* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-03-28 17:51 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-03-28 17:51 UTC (permalink / raw
To: gentoo-commits
commit: 13e21e84fe92ee3cdf7eaa7657eef3f4e172ebb0
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Fri Mar 28 17:50:45 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Fri Mar 28 17:50:45 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=13e21e84
sci-ml/datasets: drop 2.21.0-r5
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 -
sci-ml/datasets/datasets-2.21.0-r5.ebuild | 152 ------------------------------
2 files changed, 153 deletions(-)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index e2747acfecc8..731f0e890079 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -1,2 +1 @@
-DIST datasets-2.21.0.gh.tar.gz 2198480 BLAKE2B 41a5e82013d26644db4f52b16728cee31c6ff27995b8a9b2ddef1512fcda5e9386e47f6726303fec64d5ed7e40432583a6ddcc2bdfbb76096dce5f23b0540149 SHA512 f4193e29505300f98ec026b1ed052ffd3d43ae05a9b191bb56105d703ba2e44752e97c0eeb36dd1b2df6499662915591ab850cdec5248af8c7c5b5a003437c5c
DIST datasets-3.0.2.gh.tar.gz 1867037 BLAKE2B 92475ac566bd825c542dbd3e138c83e1733e6d1137f30a2cfd8f82f7dd54ed78c0566aa766182625e8787413be18f5a0c0c0a289b5823fa9112dcb2876074b7e SHA512 aa55dd2f53df7217c30999f759c76a13d78851233bce4df9b1ad85081e101a7280e905822160874d24d15ea1e1a5ecee1bae9cc66785e89af45ad16dd0339595
diff --git a/sci-ml/datasets/datasets-2.21.0-r5.ebuild b/sci-ml/datasets/datasets-2.21.0-r5.ebuild
deleted file mode 100644
index 7d13d51dcea0..000000000000
--- a/sci-ml/datasets/datasets-2.21.0-r5.ebuild
+++ /dev/null
@@ -1,152 +0,0 @@
-# Copyright 2023-2025 Gentoo Authors
-# Distributed under the terms of the GNU General Public License v2
-
-EAPI=8
-
-DISTUTILS_USE_PEP517=setuptools
-PYTHON_COMPAT=( python3_{10..12} )
-DISTUTILS_SINGLE_IMPL=1
-inherit distutils-r1
-
-DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
-HOMEPAGE="https://pypi.org/project/datasets/"
-SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
- -> ${P}.gh.tar.gz"
-
-LICENSE="Apache-2.0"
-SLOT="0"
-KEYWORDS="~amd64"
-
-IUSE="test"
-
-RDEPEND="
- ${PYTHON_DEPS}
- sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
- sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
- $(python_gen_cond_dep '
- dev-python/aiohttp[${PYTHON_USEDEP}]
- dev-python/dill[${PYTHON_USEDEP}]
- dev-python/filelock[${PYTHON_USEDEP}]
- dev-python/fsspec[${PYTHON_USEDEP}]
- dev-python/multiprocess[${PYTHON_USEDEP}]
- dev-python/numpy[${PYTHON_USEDEP}]
- dev-python/packaging[${PYTHON_USEDEP}]
- dev-python/pandas[${PYTHON_USEDEP}]
- dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
- dev-python/pyyaml[${PYTHON_USEDEP}]
- dev-python/requests[${PYTHON_USEDEP}]
- dev-python/tqdm[${PYTHON_USEDEP}]
- dev-python/xxhash[${PYTHON_USEDEP}]
- sci-ml/huggingface_hub[${PYTHON_USEDEP}]
- ')
-"
-DEPEND="${RDEPEND}"
-BDEPEND="test? (
- $(python_gen_cond_dep '
- dev-python/absl-py[${PYTHON_USEDEP}]
- dev-python/decorator[${PYTHON_USEDEP}]
- dev-python/pytest-datadir[${PYTHON_USEDEP}]
- dev-python/scikit-learn[${PYTHON_USEDEP}]
- dev-python/sqlalchemy[${PYTHON_USEDEP}]
- dev-python/zstandard[${PYTHON_USEDEP}]
- sci-ml/jiwer[${PYTHON_USEDEP}]
- sci-ml/seqeval[${PYTHON_USEDEP}]
- ')
- sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
-)"
-
-PATCHES=(
- "${FILESDIR}"/${P}-tests.patch
-)
-
-distutils_enable_tests pytest
-
-src_prepare() {
- distutils-r1_src_prepare
- sed -i -e \
- "/pyarrow_hotfix/d" \
- src/datasets/features/features.py || die
-}
-
-src_test() {
- local EPYTEST_IGNORE=(
- tests/test_upstream_hub.py
- tests/packaged_modules/test_spark.py
- tests/test_load.py
- )
-
- local EPYTEST_DESELECT=(
- "tests/features/test_audio.py::test_dataset_with_audio_feature_map_undecoded"
- "tests/features/test_audio.py::test_dataset_with_audio_feature_undecoded"
- "tests/features/test_audio.py::test_formatted_dataset_with_audio_feature_undecoded"
- "tests/io/test_parquet.py::test_parquet_read_geoparquet"
- "tests/packaged_modules/test_cache.py::test_cache_multi_configs"
- "tests/packaged_modules/test_cache.py::test_cache_single_config"
- "tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk"
- "tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk"
- "tests/test_arrow_dataset.py::TaskTemplatesTest::test_task_automatic_speech_recognition"
- "tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing"
- "tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files"
- "tests/test_distributed.py::test_torch_distributed_run"
- "tests/test_exceptions.py::test_error_deprecated"
- "tests/test_file_utils.py::TestxPath::test_xpath_rglob"
- "tests/test_file_utils.py::TestxPath::test_xpath_glob"
- "tests/test_fingerprint.py::HashingTest::test_hash_torch_compiled_module"
- "tests/test_fingerprint.py::HashingTest::test_hash_torch_generator"
- "tests/test_fingerprint.py::HashingTest::test_hash_torch_tensor"
- "tests/test_fingerprint.py::HashingTest::test_set_doesnt_depend_on_order"
- "tests/test_fingerprint.py::HashingTest::test_set_stable"
- "tests/test_fingerprint.py::RecurseHashTest::test_hash_ignores_line_definition_of_function"
- "tests/test_fingerprint.py::RecurseHashTest::test_hash_ipython_function"
- "tests/test_fingerprint.py::test_move_script_doesnt_change_hash"
