commit: edf6876c43828c4321e0f9b70398f4cc8f51da55
Author: Michał Górny <mgorny <AT> gentoo <DOT> org>
AuthorDate: Wed Jul 10 09:51:50 2024 +0000
Commit: Michał Górny <mgorny <AT> gentoo <DOT> org>
CommitDate: Wed Jul 10 10:53:03 2024 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=edf6876c
dev-python/tables: Add a patch for numpy-2 support
Signed-off-by: Michał Górny <mgorny <AT> gentoo.org>
dev-python/tables/files/tables-3.9.2-numpy-2.patch | 264 +++++++++++++++++++++
dev-python/tables/tables-3.9.2-r1.ebuild | 80 +++++++
2 files changed, 344 insertions(+)
diff --git a/dev-python/tables/files/tables-3.9.2-numpy-2.patch
b/dev-python/tables/files/tables-3.9.2-numpy-2.patch
new file mode 100644
index 000000000000..a3d56d13f1a8
--- /dev/null
+++ b/dev-python/tables/files/tables-3.9.2-numpy-2.patch
@@ -0,0 +1,264 @@
+diff --git a/setup.py b/setup.py
+index 48dc6ca5..e8957c1f 100755
+--- a/setup.py
++++ b/setup.py
+@@ -736,7 +736,10 @@ if __name__ == "__main__":
+
+ # -----------------------------------------------------------------
+
+- def_macros = [("NDEBUG", 1)]
++ def_macros = [
++ ("NDEBUG", 1),
++ ("NPY_TARGET_VERSION", "NPY_1_20_API_VERSION"),
++ ]
+
+ # Define macros for Windows platform
+ if os.name == "nt":
+diff --git a/src/utils.c b/src/utils.c
+index 15fce02d..b28dbc90 100644
+--- a/src/utils.c
++++ b/src/utils.c
+@@ -765,8 +765,8 @@ hid_t create_ieee_complex64(const char *byteorder) {
+ return float_id;
+ }
+
+- H5Tinsert(complex_id, "r", HOFFSET(npy_complex64, real), float_id);
+- H5Tinsert(complex_id, "i", HOFFSET(npy_complex64, imag), float_id);
++ H5Tinsert(complex_id, "r", 0, float_id);
++ H5Tinsert(complex_id, "i", 4, float_id);
+ H5Tclose(float_id);
+ return complex_id;
+ }
+@@ -790,8 +790,8 @@ hid_t create_ieee_complex128(const char *byteorder) {
+ return float_id;
+ }
+
+- H5Tinsert(complex_id, "r", HOFFSET(npy_complex128, real), float_id);
+- H5Tinsert(complex_id, "i", HOFFSET(npy_complex128, imag), float_id);
++ H5Tinsert(complex_id, "r", 0, float_id);
++ H5Tinsert(complex_id, "i", 8, float_id);
+ H5Tclose(float_id);
+ return complex_id;
+ }
+@@ -822,8 +822,8 @@ hid_t create_ieee_complex192(const char *byteorder) {
+ return err;
+ }
+
+- H5Tinsert(complex_id, "r", HOFFSET(npy_complex192, real), float_id);
+- H5Tinsert(complex_id, "i", HOFFSET(npy_complex192, imag), float_id);
++ H5Tinsert(complex_id, "r", 0, float_id);
++ H5Tinsert(complex_id, "i", 12, float_id);
+ H5Tclose(float_id);
+ return complex_id;
+ }
+@@ -854,8 +854,8 @@ hid_t create_ieee_complex256(const char *byteorder) {
+ return err;
+ }
+
+- H5Tinsert(complex_id, "r", HOFFSET(npy_complex256, real), float_id);
+- H5Tinsert(complex_id, "i", HOFFSET(npy_complex256, imag), float_id);
++ H5Tinsert(complex_id, "r", 0, float_id);
++ H5Tinsert(complex_id, "i", 16, float_id);
+ H5Tclose(float_id);
+ return complex_id;
+ }
+diff --git a/tables/atom.py b/tables/atom.py
+index 56ab6423..5d4fba2b 100644
+--- a/tables/atom.py
++++ b/tables/atom.py
+@@ -276,15 +276,15 @@ class Atom(metaclass=MetaAtom):
+ >>> atom1 = StringAtom(itemsize=10) # same as ``atom2``
+ >>> atom2 = Atom.from_kind('string', 10) # same as ``atom1``
+ >>> atom3 = IntAtom()
+- >>> atom1 == 'foo'
