pkgs/development/python-modules/unstructured-inference/default.nix

   1 {
   2   lib,
   3   buildPythonPackage,
   4   fetchFromGitHub,
   5   # runtime dependencies
   6   layoutparser,
   7   python-multipart,
   8   huggingface-hub,
   9   opencv-python,
  10   onnxruntime,
  11   transformers,
  12   detectron2,
  13   paddleocr,
  14   # check inputs
  15   pytestCheckHook,
  16   coverage,
  17   click,
  18   httpx,
  19   mypy,
  20   pytest-cov-stub,
  21   pdf2image,
  22 }:
  23
  24 buildPythonPackage rec {
  25   pname = "unstructured-inference";
  26   version = "0.7.37";
  27   format = "setuptools";
  28
  29   src = fetchFromGitHub {
  30     owner = "Unstructured-IO";
  31     repo = "unstructured-inference";
  32     rev = "refs/tags/${version}";
  33     hash = "sha256-2k7gFlBUevVnz2A5pvUE4eIGmXwEr5s4F8BbX6j5lzc=";
  34   };
  35
  36   propagatedBuildInputs =
  37     [
  38       layoutparser
  39       python-multipart
  40       huggingface-hub
  41       opencv-python
  42       onnxruntime
  43       transformers
  44       detectron2
  45       paddleocr
  46       # yolox
  47     ]
  48     ++ layoutparser.optional-dependencies.layoutmodels ++ layoutparser.optional-dependencies.tesseract;
  49
  50   nativeCheckInputs = [
  51     pytestCheckHook
  52     coverage
  53     click
  54     httpx
  55     mypy
  56     pytest-cov-stub
  57     pdf2image
  58     huggingface-hub
  59   ];
  60
  61   preCheck = ''
  62     export HOME=$(mktemp -d)
  63   '';
  64
  65   disabledTests = [
  66     # not sure why this fails
  67     "test_get_path_oob_move_deeply_nested"
  68     "test_get_path_oob_move_nested[False]"
  69     # requires yolox
  70     "test_yolox"
  71   ];
  72
  73   disabledTestPaths = [
  74     # network access
  75     "test_unstructured_inference/inference/test_layout.py"
  76     "test_unstructured_inference/models/test_chippermodel.py"
  77     "test_unstructured_inference/models/test_detectron2.py"
  78     "test_unstructured_inference/models/test_detectron2onnx.py"
  79     # unclear failure
  80     "test_unstructured_inference/models/test_donut.py"
  81     "test_unstructured_inference/models/test_model.py"
  82     "test_unstructured_inference/models/test_tables.py"
  83   ];
  84
  85   pythonImportsCheck = [ "unstructured_inference" ];
  86
  87   meta = with lib; {
  88     description = "hosted model inference code for layout parsing models";
  89     homepage = "https://github.com/Unstructured-IO/unstructured-inference";
  90     changelog = "https://github.com/Unstructured-IO/unstructured-inference/blob/${src.rev}/CHANGELOG.md";
  91     license = licenses.asl20;
  92     maintainers = with maintainers; [ happysalada ];
  93     platforms = [
  94       "x86_64-linux"
  95       "x86_64-darwin"
  96       "aarch64-darwin"
  97     ];
  98   };
  99 }