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

   1 {
   2   lib,
   3   fetchFromGitHub,
   4
   5   # build-system
   6   setuptools,
   7
   8   # dependencies
   9   alembic,
  10   buildPythonPackage,
  11   cachetools,
  12   click,
  13   cloudpickle,
  14   databricks-sdk,
  15   docker,
  16   flask,
  17   gitpython,
  18   graphene,
  19   gunicorn,
  20   importlib-metadata,
  21   jinja2,
  22   markdown,
  23   matplotlib,
  24   numpy,
  25   opentelemetry-api,
  26   opentelemetry-sdk,
  27   packaging,
  28   pandas,
  29   protobuf,
  30   pyarrow,
  31   pyyaml,
  32   requests,
  33   scikit-learn,
  34   scipy,
  35   sqlalchemy,
  36   sqlparse,
  37
  38   # tests
  39   aiohttp,
  40   azure-core,
  41   azure-storage-blob,
  42   azure-storage-file,
  43   boto3,
  44   botocore,
  45   catboost,
  46   datasets,
  47   fastapi,
  48   google-cloud-storage,
  49   httpx,
  50   jwt,
  51   keras,
  52   langchain,
  53   librosa,
  54   moto,
  55   opentelemetry-exporter-otlp,
  56   optuna,
  57   pydantic,
  58   pyspark,
  59   pytestCheckHook,
  60   pytorch-lightning,
  61   sentence-transformers,
  62   starlette,
  63   statsmodels,
  64   tensorflow,
  65   torch,
  66   transformers,
  67   uvicorn,
  68   xgboost,
  69 }:
  70
  71 buildPythonPackage rec {
  72   pname = "mlflow";
  73   version = "2.16.2";
  74   pyproject = true;
  75
  76   src = fetchFromGitHub {
  77     owner = "mlflow";
  78     repo = "mlflow";
  79     rev = "refs/tags/v${version}";
  80     hash = "sha256-7W1gpVgJSN/iXoW987eCHfcOeE3D/ZJ2W/eilDdzOww=";
  81   };
  82
  83   build-system = [
  84     setuptools
  85   ];
  86
  87   dependencies = [
  88     alembic
  89     cachetools
  90     click
  91     cloudpickle
  92     databricks-sdk
  93     docker
  94     flask
  95     gitpython
  96     graphene
  97     gunicorn
  98     importlib-metadata
  99     jinja2
 100     markdown
 101     matplotlib
 102     numpy
 103     opentelemetry-api
 104     opentelemetry-sdk
 105     packaging
 106     pandas
 107     protobuf
 108     pyarrow
 109     pyyaml
 110     requests
 111     scikit-learn
 112     scipy
 113     sqlalchemy
 114     sqlparse
 115   ];
 116
 117   pythonImportsCheck = [ "mlflow" ];
 118
 119   nativeCheckInputs = [
 120     aiohttp
 121     azure-core
 122     azure-storage-blob
 123     azure-storage-file
 124     boto3
 125     botocore
 126     catboost
 127     datasets
 128     fastapi
 129     google-cloud-storage
 130     httpx
 131     jwt
 132     keras
 133     langchain
 134     librosa
 135     moto
 136     opentelemetry-exporter-otlp
 137     optuna
 138     pydantic
 139     pyspark
 140     pytestCheckHook
 141     pytorch-lightning
 142     sentence-transformers
 143     starlette
 144     statsmodels
 145     tensorflow
 146     torch
 147     transformers
 148     uvicorn
 149     xgboost
 150   ];
 151
 152   disabledTestPaths = [
 153     # Requires unpackaged `autogen`
 154     "tests/autogen/test_autogen_autolog.py"
 155
 156     # Requires unpackaged `diviner`
 157     "tests/diviner/test_diviner_model_export.py"
 158
 159     # Requires unpackaged `sktime`
 160     "examples/sktime/test_sktime_model_export.py"
 161
 162     # Requires `fastai` which would cause a circular dependency
 163     "tests/fastai/test_fastai_autolog.py"
 164     "tests/fastai/test_fastai_model_export.py"
 165
 166     # Requires `spacy` which would cause a circular dependency
 167     "tests/spacy/test_spacy_model_export.py"
 168
 169     # Requires `tensorflow.keras` which is not included in our outdated version of `tensorflow` (2.13.0)
 170     "tests/gateway/providers/test_ai21labs.py"
 171     "tests/tensorflow/test_keras_model_export.py"
 172     "tests/tensorflow/test_keras_pyfunc_model_works_with_all_input_types.py"
 173     "tests/tensorflow/test_mlflow_callback.py"
 174   ];
 175
 176   # I (@GaetanLepage) gave up at enabling tests:
 177   # - They require a lot of dependencies (some unpackaged);
 178   # - Many errors occur at collection time;
 179   # - Most (all ?) tests require internet access anyway.
 180   doCheck = false;
 181
 182   meta = {
 183     description = "Open source platform for the machine learning lifecycle";
 184     mainProgram = "mlflow";
 185     homepage = "https://github.com/mlflow/mlflow";
 186     changelog = "https://github.com/mlflow/mlflow/blob/v${version}/CHANGELOG.md";
 187     license = lib.licenses.asl20;
 188     maintainers = with lib.maintainers; [ tbenst ];
 189   };
 190 }