++ >>> bool(atom1 == 'foo')
+ False
+- >>> atom1 == atom2
++ >>> bool(atom1 == atom2)
+ True
+- >>> atom2 != atom1
++ >>> bool(atom2 != atom1)
+ False
+- >>> atom1 == atom3
++ >>> bool(atom1 == atom3)
+ False
+- >>> atom3 != atom2
++ >>> bool(atom3 != atom2)
+ True
+
+ """
+diff --git a/tables/index.py b/tables/index.py
+index e8c8caf7..28ff37e2 100644
+--- a/tables/index.py
++++ b/tables/index.py
+@@ -581,7 +581,8 @@ class Index(NotLoggedMixin, Group, indexesextension.Index):
+ # Add a second offset in this case
+ # First normalize the number of rows
+ offset2 = (nrow % self.nslicesblock) * slicesize // lbucket
+- idx += offset2
++ assert offset2 < 2**(indsize*8)
++ idx += np.asarray(offset2).astype(idx.dtype)
+ # Add the last row at the beginning of arr & idx (if needed)
+ if (indsize == 8 and nelementsILR > 0):
+ # It is possible that the values in LR are already sorted.
+@@ -622,11 +623,11 @@ class Index(NotLoggedMixin, Group,
indexesextension.Index):
+ show_stats("Entering final_idx32", tref)
+ # Do an upcast first in order to add the offset.
+ idx = idx.astype('uint64')
+- idx += offset
++ idx += np.asarray(offset).astype(idx.dtype)
+ # The next partition is valid up to table sizes of
+ # 2**30 * 2**18 = 2**48 bytes, that is, 256 Tera-elements,
+ # which should be a safe figure, at least for a while.
+- idx //= self.lbucket
++ idx //= np.asarray(self.lbucket).astype(idx.dtype)
+ # After the division, we can downsize the indexes to 'uint32'
+ idx = idx.astype('uint32')
+ if profile:
+@@ -2002,7 +2003,7 @@ class Index(NotLoggedMixin, Group,
indexesextension.Index):
+ else:
+ self.indicesLR._read_index_slice(start, stop, idx)
+ if indsize == 8:
+- idx //= lbucket
++ idx //= np.asarray(lbucket).astype(idx.dtype)
+ elif indsize == 2:
+ # The chunkmap size cannot be never larger than 'int_'
+ idx = idx.astype("int_")
+diff --git a/tables/tests/common.py b/tables/tests/common.py
+index 31378a88..1992f39a 100644
+--- a/tables/tests/common.py
++++ b/tables/tests/common.py
+@@ -205,7 +205,7 @@ def allequal(a, b, flavor="numpy"):
+ return result
+
+
+-def areArraysEqual(arr1, arr2):
++def areArraysEqual(arr1, arr2, *, check_type=True):
+ """Are both `arr1` and `arr2` equal arrays?
+
+ Arguments can be regular NumPy arrays, chararray arrays or
+@@ -217,8 +217,8 @@ def areArraysEqual(arr1, arr2):
+ t1 = type(arr1)
+ t2 = type(arr2)
+
+- if not ((hasattr(arr1, 'dtype') and arr1.dtype == arr2.dtype) or
+- issubclass(t1, t2) or issubclass(t2, t1)):
++ if check_type and not ((hasattr(arr1, 'dtype') and arr1.dtype ==
arr2.dtype) or
++ issubclass(t1, t2) or issubclass(t2, t1)):
+ return False
+
+ return np.all(arr1 == arr2)
+diff --git a/tables/tests/test_expression.py b/tables/tests/test_expression.py
+index 018d4208..d9c0e990 100644
+--- a/tables/tests/test_expression.py
++++ b/tables/tests/test_expression.py
+@@ -265,9 +265,12 @@ class MixedContainersTestCase(common.TempFileMixin,
common.PyTablesTestCase):
+ if common.verbose:
+ print("Computed expression:", repr(r1), r1.dtype)
+ print("Should look like:", repr(r2), r2.dtype)
+- self.assertTrue(
+- r1.shape == r2.shape and r1.dtype == r2.dtype and r1 == r2,
+- "Evaluate is returning a wrong value.")
