_, settings_, index_granularity_) { @@ -117,7 +116,7 @@ void MergeTreeDataPartWriterWide::addStreams( column_streams[stream_name] = std::make_unique( stream_name, - data_part_storage_builder, + data_part->getDataPartStoragePtr(), stream_name, DATA_FILE_EXTENSION, stream_name, marks_file_extension, compression_codec, @@ -356,7 +355,7 @@ void MergeTreeDataPartWriterWide::writeColumn( { ISerialization::SerializeBinaryBulkSettings serialize_settings; serialize_settings.getter = createStreamGetter(name_and_type, offset_columns); - serialization->serializeBinaryBulkStatePrefix(serialize_settings, it->second); + serialization->serializeBinaryBulkStatePrefix(column, serialize_settings, it->second); } const auto & global_settings = storage.getContext()->getSettingsRef(); @@ -421,20 +420,18 @@ void MergeTreeDataPartWriterWide::validateColumnOfFixedSize(const NameAndTypePai String mrk_path = escaped_name + marks_file_extension; String bin_path = escaped_name + DATA_FILE_EXTENSION; - auto data_part_storage = data_part_storage_builder->getStorage(); - /// Some columns may be removed because of ttl. Skip them. - if (!data_part_storage->exists(mrk_path)) + if (!data_part->getDataPartStorage().exists(mrk_path)) return; - auto mrk_file_in = data_part_storage->readFile(mrk_path, {}, std::nullopt, std::nullopt); + auto mrk_file_in = data_part->getDataPartStorage().readFile(mrk_path, {}, std::nullopt, std::nullopt); std::unique_ptr mrk_in; if (data_part->index_granularity_info.mark_type.compressed) mrk_in = std::make_unique(std::move(mrk_file_in)); else mrk_in = std::move(mrk_file_in); - DB::CompressedReadBufferFromFile bin_in(data_part_storage->readFile(bin_path, {}, std::nullopt, std::nullopt)); + DB::CompressedReadBufferFromFile bin_in(data_part->getDataPartStorage().readFile(bin_path, {}, std::nullopt, std::nullopt)); bool must_be_last = false; UInt64 offset_in_compressed_file = 0; UInt64 offset_in_decompressed_block = 0; @@ -485,7 +482,7 @@ void MergeTreeDataPartWriterWide::validateColumnOfFixedSize(const NameAndTypePai if (index_granularity_rows != index_granularity.getMarkRows(mark_num)) throw Exception( ErrorCodes::LOGICAL_ERROR, "Incorrect mark rows for part {} for mark #{} (compressed offset {}, decompressed offset {}), in-memory {}, on disk {}, total marks {}", - data_part_storage_builder->getFullPath(), mark_num, offset_in_compressed_file, offset_in_decompressed_block, index_granularity.getMarkRows(mark_num), index_granularity_rows, index_granularity.getMarksCount()); + data_part->getDataPartStorage().getFullPath(), mark_num, offset_in_compressed_file, offset_in_decompressed_block, index_granularity.getMarkRows(mark_num), index_granularity_rows, index_granularity.getMarksCount()); auto column = type->createColumn(); diff --git a/src/Storages/MergeTree/MergeTreeDataPartWriterWide.h b/src/Storages/MergeTree/MergeTreeDataPartWriterWide.h index 08815d9930a..633b5119474 100644 --- a/src/Storages/MergeTree/MergeTreeDataPartWriterWide.h +++ b/src/Storages/MergeTree/MergeTreeDataPartWriterWide.h @@ -18,8 +18,7 @@ class MergeTreeDataPartWriterWide : public MergeTreeDataPartWriterOnDisk { public: MergeTreeDataPartWriterWide( - const MergeTreeData::DataPartPtr & data_part, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part, const NamesAndTypesList & columns_list, const StorageMetadataPtr & metadata_snapshot, const std::vector & indices_to_recalc, diff --git a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp index 6159d6bdf2d..afdd98b8e41 100644 --- a/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp +++ b/src/Storages/MergeTree/MergeTreeDataSelectExecutor.cpp @@ -1639,10 +1639,10 @@ MarkRanges MergeTreeDataSelectExecutor::filterMarksUsingIndex( UncompressedCache * uncompressed_cache, Poco::Logger * log) { - if (!index_helper->getDeserializedFormat(part->data_part_storage, index_helper->getFileName())) + if (!index_helper->getDeserializedFormat(part->getDataPartStorage(), index_helper->getFileName())) { LOG_DEBUG(log, "File for index {} does not exist ({}.*). Skipping it.", backQuote(index_helper->index.name), - (fs::path(part->data_part_storage->getFullPath()) / index_helper->getFileName()).string()); + (fs::path(part->getDataPartStorage().getFullPath()) / index_helper->getFileName()).string()); return ranges; } @@ -1757,7 +1757,7 @@ MarkRanges MergeTreeDataSelectExecutor::filterMarksUsingMergedIndex( { for (const auto & index_helper : indices) { - if (!part->data_part_storage->exists(index_helper->getFileName() + ".idx")) + if (!part->getDataPartStorage().exists(index_helper->getFileName() + ".idx")) { LOG_DEBUG(log, "File for index {} does not exist. Skipping it.", backQuote(index_helper->index.name)); return ranges; diff --git a/src/Storages/MergeTree/MergeTreeDataWriter.cpp b/src/Storages/MergeTree/MergeTreeDataWriter.cpp index 7b99819340e..815e62848a2 100644 --- a/src/Storages/MergeTree/MergeTreeDataWriter.cpp +++ b/src/Storages/MergeTree/MergeTreeDataWriter.cpp @@ -288,7 +288,7 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( auto columns = metadata_snapshot->getColumns().getAllPhysical().filter(block.getNames()); for (auto & column : columns) - if (isObject(column.type)) + if (column.type->hasDynamicSubcolumns()) column.type = block.getByName(column.name).type; static const String TMP_PREFIX = "tmp_insert_"; @@ -378,10 +378,7 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( data.relative_data_path, TMP_PREFIX + part_name); - auto data_part_storage_builder = std::make_shared( - data_part_volume, - data.relative_data_path, - TMP_PREFIX + part_name); + data_part_storage->beginTransaction(); auto new_data_part = data.createPart( part_name, @@ -408,15 +405,15 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( if (new_data_part->isStoredOnDisk()) { /// The name could be non-unique in case of stale files from previous runs. - String full_path = new_data_part->data_part_storage->getFullPath(); + String full_path = new_data_part->getDataPartStorage().getFullPath(); - if (new_data_part->data_part_storage->exists()) + if (new_data_part->getDataPartStorage().exists()) { LOG_WARNING(log, "Removing old temporary directory {}", full_path); - data_part_storage_builder->removeRecursive(); + data_part_storage->removeRecursive(); } - data_part_storage_builder->createDirectories(); + data_part_storage->createDirectories(); if (data.getSettings()->fsync_part_directory) { @@ -448,7 +445,7 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( auto compression_codec = data.getContext()->chooseCompressionCodec(0, 0); const auto & index_factory = MergeTreeIndexFactory::instance(); - auto out = std::make_unique(new_data_part, data_part_storage_builder, metadata_snapshot, columns, + auto out = std::make_unique(new_data_part, metadata_snapshot, columns, index_factory.getMany(metadata_snapshot->getSecondaryIndices()), compression_codec, context->getCurrentTransaction(), false, false, context->getWriteSettings()); @@ -459,9 +456,8 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( auto projection_block = projection.calculate(block, context); if (projection_block.rows()) { - auto proj_temp_part = writeProjectionPart(data, log, projection_block, projection, data_part_storage_builder, new_data_part.get()); + auto proj_temp_part = writeProjectionPart(data, log, projection_block, projection, new_data_part.get()); new_data_part->addProjectionPart(projection.name, std::move(proj_temp_part.part)); - proj_temp_part.builder->commit(); for (auto & stream : proj_temp_part.streams) temp_part.streams.emplace_back(std::move(stream)); } @@ -473,7 +469,6 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( nullptr, nullptr); temp_part.part = new_data_part; - temp_part.builder = data_part_storage_builder; temp_part.streams.emplace_back(TemporaryPart::Stream{.stream = std::move(out), .finalizer = std::move(finalizer)}); ProfileEvents::increment(ProfileEvents::MergeTreeDataWriterRows, block.rows()); @@ -485,11 +480,8 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempPart( MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( const String & part_name, - MergeTreeDataPartType part_type, - const String & relative_path, - const DataPartStorageBuilderPtr & data_part_storage_builder, bool is_temp, - const IMergeTreeDataPart * parent_part, + IMergeTreeDataPart * parent_part, const MergeTreeData & data, Poco::Logger * log, Block block, @@ -498,7 +490,23 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( TemporaryPart temp_part; const StorageMetadataPtr & metadata_snapshot = projection.metadata; MergeTreePartInfo new_part_info("all", 0, 0, 0); - auto projection_part_storage = parent_part->data_part_storage->getProjection(relative_path); + + MergeTreeDataPartType part_type; + if (parent_part->getType() == MergeTreeDataPartType::InMemory) + { + part_type = MergeTreeDataPartType::InMemory; + } + else + { + /// Size of part would not be greater than block.bytes() + epsilon + size_t expected_size = block.bytes(); + // just check if there is enough space on parent volume + data.reserveSpace(expected_size, parent_part->getDataPartStorage()); + part_type = data.choosePartTypeOnDisk(expected_size, block.rows()); + } + + auto relative_path = part_name + (is_temp ? ".tmp_proj" : ".proj"); + auto projection_part_storage = parent_part->getDataPartStorage().getProjection(relative_path); auto new_data_part = data.createPart( part_name, part_type, @@ -506,7 +514,6 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( projection_part_storage, parent_part); - auto projection_part_storage_builder = data_part_storage_builder->getProjection(relative_path); new_data_part->is_temp = is_temp; NamesAndTypesList columns = metadata_snapshot->getColumns().getAllPhysical().filter(block.getNames()); @@ -522,10 +529,10 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( if (projection_part_storage->exists()) { LOG_WARNING(log, "Removing old temporary directory {}", projection_part_storage->getFullPath()); - projection_part_storage_builder->removeRecursive(); + projection_part_storage->removeRecursive(); } - projection_part_storage_builder->createDirectories(); + projection_part_storage->createDirectories(); } /// If we need to calculate some columns to sort. @@ -569,7 +576,6 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( auto out = std::make_unique( new_data_part, - projection_part_storage_builder, metadata_snapshot, columns, MergeTreeIndices{}, @@ -580,7 +586,6 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( out->writeWithPermutation(block, perm_ptr); auto finalizer = out->finalizePartAsync(new_data_part, false); temp_part.part = new_data_part; - temp_part.builder = projection_part_storage_builder; temp_part.streams.emplace_back(TemporaryPart::Stream{.stream = std::move(out), .finalizer = std::move(finalizer)}); ProfileEvents::increment(ProfileEvents::MergeTreeDataProjectionWriterRows, block.rows()); @@ -591,98 +596,40 @@ MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPartImpl( } MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeProjectionPart( - MergeTreeData & data, + const MergeTreeData & data, Poco::Logger * log, Block block, const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part) + IMergeTreeDataPart * parent_part) { - String part_name = projection.name; - MergeTreeDataPartType part_type; - if (parent_part->getType() == MergeTreeDataPartType::InMemory) - { - part_type = MergeTreeDataPartType::InMemory; - } - else - { - /// Size of part would not be greater than block.bytes() + epsilon - size_t expected_size = block.bytes(); - // just check if there is enough space on parent volume - data.reserveSpace(expected_size, data_part_storage_builder); - part_type = data.choosePartTypeOnDisk(expected_size, block.rows()); - } - return writeProjectionPartImpl( - part_name, - part_type, - part_name + ".proj" /* relative_path */, - data_part_storage_builder, + projection.name, false /* is_temp */, parent_part, data, log, - block, + std::move(block), projection); } /// This is used for projection materialization process which may contain multiple stages of /// projection part merges. MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeTempProjectionPart( - MergeTreeData & data, - Poco::Logger * log, - Block block, - const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part, - size_t block_num) -{ - String part_name = fmt::format("{}_{}", projection.name, block_num); - MergeTreeDataPartType part_type; - if (parent_part->getType() == MergeTreeDataPartType::InMemory) - { - part_type = MergeTreeDataPartType::InMemory; - } - else - { - /// Size of part would not be greater than block.bytes() + epsilon - size_t expected_size = block.bytes(); - // just check if there is enough space on parent volume - data.reserveSpace(expected_size, data_part_storage_builder); - part_type = data.choosePartTypeOnDisk(expected_size, block.rows()); - } - - return writeProjectionPartImpl( - part_name, - part_type, - part_name + ".tmp_proj" /* relative_path */, - data_part_storage_builder, - true /* is_temp */, - parent_part, - data, - log, - block, - projection); -} - -MergeTreeDataWriter::TemporaryPart MergeTreeDataWriter::writeInMemoryProjectionPart( const MergeTreeData & data, Poco::Logger * log, Block block, const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part) + IMergeTreeDataPart * parent_part, + size_t block_num) { + String part_name = fmt::format("{}_{}", projection.name, block_num); return writeProjectionPartImpl( - projection.name, - MergeTreeDataPartType::InMemory, - projection.name + ".proj" /* relative_path */, - data_part_storage_builder, - false /* is_temp */, + part_name, + true /* is_temp */, parent_part, data, log, - block, + std::move(block), projection); } diff --git a/src/Storages/MergeTree/MergeTreeDataWriter.h b/src/Storages/MergeTree/MergeTreeDataWriter.h index 00438a29fa1..8c2bf66e8f8 100644 --- a/src/Storages/MergeTree/MergeTreeDataWriter.h +++ b/src/Storages/MergeTree/MergeTreeDataWriter.h @@ -52,7 +52,6 @@ public: struct TemporaryPart { MergeTreeData::MutableDataPartPtr part; - DataPartStorageBuilderPtr builder; struct Stream { @@ -74,31 +73,20 @@ public: /// For insertion. static TemporaryPart writeProjectionPart( - MergeTreeData & data, - Poco::Logger * log, - Block block, - const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part); - - /// For mutation: MATERIALIZE PROJECTION. - static TemporaryPart writeTempProjectionPart( - MergeTreeData & data, - Poco::Logger * log, - Block block, - const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part, - size_t block_num); - - /// For WriteAheadLog AddPart. - static TemporaryPart writeInMemoryProjectionPart( const MergeTreeData & data, Poco::Logger * log, Block block, const ProjectionDescription & projection, - const DataPartStorageBuilderPtr & data_part_storage_builder, - const IMergeTreeDataPart * parent_part); + IMergeTreeDataPart * parent_part); + + /// For mutation: MATERIALIZE PROJECTION. + static TemporaryPart writeTempProjectionPart( + const MergeTreeData & data, + Poco::Logger * log, + Block block, + const ProjectionDescription & projection, + IMergeTreeDataPart * parent_part, + size_t block_num); static Block mergeBlock( const Block & block, @@ -110,18 +98,14 @@ public: private: static TemporaryPart writeProjectionPartImpl( const String & part_name, - MergeTreeDataPartType part_type, - const String & relative_path, - const DataPartStorageBuilderPtr & data_part_storage_builder, bool is_temp, - const IMergeTreeDataPart * parent_part, + IMergeTreeDataPart * parent_part, const MergeTreeData & data, Poco::Logger * log, Block block, const ProjectionDescription & projection); MergeTreeData & data; - Poco::Logger * log; }; diff --git a/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.cpp b/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.cpp index 3dd0568107e..be7118066bb 100644 --- a/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.cpp @@ -6,11 +6,13 @@ #include #include #include +#include #include #include #include #include #include +#include #include #include #include @@ -28,19 +30,7 @@ namespace ErrorCodes namespace { -PreparedSetKey getPreparedSetKey(const ASTPtr & node, const DataTypePtr & data_type) -{ - /// If the data type is tuple, let's try unbox once - if (node->as() || node->as()) - return PreparedSetKey::forSubquery(*node); - - if (const auto * date_type_tuple = typeid_cast(&*data_type)) - return PreparedSetKey::forLiteral(*node, date_type_tuple->getElements()); - - return PreparedSetKey::forLiteral(*node, DataTypes(1, data_type)); -} - -ColumnWithTypeAndName getPreparedSetInfo(const SetPtr & prepared_set) +ColumnWithTypeAndName getPreparedSetInfo(const ConstSetPtr & prepared_set) { if (prepared_set->getDataTypes().size() == 1) return {prepared_set->getSetElements()[0], prepared_set->getElementsTypes()[0], "dummy"}; @@ -110,8 +100,22 @@ MergeTreeIndexConditionBloomFilter::MergeTreeIndexConditionBloomFilter( const SelectQueryInfo & info_, ContextPtr context_, const Block & header_, size_t hash_functions_) : WithContext(context_), header(header_), query_info(info_), hash_functions(hash_functions_) { - auto atom_from_ast = [this](auto & node, auto, auto & constants, auto & out) { return traverseAtomAST(node, constants, out); }; - rpn = std::move(RPNBuilder(info_, getContext(), atom_from_ast).extractRPN()); + ASTPtr filter_node = buildFilterNode(query_info.query); + + if (!filter_node) + { + rpn.push_back(RPNElement::FUNCTION_UNKNOWN); + return; + } + + auto block_with_constants = KeyCondition::getBlockWithConstants(query_info.query, query_info.syntax_analyzer_result, context_); + RPNBuilder builder( + filter_node, + context_, + std::move(block_with_constants), + query_info.prepared_sets, + [&](const RPNBuilderTreeNode & node, RPNElement & out) { return extractAtomFromTree(node, out); }); + rpn = std::move(builder).extractRPN(); } bool MergeTreeIndexConditionBloomFilter::alwaysUnknownOrTrue() const @@ -235,12 +239,13 @@ bool MergeTreeIndexConditionBloomFilter::mayBeTrueOnGranule(const MergeTreeIndex return rpn_stack[0].can_be_true; } -bool MergeTreeIndexConditionBloomFilter::traverseAtomAST(const ASTPtr & node, Block & block_with_constants, RPNElement & out) +bool MergeTreeIndexConditionBloomFilter::extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out) { { Field const_value; DataTypePtr const_type; - if (KeyCondition::getConstant(node, block_with_constants, const_value, const_type)) + + if (node.tryGetConstant(const_value, const_type)) { if (const_value.getType() == Field::Types::UInt64) { @@ -262,56 +267,62 @@ bool MergeTreeIndexConditionBloomFilter::traverseAtomAST(const ASTPtr & node, Bl } } - return traverseFunction(node, block_with_constants, out, nullptr); + return traverseFunction(node, out, nullptr /*parent*/); } -bool MergeTreeIndexConditionBloomFilter::traverseFunction(const ASTPtr & node, Block & block_with_constants, RPNElement & out, const ASTPtr & parent) +bool MergeTreeIndexConditionBloomFilter::traverseFunction(const RPNBuilderTreeNode & node, RPNElement & out, const RPNBuilderTreeNode * parent) { bool maybe_useful = false; - if (const auto * function = node->as()) + if (node.isFunction()) { - if (!function->arguments) - return false; + const auto function = node.toFunctionNode(); + auto arguments_size = function.getArgumentsSize(); + auto function_name = function.getFunctionName(); - const ASTs & arguments = function->arguments->children; - for (const auto & arg : arguments) + for (size_t i = 0; i < arguments_size; ++i) { - if (traverseFunction(arg, block_with_constants, out, node)) + auto argument = function.getArgumentAt(i); + if (traverseFunction(argument, out, &node)) maybe_useful = true; } - if (arguments.size() != 2) + if (arguments_size != 2) return false; - if (functionIsInOrGlobalInOperator(function->name)) - { - auto prepared_set = getPreparedSet(arguments[1]); + auto lhs_argument = function.getArgumentAt(0); + auto rhs_argument = function.getArgumentAt(1); - if (prepared_set) + if (functionIsInOrGlobalInOperator(function_name)) + { + ConstSetPtr prepared_set = rhs_argument.tryGetPreparedSet(); + + if (prepared_set && prepared_set->hasExplicitSetElements()) { - if (traverseASTIn(function->name, arguments[0], prepared_set, out)) + const auto prepared_info = getPreparedSetInfo(prepared_set); + if (traverseTreeIn(function_name, lhs_argument, prepared_set, prepared_info.type, prepared_info.column, out)) maybe_useful = true; } } - else if (function->name == "equals" || - function->name == "notEquals" || - function->name == "has" || - function->name == "mapContains" || - function->name == "indexOf" || - function->name == "hasAny" || - function->name == "hasAll") + else if (function_name == "equals" || + function_name == "notEquals" || + function_name == "has" || + function_name == "mapContains" || + function_name == "indexOf" || + function_name == "hasAny" || + function_name == "hasAll") { Field const_value; DataTypePtr const_type; - if (KeyCondition::getConstant(arguments[1], block_with_constants, const_value, const_type)) + + if (rhs_argument.tryGetConstant(const_value, const_type)) { - if (traverseASTEquals(function->name, arguments[0], const_type, const_value, out, parent)) + if (traverseTreeEquals(function_name, lhs_argument, const_type, const_value, out, parent)) maybe_useful = true; } - else if (KeyCondition::getConstant(arguments[0], block_with_constants, const_value, const_type)) + else if (lhs_argument.tryGetConstant(const_value, const_type)) { - if (traverseASTEquals(function->name, arguments[1], const_type, const_value, out, parent)) + if (traverseTreeEquals(function_name, rhs_argument, const_type, const_value, out, parent)) maybe_useful = true; } } @@ -320,28 +331,20 @@ bool MergeTreeIndexConditionBloomFilter::traverseFunction(const ASTPtr & node, B return maybe_useful; } -bool MergeTreeIndexConditionBloomFilter::traverseASTIn( +bool MergeTreeIndexConditionBloomFilter::traverseTreeIn( const String & function_name, - const ASTPtr & key_ast, - const SetPtr & prepared_set, - RPNElement & out) -{ - const auto prepared_info = getPreparedSetInfo(prepared_set); - return traverseASTIn(function_name, key_ast, prepared_set, prepared_info.type, prepared_info.column, out); -} - -bool MergeTreeIndexConditionBloomFilter::traverseASTIn( - const String & function_name, - const ASTPtr & key_ast, - const SetPtr & prepared_set, + const RPNBuilderTreeNode & key_node, + const ConstSetPtr & prepared_set, const DataTypePtr & type, const ColumnPtr & column, RPNElement & out) { - if (header.has(key_ast->getColumnName())) + auto key_node_column_name = key_node.getColumnName(); + + if (header.has(key_node_column_name)) { size_t row_size = column->size(); - size_t position = header.getPositionByName(key_ast->getColumnName()); + size_t position = header.getPositionByName(key_node_column_name); const DataTypePtr & index_type = header.getByPosition(position).type; const auto & converted_column = castColumn(ColumnWithTypeAndName{column, type, ""}, index_type); out.predicate.emplace_back(std::make_pair(position, BloomFilterHash::hashWithColumn(index_type, converted_column, 0, row_size))); @@ -355,30 +358,33 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTIn( return true; } - if (const auto * function = key_ast->as()) + if (key_node.isFunction()) { + auto key_node_function = key_node.toFunctionNode(); + auto key_node_function_name = key_node_function.getFunctionName(); + size_t key_node_function_arguments_size = key_node_function.getArgumentsSize(); + WhichDataType which(type); - if (which.isTuple() && function->name == "tuple") + if (which.isTuple() && key_node_function_name == "tuple") { const auto & tuple_column = typeid_cast(column.get()); const auto & tuple_data_type = typeid_cast(type.get()); - const ASTs & arguments = typeid_cast(*function->arguments).children; - if (tuple_data_type->getElements().size() != arguments.size() || tuple_column->getColumns().size() != arguments.size()) + if (tuple_data_type->getElements().size() != key_node_function_arguments_size || tuple_column->getColumns().size() != key_node_function_arguments_size) throw Exception("Illegal types of arguments of function " + function_name, ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); bool match_with_subtype = false; const auto & sub_columns = tuple_column->getColumns(); const auto & sub_data_types = tuple_data_type->getElements(); - for (size_t index = 0; index < arguments.size(); ++index) - match_with_subtype |= traverseASTIn(function_name, arguments[index], nullptr, sub_data_types[index], sub_columns[index], out); + for (size_t index = 0; index < key_node_function_arguments_size; ++index) + match_with_subtype |= traverseTreeIn(function_name, key_node_function.getArgumentAt(index), nullptr, sub_data_types[index], sub_columns[index], out); return match_with_subtype; } - if (function->name == "arrayElement") + if (key_node_function_name == "arrayElement") { /** Try to parse arrayElement for mapKeys index. * It is important to ignore keys like column_map['Key'] IN ('') because if key does not exists in map @@ -387,7 +393,6 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTIn( * We cannot skip keys that does not exist in map if comparison is with default type value because * that way we skip necessary granules where map key does not exists. */ - if (!prepared_set) return false; @@ -400,28 +405,26 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTIn( if (set_contain_default_value) return false; - const auto * column_ast_identifier = function->arguments.get()->children[0].get()->as(); - if (!column_ast_identifier) - return false; - - const auto & col_name = column_ast_identifier->name(); - auto map_keys_index_column_name = fmt::format("mapKeys({})", col_name); - auto map_values_index_column_name = fmt::format("mapValues({})", col_name); + auto first_argument = key_node_function.getArgumentAt(0); + const auto column_name = first_argument.getColumnName(); + auto map_keys_index_column_name = fmt::format("mapKeys({})", column_name); + auto map_values_index_column_name = fmt::format("mapValues({})", column_name); if (header.has(map_keys_index_column_name)) { /// For mapKeys we serialize key argument with bloom filter - auto & argument = function->arguments.get()->children[1]; + auto second_argument = key_node_function.getArgumentAt(1); - if (const auto * literal = argument->as()) + Field constant_value; + DataTypePtr constant_type; + + if (second_argument.tryGetConstant(constant_value, constant_type)) { size_t position = header.getPositionByName(map_keys_index_column_name); const DataTypePtr & index_type = header.getByPosition(position).type; - - auto element_key = literal->value; const DataTypePtr actual_type = BloomFilter::getPrimitiveType(index_type); - out.predicate.emplace_back(std::make_pair(position, BloomFilterHash::hashWithField(actual_type.get(), element_key))); + out.predicate.emplace_back(std::make_pair(position, BloomFilterHash::hashWithField(actual_type.get(), constant_value))); } else { @@ -459,74 +462,97 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTIn( } -static bool indexOfCanUseBloomFilter(const ASTPtr & parent) +static bool indexOfCanUseBloomFilter(const RPNBuilderTreeNode * parent) { if (!parent) return true; + if (!parent->isFunction()) + return false; + + auto function = parent->toFunctionNode(); + auto function_name = function.getFunctionName(); + /// `parent` is a function where `indexOf` is located. /// Example: `indexOf(arr, x) = 1`, parent is a function named `equals`. - if (const auto * function = parent->as()) + if (function_name == "and") { - if (function->name == "and") + return true; + } + else if (function_name == "equals" /// notEquals is not applicable + || function_name == "greater" || function_name == "greaterOrEquals" + || function_name == "less" || function_name == "lessOrEquals") + { + size_t function_arguments_size = function.getArgumentsSize(); + if (function_arguments_size != 2) + return false; + + /// We don't allow constant expressions like `indexOf(arr, x) = 1 + 0` but it's negligible. + + /// We should return true when the corresponding expression implies that the array contains the element. + /// Example: when `indexOf(arr, x)` > 10 is written, it means that arr definitely should contain the element + /// (at least at 11th position but it does not matter). + + bool reversed = false; + Field constant_value; + DataTypePtr constant_type; + + if (function.getArgumentAt(0).tryGetConstant(constant_value, constant_type)) { + reversed = true; + } + else if (function.getArgumentAt(1).tryGetConstant(constant_value, constant_type)) + { + } + else + { + return false; + } + + Field zero(0); + bool constant_equal_zero = applyVisitor(FieldVisitorAccurateEquals(), constant_value, zero); + + if (function_name == "equals" && !constant_equal_zero) + { + /// indexOf(...) = c, c != 0 return true; } - else if (function->name == "equals" /// notEquals is not applicable - || function->name == "greater" || function->name == "greaterOrEquals" - || function->name == "less" || function->name == "lessOrEquals") + else if (function_name == "notEquals" && constant_equal_zero) { - if (function->arguments->children.size() != 2) - return false; - - /// We don't allow constant expressions like `indexOf(arr, x) = 1 + 0` but it's negligible. - - /// We should return true when the corresponding expression implies that the array contains the element. - /// Example: when `indexOf(arr, x)` > 10 is written, it means that arr definitely should contain the element - /// (at least at 11th position but it does not matter). - - bool reversed = false; - const ASTLiteral * constant = nullptr; - - if (const ASTLiteral * left = function->arguments->children[0]->as()) - { - constant = left; - reversed = true; - } - else if (const ASTLiteral * right = function->arguments->children[1]->as()) - { - constant = right; - } - else - return false; - - Field zero(0); - return (function->name == "equals" /// indexOf(...) = c, c != 0 - && !applyVisitor(FieldVisitorAccurateEquals(), constant->value, zero)) - || (function->name == "notEquals" /// indexOf(...) != c, c = 0 - && applyVisitor(FieldVisitorAccurateEquals(), constant->value, zero)) - || (function->name == (reversed ? "less" : "greater") /// indexOf(...) > c, c >= 0 - && !applyVisitor(FieldVisitorAccurateLess(), constant->value, zero)) - || (function->name == (reversed ? "lessOrEquals" : "greaterOrEquals") /// indexOf(...) >= c, c > 0 - && applyVisitor(FieldVisitorAccurateLess(), zero, constant->value)); + /// indexOf(...) != c, c = 0 + return true; } + else if (function_name == (reversed ? "less" : "greater") && !applyVisitor(FieldVisitorAccurateLess(), constant_value, zero)) + { + /// indexOf(...) > c, c >= 0 + return true; + } + else if (function_name == (reversed ? "lessOrEquals" : "greaterOrEquals") && applyVisitor(FieldVisitorAccurateLess(), zero, constant_value)) + { + /// indexOf(...) >= c, c > 0 + return true; + } + + return false; } return false; } -bool MergeTreeIndexConditionBloomFilter::traverseASTEquals( +bool MergeTreeIndexConditionBloomFilter::traverseTreeEquals( const String & function_name, - const ASTPtr & key_ast, + const RPNBuilderTreeNode & key_node, const DataTypePtr & value_type, const Field & value_field, RPNElement & out, - const ASTPtr & parent) + const RPNBuilderTreeNode * parent) { - if (header.has(key_ast->getColumnName())) + auto key_column_name = key_node.getColumnName(); + + if (header.has(key_column_name)) { - size_t position = header.getPositionByName(key_ast->getColumnName()); + size_t position = header.getPositionByName(key_column_name); const DataTypePtr & index_type = header.getByPosition(position).type; const auto * array_type = typeid_cast(index_type.get()); @@ -602,13 +628,7 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTEquals( if (function_name == "mapContains" || function_name == "has") { - const auto * key_ast_identifier = key_ast.get()->as(); - if (!key_ast_identifier) - return false; - - const auto & col_name = key_ast_identifier->name(); - auto map_keys_index_column_name = fmt::format("mapKeys({})", col_name); - + auto map_keys_index_column_name = fmt::format("mapKeys({})", key_column_name); if (!header.has(map_keys_index_column_name)) return false; @@ -629,29 +649,32 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTEquals( return true; } - if (const auto * function = key_ast->as()) + if (key_node.isFunction()) { WhichDataType which(value_type); - if (which.isTuple() && function->name == "tuple") + auto key_node_function = key_node.toFunctionNode(); + auto key_node_function_name = key_node_function.getFunctionName(); + size_t key_node_function_arguments_size = key_node_function.getArgumentsSize(); + + if (which.isTuple() && key_node_function_name == "tuple") { const Tuple & tuple = value_field.get(); const auto * value_tuple_data_type = typeid_cast(value_type.get()); - const ASTs & arguments = typeid_cast(*function->arguments).children; - if (tuple.size() != arguments.size()) + if (tuple.size() != key_node_function_arguments_size) throw Exception("Illegal types of arguments of function " + function_name, ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); bool match_with_subtype = false; const DataTypes & subtypes = value_tuple_data_type->getElements(); for (size_t index = 0; index < tuple.size(); ++index) - match_with_subtype |= traverseASTEquals(function_name, arguments[index], subtypes[index], tuple[index], out, key_ast); + match_with_subtype |= traverseTreeEquals(function_name, key_node_function.getArgumentAt(index), subtypes[index], tuple[index], out, &key_node); return match_with_subtype; } - if (function->name == "arrayElement" && (function_name == "equals" || function_name == "notEquals")) + if (key_node_function_name == "arrayElement" && (function_name == "equals" || function_name == "notEquals")) { /** Try to parse arrayElement for mapKeys index. * It is important to ignore keys like column_map['Key'] = '' because if key does not exists in map @@ -663,27 +686,22 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTEquals( if (value_field == value_type->getDefault()) return false; - const auto * column_ast_identifier = function->arguments.get()->children[0].get()->as(); - if (!column_ast_identifier) - return false; + auto first_argument = key_node_function.getArgumentAt(0); + const auto column_name = first_argument.getColumnName(); - const auto & col_name = column_ast_identifier->name(); - - auto map_keys_index_column_name = fmt::format("mapKeys({})", col_name); - auto map_values_index_column_name = fmt::format("mapValues({})", col_name); + auto map_keys_index_column_name = fmt::format("mapKeys({})", column_name); + auto map_values_index_column_name = fmt::format("mapValues({})", column_name); size_t position = 0; Field const_value = value_field; + DataTypePtr const_type; if (header.has(map_keys_index_column_name)) { position = header.getPositionByName(map_keys_index_column_name); + auto second_argument = key_node_function.getArgumentAt(1); - auto & argument = function->arguments.get()->children[1]; - - if (const auto * literal = argument->as()) - const_value = literal->value; - else + if (!second_argument.tryGetConstant(const_value, const_type)) return false; } else if (header.has(map_values_index_column_name)) @@ -708,23 +726,4 @@ bool MergeTreeIndexConditionBloomFilter::traverseASTEquals( return false; } -SetPtr MergeTreeIndexConditionBloomFilter::getPreparedSet(const ASTPtr & node) -{ - if (header.has(node->getColumnName())) - { - const auto & column_and_type = header.getByName(node->getColumnName()); - auto set_key = getPreparedSetKey(node, column_and_type.type); - if (auto prepared_set = query_info.prepared_sets->get(set_key)) - return prepared_set; - } - else - { - for (const auto & set : query_info.prepared_sets->getByTreeHash(node->getTreeHash())) - if (set->hasExplicitSetElements()) - return set; - } - - return DB::SetPtr(); -} - } diff --git a/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.h b/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.h index 27fd701c67b..5d7ea371a83 100644 --- a/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.h +++ b/src/Storages/MergeTree/MergeTreeIndexConditionBloomFilter.h @@ -62,35 +62,27 @@ private: const size_t hash_functions; std::vector rpn; - SetPtr getPreparedSet(const ASTPtr & node); - bool mayBeTrueOnGranule(const MergeTreeIndexGranuleBloomFilter * granule) const; - bool traverseAtomAST(const ASTPtr & node, Block & block_with_constants, RPNElement & out); + bool extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out); - bool traverseFunction(const ASTPtr & node, Block & block_with_constants, RPNElement & out, const ASTPtr & parent); + bool traverseFunction(const RPNBuilderTreeNode & node, RPNElement & out, const RPNBuilderTreeNode * parent); - bool traverseASTIn( + bool traverseTreeIn( const String & function_name, - const ASTPtr & key_ast, - const SetPtr & prepared_set, - RPNElement & out); - - bool traverseASTIn( - const String & function_name, - const ASTPtr & key_ast, - const SetPtr & prepared_set, + const RPNBuilderTreeNode & key_node, + const ConstSetPtr & prepared_set, const DataTypePtr & type, const ColumnPtr & column, RPNElement & out); - bool traverseASTEquals( + bool traverseTreeEquals( const String & function_name, - const ASTPtr & key_ast, + const RPNBuilderTreeNode & key_node, const DataTypePtr & value_type, const Field & value_field, RPNElement & out, - const ASTPtr & parent); + const RPNBuilderTreeNode * parent); }; } diff --git a/src/Storages/MergeTree/MergeTreeIndexFullText.cpp b/src/Storages/MergeTree/MergeTreeIndexFullText.cpp index ff924290783..b96d40f5759 100644 --- a/src/Storages/MergeTree/MergeTreeIndexFullText.