|author||Wes McKinney <firstname.lastname@example.org>||Tue Sep 19 21:38:52 2017 -0400|
|committer||Wes McKinney <email@example.com>||Tue Sep 19 21:38:52 2017 -0400|
PARQUET-1100: Introduce RecordReader interface to better support nested data, refactor parquet/arrow/reader We did not have very consistent logic around reading values from leaf nodes versus reading semantic records where the repetition level is greater than zero. This introduces a reader class that reads from column chunks until it identifies the end of records. It also reads values (with spaces, if required by the schema) into internal buffers. This permitted a substantial refactoring and simplification of the code in parquet::arrow where we were handling the interpretation of batch reads as records manually. As follow up patch, we should be able to take a collection of record readers from the same "tree" in a nested type and reassemble the intermediate Arrow structure and dealing with any redundant structure information in repetition and definition levels. This should a allow a unification of our nested data read code path so that we can read arbitrary nested structures. Author: Wes McKinney <firstname.lastname@example.org> Closes #398 from wesm/PARQUET-1100 and squashes the following commits: 9ea85d9 [Wes McKinney] Revert to const args f4dc0fe [Wes McKinney] Make parquet::schema::Node non-copyable. Use const-refs instead of const-ptr for non-nullable argument 0d859cc [Wes McKinney] Code review comments, scrubbing some flakes 1368415 [Wes McKinney] Fix more MSVC warnings eccb84c [Wes McKinney] Give macro more accurate name 0eaada0 [Wes McKinney] Use int64_t instead of int for batch sizes 79c3709 [Wes McKinney] Add documentation. Remove RecordReader from public API 8fa619b [Wes McKinney] Initialize memory in DecodeSpaced to avoid undefined behavior 5a0c860 [Wes McKinney] Remove non-repeated branch from DelimitRecords c754e6e [Wes McKinney] Refactor to skip record delimiting for non-repeated data ed2a03f [Wes McKinney] Move more code into TypedRecordReader 2e934e9 [Wes McKinney] Set some integers as const 58d3a0f [Wes McKinney] Do not index into levels arrays b766371 [Wes McKinney] Add RecordReader::Reserve to preallocate, fixing perf regression. cpplint 1bf3e8f [Wes McKinney] Refactor to create stateful parquet::RecordReader class to better support nested data. Shift value buffering logic from parquet/arrow/reader into RecordReader. Fix bug described in PARQUET-1100
We use the CMake build system and require a minimum version of 3.2. If you are using an older Linux distribution, you may need to use a PPA (for apt users) or build CMake from source.
parquet-cpp requires gcc 4.8 or higher on Linux.
To build parquet-cpp out of the box, you must install some build prerequisites for the thirdparty dependencies. On Debian/Ubuntu, these can be installed with:
sudo apt-get install libboost-dev libboost-filesystem-dev \ libboost-program-options-dev libboost-regex-dev \ libboost-system-dev libboost-test-dev \ libssl-dev libtool bison flex pkg-config
You must use XCode 6 or higher. We recommend using Homebrew to install Boost, which is required for Thrift:
brew install boost
Check Windows developer guide for instructions to build parquet-cpp on Windows.
You can either install these dependencies separately, otherwise they will be built automatically as part of the build.
Symbols from Thrift, Snappy, and ZLib are statically-linked into the
libparquet shared library, so these dependencies must be built with
-fPIC on Linux and OS X. Since Linux package managers do not consistently compile the static libraries for these components with
-fPIC, you may have issues with Linux packages such as
libsnappy-dev. It may be easier to depend on the thirdparty toolchain that parquet-cpp builds automatically.
The binaries will be built to ./debug which contains the libraries to link against as well as a few example executables.
To disable the testing (which requires
For release-level builds (enable optimizations and disable debugging), pass
To build only the library with minimal dependencies, pass
cmake. Note that the executables, tests, and benchmarks should be disabled as well.
