docling: init at 2.17.0 (#359783)

pkgs/by-name/do/docling/package.nix

@@ -0,0 +1,3 @@
+{ python3Packages }:
+
+python3Packages.toPythonApplication python3Packages.docling

pkgs/development/python-modules/docling-core/default.nix

@@ -3,12 +3,18 @@
   buildPythonPackage,
   fetchFromGitHub,
   poetry-core,
+  # dependencies
   jsonref,
   jsonschema,
   pandas,
   pillow,
   pydantic,
   tabulate,
+  pyyaml,
+  typing-extensions,
+  transformers,
+  typer,
+  latex2mathml,
   jsondiff,
   requests,
   pytestCheckHook,
@@ -16,14 +22,14 @@
 
 buildPythonPackage rec {
   pname = "docling-core";
-  version = "2.3.2";
+  version = "2.16.1";
   pyproject = true;
 
   src = fetchFromGitHub {
     owner = "DS4SD";
     repo = "docling-core";
     tag = "v${version}";
-    hash = "sha256-N8rL+5bCVF4Qi5eqgkaB2r3LTYoqTVPeK4gQ6stiW/w=";
+    hash = "sha256-oW/jX9IHCpztc0FDm8/3OzDmOxM92jrkFq/JeAcI9ZA=";
   };
 
   build-system = [
@@ -31,12 +37,18 @@ buildPythonPackage rec {
   ];
 
   dependencies = [
-    jsonref
     jsonschema
+    pydantic
+    jsonref
+    tabulate
     pandas
     pillow
-    pydantic
-    tabulate
+    pyyaml
+    typing-extensions
+    transformers
+    # semchunk
+    typer
+    latex2mathml
   ];
 
   pythonRelaxDeps = [
@@ -47,6 +59,8 @@ buildPythonPackage rec {
     "docling_core"
   ];
 
+  doCheck = false;
+
   nativeCheckInputs = [
     jsondiff
     pytestCheckHook

pkgs/development/python-modules/docling-ibm-models/default.nix

@@ -3,28 +3,30 @@
   buildPythonPackage,
   fetchFromGitHub,
   poetry-core,
+  # dependencies
+  torch,
+  torchvision,
+  transformers,
   huggingface-hub,
   jsonlines,
-  mean-average-precision,
   numpy,
   opencv-python-headless,
   pillow,
-  torch,
-  torchvision,
   tqdm,
+  safetensors,
   pytestCheckHook,
 }:
 
 buildPythonPackage rec {
   pname = "docling-ibm-models";
-  version = "2.0.4";
+  version = "3.3.0";
   pyproject = true;
 
   src = fetchFromGitHub {
     owner = "DS4SD";
     repo = "docling-ibm-models";
     tag = "v${version}";
-    hash = "sha256-QZvkkazxgkGuSQKIYI+YghH7pLlDSEbCGhg89gZsOpk=";
+    hash = "sha256-wxkHd+TCBibOTWO09JOsjX6oBtUxZ/9IOmyLdeptzeQ=";
   };
 
   build-system = [
@@ -32,21 +34,23 @@ buildPythonPackage rec {
   ];
 
   dependencies = [
-    huggingface-hub
-    jsonlines
-    mean-average-precision
-    numpy
-    opencv-python-headless
-    pillow
     torch
     torchvision
+    transformers
+    numpy
+    jsonlines
+    pillow
     tqdm
+    opencv-python-headless
+    huggingface-hub
+    safetensors
   ];
 
   pythonRelaxDeps = [
-    "mean_average_precision"
     "pillow"
     "torchvision"
+    "transformers"
+    "numpy"
   ];
 
   pythonImportsCheck = [
@@ -57,10 +61,16 @@ buildPythonPackage rec {
     pytestCheckHook
   ];
 
+  preCheck = ''
+    export HOME="$TEMPDIR"
+  '';
+
   disabledTests = [
     # Requires network access
     "test_layoutpredictor"
     "test_tf_predictor"
+    "test_code_formula_predictor" # huggingface_hub.errors.LocalEntryNotFoundError
+    "test_figure_classifier" # huggingface_hub.errors.LocalEntryNotFoundError
   ];
 
   meta = {

pkgs/development/python-modules/docling-parse/default.nix

@@ -7,26 +7,30 @@
   cxxopts,
   poetry-core,
   pybind11,
-  tabulate,
   zlib,
   nlohmann_json,
   utf8cpp,
   libjpeg,
   qpdf,
   loguru-cpp,
+  # python dependencies
+  tabulate,
+  pillow,
+  pydantic,
+  docling-core,
   pytestCheckHook,
 }:
 