- "tests/test_fingerprint.py::TokenizersHashTest::test_hash_regex"
- "tests/test_fingerprint.py::TokenizersHashTest::test_hash_tokenizer"
- "tests/test_fingerprint.py::TokenizersHashTest::test_hash_tokenizer_with_cache"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_natural_questions/default@19ba7767b174ad046a84f46af056517a3910ee57"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wiki40b/en@7b21a2e64b90323b2d3d1b81aa349bb4bc76d9bf"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wiki_dpr/psgs_w100.multiset.no_index@b24a417d802a583f8922946c1c75210290e93108"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wiki_dpr/psgs_w100.nq.compressed@b24a417d802a583f8922946c1c75210290e93108"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wiki_dpr/psgs_w100.nq.no_index@b24a417d802a583f8922946c1c75210290e93108"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.de@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.en@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.fr@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.frr@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.it@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::TestDatasetOnHfGcp::test_dataset_info_available_wikipedia/20220301.simple@4d013bdd32c475c8536aae00a56efc774f061649"
- "tests/test_hf_gcp.py::test_as_dataset_from_hf_gcs"
- "tests/test_hf_gcp.py::test_as_streaming_dataset_from_hf_gcs"
- "tests/test_hub.py::test_convert_to_parquet"
- "tests/test_hub.py::test_delete_from_hub"
- "tests/test_iterable_dataset.py::test_iterable_dataset_from_file"
- "tests/test_iterable_dataset.py::test_iterable_dataset_from_hub_torch_dataloader_parallel"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_bertscore"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_bleurt"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_chrf"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_code_eval"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_competition_math"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_coval"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_cuad"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_f1"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_frugalscore"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_glue"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_google_bleu"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_mauve"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_mean_iou"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_meteor"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_mse"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_precision"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_recall"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_roc_auc"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_rouge"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_sacrebleu"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_sari"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_spearmanr"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_super_glue"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_ter"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_wiki_split"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_xnli"
- "tests/test_metric_common.py::LocalMetricTest::test_load_metric_xtreme_s"
- "tests/test_offline_util.py::test_offline_with_timeout"
- )
- distutils-r1_src_test
-}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-01 6:53 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-01 6:53 UTC (permalink / raw
To: gentoo-commits
commit: 0cf862a5939371201eddbaec870e8e47340d50fe
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Tue Apr 1 06:52:22 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Tue Apr 1 06:52:39 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0cf862a5
sci-ml/datasets: enable py3.13
Closes: https://bugs.gentoo.org/952694
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/datasets-3.0.2.ebuild | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sci-ml/datasets/datasets-3.0.2.ebuild b/sci-ml/datasets/datasets-3.0.2.ebuild
index 051a56ca8680..e6b81257061a 100644
--- a/sci-ml/datasets/datasets-3.0.2.ebuild
+++ b/sci-ml/datasets/datasets-3.0.2.ebuild
@@ -4,7 +4,7 @@
EAPI=8
DISTUTILS_USE_PEP517=setuptools
-PYTHON_COMPAT=( python3_{10..12} )
+PYTHON_COMPAT=( python3_{10..13} )
DISTUTILS_SINGLE_IMPL=1
inherit distutils-r1
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-07 9:55 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-07 9:55 UTC (permalink / raw
To: gentoo-commits
commit: 8694ae7c4374a045cad3722602c9a6955cf02579
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Mon Apr 7 09:55:07 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Mon Apr 7 09:55:34 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=8694ae7c
sci-ml/datasets: drop test that needs network
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/datasets-3.0.2.ebuild | 22 +++++++++++++++++++++-
1 file changed, 21 insertions(+), 1 deletion(-)
diff --git a/sci-ml/datasets/datasets-3.0.2.ebuild b/sci-ml/datasets/datasets-3.0.2.ebuild
index e6b81257061a..0d701b79377a 100644
--- a/sci-ml/datasets/datasets-3.0.2.ebuild
+++ b/sci-ml/datasets/datasets-3.0.2.ebuild
@@ -69,13 +69,18 @@ src_prepare() {