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexFullText.cpp @@ -11,9 +11,11 @@ #include #include #include +#include #include #include #include +#include #include #include @@ -148,13 +150,22 @@ MergeTreeConditionFullText::MergeTreeConditionFullText( , token_extractor(token_extactor_) , prepared_sets(query_info.prepared_sets) { - rpn = std::move( - RPNBuilder( - query_info, context, - [this] (const ASTPtr & node, ContextPtr /* context */, Block & block_with_constants, RPNElement & out) -> bool - { - return this->traverseAtomAST(node, block_with_constants, out); - }).extractRPN()); + ASTPtr filter_node = buildFilterNode(query_info.query); + + if (!filter_node) + { + rpn.push_back(RPNElement::FUNCTION_UNKNOWN); + return; + } + + auto block_with_constants = KeyCondition::getBlockWithConstants(query_info.query, query_info.syntax_analyzer_result, context); + RPNBuilder builder( + filter_node, + context, + std::move(block_with_constants), + query_info.prepared_sets, + [&](const RPNBuilderTreeNode & node, RPNElement & out) { return extractAtomFromTree(node, out); }); + rpn = std::move(builder).extractRPN(); } bool MergeTreeConditionFullText::alwaysUnknownOrTrue() const @@ -306,13 +317,13 @@ bool MergeTreeConditionFullText::getKey(const std::string & key_column_name, siz return true; } -bool MergeTreeConditionFullText::traverseAtomAST(const ASTPtr & node, Block & block_with_constants, RPNElement & out) +bool MergeTreeConditionFullText::extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out) { { Field const_value; DataTypePtr const_type; - if (KeyCondition::getConstant(node, block_with_constants, const_value, const_type)) + if (node.tryGetConstant(const_value, const_type)) { /// Check constant like in KeyCondition if (const_value.getType() == Field::Types::UInt64 @@ -329,53 +340,56 @@ bool MergeTreeConditionFullText::traverseAtomAST(const ASTPtr & node, Block & bl } } - if (const auto * function = node->as()) + if (node.isFunction()) { - if (!function->arguments) + auto function_node = node.toFunctionNode(); + auto function_name = function_node.getFunctionName(); + + size_t arguments_size = function_node.getArgumentsSize(); + if (arguments_size != 2) return false; - const ASTs & arguments = function->arguments->children; + auto left_argument = function_node.getArgumentAt(0); + auto right_argument = function_node.getArgumentAt(1); - if (arguments.size() != 2) - return false; - - if (functionIsInOrGlobalInOperator(function->name)) + if (functionIsInOrGlobalInOperator(function_name)) { - if (tryPrepareSetBloomFilter(arguments, out)) + if (tryPrepareSetBloomFilter(left_argument, right_argument, out)) { - if (function->name == "notIn") + if (function_name == "notIn") { out.function = RPNElement::FUNCTION_NOT_IN; return true; } - else if (function->name == "in") + else if (function_name == "in") { out.function = RPNElement::FUNCTION_IN; return true; } } } - else if (function->name == "equals" || - function->name == "notEquals" || - function->name == "has" || - function->name == "mapContains" || - function->name == "like" || - function->name == "notLike" || - function->name == "hasToken" || - function->name == "startsWith" || - function->name == "endsWith" || - function->name == "multiSearchAny") + else if (function_name == "equals" || + function_name == "notEquals" || + function_name == "has" || + function_name == "mapContains" || + function_name == "like" || + function_name == "notLike" || + function_name == "hasToken" || + function_name == "startsWith" || + function_name == "endsWith" || + function_name == "multiSearchAny") { Field const_value; DataTypePtr const_type; - if (KeyCondition::getConstant(arguments[1], block_with_constants, const_value, const_type)) + + if (right_argument.tryGetConstant(const_value, const_type)) { - if (traverseASTEquals(function->name, arguments[0], const_type, const_value, out)) + if (traverseTreeEquals(function_name, left_argument, const_type, const_value, out)) return true; } - else if (KeyCondition::getConstant(arguments[0], block_with_constants, const_value, const_type) && (function->name == "equals" || function->name == "notEquals")) + else if (left_argument.tryGetConstant(const_value, const_type) && (function_name == "equals" || function_name == "notEquals")) { - if (traverseASTEquals(function->name, arguments[1], const_type, const_value, out)) + if (traverseTreeEquals(function_name, right_argument, const_type, const_value, out)) return true; } } @@ -384,9 +398,9 @@ bool MergeTreeConditionFullText::traverseAtomAST(const ASTPtr & node, Block & bl return false; } -bool MergeTreeConditionFullText::traverseASTEquals( +bool MergeTreeConditionFullText::traverseTreeEquals( const String & function_name, - const ASTPtr & key_ast, + const RPNBuilderTreeNode & key_node, const DataTypePtr & value_type, const Field & value_field, RPNElement & out) @@ -397,13 +411,17 @@ bool MergeTreeConditionFullText::traverseASTEquals( Field const_value = value_field; + auto column_name = key_node.getColumnName(); size_t key_column_num = 0; - bool key_exists = getKey(key_ast->getColumnName(), key_column_num); - bool map_key_exists = getKey(fmt::format("mapKeys({})", key_ast->getColumnName()), key_column_num); + bool key_exists = getKey(column_name, key_column_num); + bool map_key_exists = getKey(fmt::format("mapKeys({})", column_name), key_column_num); - if (const auto * function = key_ast->as()) + if (key_node.isFunction()) { - if (function->name == "arrayElement") + auto key_function_node = key_node.toFunctionNode(); + auto key_function_node_function_name = key_function_node.getFunctionName(); + + if (key_function_node_function_name == "arrayElement") { /** Try to parse arrayElement for mapKeys index. * It is important to ignore keys like column_map['Key'] = '' because if key does not exists in map @@ -415,11 +433,8 @@ bool MergeTreeConditionFullText::traverseASTEquals( if (value_field == value_type->getDefault()) return false; - const auto * column_ast_identifier = function->arguments.get()->children[0].get()->as(); - if (!column_ast_identifier) - return false; - - const auto & map_column_name = column_ast_identifier->name(); + auto first_argument = key_function_node.getArgumentAt(0); + const auto map_column_name = first_argument.getColumnName(); size_t map_keys_key_column_num = 0; auto map_keys_index_column_name = fmt::format("mapKeys({})", map_column_name); @@ -431,12 +446,11 @@ bool MergeTreeConditionFullText::traverseASTEquals( if (map_keys_exists) { - auto & argument = function->arguments.get()->children[1]; + auto second_argument = key_function_node.getArgumentAt(1); + DataTypePtr const_type; - if (const auto * literal = argument->as()) + if (second_argument.tryGetConstant(const_value, const_type)) { - auto element_key = literal->value; - const_value = element_key; key_column_num = map_keys_key_column_num; key_exists = true; } @@ -567,23 +581,24 @@ bool MergeTreeConditionFullText::traverseASTEquals( } bool MergeTreeConditionFullText::tryPrepareSetBloomFilter( - const ASTs & args, + const RPNBuilderTreeNode & left_argument, + const RPNBuilderTreeNode & right_argument, RPNElement & out) { - const ASTPtr & left_arg = args[0]; - const ASTPtr & right_arg = args[1]; - std::vector key_tuple_mapping; DataTypes data_types; - const auto * left_arg_tuple = typeid_cast(left_arg.get()); - if (left_arg_tuple && left_arg_tuple->name == "tuple") + auto left_argument_function_node_optional = left_argument.toFunctionNodeOrNull(); + + if (left_argument_function_node_optional && left_argument_function_node_optional->getFunctionName() == "tuple") { - const auto & tuple_elements = left_arg_tuple->arguments->children; - for (size_t i = 0; i < tuple_elements.size(); ++i) + const auto & left_argument_function_node = *left_argument_function_node_optional; + size_t left_argument_function_node_arguments_size = left_argument_function_node.getArgumentsSize(); + + for (size_t i = 0; i < left_argument_function_node_arguments_size; ++i) { size_t key = 0; - if (getKey(tuple_elements[i]->getColumnName(), key)) + if (getKey(left_argument_function_node.getArgumentAt(i).getColumnName(), key)) { key_tuple_mapping.emplace_back(i, key); data_types.push_back(index_data_types[key]); @@ -593,7 +608,7 @@ bool MergeTreeConditionFullText::tryPrepareSetBloomFilter( else { size_t key = 0; - if (getKey(left_arg->getColumnName(), key)) + if (getKey(left_argument.getColumnName(), key)) { key_tuple_mapping.emplace_back(0, key); data_types.push_back(index_data_types[key]); @@ -603,19 +618,10 @@ bool MergeTreeConditionFullText::tryPrepareSetBloomFilter( if (key_tuple_mapping.empty()) return false; - PreparedSetKey set_key; - if (typeid_cast(right_arg.get()) || typeid_cast(right_arg.get())) - set_key = PreparedSetKey::forSubquery(*right_arg); - else - set_key = PreparedSetKey::forLiteral(*right_arg, data_types); - - auto prepared_set = prepared_sets->get(set_key); + auto prepared_set = right_argument.tryGetPreparedSet(data_types); if (!prepared_set) return false; - if (!prepared_set->hasExplicitSetElements()) - return false; - for (const auto & data_type : prepared_set->getDataTypes()) if (data_type->getTypeId() != TypeIndex::String && data_type->getTypeId() != TypeIndex::FixedString) return false; diff --git a/src/Storages/MergeTree/MergeTreeIndexFullText.h b/src/Storages/MergeTree/MergeTreeIndexFullText.h index bb4f52a463e..ad487816aef 100644 --- a/src/Storages/MergeTree/MergeTreeIndexFullText.h +++ b/src/Storages/MergeTree/MergeTreeIndexFullText.h @@ -122,17 +122,17 @@ private: using RPN = std::vector; - bool traverseAtomAST(const ASTPtr & node, Block & block_with_constants, RPNElement & out); + bool extractAtomFromTree(const RPNBuilderTreeNode & node, RPNElement & out); - bool traverseASTEquals( + bool traverseTreeEquals( const String & function_name, - const ASTPtr & key_ast, + const RPNBuilderTreeNode & key_node, const DataTypePtr & value_type, const Field & value_field, RPNElement & out); bool getKey(const std::string & key_column_name, size_t & key_column_num); - bool tryPrepareSetBloomFilter(const ASTs & args, RPNElement & out); + bool tryPrepareSetBloomFilter(const RPNBuilderTreeNode & left_argument, const RPNBuilderTreeNode & right_argument, RPNElement & out); static bool createFunctionEqualsCondition( RPNElement & out, const Field & value, const BloomFilterParameters & params, TokenExtractorPtr token_extractor); diff --git a/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.cpp b/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.cpp index 9c154f786f7..11e1f9efcc2 100644 --- a/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.cpp @@ -89,10 +89,10 @@ std::string MarkType::getFileExtension() const } -std::optional MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(const DataPartStoragePtr & data_part_storage) +std::optional MergeTreeIndexGranularityInfo::getMarksExtensionFromFilesystem(const IDataPartStorage & data_part_storage) { - if (data_part_storage->exists()) - for (auto it = data_part_storage->iterate(); it->isValid(); it->next()) + if (data_part_storage.exists()) + for (auto it = data_part_storage.iterate(); it->isValid(); it->next()) if (it->isFile()) if (std::string ext = fs::path(it->name()).extension(); MarkType::isMarkFileExtension(ext)) return ext; @@ -110,7 +110,7 @@ MergeTreeIndexGranularityInfo::MergeTreeIndexGranularityInfo(const MergeTreeData fixed_index_granularity = storage.getSettings()->index_granularity; } -void MergeTreeIndexGranularityInfo::changeGranularityIfRequired(const DataPartStoragePtr & data_part_storage) +void MergeTreeIndexGranularityInfo::changeGranularityIfRequired(const IDataPartStorage & data_part_storage) { auto mrk_ext = getMarksExtensionFromFilesystem(data_part_storage); if (mrk_ext && !MarkType(*mrk_ext).adaptive) diff --git a/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.h b/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.h index 883fe3c899e..aed3081d3d0 100644 --- a/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.h +++ b/src/Storages/MergeTree/MergeTreeIndexGranularityInfo.h @@ -48,7 +48,7 @@ public: MergeTreeIndexGranularityInfo(MergeTreeDataPartType type_, bool is_adaptive_, size_t index_granularity_, size_t index_granularity_bytes_); - void changeGranularityIfRequired(const DataPartStoragePtr & data_part_storage); + void changeGranularityIfRequired(const IDataPartStorage & data_part_storage); String getMarksFilePath(const String & path_prefix) const { @@ -57,7 +57,7 @@ public: size_t getMarkSizeInBytes(size_t columns_num = 1) const; - static std::optional getMarksExtensionFromFilesystem(const DataPartStoragePtr & data_part_storage); + static std::optional getMarksExtensionFromFilesystem(const IDataPartStorage & data_part_storage); }; constexpr inline auto getNonAdaptiveMrkSizeWide() { return sizeof(UInt64) * 2; } diff --git a/src/Storages/MergeTree/MergeTreeIndexMinMax.cpp b/src/Storages/MergeTree/MergeTreeIndexMinMax.cpp index b190ac2b2fd..43e655a4ee5 100644 --- a/src/Storages/MergeTree/MergeTreeIndexMinMax.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexMinMax.cpp @@ -211,11 +211,11 @@ bool MergeTreeIndexMinMax::mayBenefitFromIndexForIn(const ASTPtr & node) const return false; } -MergeTreeIndexFormat MergeTreeIndexMinMax::getDeserializedFormat(const DataPartStoragePtr & data_part_storage, const std::string & relative_path_prefix) const +MergeTreeIndexFormat MergeTreeIndexMinMax::getDeserializedFormat(const IDataPartStorage & data_part_storage, const std::string & relative_path_prefix) const { - if (data_part_storage->exists(relative_path_prefix + ".idx2")) + if (data_part_storage.exists(relative_path_prefix + ".idx2")) return {2, ".idx2"}; - else if (data_part_storage->exists(relative_path_prefix + ".idx")) + else if (data_part_storage.exists(relative_path_prefix + ".idx")) return {1, ".idx"}; return {0 /* unknown */, ""}; } diff --git a/src/Storages/MergeTree/MergeTreeIndexMinMax.h b/src/Storages/MergeTree/MergeTreeIndexMinMax.h index 0566a15d535..af420613855 100644 --- a/src/Storages/MergeTree/MergeTreeIndexMinMax.h +++ b/src/Storages/MergeTree/MergeTreeIndexMinMax.h @@ -83,7 +83,7 @@ public: bool mayBenefitFromIndexForIn(const ASTPtr & node) const override; const char* getSerializedFileExtension() const override { return ".idx2"; } - MergeTreeIndexFormat getDeserializedFormat(const DataPartStoragePtr & data_part_storage, const std::string & path_prefix) const override; /// NOLINT + MergeTreeIndexFormat getDeserializedFormat(const IDataPartStorage & data_part_storage, const std::string & path_prefix) const override; /// NOLINT }; } diff --git a/src/Storages/MergeTree/MergeTreeIndexReader.cpp b/src/Storages/MergeTree/MergeTreeIndexReader.cpp index 33106f7ab64..7d7024a8ac2 100644 --- a/src/Storages/MergeTree/MergeTreeIndexReader.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexReader.cpp @@ -20,7 +20,7 @@ std::unique_ptr makeIndexReader( auto * load_marks_threadpool = settings.read_settings.load_marks_asynchronously ? &context->getLoadMarksThreadpool() : nullptr; return std::make_unique( - part->data_part_storage, + part->getDataPartStoragePtr(), index->getFileName(), extension, marks_count, all_mark_ranges, std::move(settings), mark_cache, uncompressed_cache, @@ -44,7 +44,7 @@ MergeTreeIndexReader::MergeTreeIndexReader( MergeTreeReaderSettings settings) : index(index_) { - auto index_format = index->getDeserializedFormat(part_->data_part_storage, index->getFileName()); + auto index_format = index->getDeserializedFormat(part_->getDataPartStorage(), index->getFileName()); stream = makeIndexReader( index_format.extension, diff --git a/src/Storages/MergeTree/MergeTreeIndexSet.cpp b/src/Storages/MergeTree/MergeTreeIndexSet.cpp index 3c31deda823..0e15f2c4cb6 100644 --- a/src/Storages/MergeTree/MergeTreeIndexSet.cpp +++ b/src/Storages/MergeTree/MergeTreeIndexSet.cpp @@ -74,8 +74,9 @@ void MergeTreeIndexGranuleSet::serializeBinary(WriteBuffer & ostr) const auto serialization = type->getDefaultSerialization(); ISerialization::SerializeBinaryBulkStatePtr state; - serialization->serializeBinaryBulkStatePrefix(settings, state); - serialization->serializeBinaryBulkWithMultipleStreams(*block.getByPosition(i).column, 0, size(), settings, state); + const auto & column = *block.getByPosition(i).column; + serialization->serializeBinaryBulkStatePrefix(column, settings, state); + serialization->serializeBinaryBulkWithMultipleStreams(column, 0, size(), settings, state); serialization->serializeBinaryBulkStateSuffix(settings, state); } } diff --git a/src/Storages/MergeTree/MergeTreeIndexUtils.cpp b/src/Storages/MergeTree/MergeTreeIndexUtils.cpp new file mode 100644 index 00000000000..652f0c853d4 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeIndexUtils.cpp @@ -0,0 +1,47 @@ +#include + +#include +#include +#include + +namespace DB +{ + +ASTPtr buildFilterNode(const ASTPtr & select_query, ASTs additional_filters) +{ + auto & select_query_typed = select_query->as(); + + ASTs filters; + if (select_query_typed.where()) + filters.push_back(select_query_typed.where()); + + if (select_query_typed.prewhere()) + filters.push_back(select_query_typed.prewhere()); + + filters.insert(filters.end(), additional_filters.begin(), additional_filters.end()); + + if (filters.empty()) + return nullptr; + + ASTPtr filter_node; + + if (filters.size() == 1) + { + filter_node = filters.front(); + } + else + { + auto function = std::make_shared(); + + function->name = "and"; + function->arguments = std::make_shared(); + function->children.push_back(function->arguments); + function->arguments->children = std::move(filters); + + filter_node = std::move(function); + } + + return filter_node; +} + +} diff --git a/src/Storages/MergeTree/MergeTreeIndexUtils.h b/src/Storages/MergeTree/MergeTreeIndexUtils.h new file mode 100644 index 00000000000..6ba9725b564 --- /dev/null +++ b/src/Storages/MergeTree/MergeTreeIndexUtils.h @@ -0,0 +1,13 @@ +#pragma once + +#include + +namespace DB +{ + +/** Build AST filter node for index analysis from WHERE and PREWHERE sections of select query and additional filters. + * If select query does not have WHERE and PREWHERE and additional filters are empty null is returned. + */ +ASTPtr buildFilterNode(const ASTPtr & select_query, ASTs additional_filters = {}); + +} diff --git a/src/Storages/MergeTree/MergeTreeIndices.h b/src/Storages/MergeTree/MergeTreeIndices.h index 14002534c94..6a671c31944 100644 --- a/src/Storages/MergeTree/MergeTreeIndices.h +++ b/src/Storages/MergeTree/MergeTreeIndices.h @@ -148,9 +148,9 @@ struct IMergeTreeIndex /// Returns extension for deserialization. /// /// Return pair. - virtual MergeTreeIndexFormat getDeserializedFormat(const DataPartStoragePtr & data_part_storage, const std::string & relative_path_prefix) const + virtual MergeTreeIndexFormat getDeserializedFormat(const IDataPartStorage & data_part_storage, const std::string & relative_path_prefix) const { - if (data_part_storage->exists(relative_path_prefix + ".idx")) + if (data_part_storage.exists(relative_path_prefix + ".idx")) return {1, ".idx"}; return {0 /*unknown*/, ""}; } diff --git a/src/Storages/MergeTree/MergeTreePartition.cpp b/src/Storages/MergeTree/MergeTreePartition.cpp index 4ea6ec11ecc..10f5cc95baf 100644 --- a/src/Storages/MergeTree/MergeTreePartition.cpp +++ b/src/Storages/MergeTree/MergeTreePartition.cpp @@ -382,20 +382,20 @@ void MergeTreePartition::load(const MergeTreeData & storage, const PartMetadataM partition_key_sample.getByPosition(i).type->getDefaultSerialization()->deserializeBinary(value[i], *file); } -std::unique_ptr MergeTreePartition::store(const MergeTreeData & storage, const DataPartStorageBuilderPtr & data_part_storage_builder, MergeTreeDataPartChecksums & checksums) const +std::unique_ptr MergeTreePartition::store(const MergeTreeData & storage, IDataPartStorage & data_part_storage, MergeTreeDataPartChecksums & checksums) const { auto metadata_snapshot = storage.getInMemoryMetadataPtr(); const auto & context = storage.getContext(); const auto & partition_key_sample = adjustPartitionKey(metadata_snapshot, storage.getContext()).sample_block; - return store(partition_key_sample, data_part_storage_builder, checksums, context->getWriteSettings()); + return store(partition_key_sample, data_part_storage, checksums, context->getWriteSettings()); } -std::unique_ptr MergeTreePartition::store(const Block & partition_key_sample, const DataPartStorageBuilderPtr & data_part_storage_builder, MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const +std::unique_ptr MergeTreePartition::store(const Block & partition_key_sample, IDataPartStorage & data_part_storage, MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const { if (!partition_key_sample) return nullptr; - auto out = data_part_storage_builder->writeFile("partition.dat", DBMS_DEFAULT_BUFFER_SIZE, settings); + auto out = data_part_storage.writeFile("partition.dat", DBMS_DEFAULT_BUFFER_SIZE, settings); HashingWriteBuffer out_hashing(*out); for (size_t i = 0; i < value.size(); ++i) { diff --git a/src/Storages/MergeTree/MergeTreePartition.h b/src/Storages/MergeTree/MergeTreePartition.h index 6394641dfa3..78b141f26ec 100644 --- a/src/Storages/MergeTree/MergeTreePartition.h +++ b/src/Storages/MergeTree/MergeTreePartition.h @@ -15,10 +15,10 @@ class MergeTreeData; struct FormatSettings; struct MergeTreeDataPartChecksums; struct StorageInMemoryMetadata; -class IDataPartStorageBuilder; +class IDataPartStorage; using StorageMetadataPtr = std::shared_ptr; -using DataPartStorageBuilderPtr = std::shared_ptr; +using MutableDataPartStoragePtr = std::shared_ptr; /// This class represents a partition value of a single part and encapsulates its loading/storing logic. struct MergeTreePartition @@ -44,8 +44,8 @@ public: /// Store functions return write buffer with written but not finalized data. /// User must call finish() for returned object. - [[nodiscard]] std::unique_ptr store(const MergeTreeData & storage, const DataPartStorageBuilderPtr & data_part_storage_builder, MergeTreeDataPartChecksums & checksums) const; - [[nodiscard]] std::unique_ptr store(const Block & partition_key_sample, const DataPartStorageBuilderPtr & data_part_storage_builder, MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const; + [[nodiscard]] std::unique_ptr store(const MergeTreeData & storage, IDataPartStorage & data_part_storage, MergeTreeDataPartChecksums & checksums) const; + [[nodiscard]] std::unique_ptr store(const Block & partition_key_sample, IDataPartStorage & data_part_storage, MergeTreeDataPartChecksums & checksums, const WriteSettings & settings) const; void assign(const MergeTreePartition & other) { value = other.value; } diff --git a/src/Storages/MergeTree/MergeTreePartsMover.cpp b/src/Storages/MergeTree/MergeTreePartsMover.cpp index afeeacbe5d6..b618b068769 100644 --- a/src/Storages/MergeTree/MergeTreePartsMover.cpp +++ b/src/Storages/MergeTree/MergeTreePartsMover.cpp @@ -100,7 +100,6 @@ bool MergeTreePartsMover::selectPartsForMove( return false; std::unordered_map need_to_move; - std::unordered_set need_to_move_disks; const auto policy = data->getStoragePolicy(); const auto & volumes = policy->getVolumes(); @@ -115,10 +114,7 @@ bool MergeTreePartsMover::selectPartsForMove( UInt64 unreserved_space = disk->getUnreservedSpace(); if (unreserved_space < required_maximum_available_space && !disk->isBroken()) - { need_to_move.emplace(disk, required_maximum_available_space - unreserved_space); - need_to_move_disks.emplace(disk); - } } } } @@ -140,8 +136,16 @@ bool MergeTreePartsMover::selectPartsForMove( auto ttl_entry = selectTTLDescriptionForTTLInfos(metadata_snapshot->getMoveTTLs(), part->ttl_infos.moves_ttl, time_of_move, true); auto to_insert = need_to_move.end(); - if (auto disk_it = part->data_part_storage->isStoredOnDisk(need_to_move_disks); disk_it != need_to_move_disks.end()) - to_insert = need_to_move.find(*disk_it); + auto part_disk_name = part->getDataPartStorage().getDiskName(); + + for (auto it = need_to_move.begin(); it != need_to_move.end(); ++it) + { + if (it->first->getName() == part_disk_name) + { + to_insert = it; + break; + } + } ReservationPtr reservation; if (ttl_entry) @@ -158,9 +162,8 @@ bool MergeTreePartsMover::selectPartsForMove( /// In order to not over-move, we need to "release" required space on this disk, /// possibly to zero. if (to_insert != need_to_move.end()) - { to_insert->second.decreaseRequiredSizeAndRemoveRedundantParts(part->getBytesOnDisk()); - } + ++parts_to_move_by_ttl_rules; parts_to_move_total_size_bytes += part->getBytesOnDisk(); } @@ -173,7 +176,7 @@ bool MergeTreePartsMover::selectPartsForMove( for (auto && move : need_to_move) { - auto min_volume_index = policy->getVolumeIndexByDisk(move.first) + 1; + auto min_volume_index = policy->getVolumeIndexByDiskName(move.first->getName()) + 1; for (auto && part : move.second.getAccumulatedParts()) { auto reservation = policy->reserve(part->getBytesOnDisk(), min_volume_index); @@ -199,7 +202,7 @@ bool MergeTreePartsMover::selectPartsForMove( return false; } -MergeTreeData::DataPartPtr MergeTreePartsMover::clonePart(const MergeTreeMoveEntry & moving_part) const +MergeTreeMutableDataPartPtr MergeTreePartsMover::clonePart(const MergeTreeMoveEntry & moving_part) const { if (moves_blocker.isCancelled()) throw Exception("Cancelled moving parts.", ErrorCodes::ABORTED); @@ -207,16 +210,15 @@ MergeTreeData::DataPartPtr MergeTreePartsMover::clonePart(const MergeTreeMoveEnt auto settings = data->getSettings(); auto part = moving_part.part; auto disk = moving_part.reserved_space->getDisk(); - LOG_DEBUG(log, "Cloning part {} from '{}' to '{}'", part->name, part->data_part_storage->getDiskName(), disk->getName()); - - DataPartStoragePtr cloned_part_storage; + LOG_DEBUG(log, "Cloning part {} from '{}' to '{}'", part->name, part->getDataPartStorage().getDiskName(), disk->getName()); + MutableDataPartStoragePtr cloned_part_storage; if (disk->supportZeroCopyReplication() && settings->allow_remote_fs_zero_copy_replication) { /// Try zero-copy replication and fallback to default copy if it's not possible moving_part.part->assertOnDisk(); String path_to_clone = fs::path(data->getRelativeDataPath()) / MergeTreeData::MOVING_DIR_NAME / ""; - String relative_path = part->data_part_storage->getPartDirectory(); + String relative_path = part->getDataPartStorage().getPartDirectory(); if (disk->exists(path_to_clone + relative_path)) { LOG_WARNING(log, "Path {} already exists. Will remove it and clone again.", fullPath(disk, path_to_clone + relative_path)); @@ -230,7 +232,7 @@ MergeTreeData::DataPartPtr MergeTreePartsMover::clonePart(const MergeTreeMoveEnt if (!cloned_part_storage) { LOG_INFO(log, "Part {} was not fetched, we are the first who move it to another disk, so we will copy it", part->name); - cloned_part_storage = part->data_part_storage->clone(path_to_clone, part->data_part_storage->getPartDirectory(), disk, log); + cloned_part_storage = part->getDataPartStorage().clonePart(path_to_clone, part->getDataPartStorage().getPartDirectory(), disk, log); } } else @@ -238,18 +240,17 @@ MergeTreeData::DataPartPtr MergeTreePartsMover::clonePart(const MergeTreeMoveEnt cloned_part_storage = part->makeCloneOnDisk(disk, MergeTreeData::MOVING_DIR_NAME); } - MergeTreeData::MutableDataPartPtr cloned_part = data->createPart(part->name, cloned_part_storage); - LOG_TRACE(log, "Part {} was cloned to {}", part->name, cloned_part->data_part_storage->getFullPath()); + auto cloned_part = data->createPart(part->name, cloned_part_storage); + LOG_TRACE(log, "Part {} was cloned to {}", part->name, cloned_part->getDataPartStorage().getFullPath()); cloned_part->loadColumnsChecksumsIndexes(true, true); cloned_part->loadVersionMetadata(); - cloned_part->modification_time = cloned_part->data_part_storage->getLastModified().epochTime(); + cloned_part->modification_time = cloned_part->getDataPartStorage().getLastModified().epochTime(); return cloned_part; - } -void MergeTreePartsMover::swapClonedPart(const MergeTreeData::DataPartPtr & cloned_part) const +void MergeTreePartsMover::swapClonedPart(const MergeTreeMutableDataPartPtr & cloned_part) const { if (moves_blocker.isCancelled()) throw Exception("Cancelled moving parts.", ErrorCodes::ABORTED); @@ -259,20 +260,17 @@ void MergeTreePartsMover::swapClonedPart(const MergeTreeData::DataPartPtr & clon /// It's ok, because we don't block moving parts for merges or mutations if (!active_part || active_part->name != cloned_part->name) { - LOG_INFO(log, "Failed to swap {}. Active part doesn't exist. Possible it was merged or mutated. Will remove copy on path '{}'.", cloned_part->name, cloned_part->data_part_storage->getFullPath()); + LOG_INFO(log, "Failed to swap {}. Active part doesn't exist. Possible it was merged or mutated. Will remove copy on path '{}'.", cloned_part->name, cloned_part->getDataPartStorage().getFullPath()); return; } - auto builder = cloned_part->data_part_storage->getBuilder(); /// Don't remove new directory but throw an error because it may contain part which is currently in use. - cloned_part->renameTo(active_part->name, false, builder); - - builder->commit(); + cloned_part->renameTo(active_part->name, false); /// TODO what happen if server goes down here? data->swapActivePart(cloned_part); - LOG_TRACE(log, "Part {} was moved to {}", cloned_part->name, cloned_part->data_part_storage->getFullPath()); + LOG_TRACE(log, "Part {} was moved to {}", cloned_part->name, cloned_part->getDataPartStorage().getFullPath()); } } diff --git a/src/Storages/MergeTree/MergeTreePartsMover.h b/src/Storages/MergeTree/MergeTreePartsMover.h index 6ad658c2cb3..0266b2daa46 100644 --- a/src/Storages/MergeTree/MergeTreePartsMover.h +++ b/src/Storages/MergeTree/MergeTreePartsMover.h @@ -50,14 +50,14 @@ public: const std::lock_guard & moving_parts_lock); /// Copies part to selected reservation in detached folder. Throws exception if part already exists. - MergeTreeDataPartPtr clonePart(const MergeTreeMoveEntry & moving_part) const; + MergeTreeMutableDataPartPtr clonePart(const MergeTreeMoveEntry & moving_part) const; /// Replaces cloned part from detached directory into active data parts set. /// Replacing part changes state to DeleteOnDestroy and will be removed from disk after destructor of ///IMergeTreeDataPart called. If replacing part doesn't exists or not active (committed) than /// cloned part will be removed and log message will be reported. It may happen in case of concurrent /// merge or mutation. - void swapClonedPart(const MergeTreeDataPartPtr & cloned_parts) const; + void swapClonedPart(const MergeTreeMutableDataPartPtr & cloned_parts) const; /// Can stop background moves and moves from queries ActionBlocker moves_blocker; diff --git a/src/Storages/MergeTree/MergeTreeReadPool.cpp b/src/Storages/MergeTree/MergeTreeReadPool.cpp index 3f51673a6b1..ca9cde0ae61 100644 --- a/src/Storages/MergeTree/MergeTreeReadPool.cpp +++ b/src/Storages/MergeTree/MergeTreeReadPool.cpp @@ -263,7 +263,7 @@ void MergeTreeReadPool::fillPerThreadInfo( { PartInfo part_info{parts[i], per_part_sum_marks[i], i}; if (parts[i].data_part->isStoredOnDisk()) - parts_per_disk[parts[i].data_part->data_part_storage->getDiskName()].push_back(std::move(part_info)); + parts_per_disk[parts[i].data_part->getDataPartStorage().getDiskName()].push_back(std::move(part_info)); else parts_per_disk[""].push_back(std::move(part_info)); } diff --git a/src/Storages/MergeTree/MergeTreeReaderCompact.cpp b/src/Storages/MergeTree/MergeTreeReaderCompact.cpp index 4801c9a4058..b0488d29f8e 100644 --- a/src/Storages/MergeTree/MergeTreeReaderCompact.cpp +++ b/src/Storages/MergeTree/MergeTreeReaderCompact.cpp @@ -59,13 +59,15 @@ MergeTreeReaderCompact::MergeTreeReaderCompact( throw Exception(ErrorCodes::CANNOT_READ_ALL_DATA, "Cannot read to empty buffer."); const String path = MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION; + auto data_part_storage = data_part_info_for_read->getDataPartStorage(); + if (uncompressed_cache) { auto buffer = std::make_unique( - std::string(fs::path(data_part_info_for_read->getDataPartStorage()->getFullPath()) / path), - [this, path]() + std::string(fs::path(data_part_storage->getFullPath()) / path), + [this, path, data_part_storage]() { - return data_part_info_for_read->getDataPartStorage()->readFile( + return data_part_storage->readFile( path, settings.read_settings, std::nullopt, std::nullopt); @@ -87,7 +89,7 @@ MergeTreeReaderCompact::MergeTreeReaderCompact( { auto buffer = std::make_unique( - data_part_info_for_read->getDataPartStorage()->readFile( + data_part_storage->readFile( path, settings.read_settings, std::nullopt, std::nullopt), diff --git a/src/Storages/MergeTree/MergeTreeSink.cpp b/src/Storages/MergeTree/MergeTreeSink.cpp index 5d00db861a8..13a72c24c59 100644 --- a/src/Storages/MergeTree/MergeTreeSink.cpp +++ b/src/Storages/MergeTree/MergeTreeSink.cpp @@ -1,8 +1,8 @@ #include #include #include -#include #include +#include namespace ProfileEvents { @@ -56,8 +56,9 @@ struct MergeTreeSink::DelayedChunk void MergeTreeSink::consume(Chunk chunk) { auto block = getHeader().cloneWithColumns(chunk.detachColumns()); + if (!storage_snapshot->object_columns.empty()) + convertDynamicColumnsToTuples(block, storage_snapshot); - deduceTypesOfObjectColumns(storage_snapshot, block); auto part_blocks = storage.writer.splitBlockIntoParts(block, max_parts_per_block, metadata_snapshot, context); using DelayedPartitions = std::vector; @@ -81,7 +82,7 @@ void MergeTreeSink::consume(Chunk chunk) if (!temp_part.part) continue; - if (!support_parallel_write && temp_part.part->data_part_storage->supportParallelWrite()) + if (!support_parallel_write && temp_part.part->getDataPartStorage().supportParallelWrite()) support_parallel_write = true; if (storage.getDeduplicationLog()) @@ -160,7 +161,7 @@ void MergeTreeSink::finishDelayedChunk() } } - added = storage.renameTempPartAndAdd(part, transaction, partition.temp_part.builder, lock); + added = storage.renameTempPartAndAdd(part, transaction, lock); transaction.commit(&lock); } diff --git a/src/Storages/MergeTree/MergeTreeWriteAheadLog.cpp b/src/Storages/MergeTree/MergeTreeWriteAheadLog.cpp index 4735eae8fdd..b3625ba8e93 100644 --- a/src/Storages/MergeTree/MergeTreeWriteAheadLog.cpp +++ b/src/Storages/MergeTree/MergeTreeWriteAheadLog.cpp @@ -150,7 +150,6 @@ MergeTreeData::MutableDataPartsVector MergeTreeWriteAheadLog::restore( while (!in->eof()) { MergeTreeData::MutableDataPartPtr part; - DataPartStorageBuilderPtr data_part_storage_builder; UInt8 version; String part_name; Block block; @@ -177,7 +176,6 @@ MergeTreeData::MutableDataPartsVector MergeTreeWriteAheadLog::restore( { auto single_disk_volume = std::make_shared("volume_" + part_name, disk, 0); auto data_part_storage = std::make_shared(single_disk_volume, storage.getRelativeDataPath(), part_name); - data_part_storage_builder = std::make_shared(single_disk_volume, storage.getRelativeDataPath(), part_name); part = storage.createPart( part_name, @@ -222,7 +220,6 @@ MergeTreeData::MutableDataPartsVector MergeTreeWriteAheadLog::restore( { MergedBlockOutputStream part_out( part, - data_part_storage_builder, metadata_snapshot, block.getNamesAndTypesList(), {}, @@ -240,11 +237,12 @@ MergeTreeData::MutableDataPartsVector MergeTreeWriteAheadLog::restore( for (const auto & projection : metadata_snapshot->getProjections()) { auto projection_block = projection.calculate(block, context); - auto temp_part = MergeTreeDataWriter::writeInMemoryProjectionPart(storage, log, projection_block, projection, data_part_storage_builder, part.get()); + auto temp_part = MergeTreeDataWriter::writeProjectionPart(storage, log, projection_block, projection, part.get()); temp_part.finalize(); if (projection_block.rows()) part->addProjectionPart(projection.name, std::move(temp_part.part)); } + part_out.finalizePart(part, false); min_block_number = std::min(min_block_number, part->info.min_block); diff --git a/src/Storages/MergeTree/MergedBlockOutputStream.cpp b/src/Storages/MergeTree/MergedBlockOutputStream.cpp index 269a78977ad..991a8d359a8 100644 --- a/src/Storages/MergeTree/MergedBlockOutputStream.cpp +++ b/src/Storages/MergeTree/MergedBlockOutputStream.cpp @@ -14,8 +14,7 @@ namespace ErrorCodes MergedBlockOutputStream::MergedBlockOutputStream( - const MergeTreeDataPartPtr & data_part, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const NamesAndTypesList & columns_list_, const MergeTreeIndices & skip_indices, @@ -24,7 +23,7 @@ MergedBlockOutputStream::MergedBlockOutputStream( bool reset_columns_, bool blocks_are_granules_size, const WriteSettings & write_settings_) - : IMergedBlockOutputStream(std::move(data_part_storage_builder_), data_part, metadata_snapshot_, columns_list_, reset_columns_) + : IMergedBlockOutputStream(data_part, metadata_snapshot_, columns_list_, reset_columns_) , columns_list(columns_list_) , default_codec(default_codec_) , write_settings(write_settings_) @@ -38,7 +37,7 @@ MergedBlockOutputStream::MergedBlockOutputStream( blocks_are_granules_size); if (data_part->isStoredOnDisk()) - data_part_storage_builder->createDirectories(); + data_part_storage->createDirectories(); /// We should write version metadata on part creation to distinguish it from parts that were created without transaction. TransactionID tid = txn ? txn->tid : Tx::PrehistoricTID; @@ -47,7 +46,7 @@ MergedBlockOutputStream::MergedBlockOutputStream( data_part->version.setCreationTID(tid, nullptr); data_part->storeVersionMetadata(); - writer = data_part->getWriter(data_part_storage_builder, columns_list, metadata_snapshot, skip_indices, default_codec, writer_settings, {}); + writer = data_part->getWriter(columns_list, metadata_snapshot, skip_indices, default_codec, writer_settings, {}); } /// If data is pre-sorted. @@ -68,17 +67,17 @@ struct MergedBlockOutputStream::Finalizer::Impl { IMergeTreeDataPartWriter & writer; MergeTreeData::MutableDataPartPtr part; - DataPartStorageBuilderPtr data_part_storage_builder; NameSet files_to_remove_after_finish; std::vector> written_files; bool sync; - Impl(IMergeTreeDataPartWriter & writer_, MergeTreeData::MutableDataPartPtr part_, DataPartStorageBuilderPtr data_part_storage_builder_, const NameSet & files_to_remove_after_finish_, bool sync_) + Impl(IMergeTreeDataPartWriter & writer_, MergeTreeData::MutableDataPartPtr part_, const NameSet & files_to_remove_after_finish_, bool sync_) : writer(writer_) , part(std::move(part_)) - , data_part_storage_builder(std::move(data_part_storage_builder_)) , files_to_remove_after_finish(files_to_remove_after_finish_) - , sync(sync_) {} + , sync(sync_) + { + } void finish(); }; @@ -95,7 +94,7 @@ void MergedBlockOutputStream::Finalizer::Impl::finish() writer.finish(sync); for (const auto & file_name : files_to_remove_after_finish) - data_part_storage_builder->removeFile(file_name); + part->getDataPartStorage().removeFile(file_name); for (auto & file : written_files) { @@ -122,19 +121,19 @@ MergedBlockOutputStream::Finalizer & MergedBlockOutputStream::Finalizer::operato MergedBlockOutputStream::Finalizer::Finalizer(std::unique_ptr impl_) : impl(std::move(impl_)) {} void MergedBlockOutputStream::finalizePart( - MergeTreeData::MutableDataPartPtr & new_part, - bool sync, - const NamesAndTypesList * total_columns_list, - MergeTreeData::DataPart::Checksums * additional_column_checksums) + const MergeTreeMutableDataPartPtr & new_part, + bool sync, + const NamesAndTypesList * total_columns_list, + MergeTreeData::DataPart::Checksums * additional_column_checksums) { finalizePartAsync(new_part, sync, total_columns_list, additional_column_checksums).finish(); } MergedBlockOutputStream::Finalizer MergedBlockOutputStream::finalizePartAsync( - MergeTreeData::MutableDataPartPtr & new_part, - bool sync, - const NamesAndTypesList * total_columns_list, - MergeTreeData::DataPart::Checksums * additional_column_checksums) + const MergeTreeMutableDataPartPtr & new_part, + bool sync, + const NamesAndTypesList * total_columns_list, + MergeTreeData::DataPart::Checksums * additional_column_checksums) { /// Finish write and get checksums. MergeTreeData::DataPart::Checksums checksums; @@ -165,7 +164,7 @@ MergedBlockOutputStream::Finalizer MergedBlockOutputStream::finalizePartAsync( new_part->setColumns(part_columns, serialization_infos); } - auto finalizer = std::make_unique(*writer, new_part, data_part_storage_builder, files_to_remove_after_sync, sync); + auto finalizer = std::make_unique(*writer, new_part, files_to_remove_after_sync, sync); if (new_part->isStoredOnDisk()) finalizer->written_files = finalizePartOnDisk(new_part, checksums); @@ -184,7 +183,7 @@ MergedBlockOutputStream::Finalizer MergedBlockOutputStream::finalizePartAsync( } MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDisk( - const MergeTreeData::DataPartPtr & new_part, + const MergeTreeMutableDataPartPtr & new_part, MergeTreeData::DataPart::Checksums & checksums) { WrittenFiles written_files; @@ -192,7 +191,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis { if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING || isCompactPart(new_part)) { - auto count_out = data_part_storage_builder->writeFile("count.txt", 4096, write_settings); + auto count_out = new_part->getDataPartStorage().writeFile("count.txt", 4096, write_settings); HashingWriteBuffer count_out_hashing(*count_out); writeIntText(rows_count, count_out_hashing); count_out_hashing.next(); @@ -206,7 +205,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis { if (new_part->uuid != UUIDHelpers::Nil) { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::UUID_FILE_NAME, 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile(IMergeTreeDataPart::UUID_FILE_NAME, 4096, write_settings); HashingWriteBuffer out_hashing(*out); writeUUIDText(new_part->uuid, out_hashing); checksums.files[IMergeTreeDataPart::UUID_FILE_NAME].file_size = out_hashing.count(); @@ -217,12 +216,12 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis if (storage.format_version >= MERGE_TREE_DATA_MIN_FORMAT_VERSION_WITH_CUSTOM_PARTITIONING) { - if (auto file = new_part->partition.store(storage, data_part_storage_builder, checksums)) + if (auto file = new_part->partition.store(storage, new_part->getDataPartStorage(), checksums)) written_files.emplace_back(std::move(file)); if (new_part->minmax_idx->initialized) { - auto files = new_part->minmax_idx->store(storage, data_part_storage_builder, checksums); + auto files = new_part->minmax_idx->store(storage, new_part->getDataPartStorage(), checksums); for (auto & file : files) written_files.emplace_back(std::move(file)); } @@ -232,7 +231,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis } { - auto count_out = data_part_storage_builder->writeFile("count.txt", 4096, write_settings); + auto count_out = new_part->getDataPartStorage().writeFile("count.txt", 4096, write_settings); HashingWriteBuffer count_out_hashing(*count_out); writeIntText(rows_count, count_out_hashing); count_out_hashing.next(); @@ -246,7 +245,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis if (!new_part->ttl_infos.empty()) { /// Write a file with ttl infos in json format. - auto out = data_part_storage_builder->writeFile("ttl.txt", 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile("ttl.txt", 4096, write_settings); HashingWriteBuffer out_hashing(*out); new_part->ttl_infos.write(out_hashing); checksums.files["ttl.txt"].file_size = out_hashing.count(); @@ -257,7 +256,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis if (!new_part->getSerializationInfos().empty()) { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, 4096, write_settings); HashingWriteBuffer out_hashing(*out); new_part->getSerializationInfos().writeJSON(out_hashing); checksums.files[IMergeTreeDataPart::SERIALIZATION_FILE_NAME].file_size = out_hashing.count(); @@ -268,7 +267,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis { /// Write a file with a description of columns. - auto out = data_part_storage_builder->writeFile("columns.txt", 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile("columns.txt", 4096, write_settings); new_part->getColumns().writeText(*out); out->preFinalize(); written_files.emplace_back(std::move(out)); @@ -276,7 +275,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis if (default_codec != nullptr) { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME, 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile(IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME, 4096, write_settings); DB::writeText(queryToString(default_codec->getFullCodecDesc()), *out); out->preFinalize(); written_files.emplace_back(std::move(out)); @@ -289,7 +288,7 @@ MergedBlockOutputStream::WrittenFiles MergedBlockOutputStream::finalizePartOnDis { /// Write file with checksums. - auto out = data_part_storage_builder->writeFile("checksums.txt", 4096, write_settings); + auto out = new_part->getDataPartStorage().writeFile("checksums.txt", 4096, write_settings); checksums.write(*out); out->preFinalize(); written_files.emplace_back(std::move(out)); diff --git a/src/Storages/MergeTree/MergedBlockOutputStream.h b/src/Storages/MergeTree/MergedBlockOutputStream.h index 92dcd8dd272..ad1bb584788 100644 --- a/src/Storages/MergeTree/MergedBlockOutputStream.h +++ b/src/Storages/MergeTree/MergedBlockOutputStream.h @@ -15,8 +15,7 @@ class MergedBlockOutputStream final : public IMergedBlockOutputStream { public: MergedBlockOutputStream( - const MergeTreeDataPartPtr & data_part, - DataPartStorageBuilderPtr data_part_storage_builder_, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const NamesAndTypesList & columns_list_, const MergeTreeIndices & skip_indices, @@ -55,16 +54,16 @@ public: /// Finalize writing part and fill inner structures /// If part is new and contains projections, they should be added before invoking this method. Finalizer finalizePartAsync( - MergeTreeData::MutableDataPartPtr & new_part, - bool sync, - const NamesAndTypesList * total_columns_list = nullptr, - MergeTreeData::DataPart::Checksums * additional_column_checksums = nullptr); + const MergeTreeMutableDataPartPtr & new_part, + bool sync, + const NamesAndTypesList * total_columns_list = nullptr, + MergeTreeData::DataPart::Checksums * additional_column_checksums = nullptr); void finalizePart( - MergeTreeData::MutableDataPartPtr & new_part, - bool sync, - const NamesAndTypesList * total_columns_list = nullptr, - MergeTreeData::DataPart::Checksums * additional_column_checksums = nullptr); + const MergeTreeMutableDataPartPtr & new_part, + bool sync, + const NamesAndTypesList * total_columns_list = nullptr, + MergeTreeData::DataPart::Checksums * additional_column_checksums = nullptr); private: /** If `permutation` is given, it rearranges the values in the columns when writing. @@ -74,8 +73,8 @@ private: using WrittenFiles = std::vector>; WrittenFiles finalizePartOnDisk( - const MergeTreeData::DataPartPtr & new_part, - MergeTreeData::DataPart::Checksums & checksums); + const MergeTreeMutableDataPartPtr & new_part, + MergeTreeData::DataPart::Checksums & checksums); NamesAndTypesList columns_list; IMergeTreeDataPart::MinMaxIndex minmax_idx; diff --git a/src/Storages/MergeTree/MergedColumnOnlyOutputStream.cpp b/src/Storages/MergeTree/MergedColumnOnlyOutputStream.cpp index dd75cddd380..e4a5a0bc3ba 100644 --- a/src/Storages/MergeTree/MergedColumnOnlyOutputStream.cpp +++ b/src/Storages/MergeTree/MergedColumnOnlyOutputStream.cpp @@ -11,8 +11,7 @@ namespace ErrorCodes } MergedColumnOnlyOutputStream::MergedColumnOnlyOutputStream( - DataPartStorageBuilderPtr data_part_storage_builder_, - const MergeTreeDataPartPtr & data_part, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const Block & header_, CompressionCodecPtr default_codec, @@ -20,7 +19,7 @@ MergedColumnOnlyOutputStream::MergedColumnOnlyOutputStream( WrittenOffsetColumns * offset_columns_, const MergeTreeIndexGranularity & index_granularity, const MergeTreeIndexGranularityInfo * index_granularity_info) - : IMergedBlockOutputStream(std::move(data_part_storage_builder_), data_part, metadata_snapshot_, header_.getNamesAndTypesList(), /*reset_columns=*/ true) + : IMergedBlockOutputStream(data_part, metadata_snapshot_, header_.getNamesAndTypesList(), /*reset_columns=*/ true) , header(header_) { const auto & global_settings = data_part->storage.getContext()->getSettings(); @@ -34,7 +33,6 @@ MergedColumnOnlyOutputStream::MergedColumnOnlyOutputStream( /* rewrite_primary_key = */ false); writer = data_part->getWriter( - data_part_storage_builder, header.getNamesAndTypesList(), metadata_snapshot_, indices_to_recalc, @@ -81,7 +79,7 @@ MergedColumnOnlyOutputStream::fillChecksums( for (const String & removed_file : removed_files) { - data_part_storage_builder->removeFileIfExists(removed_file); + new_part->getDataPartStorage().removeFileIfExists(removed_file); if (all_checksums.files.contains(removed_file)) all_checksums.files.erase(removed_file); diff --git a/src/Storages/MergeTree/MergedColumnOnlyOutputStream.h b/src/Storages/MergeTree/MergedColumnOnlyOutputStream.h index 1fd1c752226..f382b0fef60 100644 --- a/src/Storages/MergeTree/MergedColumnOnlyOutputStream.h +++ b/src/Storages/MergeTree/MergedColumnOnlyOutputStream.h @@ -14,8 +14,7 @@ public: /// Pass empty 'already_written_offset_columns' first time then and pass the same object to subsequent instances of MergedColumnOnlyOutputStream /// if you want to serialize elements of Nested data structure in different instances of MergedColumnOnlyOutputStream. MergedColumnOnlyOutputStream( - DataPartStorageBuilderPtr data_part_storage_builder_, - const MergeTreeDataPartPtr & data_part, + const MergeTreeMutableDataPartPtr & data_part, const StorageMetadataPtr & metadata_snapshot_, const Block & header_, CompressionCodecPtr default_codec_, diff --git a/src/Storages/MergeTree/MutateFromLogEntryTask.cpp b/src/Storages/MergeTree/MutateFromLogEntryTask.cpp index 549c4e7373f..9e3cbb0640b 100644 --- a/src/Storages/MergeTree/MutateFromLogEntryTask.cpp +++ b/src/Storages/MergeTree/MutateFromLogEntryTask.cpp @@ -92,7 +92,7 @@ ReplicatedMergeMutateTaskBase::PrepareResult MutateFromLogEntryTask::prepare() /// Once we mutate part, we must reserve space on the same disk, because mutations can possibly create hardlinks. /// Can throw an exception. - reserved_space = storage.reserveSpace(estimated_space_for_result, source_part->data_part_storage); + reserved_space = storage.reserveSpace(estimated_space_for_result, source_part->getDataPartStorage()); table_lock_holder = storage.lockForShare( RWLockImpl::NO_QUERY, storage_settings_ptr->lock_acquire_timeout_for_background_operations); @@ -193,12 +193,7 @@ ReplicatedMergeMutateTaskBase::PrepareResult MutateFromLogEntryTask::prepare() bool MutateFromLogEntryTask::finalize(ReplicatedMergeMutateTaskBase::PartLogWriter write_part_log) { new_part = mutate_task->getFuture().get(); - auto builder = mutate_task->getBuilder(); - - if (!builder) - builder = new_part->data_part_storage->getBuilder(); - - storage.renameTempPartAndReplace(new_part, *transaction_ptr, builder); + storage.renameTempPartAndReplace(new_part, *transaction_ptr); try { diff --git a/src/Storages/MergeTree/MutatePlainMergeTreeTask.cpp b/src/Storages/MergeTree/MutatePlainMergeTreeTask.cpp index 0cf10ee1935..b1714076a46 100644 --- a/src/Storages/MergeTree/MutatePlainMergeTreeTask.cpp +++ b/src/Storages/MergeTree/MutatePlainMergeTreeTask.cpp @@ -83,14 +83,9 @@ bool MutatePlainMergeTreeTask::executeStep() new_part = mutate_task->getFuture().get(); - auto builder = mutate_task->getBuilder(); - if (!builder) - builder = new_part->data_part_storage->getBuilder(); - - MergeTreeData::Transaction transaction(storage, merge_mutate_entry->txn.get()); /// FIXME Transactions: it's too optimistic, better to lock parts before starting transaction - storage.renameTempPartAndReplace(new_part, transaction, builder); + storage.renameTempPartAndReplace(new_part, transaction); transaction.commit(); storage.updateMutationEntriesErrors(future_part, true, ""); diff --git a/src/Storages/MergeTree/MutateTask.cpp b/src/Storages/MergeTree/MutateTask.cpp index 3d964e60798..e5ba771a198 100644 --- a/src/Storages/MergeTree/MutateTask.cpp +++ b/src/Storages/MergeTree/MutateTask.cpp @@ -619,7 +619,6 @@ static NameToNameVector collectFilesForRenames( /// Initialize and write to disk new part fields like checksums, columns, etc. void finalizeMutatedPart( const MergeTreeDataPartPtr & source_part, - const DataPartStorageBuilderPtr & data_part_storage_builder, MergeTreeData::MutableDataPartPtr new_data_part, ExecuteTTLType execute_ttl_type, const CompressionCodecPtr & codec, @@ -627,7 +626,7 @@ void finalizeMutatedPart( { if (new_data_part->uuid != UUIDHelpers::Nil) { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::UUID_FILE_NAME, 4096, context->getWriteSettings()); + auto out = new_data_part->getDataPartStorage().writeFile(IMergeTreeDataPart::UUID_FILE_NAME, 4096, context->getWriteSettings()); HashingWriteBuffer out_hashing(*out); writeUUIDText(new_data_part->uuid, out_hashing); new_data_part->checksums.files[IMergeTreeDataPart::UUID_FILE_NAME].file_size = out_hashing.count(); @@ -637,7 +636,7 @@ void finalizeMutatedPart( if (execute_ttl_type != ExecuteTTLType::NONE) { /// Write a file with ttl infos in json format. - auto out_ttl = data_part_storage_builder->writeFile("ttl.txt", 4096, context->getWriteSettings()); + auto out_ttl = new_data_part->getDataPartStorage().writeFile("ttl.txt", 4096, context->getWriteSettings()); HashingWriteBuffer out_hashing(*out_ttl); new_data_part->ttl_infos.write(out_hashing); new_data_part->checksums.files["ttl.txt"].file_size = out_hashing.count(); @@ -646,7 +645,7 @@ void finalizeMutatedPart( if (!new_data_part->getSerializationInfos().empty()) { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, 4096, context->getWriteSettings()); + auto out = new_data_part->getDataPartStorage().writeFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, 4096, context->getWriteSettings()); HashingWriteBuffer out_hashing(*out); new_data_part->getSerializationInfos().writeJSON(out_hashing); new_data_part->checksums.files[IMergeTreeDataPart::SERIALIZATION_FILE_NAME].file_size = out_hashing.count(); @@ -655,18 +654,18 @@ void finalizeMutatedPart( { /// Write file with checksums. - auto out_checksums = data_part_storage_builder->writeFile("checksums.txt", 4096, context->getWriteSettings()); + auto out_checksums = new_data_part->getDataPartStorage().writeFile("checksums.txt", 4096, context->getWriteSettings()); new_data_part->checksums.write(*out_checksums); } /// close fd { - auto out = data_part_storage_builder->writeFile(IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME, 4096, context->getWriteSettings()); + auto out = new_data_part->getDataPartStorage().writeFile(IMergeTreeDataPart::DEFAULT_COMPRESSION_CODEC_FILE_NAME, 4096, context->getWriteSettings()); DB::writeText(queryToString(codec->getFullCodecDesc()), *out); } /// close fd { /// Write a file with a description of columns. - auto out_columns = data_part_storage_builder->writeFile("columns.txt", 4096, context->getWriteSettings()); + auto out_columns = new_data_part->getDataPartStorage().writeFile("columns.txt", 4096, context->getWriteSettings()); new_data_part->getColumns().writeText(*out_columns); } /// close fd @@ -734,8 +733,6 @@ struct MutationContext = MutationsInterpreter::MutationKind::MutationKindEnum::MUTATE_UNKNOWN; MergeTreeData::MutableDataPartPtr new_data_part; - DataPartStorageBuilderPtr data_part_storage_builder; - IMergedBlockOutputStreamPtr out{nullptr}; String mrk_extension; @@ -816,11 +813,9 @@ public: if (next_level_parts.empty()) { LOG_DEBUG(log, "Merged a projection part in level {}", current_level); - auto builder = selected_parts[0]->data_part_storage->getBuilder(); - selected_parts[0]->renameTo(projection.name + ".proj", true, builder); + selected_parts[0]->renameTo(projection.name + ".proj", true); selected_parts[0]->name = projection.name; selected_parts[0]->is_temp = false; - builder->commit(); ctx->new_data_part->addProjectionPart(name, std::move(selected_parts[0])); /// Task is finished @@ -865,7 +860,6 @@ public: projection_merging_params, NO_TRANSACTION_PTR, ctx->new_data_part.get(), - ctx->data_part_storage_builder.get(), ".tmp_proj"); next_level_parts.push_back(executeHere(tmp_part_merge_task)); @@ -1025,8 +1019,7 @@ bool PartMergerWriter::mutateOriginalPartAndPrepareProjections() if (projection_block) { auto tmp_part = MergeTreeDataWriter::writeTempProjectionPart( - *ctx->data, ctx->log, projection_block, projection, ctx->data_part_storage_builder, ctx->new_data_part.get(), ++block_num); - tmp_part.builder->commit(); + *ctx->data, ctx->log, projection_block, projection, ctx->new_data_part.get(), ++block_num); tmp_part.finalize(); projection_parts[projection.name].emplace_back(std::move(tmp_part.part)); } @@ -1048,8 +1041,7 @@ bool PartMergerWriter::mutateOriginalPartAndPrepareProjections() if (projection_block) { auto temp_part = MergeTreeDataWriter::writeTempProjectionPart( - *ctx->data, ctx->log, projection_block, projection, ctx->data_part_storage_builder, ctx->new_data_part.get(), ++block_num); - temp_part.builder->commit(); + *ctx->data, ctx->log, projection_block, projection, ctx->new_data_part.get(), ++block_num); temp_part.finalize(); projection_parts[projection.name].emplace_back(std::move(temp_part.part)); } @@ -1149,7 +1141,7 @@ private: void prepare() { - ctx->data_part_storage_builder->createDirectories(); + ctx->new_data_part->getDataPartStorage().createDirectories(); /// Note: this is done before creating input streams, because otherwise data.data_parts_mutex /// (which is locked in data.getTotalActiveSizeInBytes()) @@ -1184,7 +1176,6 @@ private: ctx->out = std::make_shared( ctx->new_data_part, - ctx->data_part_storage_builder, ctx->metadata_snapshot, ctx->new_data_part->getColumns(), skip_part_indices, @@ -1280,7 +1271,7 @@ private: if (ctx->execute_ttl_type != ExecuteTTLType::NONE) ctx->files_to_skip.insert("ttl.txt"); - ctx->data_part_storage_builder->createDirectories(); + ctx->new_data_part->getDataPartStorage().createDirectories(); /// We should write version metadata on part creation to distinguish it from parts that were created without transaction. TransactionID tid = ctx->txn ? ctx->txn->tid : Tx::PrehistoricTID; @@ -1291,7 +1282,7 @@ private: NameSet hardlinked_files; /// Create hardlinks for unchanged files - for (auto it = ctx->source_part->data_part_storage->iterate(); it->isValid(); it->next()) + for (auto it = ctx->source_part->getDataPartStorage().iterate(); it->isValid(); it->next()) { if (ctx->files_to_skip.contains(it->name())) continue; @@ -1317,22 +1308,22 @@ private: if (it->isFile()) { - ctx->data_part_storage_builder->createHardLinkFrom( - *ctx->source_part->data_part_storage, it->name(), destination); + ctx->new_data_part->getDataPartStorage().createHardLinkFrom( + ctx->source_part->getDataPartStorage(), it->name(), destination); hardlinked_files.insert(it->name()); } else if (!endsWith(it->name(), ".tmp_proj")) // ignore projection tmp merge dir { // it's a projection part directory - ctx->data_part_storage_builder->createProjection(destination); + ctx->new_data_part->getDataPartStorage().createProjection(destination); - auto projection_data_part_storage = ctx->source_part->data_part_storage->getProjection(destination); - auto projection_data_part_storage_builder = ctx->data_part_storage_builder->getProjection(destination); + auto projection_data_part_storage_src = ctx->source_part->getDataPartStorage().getProjection(destination); + auto projection_data_part_storage_dst = ctx->new_data_part->getDataPartStorage().getProjection(destination); - for (auto p_it = projection_data_part_storage->iterate(); p_it->isValid(); p_it->next()) + for (auto p_it = projection_data_part_storage_src->iterate(); p_it->isValid(); p_it->next()) { - projection_data_part_storage_builder->createHardLinkFrom( - *projection_data_part_storage, p_it->name(), p_it->name()); + projection_data_part_storage_dst->createHardLinkFrom( + *projection_data_part_storage_src, p_it->name(), p_it->name()); hardlinked_files.insert(p_it->name()); } } @@ -1362,7 +1353,6 @@ private: builder.addTransform(std::make_shared(builder.getHeader(), *ctx->data, ctx->metadata_snapshot, ctx->new_data_part, ctx->time_of_mutation, true)); ctx->out = std::make_shared( - ctx->data_part_storage_builder, ctx->new_data_part, ctx->metadata_snapshot, ctx->updated_header, @@ -1414,7 +1404,7 @@ private: } } - MutationHelpers::finalizeMutatedPart(ctx->source_part, ctx->data_part_storage_builder, ctx->new_data_part, ctx->execute_ttl_type, ctx->compression_codec, ctx->context); + MutationHelpers::finalizeMutatedPart(ctx->source_part, ctx->new_data_part, ctx->execute_ttl_type, ctx->compression_codec, ctx->context); } @@ -1584,10 +1574,7 @@ bool MutateTask::prepare() ctx->data->getRelativeDataPath(), tmp_part_dir_name); - ctx->data_part_storage_builder = std::make_shared( - single_disk_volume, - ctx->data->getRelativeDataPath(), - tmp_part_dir_name); + data_part_storage->beginTransaction(); ctx->new_data_part = ctx->data->createPart( ctx->future_part->name, ctx->future_part->type, ctx->future_part->part_info, data_part_storage); @@ -1690,9 +1677,4 @@ const MergeTreeData::HardlinkedFiles & MutateTask::getHardlinkedFiles() const return ctx->hardlinked_files; } -DataPartStorageBuilderPtr MutateTask::getBuilder() const -{ - return ctx->data_part_storage_builder; -} - } diff --git a/src/Storages/MergeTree/MutateTask.h b/src/Storages/MergeTree/MutateTask.h index 1f2e8a6fd20..3df30670d7f 100644 --- a/src/Storages/MergeTree/MutateTask.h +++ b/src/Storages/MergeTree/MutateTask.h @@ -46,7 +46,7 @@ public: const MergeTreeData::HardlinkedFiles & getHardlinkedFiles() const; - DataPartStorageBuilderPtr getBuilder() const; + MutableDataPartStoragePtr getBuilder() const; private: diff --git a/src/Storages/MergeTree/PartMetadataManagerOrdinary.cpp b/src/Storages/MergeTree/PartMetadataManagerOrdinary.cpp index 7eb868f7754..30823d593a2 100644 --- a/src/Storages/MergeTree/PartMetadataManagerOrdinary.cpp +++ b/src/Storages/MergeTree/PartMetadataManagerOrdinary.cpp @@ -8,20 +8,10 @@ namespace DB { -static std::unique_ptr openForReading(const DataPartStoragePtr & data_part_storage, const String & path) -{ - size_t file_size = data_part_storage->getFileSize(path); - return data_part_storage->readFile(path, ReadSettings().adjustBufferSize(file_size), file_size, std::nullopt); -} - -PartMetadataManagerOrdinary::PartMetadataManagerOrdinary(const IMergeTreeDataPart * part_) : IPartMetadataManager(part_) -{ -} - - std::unique_ptr PartMetadataManagerOrdinary::read(const String & file_name) const { - auto res = openForReading(part->data_part_storage, file_name); + size_t file_size = part->getDataPartStorage().getFileSize(file_name); + auto res = part->getDataPartStorage().readFile(file_name, ReadSettings().adjustBufferSize(file_size), file_size, std::nullopt); if (isCompressedFromFileName(file_name)) return std::make_unique(std::move(res)); @@ -31,7 +21,7 @@ std::unique_ptr PartMetadataManagerOrdinary::read(const String & fil bool PartMetadataManagerOrdinary::exists(const String & file_name) const { - return part->data_part_storage->exists(file_name); + return part->getDataPartStorage().exists(file_name); } diff --git a/src/Storages/MergeTree/PartMetadataManagerOrdinary.h b/src/Storages/MergeTree/PartMetadataManagerOrdinary.h index d86d5c54c00..428b6d4710a 100644 --- a/src/Storages/MergeTree/PartMetadataManagerOrdinary.h +++ b/src/Storages/MergeTree/PartMetadataManagerOrdinary.h @@ -8,7 +8,7 @@ namespace DB class PartMetadataManagerOrdinary : public IPartMetadataManager { public: - explicit PartMetadataManagerOrdinary(const IMergeTreeDataPart * part_); + explicit PartMetadataManagerOrdinary(const IMergeTreeDataPart * part_) : IPartMetadataManager(part_) {} ~PartMetadataManagerOrdinary() override = default; diff --git a/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp b/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp index ee0970984f9..90fd25bc4e7 100644 --- a/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp +++ b/src/Storages/MergeTree/PartMetadataManagerWithCache.cpp @@ -31,24 +31,24 @@ PartMetadataManagerWithCache::PartMetadataManagerWithCache(const IMergeTreeDataP String PartMetadataManagerWithCache::getKeyFromFilePath(const String & file_path) const { - return part->data_part_storage->getDiskName() + ":" + file_path; + return part->getDataPartStorage().getDiskName() + ":" + file_path; } String PartMetadataManagerWithCache::getFilePathFromKey(const String & key) const { - return key.substr(part->data_part_storage->getDiskName().size() + 1); + return key.substr(part->getDataPartStorage().getDiskName().size() + 1); } std::unique_ptr PartMetadataManagerWithCache::read(const String & file_name) const { - String file_path = fs::path(part->data_part_storage->getRelativePath()) / file_name; + String file_path = fs::path(part->getDataPartStorage().getRelativePath()) / file_name; String key = getKeyFromFilePath(file_path); String value; auto status = cache->get(key, value); if (!status.ok()) { ProfileEvents::increment(ProfileEvents::MergeTreeMetadataCacheMiss); - auto in = part->data_part_storage->readFile(file_name, {}, std::nullopt, std::nullopt); + auto in = part->getDataPartStorage().readFile(file_name, {}, std::nullopt, std::nullopt); std::unique_ptr reader; if (!isCompressedFromFileName(file_name)) reader = std::move(in); @@ -67,7 +67,7 @@ std::unique_ptr PartMetadataManagerWithCache::read(const String & fi bool PartMetadataManagerWithCache::exists(const String & file_name) const { - String file_path = fs::path(part->data_part_storage->getRelativePath()) / file_name; + String file_path = fs::path(part->getDataPartStorage().getRelativePath()) / file_name; String key = getKeyFromFilePath(file_path); String value; auto status = cache->get(key, value); @@ -79,7 +79,7 @@ bool PartMetadataManagerWithCache::exists(const String & file_name) const else { ProfileEvents::increment(ProfileEvents::MergeTreeMetadataCacheMiss); - return part->data_part_storage->exists(file_name); + return part->getDataPartStorage().exists(file_name); } } @@ -91,7 +91,7 @@ void PartMetadataManagerWithCache::deleteAll(bool include_projection) String value; for (const auto & file_name : file_names) { - String file_path = fs::path(part->data_part_storage->getRelativePath()) / file_name; + String file_path = fs::path(part->getDataPartStorage().getRelativePath()) / file_name; String key = getKeyFromFilePath(file_path); auto status = cache->del(key); if (!status.ok()) @@ -119,10 +119,10 @@ void PartMetadataManagerWithCache::updateAll(bool include_projection) String read_value; for (const auto & file_name : file_names) { - String file_path = fs::path(part->data_part_storage->getRelativePath()) / file_name; - if (!part->data_part_storage->exists(file_name)) + String file_path = fs::path(part->getDataPartStorage().getRelativePath()) / file_name; + if (!part->getDataPartStorage().exists(file_name)) continue; - auto in = part->data_part_storage->readFile(file_name, {}, std::nullopt, std::nullopt); + auto in = part->getDataPartStorage().readFile(file_name, {}, std::nullopt, std::nullopt); readStringUntilEOF(value, *in); String key = getKeyFromFilePath(file_path); @@ -159,7 +159,7 @@ void PartMetadataManagerWithCache::assertAllDeleted(bool include_projection) con file_name = fs::path(file_path).filename(); /// Metadata file belongs to current part - if (fs::path(part->data_part_storage->getRelativePath()) / file_name == file_path) + if (fs::path(part->getDataPartStorage().getRelativePath()) / file_name == file_path) throw Exception( ErrorCodes::LOGICAL_ERROR, "Data part {} with type {} with meta file {} still in cache", @@ -173,7 +173,7 @@ void PartMetadataManagerWithCache::assertAllDeleted(bool include_projection) con const auto & projection_parts = part->getProjectionParts(); for (const auto & [projection_name, projection_part] : projection_parts) { - if (fs::path(part->data_part_storage->getRelativePath()) / (projection_name + ".proj") / file_name == file_path) + if (fs::path(part->getDataPartStorage().getRelativePath()) / (projection_name + ".proj") / file_name == file_path) { throw Exception( ErrorCodes::LOGICAL_ERROR, @@ -190,7 +190,7 @@ void PartMetadataManagerWithCache::assertAllDeleted(bool include_projection) con void PartMetadataManagerWithCache::getKeysAndCheckSums(Strings & keys, std::vector & checksums) const { - String prefix = getKeyFromFilePath(fs::path(part->data_part_storage->getRelativePath()) / ""); + String prefix = getKeyFromFilePath(fs::path(part->getDataPartStorage().getRelativePath()) / ""); Strings values; cache->getByPrefix(prefix, keys, values); size_t size = keys.size(); @@ -225,7 +225,7 @@ std::unordered_map PartMetadataManagerWit results.emplace(file_name, cache_checksums[i]); /// File belongs to normal part - if (fs::path(part->data_part_storage->getRelativePath()) / file_name == file_path) + if (fs::path(part->getDataPartStorage().getRelativePath()) / file_name == file_path) { auto disk_checksum = part->getActualChecksumByFile(file_name); if (disk_checksum != cache_checksums[i]) diff --git a/src/Storages/MergeTree/RPNBuilder.cpp b/src/Storages/MergeTree/RPNBuilder.cpp new file mode 100644 index 00000000000..d7ea68e7d64 --- /dev/null +++ b/src/Storages/MergeTree/RPNBuilder.cpp @@ -0,0 +1,417 @@ +#include + +#include + +#include +#include +#include +#include + +#include +#include + +#include +#include + +#include + +#include + + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +namespace +{ + +void appendColumnNameWithoutAlias(const ActionsDAG::Node & node, WriteBuffer & out, bool legacy = false) +{ + switch (node.type) + { + case ActionsDAG::ActionType::INPUT: + writeString(node.result_name, out); + break; + case ActionsDAG::ActionType::COLUMN: + { + /// If it was created from ASTLiteral, then result_name can be an alias. + /// We need to convert value back to string here. + if (const auto * column_const = typeid_cast(node.column.get())) + writeString(applyVisitor(FieldVisitorToString(), column_const->getField()), out); + /// It may be possible that column is ColumnSet + else + writeString(node.result_name, out); + break; + } + case ActionsDAG::ActionType::ALIAS: + appendColumnNameWithoutAlias(*node.children.front(), out, legacy); + break; + case ActionsDAG::ActionType::ARRAY_JOIN: + writeCString("arrayJoin(", out); + appendColumnNameWithoutAlias(*node.children.front(), out, legacy); + writeChar(')', out); + break; + case ActionsDAG::ActionType::FUNCTION: + { + auto name = node.function_base->getName(); + if (legacy && name == "modulo") + writeCString("moduleLegacy", out); + else + writeString(name, out); + + writeChar('(', out); + bool first = true; + for (const auto * arg : node.children) + { + if (!first) + writeCString(", ", out); + first = false; + + appendColumnNameWithoutAlias(*arg, out, legacy); + } + writeChar(')', out); + } + } +} + +String getColumnNameWithoutAlias(const ActionsDAG::Node & node, bool legacy = false) +{ + WriteBufferFromOwnString out; + appendColumnNameWithoutAlias(node, out, legacy); + return std::move(out.str()); +} + +} + +RPNBuilderTreeContext::RPNBuilderTreeContext(ContextPtr query_context_) + : query_context(std::move(query_context_)) +{} + +RPNBuilderTreeContext::RPNBuilderTreeContext(ContextPtr query_context_, Block block_with_constants_, PreparedSetsPtr prepared_sets_) + : query_context(std::move(query_context_)) + , block_with_constants(std::move(block_with_constants_)) + , prepared_sets(std::move(prepared_sets_)) +{} + +RPNBuilderTreeNode::RPNBuilderTreeNode(const ActionsDAG::Node * dag_node_, RPNBuilderTreeContext & tree_context_) + : dag_node(dag_node_) + , tree_context(tree_context_) +{ + assert(dag_node); +} + +RPNBuilderTreeNode::RPNBuilderTreeNode(const IAST * ast_node_, RPNBuilderTreeContext & tree_context_) + : ast_node(ast_node_) + , tree_context(tree_context_) +{ + assert(ast_node); +} + +std::string RPNBuilderTreeNode::getColumnName() const +{ + if (ast_node) + return ast_node->getColumnNameWithoutAlias(); + else + return getColumnNameWithoutAlias(*dag_node); +} + +std::string RPNBuilderTreeNode::getColumnNameWithModuloLegacy() const +{ + if (ast_node) + { + auto adjusted_ast = ast_node->clone(); + KeyDescription::moduloToModuloLegacyRecursive(adjusted_ast); + return adjusted_ast->getColumnNameWithoutAlias(); + } + else + { + return getColumnNameWithoutAlias(*dag_node, true /*legacy*/); + } +} + +bool RPNBuilderTreeNode::isFunction() const +{ + if (ast_node) + return typeid_cast(ast_node); + else + return dag_node->type == ActionsDAG::ActionType::FUNCTION; +} + +bool RPNBuilderTreeNode::isConstant() const +{ + if (ast_node) + { + bool is_literal = typeid_cast(ast_node); + if (is_literal) + return true; + + String column_name = ast_node->getColumnName(); + const auto & block_with_constants = tree_context.getBlockWithConstants(); + + if (block_with_constants.has(column_name) && isColumnConst(*block_with_constants.getByName(column_name).column)) + return true; + + return false; + } + else + { + return dag_node->column && isColumnConst(*dag_node->column); + } +} + +ColumnWithTypeAndName RPNBuilderTreeNode::getConstantColumn() const +{ + if (!isConstant()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "RPNBuilderTree node is not a constant"); + + ColumnWithTypeAndName result; + + if (ast_node) + { + const auto * literal = assert_cast(ast_node); + if (literal) + { + result.type = applyVisitor(FieldToDataType(), literal->value); + result.column = result.type->createColumnConst(0, literal->value); + + return result; + } + + String column_name = ast_node->getColumnName(); + const auto & block_with_constants = tree_context.getBlockWithConstants(); + + return block_with_constants.getByName(column_name); + } + else + { + result.type = dag_node->result_type; + result.column = dag_node->column; + } + + return result; +} + +bool RPNBuilderTreeNode::tryGetConstant(Field & output_value, DataTypePtr & output_type) const +{ + if (ast_node) + { + // Constant expr should use alias names if any + String column_name = ast_node->getColumnName(); + const auto & block_with_constants = tree_context.getBlockWithConstants(); + + if (const auto * literal = ast_node->as()) + { + /// By default block_with_constants has only one column named "_dummy". + /// If block contains only constants it's may not be preprocessed by + // ExpressionAnalyzer, so try to look up in the default column. + if (!block_with_constants.has(column_name)) + column_name = "_dummy"; + + /// Simple literal + output_value = literal->value; + output_type = block_with_constants.getByName(column_name).type; + + /// If constant is not Null, we can assume it's type is not Nullable as well. + if (!output_value.isNull()) + output_type = removeNullable(output_type); + + return true; + } + else if (block_with_constants.has(column_name) && + isColumnConst(*block_with_constants.getByName(column_name).column)) + { + /// An expression which is dependent on constants only + const auto & constant_column = block_with_constants.getByName(column_name); + output_value = (*constant_column.column)[0]; + output_type = constant_column.type; + + if (!output_value.isNull()) + output_type = removeNullable(output_type); + + return true; + } + } + else + { + if (dag_node->column && isColumnConst(*dag_node->column)) + { + output_value = (*dag_node->column)[0]; + output_type = dag_node->result_type; + + if (!output_value.isNull()) + output_type = removeNullable(output_type); + + return true; + } + } + + return false; +} + +namespace +{ + +ConstSetPtr tryGetSetFromDAGNode(const ActionsDAG::Node * dag_node) +{ + if (!dag_node->column) + return {}; + + const IColumn * column = dag_node->column.get(); + if (const auto * column_const = typeid_cast(column)) + column = &column_const->getDataColumn(); + + if (const auto * column_set = typeid_cast(column)) + { + auto set = column_set->getData(); + + if (set->isCreated()) + return set; + } + + return {}; +} + +} + +ConstSetPtr RPNBuilderTreeNode::tryGetPreparedSet() const +{ + const auto & prepared_sets = getTreeContext().getPreparedSets(); + + if (ast_node && prepared_sets) + { + auto prepared_sets_with_same_hash = prepared_sets->getByTreeHash(ast_node->getTreeHash()); + for (auto & set : prepared_sets_with_same_hash) + if (set->isCreated()) + return set; + } + else if (dag_node) + { + return tryGetSetFromDAGNode(dag_node); + } + + return {}; +} + +ConstSetPtr RPNBuilderTreeNode::tryGetPreparedSet(const DataTypes & data_types) const +{ + const auto & prepared_sets = getTreeContext().getPreparedSets(); + + if (prepared_sets && ast_node) + { + if (ast_node->as() || ast_node->as()) + return prepared_sets->get(PreparedSetKey::forSubquery(*ast_node)); + + return prepared_sets->get(PreparedSetKey::forLiteral(*ast_node, data_types)); + } + else if (dag_node) + { + return tryGetSetFromDAGNode(dag_node); + } + + return nullptr; +} + +ConstSetPtr RPNBuilderTreeNode::tryGetPreparedSet( + const std::vector & indexes_mapping, + const DataTypes & data_types) const +{ + const auto & prepared_sets = getTreeContext().getPreparedSets(); + + if (prepared_sets && ast_node) + { + if (ast_node->as() || ast_node->as()) + return prepared_sets->get(PreparedSetKey::forSubquery(*ast_node)); + + /// We have `PreparedSetKey::forLiteral` but it is useless here as we don't have enough information + /// about types in left argument of the IN operator. Instead, we manually iterate through all the sets + /// and find the one for the right arg based on the AST structure (getTreeHash), after that we check + /// that the types it was prepared with are compatible with the types of the primary key. + auto types_match = [&indexes_mapping, &data_types](const SetPtr & candidate_set) + { + assert(indexes_mapping.size() == data_types.size()); + + for (size_t i = 0; i < indexes_mapping.size(); ++i) + { + if (!candidate_set->areTypesEqual(indexes_mapping[i].tuple_index, data_types[i])) + return false; + } + + return true; + }; + + auto tree_hash = ast_node->getTreeHash(); + for (const auto & set : prepared_sets->getByTreeHash(tree_hash)) + { + if (types_match(set)) + return set; + } + } + else if (dag_node->column) + { + return tryGetSetFromDAGNode(dag_node); + } + + return nullptr; +} + +RPNBuilderFunctionTreeNode RPNBuilderTreeNode::toFunctionNode() const +{ + if (!isFunction()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "RPNBuilderTree node is not a function"); + + if (this->ast_node) + return RPNBuilderFunctionTreeNode(this->ast_node, tree_context); + else + return RPNBuilderFunctionTreeNode(this->dag_node, tree_context); +} + +std::optional RPNBuilderTreeNode::toFunctionNodeOrNull() const +{ + if (!isFunction()) + return {}; + + if (this->ast_node) + return RPNBuilderFunctionTreeNode(this->ast_node, tree_context); + else + return RPNBuilderFunctionTreeNode(this->dag_node, tree_context); +} + +std::string RPNBuilderFunctionTreeNode::getFunctionName() const +{ + if (ast_node) + return assert_cast(ast_node)->name; + else + return dag_node->function_base->getName(); +} + +size_t RPNBuilderFunctionTreeNode::getArgumentsSize() const +{ + if (ast_node) + { + const auto * ast_function = assert_cast(ast_node); + return ast_function->arguments ? ast_function->arguments->children.size() : 0; + } + else + { + return dag_node->children.size(); + } +} + +RPNBuilderTreeNode RPNBuilderFunctionTreeNode::getArgumentAt(size_t index) const +{ + if (ast_node) + { + const auto * ast_function = assert_cast(ast_node); + return RPNBuilderTreeNode(ast_function->arguments->children[index].get(), tree_context); + } + else + { + return RPNBuilderTreeNode(dag_node->children[index], tree_context); + } +} + +} diff --git a/src/Storages/MergeTree/RPNBuilder.h b/src/Storages/MergeTree/RPNBuilder.h index 27b616dc301..132d3aa44e8 100644 --- a/src/Storages/MergeTree/RPNBuilder.h +++ b/src/Storages/MergeTree/RPNBuilder.h @@ -1,111 +1,266 @@ #pragma once #include -#include -#include -#include -#include -#include -#include -#include +#include +#include +#include +#include namespace DB { -/// Builds reverse polish notation -template -class RPNBuilder : WithContext +/** Context of RPNBuilderTree. + * + * For AST tree context, precalculated block with constants and prepared sets are required for index analysis. + * For DAG tree precalculated block with constants and prepared sets are not required, because constants and sets already + * calculated inside COLUMN actions dag node. + */ +class RPNBuilderTreeContext { public: - using RPN = std::vector; - using AtomFromASTFunc = std::function< - bool(const ASTPtr & node, ContextPtr context, Block & block_with_constants, RPNElement & out)>; + /// Construct RPNBuilderTreeContext for ActionsDAG tree + explicit RPNBuilderTreeContext(ContextPtr query_context_); - RPNBuilder(const SelectQueryInfo & query_info, ContextPtr context_, const AtomFromASTFunc & atom_from_ast_) - : WithContext(context_), atom_from_ast(atom_from_ast_) + /// Construct RPNBuilderTreeContext for AST tree + explicit RPNBuilderTreeContext(ContextPtr query_context_, Block block_with_constants_, PreparedSetsPtr prepared_sets_); + + /// Get query context + const ContextPtr & getQueryContext() const { - /** Evaluation of expressions that depend only on constants. - * For the index to be used, if it is written, for example `WHERE Date = toDate(now())`. - */ - block_with_constants = KeyCondition::getBlockWithConstants(query_info.query, query_info.syntax_analyzer_result, getContext()); - - /// Transform WHERE section to Reverse Polish notation - const ASTSelectQuery & select = typeid_cast(*query_info.query); - if (select.where()) - { - traverseAST(select.where()); - - if (select.prewhere()) - { - traverseAST(select.prewhere()); - rpn.emplace_back(RPNElement::FUNCTION_AND); - } - } - else if (select.prewhere()) - { - traverseAST(select.prewhere()); - } - else - { - rpn.emplace_back(RPNElement::FUNCTION_UNKNOWN); - } + return query_context; } - RPN && extractRPN() { return std::move(rpn); } + /** Get block with constants. + * Valid only for AST tree. + */ + const Block & getBlockWithConstants() const + { + return block_with_constants; + } + + /** Get prepared sets. + * Valid only for AST tree. + */ + const PreparedSetsPtr & getPreparedSets() const + { + return prepared_sets; + } private: - void traverseAST(const ASTPtr & node) + /// Valid for both AST and ActionDAG tree + ContextPtr query_context; + + /// Valid only for AST tree + Block block_with_constants; + + /// Valid only for AST tree + PreparedSetsPtr prepared_sets; +}; + +class RPNBuilderFunctionTreeNode; + +/** RPNBuilderTreeNode is wrapper around DAG or AST node. + * It defines unified interface for index analysis. + */ +class RPNBuilderTreeNode +{ +public: + /// Construct RPNBuilderTreeNode with non null dag node and tree context + explicit RPNBuilderTreeNode(const ActionsDAG::Node * dag_node_, RPNBuilderTreeContext & tree_context_); + + /// Construct RPNBuilderTreeNode with non null ast node and tree context + explicit RPNBuilderTreeNode(const IAST * ast_node_, RPNBuilderTreeContext & tree_context_); + + /// Get column name + std::string getColumnName() const; + + /** Get column name. + * Function `modulo` is replaced with `moduloLegacy`. + */ + std::string getColumnNameWithModuloLegacy() const; + + /// Is node function + bool isFunction() const; + + /// Is node constant + bool isConstant() const; + + /** Get constant as constant column. + * Node must be constant before calling these method, otherwise logical exception is thrown. + */ + ColumnWithTypeAndName getConstantColumn() const; + + /** Try get constant from node. If node is constant returns true, and constant value and constant type output parameters are set. + * Otherwise false is returned. + */ + bool tryGetConstant(Field & output_value, DataTypePtr & output_type) const; + + /// Try get prepared set from node + ConstSetPtr tryGetPreparedSet() const; + + /// Try get prepared set from node that match data types + ConstSetPtr tryGetPreparedSet(const DataTypes & data_types) const; + + /// Try get prepared set from node that match indexes mapping and data types + ConstSetPtr tryGetPreparedSet( + const std::vector & indexes_mapping, + const DataTypes & data_types) const; + + /** Convert node to function node. + * Node must be function before calling these method, otherwise exception is thrown. + */ + RPNBuilderFunctionTreeNode toFunctionNode() const; + + /// Convert node to function node or null optional + std::optional toFunctionNodeOrNull() const; + + /// Get tree context + const RPNBuilderTreeContext & getTreeContext() const + { + return tree_context; + } + + /// Get tree context + RPNBuilderTreeContext & getTreeContext() + { + return tree_context; + } + +protected: + const IAST * ast_node = nullptr; + const ActionsDAG::Node * dag_node = nullptr; + RPNBuilderTreeContext & tree_context; +}; + +/** RPNBuilderFunctionTreeNode is wrapper around RPNBuilderTreeNode with function type. + * It provide additional functionality that is specific for function. + */ +class RPNBuilderFunctionTreeNode : public RPNBuilderTreeNode +{ +public: + using RPNBuilderTreeNode::RPNBuilderTreeNode; + + /// Get function name + std::string getFunctionName() const; + + /// Get function arguments size + size_t getArgumentsSize() const; + + /// Get function argument at index + RPNBuilderTreeNode getArgumentAt(size_t index) const; +}; + +/** RPN Builder build stack of reverse polish notation elements (RPNElements) required for index analysis. + * + * RPNBuilder client must provide RPNElement type that has following interface: + * + * struct RPNElementInterface + * { + * enum Function + * { + * FUNCTION_UNKNOWN, /// Can take any value. + * /// Operators of the logical expression. + * FUNCTION_NOT, + * FUNCTION_AND, + * FUNCTION_OR, + * ... + * }; + * + * RPNElementInterface(); + * + * Function function = FUNCTION_UNKNOWN; + * + * } + * + * RPNBuilder take care of building stack of RPNElements with `NOT`, `AND`, `OR` types. + * In addition client must provide ExtractAtomFromTreeFunction that returns true and RPNElement as output parameter, + * if it can convert RPNBuilderTree node to RPNElement, false otherwise. + */ +template +class RPNBuilder +{ +public: + using RPNElements = std::vector; + using ExtractAtomFromTreeFunction = std::function; + + explicit RPNBuilder(const ActionsDAG::Node * filter_actions_dag_node, + ContextPtr query_context_, + const ExtractAtomFromTreeFunction & extract_atom_from_tree_function_) + : tree_context(std::move(query_context_)) + , extract_atom_from_tree_function(extract_atom_from_tree_function_) + { + traverseTree(RPNBuilderTreeNode(filter_actions_dag_node, tree_context)); + } + + RPNBuilder(const ASTPtr & filter_node, + ContextPtr query_context_, + Block block_with_constants_, + PreparedSetsPtr prepared_sets_, + const ExtractAtomFromTreeFunction & extract_atom_from_tree_function_) + : tree_context(std::move(query_context_), std::move(block_with_constants_), std::move(prepared_sets_)) + , extract_atom_from_tree_function(extract_atom_from_tree_function_) + { + traverseTree(RPNBuilderTreeNode(filter_node.get(), tree_context)); + } + + RPNElements && extractRPN() && { return std::move(rpn_elements); } + +private: + void traverseTree(const RPNBuilderTreeNode & node) { RPNElement element; - if (ASTFunction * func = typeid_cast(&*node)) + if (node.isFunction()) { - if (operatorFromAST(func, element)) + auto function_node = node.toFunctionNode(); + + if (extractLogicalOperatorFromTree(function_node, element)) { - auto & args = typeid_cast(*func->arguments).children; - for (size_t i = 0, size = args.size(); i < size; ++i) + size_t arguments_size = function_node.getArgumentsSize(); + + for (size_t argument_index = 0; argument_index < arguments_size; ++argument_index) { - traverseAST(args[i]); + auto function_node_argument = function_node.getArgumentAt(argument_index); + traverseTree(function_node_argument); /** The first part of the condition is for the correct support of `and` and `or` functions of arbitrary arity * - in this case `n - 1` elements are added (where `n` is the number of arguments). */ - if (i != 0 || element.function == RPNElement::FUNCTION_NOT) - rpn.emplace_back(std::move(element)); + if (argument_index != 0 || element.function == RPNElement::FUNCTION_NOT) + rpn_elements.emplace_back(std::move(element)); } return; } } - if (!atom_from_ast(node, getContext(), block_with_constants, element)) - { + if (!extract_atom_from_tree_function(node, element)) element.function = RPNElement::FUNCTION_UNKNOWN; - } - rpn.emplace_back(std::move(element)); + rpn_elements.emplace_back(std::move(element)); } - bool operatorFromAST(const ASTFunction * func, RPNElement & out) + bool extractLogicalOperatorFromTree(const RPNBuilderFunctionTreeNode & function_node, RPNElement & out) { - /// Functions AND, OR, NOT. - /// Also a special function `indexHint` - works as if instead of calling a function there are just parentheses - /// (or, the same thing - calling the function `and` from one argument). - const ASTs & args = typeid_cast(*func->arguments).children; + /** Functions AND, OR, NOT. + * Also a special function `indexHint` - works as if instead of calling a function there are just parentheses + * (or, the same thing - calling the function `and` from one argument). + */ - if (func->name == "not") + auto function_name = function_node.getFunctionName(); + if (function_name == "not") { - if (args.size() != 1) + if (function_node.getArgumentsSize() != 1) return false; out.function = RPNElement::FUNCTION_NOT; } else { - if (func->name == "and" || func->name == "indexHint") + if (function_name == "and" || function_name == "indexHint") out.function = RPNElement::FUNCTION_AND; - else if (func->name == "or") + else if (function_name == "or") out.function = RPNElement::FUNCTION_OR; else return false; @@ -114,10 +269,9 @@ private: return true; } - const AtomFromASTFunc & atom_from_ast; - Block block_with_constants; - RPN rpn; + RPNBuilderTreeContext tree_context; + const ExtractAtomFromTreeFunction & extract_atom_from_tree_function; + RPNElements rpn_elements; }; - } diff --git a/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp b/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp index 158cbfca9fd..dbc2bd98e20 100644 --- a/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp +++ b/src/Storages/MergeTree/ReplicatedMergeTreeSink.cpp @@ -1,10 +1,10 @@ #include #include #include -#include #include #include #include +#include #include #include @@ -165,7 +165,9 @@ void ReplicatedMergeTreeSink::consume(Chunk chunk) */ size_t replicas_num = checkQuorumPrecondition(zookeeper); - deduceTypesOfObjectColumns(storage_snapshot, block); + if (!storage_snapshot->object_columns.empty()) + convertDynamicColumnsToTuples(block, storage_snapshot); + auto part_blocks = storage.writer.splitBlockIntoParts(block, max_parts_per_block, metadata_snapshot, context); using DelayedPartitions = std::vector; @@ -268,7 +270,7 @@ void ReplicatedMergeTreeSink::finishDelayedChunk(zkutil::ZooKeeperPtr & zookeepe try { - commitPart(zookeeper, part, partition.block_id, partition.temp_part.builder, delayed_chunk->replicas_num); + commitPart(zookeeper, part, partition.block_id, delayed_chunk->replicas_num); last_block_is_duplicate = last_block_is_duplicate || part->is_duplicate; @@ -301,7 +303,7 @@ void ReplicatedMergeTreeSink::writeExistingPart(MergeTreeData::MutableDataPartPt try { part->version.setCreationTID(Tx::PrehistoricTID, nullptr); - commitPart(zookeeper, part, "", part->data_part_storage->getBuilder(), replicas_num); + commitPart(zookeeper, part, "", replicas_num); PartLog::addNewPart(storage.getContext(), part, watch.elapsed()); } catch (...) @@ -315,7 +317,6 @@ void ReplicatedMergeTreeSink::commitPart( zkutil::ZooKeeperPtr & zookeeper, MergeTreeData::MutableDataPartPtr & part, const String & block_id, - DataPartStorageBuilderPtr builder, size_t replicas_num) { /// It is possible that we alter a part with different types of source columns. @@ -326,7 +327,7 @@ void ReplicatedMergeTreeSink::commitPart( assertSessionIsNotExpired(zookeeper); - String temporary_part_relative_path = part->data_part_storage->getPartDirectory(); + String temporary_part_relative_path = part->getDataPartStorage().getPartDirectory(); /// There is one case when we need to retry transaction in a loop. /// But don't do it too many times - just as defensive measure. @@ -499,7 +500,7 @@ void ReplicatedMergeTreeSink::commitPart( try { auto lock = storage.lockParts(); - renamed = storage.renameTempPartAndAdd(part, transaction, builder, lock); + renamed = storage.renameTempPartAndAdd(part, transaction, lock); } catch (const Exception & e) { @@ -563,8 +564,7 @@ void ReplicatedMergeTreeSink::commitPart( transaction.rollbackPartsToTemporaryState(); part->is_temp = true; - part->renameTo(temporary_part_relative_path, false, builder); - builder->commit(); + part->renameTo(temporary_part_relative_path, false); /// If this part appeared on other replica than it's better to try to write it locally one more time. If it's our part /// than it will be ignored on the next itration. diff --git a/src/Storages/MergeTree/ReplicatedMergeTreeSink.h b/src/Storages/MergeTree/ReplicatedMergeTreeSink.h index ab729e6edec..da87ddc0d63 100644 --- a/src/Storages/MergeTree/ReplicatedMergeTreeSink.h +++ b/src/Storages/MergeTree/ReplicatedMergeTreeSink.h @@ -79,7 +79,6 @@ private: zkutil::ZooKeeperPtr & zookeeper, MergeTreeData::MutableDataPartPtr & part, const String & block_id, - DataPartStorageBuilderPtr part_builder, size_t replicas_num); /// Wait for quorum to be satisfied on path (quorum_path) form part (part_name) diff --git a/src/Storages/MergeTree/StorageFromMergeTreeDataPart.h b/src/Storages/MergeTree/StorageFromMergeTreeDataPart.h index a3d578cf5f2..7bad9947a88 100644 --- a/src/Storages/MergeTree/StorageFromMergeTreeDataPart.h +++ b/src/Storages/MergeTree/StorageFromMergeTreeDataPart.h @@ -47,10 +47,10 @@ public: const StorageMetadataPtr & metadata_snapshot, ContextPtr /*query_context*/) const override { const auto & storage_columns = metadata_snapshot->getColumns(); - if (!hasObjectColumns(storage_columns)) + if (!hasDynamicSubcolumns(storage_columns)) return std::make_shared(*this, metadata_snapshot); - auto object_columns = getObjectColumns( + auto object_columns = getConcreteObjectColumns( parts.begin(), parts.end(), storage_columns, [](const auto & part) -> const auto & { return part->getColumns(); }); diff --git a/src/Storages/MergeTree/checkDataPart.cpp b/src/Storages/MergeTree/checkDataPart.cpp index d5a838668d2..6f9f16b6155 100644 --- a/src/Storages/MergeTree/checkDataPart.cpp +++ b/src/Storages/MergeTree/checkDataPart.cpp @@ -1,3 +1,4 @@ +#include "Storages/MergeTree/IDataPartStorage.h" #include #include @@ -46,7 +47,7 @@ bool isNotEnoughMemoryErrorCode(int code) IMergeTreeDataPart::Checksums checkDataPart( MergeTreeData::DataPartPtr data_part, - const DataPartStoragePtr & data_part_storage, + const IDataPartStorage & data_part_storage, const NamesAndTypesList & columns_list, const MergeTreeDataPartType & part_type, const NameSet & files_without_checksums, @@ -64,13 +65,13 @@ IMergeTreeDataPart::Checksums checkDataPart( NamesAndTypesList columns_txt; { - auto buf = data_part_storage->readFile("columns.txt", {}, std::nullopt, std::nullopt); + auto buf = data_part_storage.readFile("columns.txt", {}, std::nullopt, std::nullopt); columns_txt.readText(*buf); assertEOF(*buf); } if (columns_txt != columns_list) - throw Exception("Columns doesn't match in part " + data_part_storage->getFullPath() + throw Exception("Columns doesn't match in part " + data_part_storage.getFullPath() + ". Expected: " + columns_list.toString() + ". Found: " + columns_txt.toString(), ErrorCodes::CORRUPTED_DATA); @@ -78,9 +79,9 @@ IMergeTreeDataPart::Checksums checkDataPart( IMergeTreeDataPart::Checksums checksums_data; /// This function calculates checksum for both compressed and decompressed contents of compressed file. - auto checksum_compressed_file = [](const DataPartStoragePtr & data_part_storage_, const String & file_path) + auto checksum_compressed_file = [](const IDataPartStorage & data_part_storage_, const String & file_path) { - auto file_buf = data_part_storage_->readFile(file_path, {}, std::nullopt, std::nullopt); + auto file_buf = data_part_storage_.readFile(file_path, {}, std::nullopt, std::nullopt); HashingReadBuffer compressed_hashing_buf(*file_buf); CompressedReadBuffer uncompressing_buf(compressed_hashing_buf); HashingReadBuffer uncompressed_hashing_buf(uncompressing_buf); @@ -96,9 +97,9 @@ IMergeTreeDataPart::Checksums checkDataPart( auto ratio_of_defaults = data_part->storage.getSettings()->ratio_of_defaults_for_sparse_serialization; SerializationInfoByName serialization_infos(columns_txt, SerializationInfo::Settings{ratio_of_defaults, false}); - if (data_part_storage->exists(IMergeTreeDataPart::SERIALIZATION_FILE_NAME)) + if (data_part_storage.exists(IMergeTreeDataPart::SERIALIZATION_FILE_NAME)) { - auto serialization_file = data_part_storage->readFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, {}, std::nullopt, std::nullopt); + auto serialization_file = data_part_storage.readFile(IMergeTreeDataPart::SERIALIZATION_FILE_NAME, {}, std::nullopt, std::nullopt); serialization_infos.readJSON(*serialization_file); } @@ -111,98 +112,17 @@ IMergeTreeDataPart::Checksums checkDataPart( }; /// This function calculates only checksum of file content (compressed or uncompressed). - /// It also calculates checksum of projections. auto checksum_file = [&](const String & file_name) { - if (data_part_storage->isDirectory(file_name) && endsWith(file_name, ".proj")) - { - auto projection_name = file_name.substr(0, file_name.size() - sizeof(".proj") + 1); - auto pit = data_part->getProjectionParts().find(projection_name); - if (pit == data_part->getProjectionParts().end()) - { - if (require_checksums) - throw Exception("Unexpected file " + file_name + " in data part", ErrorCodes::UNEXPECTED_FILE_IN_DATA_PART); - else - return; - } - - const auto & projection = pit->second; - IMergeTreeDataPart::Checksums projection_checksums_data; - - auto projection_part_storage = data_part_storage->getProjection(file_name); - - if (projection->getType() == MergeTreeDataPartType::Compact) - { - auto file_buf = projection_part_storage->readFile(MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION, {}, std::nullopt, std::nullopt); - HashingReadBuffer hashing_buf(*file_buf); - hashing_buf.ignoreAll(); - projection_checksums_data.files[MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION] - = IMergeTreeDataPart::Checksums::Checksum(hashing_buf.count(), hashing_buf.getHash()); - } - else - { - const NamesAndTypesList & projection_columns_list = projection->getColumns(); - for (const auto & projection_column : projection_columns_list) - { - get_serialization(projection_column)->enumerateStreams( - [&](const ISerialization::SubstreamPath & substream_path) - { - String projection_file_name = ISerialization::getFileNameForStream(projection_column, substream_path) + ".bin"; - projection_checksums_data.files[projection_file_name] = checksum_compressed_file(projection_part_storage, projection_file_name); - }); - } - } - - IMergeTreeDataPart::Checksums projection_checksums_txt; - - if (require_checksums || projection_part_storage->exists("checksums.txt")) - { - auto buf = projection_part_storage->readFile("checksums.txt", {}, std::nullopt, std::nullopt); - projection_checksums_txt.read(*buf); - assertEOF(*buf); - } - - const auto & projection_checksum_files_txt = projection_checksums_txt.files; - for (auto projection_it = projection_part_storage->iterate(); projection_it->isValid(); projection_it->next()) - { - const String & projection_file_name = projection_it->name(); - auto projection_checksum_it = projection_checksums_data.files.find(projection_file_name); - - /// Skip files that we already calculated. Also skip metadata files that are not checksummed. - if (projection_checksum_it == projection_checksums_data.files.end() && !files_without_checksums.contains(projection_file_name)) - { - auto projection_txt_checksum_it = projection_checksum_files_txt.find(file_name); - if (projection_txt_checksum_it == projection_checksum_files_txt.end() - || projection_txt_checksum_it->second.uncompressed_size == 0) - { - auto projection_file_buf = projection_part_storage->readFile(projection_file_name, {}, std::nullopt, std::nullopt); - HashingReadBuffer projection_hashing_buf(*projection_file_buf); - projection_hashing_buf.ignoreAll(); - projection_checksums_data.files[projection_file_name] = IMergeTreeDataPart::Checksums::Checksum( - projection_hashing_buf.count(), projection_hashing_buf.getHash()); - } - else - { - projection_checksums_data.files[projection_file_name] = checksum_compressed_file(projection_part_storage, projection_file_name); - } - } - } - checksums_data.files[file_name] = IMergeTreeDataPart::Checksums::Checksum( - projection_checksums_data.getTotalSizeOnDisk(), projection_checksums_data.getTotalChecksumUInt128()); - - if (require_checksums || !projection_checksums_txt.files.empty()) - projection_checksums_txt.checkEqual(projection_checksums_data, false); - } - else - { - auto file_buf = data_part_storage->readFile(file_name, {}, std::nullopt, std::nullopt); - HashingReadBuffer hashing_buf(*file_buf); - hashing_buf.ignoreAll(); - checksums_data.files[file_name] = IMergeTreeDataPart::Checksums::Checksum(hashing_buf.count(), hashing_buf.getHash()); - } + auto file_buf = data_part_storage.readFile(file_name, {}, std::nullopt, std::nullopt); + HashingReadBuffer hashing_buf(*file_buf); + hashing_buf.ignoreAll(); + checksums_data.files[file_name] = IMergeTreeDataPart::Checksums::Checksum(hashing_buf.count(), hashing_buf.getHash()); }; - bool check_uncompressed = true; + /// Do not check uncompressed for projections. But why? + bool check_uncompressed = !data_part->isProjectionPart(); + /// First calculate checksums for columns data if (part_type == MergeTreeDataPartType::Compact) { @@ -224,23 +144,32 @@ IMergeTreeDataPart::Checksums checkDataPart( } else { - throw Exception("Unknown type in part " + data_part_storage->getFullPath(), ErrorCodes::UNKNOWN_PART_TYPE); + throw Exception("Unknown type in part " + data_part_storage.getFullPath(), ErrorCodes::UNKNOWN_PART_TYPE); } /// Checksums from the rest files listed in checksums.txt. May be absent. If present, they are subsequently compared with the actual data checksums. IMergeTreeDataPart::Checksums checksums_txt; - if (require_checksums || data_part_storage->exists("checksums.txt")) + if (require_checksums || data_part_storage.exists("checksums.txt")) { - auto buf = data_part_storage->readFile("checksums.txt", {}, std::nullopt, std::nullopt); + auto buf = data_part_storage.readFile("checksums.txt", {}, std::nullopt, std::nullopt); checksums_txt.read(*buf); assertEOF(*buf); } + NameSet projections_on_disk; const auto & checksum_files_txt = checksums_txt.files; - for (auto it = data_part_storage->iterate(); it->isValid(); it->next()) + for (auto it = data_part_storage.iterate(); it->isValid(); it->next()) { - const String & file_name = it->name(); + auto file_name = it->name(); + + /// We will check projections later. + if (data_part_storage.isDirectory(file_name) && endsWith(file_name, ".proj")) + { + projections_on_disk.insert(file_name); + continue; + } + auto checksum_it = checksums_data.files.find(file_name); /// Skip files that we already calculated. Also skip metadata files that are not checksummed. @@ -259,11 +188,38 @@ IMergeTreeDataPart::Checksums checkDataPart( } } + for (const auto & [name, projection] : data_part->getProjectionParts()) + { + if (is_cancelled()) + return {}; + + auto projection_file = name + ".proj"; + auto projection_checksums = checkDataPart( + projection, *data_part_storage.getProjection(projection_file), + projection->getColumns(), projection->getType(), + projection->getFileNamesWithoutChecksums(), + require_checksums, is_cancelled); + + checksums_data.files[projection_file] = IMergeTreeDataPart::Checksums::Checksum( + projection_checksums.getTotalSizeOnDisk(), + projection_checksums.getTotalChecksumUInt128()); + + projections_on_disk.erase(projection_file); + } + + if (require_checksums && !projections_on_disk.empty()) + { + throw Exception(ErrorCodes::UNEXPECTED_FILE_IN_DATA_PART, + "Found unexpected projection directories: {}", + fmt::join(projections_on_disk, ",")); + } + if (is_cancelled()) return {}; if (require_checksums || !checksums_txt.files.empty()) checksums_txt.checkEqual(checksums_data, check_uncompressed); + return checksums_data; } @@ -285,7 +241,7 @@ IMergeTreeDataPart::Checksums checkDataPart( return checkDataPart( data_part, - data_part->data_part_storage, + data_part->getDataPartStorage(), data_part->getColumns(), data_part->getType(), data_part->getFileNamesWithoutChecksums(), diff --git a/src/Storages/StorageDistributed.cpp b/src/Storages/StorageDistributed.cpp index 8e4715db483..f7f68eba30f 100644 --- a/src/Storages/StorageDistributed.cpp +++ b/src/Storages/StorageDistributed.cpp @@ -598,7 +598,7 @@ std::optional StorageDistributed::getOptimizedQueryP static bool requiresObjectColumns(const ColumnsDescription & all_columns, ASTPtr query) { - if (!hasObjectColumns(all_columns)) + if (!hasDynamicSubcolumns(all_columns)) return false; if (!query) @@ -613,7 +613,7 @@ static bool requiresObjectColumns(const ColumnsDescription & all_columns, ASTPtr auto name_in_storage = Nested::splitName(required_column).first; auto column_in_storage = all_columns.tryGetPhysical(name_in_storage); - if (column_in_storage && isObject(column_in_storage->type)) + if (column_in_storage && column_in_storage->type->hasDynamicSubcolumns()) return true; } @@ -640,7 +640,7 @@ StorageSnapshotPtr StorageDistributed::getStorageSnapshotForQuery( metadata_snapshot->getColumns(), getContext()); - auto object_columns = DB::getObjectColumns( + auto object_columns = DB::getConcreteObjectColumns( snapshot_data->objects_by_shard.begin(), snapshot_data->objects_by_shard.end(), metadata_snapshot->getColumns(), diff --git a/src/Storages/StorageInMemoryMetadata.cpp b/src/Storages/StorageInMemoryMetadata.cpp index 66dcc938aef..a80f21834db 100644 --- a/src/Storages/StorageInMemoryMetadata.cpp +++ b/src/Storages/StorageInMemoryMetadata.cpp @@ -526,7 +526,7 @@ void StorageInMemoryMetadata::check(const NamesAndTypesList & provided_columns) const auto * available_type = it->getMapped(); - if (!isObject(*available_type) + if (!available_type->hasDynamicSubcolumns() && !column.type->equals(*available_type) && !isCompatibleEnumTypes(available_type, column.type.get())) throw Exception( @@ -575,7 +575,7 @@ void StorageInMemoryMetadata::check(const NamesAndTypesList & provided_columns, const auto * provided_column_type = it->getMapped(); const auto * available_column_type = jt->getMapped(); - if (!isObject(*provided_column_type) + if (!provided_column_type->hasDynamicSubcolumns() && !provided_column_type->equals(*available_column_type) && !isCompatibleEnumTypes(available_column_type, provided_column_type)) throw Exception( @@ -619,7 +619,7 @@ void StorageInMemoryMetadata::check(const Block & block, bool need_all) const listOfColumns(available_columns)); const auto * available_type = it->getMapped(); - if (!isObject(*available_type) + if (!available_type->hasDynamicSubcolumns() && !column.type->equals(*available_type) && !isCompatibleEnumTypes(available_type, column.type.get())) throw Exception( diff --git a/src/Storages/StorageLog.cpp b/src/Storages/StorageLog.cpp index 9909489d901..8ed33220507 100644 --- a/src/Storages/StorageLog.cpp +++ b/src/Storages/StorageLog.cpp @@ -462,7 +462,7 @@ void LogSink::writeData(const NameAndTypePair & name_and_type, const IColumn & c settings.getter = createStreamGetter(name_and_type); if (!serialize_states.contains(name)) - serialization->serializeBinaryBulkStatePrefix(settings, serialize_states[name]); + serialization->serializeBinaryBulkStatePrefix(column, settings, serialize_states[name]); if (storage.use_marks_file) { diff --git a/src/Storages/StorageMemory.cpp b/src/Storages/StorageMemory.cpp index 957aae450c8..881cbc18b10 100644 --- a/src/Storages/StorageMemory.cpp +++ b/src/Storages/StorageMemory.cpp @@ -146,7 +146,7 @@ public: auto extended_storage_columns = storage_snapshot->getColumns( GetColumnsOptions(GetColumnsOptions::AllPhysical).withExtendedObjects()); - convertObjectsToTuples(block, extended_storage_columns); + convertDynamicColumnsToTuples(block, storage_snapshot); } if (storage.compress) @@ -212,10 +212,10 @@ StorageSnapshotPtr StorageMemory::getStorageSnapshot(const StorageMetadataPtr & auto snapshot_data = std::make_unique(); snapshot_data->blocks = data.get(); - if (!hasObjectColumns(metadata_snapshot->getColumns())) + if (!hasDynamicSubcolumns(metadata_snapshot->getColumns())) return std::make_shared(*this, metadata_snapshot, ColumnsDescription{}, std::move(snapshot_data)); - auto object_columns = getObjectColumns( + auto object_columns = getConcreteObjectColumns( snapshot_data->blocks->begin(), snapshot_data->blocks->end(), metadata_snapshot->getColumns(), diff --git a/src/Storages/StorageMergeTree.cpp b/src/Storages/StorageMergeTree.cpp index aac5372a83e..a450a9ef3a9 100644 --- a/src/Storages/StorageMergeTree.cpp +++ b/src/Storages/StorageMergeTree.cpp @@ -1,4 +1,5 @@ #include "StorageMergeTree.h" +#include "Storages/MergeTree/IMergeTreeDataPart.h" #include @@ -378,7 +379,9 @@ CurrentlyMergingPartsTagger::CurrentlyMergingPartsTagger( /// if we mutate part, than we should reserve space on the same disk, because mutations possible can create hardlinks if (is_mutation) - reserved_space = storage.tryReserveSpace(total_size, future_part->parts[0]->data_part_storage); + { + reserved_space = storage.tryReserveSpace(total_size, future_part->parts[0]->getDataPartStorage()); + } else { IMergeTreeDataPart::TTLInfos ttl_infos; @@ -386,7 +389,9 @@ CurrentlyMergingPartsTagger::CurrentlyMergingPartsTagger( for (auto & part_ptr : future_part->parts) { ttl_infos.update(part_ptr->ttl_infos); - max_volume_index = std::max(max_volume_index, part_ptr->data_part_storage->getVolumeIndex(*storage.getStoragePolicy())); + auto disk_name = part_ptr->getDataPartStorage().getDiskName(); + size_t volume_index = storage.getStoragePolicy()->getVolumeIndexByDiskName(disk_name); + max_volume_index = std::max(max_volume_index, volume_index); } reserved_space = storage.balancedReservation( @@ -1474,7 +1479,7 @@ void StorageMergeTree::dropPartsImpl(DataPartsVector && parts_to_remove, bool de /// NOTE: no race with background cleanup until we hold pointers to parts for (const auto & part : parts_to_remove) { - LOG_INFO(log, "Detaching {}", part->data_part_storage->getPartDirectory()); + LOG_INFO(log, "Detaching {}", part->getDataPartStorage().getPartDirectory()); part->makeCloneInDetached("", metadata_snapshot); } } @@ -1519,9 +1524,8 @@ PartitionCommandsResultInfo StorageMergeTree::attachPartition( MergeTreeData::Transaction transaction(*this, local_context->getCurrentTransaction().get()); { auto lock = lockParts(); - auto builder = loaded_parts[i]->data_part_storage->getBuilder(); fillNewPartName(loaded_parts[i], lock); - renameTempPartAndAdd(loaded_parts[i], transaction, builder, lock); + renameTempPartAndAdd(loaded_parts[i], transaction, lock); transaction.commit(&lock); } @@ -1604,9 +1608,7 @@ void StorageMergeTree::replacePartitionFrom(const StoragePtr & source_table, con for (auto part : dst_parts) { fillNewPartName(part, data_parts_lock); - - auto builder = part->data_part_storage->getBuilder(); - renameTempPartAndReplaceUnlocked(part, transaction, builder, data_parts_lock); + renameTempPartAndReplaceUnlocked(part, transaction, data_parts_lock); } /// Populate transaction transaction.commit(&data_parts_lock); @@ -1685,9 +1687,8 @@ void StorageMergeTree::movePartitionToTable(const StoragePtr & dest_table, const for (auto & part : dst_parts) { - auto builder = part->data_part_storage->getBuilder(); dest_table_storage->fillNewPartName(part, dest_data_parts_lock); - dest_table_storage->renameTempPartAndReplaceUnlocked(part, transaction, builder, dest_data_parts_lock); + dest_table_storage->renameTempPartAndReplaceUnlocked(part, transaction, dest_data_parts_lock); } @@ -1741,16 +1742,16 @@ CheckResults StorageMergeTree::checkData(const ASTPtr & query, ContextPtr local_ for (auto & part : data_parts) { /// If the checksums file is not present, calculate the checksums and write them to disk. - String checksums_path = "checksums.txt"; - String tmp_checksums_path = "checksums.txt.tmp"; - if (part->isStoredOnDisk() && !part->data_part_storage->exists(checksums_path)) + static constexpr auto checksums_path = "checksums.txt"; + if (part->isStoredOnDisk() && !part->getDataPartStorage().exists(checksums_path)) { try { auto calculated_checksums = checkDataPart(part, false); calculated_checksums.checkEqual(part->checksums, true); - part->data_part_storage->writeChecksums(part->checksums, local_context->getWriteSettings()); + auto & part_mutable = const_cast(*part); + part_mutable.writeChecksums(part->checksums, local_context->getWriteSettings()); part->checkMetadata(); results.emplace_back(part->name, true, "Checksums recounted and written to disk."); @@ -1810,17 +1811,15 @@ BackupEntries StorageMergeTree::backupMutations(UInt64 version, const String & d void StorageMergeTree::attachRestoredParts(MutableDataPartsVector && parts) { - for (auto part : parts) { /// It's important to create it outside of lock scope because /// otherwise it can lock parts in destructor and deadlock is possible. MergeTreeData::Transaction transaction(*this, NO_TRANSACTION_RAW); - auto builder = part->data_part_storage->getBuilder(); { auto lock = lockParts(); fillNewPartName(part, lock); - renameTempPartAndAdd(part, transaction, builder, lock); + renameTempPartAndAdd(part, transaction, lock); transaction.commit(&lock); } } diff --git a/src/Storages/StorageReplicatedMergeTree.cpp b/src/Storages/StorageReplicatedMergeTree.cpp index ef56ade2d58..3c0fbb162bc 100644 --- a/src/Storages/StorageReplicatedMergeTree.cpp +++ b/src/Storages/StorageReplicatedMergeTree.cpp @@ -1,5 +1,6 @@ #include +#include #include #include "Common/hex.h" #include @@ -1455,6 +1456,7 @@ MergeTreeData::MutableDataPartPtr StorageReplicatedMergeTree::attachPartHelperFo const String part_new_name = actual_part_info.getPartName(); for (const DiskPtr & disk : getStoragePolicy()->getDisks()) + { for (const auto it = disk->iterateDirectory(fs::path(relative_data_path) / "detached/"); it->isValid(); it->next()) { const auto part_info = MergeTreePartInfo::tryParsePartName(it->name(), format_version); @@ -1491,6 +1493,7 @@ MergeTreeData::MutableDataPartPtr StorageReplicatedMergeTree::attachPartHelperFo return part; } } + } return {}; } @@ -1541,8 +1544,7 @@ bool StorageReplicatedMergeTree::executeLogEntry(LogEntry & entry) Transaction transaction(*this, NO_TRANSACTION_RAW); part->version.setCreationTID(Tx::PrehistoricTID, nullptr); - auto builder = part->data_part_storage->getBuilder(); - renameTempPartAndReplace(part, transaction, builder); + renameTempPartAndReplace(part, transaction); checkPartChecksumsAndCommit(transaction, part); writePartLog(PartLogElement::Type::NEW_PART, {}, 0 /** log entry is fake so we don't measure the time */, @@ -1781,7 +1783,7 @@ bool StorageReplicatedMergeTree::executeFetch(LogEntry & entry, bool need_to_che } -DataPartStoragePtr StorageReplicatedMergeTree::executeFetchShared( +MutableDataPartStoragePtr StorageReplicatedMergeTree::executeFetchShared( const String & source_replica, const String & new_part_name, const DiskPtr & disk, @@ -1825,7 +1827,7 @@ void StorageReplicatedMergeTree::executeDropRange(const LogEntry & entry) /// Therefore, we use all data parts. auto metadata_snapshot = getInMemoryMetadataPtr(); - DataPartsVector parts_to_remove; + PartsToRemoveFromZooKeeper parts_to_remove; { auto data_parts_lock = lockParts(); parts_to_remove = removePartsInRangeFromWorkingSetAndGetPartsToRemoveFromZooKeeper(NO_TRANSACTION_RAW, drop_range_info, data_parts_lock); @@ -1847,8 +1849,11 @@ void StorageReplicatedMergeTree::executeDropRange(const LogEntry & entry) /// If DETACH clone parts to detached/ directory for (const auto & part : parts_to_remove) { - LOG_INFO(log, "Detaching {}", part->data_part_storage->getPartDirectory()); - part->makeCloneInDetached("", metadata_snapshot); + if (auto part_to_detach = part.getPartIfItWasActive()) + { + LOG_INFO(log, "Detaching {}", part_to_detach->getDataPartStorage().getPartDirectory()); + part_to_detach->makeCloneInDetached("", metadata_snapshot); + } } } @@ -1939,7 +1944,7 @@ bool StorageReplicatedMergeTree::executeReplaceRange(const LogEntry & entry) PartDescriptions all_parts; PartDescriptions parts_to_add; - DataPartsVector parts_to_remove; + PartsToRemoveFromZooKeeper parts_to_remove; auto table_lock_holder_dst_table = lockForShare( RWLockImpl::NO_QUERY, getSettings()->lock_acquire_timeout_for_background_operations); @@ -1970,7 +1975,7 @@ bool StorageReplicatedMergeTree::executeReplaceRange(const LogEntry & entry) String parts_to_remove_str; for (const auto & part : parts_to_remove) { - parts_to_remove_str += part->name; + parts_to_remove_str += part.getPartName(); parts_to_remove_str += " "; } LOG_TRACE(log, "Replacing {} parts {}with empty set", parts_to_remove.size(), parts_to_remove_str); @@ -2226,8 +2231,7 @@ bool StorageReplicatedMergeTree::executeReplaceRange(const LogEntry & entry) Coordination::Requests ops; for (PartDescriptionPtr & part_desc : final_parts) { - auto builder = part_desc->res_part->data_part_storage->getBuilder(); - renameTempPartAndReplace(part_desc->res_part, transaction, builder); + renameTempPartAndReplace(part_desc->res_part, transaction); getCommitPartOps(ops, part_desc->res_part); lockSharedData(*part_desc->res_part, false, part_desc->hardlinked_files); @@ -2247,7 +2251,7 @@ bool StorageReplicatedMergeTree::executeReplaceRange(const LogEntry & entry) String parts_to_remove_str; for (const auto & part : parts_to_remove) { - parts_to_remove_str += part->name; + parts_to_remove_str += part.getPartName(); parts_to_remove_str += " "; } LOG_TRACE(log, "Replacing {} parts {}with {} parts {}", parts_to_remove.size(), parts_to_remove_str, @@ -2324,9 +2328,7 @@ void StorageReplicatedMergeTree::executeClonePartFromShard(const LogEntry & entr part = get_part(); // The fetched part is valuable and should not be cleaned like a temp part. part->is_temp = false; - auto builder = part->data_part_storage->getBuilder(); - part->renameTo("detached/" + entry.new_part_name, true, builder); - builder->commit(); + part->renameTo("detached/" + entry.new_part_name, true); LOG_INFO(log, "Cloned part {} to detached directory", part->name); } @@ -2541,7 +2543,7 @@ void StorageReplicatedMergeTree::cloneReplica(const String & source_replica, Coo for (const auto & part : parts_to_remove_from_working_set) { - LOG_INFO(log, "Detaching {}", part->data_part_storage->getPartDirectory()); + LOG_INFO(log, "Detaching {}", part->getDataPartStorage().getPartDirectory()); part->makeCloneInDetached("clone", metadata_snapshot); } } @@ -3893,7 +3895,7 @@ bool StorageReplicatedMergeTree::fetchPart(const String & part_name, const Stora auto source_part = getActiveContainingPart(covered_part_info); /// Fetch for zero-copy replication is cheap and straightforward, so we don't use local clone here - if (source_part && (!settings_ptr->allow_remote_fs_zero_copy_replication || !source_part->data_part_storage->supportZeroCopyReplication())) + if (source_part && (!settings_ptr->allow_remote_fs_zero_copy_replication || !source_part->getDataPartStorage().supportZeroCopyReplication())) { auto source_part_header = ReplicatedMergeTreePartHeader::fromColumnsAndChecksums( source_part->getColumns(), source_part->checksums); @@ -3991,11 +3993,10 @@ bool StorageReplicatedMergeTree::fetchPart(const String & part_name, const Stora { part = get_part(); - auto builder = part->data_part_storage->getBuilder(); if (!to_detached) { Transaction transaction(*this, NO_TRANSACTION_RAW); - renameTempPartAndReplace(part, transaction, builder); + renameTempPartAndReplace(part, transaction); replaced_parts = checkPartChecksumsAndCommit(transaction, part, hardlinked_files); @@ -4037,8 +4038,7 @@ bool StorageReplicatedMergeTree::fetchPart(const String & part_name, const Stora { // The fetched part is valuable and should not be cleaned like a temp part. part->is_temp = false; - part->renameTo(fs::path("detached") / part_name, true, builder); - builder->commit(); + part->renameTo(fs::path("detached") / part_name, true); } } catch (const Exception & e) @@ -4072,7 +4072,7 @@ bool StorageReplicatedMergeTree::fetchPart(const String & part_name, const Stora } -DataPartStoragePtr StorageReplicatedMergeTree::fetchExistsPart( +MutableDataPartStoragePtr StorageReplicatedMergeTree::fetchExistsPart( const String & part_name, const StorageMetadataPtr & metadata_snapshot, const String & source_replica_path, @@ -4147,14 +4147,11 @@ DataPartStoragePtr StorageReplicatedMergeTree::fetchExistsPart( { part = get_part(); - if (part->data_part_storage->getDiskName() != replaced_disk->getName()) - throw Exception("Part " + part->name + " fetched on wrong disk " + part->data_part_storage->getDiskName(), ErrorCodes::LOGICAL_ERROR); + if (part->getDataPartStorage().getDiskName() != replaced_disk->getName()) + throw Exception("Part " + part->name + " fetched on wrong disk " + part->getDataPartStorage().getDiskName(), ErrorCodes::LOGICAL_ERROR); auto replaced_path = fs::path(replaced_part_path); - auto builder = part->data_part_storage->getBuilder(); - builder->rename(replaced_path.parent_path(), replaced_path.filename(), nullptr, true, false); - part->data_part_storage->onRename(replaced_path.parent_path(), replaced_path.filename()); - builder->commit(); + part->getDataPartStorage().rename(replaced_path.parent_path(), replaced_path.filename(), nullptr, true, false); } catch (const Exception & e) { @@ -4177,8 +4174,7 @@ DataPartStoragePtr StorageReplicatedMergeTree::fetchExistsPart( ProfileEvents::increment(ProfileEvents::ReplicatedPartFetches); LOG_DEBUG(log, "Fetched part {} from {}", part_name, source_replica_path); - - return part->data_part_storage; + return part->getDataPartStoragePtr(); } void StorageReplicatedMergeTree::startup() @@ -6237,11 +6233,11 @@ void StorageReplicatedMergeTree::clearOldPartsAndRemoveFromZK() } -void StorageReplicatedMergeTree::removePartsFromZooKeeperWithRetries(DataPartsVector & parts, size_t max_retries) +void StorageReplicatedMergeTree::removePartsFromZooKeeperWithRetries(PartsToRemoveFromZooKeeper & parts, size_t max_retries) { Strings part_names_to_remove; for (const auto & part : parts) - part_names_to_remove.emplace_back(part->name); + part_names_to_remove.emplace_back(part.getPartName()); return removePartsFromZooKeeperWithRetries(part_names_to_remove, max_retries); } @@ -6568,7 +6564,7 @@ void StorageReplicatedMergeTree::replacePartitionFrom( if (replace) clearBlocksInPartition(*zookeeper, drop_range.partition_id, drop_range.max_block, drop_range.max_block); - DataPartsVector parts_to_remove; + PartsToRemoveFromZooKeeper parts_to_remove; Coordination::Responses op_results; try @@ -6594,10 +6590,7 @@ void StorageReplicatedMergeTree::replacePartitionFrom( { auto data_parts_lock = lockParts(); for (auto & part : dst_parts) - { - auto builder = part->data_part_storage->getBuilder(); - renameTempPartAndReplaceUnlocked(part, transaction, builder, data_parts_lock); - } + renameTempPartAndReplaceUnlocked(part, transaction, data_parts_lock); } for (size_t i = 0; i < dst_parts.size(); ++i) @@ -6807,7 +6800,7 @@ void StorageReplicatedMergeTree::movePartitionToTable(const StoragePtr & dest_ta clearBlocksInPartition(*zookeeper, drop_range.partition_id, drop_range.max_block, drop_range.max_block); - DataPartsVector parts_to_remove; + PartsToRemoveFromZooKeeper parts_to_remove; Coordination::Responses op_results; try @@ -6833,10 +6826,7 @@ void StorageReplicatedMergeTree::movePartitionToTable(const StoragePtr & dest_ta auto dest_data_parts_lock = dest_table_storage->lockParts(); for (auto & part : dst_parts) - { - auto builder = part->data_part_storage->getBuilder(); - dest_table_storage->renameTempPartAndReplaceUnlocked(part, transaction, builder, dest_data_parts_lock); - } + dest_table_storage->renameTempPartAndReplaceUnlocked(part, transaction, dest_data_parts_lock); for (size_t i = 0; i < dst_parts.size(); ++i) dest_table_storage->lockSharedData(*dst_parts[i], false, hardlinked_files_for_parts[i]); @@ -7425,7 +7415,7 @@ void StorageReplicatedMergeTree::checkBrokenDisks() for (auto & part : *parts) { - if (part->data_part_storage && part->data_part_storage->getDiskName() == disk_ptr->getName()) + if (part->getDataPartStorage().getDiskName() == disk_ptr->getName()) broken_part_callback(part->name); } continue; @@ -7588,10 +7578,10 @@ void StorageReplicatedMergeTree::lockSharedData(const IMergeTreeDataPart & part, { auto settings = getSettings(); - if (!part.data_part_storage || !part.isStoredOnDisk() || !settings->allow_remote_fs_zero_copy_replication) + if (!part.isStoredOnDisk() || !settings->allow_remote_fs_zero_copy_replication) return; - if (!part.data_part_storage->supportZeroCopyReplication()) + if (!part.getDataPartStorage().supportZeroCopyReplication()) return; zkutil::ZooKeeperPtr zookeeper = tryGetZooKeeper(); @@ -7602,7 +7592,7 @@ void StorageReplicatedMergeTree::lockSharedData(const IMergeTreeDataPart & part, boost::replace_all(id, "/", "_"); Strings zc_zookeeper_paths = getZeroCopyPartPath( - *getSettings(), part.data_part_storage->getDiskType(), getTableSharedID(), + *getSettings(), part.getDataPartStorage().getDiskType(), getTableSharedID(), part.name, zookeeper_path); String path_to_set_hardlinked_files; @@ -7611,7 +7601,7 @@ void StorageReplicatedMergeTree::lockSharedData(const IMergeTreeDataPart & part, if (hardlinked_files.has_value() && !hardlinked_files->hardlinks_from_source_part.empty()) { path_to_set_hardlinked_files = getZeroCopyPartPath( - *getSettings(), part.data_part_storage->getDiskType(), hardlinked_files->source_table_shared_id, + *getSettings(), part.getDataPartStorage().getDiskType(), hardlinked_files->source_table_shared_id, hardlinked_files->source_part_name, zookeeper_path)[0]; hardlinks = hardlinked_files->hardlinks_from_source_part; @@ -7635,25 +7625,22 @@ std::pair StorageReplicatedMergeTree::unlockSharedData(const IMer if (!settings->allow_remote_fs_zero_copy_replication) return std::make_pair(true, NameSet{}); - if (!part.data_part_storage) - LOG_WARNING(log, "Datapart storage for part {} (temp: {}) is not initialzied", part.name, part.is_temp); - - if (!part.data_part_storage || !part.isStoredOnDisk()) + if (!part.isStoredOnDisk()) { LOG_TRACE(log, "Part {} is not stored on disk, blobs can be removed", part.name); return std::make_pair(true, NameSet{}); } - if (!part.data_part_storage || !part.data_part_storage->supportZeroCopyReplication()) + if (!part.getDataPartStorage().supportZeroCopyReplication()) { LOG_TRACE(log, "Part {} is not stored on zero-copy replicated disk, blobs can be removed", part.name); return std::make_pair(true, NameSet{}); } /// If part is temporary refcount file may be absent - if (part.data_part_storage->exists(IMergeTreeDataPart::FILE_FOR_REFERENCES_CHECK)) + if (part.getDataPartStorage().exists(IMergeTreeDataPart::FILE_FOR_REFERENCES_CHECK)) { - auto ref_count = part.data_part_storage->getRefCount(IMergeTreeDataPart::FILE_FOR_REFERENCES_CHECK); + auto ref_count = part.getDataPartStorage().getRefCount(IMergeTreeDataPart::FILE_FOR_REFERENCES_CHECK); if (ref_count > 0) /// Keep part shard info for frozen backups { LOG_TRACE(log, "Part {} has more than zero local references ({}), blobs cannot be removed", part.name, ref_count); @@ -7691,7 +7678,7 @@ std::pair StorageReplicatedMergeTree::unlockSharedData(const IMer return unlockSharedDataByID( part.getUniqueId(), getTableSharedID(), part.name, replica_name, - part.data_part_storage->getDiskType(), zookeeper, *getSettings(), log, zookeeper_path, format_version); + part.getDataPartStorage().getDiskType(), zookeeper, *getSettings(), log, zookeeper_path, format_version); } namespace @@ -7890,7 +7877,7 @@ std::pair StorageReplicatedMergeTree::unlockSharedDataByID( } -DataPartStoragePtr StorageReplicatedMergeTree::tryToFetchIfShared( +MutableDataPartStoragePtr StorageReplicatedMergeTree::tryToFetchIfShared( const IMergeTreeDataPart & part, const DiskPtr & disk, const String & path) @@ -8125,15 +8112,13 @@ bool StorageReplicatedMergeTree::createEmptyPartInsteadOfLost(zkutil::ZooKeeperP minmax_idx->update(block, getMinMaxColumnsNames(metadata_snapshot->getPartitionKey())); auto new_volume = createVolumeFromReservation(reservation, volume); + auto data_part_storage = std::make_shared( new_volume, relative_data_path, TMP_PREFIX + lost_part_name); - DataPartStorageBuilderPtr data_part_storage_builder = std::make_shared( - new_volume, - relative_data_path, - TMP_PREFIX + lost_part_name); + data_part_storage->beginTransaction(); auto new_data_part = createPart( lost_part_name, @@ -8176,16 +8161,16 @@ bool StorageReplicatedMergeTree::createEmptyPartInsteadOfLost(zkutil::ZooKeeperP if (new_data_part->isStoredOnDisk()) { /// The name could be non-unique in case of stale files from previous runs. - if (data_part_storage_builder->exists()) + if (data_part_storage->exists()) { - LOG_WARNING(log, "Removing old temporary directory {}", new_data_part->data_part_storage->getFullPath()); - data_part_storage_builder->removeRecursive(); + LOG_WARNING(log, "Removing old temporary directory {}", new_data_part->getDataPartStorage().getFullPath()); + data_part_storage->removeRecursive(); } - data_part_storage_builder->createDirectories(); + data_part_storage->createDirectories(); if (getSettings()->fsync_part_directory) - sync_guard = data_part_storage_builder->getDirectorySyncGuard(); + sync_guard = data_part_storage->getDirectorySyncGuard(); } /// This effectively chooses minimal compression method: @@ -8193,7 +8178,7 @@ bool StorageReplicatedMergeTree::createEmptyPartInsteadOfLost(zkutil::ZooKeeperP auto compression_codec = getContext()->chooseCompressionCodec(0, 0); const auto & index_factory = MergeTreeIndexFactory::instance(); - MergedBlockOutputStream out(new_data_part, data_part_storage_builder, metadata_snapshot, columns, + MergedBlockOutputStream out(new_data_part, metadata_snapshot, columns, index_factory.getMany(metadata_snapshot->getSecondaryIndices()), compression_codec, NO_TRANSACTION_PTR); bool sync_on_insert = settings->fsync_after_insert; @@ -8207,7 +8192,7 @@ bool StorageReplicatedMergeTree::createEmptyPartInsteadOfLost(zkutil::ZooKeeperP try { MergeTreeData::Transaction transaction(*this, NO_TRANSACTION_RAW); - auto replaced_parts = renameTempPartAndReplace(new_data_part, transaction, data_part_storage_builder); + auto replaced_parts = renameTempPartAndReplace(new_data_part, transaction); if (!replaced_parts.empty()) { diff --git a/src/Storages/StorageReplicatedMergeTree.h b/src/Storages/StorageReplicatedMergeTree.h index 436defebe1d..323b1ce02bf 100644 --- a/src/Storages/StorageReplicatedMergeTree.h +++ b/src/Storages/StorageReplicatedMergeTree.h @@ -263,7 +263,7 @@ public: bool canExecuteFetch(const ReplicatedMergeTreeLogEntry & entry, String & disable_reason) const; /// Fetch part only when it stored on shared storage like S3 - DataPartStoragePtr executeFetchShared(const String & source_replica, const String & new_part_name, const DiskPtr & disk, const String & path); + MutableDataPartStoragePtr executeFetchShared(const String & source_replica, const String & new_part_name, const DiskPtr & disk, const String & path); /// Lock part in zookeeper for use shared data in several nodes void lockSharedData(const IMergeTreeDataPart & part, bool replace_existing_lock, std::optional hardlinked_files) const override; @@ -283,7 +283,7 @@ public: const String & zookeeper_path_old, MergeTreeDataFormatVersion data_format_version); /// Fetch part only if some replica has it on shared storage like S3 - DataPartStoragePtr tryToFetchIfShared(const IMergeTreeDataPart & part, const DiskPtr & disk, const String & path) override; + MutableDataPartStoragePtr tryToFetchIfShared(const IMergeTreeDataPart & part, const DiskPtr & disk, const String & path) override; /// Get best replica having this partition on a same type remote disk String getSharedDataReplica(const IMergeTreeDataPart & part, DataSourceType data_source_type) const; @@ -549,7 +549,7 @@ private: /// Remove parts from ZooKeeper, throw exception if unable to do so after max_retries. void removePartsFromZooKeeperWithRetries(const Strings & part_names, size_t max_retries = 5); - void removePartsFromZooKeeperWithRetries(DataPartsVector & parts, size_t max_retries = 5); + void removePartsFromZooKeeperWithRetries(PartsToRemoveFromZooKeeper & parts, size_t max_retries = 5); /// Removes a part from ZooKeeper and adds a task to the queue to download it. It is supposed to do this with broken parts. void removePartAndEnqueueFetch(const String & part_name); @@ -682,7 +682,7 @@ private: * Used for replace local part on the same s3-shared part in hybrid storage. * Returns false if part is already fetching right now. */ - DataPartStoragePtr fetchExistsPart( + MutableDataPartStoragePtr fetchExistsPart( const String & part_name, const StorageMetadataPtr & metadata_snapshot, const String & replica_path, diff --git a/src/Storages/StorageS3.cpp b/src/Storages/StorageS3.cpp index 4a41659a3fb..9bbccf5f582 100644 --- a/src/Storages/StorageS3.cpp +++ b/src/Storages/StorageS3.cpp @@ -364,39 +364,6 @@ String StorageS3Source::KeysIterator::next() return pimpl->next(); } -class StorageS3Source::ReadTasksIterator::Impl -{ -public: - explicit Impl(const std::vector & read_tasks_, const ReadTaskCallback & new_read_tasks_callback_) - : read_tasks(read_tasks_), new_read_tasks_callback(new_read_tasks_callback_) - { - } - - String next() - { - size_t current_index = index.fetch_add(1, std::memory_order_relaxed); - if (current_index >= read_tasks.size()) - return new_read_tasks_callback(); - return read_tasks[current_index]; - } - -private: - std::atomic_size_t index = 0; - std::vector read_tasks; - ReadTaskCallback new_read_tasks_callback; -}; - -StorageS3Source::ReadTasksIterator::ReadTasksIterator( - const std::vector & read_tasks_, const ReadTaskCallback & new_read_tasks_callback_) - : pimpl(std::make_shared(read_tasks_, new_read_tasks_callback_)) -{ -} - -String StorageS3Source::ReadTasksIterator::next() -{ - return pimpl->next(); -} - Block StorageS3Source::getHeader(Block sample_block, const std::vector & requested_virtual_columns) { for (const auto & virtual_column : requested_virtual_columns) @@ -806,8 +773,7 @@ StorageS3::StorageS3( distributed_processing_, is_key_with_globs, format_settings, - context_, - &read_tasks_used_in_schema_inference); + context_); storage_metadata.setColumns(columns); } else @@ -835,19 +801,14 @@ std::shared_ptr StorageS3::createFileIterator( ContextPtr local_context, ASTPtr query, const Block & virtual_block, - const std::vector & read_tasks, std::unordered_map * object_infos, Strings * read_keys) { if (distributed_processing) { return std::make_shared( - [read_tasks_iterator = std::make_shared(read_tasks, local_context->getReadTaskCallback()), read_keys]() -> String - { - auto key = read_tasks_iterator->next(); - if (read_keys) - read_keys->push_back(key); - return key; + [callback = local_context->getReadTaskCallback()]() -> String { + return callback(); }); } else if (is_key_with_globs) @@ -907,7 +868,6 @@ Pipe StorageS3::read( local_context, query_info.query, virtual_block, - read_tasks_used_in_schema_inference, &object_infos); ColumnsDescription columns_description; @@ -1205,7 +1165,7 @@ ColumnsDescription StorageS3::getTableStructureFromData( return getTableStructureFromDataImpl( configuration.format, s3_configuration, configuration.compression_method, distributed_processing, - s3_configuration.uri.key.find_first_of("*?{") != std::string::npos, format_settings, ctx, nullptr, object_infos); + s3_configuration.uri.key.find_first_of("*?{") != std::string::npos, format_settings, ctx, object_infos); } ColumnsDescription StorageS3::getTableStructureFromDataImpl( @@ -1216,13 +1176,12 @@ ColumnsDescription StorageS3::getTableStructureFromDataImpl( bool is_key_with_globs, const std::optional & format_settings, ContextPtr ctx, - std::vector * read_keys_in_distributed_processing, std::unordered_map * object_infos) { std::vector read_keys; auto file_iterator - = createFileIterator(s3_configuration, {s3_configuration.uri.key}, is_key_with_globs, distributed_processing, ctx, nullptr, {}, {}, object_infos, &read_keys); + = createFileIterator(s3_configuration, {s3_configuration.uri.key}, is_key_with_globs, distributed_processing, ctx, nullptr, {}, object_infos, &read_keys); std::optional columns_from_cache; size_t prev_read_keys_size = read_keys.size(); @@ -1275,9 +1234,6 @@ ColumnsDescription StorageS3::getTableStructureFromDataImpl( if (ctx->getSettingsRef().schema_inference_use_cache_for_s3) addColumnsToCache(read_keys, s3_configuration, columns, format, format_settings, ctx); - if (distributed_processing && read_keys_in_distributed_processing) - *read_keys_in_distributed_processing = std::move(read_keys); - return columns; } diff --git a/src/Storages/StorageS3.h b/src/Storages/StorageS3.h index 8e79516ba4c..23947a32092 100644 --- a/src/Storages/StorageS3.h +++ b/src/Storages/StorageS3.h @@ -66,18 +66,6 @@ public: std::shared_ptr pimpl; }; - class ReadTasksIterator - { - public: - ReadTasksIterator(const std::vector & read_tasks_, const ReadTaskCallback & new_read_tasks_callback_); - String next(); - - private: - class Impl; - /// shared_ptr to have copy constructor - std::shared_ptr pimpl; - }; - using IteratorWrapper = std::function; static Block getHeader(Block sample_block, const std::vector & requested_virtual_columns); @@ -238,8 +226,6 @@ private: ASTPtr partition_by; bool is_key_with_globs = false; - std::vector read_tasks_used_in_schema_inference; - std::unordered_map object_infos; static void updateS3Configuration(ContextPtr, S3Configuration &); @@ -252,7 +238,6 @@ private: ContextPtr local_context, ASTPtr query, const Block & virtual_block, - const std::vector & read_tasks = {}, std::unordered_map * object_infos = nullptr, Strings * read_keys = nullptr); @@ -264,7 +249,6 @@ private: bool is_key_with_globs, const std::optional & format_settings, ContextPtr ctx, - std::vector * read_keys_in_distributed_processing = nullptr, std::unordered_map * object_infos = nullptr); bool supportsSubsetOfColumns() const override; diff --git a/src/Storages/StorageS3Cluster.cpp b/src/Storages/StorageS3Cluster.cpp index 0e4e51f7926..3b8c8b1cb92 100644 --- a/src/Storages/StorageS3Cluster.cpp +++ b/src/Storages/StorageS3Cluster.cpp @@ -5,46 +5,40 @@ #if USE_AWS_S3 #include "Common/Exception.h" -#include #include "Client/Connection.h" #include "Core/QueryProcessingStage.h" -#include -#include -#include #include -#include #include #include #include #include #include #include -#include #include #include #include -#include "Processors/ISource.h" #include #include #include #include +#include #include #include #include +#include +#include #include #include #include #include -#include #include #include -#include -#include namespace DB { + StorageS3Cluster::StorageS3Cluster( const StorageS3ClusterConfiguration & configuration_, const StorageID & table_id_, @@ -72,6 +66,7 @@ StorageS3Cluster::StorageS3Cluster( auto columns = StorageS3::getTableStructureFromDataImpl(format_name, s3_configuration, compression_method, /*distributed_processing_*/false, is_key_with_globs, /*format_settings=*/std::nullopt, context_); storage_metadata.setColumns(columns); + add_columns_structure_to_query = true; } else storage_metadata.setColumns(columns_); @@ -117,6 +112,11 @@ Pipe StorageS3Cluster::read( const bool add_agg_info = processed_stage == QueryProcessingStage::WithMergeableState; + ASTPtr query_to_send = query_info.original_query->clone(); + if (add_columns_structure_to_query) + addColumnsStructureToQueryWithClusterEngine( + query_to_send, StorageDictionary::generateNamesAndTypesDescription(storage_snapshot->metadata->getColumns().getAll()), 5, getName()); + for (const auto & replicas : cluster->getShardsAddresses()) { /// There will be only one replica, because we consider each replica as a shard @@ -135,7 +135,7 @@ Pipe StorageS3Cluster::read( /// So, task_identifier is passed as constructor argument. It is more obvious. auto remote_query_executor = std::make_shared( connection, - queryToString(query_info.original_query), + queryToString(query_to_send), header, context, /*throttler=*/nullptr, diff --git a/src/Storages/StorageS3Cluster.h b/src/Storages/StorageS3Cluster.h index 194c2ed0103..3a3942f4222 100644 --- a/src/Storages/StorageS3Cluster.h +++ b/src/Storages/StorageS3Cluster.h @@ -46,6 +46,7 @@ private: String compression_method; NamesAndTypesList virtual_columns; Block virtual_block; + bool add_columns_structure_to_query = false; }; diff --git a/src/Storages/StorageSnapshot.cpp b/src/Storages/StorageSnapshot.cpp index a99fec8c154..48851f0974d 100644 --- a/src/Storages/StorageSnapshot.cpp +++ b/src/Storages/StorageSnapshot.cpp @@ -76,7 +76,7 @@ std::optional StorageSnapshot::tryGetColumn(const GetColumnsOpt { const auto & columns = getMetadataForQuery()->getColumns(); auto column = columns.tryGetColumn(options, column_name); - if (column && (!isObject(column->type) || !options.with_extended_objects)) + if (column && (!column->type->hasDynamicSubcolumns() || !options.with_extended_objects)) return column; if (options.with_extended_objects) diff --git a/src/Storages/System/StorageSystemParts.cpp b/src/Storages/System/StorageSystemParts.cpp index d788efd8860..fa1c26b623d 100644 --- a/src/Storages/System/StorageSystemParts.cpp +++ b/src/Storages/System/StorageSystemParts.cpp @@ -198,9 +198,9 @@ void StorageSystemParts::processNextStorage( if (part->isStoredOnDisk()) { if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getDiskName()); + columns[res_index++]->insert(part->getDataPartStorage().getDiskName()); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getFullPath()); + columns[res_index++]->insert(part->getDataPartStorage().getFullPath()); } else { diff --git a/src/Storages/System/StorageSystemPartsColumns.cpp b/src/Storages/System/StorageSystemPartsColumns.cpp index cc6e69b160f..cd51c767eae 100644 --- a/src/Storages/System/StorageSystemPartsColumns.cpp +++ b/src/Storages/System/StorageSystemPartsColumns.cpp @@ -190,9 +190,9 @@ void StorageSystemPartsColumns::processNextStorage( if (columns_mask[src_index++]) columns[res_index++]->insert(info.engine); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getDiskName()); + columns[res_index++]->insert(part->getDataPartStorage().getDiskName()); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getFullPath()); + columns[res_index++]->insert(part->getDataPartStorage().getFullPath()); if (columns_mask[src_index++]) columns[res_index++]->insert(column.name); diff --git a/src/Storages/System/StorageSystemProjectionParts.cpp b/src/Storages/System/StorageSystemProjectionParts.cpp index 3934e7c9623..37c62ba5eb0 100644 --- a/src/Storages/System/StorageSystemProjectionParts.cpp +++ b/src/Storages/System/StorageSystemProjectionParts.cpp @@ -200,9 +200,9 @@ void StorageSystemProjectionParts::processNextStorage( if (part->isStoredOnDisk()) { if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getDiskName()); + columns[res_index++]->insert(part->getDataPartStorage().getDiskName()); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getFullPath()); + columns[res_index++]->insert(part->getDataPartStorage().getFullPath()); } else { diff --git a/src/Storages/System/StorageSystemProjectionPartsColumns.cpp b/src/Storages/System/StorageSystemProjectionPartsColumns.cpp index 0847010faaa..a5968597885 100644 --- a/src/Storages/System/StorageSystemProjectionPartsColumns.cpp +++ b/src/Storages/System/StorageSystemProjectionPartsColumns.cpp @@ -211,9 +211,9 @@ void StorageSystemProjectionPartsColumns::processNextStorage( if (columns_mask[src_index++]) columns[res_index++]->insert(info.engine); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getDiskName()); + columns[res_index++]->insert(part->getDataPartStorage().getDiskName()); if (columns_mask[src_index++]) - columns[res_index++]->insert(part->data_part_storage->getFullPath()); + columns[res_index++]->insert(part->getDataPartStorage().getFullPath()); if (columns_mask[src_index++]) columns[res_index++]->insert(column.name); diff --git a/src/Storages/addColumnsStructureToQueryWithClusterEngine.cpp b/src/Storages/addColumnsStructureToQueryWithClusterEngine.cpp new file mode 100644 index 00000000000..31f49fa5490 --- /dev/null +++ b/src/Storages/addColumnsStructureToQueryWithClusterEngine.cpp @@ -0,0 +1,51 @@ +#include +#include +#include +#include +#include +#include +#include + +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +static ASTExpressionList * extractTableFunctionArgumentsFromSelectQuery(ASTPtr & query) +{ + auto * select_query = query->as(); + if (!select_query || !select_query->tables()) + return nullptr; + + auto * tables = select_query->tables()->as(); + auto * table_expression = tables->children[0]->as()->table_expression->as(); + if (!table_expression->table_function) + return nullptr; + + auto * table_function = table_expression->table_function->as(); + return table_function->arguments->as(); +} + +void addColumnsStructureToQueryWithClusterEngine(ASTPtr & query, const String & structure, size_t max_arguments, const String & function_name) +{ + ASTExpressionList * expression_list = extractTableFunctionArgumentsFromSelectQuery(query); + if (!expression_list) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Expected SELECT query from table function {}, got '{}'", function_name, queryToString(query)); + auto structure_literal = std::make_shared(structure); + + if (expression_list->children.size() < 2 || expression_list->children.size() > max_arguments) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Expected 