 buildPythonPackage rec {
   pname = "docling-parse";
-  version = "2.0.3";
+  version = "3.1.2";
   pyproject = true;
 
   src = fetchFromGitHub {
     owner = "DS4SD";
     repo = "docling-parse";
     tag = "v${version}";
-    hash = "sha256-pZJ7lneg4ftAoWS5AOflkkKCwZGF4TJIuqDjq4W4VBw=";
+    hash = "sha256-SgVLk1kruUSjtzuo/5YFY4Keha8zMzovm/UeCtfGaNY=";
   };
 
   dontUseCmakeConfigure = true;
@@ -61,6 +65,14 @@ buildPythonPackage rec {
 
   dependencies = [
     tabulate
+    pillow
+    pydantic
+    docling-core
+  ];
+
+  pythonRelaxDeps = [
+    "pydantic"
+    "pillow"
   ];
 
   pythonImportsCheck = [

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

@@ -0,0 +1,150 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  # dependencies
+  pydantic,
+  docling-core,
+  docling-ibm-models,
+  deepsearch-glm,
+  docling-parse,
+  filetype,
+  pypdfium2,
+  pydantic-settings,
+  huggingface-hub,
+  requests,
+  easyocr,
+  tesserocr,
+  certifi,
+  rtree,
+  scipy,
+  typer,
+  python-docx,
+  python-pptx,
+  beautifulsoup4,
+  pandas,
+  marko,
+  openpyxl,
+  lxml,
+  # ocrmac # not yet packaged
+  rapidocr-onnxruntime,
+  onnxruntime,
+  pillow,
+  pyarrow,
+  # build system
+  poetry-core,
+  # optional dependencies
+  mkdocs-material,
+  mkdocs-jupyter,
+  # mkdocs-click # not yet packaged
+  mkdocstrings,
+  # native check inputs
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "docling";
+  version = "2.17.0";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "DS4SD";
+    repo = "docling";
+    tag = "v${version}";
+    hash = "sha256-OtUFQRNqyTGT1Z41tHziwM5hqbk+tg/97bxhtPVtmN0=";
+  };
+
+  build-system = [
+    poetry-core
+  ];
+
+  dependencies = [
+    pydantic
+    docling-core
+    docling-ibm-models
+    deepsearch-glm
+    docling-parse
+    filetype
+    pypdfium2
+    pydantic-settings
+    huggingface-hub
+    requests
+    easyocr
+    tesserocr
+    certifi
+    rtree
+    scipy
+    typer
+    python-docx
+    python-pptx
+    beautifulsoup4
+    pandas
+    marko
+    openpyxl
+    lxml
+    # ocrmac # not yet packaged
+    rapidocr-onnxruntime
+    onnxruntime
+    pillow
+    pyarrow
+  ];
+
+  pythonRelaxDeps = [
+    "pillow"
+  ];
+
+  optional-dependencies = {
+    ocrmac = [
+      # ocrmac # not yet packaged
+    ];
+    rapidocr = [
+      onnxruntime
+      rapidocr-onnxruntime
+    ];
+    tesserocr = [
+      tesserocr
+    ];
+
+    docs = [
+      mkdocs-material
+      mkdocs-jupyter
+      # mkdocs-click # not yet packaged
+      mkdocstrings
+      # griffle-pydantic
+    ];
+  };
+
+  preCheck = ''
+    export HOME="$TEMPDIR"
+  '';
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  pythonImportsCheck = [
+    "docling"
+  ];
+
+  disabledTests = [
+    "test_e2e_pdfs_conversions" # AssertionError: ## TableFormer: Table Structure Understanding with Transf
+    "test_e2e_conversions" # RuntimeError: Tesseract is not available
+    # huggingface_hub.errors.LocalEntryNotFoundError: An error happened
+    "test_cli_convert"
+    "test_code_and_formula_conversion"
+    "test_picture_classifier"
+    "test_convert_path"
+    "test_convert_stream"
+    "test_compare_legacy_output"
+    "test_ocr_coverage_threshold"
+  ];
+
+  meta = {
+    description = "Get your documents ready for gen AI";
+    homepage = "https://github.com/DS4SD/docling";
+    changelog = "https://github.com/DS4SD/docling/blob/${src.rev}/CHANGELOG.md";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ happysalada ];
+    mainProgram = "docling";
+  };
+}

pkgs/top-level/python-packages.nix

@@ -3804,6 +3804,8 @@ self: super: with self; {
 
   dockerspawner = callPackage ../development/python-modules/dockerspawner { };
 
+  docling = callPackage ../development/python-modules/docling { };
+
   docling-core = callPackage ../development/python-modules/docling-core { };
 
   docling-ibm-models = callPackage ../development/python-modules/docling-ibm-models { };