src_test() {
local EPYTEST_IGNORE=(
tests/features/test_audio.py
- tests/test_fingerprint.py
tests/packaged_modules/test_audiofolder.py
tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
)
local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
tests/packaged_modules/test_cache.py::test_cache_multi_configs
tests/packaged_modules/test_cache.py::test_cache_single_config
tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
@@ -83,7 +88,22 @@ src_test() {
tests/test_distributed.py::test_torch_distributed_run
tests/test_file_utils.py::TestxPath::test_xpath_rglob
tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
)
distutils-r1_src_test
}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-08 19:07 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-08 19:07 UTC (permalink / raw
To: gentoo-commits
commit: cc75a28a6d68bc5b6aaa9fe13fb346d6d05f9eb6
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Tue Apr 8 19:07:27 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Tue Apr 8 19:07:27 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=cc75a28a
sci-ml/datasets: add test deps
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/{datasets-3.0.2.ebuild => datasets-3.0.2-r1.ebuild} | 2 ++
1 file changed, 2 insertions(+)
diff --git a/sci-ml/datasets/datasets-3.0.2.ebuild b/sci-ml/datasets/datasets-3.0.2-r1.ebuild
similarity index 97%
rename from sci-ml/datasets/datasets-3.0.2.ebuild
rename to sci-ml/datasets/datasets-3.0.2-r1.ebuild
index 0d701b79377a..4b438c0f0888 100644
--- a/sci-ml/datasets/datasets-3.0.2.ebuild
+++ b/sci-ml/datasets/datasets-3.0.2-r1.ebuild
@@ -44,8 +44,10 @@ BDEPEND="test? (
$(python_gen_cond_dep '
dev-python/absl-py[${PYTHON_USEDEP}]
dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
dev-python/pytest-datadir[${PYTHON_USEDEP}]
dev-python/scikit-learn[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
dev-python/sqlalchemy[${PYTHON_USEDEP}]
dev-python/zstandard[${PYTHON_USEDEP}]
sci-ml/jiwer[${PYTHON_USEDEP}]
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-09 16:58 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-09 16:58 UTC (permalink / raw
To: gentoo-commits
commit: 55dce08b170e3f5212c843d1d0b41b9d59d37202
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Wed Apr 9 16:57:09 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Wed Apr 9 16:58:12 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=55dce08b
sci-ml/datasets: drop test with elasticsearch (not compatible)
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/datasets-3.0.2-r1.ebuild | 1 +
1 file changed, 1 insertion(+)
diff --git a/sci-ml/datasets/datasets-3.0.2-r1.ebuild b/sci-ml/datasets/datasets-3.0.2-r1.ebuild
index 4b438c0f0888..3a4fb6545732 100644
--- a/sci-ml/datasets/datasets-3.0.2-r1.ebuild
+++ b/sci-ml/datasets/datasets-3.0.2-r1.ebuild
@@ -106,6 +106,7 @@ src_test() {
tests/test_file_utils.py::test_xopen_remote
tests/test_hub.py::test_delete_from_hub
tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
)
distutils-r1_src_test
}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-09 17:49 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-09 17:49 UTC (permalink / raw
To: gentoo-commits
commit: c3a24a8b818da060648b7440e82f731bbafe0681
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Wed Apr 9 17:48:53 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Wed Apr 9 17:49:04 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=c3a24a8b
sci-ml/datasets: add 3.1.0
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 +
sci-ml/datasets/datasets-3.1.0.ebuild | 112 ++++++++++++++++++++++++++++++++++
2 files changed, 113 insertions(+)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index 731f0e890079..5b7eec02ccc8 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -1 +1,2 @@
DIST datasets-3.0.2.gh.tar.gz 1867037 BLAKE2B 92475ac566bd825c542dbd3e138c83e1733e6d1137f30a2cfd8f82f7dd54ed78c0566aa766182625e8787413be18f5a0c0c0a289b5823fa9112dcb2876074b7e SHA512 aa55dd2f53df7217c30999f759c76a13d78851233bce4df9b1ad85081e101a7280e905822160874d24d15ea1e1a5ecee1bae9cc66785e89af45ad16dd0339595
+DIST datasets-3.1.0.gh.tar.gz 1877057 BLAKE2B 818cbf10282f775cda5956ae37607bb5357a753229fa91788b19437eb28df609a99c9000f283a8b9b27a920efb3b32f991d729db94a0acf08aaca8cdcd6c4f4b SHA512 7e2a22b6ca423f4358153137f50d963075d37ad01e834c78176dc432f1b3adc3b80df44ac35b22d727db4e1aa8331c938efca1ab2d678e58401e8e7ce571abfc
diff --git a/sci-ml/datasets/datasets-3.1.0.ebuild b/sci-ml/datasets/datasets-3.1.0.ebuild
new file mode 100644
index 000000000000..206da6ebe884
--- /dev/null
+++ b/sci-ml/datasets/datasets-3.1.0.ebuild
@@ -0,0 +1,112 @@
+# Copyright 2023-2025 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( python3_{10..13} )
+DISTUTILS_SINGLE_IMPL=1
+inherit distutils-r1
+
+DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
+HOMEPAGE="https://pypi.org/project/datasets/"
+SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
+ -> ${P}.gh.tar.gz"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~amd64"
+
+RDEPEND="
+ ${PYTHON_DEPS}
+ sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
+ sci-ml/huggingface_hub[${PYTHON_SINGLE_USEDEP}]
+ sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/aiohttp[${PYTHON_USEDEP}]
+ dev-python/dill[${PYTHON_USEDEP}]
+ dev-python/filelock[${PYTHON_USEDEP}]
