python3Packages.docling-jobkit: 1.8.1 -> 3.2.0

Diff: https://github.com/docling-project/docling-jobkit/compare/v1.8.1...v3.2.0

Changelog: https://github.com/docling-project/docling-jobkit/blob/v3.2.0/CHANGELOG.md

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

@@ -3,92 +3,131 @@
   stdenv,
   buildPythonPackage,
   fetchFromGitHub,
+  pythonAtLeast,
+  pythonOlder,
 
   # build-system
   hatchling,
-  poetry-core,
 
   # dependencies
-  docling,
-  pydantic-settings,
-  typer,
   boto3,
-  pandas,
-  fastparquet,
-  pyarrow,
+  docling-slim,
   httpx,
+  pandas,
+  pydantic-settings,
 
-  # optional dependencies
-  ray,
+  # optional-dependencies
+  # rq:
   rq,
   msgpack,
+  # ray:
+  redis,
+  psutil,
+  ray,
+  # gdrive:
+  google-api-python-client,
+  google-auth-oauthlib,
+  # gcloudstorage:
+  google-cloud-storage,
+  # azure:
+  azure-storage-blob,
+  # opensearch:
+  opensearch-py,
 
   # tests
-  pytestCheckHook,
+  aiohttp,
   pytest-asyncio,
+  pytestCheckHook,
+  starlette,
+  redisTestHook,
   writableTmpDirAsHomeHook,
 }:
 
-buildPythonPackage rec {
+buildPythonPackage (finalAttrs: {
   pname = "docling-jobkit";
-  version = "1.8.1";
+  version = "3.2.0";
   pyproject = true;
+  __structuredAttrs = true;
 
   src = fetchFromGitHub {
     owner = "docling-project";
     repo = "docling-jobkit";
-    tag = "v${version}";
-    hash = "sha256-9DzQY/XMmx/8XP1bMYZYl+Bp7AVcYfuv3MtO6lvQ/24=";
+    tag = "v${finalAttrs.version}";
+    hash = "sha256-4OUNDn90jGWqHcs9QPpOKqYnKDxH+N/YUeszY21nhMo=";
   };
 
   build-system = [
     hatchling
-    poetry-core
   ];
 
-  dependencies = [
-    docling
-    pydantic-settings
-    typer
-    boto3
-    pandas
-    fastparquet
-    pyarrow
-    httpx
+  pythonRelaxDeps = [
+    "pandas"
   ];
+  dependencies = [
+    boto3
+    docling-slim
+    httpx
+    pandas
+    pydantic-settings
+  ]
+  ++ docling-slim.optional-dependencies.standard;
 
   optional-dependencies = {
-    ray = [ ray ];
+    vlm = docling-slim.optional-dependencies.models-vlm-inline;
     rq = [
       rq
       msgpack
     ];
+    ray = [
+      msgpack
+      psutil
+      ray
+    ]
+    ++ ray.optional-dependencies.serve
+    ++ redis.optional-dependencies.hiredis
+    ++ lib.optionals ((pythonAtLeast "3.11") && (pythonOlder "3.14")) [
+      # cloudflare-sdk
+    ];
+    gdrive = [
+      google-api-python-client
+      google-auth-oauthlib
+    ];
+    gcloudstorage = [
+      google-cloud-storage
+    ];
+    azure = [
+      azure-storage-blob
+    ];
+    opensearch = [
+      opensearch-py
+    ];
   };
 
-  pythonRelaxDeps = [
-    "boto3"
-    "pandas"
-    "pyarrow"
-  ];
-
   pythonImportsCheck = [
     "docling"
     "docling_jobkit"
   ];
 
   nativeCheckInputs = [
-    pytestCheckHook
+    aiohttp
     pytest-asyncio
+    pytestCheckHook
+    ray
+    redisTestHook
+    starlette
     writableTmpDirAsHomeHook
   ]
-  ++ optional-dependencies.rq;
+  ++ lib.concatAttrValues finalAttrs.passthru.optional-dependencies;
 
   disabledTests = [
     # requires network access
     "test_chunk_file"
+    "test_chunker_manager_shared_across_workers"
+    "test_clear_converters_clears_caches"
     "test_convert_file"
-    "test_convert_warmup"
     "test_convert_url"
+    "test_convert_warmup"
+    "test_convert_with_callbacks"
     "test_replicated_convert"
   ]
   ++ lib.optionals stdenv.hostPlatform.isDarwin [
@@ -98,10 +137,10 @@ buildPythonPackage rec {
   ];
 
   meta = {
-    changelog = "https://github.com/docling-project/docling-jobkit/blob/${src.tag}/CHANGELOG.md";
     description = "Running a distributed job processing documents with Docling";
     homepage = "https://github.com/docling-project/docling-jobkit";
+    changelog = "https://github.com/docling-project/docling-jobkit/blob/${finalAttrs.src.tag}/CHANGELOG.md";
     license = lib.licenses.mit;
     maintainers = with lib.maintainers; [ codgician ];
   };
-}
+})