python312Packages.markitdown: init at 0-unstable-2024-12-18

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

@@ -0,0 +1,51 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  pytestCheckHook,
+  gitUpdater,
+}:
+
+buildPythonPackage rec {
+  pname = "cobble";
+  version = "0.1.4";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "mwilliamson";
+    repo = "python-cobble";
+    tag = version;
+    hash = "sha256-xi6cCSUnMYc5Tp6+TQlC9Oo9kpam5C7QGCul/IoTW1k=";
+  };
+
+  build-system = [ setuptools ];
+
+  pythonImportsCheck = [ "cobble" ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  pytestFlagsArray = [ "tests.py" ];
+
+  disabledTests = [
+    # Broken tests
+    #
+    # left = "Can't instantiate abstract class Evaluator with abstract method visit_add"
+    # right = "Can't instantiate abstract class Evaluator without an implementation for abstract method 'visit_add'"
+    "test_error_if_visitor_is_missing_methods"
+    # left = "Can't instantiate abstract class Evaluator with abstract method visit_literal"
+    # right = "Can't instantiate abstract class Evaluator without an implementation for abstract method 'visit_literal'"
+    "test_sub_sub_classes_are_included_in_abc"
+  ];
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "Create Python data objects";
+    homepage = "https://github.com/mwilliamson/python-cobble";
+    license = lib.licenses.bsd2;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,40 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  precisely,
+  gitUpdater,
+}:
+
+buildPythonPackage rec {
+  pname = "funk";
+  version = "0.5.0";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "mwilliamson";
+    repo = "funk";
+    tag = version;
+    hash = "sha256-dEq3zyA8rtNt0sui2TfQ3OUSCZ0XDMOdthcqt/QrCsU=";
+  };
+
+  build-system = [ setuptools ];
+
+  dependencies = [ precisely ];
+
+  pythonImportsCheck = [ "funk" ];
+
+  # Disabling tests, they rely on Nose which is outdated and not supported
+  doCheck = false;
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "A mocking framework for Python, influenced by JMock";
+    homepage = "https://github.com/mwilliamson/funk";
+    changelog = "https://github.com/mwilliamson/funk/blob/${src.tag}/NEWS";
+    license = lib.licenses.bsd2;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,57 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  cobble,
+  funk,
+  pytestCheckHook,
+  spur,
+  tempman,
+  gitUpdater,
+}:
+
+buildPythonPackage rec {
+  pname = "mammoth";
+  version = "1.8.0";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "mwilliamson";
+    repo = "python-mammoth";
+    tag = version;
+    hash = "sha256-mIvzEyKpy1j7tZti+QTXgsbqqZAhYs7N+p5sQFsLhBg=";
+  };
+
+  postPatch = ''
+    substituteInPlace setup.py \
+      --replace-fail 'read("README")' '""'
+  '';
+
+  build-system = [ setuptools ];
+
+  dependencies = [ cobble ];
+
+  pythonImportsCheck = [ "mammoth" ];
+
+  nativeCheckInputs = [
+    funk
+    pytestCheckHook
+    spur
+    tempman
+  ];
+
+  preCheck = ''
+    export PATH=$out/bin:$PATH
+  '';
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "Convert Word documents (.docx files) to HTML";
+    homepage = "https://github.com/mwilliamson/python-mammoth";
+    changelog = "https://github.com/mwilliamson/python-mammoth/blob/${src.tag}/NEWS";
+    license = lib.licenses.bsd2;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,76 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  hatchling,
+  beautifulsoup4,
+  ffmpeg-headless,
+  mammoth,
+  markdownify,
+  numpy,
+  openai,
+  openpyxl,
+  pandas,
+  pathvalidate,
+  pdfminer-six,
+  puremagic,
+  pydub,
+  python-pptx,
+  requests,
+  speechrecognition,
+  youtube-transcript-api,
+  pytestCheckHook,
+  gitUpdater,
+}:
+
+buildPythonPackage {
+  pname = "markitdown";
+  version = "unstable-2024-12-18";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "microsoft";
+    repo = "markitdown";
+    rev = "3ce21a47abed0e4db162de1088d661887ae076ff";
+    hash = "sha256-5YafFL8OHNcGgB/qH6CmX0rTith1ZSRNIa+ktl4Ffvg=";
+  };
+
+  build-system = [ hatchling ];
+
+  dependencies = [
+    beautifulsoup4
+    ffmpeg-headless
+    mammoth
+    markdownify
+    numpy
+    openai
+    openpyxl
+    pandas
+    pathvalidate
+    pdfminer-six
+    puremagic
+    pydub
+    python-pptx
+    requests
+    speechrecognition
+    youtube-transcript-api
+  ];
+
+  pythonImportsCheck = [ "markitdown" ];
+
+  nativeCheckInputs = [ pytestCheckHook ];
+
+  disabledTests = [
+    # Require network access
+    "test_markitdown_remote"
+  ];
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "Python tool for converting files and office documents to Markdown";
+    homepage = "https://github.com/microsoft/markitdown";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,37 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  pytestCheckHook,
+  gitUpdater,
+}:
+
+buildPythonPackage rec {
+  pname = "precisely";
+  version = "0.1.9";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "mwilliamson";
+    repo = "python-precisely";
+    tag = version;
+    hash = "sha256-jvvRreSGpRgDk1bbqC8Z/UEfvxwKilfc/sm7nxdJU6k=";
+  };
+
+  build-system = [ setuptools ];
+
+  pythonImportsCheck = [ "precisely" ];
+
+  # Tests are outdated and based on Nose, which is not supported anymore.
+  doCheck = false;
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "Matcher library for Python";
+    homepage = "https://github.com/mwilliamson/python-precisely";
+    license = lib.licenses.bsd2;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,42 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  setuptools,
+  pytestCheckHook,
+  gitUpdater,
+}:
+
+buildPythonPackage rec {
+  pname = "tempman";
+  version = "0.1.3";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "mwilliamson";
+    repo = "python-tempman";
+    tag = version;
+    hash = "sha256-EHTnlT3vcmyjyyS3QCJXjAuZqOEc0i11rEb6zfX6rDY=";
+  };
+
+  postPatch = ''
+    substituteInPlace setup.py \
+      --replace-fail 'read("README")' '""'
+  '';
+
+  build-system = [ setuptools ];
+
+  pythonImportsCheck = [ "tempman" ];
+
+  # Disabling tests, they rely on dependencies that are outdated and not supported
+  doCheck = false;
+
+  passthru.updateScripts = gitUpdater { };
+
+  meta = {
+    description = "Create and clean up temporary directories";
+    homepage = "https://github.com/mwilliamson/python-tempman";
+    license = lib.licenses.bsd2;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

