pkgs/development/python-modules/tensorflow-datasets/default.nix

   1 { apache-beam
   2 , attrs
   3 , beautifulsoup4
   4 , buildPythonPackage
   5 , dill
   6 , dm-tree
   7 , fetchFromGitHub
   8 , ffmpeg
   9 , future
  10 , imagemagick
  11 , importlib-resources
  12 , jax
  13 , jaxlib
  14 , jinja2
  15 , langdetect
  16 , lib
  17 , matplotlib
  18 , mwparserfromhell
  19 , networkx
  20 , nltk
  21 , numpy
  22 , opencv4
  23 , pandas
  24 , pillow
  25 , promise
  26 , protobuf
  27 , pycocotools
  28 , pydub
  29 , pytest-xdist
  30 , pytestCheckHook
  31 , requests
  32 , scikitimage
  33 , scipy
  34 , six
  35 , tensorflow
  36 , tensorflow-metadata
  37 , termcolor
  38 , tifffile
  39 , tqdm
  40 , zarr
  41 }:
  42
  43 buildPythonPackage rec {
  44   pname = "tensorflow-datasets";
  45   version = "4.6.0";
  46
  47   src = fetchFromGitHub {
  48     owner = "tensorflow";
  49     repo = "datasets";
  50     rev = "refs/tags/v${version}";
  51     sha256 = "sha256-z52UZz9d1AaZklLOPbWuzByEl1hJ6ra4Hoz6eNGD+hg=";
  52   };
  53
  54   patches = [
  55     # addresses https://github.com/tensorflow/datasets/issues/3673
  56     ./corruptions.patch
  57   ];
  58
  59   propagatedBuildInputs = [
  60     attrs
  61     dill
  62     dm-tree
  63     future
  64     importlib-resources
  65     numpy
  66     promise
  67     protobuf
  68     requests
  69     six
  70     tensorflow-metadata
  71     termcolor
  72     tqdm
  73   ];
  74
  75   pythonImportsCheck = [
  76     "tensorflow_datasets"
  77   ];
  78
  79   checkInputs = [
  80     apache-beam
  81     beautifulsoup4
  82     ffmpeg
  83     imagemagick
  84     jax
  85     jaxlib
  86     jinja2
  87     langdetect
  88     matplotlib
  89     mwparserfromhell
  90     networkx
  91     nltk
  92     opencv4
  93     pandas
  94     pillow
  95     pycocotools
  96     pydub
  97     pytest-xdist
  98     pytestCheckHook
  99     scikitimage
 100     scipy
 101     tensorflow
 102     tifffile
 103     zarr
 104   ];
 105
 106   disabledTestPaths = [
 107     # Sandbox violations: network access, filesystem write attempts outside of build dir, ...
 108     "tensorflow_datasets/core/dataset_builder_test.py"
 109     "tensorflow_datasets/core/dataset_info_test.py"
 110     "tensorflow_datasets/core/features/features_test.py"
 111     "tensorflow_datasets/core/github_api/github_path_test.py"
 112     "tensorflow_datasets/core/utils/gcs_utils_test.py"
 113     "tensorflow_datasets/scripts/cli/build_test.py"
 114
 115     # Requires `pretty_midi` which is not packaged in `nixpkgs`.
 116     "tensorflow_datasets/audio/groove_test.py"
 117
 118     # Requires `crepe` which is not packaged in `nixpkgs`.
 119     "tensorflow_datasets/audio/nsynth_test.py"
 120
 121     # Requires `gcld3` and `pretty_midi` which are not packaged in `nixpkgs`.
 122     "tensorflow_datasets/core/lazy_imports_lib_test.py"
 123
 124     # Requires `tensorflow_io` which is not packaged in `nixpkgs`.
 125     "tensorflow_datasets/image/lsun_test.py"
 126
 127     # Requires `envlogger` which is not packaged in `nixpkgs`.
 128     "tensorflow_datasets/rlds/locomotion/locomotion_test.py"
 129     "tensorflow_datasets/rlds/robosuite_panda_pick_place_can/robosuite_panda_pick_place_can_test.py"
 130
 131     # Fails with `TypeError: Constant constructor takes either 0 or 2 positional arguments`
 132     # deep in TF AutoGraph. Doesn't reproduce in Docker with Ubuntu 22.04 => might be related
 133     # to the differences in some of the dependencies?
 134     "tensorflow_datasets/rl_unplugged/rlu_atari/rlu_atari_test.py"
 135
 136     # Requires `tensorflow_docs` which is not packaged in `nixpkgs` and the test is for documentation anyway.
 137     "tensorflow_datasets/scripts/documentation/build_api_docs_test.py"
 138
 139     # Not a test, should not be executed.
 140     "tensorflow_datasets/testing/test_utils.py"
 141
 142     # Require `gcld3` and `nltk.punkt` which are not packaged in `nixpkgs`.
 143     "tensorflow_datasets/text/c4_test.py"
 144     "tensorflow_datasets/text/c4_utils_test.py"
 145   ];
 146
 147   meta = with lib; {
 148     description = "Library of datasets ready to use with TensorFlow";
 149     homepage = "https://www.tensorflow.org/datasets/overview";
 150     license = licenses.asl20;
 151     maintainers = with maintainers; [ ndl ];
 152   };
 153 }