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

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