+ dev-python/fsspec[${PYTHON_USEDEP}]
+ dev-python/multiprocess[${PYTHON_USEDEP}]
+ dev-python/numpy[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ dev-python/pandas[${PYTHON_USEDEP}]
+ dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
+ dev-python/pyyaml[${PYTHON_USEDEP}]
+ dev-python/requests[${PYTHON_USEDEP}]
+ dev-python/tqdm[${PYTHON_USEDEP}]
+ dev-python/xxhash[${PYTHON_USEDEP}]
+ ')
+"
+DEPEND="${RDEPEND}"
+BDEPEND="test? (
+ sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/absl-py[${PYTHON_USEDEP}]
+ dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
+ dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/scikit-learn[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
+ dev-python/sqlalchemy[${PYTHON_USEDEP}]
+ dev-python/zstandard[${PYTHON_USEDEP}]
+ sci-ml/jiwer[${PYTHON_USEDEP}]
+ sci-ml/seqeval[${PYTHON_USEDEP}]
+ ')
+)"
+
+PATCHES=(
+ "${FILESDIR}"/${PN}-3.0.2-tests.patch
+)
+
+distutils_enable_tests pytest
+
+src_prepare() {
+ distutils-r1_src_prepare
+ sed -i -e \
+ "/pyarrow_hotfix/d" \
+ src/datasets/features/features.py || die
+}
+
+src_test() {
+ local EPYTEST_IGNORE=(
+ tests/features/test_audio.py
+ tests/packaged_modules/test_audiofolder.py
+ tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
+ tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
+ )
+
+ local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
+ tests/packaged_modules/test_cache.py::test_cache_multi_configs
+ tests/packaged_modules/test_cache.py::test_cache_single_config
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk
+ tests/test_distributed.py::test_torch_distributed_run
+ tests/test_file_utils.py::TestxPath::test_xpath_rglob
+ tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
+ tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
+ )
+ distutils-r1_src_test
+}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-14 18:12 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-14 18:12 UTC (permalink / raw
To: gentoo-commits
commit: da7700303f28481c4766716f0e6a1de721f5d2f8
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Mon Apr 14 18:12:07 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Mon Apr 14 18:12:22 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=da770030
sci-ml/datasets: add 3.2.0
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 +
sci-ml/datasets/datasets-3.2.0.ebuild | 112 ++++++++++++++++++++++++++++++++++
2 files changed, 113 insertions(+)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index 5b7eec02ccc8..86d90f578b59 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -1,2 +1,3 @@
DIST datasets-3.0.2.gh.tar.gz 1867037 BLAKE2B 92475ac566bd825c542dbd3e138c83e1733e6d1137f30a2cfd8f82f7dd54ed78c0566aa766182625e8787413be18f5a0c0c0a289b5823fa9112dcb2876074b7e SHA512 aa55dd2f53df7217c30999f759c76a13d78851233bce4df9b1ad85081e101a7280e905822160874d24d15ea1e1a5ecee1bae9cc66785e89af45ad16dd0339595
DIST datasets-3.1.0.gh.tar.gz 1877057 BLAKE2B 818cbf10282f775cda5956ae37607bb5357a753229fa91788b19437eb28df609a99c9000f283a8b9b27a920efb3b32f991d729db94a0acf08aaca8cdcd6c4f4b SHA512 7e2a22b6ca423f4358153137f50d963075d37ad01e834c78176dc432f1b3adc3b80df44ac35b22d727db4e1aa8331c938efca1ab2d678e58401e8e7ce571abfc
+DIST datasets-3.2.0.gh.tar.gz 1877261 BLAKE2B 0169ee27e1128f8b64fda0e3a4dd63d5d603874cdbf703cd73f100ad1ea69294c300b93cddaa4f2dd8c795570b76adb34e40475d38730f4733679008fb593d84 SHA512 2e381d6919ea7f2e934e832d6bff30a4b23cfe929c0b4345703ce39c7882daf8a234fe71ca89d1d9bcd75000e35845156f7036f9604504ac584feb1b8d3239b1
diff --git a/sci-ml/datasets/datasets-3.2.0.ebuild b/sci-ml/datasets/datasets-3.2.0.ebuild
new file mode 100644
index 000000000000..206da6ebe884
--- /dev/null
+++ b/sci-ml/datasets/datasets-3.2.0.ebuild
@@ -0,0 +1,112 @@
+# Copyright 2023-2025 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( python3_{10..13} )
+DISTUTILS_SINGLE_IMPL=1
+inherit distutils-r1
+
+DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
+HOMEPAGE="https://pypi.org/project/datasets/"
+SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
+ -> ${P}.gh.tar.gz"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~amd64"
+
+RDEPEND="
+ ${PYTHON_DEPS}
+ sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
+ sci-ml/huggingface_hub[${PYTHON_SINGLE_USEDEP}]
+ sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/aiohttp[${PYTHON_USEDEP}]
+ dev-python/dill[${PYTHON_USEDEP}]
+ dev-python/filelock[${PYTHON_USEDEP}]
+ dev-python/fsspec[${PYTHON_USEDEP}]
+ dev-python/multiprocess[${PYTHON_USEDEP}]
+ dev-python/numpy[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ dev-python/pandas[${PYTHON_USEDEP}]
+ dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
+ dev-python/pyyaml[${PYTHON_USEDEP}]
+ dev-python/requests[${PYTHON_USEDEP}]
+ dev-python/tqdm[${PYTHON_USEDEP}]
+ dev-python/xxhash[${PYTHON_USEDEP}]
+ ')
+"
+DEPEND="${RDEPEND}"
+BDEPEND="test? (
+ sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/absl-py[${PYTHON_USEDEP}]
+ dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
+ dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/scikit-learn[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
+ dev-python/sqlalchemy[${PYTHON_USEDEP}]
+ dev-python/zstandard[${PYTHON_USEDEP}]
+ sci-ml/jiwer[${PYTHON_USEDEP}]
