# Contributor: Duncan Bellamy <dunk@denkimushi.com>
# Maintainer: Duncan Bellamy <dunk@denkimushi.com>
# based on arch linux PKGBUILD
pkgname=apache-arrow
pkgver=19.0.1
pkgrel=6
_dotver=${pkgver//./}
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-gpep517
	py3-numpy-dev
	py3-setuptools
	py3-setuptools_scm
	py3-typing-extensions
	py3-wheel
	python3-dev
	"
makedepends="
	abseil-cpp-dev
	apache-orc-dev
	boost-dev
	brotli-dev
	bzip2-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-xdist
	"
checkdepends="bash grep gzip perl python3 tzdata $_py3checkdepends"
somask="libarrow_python.so.$_dotver libarrow_python_flight.so.$_dotver libarrow_python_parquet_encryption.so.$_dotver"
subpackages="
	$pkgname-dev
	$pkgname-doc
	$pkgname-gdb
	py3-pyarrow-pyc
	py3-pyarrow:python_arrow
	libarrow:lib
	libarrow_acero:lib
	libarrow_dataset:lib
	libarrow_flight:lib
	libparquet:lib
	"
_arrowsha="e81d0c6de35948b3be7984af8e00413b314cde6e"
_parquetsha="d79a0101d90dfa3bbb10337626f57a3e8c4b5363"
source="https://archive.apache.org/dist/arrow/arrow-$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 net" # 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

	export CFLAGS="$CFLAGS -O2 -flto=auto -DNDEBUG"
	export CXXFLAGS="$CXXFLAGS -O2 -flto=auto -DNDEBUG"
	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_RUNTIME_SIMD_LEVEL="$_SIMD" \
		-DARROW_BUILD_EXAMPLES=OFF \
		-DARROW_BUILD_TESTS="$(want_check && echo ON || echo OFF)" \
		-DARROW_COMPUTE=ON \
		-DARROW_CSV=ON \
		-DARROW_DATASET=ON \
		-DARROW_FILESYSTEM=ON \
		-DARROW_FLIGHT=ON \
		-DARROW_HDFS=ON \
		-DARROW_JEMALLOC=OFF \
		-DARROW_JSON=ON \
		-DARROW_PARQUET=ON \
		-DARROW_SIMD_LEVEL="NONE" \
		-DARROW_TENSORFLOW=ON \
		-DARROW_USE_GLOG=ON \
		-DARROW_ORC=ON \
		-DARROW_WITH_BROTLI=ON \
		-DARROW_WITH_BZ2=ON \
		-DARROW_WITH_LZ4=ON \
		-DARROW_WITH_MUSL=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 ArrowAcero_DIR="$Arrow_DIR"
	export ArrowDataset_DIR="$Arrow_DIR"
	export ArrowFlight_DIR="$Arrow_DIR"
	export Parquet_DIR="$Arrow_DIR"
	export PYARROW_CMAKE_OPTIONS="-DARROW_RUNTIME_SIMD_LEVEL=$_SIMD -DARROW_CPU_FLAG=$arrowcpu"
	gpep517 build-wheel \
		--wheel-dir .dist \
		--output-fd 3 3>&1 >&2
}

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 -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"

	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 -m installer -d "$pkgdir" .dist/*.whl

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

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*
}

gdb() {
	pkgdesc="$pkgdesc (gdb integration)"
	install_if="$pkgname=$pkgver-r$pkgrel gdb"

	amove \
		usr/share/arrow/gdb/ \
		usr/share/gdb/
}

lib() {
	pkgdesc="$pkgdesc ($subpkgname library)"
	depends="$pkgname=$pkgver-r$pkgrel"

	amove usr/lib/$subpkgname.so.*
}

sha512sums="
524187c6f12bbb677b5d4a53e1608c69b56c83a5c8667bfe82d5a10232c33f151a8a7b5e6e26af48d1e0ca25a5d3dc885b27d9b5e798c7e07f115df8f0779516  apache-arrow-19.0.1.tar.gz
2c31dd48fc070c3b1b25cdd1d58615accf73e3ee864d58b9a92a92c71a531174abeae42c2d0e6a70bcce47fe01c6cf43b1c17fab0636fb51524db8d035efce20  apache-arrow-arrow-testing-e81d0c6de35948b3be7984af8e00413b314cde6e.tar.gz
5bf4f1341dfab0f71d61d00ef89e2f6a49831af3e5ade1209b9f35abf61d5edcbe311bebb2bb5d8611486e16fe97bfa8c85974dc2ea11325b58fb74c1cc09e3a  apache-arrow-parquet-testing-d79a0101d90dfa3bbb10337626f57a3e8c4b5363.tar.gz
"