2 to {} arguments in {} table functions, got {}", function_name, max_arguments, expression_list->children.size()); + + if (expression_list->children.size() == 2 || expression_list->children.size() == max_arguments - 1) + { + auto format_literal = std::make_shared("auto"); + expression_list->children.push_back(format_literal); + } + + expression_list->children.push_back(structure_literal); +} + +} diff --git a/src/Storages/addColumnsStructureToQueryWithClusterEngine.h b/src/Storages/addColumnsStructureToQueryWithClusterEngine.h new file mode 100644 index 00000000000..f39f3a31630 --- /dev/null +++ b/src/Storages/addColumnsStructureToQueryWithClusterEngine.h @@ -0,0 +1,11 @@ +#pragma once + +#include + +namespace DB +{ + +/// Add structure argument for queries with s3Cluster/hdfsCluster table function. +void addColumnsStructureToQueryWithClusterEngine(ASTPtr & query, const String & structure, size_t max_arguments, const String & function_name); + +} diff --git a/src/Storages/getStructureOfRemoteTable.cpp b/src/Storages/getStructureOfRemoteTable.cpp index a93a480adb0..7bd5e629c39 100644 --- a/src/Storages/getStructureOfRemoteTable.cpp +++ b/src/Storages/getStructureOfRemoteTable.cpp @@ -200,7 +200,7 @@ ColumnsDescriptionByShardNum getExtendedObjectsOfRemoteTables( auto type_name = type_col[i].get(); auto storage_column = storage_columns.tryGetPhysical(name); - if (storage_column && isObject(storage_column->type)) + if (storage_column && storage_column->type->hasDynamicSubcolumns()) res.add(ColumnDescription(std::move(name), DataTypeFactory::instance().get(type_name))); } } diff --git a/src/TableFunctions/TableFunctionHDFSCluster.cpp b/src/TableFunctions/TableFunctionHDFSCluster.cpp index 26fcb514cca..73b77f770b2 100644 --- a/src/TableFunctions/TableFunctionHDFSCluster.cpp +++ b/src/TableFunctions/TableFunctionHDFSCluster.cpp @@ -48,7 +48,7 @@ void TableFunctionHDFSCluster::parseArguments(const ASTPtr & ast_function, Conte const auto message = fmt::format( "The signature of table function {} shall be the following:\n" \ " - cluster, uri\n",\ - " - cluster, format\n",\ + " - cluster, uri, format\n",\ " - cluster, uri, format, structure\n",\ " - cluster, uri, format, structure, compression_method", getName()); diff --git a/src/TableFunctions/TableFunctionS3.cpp b/src/TableFunctions/TableFunctionS3.cpp index b8e4fcb67fa..4c0b5352545 100644 --- a/src/TableFunctions/TableFunctionS3.cpp +++ b/src/TableFunctions/TableFunctionS3.cpp @@ -64,7 +64,7 @@ void TableFunctionS3::parseArgumentsImpl(const String & error_message, ASTs & ar if (args.size() == 4) { auto second_arg = checkAndGetLiteralArgument(args[1], "format/access_key_id"); - if (FormatFactory::instance().getAllFormats().contains(second_arg)) + if (second_arg == "auto" || FormatFactory::instance().getAllFormats().contains(second_arg)) args_to_idx = {{"format", 1}, {"structure", 2}, {"compression_method", 3}}; else @@ -77,7 +77,7 @@ void TableFunctionS3::parseArgumentsImpl(const String & error_message, ASTs & ar { auto second_arg = checkAndGetLiteralArgument(args[1], "format/access_key_id"); - if (FormatFactory::instance().getAllFormats().contains(second_arg)) + if (second_arg == "auto" || FormatFactory::instance().getAllFormats().contains(second_arg)) args_to_idx = {{"format", 1}, {"structure", 2}}; else args_to_idx = {{"access_key_id", 1}, {"secret_access_key", 2}}; diff --git a/tests/ci/bugfix_validate_check.py b/tests/ci/bugfix_validate_check.py index 4e6001aaa74..e5f37f2940b 100644 --- a/tests/ci/bugfix_validate_check.py +++ b/tests/ci/bugfix_validate_check.py @@ -3,14 +3,21 @@ import argparse import csv import itertools +import logging import os -import sys + +from github import Github + +from s3_helper import S3Helper +from get_robot_token import get_best_robot_token +from pr_info import PRInfo +from upload_result_helper import upload_results +from commit_status_helper import post_commit_status def parse_args(): parser = argparse.ArgumentParser() - parser.add_argument("report1") - parser.add_argument("report2") + parser.add_argument("status", nargs="+", help="Path to status file") return parser.parse_args() @@ -26,20 +33,63 @@ def post_commit_status_from_file(file_path): return res[0] -def process_results(file_path): +def process_result(file_path): + test_results = [] state, report_url, description = post_commit_status_from_file(file_path) prefix = os.path.basename(os.path.dirname(file_path)) - print( - f"::notice:: bugfix check: {prefix} - {state}: {description} Report url: {report_url}" - ) - return state == "success" + is_ok = state == "success" + if is_ok and report_url == "null": + return is_ok, None + + status = f'OK: Bug reproduced (Report' + if not is_ok: + status = f'Bug is not reproduced (Report)' + test_results.append([f"{prefix}: {description}", status]) + return is_ok, test_results + + +def process_all_results(file_paths): + any_ok = False + all_results = [] + for status_path in file_paths: + is_ok, test_results = process_result(status_path) + any_ok = any_ok or is_ok + if test_results is not None: + all_results.extend(test_results) + + return any_ok, all_results def main(args): - is_ok = False - is_ok = process_results(args.report1) or is_ok - is_ok = process_results(args.report2) or is_ok - sys.exit(0 if is_ok else 1) + logging.basicConfig(level=logging.INFO) + + check_name_with_group = "Bugfix validate check" + + is_ok, test_results = process_all_results(args.status) + + if not test_results: + logging.info("No results to upload") + return + + pr_info = PRInfo() + report_url = upload_results( + S3Helper(), + pr_info.number, + pr_info.sha, + test_results, + [], + check_name_with_group, + ) + + gh = Github(get_best_robot_token(), per_page=100) + post_commit_status( + gh, + pr_info.sha, + check_name_with_group, + "" if is_ok else "Changed tests doesn't reproduce the bug", + "success" if is_ok else "error", + report_url, + ) if __name__ == "__main__": diff --git a/tests/ci/cancel_and_rerun_workflow_lambda/app.py b/tests/ci/cancel_and_rerun_workflow_lambda/app.py index 813ee9d1ab7..21a5ce517f6 100644 --- a/tests/ci/cancel_and_rerun_workflow_lambda/app.py +++ b/tests/ci/cancel_and_rerun_workflow_lambda/app.py @@ -15,7 +15,7 @@ import boto3 # type: ignore NEED_RERUN_OR_CANCELL_WORKFLOWS = { "PullRequestCI", "DocsCheck", - "DocsRelease", + "DocsReleaseChecks", "BackportPR", } diff --git a/tests/ci/clickhouse_helper.py b/tests/ci/clickhouse_helper.py index a81334860d1..c82d9da05e9 100644 --- a/tests/ci/clickhouse_helper.py +++ b/tests/ci/clickhouse_helper.py @@ -37,12 +37,8 @@ class ClickHouseHelper: url, params=params, data=json_str, headers=auth ) except Exception as e: - logging.warning( - "Received exception while sending data to %s on %s attempt: %s", - url, - i, - e, - ) + error = f"Received exception while sending data to {url} on {i} attempt: {e}" + logging.warning(error) continue logging.info("Response content '%s'", response.content) diff --git a/tests/ci/functional_test_check.py b/tests/ci/functional_test_check.py index 388f93f34ec..f7d3288c316 100644 --- a/tests/ci/functional_test_check.py +++ b/tests/ci/functional_test_check.py @@ -210,7 +210,10 @@ if __name__ == "__main__": run_changed_tests = flaky_check or validate_bugix_check gh = Github(get_best_robot_token(), per_page=100) - pr_info = PRInfo(need_changed_files=run_changed_tests) + # For validate_bugix_check we need up to date information about labels, so pr_event_from_api is used + pr_info = PRInfo( + need_changed_files=run_changed_tests, pr_event_from_api=validate_bugix_check + ) atexit.register(update_mergeable_check, gh, pr_info, check_name) @@ -221,11 +224,11 @@ if __name__ == "__main__": if args.post_commit_status == "file": post_commit_status_to_file( os.path.join(temp_path, "post_commit_status.tsv"), - "Skipped (no pr-bugfix)", + f"Skipped (no pr-bugfix in {pr_info.labels})", "success", "null", ) - logging.info("Skipping '%s' (no pr-bugfix)", check_name) + logging.info("Skipping '%s' (no pr-bugfix in %s)", check_name, pr_info.labels) sys.exit(0) if "RUN_BY_HASH_NUM" in os.environ: @@ -320,7 +323,7 @@ if __name__ == "__main__": state, description, test_results, additional_logs = process_results( result_path, server_log_path ) - state = override_status(state, check_name, validate_bugix_check) + state = override_status(state, check_name, invert=validate_bugix_check) ch_helper = ClickHouseHelper() mark_flaky_tests(ch_helper, check_name, test_results) diff --git a/tests/ci/integration_test_check.py b/tests/ci/integration_test_check.py index 3709a7271d7..cba428cbcf5 100644 --- a/tests/ci/integration_test_check.py +++ b/tests/ci/integration_test_check.py @@ -167,17 +167,22 @@ if __name__ == "__main__": os.makedirs(temp_path) is_flaky_check = "flaky" in check_name - pr_info = PRInfo(need_changed_files=is_flaky_check or validate_bugix_check) + + # For validate_bugix_check we need up to date information about labels, so pr_event_from_api is used + pr_info = PRInfo( + need_changed_files=is_flaky_check or validate_bugix_check, + pr_event_from_api=validate_bugix_check, + ) if validate_bugix_check and "pr-bugfix" not in pr_info.labels: if args.post_commit_status == "file": post_commit_status_to_file( os.path.join(temp_path, "post_commit_status.tsv"), - "Skipped (no pr-bugfix)", + f"Skipped (no pr-bugfix in {pr_info.labels})", "success", "null", ) - logging.info("Skipping '%s' (no pr-bugfix)", check_name) + logging.info("Skipping '%s' (no pr-bugfix in '%s')", check_name, pr_info.labels) sys.exit(0) gh = Github(get_best_robot_token(), per_page=100) @@ -244,7 +249,7 @@ if __name__ == "__main__": subprocess.check_call(f"sudo chown -R ubuntu:ubuntu {temp_path}", shell=True) state, description, test_results, additional_logs = process_results(result_path) - state = override_status(state, check_name, validate_bugix_check) + state = override_status(state, check_name, invert=validate_bugix_check) ch_helper = ClickHouseHelper() mark_flaky_tests(ch_helper, check_name, test_results) diff --git a/tests/ci/workflow_approve_rerun_lambda/app.py b/tests/ci/workflow_approve_rerun_lambda/app.py index 39bd9cfb283..f2b785840d8 100644 --- a/tests/ci/workflow_approve_rerun_lambda/app.py +++ b/tests/ci/workflow_approve_rerun_lambda/app.py @@ -61,11 +61,11 @@ TRUSTED_WORKFLOW_IDS = { NEED_RERUN_WORKFLOWS = { "BackportPR", - "Docs", - "DocsRelease", + "DocsCheck", + "DocsReleaseChecks", "MasterCI", "PullRequestCI", - "ReleaseCI", + "ReleaseBranchCI", } # Individual trusted contirbutors who are not in any trusted organization. diff --git a/tests/integration/test_attach_backup_from_s3_plain/__init__.py b/tests/integration/test_attach_backup_from_s3_plain/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/integration/test_attach_backup_from_s3_plain/configs/disk_s3.xml b/tests/integration/test_attach_backup_from_s3_plain/configs/disk_s3.xml new file mode 100644 index 00000000000..67278694d39 --- /dev/null +++ b/tests/integration/test_attach_backup_from_s3_plain/configs/disk_s3.xml @@ -0,0 +1,34 @@ + + + + + + s3_plain + http://minio1:9001/root/data/disks/disk_s3_plain/ + minio + minio123 + 33554432 + + + s3_plain + + http://minio1:9001/root/data/disks/disk_s3_plain/backup/ + minio + minio123 + 33554432 + + + + + +
+ attach_disk_s3_plain +
+ + + + + + backup_disk_s3_plain + + diff --git a/tests/integration/test_attach_backup_from_s3_plain/test.py b/tests/integration/test_attach_backup_from_s3_plain/test.py new file mode 100644 index 00000000000..35d53d5b8bd --- /dev/null +++ b/tests/integration/test_attach_backup_from_s3_plain/test.py @@ -0,0 +1,40 @@ +# pylint: disable=global-statement +# pylint: disable=line-too-long + +import pytest +from helpers.cluster import ClickHouseCluster + +cluster = ClickHouseCluster(__file__) +node = cluster.add_instance( + "node", + main_configs=["configs/disk_s3.xml"], + with_minio=True, +) + + +@pytest.fixture(scope="module", autouse=True) +def start_cluster(): + try: + cluster.start() + yield + finally: + cluster.shutdown() + + +def test_attach_backup(): + node.query( + f""" + -- BACKUP writes Ordinary like structure + set allow_deprecated_database_ordinary=1; + create database ordinary engine=Ordinary; + + create table ordinary.test_backup_attach engine=MergeTree() order by tuple() as select * from numbers(100); + -- NOTE: name of backup ("backup") is significant. + backup table ordinary.test_backup_attach TO Disk('backup_disk_s3_plain', 'backup'); + + drop table ordinary.test_backup_attach; + attach table ordinary.test_backup_attach (number UInt64) engine=MergeTree() order by tuple() settings storage_policy='attach_policy_s3_plain'; + """ + ) + + assert int(node.query("select count() from ordinary.test_backup_attach")) == 100 diff --git a/tests/integration/test_disk_over_web_server/test.py b/tests/integration/test_disk_over_web_server/test.py index ea6e407a18f..2ccc17db4f4 100644 --- a/tests/integration/test_disk_over_web_server/test.py +++ b/tests/integration/test_disk_over_web_server/test.py @@ -129,6 +129,9 @@ def test_incorrect_usage(cluster): result = node2.query_and_get_error("TRUNCATE TABLE test0") assert "Table is read-only" in result + result = node2.query_and_get_error("OPTIMIZE TABLE test0 FINAL") + assert "Only read-only operations are supported" in result + node2.query("DROP TABLE test0 SYNC") diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/__init__.py b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/config.xml b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/config.xml new file mode 100644 index 00000000000..42a1f962705 --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/config.xml @@ -0,0 +1,4 @@ + + 1 + 250 + diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/host_regexp.xml b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/host_regexp.xml new file mode 100644 index 00000000000..7a2141e6c7e --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/host_regexp.xml @@ -0,0 +1,11 @@ + + + + + + test1\.example\.com$ + + default + + + \ No newline at end of file diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/listen_host.xml b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/listen_host.xml new file mode 100644 index 00000000000..58ef55cd3f3 --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/configs/listen_host.xml @@ -0,0 +1,5 @@ + + :: + 0.0.0.0 + 1 + diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/Corefile b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/Corefile new file mode 100644 index 00000000000..0dd198441dc --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/Corefile @@ -0,0 +1,8 @@ +. { + hosts /example.com { + reload "200ms" + fallthrough + } + forward . 127.0.0.11 + log +} diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/example.com b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/example.com new file mode 100644 index 00000000000..9beb415c290 --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/coredns_config/example.com @@ -0,0 +1 @@ +filled in runtime, but needs to exist in order to be volume mapped in docker \ No newline at end of file diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/scripts/stress_test.py b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/scripts/stress_test.py new file mode 100644 index 00000000000..b8bafb3d0c1 --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/scripts/stress_test.py @@ -0,0 +1,63 @@ +import pycurl +import threading +from io import BytesIO +import sys + +client_ip = sys.argv[1] +server_ip = sys.argv[2] + +mutex = threading.Lock() +success_counter = 0 +number_of_threads = 100 +number_of_iterations = 100 + + +def perform_request(): + + buffer = BytesIO() + crl = pycurl.Curl() + crl.setopt(pycurl.INTERFACE, client_ip) + crl.setopt(crl.WRITEDATA, buffer) + crl.setopt(crl.URL, f"http://{server_ip}:8123/?query=select+1&user=test_dns") + + crl.perform() + + # End curl session + crl.close() + + str_response = buffer.getvalue().decode("iso-8859-1") + expected_response = "1\n" + + mutex.acquire() + + global success_counter + + if str_response == expected_response: + success_counter += 1 + + mutex.release() + + +def perform_multiple_requests(n): + for request_number in range(n): + perform_request() + + +threads = [] + + +for i in range(number_of_threads): + thread = threading.Thread( + target=perform_multiple_requests, args=(number_of_iterations,) + ) + thread.start() + threads.append(thread) + +for thread in threads: + thread.join() + + +if success_counter == number_of_threads * number_of_iterations: + exit(0) + +exit(1) diff --git a/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/test.py b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/test.py new file mode 100644 index 00000000000..62f47579612 --- /dev/null +++ b/tests/integration/test_host_regexp_multiple_ptr_records_concurrent/test.py @@ -0,0 +1,71 @@ +import pytest +from helpers.cluster import ClickHouseCluster, get_docker_compose_path, run_and_check +from time import sleep +import os + +DOCKER_COMPOSE_PATH = get_docker_compose_path() +SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__)) + +cluster = ClickHouseCluster(__file__) + +ch_server = cluster.add_instance( + "clickhouse-server", + with_coredns=True, + main_configs=["configs/config.xml", "configs/listen_host.xml"], + user_configs=["configs/host_regexp.xml"], +) + +client = cluster.add_instance( + "clickhouse-client", +) + + +@pytest.fixture(scope="module") +def started_cluster(): + global cluster + try: + cluster.start() + yield cluster + + finally: + cluster.shutdown() + + +def setup_dns_server(ip): + domains_string = "test3.example.com test2.example.com test1.example.com" + example_file_path = f'{ch_server.env_variables["COREDNS_CONFIG_DIR"]}/example.com' + run_and_check(f"echo '{ip} {domains_string}' > {example_file_path}", shell=True) + + +def setup_ch_server(dns_server_ip): + ch_server.exec_in_container( + (["bash", "-c", f"echo 'nameserver {dns_server_ip}' > /etc/resolv.conf"]) + ) + ch_server.exec_in_container( + (["bash", "-c", "echo 'options ndots:0' >> /etc/resolv.conf"]) + ) + ch_server.query("SYSTEM DROP DNS CACHE") + + +def build_endpoint_v4(ip): + return f"'http://{ip}:8123/?query=SELECT+1&user=test_dns'" + + +def build_endpoint_v6(ip): + return build_endpoint_v4(f"[{ip}]") + + +def test_host_regexp_multiple_ptr_v4(started_cluster): + server_ip = cluster.get_instance_ip("clickhouse-server") + client_ip = cluster.get_instance_ip("clickhouse-client") + dns_server_ip = cluster.get_instance_ip(cluster.coredns_host) + + setup_dns_server(client_ip) + setup_ch_server(dns_server_ip) + + current_dir = os.path.dirname(__file__) + client.copy_file_to_container( + os.path.join(current_dir, "scripts", "stress_test.py"), "stress_test.py" + ) + + client.exec_in_container(["python3", f"stress_test.py", client_ip, server_ip]) diff --git a/tests/performance/line_as_string_parsing.xml b/tests/performance/line_as_string_parsing.xml new file mode 100644 index 00000000000..d9fa1d4fa6e --- /dev/null +++ b/tests/performance/line_as_string_parsing.xml @@ -0,0 +1,9 @@ + + +INSERT INTO FUNCTION file(test_line_as_string.tsv) SELECT randomString(1000) FROM numbers(1000000) SETTINGS engine_file_truncate_on_insert=1 + +SELECT * FROM file(test_line_as_string.tsv, LineAsString) FORMAT Null + +INSERT INTO FUNCTION file(test_line_as_string.tsv) SELECT * FROM numbers(0) SETTINGS engine_file_truncate_on_insert=1 + + diff --git a/tests/performance/url_hits.xml b/tests/performance/url_hits.xml index 4a07c38b83f..46b39f3a6e9 100644 --- a/tests/performance/url_hits.xml +++ b/tests/performance/url_hits.xml @@ -13,10 +13,14 @@ protocol domain + domainRFC domainWithoutWWW + domainWithoutWWWRFC topLevelDomain firstSignificantSubdomain + firstSignificantSubdomainRFC cutToFirstSignificantSubdomain + cutToFirstSignificantSubdomainRFC path pathFull queryString diff --git a/tests/queries/0_stateless/00398_url_functions.reference b/tests/queries/0_stateless/00398_url_functions.reference index 2e5a97b380e..39d740e55cd 100644 --- a/tests/queries/0_stateless/00398_url_functions.reference +++ b/tests/queries/0_stateless/00398_url_functions.reference @@ -124,8 +124,25 @@ example.com example.com com +example.com +example.com +example.com +example.com +example.com +example.com +example.com +example.com +example.com +com + ====CUT TO FIRST SIGNIFICANT SUBDOMAIN WITH WWW==== +www.com +example.com +example.com +example.com +example.com + www.com example.com example.com diff --git a/tests/queries/0_stateless/00398_url_functions.sql b/tests/queries/0_stateless/00398_url_functions.sql.j2 similarity index 69% rename from tests/queries/0_stateless/00398_url_functions.sql rename to tests/queries/0_stateless/00398_url_functions.sql.j2 index cbefde7515a..dd7da2ce6ad 100644 --- a/tests/queries/0_stateless/00398_url_functions.sql +++ b/tests/queries/0_stateless/00398_url_functions.sql.j2 @@ -7,42 +7,28 @@ SELECT protocol('http://127.0.0.1:443/') AS Scheme; SELECT protocol('//127.0.0.1:443/') AS Scheme; SELECT '====HOST===='; -SELECT domain('http://paul@www.example.com:80/') AS Host; -SELECT domain('user:password@example.com:8080') AS Host; -SELECT domain('http://user:password@example.com:8080') AS Host; -SELECT domain('http://user:password@example.com:8080/path?query=value#fragment') AS Host; -SELECT domain('newuser:@example.com') AS Host; -SELECT domain('http://:pass@example.com') AS Host; -SELECT domain(':newpass@example.com') AS Host; -SELECT domain('http://user:pass@example@.com') AS Host; -SELECT domain('http://user:pass:example.com') AS Host; -SELECT domain('http:/paul/example/com') AS Host; -SELECT domain('http://www.example.com?q=4') AS Host; -SELECT domain('http://127.0.0.1:443/') AS Host; -SELECT domain('//www.example.com') AS Host; -SELECT domain('//paul@www.example.com') AS Host; -SELECT domain('www.example.com') as Host; -SELECT domain('example.com') as Host; -SELECT domainWithoutWWW('//paul@www.example.com') AS Host; -SELECT domainWithoutWWW('http://paul@www.example.com:80/') AS Host; -SELECT domainRFC('http://paul@www.example.com:80/') AS Host; -SELECT domainRFC('user:password@example.com:8080') AS Host; -SELECT domainRFC('http://user:password@example.com:8080') AS Host; -SELECT domainRFC('http://user:password@example.com:8080/path?query=value#fragment') AS Host; -SELECT domainRFC('newuser:@example.com') AS Host; -SELECT domainRFC('http://:pass@example.com') AS Host; -SELECT domainRFC(':newpass@example.com') AS Host; -SELECT domainRFC('http://user:pass@example@.com') AS Host; -SELECT domainRFC('http://user:pass:example.com') AS Host; -SELECT domainRFC('http:/paul/example/com') AS Host; -SELECT domainRFC('http://www.example.com?q=4') AS Host; -SELECT domainRFC('http://127.0.0.1:443/') AS Host; -SELECT domainRFC('//www.example.com') AS Host; -SELECT domainRFC('//paul@www.example.com') AS Host; -SELECT domainRFC('www.example.com') as Host; -SELECT domainRFC('example.com') as Host; -SELECT domainWithoutWWWRFC('//paul@www.example.com') AS Host; -SELECT domainWithoutWWWRFC('http://paul@www.example.com:80/') AS Host; +{% for suffix in ['', 'RFC'] -%} + +SELECT domain{{ suffix }}('http://paul@www.example.com:80/') AS Host; +SELECT domain{{ suffix }}('user:password@example.com:8080') AS Host; +SELECT domain{{ suffix }}('http://user:password@example.com:8080') AS Host; +SELECT domain{{ suffix }}('http://user:password@example.com:8080/path?query=value#fragment') AS Host; +SELECT domain{{ suffix }}('newuser:@example.com') AS Host; +SELECT domain{{ suffix }}('http://:pass@example.com') AS Host; +SELECT domain{{ suffix }}(':newpass@example.com') AS Host; +SELECT domain{{ suffix }}('http://user:pass@example@.com') AS Host; +SELECT domain{{ suffix }}('http://user:pass:example.com') AS Host; +SELECT domain{{ suffix }}('http:/paul/example/com') AS Host; +SELECT domain{{ suffix }}('http://www.example.com?q=4') AS Host; +SELECT domain{{ suffix }}('http://127.0.0.1:443/') AS Host; +SELECT domain{{ suffix }}('//www.example.com') AS Host; +SELECT domain{{ suffix }}('//paul@www.example.com') AS Host; +SELECT domain{{ suffix }}('www.example.com') as Host; +SELECT domain{{ suffix }}('example.com') as Host; +SELECT domainWithoutWWW{{ suffix }}('//paul@www.example.com') AS Host; +SELECT domainWithoutWWW{{ suffix }}('http://paul@www.example.com:80/') AS Host; + +{% endfor %} SELECT '====NETLOC===='; SELECT netloc('http://paul@www.example.com:80/') AS Netloc; @@ -121,25 +107,31 @@ SELECT decodeURLComponent(encodeURLComponent('http://paul@127.0.0.1/?query=hello SELECT decodeURLFormComponent(encodeURLFormComponent('http://paul@127.0.0.1/?query=hello world foo+bar#a=b')); SELECT '====CUT TO FIRST SIGNIFICANT SUBDOMAIN===='; -SELECT cutToFirstSignificantSubdomain('http://www.example.com'); -SELECT cutToFirstSignificantSubdomain('http://www.example.com:1234'); -SELECT cutToFirstSignificantSubdomain('http://www.example.com/a/b/c'); -SELECT cutToFirstSignificantSubdomain('http://www.example.com/a/b/c?a=b'); -SELECT cutToFirstSignificantSubdomain('http://www.example.com/a/b/c?a=b#d=f'); -SELECT cutToFirstSignificantSubdomain('http://paul@www.example.com/a/b/c?a=b#d=f'); -SELECT cutToFirstSignificantSubdomain('//paul@www.example.com/a/b/c?a=b#d=f'); -SELECT cutToFirstSignificantSubdomain('www.example.com'); -SELECT cutToFirstSignificantSubdomain('example.com'); -SELECT cutToFirstSignificantSubdomain('www.com'); -SELECT cutToFirstSignificantSubdomain('com'); + +{% for suffix in ['', 'RFC'] -%} +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://www.example.com'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://www.example.com:1234'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://www.example.com/a/b/c'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://www.example.com/a/b/c?a=b'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://www.example.com/a/b/c?a=b#d=f'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('http://paul@www.example.com/a/b/c?a=b#d=f'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('//paul@www.example.com/a/b/c?a=b#d=f'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('www.example.com'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('example.com'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('www.com'); +SELECT cutToFirstSignificantSubdomain{{ suffix }}('com'); +{% endfor %} SELECT '====CUT TO FIRST SIGNIFICANT SUBDOMAIN WITH WWW===='; -SELECT cutToFirstSignificantSubdomainWithWWW('http://com'); -SELECT cutToFirstSignificantSubdomainWithWWW('http://www.com'); -SELECT cutToFirstSignificantSubdomainWithWWW('http://www.example.com'); -SELECT cutToFirstSignificantSubdomainWithWWW('http://www.foo.example.com'); -SELECT cutToFirstSignificantSubdomainWithWWW('http://www.example.com:1'); -SELECT cutToFirstSignificantSubdomainWithWWW('http://www.example.com/'); + +{% for suffix in ['', 'RFC'] -%} +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://com'); +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://www.com'); +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://www.example.com'); +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://www.foo.example.com'); +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://www.example.com:1'); +SELECT cutToFirstSignificantSubdomainWithWWW{{ suffix }}('http://www.example.com/'); +{% endfor %} SELECT '====CUT WWW===='; SELECT cutWWW('http://www.example.com'); diff --git a/tests/queries/0_stateless/00700_to_decimal_or_something.reference b/tests/queries/0_stateless/00700_to_decimal_or_something.reference index 89ded7bd6d4..dec36ed5df5 100644 --- a/tests/queries/0_stateless/00700_to_decimal_or_something.reference +++ b/tests/queries/0_stateless/00700_to_decimal_or_something.reference @@ -1,5 +1,5 @@ 1.1 1.1 1.1 -0 +1 0 0.42 0 0.42 0 0.42 @@ -13,7 +13,7 @@ 0 ---- 1.1 1.1 1.1 -\N +1 \N -0.42 \N -0.42 \N -0.42 diff --git a/tests/queries/0_stateless/00718_format_datetime.reference b/tests/queries/0_stateless/00718_format_datetime.reference index 4f12a46d7c0..bc98dd59d5f 100644 --- a/tests/queries/0_stateless/00718_format_datetime.reference +++ b/tests/queries/0_stateless/00718_format_datetime.reference @@ -1,33 +1,34 @@ -20 +20 20 +02 02 +01/02/18 01/02/18 + 2 2 +2018-01-02 2018-01-02 +22 00 02 -01/02/18 - 2 -2018-01-02 -22 -02 -10 +10 12 11 12 -001 -366 -01 -33 -\n -AM +001 001 +366 366 +01 01 +33 00 +\n \n +AM AM AM PM -22:33 -44 -\t -22:33:44 -1 7 -01 01 53 52 -1 0 -18 -2018 -% -no formatting pattern +22:33 00:00 +44 00 +\t \t +22:33:44 00:00:00 +1 7 1 7 +01 01 53 52 01 01 53 52 +1 0 1 0 +18 18 +2018 2018 +% % +no formatting pattern no formatting pattern 2018-01-01 00:00:00 +1927-01-01 00:00:00 2018-01-01 01:00:00 2018-01-01 04:00:00 +0000 -1100 diff --git a/tests/queries/0_stateless/00718_format_datetime.sql b/tests/queries/0_stateless/00718_format_datetime.sql index 7ed1f0abea4..deb5fb96c6c 100644 --- a/tests/queries/0_stateless/00718_format_datetime.sql +++ b/tests/queries/0_stateless/00718_format_datetime.sql @@ -8,38 +8,44 @@ SELECT formatDateTime(now(), 'unescaped %'); -- { serverError 36 } SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%U'); -- { serverError 48 } SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%W'); -- { serverError 48 } -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%C'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%d'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%D'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%e'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%F'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%H'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%C'), formatDateTime(toDate32('2018-01-02'), '%C'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%d'), formatDateTime(toDate32('2018-01-02'), '%d'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%D'), formatDateTime(toDate32('2018-01-02'), '%D'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%e'), formatDateTime(toDate32('2018-01-02'), '%e'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%F'), formatDateTime(toDate32('2018-01-02'), '%F'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%H'), formatDateTime(toDate32('2018-01-02'), '%H'); SELECT formatDateTime(toDateTime('2018-01-02 02:33:44'), '%H'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%I'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%I'), formatDateTime(toDate32('2018-01-02'), '%I'); SELECT formatDateTime(toDateTime('2018-01-02 11:33:44'), '%I'); SELECT formatDateTime(toDateTime('2018-01-02 00:33:44'), '%I'); -SELECT formatDateTime(toDateTime('2018-01-01 00:33:44'), '%j'); -SELECT formatDateTime(toDateTime('2000-12-31 00:33:44'), '%j'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%m'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%M'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%n'); -SELECT formatDateTime(toDateTime('2018-01-02 00:33:44'), '%p'); +SELECT formatDateTime(toDateTime('2018-01-01 00:33:44'), '%j'), formatDateTime(toDate32('2018-01-01'), '%j'); +SELECT formatDateTime(toDateTime('2000-12-31 00:33:44'), '%j'), formatDateTime(toDate32('2000-12-31'), '%j'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%m'), formatDateTime(toDate32('2018-01-02'), '%m'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%M'), formatDateTime(toDate32('2018-01-02'), '%M'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%n'), formatDateTime(toDate32('2018-01-02'), '%n'); +SELECT formatDateTime(toDateTime('2018-01-02 00:33:44'), '%p'), formatDateTime(toDateTime('2018-01-02'), '%p'); SELECT formatDateTime(toDateTime('2018-01-02 11:33:44'), '%p'); SELECT formatDateTime(toDateTime('2018-01-02 12:33:44'), '%p'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%R'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%S'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%t'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%T'); -SELECT formatDateTime(toDateTime('2018-01-01 22:33:44'), '%u'), formatDateTime(toDateTime('2018-01-07 22:33:44'), '%u'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%R'), formatDateTime(toDate32('2018-01-02'), '%R'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%S'), formatDateTime(toDate32('2018-01-02'), '%S'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%t'), formatDateTime(toDate32('2018-01-02'), '%t'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%T'), formatDateTime(toDate32('2018-01-02'), '%T'); +SELECT formatDateTime(toDateTime('2018-01-01 22:33:44'), '%u'), formatDateTime(toDateTime('2018-01-07 22:33:44'), '%u'), + formatDateTime(toDate32('2018-01-01'), '%u'), formatDateTime(toDate32('2018-01-07'), '%u'); SELECT formatDateTime(toDateTime('1996-01-01 22:33:44'), '%V'), formatDateTime(toDateTime('1996-12-31 22:33:44'), '%V'), - formatDateTime(toDateTime('1999-01-01 22:33:44'), '%V'), formatDateTime(toDateTime('1999-12-31 22:33:44'), '%V'); -SELECT formatDateTime(toDateTime('2018-01-01 22:33:44'), '%w'), formatDateTime(toDateTime('2018-01-07 22:33:44'), '%w'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%y'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%Y'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%%'); -SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), 'no formatting pattern'); + formatDateTime(toDateTime('1999-01-01 22:33:44'), '%V'), formatDateTime(toDateTime('1999-12-31 22:33:44'), '%V'), + formatDateTime(toDate32('1996-01-01'), '%V'), formatDateTime(toDate32('1996-12-31'), '%V'), + formatDateTime(toDate32('1999-01-01'), '%V'), formatDateTime(toDate32('1999-12-31'), '%V'); +SELECT formatDateTime(toDateTime('2018-01-01 22:33:44'), '%w'), formatDateTime(toDateTime('2018-01-07 22:33:44'), '%w'), + formatDateTime(toDate32('2018-01-01'), '%w'), formatDateTime(toDate32('2018-01-07'), '%w'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%y'), formatDateTime(toDate32('2018-01-02'), '%y'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%Y'), formatDateTime(toDate32('2018-01-02'), '%Y'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), '%%'), formatDateTime(toDate32('2018-01-02'), '%%'); +SELECT formatDateTime(toDateTime('2018-01-02 22:33:44'), 'no formatting pattern'), formatDateTime(toDate32('2018-01-02'), 'no formatting pattern'); SELECT formatDateTime(toDate('2018-01-01'), '%F %T'); +SELECT formatDateTime(toDate32('1927-01-01'), '%F %T'); + SELECT formatDateTime(toDateTime('2018-01-01 01:00:00', 'UTC'), '%F %T', 'UTC'), formatDateTime(toDateTime('2018-01-01 01:00:00', 'UTC'), '%F %T', 'Asia/Istanbul'); diff --git a/tests/queries/0_stateless/00732_base64_functions.reference b/tests/queries/0_stateless/00732_base64_functions.reference index b22ae4e7e24..5dc1ba03b89 100644 --- a/tests/queries/0_stateless/00732_base64_functions.reference +++ b/tests/queries/0_stateless/00732_base64_functions.reference @@ -14,3 +14,5 @@ fooba foobar 1 1 +Zm9v +foo diff --git a/tests/queries/0_stateless/00732_base64_functions.sql b/tests/queries/0_stateless/00732_base64_functions.sql index 4ed86e20913..adba0cdebbd 100644 --- a/tests/queries/0_stateless/00732_base64_functions.sql +++ b/tests/queries/0_stateless/00732_base64_functions.sql @@ -14,3 +14,6 @@ SELECT base64Decode(val, 'excess argument') FROM (select arrayJoin(['', 'Zg==', SELECT tryBase64Decode('Zm9vYmF=Zm9v', 'excess argument'); -- { serverError 42 } SELECT base64Decode('Zm9vYmF=Zm9v'); -- { serverError 117 } + +select base64Encode(toFixedString('foo', 3)); +select base64Decode(toFixedString('Zm9v', 4)); diff --git a/tests/queries/0_stateless/00918_json_functions.reference b/tests/queries/0_stateless/00918_json_functions.reference index 8e6fc3914e0..fc03457c677 100644 --- a/tests/queries/0_stateless/00918_json_functions.reference +++ b/tests/queries/0_stateless/00918_json_functions.reference @@ -61,11 +61,47 @@ Friday (1,'417ddc5d-e556-4d27-95dd-a34d84e46a50') hello (3333.6,'test') +(3333.6,'test') +(3333.6333333333,'test') (3333.6333333333,'test') 123456.1234 Decimal(20, 4) +123456.1234 Decimal(20, 4) +123456789012345.12 Decimal(30, 4) +(1234567890.1234567890123456789,'test') Tuple(a Decimal(35, 20), b LowCardinality(String)) +(1234567890.12345678901234567890123456789,'test') Tuple(a Decimal(45, 30), b LowCardinality(String)) 123456789012345.1136 123456789012345.1136 1234567890.12345677879616925706 (1234567890.12345677879616925706,'test') 1234567890.123456695758468374595199311875 (1234567890.123456695758468374595199311875,'test') +-1234567890 Int32 +1234567890 UInt32 +-1234567890123456789 Int64 +1234567890123456789 UInt64 +-1234567890123456789 Int128 +1234567890123456789 UInt128 +-1234567890123456789 Int256 +1234567890123456789 UInt256 +-123456789 Int32 +123456789 UInt32 +-123456789012 Int64 +123456789012 UInt64 +-123456789012 Int128 +123456789012 UInt128 +-123456789012 Int256 +123456789012 UInt256 +-123456789 Int32 +123456789 UInt32 +-1234567890123456789 Int64 +1234567890123456789 