python312Packages.docling-core: add semchunk dependency (#379628)

Author: happysalada

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

@@ -11,6 +11,7 @@
   pydantic,
   tabulate,
   pyyaml,
+  semchunk,
   typing-extensions,
   transformers,
   typer,
@@ -46,7 +47,7 @@ buildPythonPackage rec {
     pyyaml
     typing-extensions
     transformers
-    # semchunk
+    semchunk
     typer
     latex2mathml
   ];
@@ -60,14 +61,17 @@ buildPythonPackage rec {
     "docling_core"
   ];
 
-  doCheck = false;
-
   nativeCheckInputs = [
     jsondiff
     pytestCheckHook
     requests
   ];
 
+  disabledTestPaths = [
+    # attempts to download models
+    "test/test_hybrid_chunker.py"
+  ];
+
   meta = {
     changelog = "https://github.com/DS4SD/docling-core/blob/${src.tag}/CHANGELOG.md";
     description = "Python library to define and validate data types in Docling";

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

@@ -0,0 +1,84 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  wheel,
+  importlib-resources,
+  pygments,
+  tqdm,
+  flask,
+  multiprocess,
+  docutils,
+  sphinx,
+  sphinx-autodoc-typehints,
+  sphinx-rtd-theme,
+  sphinx-versions,
+  sphinxcontrib-images,
+  ipywidgets,
+  numpy,
+  rich,
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "mpire";
+  version = "2.10.2";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "sybrenjansen";
+    repo = "mpire";
+    tag = "v${version}";
+    hash = "sha256-6O+k8gSMCu4zhj7KzbsC5UUCU/TG/g3dYsGVuvcy25E=";
+  };
+
+  build-system = [
+    setuptools
+    wheel
+  ];
+
+  dependencies = [
+    importlib-resources
+    pygments
+    tqdm
+  ];
+
+  optional-dependencies = {
+    dashboard = [
+      flask
+    ];
+    dill = [
+      multiprocess
+    ];
+    docs = [
+      docutils
+      sphinx
+      sphinx-autodoc-typehints
+      sphinx-rtd-theme
+      sphinx-versions
+      sphinxcontrib-images
+    ];
+    testing = [
+      ipywidgets
+      multiprocess
+      numpy
+      rich
+    ];
+  };
+
+  pythonImportsCheck = [
+    "mpire"
+  ];
+
+  nativeCheckInputs = [ pytestCheckHook ] ++ optional-dependencies.testing;
+
+  pytestFlagsArray = [ "tests" ];
+
+  meta = {
+    description = "A Python package for easy multiprocessing, but faster than multiprocessing";
+    homepage = "https://pypi.org/project/mpire/";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ booxter ];
+  };
+}

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

@@ -0,0 +1,39 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchPypi,
+  hatchling,
+  mpire,
+  tqdm,
+}:
+
+buildPythonPackage rec {
+  pname = "semchunk";
+  version = "3.0.1";
+  pyproject = true;
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-UP9nHLHGYNZm5eXHfNufDYhd9pPvrmp3HcVUFAjcAZw=";
+  };
+
+  build-system = [
+    hatchling
+  ];
+
+  dependencies = [
+    mpire
+    tqdm
+  ];
+
+  pythonImportsCheck = [
+    "semchunk"
+  ];
+
+  meta = {
+    description = "A fast, lightweight and easy-to-use Python library for splitting text into semantically meaningful chunks";
+    homepage = "https://pypi.org/project/semchunk/";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ booxter ];
+  };
+}

pkgs/development/python-modules/sphinx-versions/default.nix

@@ -0,0 +1,43 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchPypi,
+  setuptools,
+  wheel,
+  click,
+  colorclass,
+  sphinx,
+}:
+
+buildPythonPackage rec {
+  pname = "sphinx-versions";
+  version = "1.1.3";
+  pyproject = true;
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-9ROFEjET+d2Dfg4DHx0IqUN34oGwY4AGbi7teK4YmR8=";
+  };
+
+  build-system = [
+    setuptools
+    wheel
+  ];
+
+  dependencies = [
+    click
+    colorclass
+    sphinx
+  ];
+
+  pythonImportsCheck = [
+    "sphinxcontrib.versioning"
+  ];
+
+  meta = {
+    description = "Sphinx extension that allows building versioned docs for self-hosting";
+    homepage = "https://pypi.org/project/sphinx-versions/";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ booxter ];
+  };
+}

pkgs/development/python-modules/sphinxcontrib-images/default.nix

@@ -0,0 +1,41 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchPypi,
+  setuptools,
+  wheel,
+  requests,
+  sphinx,
+}:
+
+buildPythonPackage rec {
+  pname = "sphinxcontrib-images";
+  version = "0.9.4";
+  pyproject = true;
+
+  src = fetchPypi {
+    inherit pname version;
+    hash = "sha256-9sI30EMHk+ZdkdvdsTsfsmos+DgECp3utSESlp+8Sks=";
+  };
+
+  build-system = [
+    setuptools
+    wheel
+  ];
+
+  dependencies = [
+    requests
+    sphinx
+  ];
+
+  pythonImportsCheck = [
+    "sphinxcontrib.images"
+  ];
+
+  meta = {
+    description = "Sphinx extension for thumbnails";
+    homepage = "https://pypi.org/project/sphinxcontrib-images/";
+    license = lib.licenses.asl20;
+    maintainers = with lib.maintainers; [ booxter ];
+  };
+}

pkgs/top-level/python-packages.nix

@@ -8556,6 +8556,8 @@ self: super: with self; {
 
   mpi4py = callPackage ../development/python-modules/mpi4py { };
 
+  mpire = callPackage ../development/python-modules/mpire { };
+
   mpldatacursor = callPackage ../development/python-modules/mpldatacursor { };
 
   mplcursors = callPackage ../development/python-modules/mplcursors { };
@@ -14763,6 +14765,8 @@ self: super: with self; {
 
   semantic-version = callPackage ../development/python-modules/semantic-version { };
 
+  semchunk = callPackage ../development/python-modules/semchunk { };
+
   semgrep = callPackage ../development/python-modules/semgrep {
     semgrep-core = callPackage ../development/python-modules/semgrep/semgrep-core.nix { };
   };
@@ -15375,6 +15379,8 @@ self: super: with self; {
 
   sphinx-comments = callPackage ../development/python-modules/sphinx-comments { };
 
+  sphinxcontrib-images = callPackage ../development/python-modules/sphinxcontrib-images { };
+
   sphinx-design = callPackage ../development/python-modules/sphinx-design { };
 
   sphinx-external-toc = callPackage ../development/python-modules/sphinx-external-toc { };
@@ -15407,6 +15413,8 @@ self: super: with self; {
 
   sphinx-togglebutton = callPackage ../development/python-modules/sphinx-togglebutton { };
 
+  sphinx-versions = callPackage ../development/python-modules/sphinx-versions { };
+
   sphinxawesome-theme = callPackage ../development/python-modules/sphinxawesome-theme { };
 
   sphinxcontrib-actdiag = callPackage ../development/python-modules/sphinxcontrib-actdiag { };