+ sci-ml/seqeval[${PYTHON_USEDEP}]
+ ')
+)"
+
+PATCHES=(
+ "${FILESDIR}"/${PN}-3.0.2-tests.patch
+)
+
+distutils_enable_tests pytest
+
+src_prepare() {
+ distutils-r1_src_prepare
+ sed -i -e \
+ "/pyarrow_hotfix/d" \
+ src/datasets/features/features.py || die
+}
+
+src_test() {
+ local EPYTEST_IGNORE=(
+ tests/features/test_audio.py
+ tests/packaged_modules/test_audiofolder.py
+ tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
+ tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
+ )
+
+ local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
+ tests/packaged_modules/test_cache.py::test_cache_multi_configs
+ tests/packaged_modules/test_cache.py::test_cache_single_config
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk
+ tests/test_distributed.py::test_torch_distributed_run
+ tests/test_file_utils.py::TestxPath::test_xpath_rglob
+ tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
+ tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
+ )
+ distutils-r1_src_test
+}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-15 6:40 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-15 6:40 UTC (permalink / raw
To: gentoo-commits
commit: 12c295157c2ff82f3ca8fde82d1345f24e8abcf2
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Tue Apr 15 06:40:08 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Tue Apr 15 06:40:18 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=12c29515
sci-ml/datasets: add 3.3.2
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 +
sci-ml/datasets/datasets-3.3.2.ebuild | 108 ++++++++++++++++++++++++++++++++++
2 files changed, 109 insertions(+)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index 86d90f578b59..8f5dd482a610 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -1,3 +1,4 @@
DIST datasets-3.0.2.gh.tar.gz 1867037 BLAKE2B 92475ac566bd825c542dbd3e138c83e1733e6d1137f30a2cfd8f82f7dd54ed78c0566aa766182625e8787413be18f5a0c0c0a289b5823fa9112dcb2876074b7e SHA512 aa55dd2f53df7217c30999f759c76a13d78851233bce4df9b1ad85081e101a7280e905822160874d24d15ea1e1a5ecee1bae9cc66785e89af45ad16dd0339595
DIST datasets-3.1.0.gh.tar.gz 1877057 BLAKE2B 818cbf10282f775cda5956ae37607bb5357a753229fa91788b19437eb28df609a99c9000f283a8b9b27a920efb3b32f991d729db94a0acf08aaca8cdcd6c4f4b SHA512 7e2a22b6ca423f4358153137f50d963075d37ad01e834c78176dc432f1b3adc3b80df44ac35b22d727db4e1aa8331c938efca1ab2d678e58401e8e7ce571abfc
DIST datasets-3.2.0.gh.tar.gz 1877261 BLAKE2B 0169ee27e1128f8b64fda0e3a4dd63d5d603874cdbf703cd73f100ad1ea69294c300b93cddaa4f2dd8c795570b76adb34e40475d38730f4733679008fb593d84 SHA512 2e381d6919ea7f2e934e832d6bff30a4b23cfe929c0b4345703ce39c7882daf8a234fe71ca89d1d9bcd75000e35845156f7036f9604504ac584feb1b8d3239b1
+DIST datasets-3.3.2.gh.tar.gz 1886396 BLAKE2B bb51d8e929fbadd9d04f7ca6c8894dc06b88940643f02743b10cea3c2dd5b91b720358d2229a2263e236a6b4911d9dc60e8698da4369393a63f2603274a4f599 SHA512 4c6c4a4e748b5018574c0968929d1dac2aee96bc6f964dedf54c63e872808ca4a4958ed59e0240c958f352ca7a8d333de7ac44095e7c9525e2d9e703afbaeabf
diff --git a/sci-ml/datasets/datasets-3.3.2.ebuild b/sci-ml/datasets/datasets-3.3.2.ebuild
new file mode 100644
index 000000000000..77434c0228ec
--- /dev/null
+++ b/sci-ml/datasets/datasets-3.3.2.ebuild
@@ -0,0 +1,108 @@
+# Copyright 2023-2025 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( python3_{10..13} )
+DISTUTILS_SINGLE_IMPL=1
+inherit distutils-r1
+
+DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
+HOMEPAGE="https://pypi.org/project/datasets/"
+SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
+ -> ${P}.gh.tar.gz"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~amd64"
+
+RDEPEND="
+ ${PYTHON_DEPS}
+ sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
+ sci-ml/huggingface_hub[${PYTHON_SINGLE_USEDEP}]
+ sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/aiohttp[${PYTHON_USEDEP}]
+ dev-python/dill[${PYTHON_USEDEP}]
+ dev-python/filelock[${PYTHON_USEDEP}]
+ dev-python/fsspec[${PYTHON_USEDEP}]
+ dev-python/multiprocess[${PYTHON_USEDEP}]
+ dev-python/numpy[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ dev-python/pandas[${PYTHON_USEDEP}]
+ dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
+ dev-python/pyyaml[${PYTHON_USEDEP}]
+ dev-python/requests[${PYTHON_USEDEP}]
+ dev-python/tqdm[${PYTHON_USEDEP}]
+ dev-python/xxhash[${PYTHON_USEDEP}]
+ ')
+"
+DEPEND="${RDEPEND}"
+BDEPEND="test? (
+ sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/absl-py[${PYTHON_USEDEP}]
+ dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
+ dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/scikit-learn[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
+ dev-python/sqlalchemy[${PYTHON_USEDEP}]
+ dev-python/zstandard[${PYTHON_USEDEP}]
+ sci-ml/jiwer[${PYTHON_USEDEP}]
+ sci-ml/seqeval[${PYTHON_USEDEP}]
+ ')
+)"
+
+distutils_enable_tests pytest
+
+src_prepare() {
+ distutils-r1_src_prepare
+ sed -i -e \
+ "/pyarrow_hotfix/d" \
+ src/datasets/features/features.py || die
+}
+
+src_test() {
+ local EPYTEST_IGNORE=(
+ tests/features/test_audio.py
+ tests/packaged_modules/test_audiofolder.py
+ tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
+ tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
+ )
+
+ local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
+ tests/packaged_modules/test_cache.py::test_cache_multi_configs