UInt64 +-12345678901234567890123456789012345678 Int128 +12345678901234567890123456789012345678 UInt128 +-11345678901234567890123456789012345678901234567890123456789012345678901234567 Int256 +11345678901234567890123456789012345678901234567890123456789012345678901234567 UInt256 +0 Int32 +0 UInt32 +0 Int64 +0 UInt64 +false Bool +true Bool --JSONExtractKeysAndValues-- [('a','hello'),('b','[-100,200,300]')] [('b',[-100,200,300])] @@ -217,3 +253,4 @@ e u v --show error: type should be const string +--show error: index type should be integer diff --git a/tests/queries/0_stateless/00918_json_functions.sql b/tests/queries/0_stateless/00918_json_functions.sql index 87682587c8e..3105994ce20 100644 --- a/tests/queries/0_stateless/00918_json_functions.sql +++ b/tests/queries/0_stateless/00918_json_functions.sql @@ -72,11 +72,47 @@ SELECT JSONExtract('{"a":123456, "b":3.55}', 'Tuple(a LowCardinality(Int32), b D SELECT JSONExtract('{"a":1, "b":"417ddc5d-e556-4d27-95dd-a34d84e46a50"}', 'Tuple(a Int8, b UUID)'); SELECT JSONExtract('{"a": "hello", "b": [-100, 200.0, 300]}', 'a', 'LowCardinality(String)'); SELECT JSONExtract('{"a":3333.6333333333333333333333, "b":"test"}', 'Tuple(a Decimal(10,1), b LowCardinality(String))'); +SELECT JSONExtract('{"a":"3333.6333333333333333333333", "b":"test"}', 'Tuple(a Decimal(10,1), b LowCardinality(String))'); SELECT JSONExtract('{"a":3333.6333333333333333333333, "b":"test"}', 'Tuple(a Decimal(20,10), b LowCardinality(String))'); +SELECT JSONExtract('{"a":"3333.6333333333333333333333", "b":"test"}', 'Tuple(a Decimal(20,10), b LowCardinality(String))'); SELECT JSONExtract('{"a":123456.123456}', 'a', 'Decimal(20, 4)') as a, toTypeName(a); +SELECT JSONExtract('{"a":"123456.123456"}', 'a', 'Decimal(20, 4)') as a, toTypeName(a); +SELECT JSONExtract('{"a":"123456789012345.12"}', 'a', 'Decimal(30, 4)') as a, toTypeName(a); +SELECT JSONExtract('{"a":"1234567890.12345678901234567890", "b":"test"}', 'Tuple(a Decimal(35,20), b LowCardinality(String))') as a, toTypeName(a); +SELECT JSONExtract('{"a":"1234567890.123456789012345678901234567890", "b":"test"}', 'Tuple(a Decimal(45,30), b LowCardinality(String))') as a, toTypeName(a); SELECT toDecimal64(123456789012345.12, 4), JSONExtract('{"a":123456789012345.12}', 'a', 'Decimal(30, 4)'); SELECT toDecimal128(1234567890.12345678901234567890, 20), JSONExtract('{"a":1234567890.12345678901234567890, "b":"test"}', 'Tuple(a Decimal(35,20), b LowCardinality(String))'); SELECT toDecimal256(1234567890.123456789012345678901234567890, 30), JSONExtract('{"a":1234567890.12345678901234567890, "b":"test"}', 'Tuple(a Decimal(45,30), b LowCardinality(String))'); +SELECT JSONExtract('{"a":-1234567890}', 'a', 'Int32') as a, toTypeName(a); +SELECT JSONExtract('{"a":1234567890}', 'a', 'UInt32') as a, toTypeName(a); +SELECT JSONExtract('{"a":-1234567890123456789}', 'a', 'Int64') as a, toTypeName(a); +SELECT JSONExtract('{"a":1234567890123456789}', 'a', 'UInt64') as a, toTypeName(a); +SELECT JSONExtract('{"a":-1234567890123456789}', 'a', 'Int128') as a, toTypeName(a); +SELECT JSONExtract('{"a":1234567890123456789}', 'a', 'UInt128') as a, toTypeName(a); +SELECT JSONExtract('{"a":-1234567890123456789}', 'a', 'Int256') as a, toTypeName(a); +SELECT JSONExtract('{"a":1234567890123456789}', 'a', 'UInt256') as a, toTypeName(a); +SELECT JSONExtract('{"a":-123456789.345}', 'a', 'Int32') as a, toTypeName(a); +SELECT JSONExtract('{"a":123456789.345}', 'a', 'UInt32') as a, toTypeName(a); +SELECT JSONExtract('{"a":-123456789012.345}', 'a', 'Int64') as a, toTypeName(a); +SELECT JSONExtract('{"a":123456789012.345}', 'a', 'UInt64') as a, toTypeName(a); +SELECT JSONExtract('{"a":-123456789012.345}', 'a', 'Int128') as a, toTypeName(a); +SELECT JSONExtract('{"a":123456789012.345}', 'a', 'UInt128') as a, toTypeName(a); +SELECT JSONExtract('{"a":-123456789012.345}', 'a', 'Int256') as a, toTypeName(a); +SELECT JSONExtract('{"a":123456789012.345}', 'a', 'UInt256') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-123456789"}', 'a', 'Int32') as a, toTypeName(a); +SELECT JSONExtract('{"a":"123456789"}', 'a', 'UInt32') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-1234567890123456789"}', 'a', 'Int64') as a, toTypeName(a); +SELECT JSONExtract('{"a":"1234567890123456789"}', 'a', 'UInt64') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-12345678901234567890123456789012345678"}', 'a', 'Int128') as a, toTypeName(a); +SELECT JSONExtract('{"a":"12345678901234567890123456789012345678"}', 'a', 'UInt128') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-11345678901234567890123456789012345678901234567890123456789012345678901234567"}', 'a', 'Int256') as a, toTypeName(a); +SELECT JSONExtract('{"a":"11345678901234567890123456789012345678901234567890123456789012345678901234567"}', 'a', 'UInt256') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-1234567899999"}', 'a', 'Int32') as a, toTypeName(a); +SELECT JSONExtract('{"a":"1234567899999"}', 'a', 'UInt32') as a, toTypeName(a); +SELECT JSONExtract('{"a":"-1234567890123456789999"}', 'a', 'Int64') as a, toTypeName(a); +SELECT JSONExtract('{"a":"1234567890123456789999"}', 'a', 'UInt64') as a, toTypeName(a); +SELECT JSONExtract('{"a":0}', 'a', 'Bool') as a, toTypeName(a); +SELECT JSONExtract('{"a":1}', 'a', 'Bool') as a, toTypeName(a); SELECT '--JSONExtractKeysAndValues--'; SELECT JSONExtractKeysAndValues('{"a": "hello", "b": [-100, 200.0, 300]}', 'String'); @@ -244,3 +280,6 @@ SELECT JSONExtractString(json, 's') FROM (SELECT arrayJoin(['{"s":"u"}', '{"s":" SELECT '--show error: type should be const string'; SELECT JSONExtractKeysAndValues([], JSONLength('^?V{LSwp')); -- { serverError 44 } WITH '{"i": 1, "f": 1.2}' AS json SELECT JSONExtract(json, 'i', JSONType(json, 'i')); -- { serverError 44 } + +SELECT '--show error: index type should be integer'; +SELECT JSONExtract('[]', JSONExtract('0', 'UInt256'), 'UInt256'); -- { serverError 43 } diff --git a/tests/queries/0_stateless/01186_conversion_to_nullable.reference b/tests/queries/0_stateless/01186_conversion_to_nullable.reference index 86fa0afff20..e4c1fd7c40b 100644 --- a/tests/queries/0_stateless/01186_conversion_to_nullable.reference +++ b/tests/queries/0_stateless/01186_conversion_to_nullable.reference @@ -26,7 +26,7 @@ \N 42 \N -\N +3.14 42 \N 3.14159 diff --git a/tests/queries/0_stateless/01284_port.reference b/tests/queries/0_stateless/01284_port.reference index 7e776595065..5b7b58bc7e4 100644 --- a/tests/queries/0_stateless/01284_port.reference +++ b/tests/queries/0_stateless/01284_port.reference @@ -22,3 +22,27 @@ ipv6 0 host-no-dot 0 +ipv4 +0 +80 +80 +80 +80 +hostname +0 +80 +80 +80 +80 +default-port +80 +80 +ipv6 +0 +0 +0 +0 +0 +0 +host-no-dot +0 diff --git a/tests/queries/0_stateless/01284_port.sql b/tests/queries/0_stateless/01284_port.sql deleted file mode 100644 index 9c31a5d42ad..00000000000 --- a/tests/queries/0_stateless/01284_port.sql +++ /dev/null @@ -1,34 +0,0 @@ -select 'ipv4'; -select port('http://127.0.0.1/'); -select port('http://127.0.0.1:80'); -select port('http://127.0.0.1:80/'); -select port('//127.0.0.1:80/'); -select port('127.0.0.1:80'); -select 'hostname'; -select port('http://foobar.com/'); -select port('http://foobar.com:80'); -select port('http://foobar.com:80/'); -select port('//foobar.com:80/'); -select port('foobar.com:80'); - -select 'default-port'; -select port('http://127.0.0.1/', toUInt16(80)); -select port('http://foobar.com/', toUInt16(80)); - --- unsupported -/* ILLEGAL_TYPE_OF_ARGUMENT */ select port(toFixedString('', 1)); -- { serverError 43; } -/* ILLEGAL_TYPE_OF_ARGUMENT */ select port('', 1); -- { serverError 43; } -/* NUMBER_OF_ARGUMENTS_DOESNT_MATCH */ select port('', 1, 1); -- { serverError 42; } - --- --- Known limitations of domain() (getURLHost()) --- -select 'ipv6'; -select port('http://[2001:db8::8a2e:370:7334]/'); -select port('http://[2001:db8::8a2e:370:7334]:80'); -select port('http://[2001:db8::8a2e:370:7334]:80/'); -select port('//[2001:db8::8a2e:370:7334]:80/'); -select port('[2001:db8::8a2e:370:7334]:80'); -select port('2001:db8::8a2e:370:7334:80'); -select 'host-no-dot'; -select port('//foobar:80/'); diff --git a/tests/queries/0_stateless/01284_port.sql.j2 b/tests/queries/0_stateless/01284_port.sql.j2 new file mode 100644 index 00000000000..6f78b3b8e3b --- /dev/null +++ b/tests/queries/0_stateless/01284_port.sql.j2 @@ -0,0 +1,39 @@ +{% for suffix in ['', 'RFC'] -%} + +select 'ipv4'; +select port{{ suffix }}('http://127.0.0.1/'); +select port{{ suffix }}('http://127.0.0.1:80'); +select port{{ suffix }}('http://127.0.0.1:80/'); +select port{{ suffix }}('//127.0.0.1:80/'); +select port{{ suffix }}('127.0.0.1:80'); + +select 'hostname'; +select port{{ suffix }}('http://foobar.com/'); +select port{{ suffix }}('http://foobar.com:80'); +select port{{ suffix }}('http://foobar.com:80/'); +select port{{ suffix }}('//foobar.com:80/'); +select port{{ suffix }}('foobar.com:80'); + +select 'default-port'; +select port{{ suffix }}('http://127.0.0.1/', toUInt16(80)); +select port{{ suffix }}('http://foobar.com/', toUInt16(80)); + +-- unsupported +/* ILLEGAL_TYPE_OF_ARGUMENT */ select port(toFixedString('', 1)); -- { serverError 43; } +/* ILLEGAL_TYPE_OF_ARGUMENT */ select port{{ suffix }}('', 1); -- { serverError 43; } +/* NUMBER_OF_ARGUMENTS_DOESNT_MATCH */ select port{{ suffix }}('', 1, 1); -- { serverError 42; } + +-- +-- Known limitations of domain() (getURLHost()) +-- +select 'ipv6'; +select port{{ suffix }}('http://[2001:db8::8a2e:370:7334]/'); +select port{{ suffix }}('http://[2001:db8::8a2e:370:7334]:80'); +select port{{ suffix }}('http://[2001:db8::8a2e:370:7334]:80/'); +select port{{ suffix }}('//[2001:db8::8a2e:370:7334]:80/'); +select port{{ suffix }}('[2001:db8::8a2e:370:7334]:80'); +select port{{ suffix }}('2001:db8::8a2e:370:7334:80'); +select 'host-no-dot'; +select port{{ suffix }}('//foobar:80/'); + +{%- endfor %} diff --git a/tests/queries/0_stateless/01288_shard_max_network_bandwidth.sql b/tests/queries/0_stateless/01288_shard_max_network_bandwidth.sql index 969bb0a126c..d2daf48a1cb 100644 --- a/tests/queries/0_stateless/01288_shard_max_network_bandwidth.sql +++ b/tests/queries/0_stateless/01288_shard_max_network_bandwidth.sql @@ -1,7 +1,7 @@ -- Tags: shard --- Limit to 10 MB/sec -SET max_network_bandwidth = 10000000; +-- Limit to 100 KB/sec +SET max_network_bandwidth = 100000; -- Lower max_block_size, so we can start throttling sooner. Otherwise query will be executed too quickly. SET max_block_size = 100; @@ -11,7 +11,7 @@ CREATE TEMPORARY TABLE times (t DateTime); -- rand64 is uncompressable data. Each number will take 8 bytes of bandwidth. -- This query should execute in no less than 1.6 seconds if throttled. INSERT INTO times SELECT now(); -SELECT sum(ignore(*)) FROM (SELECT rand64() FROM remote('127.0.0.{2,3}', numbers(2000000))); +SELECT sum(ignore(*)) FROM (SELECT rand64() FROM remote('127.0.0.{2,3}', numbers(20000))); INSERT INTO times SELECT now(); SELECT max(t) - min(t) >= 1 FROM times; diff --git a/tests/queries/0_stateless/01411_from_unixtime.reference b/tests/queries/0_stateless/01411_from_unixtime.reference index 1bc7519e668..17086e8c58b 100644 --- a/tests/queries/0_stateless/01411_from_unixtime.reference +++ b/tests/queries/0_stateless/01411_from_unixtime.reference @@ -5,25 +5,25 @@ 11 1970-01-15 1970-01-15 06:52:36 -20 +20 20 +02 02 +01/02/18 01/02/18 + 2 2 +2018-01-02 2018-01-02 +22 00 02 -01/02/18 - 2 -2018-01-02 -22 -02 -10 +10 12 11 12 -001 -366 -01 -33 -\n -AM +001 001 +366 366 +01 01 +33 00 +\n \n +AM AM AM PM -22:33 -44 -\t -22:33:44 +22:33 00:00 +44 00 +\t \t +22:33:44 00:00:00 diff --git a/tests/queries/0_stateless/01411_from_unixtime.sql b/tests/queries/0_stateless/01411_from_unixtime.sql index ec7b4d65b57..9a6655768e0 100644 --- a/tests/queries/0_stateless/01411_from_unixtime.sql +++ b/tests/queries/0_stateless/01411_from_unixtime.sql @@ -5,25 +5,25 @@ SELECT FROM_UNIXTIME(5345345, '%C', 'UTC'); SELECT FROM_UNIXTIME(645123, '%H', 'UTC'); SELECT FROM_UNIXTIME(1232456, '%Y-%m-%d', 'UTC'); SELECT FROM_UNIXTIME(1234356, '%Y-%m-%d %R:%S', 'UTC'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%C'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%d'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%D'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%e'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%F'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%H'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%C'), FROM_UNIXTIME(toDate32('2018-01-02'), '%C'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%d'), FROM_UNIXTIME(toDate32('2018-01-02'), '%d'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%D'), FROM_UNIXTIME(toDate32('2018-01-02'), '%D'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%e'), FROM_UNIXTIME(toDate32('2018-01-02'), '%e'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%F'), FROM_UNIXTIME(toDate32('2018-01-02'), '%F'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%H'), FROM_UNIXTIME(toDate32('2018-01-02'), '%H'); SELECT FROM_UNIXTIME(toDateTime('2018-01-02 02:33:44'), '%H'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%I'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%I'), FROM_UNIXTIME(toDate32('2018-01-02'), '%I'); SELECT FROM_UNIXTIME(toDateTime('2018-01-02 11:33:44'), '%I'); SELECT FROM_UNIXTIME(toDateTime('2018-01-02 00:33:44'), '%I'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-01 00:33:44'), '%j'); -SELECT FROM_UNIXTIME(toDateTime('2000-12-31 00:33:44'), '%j'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%m'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%M'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%n'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 00:33:44'), '%p'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-01 00:33:44'), '%j'), FROM_UNIXTIME(toDate32('2018-01-01'), '%j'); +SELECT FROM_UNIXTIME(toDateTime('2000-12-31 00:33:44'), '%j'), FROM_UNIXTIME(toDate32('2000-12-31'), '%j'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%m'), FROM_UNIXTIME(toDate32('2018-01-02'), '%m'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%M'), FROM_UNIXTIME(toDate32('2018-01-02'), '%M'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%n'), FROM_UNIXTIME(toDate32('2018-01-02'), '%n'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 00:33:44'), '%p'), FROM_UNIXTIME(toDate32('2018-01-02'), '%p'); SELECT FROM_UNIXTIME(toDateTime('2018-01-02 11:33:44'), '%p'); SELECT FROM_UNIXTIME(toDateTime('2018-01-02 12:33:44'), '%p'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%R'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%S'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%t'); -SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%T'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%R'), FROM_UNIXTIME(toDate32('2018-01-02'), '%R'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%S'), FROM_UNIXTIME(toDate32('2018-01-02'), '%S'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%t'), FROM_UNIXTIME(toDate32('2018-01-02'), '%t'); +SELECT FROM_UNIXTIME(toDateTime('2018-01-02 22:33:44'), '%T'), FROM_UNIXTIME(toDate32('2018-01-02'), '%T'); diff --git a/tests/queries/0_stateless/01601_custom_tld.reference b/tests/queries/0_stateless/01601_custom_tld.reference index 981067606a2..7ef6eb7d3a2 100644 --- a/tests/queries/0_stateless/01601_custom_tld.reference +++ b/tests/queries/0_stateless/01601_custom_tld.reference @@ -89,3 +89,92 @@ select cutToFirstSignificantSubdomainCustom('city.kawasaki.jp', 'public_suffix_l city.kawasaki.jp select cutToFirstSignificantSubdomainCustom('some.city.kawasaki.jp', 'public_suffix_list'); city.kawasaki.jp +select '-- no-tld'; +-- no-tld +-- even if there is no TLD, 2-nd level by default anyway +-- FIXME: make this behavior optional (so that TLD for host never changed, either empty or something real) +select cutToFirstSignificantSubdomainRFC('there-is-no-such-domain'); + +select cutToFirstSignificantSubdomainRFC('foo.there-is-no-such-domain'); +foo.there-is-no-such-domain +select cutToFirstSignificantSubdomainRFC('bar.foo.there-is-no-such-domain'); +foo.there-is-no-such-domain +select cutToFirstSignificantSubdomainCustomRFC('there-is-no-such-domain', 'public_suffix_list'); + +select cutToFirstSignificantSubdomainCustomRFC('foo.there-is-no-such-domain', 'public_suffix_list'); +foo.there-is-no-such-domain +select cutToFirstSignificantSubdomainCustomRFC('bar.foo.there-is-no-such-domain', 'public_suffix_list'); +foo.there-is-no-such-domain +select firstSignificantSubdomainCustomRFC('bar.foo.there-is-no-such-domain', 'public_suffix_list'); +foo +select '-- generic'; +-- generic +select firstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel +kernel +select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss +kernel.biz.ss +select '-- difference'; +-- difference +-- biz.ss is not in the default TLD list, hence: +select cutToFirstSignificantSubdomainRFC('foo.kernel.biz.ss'); -- biz.ss +biz.ss +select cutToFirstSignificantSubdomainCustomRFC('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss +kernel.biz.ss +select '-- 3+level'; +-- 3+level +select cutToFirstSignificantSubdomainCustomRFC('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at +xx.blogspot.co.at +select firstSignificantSubdomainCustomRFC('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot +blogspot +select cutToFirstSignificantSubdomainCustomRFC('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at +xx.blogspot.co.at +select firstSignificantSubdomainCustomRFC('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot +blogspot +select '-- url'; +-- url +select cutToFirstSignificantSubdomainCustomRFC('http://foobar.com', 'public_suffix_list'); +foobar.com +select cutToFirstSignificantSubdomainCustomRFC('http://foobar.com/foo', 'public_suffix_list'); +foobar.com +select cutToFirstSignificantSubdomainCustomRFC('http://bar.foobar.com/foo', 'public_suffix_list'); +foobar.com +select cutToFirstSignificantSubdomainCustomRFC('http://xx.blogspot.co.at', 'public_suffix_list'); +xx.blogspot.co.at +select '-- www'; +-- www +select cutToFirstSignificantSubdomainCustomWithWWWRFC('http://www.foo', 'public_suffix_list'); +www.foo +select cutToFirstSignificantSubdomainCustomRFC('http://www.foo', 'public_suffix_list'); +foo +select '-- vector'; +-- vector +select cutToFirstSignificantSubdomainCustomRFC('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1); +xx.blogspot.co.at +select cutToFirstSignificantSubdomainCustomRFC('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1); + +select '-- no new line'; +-- no new line +select cutToFirstSignificantSubdomainCustomRFC('foo.bar', 'no_new_line_list'); +foo.bar +select cutToFirstSignificantSubdomainCustomRFC('a.foo.bar', 'no_new_line_list'); +a.foo.bar +select cutToFirstSignificantSubdomainCustomRFC('a.foo.baz', 'no_new_line_list'); +foo.baz +select '-- asterisk'; +-- asterisk +select cutToFirstSignificantSubdomainCustomRFC('foo.something.sheffield.sch.uk', 'public_suffix_list'); +something.sheffield.sch.uk +select cutToFirstSignificantSubdomainCustomRFC('something.sheffield.sch.uk', 'public_suffix_list'); +something.sheffield.sch.uk +select cutToFirstSignificantSubdomainCustomRFC('sheffield.sch.uk', 'public_suffix_list'); +sheffield.sch.uk +select '-- exclamation mark'; +-- exclamation mark +select cutToFirstSignificantSubdomainCustomRFC('foo.kawasaki.jp', 'public_suffix_list'); +foo.kawasaki.jp +select cutToFirstSignificantSubdomainCustomRFC('foo.foo.kawasaki.jp', 'public_suffix_list'); +foo.foo.kawasaki.jp +select cutToFirstSignificantSubdomainCustomRFC('city.kawasaki.jp', 'public_suffix_list'); +city.kawasaki.jp +select cutToFirstSignificantSubdomainCustomRFC('some.city.kawasaki.jp', 'public_suffix_list'); +city.kawasaki.jp diff --git a/tests/queries/0_stateless/01601_custom_tld.sql b/tests/queries/0_stateless/01601_custom_tld.sql deleted file mode 100644 index 69ae209af2c..00000000000 --- a/tests/queries/0_stateless/01601_custom_tld.sql +++ /dev/null @@ -1,57 +0,0 @@ --- { echo } - -select '-- no-tld'; --- even if there is no TLD, 2-nd level by default anyway --- FIXME: make this behavior optional (so that TLD for host never changed, either empty or something real) -select cutToFirstSignificantSubdomain('there-is-no-such-domain'); -select cutToFirstSignificantSubdomain('foo.there-is-no-such-domain'); -select cutToFirstSignificantSubdomain('bar.foo.there-is-no-such-domain'); -select cutToFirstSignificantSubdomainCustom('there-is-no-such-domain', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('foo.there-is-no-such-domain', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list'); -select firstSignificantSubdomainCustom('bar.foo.there-is-no-such-domain', 'public_suffix_list'); - -select '-- generic'; -select firstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel -select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss - -select '-- difference'; --- biz.ss is not in the default TLD list, hence: -select cutToFirstSignificantSubdomain('foo.kernel.biz.ss'); -- biz.ss -select cutToFirstSignificantSubdomainCustom('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss - -select '-- 3+level'; -select cutToFirstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at -select firstSignificantSubdomainCustom('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot -select cutToFirstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at -select firstSignificantSubdomainCustom('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot - -select '-- url'; -select cutToFirstSignificantSubdomainCustom('http://foobar.com', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('http://foobar.com/foo', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('http://bar.foobar.com/foo', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at', 'public_suffix_list'); - -select '-- www'; -select cutToFirstSignificantSubdomainCustomWithWWW('http://www.foo', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('http://www.foo', 'public_suffix_list'); - -select '-- vector'; -select cutToFirstSignificantSubdomainCustom('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1); -select cutToFirstSignificantSubdomainCustom('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1); - -select '-- no new line'; -select cutToFirstSignificantSubdomainCustom('foo.bar', 'no_new_line_list'); -select cutToFirstSignificantSubdomainCustom('a.foo.bar', 'no_new_line_list'); -select cutToFirstSignificantSubdomainCustom('a.foo.baz', 'no_new_line_list'); - -select '-- asterisk'; -select cutToFirstSignificantSubdomainCustom('foo.something.sheffield.sch.uk', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('something.sheffield.sch.uk', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('sheffield.sch.uk', 'public_suffix_list'); - -select '-- exclamation mark'; -select cutToFirstSignificantSubdomainCustom('foo.kawasaki.jp', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('foo.foo.kawasaki.jp', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('city.kawasaki.jp', 'public_suffix_list'); -select cutToFirstSignificantSubdomainCustom('some.city.kawasaki.jp', 'public_suffix_list'); diff --git a/tests/queries/0_stateless/01601_custom_tld.sql.j2 b/tests/queries/0_stateless/01601_custom_tld.sql.j2 new file mode 100644 index 00000000000..1e0982ea1b7 --- /dev/null +++ b/tests/queries/0_stateless/01601_custom_tld.sql.j2 @@ -0,0 +1,61 @@ +-- { echo } + +{% for suffix in ['', 'RFC'] -%} + +select '-- no-tld'; +-- even if there is no TLD, 2-nd level by default anyway +-- FIXME: make this behavior optional (so that TLD for host never changed, either empty or something real) +select cutToFirstSignificantSubdomain{{ suffix }}('there-is-no-such-domain'); +select cutToFirstSignificantSubdomain{{ suffix }}('foo.there-is-no-such-domain'); +select cutToFirstSignificantSubdomain{{ suffix }}('bar.foo.there-is-no-such-domain'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('there-is-no-such-domain', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.there-is-no-such-domain', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('bar.foo.there-is-no-such-domain', 'public_suffix_list'); +select firstSignificantSubdomainCustom{{ suffix }}('bar.foo.there-is-no-such-domain', 'public_suffix_list'); + +select '-- generic'; +select firstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss + +select '-- difference'; +-- biz.ss is not in the default TLD list, hence: +select cutToFirstSignificantSubdomain{{ suffix }}('foo.kernel.biz.ss'); -- biz.ss +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kernel.biz.ss', 'public_suffix_list'); -- kernel.biz.ss + +select '-- 3+level'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at +select firstSignificantSubdomainCustom{{ suffix }}('xx.blogspot.co.at', 'public_suffix_list'); -- blogspot +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- xx.blogspot.co.at +select firstSignificantSubdomainCustom{{ suffix }}('foo.bar.xx.blogspot.co.at', 'public_suffix_list'); -- blogspot + +select '-- url'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://foobar.com', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://foobar.com/foo', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://bar.foobar.com/foo', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://xx.blogspot.co.at', 'public_suffix_list'); + +select '-- www'; +select cutToFirstSignificantSubdomainCustomWithWWW{{ suffix }}('http://www.foo', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://www.foo', 'public_suffix_list'); + +select '-- vector'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('http://xx.blogspot.co.at/' || toString(number), 'public_suffix_list') from numbers(1); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('there-is-no-such-domain' || toString(number), 'public_suffix_list') from numbers(1); + +select '-- no new line'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.bar', 'no_new_line_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('a.foo.bar', 'no_new_line_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('a.foo.baz', 'no_new_line_list'); + +select '-- asterisk'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.something.sheffield.sch.uk', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('something.sheffield.sch.uk', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('sheffield.sch.uk', 'public_suffix_list'); + +select '-- exclamation mark'; +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.kawasaki.jp', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('foo.foo.kawasaki.jp', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('city.kawasaki.jp', 'public_suffix_list'); +select cutToFirstSignificantSubdomainCustom{{ suffix }}('some.city.kawasaki.jp', 'public_suffix_list'); + +{% endfor %} diff --git a/tests/queries/0_stateless/01685_json_extract_double_as_float.reference b/tests/queries/0_stateless/01685_json_extract_double_as_float.reference index f3f4206b425..a24f6569f44 100644 --- a/tests/queries/0_stateless/01685_json_extract_double_as_float.reference +++ b/tests/queries/0_stateless/01685_json_extract_double_as_float.reference @@ -1,7 +1,7 @@ 1.1 1.1 1.1 1.1 0.01 0.01 0.01 0.01 -0 -\N +1 +1 -1e300 -inf 0 diff --git a/tests/queries/0_stateless/01744_fuse_sum_count_aggregate.sql b/tests/queries/0_stateless/01744_fuse_sum_count_aggregate.sql index 5b6ed440ba4..375662eb405 100644 --- a/tests/queries/0_stateless/01744_fuse_sum_count_aggregate.sql +++ b/tests/queries/0_stateless/01744_fuse_sum_count_aggregate.sql @@ -10,4 +10,5 @@ EXPLAIN SYNTAX SELECT sum(a), sum(b), count(b) from fuse_tbl; SELECT '---------NOT trigger fuse--------'; SELECT sum(a), avg(b) from fuse_tbl; EXPLAIN SYNTAX SELECT sum(a), avg(b) from fuse_tbl; + DROP TABLE fuse_tbl; diff --git a/tests/queries/0_stateless/01825_type_json_in_array.reference b/tests/queries/0_stateless/01825_type_json_in_array.reference new file mode 100644 index 00000000000..c36a22e6951 --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_in_array.reference @@ -0,0 +1,23 @@ +{"id":1,"arr":[{"k1":1,"k2":{"k3":2,"k4":3,"k5":""}},{"k1":2,"k2":{"k3":0,"k4":0,"k5":"foo"}}]} +{"id":2,"arr":[{"k1":3,"k2":{"k3":4,"k4":5,"k5":""}}]} +1 [1,2] [2,0] [3,0] ['','foo'] +2 [3] [4] [5] [''] +{"arr":{"k1":1,"k2":{"k3":2,"k4":3,"k5":""}}} +{"arr":{"k1":2,"k2":{"k3":0,"k4":0,"k5":"foo"}}} +{"arr":{"k1":3,"k2":{"k3":4,"k4":5,"k5":""}}} +Array(Tuple(k1 Int8, k2 Tuple(k3 Int8, k4 Int8, k5 String))) +{"id":1,"arr":[{"k1":[{"k2":"aaa","k3":"bbb","k4":0},{"k2":"ccc","k3":"","k4":0}],"k5":{"k6":""}}]} +{"id":2,"arr":[{"k1":[{"k2":"","k3":"ddd","k4":10},{"k2":"","k3":"","k4":20}],"k5":{"k6":"foo"}}]} +1 [['aaa','ccc']] [['bbb','']] [[0,0]] [''] +2 [['','']] [['ddd','']] [[10,20]] ['foo'] +{"k1":{"k2":"","k3":"","k4":20}} +{"k1":{"k2":"","k3":"ddd","k4":10}} +{"k1":{"k2":"aaa","k3":"bbb","k4":0}} +{"k1":{"k2":"ccc","k3":"","k4":0}} +Tuple(k2 String, k3 String, k4 Int8) +{"arr":[{"x":1}]} +{"arr":{"x":{"y":1},"t":{"y":2}}} +{"arr":[1,{"y":1}]} +{"arr":[2,{"y":2}]} +{"arr":[{"x":"aaa","y":[1,2,3]}]} +{"arr":[{"x":1}]} diff --git a/tests/queries/0_stateless/01825_type_json_in_array.sql b/tests/queries/0_stateless/01825_type_json_in_array.sql new file mode 100644 index 00000000000..e5c20d7ba6b --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_in_array.sql @@ -0,0 +1,35 @@ +-- Tags: no-fasttest + +SET allow_experimental_object_type = 1; +DROP TABLE IF EXISTS t_json_array; + +CREATE TABLE t_json_array (id UInt32, arr Array(JSON)) ENGINE = MergeTree ORDER BY id; + +INSERT INTO t_json_array FORMAT JSONEachRow {"id": 1, "arr": [{"k1": 1, "k2": {"k3": 2, "k4": 3}}, {"k1": 2, "k2": {"k5": "foo"}}]} +INSERT INTO t_json_array FORMAT JSONEachRow {"id": 2, "arr": [{"k1": 3, "k2": {"k3": 4, "k4": 5}}]} + +SET output_format_json_named_tuples_as_objects = 1; + +SELECT * FROM t_json_array ORDER BY id FORMAT JSONEachRow; +SELECT id, arr.k1, arr.k2.k3, arr.k2.k4, arr.k2.k5 FROM t_json_array ORDER BY id; +SELECT arr FROM t_json_array ARRAY JOIN arr ORDER BY arr.k1 FORMAT JSONEachRow; +SELECT toTypeName(arr) FROM t_json_array LIMIT 1; + +TRUNCATE TABLE t_json_array; + +INSERT INTO t_json_array FORMAT JSONEachRow {"id": 1, "arr": [{"k1": [{"k2": "aaa", "k3": "bbb"}, {"k2": "ccc"}]}]} +INSERT INTO t_json_array FORMAT JSONEachRow {"id": 2, "arr": [{"k1": [{"k3": "ddd", "k4": 10}, {"k4": 20}], "k5": {"k6": "foo"}}]} + +SELECT * FROM t_json_array ORDER BY id FORMAT JSONEachRow; +SELECT id, arr.k1.k2, arr.k1.k3, arr.k1.k4, arr.k5.k6 FROM t_json_array ORDER BY id; + +SELECT arrayJoin(arrayJoin(arr.k1)) AS k1 FROM t_json_array ORDER BY k1 FORMAT JSONEachRow; +SELECT toTypeName(arrayJoin(arrayJoin(arr.k1))) AS arr FROM t_json_array LIMIT 1; + +DROP TABLE t_json_array; + +SELECT * FROM values('arr Array(JSON)', '[\'{"x" : 1}\']') FORMAT JSONEachRow; +SELECT * FROM values('arr Map(String, JSON)', '{\'x\' : \'{"y" : 1}\', \'t\' : \'{"y" : 2}\'}') FORMAT JSONEachRow; +SELECT * FROM values('arr Tuple(Int32, JSON)', '(1, \'{"y" : 1}\')', '(2, \'{"y" : 2}\')') FORMAT JSONEachRow; +SELECT * FROM format(JSONEachRow, '{"arr" : [{"x" : "aaa", "y" : [1,2,3]}]}') FORMAT JSONEachRow; +SELECT * FROM values('arr Array(JSON)', '[\'{"x" : 1}\']') FORMAT JSONEachRow; diff --git a/tests/queries/0_stateless/01825_type_json_in_other_types.reference b/tests/queries/0_stateless/01825_type_json_in_other_types.reference new file mode 100644 index 00000000000..b94885a65ab --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_in_other_types.reference @@ -0,0 +1,17 @@ +Tuple(String, Map(String, Array(Tuple(k1 Nested(k2 Int8, k3 Int8, k5 String), k4 String))), Tuple(k1 String, k2 Tuple(k3 String, k4 String))) +============= +{"id":1,"data":["foo",{"aa":[{"k1":[{"k2":1,"k3":2,"k5":""},{"k2":0,"k3":3,"k5":""}],"k4":""},{"k1":[{"k2":4,"k3":0,"k5":""},{"k2":0,"k3":5,"k5":""},{"k2":6,"k3":0,"k5":""}],"k4":"qqq"}],"bb":[{"k1":[],"k4":"www"},{"k1":[{"k2":7,"k3":8,"k5":""},{"k2":9,"k3":10,"k5":""},{"k2":11,"k3":12,"k5":""}],"k4":""}]},{"k1":"aa","k2":{"k3":"bb","k4":"c"}}]} +{"id":2,"data":["bar",{"aa":[{"k1":[{"k2":13,"k3":14,"k5":""},{"k2":15,"k3":16,"k5":""}],"k4":"www"}]},{"k1":"","k2":{"k3":"","k4":""}}]} +{"id":3,"data":["some",{"aa":[{"k1":[{"k2":0,"k3":20,"k5":"some"}],"k4":""}]},{"k1":"eee","k2":{"k3":"","k4":""}}]} +============= +{"aa":[{"k1":[{"k2":1,"k3":2,"k5":""},{"k2":0,"k3":3,"k5":""}],"k4":""},{"k1":[{"k2":4,"k3":0,"k5":""},{"k2":0,"k3":5,"k5":""},{"k2":6,"k3":0,"k5":""}],"k4":"qqq"}],"bb":[{"k1":[],"k4":"www"},{"k1":[{"k2":7,"k3":8,"k5":""},{"k2":9,"k3":10,"k5":""},{"k2":11,"k3":12,"k5":""}],"k4":""}]} +{"aa":[{"k1":[{"k2":13,"k3":14,"k5":""},{"k2":15,"k3":16,"k5":""}],"k4":"www"}],"bb":[]} +{"aa":[{"k1":[{"k2":0,"k3":20,"k5":"some"}],"k4":""}],"bb":[]} +============= +{"k1":[[{"k2":1,"k3":2,"k5":""},{"k2":0,"k3":3,"k5":""}],[{"k2":4,"k3":0,"k5":""},{"k2":0,"k3":5,"k5":""},{"k2":6,"k3":0,"k5":""}]],"k4":["","qqq"]} +{"k1":[[{"k2":13,"k3":14,"k5":""},{"k2":15,"k3":16,"k5":""}]],"k4":["www"]} +{"k1":[[{"k2":0,"k3":20,"k5":"some"}]],"k4":[""]} +============= +{"obj":{"k1":"aa","k2":{"k3":"bb","k4":"c"}}} +{"obj":{"k1":"","k2":{"k3":"","k4":""}}} +{"obj":{"k1":"eee","k2":{"k3":"","k4":""}}} diff --git a/tests/queries/0_stateless/01825_type_json_in_other_types.sh b/tests/queries/0_stateless/01825_type_json_in_other_types.sh new file mode 100755 index 00000000000..e9cf0bcaca1 --- /dev/null +++ b/tests/queries/0_stateless/01825_type_json_in_other_types.sh @@ -0,0 +1,91 @@ +#!