python3Packages.datasets: 3.2.0 -> 3.4.1 (#391372)

This commit is contained in:
Peder Bergebakken Sundt 2025-03-21 08:17:12 +01:00 committed by GitHub
commit fa6ab1d7fd
No known key found for this signature in database
GPG key ID: B5690EEEBB952194

View file

@ -15,31 +15,27 @@
pythonOlder, pythonOlder,
requests, requests,
responses, responses,
setuptools,
tqdm, tqdm,
xxhash, xxhash,
}: }:
buildPythonPackage rec { buildPythonPackage rec {
pname = "datasets"; pname = "datasets";
version = "3.2.0"; version = "3.4.1";
format = "setuptools"; pyproject = true;
disabled = pythonOlder "3.8";
src = fetchFromGitHub { src = fetchFromGitHub {
owner = "huggingface"; owner = "huggingface";
repo = pname; repo = "datasets";
tag = version; tag = version;
hash = "sha256-3Q4tNLA9qUb7XdxP1NftYDcVUgq5ol9OZfklhmadk5I="; hash = "sha256-a0c5E4N1X+PtO4+UZn8l1JcLGTNpLPyfEkrrxNsjfLA=";
}; };
# remove pyarrow<14.0.1 vulnerability fix build-system = [
postPatch = '' setuptools
substituteInPlace src/datasets/features/features.py \ ];
--replace "import pyarrow_hotfix" "#import pyarrow_hotfix"
'';
propagatedBuildInputs = [ dependencies = [
aiohttp aiohttp
dill dill
fsspec fsspec
@ -53,7 +49,17 @@ buildPythonPackage rec {
responses responses
tqdm tqdm
xxhash xxhash
] ++ lib.optionals (pythonOlder "3.8") [ importlib-metadata ]; ];
pythonRelaxDeps = [
# https://github.com/huggingface/datasets/blob/a256b85cbc67aa3f0e75d32d6586afc507cf535b/setup.py#L117
# "pin until dill has official support for determinism"
"dill"
"multiprocess"
# https://github.com/huggingface/datasets/blob/a256b85cbc67aa3f0e75d32d6586afc507cf535b/setup.py#L129
# "to support protocol=kwargs in fsspec's `open`, `get_fs_token_paths`"
"fsspec"
];
# Tests require pervasive internet access # Tests require pervasive internet access
doCheck = false; doCheck = false;
@ -63,13 +69,12 @@ buildPythonPackage rec {
pythonImportsCheck = [ "datasets" ]; pythonImportsCheck = [ "datasets" ];
meta = with lib; { meta = {
description = "Open-access datasets and evaluation metrics for natural language processing"; description = "Open-access datasets and evaluation metrics for natural language processing";
mainProgram = "datasets-cli"; mainProgram = "datasets-cli";
homepage = "https://github.com/huggingface/datasets"; homepage = "https://github.com/huggingface/datasets";
changelog = "https://github.com/huggingface/datasets/releases/tag/${src.tag}"; changelog = "https://github.com/huggingface/datasets/releases/tag/${src.tag}";
license = licenses.asl20; license = lib.licenses.asl20;
platforms = platforms.unix; maintainers = with lib.maintainers; [ osbm ];
maintainers = [ ];
}; };
} }