+ tests/packaged_modules/test_cache.py::test_cache_single_config
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk
+ tests/test_distributed.py::test_torch_distributed_run
+ tests/test_file_utils.py::TestxPath::test_xpath_rglob
+ tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
+ tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
+ )
+ distutils-r1_src_test
+}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-16 18:29 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-16 18:29 UTC (permalink / raw
To: gentoo-commits
commit: 2ae4216633e94c24652160c047737564077e4a80
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Wed Apr 16 18:28:34 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Wed Apr 16 18:28:44 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=2ae42166
sci-ml/datasets: add 3.4.1
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 +
sci-ml/datasets/datasets-3.4.1.ebuild | 119 ++++++++++++++++++++++++++++++++++
2 files changed, 120 insertions(+)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index 8f5dd482a610..bc908dca6594 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -2,3 +2,4 @@ DIST datasets-3.0.2.gh.tar.gz 1867037 BLAKE2B 92475ac566bd825c542dbd3e138c83e173
DIST datasets-3.1.0.gh.tar.gz 1877057 BLAKE2B 818cbf10282f775cda5956ae37607bb5357a753229fa91788b19437eb28df609a99c9000f283a8b9b27a920efb3b32f991d729db94a0acf08aaca8cdcd6c4f4b SHA512 7e2a22b6ca423f4358153137f50d963075d37ad01e834c78176dc432f1b3adc3b80df44ac35b22d727db4e1aa8331c938efca1ab2d678e58401e8e7ce571abfc
DIST datasets-3.2.0.gh.tar.gz 1877261 BLAKE2B 0169ee27e1128f8b64fda0e3a4dd63d5d603874cdbf703cd73f100ad1ea69294c300b93cddaa4f2dd8c795570b76adb34e40475d38730f4733679008fb593d84 SHA512 2e381d6919ea7f2e934e832d6bff30a4b23cfe929c0b4345703ce39c7882daf8a234fe71ca89d1d9bcd75000e35845156f7036f9604504ac584feb1b8d3239b1
DIST datasets-3.3.2.gh.tar.gz 1886396 BLAKE2B bb51d8e929fbadd9d04f7ca6c8894dc06b88940643f02743b10cea3c2dd5b91b720358d2229a2263e236a6b4911d9dc60e8698da4369393a63f2603274a4f599 SHA512 4c6c4a4e748b5018574c0968929d1dac2aee96bc6f964dedf54c63e872808ca4a4958ed59e0240c958f352ca7a8d333de7ac44095e7c9525e2d9e703afbaeabf
+DIST datasets-3.4.1.gh.tar.gz 1888651 BLAKE2B 836f0e0bb55cefd5a0616785c87c867292cc46ec609934b4f7688462aa7ec95b248f06018cd99e68a88b4e179e0aa48dd601a2fc74bc1b809c336bd70ae0bd6a SHA512 c3654fec241b5da1fb2c9349ff2d0f309e184e5158a0bd19ca0477f1bfb7570d9268e93d1828e1648132563c04ba62ec4eabead2f29d8ef1a29b593aec1c62b1
diff --git a/sci-ml/datasets/datasets-3.4.1.ebuild b/sci-ml/datasets/datasets-3.4.1.ebuild
new file mode 100644
index 000000000000..4bd238519b7b
--- /dev/null
+++ b/sci-ml/datasets/datasets-3.4.1.ebuild
@@ -0,0 +1,119 @@
+# Copyright 2023-2025 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( python3_{11..13} )
+DISTUTILS_SINGLE_IMPL=1
+inherit distutils-r1
+
+DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
+HOMEPAGE="https://pypi.org/project/datasets/"
+SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
+ -> ${P}.gh.tar.gz"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~amd64"
+
+RDEPEND="
+ sci-ml/huggingface_hub[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/aiohttp[${PYTHON_USEDEP}]
+ dev-python/dill[${PYTHON_USEDEP}]
+ dev-python/filelock[${PYTHON_USEDEP}]
+ dev-python/fsspec[${PYTHON_USEDEP}]
+ dev-python/multiprocess[${PYTHON_USEDEP}]
+ dev-python/numpy[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ dev-python/pandas[${PYTHON_USEDEP}]
+ dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
+ dev-python/pyyaml[${PYTHON_USEDEP}]
+ dev-python/requests[${PYTHON_USEDEP}]
+ dev-python/tqdm[${PYTHON_USEDEP}]
+ dev-python/xxhash[${PYTHON_USEDEP}]
+ ')
+"
+DEPEND="${RDEPEND}"
+# Missing x test:
+# elastisearch
+# joblib
+# joblibspark
+# faiss-cpu
+# jax
+# jaxlib
+# polars
+# pyav
+# pyspark
+# py7zr
+# pytest-xdist
+# s3fs
+# tensorflow
+# tiktoken
+# torchdata
+# transformers
+BDEPEND="test? (
+ sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
+ sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
+ sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/absl-py[${PYTHON_USEDEP}]
+ dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
+ dev-python/moto[${PYTHON_USEDEP}]
+ dev-python/protobuf:=[${PYTHON_USEDEP}]
+ dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
+ dev-python/sqlalchemy[${PYTHON_USEDEP}]
+ dev-python/zstandard[${PYTHON_USEDEP}]
+ ')
+)"
+
+distutils_enable_tests pytest
+
+src_test() {
+ local EPYTEST_IGNORE=(
+ tests/features/test_audio.py
+ tests/packaged_modules/test_audiofolder.py
+ tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
+ tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
+ )
+
+ local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/features/test_video.py::test_video_feature_encode_example
+ tests/features/test_video.py::test_dataset_with_video_feature
+ tests/features/test_video.py::test_dataset_with_video_map_and_formatted
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
+ tests/packaged_modules/test_cache.py::test_cache_multi_configs
+ tests/packaged_modules/test_cache.py::test_cache_single_config
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk
+ tests/test_distributed.py::test_torch_distributed_run
+ tests/test_file_utils.py::TestxPath::test_xpath_rglob
+ tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
+ tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
+ )
+ distutils-r1_src_test
+}
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-16 18:33 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-16 18:33 UTC (permalink / raw
To: gentoo-commits
commit: 923307c8dcba5966034a3a3f7e0648c75d9e3c82
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Wed Apr 16 18:33:17 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Wed Apr 16 18:33:17 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=923307c8
sci-ml/datasets: add bdep on elasticsearch
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/datasets-3.4.1.ebuild | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sci-ml/datasets/datasets-3.4.1.ebuild b/sci-ml/datasets/datasets-3.4.1.ebuild
index 4bd238519b7b..a5dd2af2a359 100644
--- a/sci-ml/datasets/datasets-3.4.1.ebuild
+++ b/sci-ml/datasets/datasets-3.4.1.ebuild
@@ -37,7 +37,6 @@ RDEPEND="
"
DEPEND="${RDEPEND}"
# Missing x test:
-# elastisearch
# joblib
# joblibspark
# faiss-cpu
@@ -60,6 +59,7 @@ BDEPEND="test? (
$(python_gen_cond_dep '
dev-python/absl-py[${PYTHON_USEDEP}]
dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/elasticsearch[${PYTHON_USEDEP}]
dev-python/lz4[${PYTHON_USEDEP}]
dev-python/moto[${PYTHON_USEDEP}]
dev-python/protobuf:=[${PYTHON_USEDEP}]
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-16 18:37 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-16 18:37 UTC (permalink / raw
To: gentoo-commits
commit: 41106bc4ae15a10d785d4855b95aa3c52edbedc2
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Wed Apr 16 18:37:08 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Wed Apr 16 18:37:28 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=41106bc4
sci-ml/datasets: add bdep on pytest-xdist
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/datasets-3.4.1.ebuild | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sci-ml/datasets/datasets-3.4.1.ebuild b/sci-ml/datasets/datasets-3.4.1.ebuild
index a5dd2af2a359..7941f2f45941 100644
--- a/sci-ml/datasets/datasets-3.4.1.ebuild
+++ b/sci-ml/datasets/datasets-3.4.1.ebuild
@@ -46,7 +46,6 @@ DEPEND="${RDEPEND}"
# pyav
# pyspark
# py7zr
-# pytest-xdist
# s3fs
# tensorflow
# tiktoken
@@ -64,6 +63,7 @@ BDEPEND="test? (
dev-python/moto[${PYTHON_USEDEP}]
dev-python/protobuf:=[${PYTHON_USEDEP}]
dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/pytest-xdist[${PYTHON_USEDEP}]
dev-python/soundfile[${PYTHON_USEDEP}]
dev-python/sqlalchemy[${PYTHON_USEDEP}]
dev-python/zstandard[${PYTHON_USEDEP}]
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/
@ 2025-04-17 12:14 Alfredo Tupone
0 siblings, 0 replies; 12+ messages in thread
From: Alfredo Tupone @ 2025-04-17 12:14 UTC (permalink / raw
To: gentoo-commits
commit: a6d51389bf37280164889b1379d29cd5186c6c85
Author: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
AuthorDate: Thu Apr 17 12:13:01 2025 +0000
Commit: Alfredo Tupone <tupone <AT> gentoo <DOT> org>
CommitDate: Thu Apr 17 12:13:24 2025 +0000
URL: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=a6d51389
sci-ml/datasets: add 3.5.0
Signed-off-by: Alfredo Tupone <tupone <AT> gentoo.org>
sci-ml/datasets/Manifest | 1 +
sci-ml/datasets/datasets-3.5.0.ebuild | 126 ++++++++++++++++++++++++++++++++++
sci-ml/datasets/metadata.xml | 4 ++
3 files changed, 131 insertions(+)
diff --git a/sci-ml/datasets/Manifest b/sci-ml/datasets/Manifest
index bc908dca6594..450673b4be4f 100644
--- a/sci-ml/datasets/Manifest
+++ b/sci-ml/datasets/Manifest
@@ -3,3 +3,4 @@ DIST datasets-3.1.0.gh.tar.gz 1877057 BLAKE2B 818cbf10282f775cda5956ae37607bb535
DIST datasets-3.2.0.gh.tar.gz 1877261 BLAKE2B 0169ee27e1128f8b64fda0e3a4dd63d5d603874cdbf703cd73f100ad1ea69294c300b93cddaa4f2dd8c795570b76adb34e40475d38730f4733679008fb593d84 SHA512 2e381d6919ea7f2e934e832d6bff30a4b23cfe929c0b4345703ce39c7882daf8a234fe71ca89d1d9bcd75000e35845156f7036f9604504ac584feb1b8d3239b1
DIST datasets-3.3.2.gh.tar.gz 1886396 BLAKE2B bb51d8e929fbadd9d04f7ca6c8894dc06b88940643f02743b10cea3c2dd5b91b720358d2229a2263e236a6b4911d9dc60e8698da4369393a63f2603274a4f599 SHA512 4c6c4a4e748b5018574c0968929d1dac2aee96bc6f964dedf54c63e872808ca4a4958ed59e0240c958f352ca7a8d333de7ac44095e7c9525e2d9e703afbaeabf
DIST datasets-3.4.1.gh.tar.gz 1888651 BLAKE2B 836f0e0bb55cefd5a0616785c87c867292cc46ec609934b4f7688462aa7ec95b248f06018cd99e68a88b4e179e0aa48dd601a2fc74bc1b809c336bd70ae0bd6a SHA512 c3654fec241b5da1fb2c9349ff2d0f309e184e5158a0bd19ca0477f1bfb7570d9268e93d1828e1648132563c04ba62ec4eabead2f29d8ef1a29b593aec1c62b1
+DIST datasets-3.5.0.gh.tar.gz 1906212 BLAKE2B 2c26b907230eb9256a6056878dc80eae3296869f24597298f262d49a5fc0a31dbcd6ebb8dfaca9cf27faece019a4f39ab0dd07db772659d414e1d3f5ecfe2258 SHA512 ca053187b6b80140136991e5b220136cf22f3cd98d073323e0a72d77211248d52986aa49b0e595759ba023b4551a9dc01ab1cac853c855f032acbd9a023ecd5a
diff --git a/sci-ml/datasets/datasets-3.5.0.ebuild b/sci-ml/datasets/datasets-3.5.0.ebuild
new file mode 100644
index 000000000000..98f3234c789b
--- /dev/null
+++ b/sci-ml/datasets/datasets-3.5.0.ebuild
@@ -0,0 +1,126 @@
+# Copyright 2023-2025 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+EAPI=8
+
+DISTUTILS_USE_PEP517=setuptools
+PYTHON_COMPAT=( python3_{11..13} )
+DISTUTILS_SINGLE_IMPL=1