/usr/bin/env bash +# Tags: no-fasttest + +CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd) +# shellcheck source=../shell_config.sh +. "$CUR_DIR"/../shell_config.sh + +${CLICKHOUSE_CLIENT} -q "SET allow_experimental_object_type = 1" +${CLICKHOUSE_CLIENT} -q "DROP TABLE IF EXISTS t_json_nested" + +${CLICKHOUSE_CLIENT} -q " + CREATE TABLE t_json_nested + ( + id UInt32, + data Tuple(String, Map(String, Array(JSON)), JSON) + ) + ENGINE = MergeTree ORDER BY id" --allow_experimental_object_type 1 + +cat < x.1, cg ) != ['y', 'n'], 'y', 'n') as so, + arrayFilter( x -> x.1 = so , cg) as r + ); + +select + r +from + (select [('y',0),('n',1)] as cg, + if( arrayMap( x -> x.1, cg ) != ['y', 'n'], 'y', 'n') as so, + arrayFilter( x -> x.1 = so , cg) as r + ); diff --git a/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.reference b/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.reference new file mode 100644 index 00000000000..bec52aa46b6 --- /dev/null +++ b/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.reference @@ -0,0 +1,2 @@ +2.1999999999999997 289.99999999999994 [1,2,3,4] [0.1,0.2,0.1,0.2] +2.1999999999999997 289.99999999999994 [1,2,3,4] [0.1,0.2,0.1,0.2] diff --git a/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.sql b/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.sql new file mode 100644 index 00000000000..b07f3f33ac3 --- /dev/null +++ b/tests/queries/0_stateless/02343_analyzer_lambdas_issue_36677.sql @@ -0,0 +1,14 @@ +SET allow_experimental_analyzer = 1; + +SELECT + arraySum(x -> ((x.1) / ((x.2) * (x.2))), arrayZip(mag, magerr)) / arraySum(x -> (1. / (x * x)), magerr) AS weightedmeanmag, + arraySum(x -> ((((x.1) - weightedmeanmag) * ((x.1) - weightedmeanmag)) / ((x.2) * (x.2))), arrayZip(mag, magerr)) AS chi2, + [1, 2, 3, 4] AS mag, + [0.1, 0.2, 0.1, 0.2] AS magerr; + +SELECT + arraySum(x -> ((x.1) / ((x.2) * (x.2))), arrayZip(mag, magerr)) / arraySum(x -> (1. / (x * x)), magerr) AS weightedmeanmag, + arraySum(x -> ((((x.1) - weightedmeanmag) * ((x.1) - weightedmeanmag)) / ((x.2) * (x.2))), arrayZip(mag, magerr)) AS chi2, + [1, 2, 3, 4] AS mag, + [0.1, 0.2, 0.1, 0.2] AS magerr +WHERE isFinite(chi2) diff --git a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference index 3fd12051f4a..040a8c8d317 100644 --- a/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference +++ b/tests/queries/0_stateless/02415_all_new_functions_must_be_documented.reference @@ -219,14 +219,6 @@ cutFragment cutIPv6 cutQueryString cutQueryStringAndFragment -cutToFirstSignificantSubdomain -cutToFirstSignificantSubdomainCustom -cutToFirstSignificantSubdomainCustomRFC -cutToFirstSignificantSubdomainCustomWithWWW -cutToFirstSignificantSubdomainCustomWithWWWRFC -cutToFirstSignificantSubdomainRFC -cutToFirstSignificantSubdomainWithWWW -cutToFirstSignificantSubdomainWithWWWRFC cutURLParameter cutWWW dateDiff @@ -284,10 +276,6 @@ dictGetUUIDOrDefault dictHas dictIsIn divide -domain -domainRFC -domainWithoutWWW -domainWithoutWWWRFC dotProduct dumpColumnStructure e @@ -336,10 +324,8 @@ filesystemAvailable filesystemCapacity filesystemFree finalizeAggregation -firstSignificantSubdomain firstSignificantSubdomainCustom firstSignificantSubdomainCustomRFC -firstSignificantSubdomainRFC flattenTuple floor format @@ -600,8 +586,6 @@ polygonsUnionCartesian polygonsUnionSpherical polygonsWithinCartesian polygonsWithinSpherical -port -portRFC position positionCaseInsensitive positionCaseInsensitiveUTF8 @@ -906,8 +890,6 @@ toYear toYearWeek today tokens -topLevelDomain -topLevelDomainRFC transactionID transactionLatestSnapshot transactionOldestSnapshot diff --git a/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.reference b/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.reference new file mode 100644 index 00000000000..13b65c29f05 --- /dev/null +++ b/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.reference @@ -0,0 +1 @@ +printer1 diff --git a/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.sql b/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.sql new file mode 100644 index 00000000000..690ec6c70e0 --- /dev/null +++ b/tests/queries/0_stateless/02457_key_condition_with_types_that_cannot_be_nullable.sql @@ -0,0 +1,9 @@ +drop table if exists test; + +create table test (Printer LowCardinality(String), IntervalStart DateTime) engine MergeTree partition by (hiveHash(Printer), toYear(IntervalStart)) order by (Printer, IntervalStart); + +insert into test values ('printer1', '2006-02-07 06:28:15'); + +select Printer from test where Printer='printer1'; + +drop table test; diff --git a/tests/queries/0_stateless/02457_s3_cluster_schema_inference.reference b/tests/queries/0_stateless/02457_s3_cluster_schema_inference.reference new file mode 100644 index 00000000000..b918bf2b155 --- /dev/null +++ b/tests/queries/0_stateless/02457_s3_cluster_schema_inference.reference @@ -0,0 +1,44 @@ +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +0 0 0 +0 0 0 +1 2 3 +4 5 6 +7 8 9 +10 11 12 +13 14 15 +16 17 18 +0 0 0 +0 0 0 +1 2 3 +4 5 6 +7 8 9 +10 11 12 +13 14 15 +16 17 18 +0 0 0 +0 0 0 +1 2 3 +4 5 6 +7 8 9 +10 11 12 +13 14 15 +16 17 18 +0 0 0 +0 0 0 +1 2 3 +4 5 6 +7 8 9 +10 11 12 +13 14 15 +16 17 18 diff --git a/tests/queries/0_stateless/02457_s3_cluster_schema_inference.sql b/tests/queries/0_stateless/02457_s3_cluster_schema_inference.sql new file mode 100644 index 00000000000..03e8785b24b --- /dev/null +++ b/tests/queries/0_stateless/02457_s3_cluster_schema_inference.sql @@ -0,0 +1,13 @@ +-- Tags: no-fasttest +-- Tag no-fasttest: Depends on AWS + +desc s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv'); +desc s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'TSV'); +desc s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'test', 'testtest'); +desc s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'test', 'testtest', 'TSV'); + +select * from s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv') order by c1, c2, c3; +select * from s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'TSV') order by c1, c2, c3; +select * from s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'test', 'testtest') order by c1, c2, c3; +select * from s3Cluster('test_cluster_one_shard_three_replicas_localhost', 'http://localhost:11111/test/{a,b}.tsv', 'test', 'testtest', 'TSV') order by c1, c2, c3; + diff --git a/tests/queries/0_stateless/02458_datediff_date32.reference b/tests/queries/0_stateless/02458_datediff_date32.reference index 9e93af03896..fdb8273a74b 100644 --- a/tests/queries/0_stateless/02458_datediff_date32.reference +++ b/tests/queries/0_stateless/02458_datediff_date32.reference @@ -1,169 +1,169 @@ -- { echo } -- Date32 vs Date32 -SELECT dateDiff('second', toDate32('1927-01-01'), toDate32('1927-01-02')); +SELECT dateDiff('second', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDate32('1927-01-01'), toDate32('1927-01-02')); +SELECT dateDiff('minute', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDate32('1927-01-01'), toDate32('1927-01-02')); +SELECT dateDiff('hour', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDate32('1927-01-01'), toDate32('1927-01-02')); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDate32('1927-01-01'), toDate32('1927-01-08')); +SELECT dateDiff('week', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-08', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDate32('1927-01-01'), toDate32('1927-02-01')); +SELECT dateDiff('month', toDate32('1927-01-01', 'UTC'), toDate32('1927-02-01', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDate32('1927-01-01'), toDate32('1927-04-01')); +SELECT dateDiff('quarter', toDate32('1927-01-01', 'UTC'), toDate32('1927-04-01', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDate32('1927-01-01'), toDate32('1928-01-01')); +SELECT dateDiff('year', toDate32('1927-01-01', 'UTC'), toDate32('1928-01-01', 'UTC'), 'UTC'); 1 -- With DateTime64 -- Date32 vs DateTime64 -SELECT dateDiff('second', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); +SELECT dateDiff('second', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); +SELECT dateDiff('minute', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); +SELECT dateDiff('hour', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDate32('1927-01-01'), toDateTime64('1927-01-08 00:00:00', 3)); +SELECT dateDiff('week', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-08 00:00:00', 3, 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDate32('1927-01-01'), toDateTime64('1927-02-01 00:00:00', 3)); +SELECT dateDiff('month', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-02-01 00:00:00', 3, 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDate32('1927-01-01'), toDateTime64('1927-04-01 00:00:00', 3)); +SELECT dateDiff('quarter', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-04-01 00:00:00', 3, 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDate32('1927-01-01'), toDateTime64('1928-01-01 00:00:00', 3)); +SELECT dateDiff('year', toDate32('1927-01-01', 'UTC'), toDateTime64('1928-01-01 00:00:00', 3, 'UTC'), 'UTC'); 1 -- DateTime64 vs Date32 -SELECT dateDiff('second', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); +SELECT dateDiff('second', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); +SELECT dateDiff('minute', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); +SELECT dateDiff('hour', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); +SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-08')); +SELECT dateDiff('week', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-08', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-02-01')); +SELECT dateDiff('month', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-02-01', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-04-01')); +SELECT dateDiff('quarter', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-04-01', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1928-01-01')); +SELECT dateDiff('year', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1928-01-01', 'UTC'), 'UTC'); 1 -- With DateTime -- Date32 vs DateTime -SELECT dateDiff('second', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); +SELECT dateDiff('second', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); +SELECT dateDiff('minute', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); +SELECT dateDiff('hour', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDate32('2015-08-18'), toDateTime('2015-08-25 00:00:00')); +SELECT dateDiff('week', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-25 00:00:00', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDate32('2015-08-18'), toDateTime('2015-09-18 00:00:00')); +SELECT dateDiff('month', toDate32('2015-08-18', 'UTC'), toDateTime('2015-09-18 00:00:00', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDate32('2015-08-18'), toDateTime('2015-11-18 00:00:00')); +SELECT dateDiff('quarter', toDate32('2015-08-18', 'UTC'), toDateTime('2015-11-18 00:00:00', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDate32('2015-08-18'), toDateTime('2016-08-18 00:00:00')); +SELECT dateDiff('year', toDate32('2015-08-18', 'UTC'), toDateTime('2016-08-18 00:00:00', 'UTC'), 'UTC'); 1 -- DateTime vs Date32 -SELECT dateDiff('second', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); +SELECT dateDiff('second', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); +SELECT dateDiff('minute', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); +SELECT dateDiff('hour', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); +SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-25')); +SELECT dateDiff('week', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-25', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDateTime('2015-08-18 00:00:00'), toDate32('2015-09-18')); +SELECT dateDiff('month', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-09-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDateTime('2015-08-18 00:00:00'), toDate32('2015-11-18')); +SELECT dateDiff('quarter', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-11-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDateTime('2015-08-18 00:00:00'), toDate32('2016-08-18')); +SELECT dateDiff('year', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2016-08-18', 'UTC'), 'UTC'); 1 -- With Date -- Date32 vs Date -SELECT dateDiff('second', toDate32('2015-08-18'), toDate('2015-08-19')); +SELECT dateDiff('second', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDate32('2015-08-18'), toDate('2015-08-19')); +SELECT dateDiff('minute', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDate32('2015-08-18'), toDate('2015-08-19')); +SELECT dateDiff('hour', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDate32('2015-08-18'), toDate('2015-08-19')); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDate32('2015-08-18'), toDate('2015-08-25')); +SELECT dateDiff('week', toDate32('2015-08-18', 'UTC'), toDate('2015-08-25', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDate32('2015-08-18'), toDate('2015-09-18')); +SELECT dateDiff('month', toDate32('2015-08-18', 'UTC'), toDate('2015-09-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDate32('2015-08-18'), toDate('2015-11-18')); +SELECT dateDiff('quarter', toDate32('2015-08-18', 'UTC'), toDate('2015-11-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDate32('2015-08-18'), toDate('2016-08-18')); +SELECT dateDiff('year', toDate32('2015-08-18', 'UTC'), toDate('2016-08-18', 'UTC'), 'UTC'); 1 -- Date vs Date32 -SELECT dateDiff('second', toDate('2015-08-18'), toDate32('2015-08-19')); +SELECT dateDiff('second', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 86400 -SELECT dateDiff('minute', toDate('2015-08-18'), toDate32('2015-08-19')); +SELECT dateDiff('minute', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 1440 -SELECT dateDiff('hour', toDate('2015-08-18'), toDate32('2015-08-19')); +SELECT dateDiff('hour', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 24 -SELECT dateDiff('day', toDate('2015-08-18'), toDate32('2015-08-19')); +SELECT dateDiff('day', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); 1 -SELECT dateDiff('week', toDate('2015-08-18'), toDate32('2015-08-25')); +SELECT dateDiff('week', toDate('2015-08-18', 'UTC'), toDate32('2015-08-25', 'UTC'), 'UTC'); 1 -SELECT dateDiff('month', toDate('2015-08-18'), toDate32('2015-09-18')); +SELECT dateDiff('month', toDate('2015-08-18', 'UTC'), toDate32('2015-09-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('quarter', toDate('2015-08-18'), toDate32('2015-11-18')); +SELECT dateDiff('quarter', toDate('2015-08-18', 'UTC'), toDate32('2015-11-18', 'UTC'), 'UTC'); 1 -SELECT dateDiff('year', toDate('2015-08-18'), toDate32('2016-08-18')); +SELECT dateDiff('year', toDate('2015-08-18', 'UTC'), toDate32('2016-08-18', 'UTC'), 'UTC'); 1 -- Const vs non-const columns -SELECT dateDiff('day', toDate32('1927-01-01'), materialize(toDate32('1927-01-02'))); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDate32('1927-01-01'), materialize(toDateTime64('1927-01-02 00:00:00', 3))); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), materialize(toDateTime64('1927-01-02 00:00:00', 3, 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3), materialize(toDate32('1927-01-02'))); +SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDate32('2015-08-18'), materialize(toDateTime('2015-08-19 00:00:00'))); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), materialize(toDateTime('2015-08-19 00:00:00', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00'), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00', 'UTC'), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDate32('2015-08-18'), materialize(toDate('2015-08-19'))); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), materialize(toDate('2015-08-19', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', toDate('2015-08-18'), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', toDate('2015-08-18', 'UTC'), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); 1 -- Non-const vs const columns -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), toDate32('1927-01-02')); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), toDate32('1927-01-02', 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), toDateTime64('1927-01-02 00:00:00', 3)); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3)), toDate32('1927-01-02')); +SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3, 'UTC')), toDate32('1927-01-02', 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), toDateTime('2015-08-19 00:00:00')); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00')), toDate32('2015-08-19')); +SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00', 'UTC')), toDate32('2015-08-19', 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), toDate('2015-08-19')); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), toDate('2015-08-19', 'UTC'), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate('2015-08-18')), toDate32('2015-08-19')); +SELECT dateDiff('day', materialize(toDate('2015-08-18', 'UTC')), toDate32('2015-08-19', 'UTC'), 'UTC'); 1 -- Non-const vs non-const columns -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), materialize(toDate32('1927-01-02'))); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), materialize(toDateTime64('1927-01-02 00:00:00', 3))); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), materialize(toDateTime64('1927-01-02 00:00:00', 3, 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3)), materialize(toDate32('1927-01-02'))); +SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3, 'UTC')), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), materialize(toDateTime('2015-08-19 00:00:00'))); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), materialize(toDateTime('2015-08-19 00:00:00', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00')), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00', 'UTC')), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), materialize(toDate('2015-08-19'))); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), materialize(toDate('2015-08-19', 'UTC')), 'UTC'); 1 -SELECT dateDiff('day', materialize(toDate('2015-08-18')), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', materialize(toDate('2015-08-18', 'UTC')), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); 1 diff --git a/tests/queries/0_stateless/02458_datediff_date32.sql b/tests/queries/0_stateless/02458_datediff_date32.sql index b4cb203080e..e41070e8146 100644 --- a/tests/queries/0_stateless/02458_datediff_date32.sql +++ b/tests/queries/0_stateless/02458_datediff_date32.sql @@ -1,101 +1,101 @@ -- { echo } -- Date32 vs Date32 -SELECT dateDiff('second', toDate32('1927-01-01'), toDate32('1927-01-02')); -SELECT dateDiff('minute', toDate32('1927-01-01'), toDate32('1927-01-02')); -SELECT dateDiff('hour', toDate32('1927-01-01'), toDate32('1927-01-02')); -SELECT dateDiff('day', toDate32('1927-01-01'), toDate32('1927-01-02')); -SELECT dateDiff('week', toDate32('1927-01-01'), toDate32('1927-01-08')); -SELECT dateDiff('month', toDate32('1927-01-01'), toDate32('1927-02-01')); -SELECT dateDiff('quarter', toDate32('1927-01-01'), toDate32('1927-04-01')); -SELECT dateDiff('year', toDate32('1927-01-01'), toDate32('1928-01-01')); +SELECT dateDiff('second', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDate32('1927-01-01', 'UTC'), toDate32('1927-01-08', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDate32('1927-01-01', 'UTC'), toDate32('1927-02-01', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDate32('1927-01-01', 'UTC'), toDate32('1927-04-01', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDate32('1927-01-01', 'UTC'), toDate32('1928-01-01', 'UTC'), 'UTC'); -- With DateTime64 -- Date32 vs DateTime64 -SELECT dateDiff('second', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); -SELECT dateDiff('minute', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); -SELECT dateDiff('hour', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); -SELECT dateDiff('day', toDate32('1927-01-01'), toDateTime64('1927-01-02 00:00:00', 3)); -SELECT dateDiff('week', toDate32('1927-01-01'), toDateTime64('1927-01-08 00:00:00', 3)); -SELECT dateDiff('month', toDate32('1927-01-01'), toDateTime64('1927-02-01 00:00:00', 3)); -SELECT dateDiff('quarter', toDate32('1927-01-01'), toDateTime64('1927-04-01 00:00:00', 3)); -SELECT dateDiff('year', toDate32('1927-01-01'), toDateTime64('1928-01-01 00:00:00', 3)); +SELECT dateDiff('second', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('week', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-01-08 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('month', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-02-01 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDate32('1927-01-01', 'UTC'), toDateTime64('1927-04-01 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('year', toDate32('1927-01-01', 'UTC'), toDateTime64('1928-01-01 00:00:00', 3, 'UTC'), 'UTC'); -- DateTime64 vs Date32 -SELECT dateDiff('second', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); -SELECT dateDiff('minute', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); -SELECT dateDiff('hour', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); -SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-02')); -SELECT dateDiff('week', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-01-08')); -SELECT dateDiff('month', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-02-01')); -SELECT dateDiff('quarter', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1927-04-01')); -SELECT dateDiff('year', toDateTime64('1927-01-01 00:00:00', 3), toDate32('1928-01-01')); +SELECT dateDiff('second', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-01-08', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-02-01', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1927-04-01', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), toDate32('1928-01-01', 'UTC'), 'UTC'); -- With DateTime -- Date32 vs DateTime -SELECT dateDiff('second', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); -SELECT dateDiff('minute', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); -SELECT dateDiff('hour', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); -SELECT dateDiff('day', toDate32('2015-08-18'), toDateTime('2015-08-19 00:00:00')); -SELECT dateDiff('week', toDate32('2015-08-18'), toDateTime('2015-08-25 00:00:00')); -SELECT dateDiff('month', toDate32('2015-08-18'), toDateTime('2015-09-18 00:00:00')); -SELECT dateDiff('quarter', toDate32('2015-08-18'), toDateTime('2015-11-18 00:00:00')); -SELECT dateDiff('year', toDate32('2015-08-18'), toDateTime('2016-08-18 00:00:00')); +SELECT dateDiff('second', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDate32('2015-08-18', 'UTC'), toDateTime('2015-08-25 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDate32('2015-08-18', 'UTC'), toDateTime('2015-09-18 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDate32('2015-08-18', 'UTC'), toDateTime('2015-11-18 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDate32('2015-08-18', 'UTC'), toDateTime('2016-08-18 00:00:00', 'UTC'), 'UTC'); -- DateTime vs Date32 -SELECT dateDiff('second', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); -SELECT dateDiff('minute', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); -SELECT dateDiff('hour', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); -SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-19')); -SELECT dateDiff('week', toDateTime('2015-08-18 00:00:00'), toDate32('2015-08-25')); -SELECT dateDiff('month', toDateTime('2015-08-18 00:00:00'), toDate32('2015-09-18')); -SELECT dateDiff('quarter', toDateTime('2015-08-18 00:00:00'), toDate32('2015-11-18')); -SELECT dateDiff('year', toDateTime('2015-08-18 00:00:00'), toDate32('2016-08-18')); +SELECT dateDiff('second', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-08-25', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-09-18', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2015-11-18', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDateTime('2015-08-18 00:00:00', 'UTC'), toDate32('2016-08-18', 'UTC'), 'UTC'); -- With Date -- Date32 vs Date -SELECT dateDiff('second', toDate32('2015-08-18'), toDate('2015-08-19')); -SELECT dateDiff('minute', toDate32('2015-08-18'), toDate('2015-08-19')); -SELECT dateDiff('hour', toDate32('2015-08-18'), toDate('2015-08-19')); -SELECT dateDiff('day', toDate32('2015-08-18'), toDate('2015-08-19')); -SELECT dateDiff('week', toDate32('2015-08-18'), toDate('2015-08-25')); -SELECT dateDiff('month', toDate32('2015-08-18'), toDate('2015-09-18')); -SELECT dateDiff('quarter', toDate32('2015-08-18'), toDate('2015-11-18')); -SELECT dateDiff('year', toDate32('2015-08-18'), toDate('2016-08-18')); +SELECT dateDiff('second', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), toDate('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDate32('2015-08-18', 'UTC'), toDate('2015-08-25', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDate32('2015-08-18', 'UTC'), toDate('2015-09-18', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDate32('2015-08-18', 'UTC'), toDate('2015-11-18', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDate32('2015-08-18', 'UTC'), toDate('2016-08-18', 'UTC'), 'UTC'); -- Date vs Date32 -SELECT dateDiff('second', toDate('2015-08-18'), toDate32('2015-08-19')); -SELECT dateDiff('minute', toDate('2015-08-18'), toDate32('2015-08-19')); -SELECT dateDiff('hour', toDate('2015-08-18'), toDate32('2015-08-19')); -SELECT dateDiff('day', toDate('2015-08-18'), toDate32('2015-08-19')); -SELECT dateDiff('week', toDate('2015-08-18'), toDate32('2015-08-25')); -SELECT dateDiff('month', toDate('2015-08-18'), toDate32('2015-09-18')); -SELECT dateDiff('quarter', toDate('2015-08-18'), toDate32('2015-11-18')); -SELECT dateDiff('year', toDate('2015-08-18'), toDate32('2016-08-18')); +SELECT dateDiff('second', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('minute', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('hour', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('day', toDate('2015-08-18', 'UTC'), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('week', toDate('2015-08-18', 'UTC'), toDate32('2015-08-25', 'UTC'), 'UTC'); +SELECT dateDiff('month', toDate('2015-08-18', 'UTC'), toDate32('2015-09-18', 'UTC'), 'UTC'); +SELECT dateDiff('quarter', toDate('2015-08-18', 'UTC'), toDate32('2015-11-18', 'UTC'), 'UTC'); +SELECT dateDiff('year', toDate('2015-08-18', 'UTC'), toDate32('2016-08-18', 'UTC'), 'UTC'); -- Const vs non-const columns -SELECT dateDiff('day', toDate32('1927-01-01'), materialize(toDate32('1927-01-02'))); -SELECT dateDiff('day', toDate32('1927-01-01'), materialize(toDateTime64('1927-01-02 00:00:00', 3))); -SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3), materialize(toDate32('1927-01-02'))); -SELECT dateDiff('day', toDate32('2015-08-18'), materialize(toDateTime('2015-08-19 00:00:00'))); -SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00'), materialize(toDate32('2015-08-19'))); -SELECT dateDiff('day', toDate32('2015-08-18'), materialize(toDate('2015-08-19'))); -SELECT dateDiff('day', toDate('2015-08-18'), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); +SELECT dateDiff('day', toDate32('1927-01-01', 'UTC'), materialize(toDateTime64('1927-01-02 00:00:00', 3, 'UTC')), 'UTC'); +SELECT dateDiff('day', toDateTime64('1927-01-01 00:00:00', 3, 'UTC'), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), materialize(toDateTime('2015-08-19 00:00:00', 'UTC')), 'UTC'); +SELECT dateDiff('day', toDateTime('2015-08-18 00:00:00', 'UTC'), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); +SELECT dateDiff('day', toDate32('2015-08-18', 'UTC'), materialize(toDate('2015-08-19', 'UTC')), 'UTC'); +SELECT dateDiff('day', toDate('2015-08-18', 'UTC'), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); -- Non-const vs const columns -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), toDate32('1927-01-02')); -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), toDateTime64('1927-01-02 00:00:00', 3)); -SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3)), toDate32('1927-01-02')); -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), toDateTime('2015-08-19 00:00:00')); -SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00')), toDate32('2015-08-19')); -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), toDate('2015-08-19')); -SELECT dateDiff('day', materialize(toDate('2015-08-18')), toDate32('2015-08-19')); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), toDateTime64('1927-01-02 00:00:00', 3, 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3, 'UTC')), toDate32('1927-01-02', 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), toDateTime('2015-08-19 00:00:00', 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00', 'UTC')), toDate32('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), toDate('2015-08-19', 'UTC'), 'UTC'); +SELECT dateDiff('day', materialize(toDate('2015-08-18', 'UTC')), toDate32('2015-08-19', 'UTC'), 'UTC'); -- Non-const vs non-const columns -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), materialize(toDate32('1927-01-02'))); -SELECT dateDiff('day', materialize(toDate32('1927-01-01')), materialize(toDateTime64('1927-01-02 00:00:00', 3))); -SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3)), materialize(toDate32('1927-01-02'))); -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), materialize(toDateTime('2015-08-19 00:00:00'))); -SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00')), materialize(toDate32('2015-08-19'))); -SELECT dateDiff('day', materialize(toDate32('2015-08-18')), materialize(toDate('2015-08-19'))); -SELECT dateDiff('day', materialize(toDate('2015-08-18')), materialize(toDate32('2015-08-19'))); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('1927-01-01', 'UTC')), materialize(toDateTime64('1927-01-02 00:00:00', 3, 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDateTime64('1927-01-01 00:00:00', 3, 'UTC')), materialize(toDate32('1927-01-02', 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), materialize(toDateTime('2015-08-19 00:00:00', 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDateTime('2015-08-18 00:00:00', 'UTC')), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDate32('2015-08-18', 'UTC')), materialize(toDate('2015-08-19', 'UTC')), 'UTC'); +SELECT dateDiff('day', materialize(toDate('2015-08-18', 'UTC')), materialize(toDate32('2015-08-19', 'UTC')), 'UTC'); diff --git a/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.reference b/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.reference new file mode 100644 index 00000000000..a812e64a642 --- /dev/null +++ b/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.reference @@ -0,0 +1,10 @@ +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +c1 Nullable(Int64) +c2 Nullable(Int64) +c3 Nullable(Int64) +1 2 3 +4 5 6 +1 2 3 +4 5 6 diff --git a/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.sql b/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.sql new file mode 100644 index 00000000000..42e88fc44b2 --- /dev/null +++ b/tests/queries/0_stateless/02458_hdfs_cluster_schema_inference.sql @@ -0,0 +1,12 @@ +-- Tags: no-fasttest, no-parallel, no-cpu-aarch64 +-- Tag no-fasttest: Depends on Java + +insert into table function hdfs('hdfs://localhost:12222/test_02458_1.tsv', 'TSV', 'column1 UInt32, column2 UInt32, column3 UInt32') select 1, 2, 3 settings hdfs_truncate_on_insert=1; +insert into table function hdfs('hdfs://localhost:12222/test_02458_2.tsv', 'TSV', 'column1 UInt32, column2 UInt32, column3 UInt32') select 4, 5, 6 settings hdfs_truncate_on_insert=1; + +desc hdfsCluster('test_cluster_one_shard_three_replicas_localhost', 'hdfs://localhost:12222/test_02458_{1,2}.tsv'); +desc hdfsCluster('test_cluster_one_shard_three_replicas_localhost', 'hdfs://localhost:12222/test_02458_{1,2}.tsv', 'TSV'); + +select * from hdfsCluster('test_cluster_one_shard_three_replicas_localhost', 'hdfs://localhost:12222/test_02458_{1,2}.tsv') order by c1, c2, c3; +select * from hdfsCluster('test_cluster_one_shard_three_replicas_localhost', 'hdfs://localhost:12222/test_02458_{1,2}.tsv', 'TSV') order by c1, c2, c3; + diff --git a/tests/queries/0_stateless/02462_match_regexp_pk.reference b/tests/queries/0_stateless/02462_match_regexp_pk.reference new file mode 100644 index 00000000000..428d6556f4c --- /dev/null +++ b/tests/queries/0_stateless/02462_match_regexp_pk.reference @@ -0,0 +1,5 @@ +4 +1 +3 +4 +4 diff --git a/tests/queries/0_stateless/02462_match_regexp_pk.sql b/tests/queries/0_stateless/02462_match_regexp_pk.sql new file mode 100644 index 00000000000..1a944b96196 --- /dev/null +++ b/tests/queries/0_stateless/02462_match_regexp_pk.sql @@ -0,0 +1,9 @@ +CREATE TABLE mt_match_pk (v String) ENGINE = MergeTree ORDER BY v SETTINGS index_granularity = 1; +INSERT INTO mt_match_pk VALUES ('a'), ('aaa'), ('aba'), ('bac'), ('acccca'); + +SET force_primary_key = 1; +SELECT count() FROM mt_match_pk WHERE match(v, '^a'); +SELECT count() FROM mt_match_pk WHERE match(v, '^ab'); +SELECT count() FROM mt_match_pk WHERE match(v, '^a.'); +SELECT count() FROM mt_match_pk WHERE match(v, '^ab*'); +SELECT count() FROM mt_match_pk WHERE match(v, '^ac?'); diff --git a/tests/queries/0_stateless/02473_prewhere_with_bigint.reference b/tests/queries/0_stateless/02473_prewhere_with_bigint.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02473_prewhere_with_bigint.sql b/tests/queries/0_stateless/02473_prewhere_with_bigint.sql new file mode 100644 index 00000000000..29c6f0da2a1 --- /dev/null +++ b/tests/queries/0_stateless/02473_prewhere_with_bigint.sql @@ -0,0 +1,24 @@ +DROP TABLE IF EXISTS prewhere_int128; +DROP TABLE IF EXISTS prewhere_int256; +DROP TABLE IF EXISTS prewhere_uint128; +DROP TABLE IF EXISTS prewhere_uint256; + +CREATE TABLE prewhere_int128 (a Int128) ENGINE=MergeTree ORDER BY a; +INSERT INTO prewhere_int128 VALUES (1); +SELECT a FROM prewhere_int128 PREWHERE a; -- { serverError 59 } +DROP TABLE prewhere_int128; + +CREATE TABLE prewhere_int256 (a Int256) ENGINE=MergeTree ORDER BY a; +INSERT INTO prewhere_int256 VALUES (1); +SELECT a FROM prewhere_int256 PREWHERE a; -- { serverError 59 } +DROP TABLE prewhere_int256; + +CREATE TABLE prewhere_uint128 (a UInt128) ENGINE=MergeTree ORDER BY a; +INSERT INTO prewhere_uint128 VALUES (1); +SELECT a FROM prewhere_uint128 PREWHERE a; -- { serverError 59 } +DROP TABLE prewhere_uint128; + +CREATE TABLE prewhere_uint256 (a UInt256) ENGINE=MergeTree ORDER BY a; +INSERT INTO prewhere_uint256 VALUES (1); +SELECT a FROM prewhere_uint256 PREWHERE a; -- { serverError 59 } +DROP TABLE prewhere_uint256; diff --git a/tests/queries/0_stateless/02474_analyzer_subqueries_table_expression_modifiers.reference b/tests/queries/0_stateless/02474_analyzer_subqueries_table_expression_modifiers.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02474_analyzer_subqueries_table_expression_modifiers.sql b/tests/queries/0_stateless/02474_analyzer_subqueries_table_expression_modifiers.sql new file mode 100644 index 00000000000..456783cad26 --- /dev/null +++ b/tests/queries/0_stateless/02474_analyzer_subqueries_table_expression_modifiers.sql @@ -0,0 +1,17 @@ +SET allow_experimental_analyzer = 1; + +SELECT * FROM (SELECT 1) FINAL; -- { serverError 1 } +SELECT * FROM (SELECT 1) SAMPLE 1/2; -- { serverError 1 } +SELECT * FROM (SELECT 1) FINAL SAMPLE 1/2; -- { serverError 1 } + +WITH cte_subquery AS (SELECT 1) SELECT * FROM cte_subquery FINAL; -- { serverError 1 } +WITH cte_subquery AS (SELECT 1) SELECT * FROM cte_subquery SAMPLE 1/2; -- { serverError 1 } +WITH cte_subquery AS (SELECT 1) SELECT * FROM cte_subquery FINAL SAMPLE 1/2; -- { serverError 1 } + +SELECT * FROM (SELECT 1 UNION ALL SELECT 1) FINAL; -- { serverError 1 } +SELECT * FROM (SELECT 1 UNION ALL SELECT 1) SAMPLE 1/2; -- { serverError 1 } +SELECT * FROM (SELECT 1 UNION ALL SELECT 1) FINAL SAMPLE 1/2; -- { serverError 1 } + +WITH cte_subquery AS (SELECT 1 UNION ALL SELECT 1) SELECT * FROM cte_subquery FINAL; -- { serverError 1 } +WITH cte_subquery AS (SELECT 1 UNION ALL SELECT 1) SELECT * FROM cte_subquery SAMPLE 1/2; -- { serverError 1 } +WITH cte_subquery AS (SELECT 1 UNION ALL SELECT 1) SELECT * FROM cte_subquery FINAL SAMPLE 1/2; -- { serverError 1 } diff --git a/tests/queries/0_stateless/02474_create_user_query_fuzzer_bug.reference b/tests/queries/0_stateless/02474_create_user_query_fuzzer_bug.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02474_create_user_query_fuzzer_bug.sql b/tests/queries/0_stateless/02474_create_user_query_fuzzer_bug.sql new file mode 100644 index 00000000000..3ef1469cf1b --- /dev/null +++ b/tests/queries/0_stateless/02474_create_user_query_fuzzer_bug.sql @@ -0,0 +1 @@ +EXPLAIN AST ALTER user WITH a; -- { clientError SYNTAX_ERROR } diff --git a/tests/queries/0_stateless/02474_fix_function_parser_bug.reference b/tests/queries/0_stateless/02474_fix_function_parser_bug.reference new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/queries/0_stateless/02474_fix_function_parser_bug.sql b/tests/queries/0_stateless/02474_fix_function_parser_bug.sql new file mode 100644 index 00000000000..67d97aa1c25 --- /dev/null +++ b/tests/queries/0_stateless/02474_fix_function_parser_bug.sql @@ -0,0 +1 @@ +CREATE DATABASE conv_mian ENGINE QALL(COLUMNS('|T.D'),¸mp} -- { clientError SYNTAX_ERROR } diff --git a/tests/queries/0_stateless/02475_join_bug_42832.reference b/tests/queries/0_stateless/02475_join_bug_42832.reference new file mode 100644 index 00000000000..e5310261d0a --- /dev/null +++ b/tests/queries/0_stateless/02475_join_bug_42832.reference @@ -0,0 +1,2 @@ +4 6 +4 4 diff --git a/tests/queries/0_stateless/02475_join_bug_42832.sql b/tests/queries/0_stateless/02475_join_bug_42832.sql new file mode 100644 index 00000000000..e383949fb22 --- /dev/null +++ b/tests/queries/0_stateless/02475_join_bug_42832.sql @@ -0,0 +1,16 @@ +DROP TABLE IF EXISTS tab1; +DROP TABLE IF EXISTS tab2; + +SET allow_suspicious_low_cardinality_types = 1; + +CREATE TABLE tab1 (a1 Int32, b1 Int32, val UInt64) ENGINE = MergeTree ORDER BY a1; +CREATE TABLE tab2 (a2 LowCardinality(Int32), b2 Int32) ENGINE = MergeTree ORDER BY a2; + +INSERT INTO tab1 SELECT number, number, 1 from numbers(4); +INSERT INTO tab2 SELECT number + 2, number + 2 from numbers(4); + +SELECT sum(val), count(val) FROM tab1 FULL OUTER JOIN tab2 ON b1 - 2 = a2 OR a1 = b2 SETTINGS join_use_nulls = 0; +SELECT sum(val), count(val) FROM tab1 FULL OUTER JOIN tab2 ON b1 - 2 = a2 OR a1 = b2 SETTINGS join_use_nulls = 1; + +DROP TABLE IF EXISTS tab1; +DROP TABLE IF EXISTS tab2; diff --git a/utils/iotest/iotest_nonblock.cpp b/utils/iotest/iotest_nonblock.cpp index 6db00045e03..32c86282743 100644 --- a/utils/iotest/iotest_nonblock.cpp +++ b/utils/iotest/iotest_nonblock.cpp @@ -101,7 +101,7 @@ int mainImpl(int argc, char ** argv) size_t ops = 0; while (ops < count) { - if (poll(polls.data(), static_cast(descriptors), -1) <= 0) + if (poll(polls.data(), static_cast(descriptors), -1) <= 0) throwFromErrno("poll failed", ErrorCodes::SYSTEM_ERROR); for (size_t i = 0; i < descriptors; ++i) { diff --git a/utils/list-versions/version_date.tsv b/utils/list-versions/version_date.tsv index e72fce63fda..47dbec5a5f8 100644 --- a/utils/list-versions/version_date.tsv +++ b/utils/list-versions/version_date.tsv @@ -1,13 +1,18 @@ +v22.10.2.11-stable 2022-11-01 v22.10.1.1877-stable 2022-10-26 +v22.9.4.32-stable 2022-10-26 v22.9.3.18-stable 2022-09-30 v22.9.2.7-stable 2022-09-23 v22.9.1.2603-stable 2022-09-22 +v22.8.8.3-lts 2022-10-27 +v22.8.7.34-lts 2022-10-26 v22.8.6.71-lts 2022-09-30 v22.8.5.29-lts 2022-09-13 v22.8.4.7-lts 2022-08-31 v22.8.3.13-lts 2022-08-29 v22.8.2.11-lts 2022-08-23 v22.8.1.2097-lts 2022-08-18 +v22.7.7.24-stable 2022-10-26 v22.7.6.74-stable 2022-09-30 v22.7.5.13-stable 2022-08-29 v22.7.4.16-stable 2022-08-23 @@ -32,6 +37,7 @@ v22.4.5.9-stable 2022-05-06 v22.4.4.7-stable 2022-04-29 v22.4.3.3-stable 2022-04-26 v22.4.2.1-stable 2022-04-22 +v22.3.14.23-lts 2022-10-28 v22.3.13.80-lts 2022-09-30 v22.3.12.19-lts 2022-08-29 v22.3.11.12-lts 2022-08-10