++ msg = f"Evaluate is returning a wrong value: {expr_str}\n{r1=}\n{r2=}"
++ self.assertEqual(r1.shape, r2.shape, msg=msg)
++ # In something like 2 * np.in16(3) + np.int16(2) the result is still a
++ # np.int16 in NumPy 2.0, so we shouldn't actually check this:
++ # self.assertEqual(r1.dtype, r2.dtype, msg=msg)
++ self.assertEqual(r1, r2, msg=msg)
+
+ def test01a_out(self):
+ """Checking expressions with mixed objects (`out` param)"""
+@@ -305,8 +308,9 @@ class MixedContainersTestCase(common.TempFileMixin,
common.PyTablesTestCase):
+ if common.verbose:
+ print("Computed expression:", repr(r1), r1.dtype)
+ print("Should look like:", repr(r2), r2.dtype)
+- self.assertTrue(common.areArraysEqual(r1, r2),
+- "Evaluate is returning a wrong value.")
++ msg = f"Evaluate is returning a wrong value:
{expr_str}\n{r1=}\n{r2=}"
++ # On NumPy 2 type promotion is different so don't check type here
++ self.assertTrue(common.areArraysEqual(r1, r2, check_type=False),
msg=msg)
+
+ def test02a_sss(self):
+ """Checking mixed objects and start, stop, step (I)"""
+diff --git a/tables/tests/test_indexvalues.py
b/tables/tests/test_indexvalues.py
+index fac33af8..85ca38c3 100644
+--- a/tables/tests/test_indexvalues.py
++++ b/tables/tests/test_indexvalues.py
+@@ -2296,6 +2296,11 @@ class SelectValuesTestCase(common.TempFileMixin,
common.PyTablesTestCase):
+ self.assertFalse(t1var3.index.dirty)
+ self.assertFalse(t1var4.index.dirty)
+
++ # TODO: IT IS DIRTY BECAUSE THIS FIXES THINGS FOR FINSV2aTestCase,
++ # which otherwise fails a test a few lines below!
++ for col in table1.colinstances.values():
++ col.reindex()
++
+ # Do some selections and check the results
+ # First selection: string
+ # Convert the limits to the appropriate type
+@@ -2318,11 +2323,15 @@ class SelectValuesTestCase(common.TempFileMixin,
common.PyTablesTestCase):
+ # Second selection: bool
+ results1 = [p["var2"] for p in table1.where('t1var2 == True')]
+ results2 = [p["var2"] for p in table2 if p["var2"] is True]
+- if common.verbose:
+- print("Length results:", len(results1))
+- print("Should be:", len(results2))
+- self.assertEqual(len(results1), len(results2))
+- self.assertEqual(results1, results2)
++ t2var1_vals = [p["var1"] for p in table2]
++ t2var2_vals = [p["var2"] for p in table2]
++ msg = (
++ f"Incorrect results for t1var2[n] == True where\n"
++
f"t2var1_vals={repr(t2var1_vals)}\nt2var2_vals={repr(t2var2_vals)}\n"
++ f"\n{results1=}\n{results2=}"
++ )
++ self.assertEqual(len(results1), len(results2), msg=msg)
++ self.assertEqual(results1, results2, msg=msg)
+
+ # Third selection: int
+ # Convert the limits to the appropriate type
+@@ -3228,7 +3237,9 @@ class LastRowReuseBuffers(common.PyTablesTestCase):
+
+
+ normal_tests = (
+- "SV1aTestCase", "SV2aTestCase", "SV3aTestCase",
++ "SV1aTestCase",
++ "SV2aTestCase",
++ "SV3aTestCase",
+ )
+
+ heavy_tests = (
+diff --git a/tables/utils.py b/tables/utils.py
+index e11e5ba7..7d786e32 100644
+--- a/tables/utils.py
++++ b/tables/utils.py
+@@ -25,6 +25,11 @@ byteorders = {
+ SizeType = np.int64
+
+
++copy_if_needed = (
++ None if np.lib.NumpyVersion(np.__version__) >= "2.0.0" else False
++)
++
++
+ def correct_byteorder(ptype, byteorder):
+ """Fix the byteorder depending on the PyTables types."""
