pkgs/development/python-modules/snorkel/default.nix

   1 {
   2   lib,
   3   buildPythonPackage,
   4   fetchFromGitHub,
   5   pythonOlder,
   6   tensorboard,
   7   scipy,
   8   tqdm,
   9   scikit-learn,
  10   munkres,
  11   networkx,
  12   torch,
  13   pandas,
  14   # test dependencies
  15   pytestCheckHook,
  16   spacy,
  17   pyspark,
  18   dill,
  19   dask,
  20   spacy-models,
  21 }:
  22 let
  23   pname = "snorkel";
  24   version = "0.10.0";
  25 in
  26 buildPythonPackage {
  27   inherit pname version;
  28   format = "setuptools";
  29
  30   disabled = pythonOlder "3.8";
  31
  32   src = fetchFromGitHub {
  33     owner = "snorkel-team";
  34     repo = pname;
  35     rev = "refs/tags/v${version}";
  36     hash = "sha256-1DgkMHYToiI3266yCND1bXiui80x8AaBttxM83kJImw=";
  37   };
  38
  39   propagatedBuildInputs = [
  40     tensorboard
  41     scipy
  42     tqdm
  43     scikit-learn
  44     munkres
  45     networkx
  46     torch
  47     pandas
  48   ];
  49
  50   # test fail because of some wrong path references
  51   # things like No such file or directory: '/nix/store/8r9x7xv9nfwmd36ca28a39xaharcjdzj-python3.10-pyspark-3.4.0/lib/python3.10/site-packages/pyspark/./bin/spark-submit'
  52   doCheck = false;
  53
  54   nativeCheckInputs = [
  55     pytestCheckHook
  56     spacy
  57     pyspark
  58     dill
  59     dask
  60     spacy-models.en_core_web_sm
  61   ] ++ dask.optional-dependencies.distributed;
  62
  63   meta = with lib; {
  64     description = "System for quickly generating training data with weak supervision";
  65     homepage = "https://github.com/snorkel-team/snorkel";
  66     changelog = "https://github.com/snorkel/snorkel/releases/tag/v${version}";
  67     license = licenses.asl20;
  68     maintainers = with maintainers; [ happysalada ];
  69   };
  70 }