python312Packages.docling: 2.23.0 -> 2.25.1 (#386660)

Author: happysalada

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

@@ -2,53 +2,62 @@
   lib,
   buildPythonPackage,
   fetchFromGitHub,
+
+  # build-system
   poetry-core,
+
   # dependencies
-  torch,
-  torchvision,
-  transformers,
+  docling-core,
   huggingface-hub,
   jsonlines,
   numpy,
   opencv-python-headless,
   pillow,
-  tqdm,
+  pydantic,
   safetensors,
+  torch,
+  torchvision,
+  tqdm,
+  transformers,
+
+  # tests
+  datasets,
   pytestCheckHook,
+  writableTmpDirAsHomeHook,
 }:
 
 buildPythonPackage rec {
   pname = "docling-ibm-models";
-  version = "3.3.2";
+  version = "3.4.1";
   pyproject = true;
 
   src = fetchFromGitHub {
     owner = "DS4SD";
     repo = "docling-ibm-models";
     tag = "v${version}";
-    hash = "sha256-8mqDgbTj5g6jXEumj16Me9NjHLCOdR+pXmAwn2dghfg=";
+    hash = "sha256-IgKP+4en0N6xLBjzwmddDk8z5P3J1vo+UMW+65a01Ww=";
   };
 
   build-system = [
     poetry-core
   ];
 
   dependencies = [
-    torch
-    torchvision
-    transformers
-    numpy
+    docling-core
+    huggingface-hub
     jsonlines
-    pillow
-    tqdm
+    numpy
     opencv-python-headless
-    huggingface-hub
+    pillow
+    pydantic
     safetensors
+    torch
+    torchvision
+    tqdm
+    transformers
   ];
 
   pythonRelaxDeps = [
-    "pillow"
-    "torchvision"
     "transformers"
     "numpy"
   ];
@@ -58,19 +67,18 @@ buildPythonPackage rec {
   ];
 
   nativeCheckInputs = [
+    datasets
     pytestCheckHook
+    writableTmpDirAsHomeHook
   ];
 
-  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
+    "test_layoutpredictor"
+    "test_readingorder"
+    "test_tf_predictor"
   ];
 
   meta = {

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

@@ -9,7 +9,6 @@
   # dependencies
   beautifulsoup4,
   certifi,
-  deepsearch-glm,
   docling-core,
   docling-ibm-models,
   docling-parse,
@@ -49,14 +48,14 @@
 
 buildPythonPackage rec {
   pname = "docling";
-  version = "2.23.0";
+  version = "2.25.1";
   pyproject = true;
 
   src = fetchFromGitHub {
     owner = "DS4SD";
     repo = "docling";
     tag = "v${version}";
-    hash = "sha256-ySywKaLxjtgQM7RtzJrxZDS3z8uMwAwPDYO51uKHT28=";
+    hash = "sha256-uJZIwXVwGd9xDBQkZ/uz2XLJ+lL1ZAZHJv/jHtW/LCk=";
   };
 
   build-system = [
@@ -66,7 +65,6 @@ buildPythonPackage rec {
   dependencies = [
     beautifulsoup4
     certifi
-    deepsearch-glm
     docling-core
     docling-ibm-models
     docling-parse
@@ -133,6 +131,19 @@ buildPythonPackage rec {
     "test_e2e_pdfs_conversions" # AssertionError: ## TableFormer: Table Structure Understanding with Transf
     "test_e2e_conversions" # RuntimeError: Tesseract is not available
 
+    # AssertionError
+    # assert doc.export_to_markdown() == pair[1], f"Error in case {idx}"
+    "test_ordered_lists"
+
+    # AssertionError: export to md
+    "test_e2e_html_conversions"
+
+    # AssertionError: assert 'Unordered li...d code block:' == 'Unordered li...d code block:'
+    "test_convert_valid"
+
+    # AssertionError: Markdown file mismatch against groundtruth pftaps057006474.md
+    "test_patent_groundtruth"
+
     # huggingface_hub.errors.LocalEntryNotFoundError: An error happened
     "test_cli_convert"
     "test_code_and_formula_conversion"
@@ -152,7 +163,7 @@ buildPythonPackage rec {
   meta = {
     description = "Get your documents ready for gen AI";
     homepage = "https://github.com/DS4SD/docling";
-    changelog = "https://github.com/DS4SD/docling/blob/${src.rev}/CHANGELOG.md";
+    changelog = "https://github.com/DS4SD/docling/blob/v${version}/CHANGELOG.md";
     license = lib.licenses.mit;
     maintainers = with lib.maintainers; [ happysalada ];
     mainProgram = "docling";