+
+@@ -78,7 +83,7 @@ def idx2long(index):
+ # with atom from a generic python type. If copy is stated as True, it
+ # is assured that it will return a copy of the object and never the same
+ # object or a new one sharing the same memory.
+-def convert_to_np_atom(arr, atom, copy=False):
++def convert_to_np_atom(arr, atom, copy=copy_if_needed):
+ """Convert a generic object into a NumPy object compliant with atom."""
+
+ # First, convert the object into a NumPy array
+@@ -112,7 +117,7 @@ def convert_to_np_atom2(object, atom):
+
+ # Check whether the object needs to be copied to make the operation
+ # safe to in-place conversion.
+- copy = atom.type in ['time64']
++ copy = True if atom.type in ['time64'] else copy_if_needed
+ nparr = convert_to_np_atom(object, atom, copy)
+ # Finally, check the byteorder and change it if needed
+ byteorder = byteorders[nparr.dtype.byteorder]
diff --git a/dev-python/tables/tables-3.9.2-r1.ebuild
b/dev-python/tables/tables-3.9.2-r1.ebuild
new file mode 100644
index 000000000000..bdbc563361d3
--- /dev/null
+++ b/dev-python/tables/tables-3.9.2-r1.ebuild
@@ -0,0 +1,80 @@
+# Copyright 1999-2024 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_EXT=1
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( pypy3 python3_{10..12} )
+PYTHON_REQ_USE="threads(+)"
+
+inherit distutils-r1 prefix pypi
+
+DESCRIPTION="Hierarchical datasets for Python"
+HOMEPAGE="
+ https://www.pytables.org/
+ https://github.com/PyTables/PyTables/
+ https://pypi.org/project/tables/
+"
+
+LICENSE="BSD"
+SLOT="0"
+KEYWORDS="~amd64 ~arm ~arm64 ~ia64 ~loong ~ppc64 ~riscv ~sparc ~x86
~amd64-linux ~x86-linux"
+IUSE="+cpudetection examples test"
+RESTRICT="!test? ( test )"
+
+DEPEND="
+ app-arch/bzip2:0=
+ app-arch/lz4:0=
+ >=app-arch/zstd-1.0.0:=
+ >=dev-libs/c-blosc-1.11.1:0=
+ >=dev-libs/c-blosc2-2.11.0:=
+ dev-libs/lzo:2=
+ >=dev-python/numpy-1.19.0:=[${PYTHON_USEDEP}]
+ >=sci-libs/hdf5-1.8.4:=
+"
+RDEPEND="
+ ${DEPEND}
+ >=dev-python/numexpr-2.6.2[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ cpudetection? ( dev-python/py-cpuinfo[${PYTHON_USEDEP}] )
+"
+BDEPEND="
+ >=dev-python/cython-0.21[${PYTHON_USEDEP}]
+ virtual/pkgconfig
+ cpudetection? ( dev-python/py-cpuinfo[${PYTHON_USEDEP}] )
+ test? (
+ ${RDEPEND}
+ )
+"
+
+python_prepare_all() {
+ local PATCHES=(
+ # https://github.com/PyTables/PyTables/pull/1176
+ "${FILESDIR}/${P}-numpy-2.patch"
+ )
+
+ rm -r c-blosc/{blosc,internal-complibs} || die
+
+ distutils-r1_python_prepare_all
+
+ sed -i -e '/blosc2/d' pyproject.toml || die
+ hprefixify -w '/prefixes =/' setup.py
+
+ export PYTABLES_NO_EMBEDDED_LIBS=1
+ export USE_PKGCONFIG=TRUE
+}
+
+python_test() {
+ cd "${BUILD_DIR}/install$(python_get_sitedir)" || die
+ "${EPYTHON}" tables/tests/test_all.py -v || die
+}
+
+python_install_all() {
+ distutils-r1_python_install_all
+
+ if use examples; then
+ dodoc -r contrib examples
+ docompress -x /usr/share/doc/${PF}/{contrib,examples}
+ fi
+}