Incremental builds can be done afterwords with just
Arrow provides some of the memory management and IO interfaces that we use in parquet-cpp. By default, Parquet links to Arrow's shared libraries. If you wish to statically-link the Arrow symbols instead, pass
This library uses Google's
googletest unit test framework. After building with
make, you can run the test suite by running
The test suite relies on an environment variable
PARQUET_TEST_DATA pointing to the
data directory in the source checkout, for example:
ctest --help for configuration details about ctest. On GNU/Linux systems, you can use valgrind with ctest to look for memory leaks:
valgrind --tool=memcheck --leak-check=yes ctest
Follow the directions for simple build except run cmake with the
--PARQUET_BUILD_BENCHMARKS parameter set correctly:
cmake -DPARQUET_BUILD_BENCHMARKS=ON ..
and instead of make unittest run either
make; ctest to run both unit tests and benchmarks or
make runbenchmark to run only the benchmark tests.
Benchmark logs will be placed in the build directory under
parquet-cpp supports out of source builds. For example:
mkdir test-build cd test-build cmake .. make ctest -L unittest
By using out-of-source builds you can preserve your current build state in case you need to switch to another git branch.
The library consists of 3 layers that map to the 3 units in the parquet format.
The first is the encodings which correspond to data pages. The APIs at this level return single values.
The second layer is the column reader which corresponds to column chunks. The APIs at this level return a triple: definition level, repetition level and value. It also handles reading pages, compression and managing encodings.
The 3rd layer would handle reading/writing records.
The project adheres to the google coding convention: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml with two notable exceptions. We do not encourage anonymous namespaces and the line length is 90 characters.
You can run
cpplint through the build system with
The project prefers the use of C++ style memory management. new/delete should be used over malloc/free. new/delete should be avoided whenever possible by using stl/boost where possible. For example, scoped_ptr instead of explicit new/delete and using std::vector instead of allocated buffers. Currently, c++11 features are not used.
For error handling, this project uses exceptions.
In general, many of the APIs at the layers are interface based for extensibility. To minimize the cost of virtual calls, the APIs should be batch-centric. For example, encoding should operate on batches of values rather than a single value.
Suppose you are building libraries with a thirdparty gcc toolchain (not a built-in system one) on Linux. To use clang for development while linking to the proper toolchain, you can do (for out of source builds):
export CMAKE_CLANG_OPTIONS=--gcc-toolchain=$TOOLCHAIN/gcc-4.9.2 export CC=$TOOLCHAIN/llvm-3.7.0/bin/clang export CXX=$TOOLCHAIN/llvm-3.7.0/bin/clang++ cmake -DCMAKE_CLANG_OPTIONS=$CMAKE_CLANG_OPTIONS \ -DCMAKE_CXX_FLAGS="-Werror" ..
First, build the project with coverage and run the test suite
cd $PARQUET_HOME mkdir coverage-build cd coverage-build cmake -DPARQUET_GENERATE_COVERAGE=1 make -j$PARALLEL ctest -L unittest
gcov artifacts are not located in a place that works well with either coveralls or codecov, so there is a helper script you need to run
mkdir coverage_artifacts python ../build-support/collect_coverage.py CMakeFiles/parquet.dir/src/ coverage_artifacts
For codecov.io (using the provided project token -- be sure to keep this private):
cd coverage_artifacts codecov --token $PARQUET_CPP_CODECOV_TOKEN --gcov-args '\-l' --root $PARQUET_ROOT
For coveralls, install
pip install cpp_coveralls
And the coveralls upload script:
coveralls -t $PARQUET_CPP_COVERAGE_TOKEN --gcov-options '\-l' -r $PARQUET_ROOT --exclude $PARQUET_ROOT/thirdparty --exclude $PARQUET_ROOT/build --exclude $NATIVE_TOOLCHAIN --exclude $PARQUET_ROOT/src/parquet/thrift
gcov throws off artifacts from the STL, so I excluded my toolchain root stored in
$NATIVE_TOOLCHAIN to avoid a cluttered coverage report.