+inherit distutils-r1
+
+DESCRIPTION="Access and share datasets for Audio, Computer Vision, and NLP tasks"
+HOMEPAGE="https://pypi.org/project/datasets/"
+SRC_URI="https://github.com/huggingface/${PN}/archive/refs/tags/${PV}.tar.gz
+ -> ${P}.gh.tar.gz"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~amd64"
+IUSE="torch vision"
+REQUIRES_USE="test? ( torch vision )"
+
+RDEPEND="
+ sci-ml/huggingface_hub[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/aiohttp[${PYTHON_USEDEP}]
+ dev-python/dill[${PYTHON_USEDEP}]
+ dev-python/filelock[${PYTHON_USEDEP}]
+ dev-python/fsspec[${PYTHON_USEDEP}]
+ dev-python/multiprocess[${PYTHON_USEDEP}]
+ dev-python/numpy[${PYTHON_USEDEP}]
+ dev-python/packaging[${PYTHON_USEDEP}]
+ dev-python/pandas[${PYTHON_USEDEP}]
+ dev-python/pyarrow[${PYTHON_USEDEP},parquet,snappy]
+ dev-python/pyyaml[${PYTHON_USEDEP}]
+ dev-python/requests[${PYTHON_USEDEP}]
+ dev-python/tqdm[${PYTHON_USEDEP}]
+ dev-python/xxhash[${PYTHON_USEDEP}]
+ vision? (
+ dev-python/pillow[${PYTHON_USEDEP}]
+ )
+ ')
+ torch? (
+ sci-ml/caffe2[${PYTHON_SINGLE_USEDEP},numpy]
+ sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
+ )
+"
+DEPEND="${RDEPEND}"
+# Missing x test:
+# joblib
+# joblibspark
+# faiss-cpu
+# jax
+# jaxlib
+# polars
+# pyav
+# pyspark
+# py7zr
+# s3fs
+# tensorflow
+# tiktoken
+# torchdata
+# transformers
+BDEPEND="test? (
+ sci-ml/torchvision[${PYTHON_SINGLE_USEDEP}]
+ $(python_gen_cond_dep '
+ dev-python/absl-py[${PYTHON_USEDEP}]
+ dev-python/decorator[${PYTHON_USEDEP}]
+ dev-python/elasticsearch[${PYTHON_USEDEP}]
+ dev-python/lz4[${PYTHON_USEDEP}]
+ dev-python/moto[${PYTHON_USEDEP}]
+ dev-python/protobuf:=[${PYTHON_USEDEP}]
+ dev-python/pytest-datadir[${PYTHON_USEDEP}]
+ dev-python/pytest-xdist[${PYTHON_USEDEP}]
+ dev-python/soundfile[${PYTHON_USEDEP}]
+ dev-python/sqlalchemy[${PYTHON_USEDEP}]
+ dev-python/zstandard[${PYTHON_USEDEP}]
+ ')
+)"
+
+distutils_enable_tests pytest
+
+src_test() {
+ local EPYTEST_IGNORE=(
+ tests/features/test_audio.py
+ tests/packaged_modules/test_audiofolder.py
+ tests/packaged_modules/test_spark.py
+ tests/test_fingerprint.py
+ tests/test_iterable_dataset.py
+ tests/test_inspect.py
+ tests/test_load.py
+ tests/test_upstream_hub.py
+ )
+
+ local EPYTEST_DESELECT=(
+ tests/commands/test_test.py::test_test_command
+ tests/features/test_video.py::test_video_feature_encode_example
+ tests/features/test_video.py::test_dataset_with_video_feature
+ tests/features/test_video.py::test_dataset_with_video_map_and_formatted
+ tests/io/test_parquet.py::test_parquet_read_geoparquet
+ tests/packaged_modules/test_cache.py::test_cache_multi_configs
+ tests/packaged_modules/test_cache.py::test_cache_single_config
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_filter_caching_on_disk
+ tests/test_arrow_dataset.py::BaseDatasetTest::test_map_caching_on_disk
+ tests/test_distributed.py::test_torch_distributed_run
+ tests/test_file_utils.py::TestxPath::test_xpath_rglob
+ tests/test_file_utils.py::TestxPath::test_xpath_glob
+ tests/test_file_utils.py::test_xexists_private
+ tests/test_file_utils.py::test_xlistdir_private
+ tests/test_file_utils.py::test_xisdir_private
+ tests/test_file_utils.py::test_xisfile_private
+ tests/test_file_utils.py::test_xgetsize_private
+ tests/test_file_utils.py::test_xglob_private
+ tests/test_file_utils.py::test_xwalk_private
+ tests/test_hub.py::test_convert_to_parquet
+ tests/packaged_modules/test_cache.py::test_cache_capital_letters
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_different_levels_no_metadata
+ tests/packaged_modules/test_folder_based_builder.py::test_data_files_with_one_label_no_metadata
+ tests/test_data_files.py::test_DataFilesList_from_patterns_locally_with_extra_files
+ tests/test_data_files.py::test_DataFilesDict_from_patterns_locally_or_remote_hashing
+ tests/test_file_utils.py::test_xopen_remote
+ tests/test_hub.py::test_delete_from_hub
+ tests/test_offline_util.py::test_offline_with_timeout
+ tests/test_search.py::ElasticSearchIndexTest::test_elasticsearch
+ )
+ distutils-r1_src_test
+}
diff --git a/sci-ml/datasets/metadata.xml b/sci-ml/datasets/metadata.xml
index 94c112402049..f7e5d145210a 100644
--- a/sci-ml/datasets/metadata.xml
+++ b/sci-ml/datasets/metadata.xml
@@ -5,6 +5,10 @@
<email>tupone@gentoo.org</email>
<name>Tupone Alfredo</name>
</maintainer>
+ <use>
+ <flag name="torch">Use pytorch</flag>
+ <flag name="vision">Support vision</flag>
+ </use>
<upstream>
<remote-id type="github">huggingface/datasets</remote-id>
<remote-id type="pypi">datasets</remote-id>
^ permalink raw reply related [flat|nested] 12+ messages in thread
end of thread, other threads:[~2025-04-17 12:14 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-03-28 17:51 [gentoo-commits] repo/gentoo:master commit in: sci-ml/datasets/ Alfredo Tupone
-- strict thread matches above, loose matches on Subject: below --
2025-04-01 6:53 Alfredo Tupone
2025-04-07 9:55 Alfredo Tupone
2025-04-08 19:07 Alfredo Tupone
2025-04-09 16:58 Alfredo Tupone
2025-04-09 17:49 Alfredo Tupone
2025-04-14 18:12 Alfredo Tupone
2025-04-15 6:40 Alfredo Tupone
2025-04-16 18:29 Alfredo Tupone
2025-04-16 18:33 Alfredo Tupone
2025-04-16 18:37 Alfredo Tupone
2025-04-17 12:14 Alfredo Tupone
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox