python312Packages.docling-parse: init at 2.0.2 (#353183)

pkgs/by-name/lo/loguru/package.nix

@@ -0,0 +1,49 @@
+{
+  lib,
+  stdenv,
+  fetchFromGitHub,
+  fetchpatch,
+  cmake,
+}:
+
+stdenv.mkDerivation {
+  pname = "loguru";
+  version = "2.1.0-unstable-2023-04-06";
+
+  src = fetchFromGitHub {
+    owner = "emilk";
+    repo = "loguru";
+    rev = "4adaa185883e3c04da25913579c451d3c32cfac1";
+    hash = "sha256-NpMKyjCC06bC5B3xqgDr2NgA9RsPEeiWr9GbHrHHzZ8=";
+  };
+
+  patches = [
+    # See https://github.com/emilk/loguru/issues/249
+    # The following patches are coming from a fork and fix builds on Darwin
+    # Hopefully they will be merged in the main repository soon.
+    (fetchpatch {
+      url = "https://github.com/virtuosonic/loguru/commit/e1ffdc4149083cc221d44b666a0f7e3ec4a87259.patch";
+      hash = "sha256-fYdS8+qfgyj1J+T6H434jDGK/L+VYq+L22CQ7M/uiXE=";
+    })
+    (fetchpatch {
+      url = "https://github.com/virtuosonic/loguru/commit/743777bea361642349d4673e6a0a55912849c14f.patch";
+      hash = "sha256-3FhH7zdkzHuXSirSCr8A3uHg8UpSfEM02AkR0ZSG0Yw=";
+    })
+  ];
+
+  cmakeFlags = [
+    "-DLOGURU_WITH_STREAMS=1"
+  ];
+
+  nativeBuildInputs = [
+    cmake
+  ];
+
+  meta = {
+    description = "Lightweight C++ logging library";
+    homepage = "https://github.com/emilk/loguru";
+    license = lib.licenses.unlicense;
+    maintainers = with lib.maintainers; [ drupol ];
+    platforms = lib.platforms.all;
+  };
+}

pkgs/development/python-modules/deepsearch-toolkit/default.nix

@@ -0,0 +1,93 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  poetry-core,
+  certifi,
+  docling-core,
+  platformdirs,
+  pluggy,
+  pydantic,
+  pydantic-settings,
+  python-dateutil,
+  python-dotenv,
+  requests,
+  six,
+  tabulate,
+  tqdm,
+  typer,
+  urllib3,
+  anyio,
+  fastapi,
+  uvicorn,
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "deepsearch-toolkit";
+  version = "1.1.1";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "DS4SD";
+    repo = "deepsearch-toolkit";
+    rev = "refs/tags/v${version}";
+    hash = "sha256-7XiI/VtXX4lRMreqUh6hJvdIULGvsCEdrd+zV5Jrne0=";
+  };
+
+  build-system = [
+    poetry-core
+  ];
+
+  dependencies = [
+    certifi
+    docling-core
+    platformdirs
+    pluggy
+    pydantic
+    pydantic-settings
+    python-dateutil
+    python-dotenv
+    requests
+    six
+    tabulate
+    tqdm
+    typer
+    urllib3
+  ];
+
+  pythonRelaxDeps = [
+    "urllib3"
+  ];
+
+  optional-dependencies = rec {
+    all = api;
+    api = [
+      anyio
+      fastapi
+      uvicorn
+    ];
+  };
+
+  pythonImportsCheck = [
+    "deepsearch"
+  ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  disabledTests = [
+    # Tests require the creation of a deepsearch profile
+    "test_project_listing"
+    "test_system_info"
+  ];
+
+  meta = {
+    changelog = "https://github.com/DS4SD/deepsearch-toolkit/blob/${src.rev}/CHANGELOG.md";
+    description = "Interact with the Deep Search platform for new knowledge explorations and discoveries";
+    homepage = "https://github.com/DS4SD/deepsearch-toolkit";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,59 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  poetry-core,
+  jsonref,
+  jsonschema,
+  pandas,
+  pillow,
+  pydantic,
+  tabulate,
+  jsondiff,
+  requests,
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "docling-core";
+  version = "2.3.1";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "DS4SD";
+    repo = "docling-core";
+    rev = "refs/tags/v${version}";
+    hash = "sha256-yNsmMAeR1sbZsddpjMFWZy2UAbIWWiZmdW4/lwLvCbM=";
+  };
+
+  build-system = [
+    poetry-core
+  ];
+
+  dependencies = [
+    jsonref
+    jsonschema
+    pandas
+    pillow
+    pydantic
+    tabulate
+  ];
+
+  pythonImportsCheck = [
+    "docling_core"
+  ];
+
+  nativeCheckInputs = [
+    jsondiff
+    pytestCheckHook
+    requests
+  ];
+
+  meta = {
+    changelog = "https://github.com/DS4SD/docling-core/blob/${version}/CHANGELOG.md";
+    description = "Python library to define and validate data types in Docling";
+    homepage = "https://github.com/DS4SD/docling-core";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,75 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  poetry-core,
+  huggingface-hub,
+  jsonlines,
+  lxml,
+  mean-average-precision,
+  numpy,
+  opencv-python-headless,
+  pillow,
+  torch,
+  torchvision,
+  tqdm,
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "docling-ibm-models";
+  version = "2.0.3";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "DS4SD";
+    repo = "docling-ibm-models";
+    rev = "refs/tags/v${version}";
+    hash = "sha256-vSEW1+mFTjUvjjUOoX3aGgT/y8iwP3JGIZaPh9RbX5I=";
+  };
+
+  build-system = [
+    poetry-core
+  ];
+
+  dependencies = [
+    huggingface-hub
+    jsonlines
+    lxml
+    mean-average-precision
+    numpy
+    opencv-python-headless
+    pillow
+    torch
+    torchvision
+    tqdm
+  ];
+
+  pythonRelaxDeps = [
+    "lxml"
+    "mean_average_precision"
+    "torchvision"
+  ];
+
+  pythonImportsCheck = [
+    "docling_ibm_models"
+  ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  disabledTests = [
+    # Requires network access
+    "test_layoutpredictor"
+    "test_tf_predictor"
+  ];
+
+  meta = {
+    changelog = "https://github.com/DS4SD/docling-ibm-models/blob/${src.rev}/CHANGELOG.md";
+    description = "Docling IBM models";
+    homepage = "https://github.com/DS4SD/docling-ibm-models";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}

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

@@ -0,0 +1,81 @@
+{
+  lib,
+  buildPythonPackage,
+  fetchFromGitHub,
+  cmake,
+  pkg-config,
+  cxxopts,
+  poetry-core,
+  pybind11,
+  tabulate,
+  zlib,
+  nlohmann_json,
+  utf8cpp,
+  libjpeg,
+  qpdf,
+  loguru-cpp,
+  pytestCheckHook,
+}:
+
+buildPythonPackage rec {
+  pname = "docling-parse";
+  version = "2.0.2";
+  pyproject = true;
+
+  src = fetchFromGitHub {
+    owner = "DS4SD";
+    repo = "docling-parse";
+    rev = "refs/tags/v${version}";
+    hash = "sha256-unXGmMp5xyRCqSoFmqcQAZOBzpE0EzgEEBIfZUHhRcQ=";
+  };
+
+  dontUseCmakeConfigure = true;
+
+  nativeBuildInputs = [
+    cmake
+    pkg-config
+  ];
+
+  build-system = [
+    poetry-core
+  ];
+
+  env.NIX_CFLAGS_COMPILE = "-I${lib.getDev utf8cpp}/include/utf8cpp";
+
+  buildInputs = [
+    pybind11
+    cxxopts
+    libjpeg
+    loguru-cpp
+    nlohmann_json
+    qpdf
+    utf8cpp
+    zlib
+  ];
+
+  env.USE_SYSTEM_DEPS = true;
+
+  cmakeFlags = [
+    "-DUSE_SYSTEM_DEPS=True"
+  ];
+
+  dependencies = [
+    tabulate
+  ];
+
+  pythonImportsCheck = [
+    "docling_parse"
+  ];
+
+  nativeCheckInputs = [
+    pytestCheckHook
+  ];
+
+  meta = {
+    changelog = "https://github.com/DS4SD/docling-parse/blob/${src.rev}/CHANGELOG.md";
+    description = "Simple package to extract text with coordinates from programmatic PDFs";
+    homepage = "https://github.com/DS4SD/docling-parse";
+    license = lib.licenses.mit;
+    maintainers = with lib.maintainers; [ drupol ];
+  };
+}