aports/community/apache-arrow/APKBUILD

# Contributor: Duncan Bellamy <dunk@denkimushi.com>
# Maintainer: Duncan Bellamy <dunk@denkimushi.com>
# based on arch linux PKGBUILD
pkgname=apache-arrow
pkgver=11.0.0
pkgrel=6
pkgdesc="multi-language toolbox for accelerated data interchange and in-memory processing"
url="https://arrow.apache.org/"
arch="all"
license="Apache-2.0"
_py3depends="
	cython
	py3-numpy-dev
	py3-setuptools
	py3-setuptools_scm
	py3-wheel
	python3-dev
	"
makedepends="
	apache-orc-dev
	boost-dev
	brotli-dev
	bzip2-dev
	c-ares-dev
	cmake
	glog-dev
	grpc-dev
	gtest-dev
	lz4-dev
	openssl-dev>3
	protobuf-dev
	rapidjson-dev
	re2-dev
	samurai
	snappy-dev
	thrift-dev
	utf8proc-dev
	zlib-dev
	zstd-dev
	$_py3depends
	"
_py3checkdepends="
	py3-cffi
	py3-hypothesis
	py3-pandas
	py3-pytest
	py3-pytest-lazy-fixture
	py3-pytest-xdist
	"
checkdepends="bash grep gzip perl python3 tzdata $_py3checkdepends"
subpackages="$pkgname-dev $pkgname-tools $pkgname-doc py3-pyarrow:python_arrow"
_arrowsha="d2c73bf78246331d8e58b6f11aa8aa199cbb5929"
_parquetsha="5b82793ef7196f7b3583e85669ced211cd8b5ff2"
source="https://downloads.apache.org/arrow/arrow-$pkgver/apache-arrow-$pkgver.tar.gz
	$pkgname-arrow-testing-$_arrowsha.tar.gz::https://github.com/apache/arrow-testing/archive/$_arrowsha.tar.gz
	$pkgname-parquet-testing-$_parquetsha.tar.gz::https://github.com/apache/parquet-testing/archive/$_parquetsha.tar.gz
	"
options="!check" # fail with py3.11

case "$CARCH" in
aarch64|x86|x86_64)
	_SIMD="MAX"
	makedepends="$makedepends xsimd-dev"
	;;
*)
	_SIMD="NONE"
	;;
esac

build() {
	if [ "$CBUILD" != "$CHOST" ]; then
		CMAKE_CROSSOPTS="-DCMAKE_SYSTEM_NAME=Linux -DCMAKE_HOST_SYSTEM_NAME=Linux"
	fi

	case "$CARCH" in
	arm*)
		local arrowcpu="armv7"
		;;
	esac

	cmake -B build-cpp -G Ninja \
		-DCMAKE_INSTALL_PREFIX=/usr \
		-DCMAKE_INSTALL_LIBDIR=lib \
		-DBUILD_SHARED_LIBS=True \
		-DCMAKE_BUILD_TYPE=MinSizeRel \
		-DARROW_BUILD_STATIC=OFF \
		-DARROW_DEPENDENCY_SOURCE=SYSTEM \
		-DARROW_SIMD_LEVEL="NONE" \
		-DARROW_RUNTIME_SIMD_LEVEL="$_SIMD" \
		-DARROW_BUILD_EXAMPLES=OFF \
		-DARROW_BUILD_TESTS="$(want_check && echo ON || echo OFF)" \
		-DARROW_COMPUTE=ON \
		-DARROW_FLIGHT=ON \
		-DARROW_GANDIVA=OFF \
		-DARROW_HDFS=ON \
		-DARROW_IPC=ON \
		-DARROW_JEMALLOC=OFF \
		-DARROW_ORC=OFF \
		-DARROW_PARQUET=ON \
		-DARROW_PLASMA=ON \
		-DARROW_PYTHON=ON \
		-DARROW_TENSORFLOW=ON \
		-DARROW_USE_GLOG=ON \
		-DARROW_ORC=ON \
		-DARROW_WITH_BROTLI=ON \
		-DARROW_WITH_BZ2=ON \
		-DARROW_WITH_LZ4=ON \
		-DARROW_WITH_SNAPPY=ON \
		-DARROW_WITH_ZLIB=ON \
		-DARROW_WITH_ZSTD=ON \
		-DARROW_CPU_FLAG=$arrowcpu \
		-DPARQUET_REQUIRE_ENCRYPTION=ON \
		-S cpp \
		$CMAKE_CROSSOPTS
	cmake --build build-cpp

	# install in Arrow_DIR for python build to find
	DESTDIR="$builddir/dist-cpp" cmake --install build-cpp

	cd python
	export SETUPTOOLS_SCM_PRETEND_VERSION=$pkgver
	export Arrow_DIR="$builddir/dist-cpp/usr"
	export ArrowFlight_DIR="$Arrow_DIR"
	export Parquet_DIR="$Arrow_DIR"
	python3 setup.py build_ext --inplace \
		--extra-cmake-args="-DARROW_SIMD_LEVEL=NONE -DARROW_RUNTIME_SIMD_LEVEL=$_SIMD -DPYARROW_BUILD_DATASET=ON -DARROW_CPU_FLAG=$arrowcpu" \
		--cmake-generator Ninja
	# TODO: why does it not work without this?
	python3 setup.py install --root="$builddir/throw"
}

check() {
	cd build-cpp
	export PARQUET_TEST_DATA="$srcdir/parquet-testing-$_parquetsha/data"
	export ARROW_TEST_DATA="$srcdir/arrow-testing-$_arrowsha/data"
	# exclude broken tests
	ctest -j4 --output-on-failure -E "arrow-buffer-test|arrow-misc-test|\
arrow-utility-test|arrow-csv-test|arrow-compute-aggregate-test|arrow-flight-test|arrow-compute-scalar-test|\
arrow-dataset-file-ipc-test|arrow-dataset-scanner-test|plasma-client-tests"

	cd ../python
	ARROW_HOME="$builddir/dist-cpp/usr" \
	python3 setup.py install --root="$PWD/dist-python"
	PYTHONPATH="$(echo $PWD/dist-python/usr/lib/python3*/site-packages)" \
	LIBRARY_PATH="$builddir/dist-cpp/usr/lib:$LIBRARY_PATH" \
	LD_LIBRARY_PATH="$builddir/dist-cpp/usr/lib:$LD_LIBRARY_PATH" \
	pytest -n 4 pyarrow --deselect=pyarrow/tests/test_memory.py \
		--deselect=pyarrow/tests/test_csv.py \
		--deselect=pyarrow/tests/parquet/test_data_types.py \
		--deselect=pyarrow/tests/test_array.py::test_dictionary_to_numpy \
		--deselect=pyarrow/tests/test_io.py::test_python_file_large_seeks \
		--deselect=pyarrow/tests/test_io.py::test_foreign_buffer \
		--deselect=pyarrow/tests/test_io.py::test_memory_map_large_seeks \
		--deselect=pyarrow/tests/test_pandas.py::TestConvertStructTypes::test_from_numpy_nested \
		--deselect=pyarrow/tests/test_schema.py::test_schema_sizeof \
		--deselect=pyarrow/tests/test_serialization.py::test_primitive_serialization \
		--deselect=pyarrow/tests/test_serialization.py::test_integer_limits \
		--deselect=pyarrow/tests/parquet/test_dataset.py::test_partitioned_dataset
}

package() {
	DESTDIR="$pkgdir" cmake --install build-cpp

	cd python
	python3 setup.py install --skip-build --root="$pkgdir"

	rm -r "$pkgdir"/usr/lib/python3*/site-packages/pyarrow/tests
}

tools() {
	pkgdesc="$pkgdesc (extra tools)"

	amove usr/bin/plasma-store-server
}

python_arrow() {
	pkgdesc="$pkgdesc (python module)"
	depends="python3 py3-cffi py3-numpy"

	# renamed from this to avoid confusion as in python this is "pyarrow"
	provides="py3-apache-arrow=$pkgver-r$pkgrel"
	replaces="py3-apache-arrow"

	amove usr/lib/python3* usr/bin/plasma_store
}

sha512sums="
46df4fb5a703d38d0a74fde9838e9f9702b24b442cb225517516c335a5ab18955699000bf0b2fc7d1698ada6d2e890ba3860933b6280f5160b0fce8a07484d0e  apache-arrow-11.0.0.tar.gz
0be19960b0d22fc2e07bf84f11148e69d6fa82f10627eb2a5e4b762b1d4cf4e151a57c5a1dc761d6d8ed29881ea589e4cfe0cd47d5c7cbf7b6107143b4a45e82  apache-arrow-arrow-testing-d2c73bf78246331d8e58b6f11aa8aa199cbb5929.tar.gz
b761ccca140ce109fd39320a0e0ab9b9cae07f0f28c0993eada332d6fd5f7a3854230c260467a9fd9f5b9a8941a375280f473f929480dca370ae45e31525b13d  apache-arrow-parquet-testing-5b82793ef7196f7b3583e85669ced211cd8b5ff2.tar.gz
"