pkgs/top-level/python-packages.nix

@@ -2480,6 +2480,8 @@ self: super: with self; {
 
   coapthon3 = callPackage ../development/python-modules/coapthon3 { };
 
+  cobble = callPackage ../development/python-modules/cobble { };
+
   coconut = callPackage ../development/python-modules/coconut { };
 
   cocotb = callPackage ../development/python-modules/cocotb { };
@@ -4965,6 +4967,8 @@ self: super: with self; {
 
   funcy = callPackage ../development/python-modules/funcy { };
 
+  funk = callPackage ../development/python-modules/funk { };
+
   funsor = callPackage ../development/python-modules/funsor { };
 
   furl = callPackage ../development/python-modules/furl { };
@@ -7803,6 +7807,8 @@ self: super: with self; {
 
   malduck = callPackage ../development/python-modules/malduck { };
 
+  mammoth = callPackage ../development/python-modules/mammoth { };
+
   mamba-ssm = callPackage ../development/python-modules/mamba-ssm { };
 
   managesieve = callPackage ../development/python-modules/managesieve { };
@@ -7863,6 +7869,8 @@ self: super: with self; {
 
   markdownify  = callPackage ../development/python-modules/markdownify { };
 
+  markitdown  = callPackage ../development/python-modules/markitdown { };
+
   marko = callPackage ../development/python-modules/marko { };
 
   markupsafe = callPackage ../development/python-modules/markupsafe { };
@@ -10404,6 +10412,8 @@ self: super: with self; {
 
   ppk2-api = callPackage ../development/python-modules/ppk2-api { };
 
+  precisely = callPackage ../development/python-modules/precisely { };
+
   primp = callPackage ../development/python-modules/primp {
     inherit (pkgs.darwin.apple_sdk.frameworks) SystemConfiguration;
   };
@@ -15749,6 +15759,8 @@ self: super: with self; {
 
   templateflow = callPackage ../development/python-modules/templateflow { };
 
+  tempman = callPackage ../development/python-modules/tempman { };
+
   tempora = callPackage ../development/python-modules/tempora { };
 
   tenacity = callPackage ../development/python-modules/tenacity { };