fix: make docling pdf processor backward-compatible

Author: DhruvGarg111

src/raglight/document_processing/docling_pdf_processor.py

@@ -9,6 +9,32 @@
 from .document_processor import DocumentProcessor
 
 
+def _set_pipeline_option_if_supported(
+    pipeline_options: PdfPipelineOptions, option_name: str, value: bool
+) -> None:
+    """
+    Safely set a Docling PdfPipelineOptions field when available.
+
+    Older Docling versions may not expose newer enrichment flags.
+    """
+    model_fields = getattr(type(pipeline_options), "model_fields", None)
+    if isinstance(model_fields, dict) and option_name not in model_fields:
+        logging.info(
+            "Docling option '%s' is not available in this version, skipping.",
+            option_name,
+        )
+        return
+
+    try:
+        setattr(pipeline_options, option_name, value)
+    except (AttributeError, ValueError) as e:
+        logging.warning(
+            "Unable to set Docling option '%s' (continuing without it): %s",
+            option_name,
+            e,
+        )
+
+
 class DoclingPDFProcessor(DocumentProcessor):
     """
     Advanced PDF processor using IBM's Docling for high-fidelity document parsing.
@@ -18,10 +44,18 @@ class DoclingPDFProcessor(DocumentProcessor):
 
     def __init__(self):
         pipeline_options = PdfPipelineOptions()
-        pipeline_options.do_table_structure = True
-        pipeline_options.do_formula_enrichment = True
-        pipeline_options.do_code_enrichment = True
-        pipeline_options.do_chart_extraction = True
+        _set_pipeline_option_if_supported(
+            pipeline_options, option_name="do_table_structure", value=True
+        )
+        _set_pipeline_option_if_supported(
+            pipeline_options, option_name="do_formula_enrichment", value=True
+        )
+        _set_pipeline_option_if_supported(
+            pipeline_options, option_name="do_code_enrichment", value=True
+        )
+        _set_pipeline_option_if_supported(
+            pipeline_options, option_name="do_chart_extraction", value=True
+        )
 
         self.converter = DocumentConverter(
             format_options={

src/raglight/document_processing/document_processor_factory.py

@@ -1,3 +1,4 @@
+import logging
 from typing import Dict, Optional
 
 from .document_processor import DocumentProcessor
@@ -13,6 +14,19 @@
     HAS_DOCLING = False
 
 
+def _build_default_pdf_processor() -> DocumentProcessor:
+    if not HAS_DOCLING:
+        return PDFProcessor()
+
+    try:
+        return DoclingPDFProcessor()
+    except Exception as e:
+        logging.warning(
+            "Docling initialization failed, falling back to PDFProcessor. Error: %s", e
+        )
+        return PDFProcessor()
+
+
 class DocumentProcessorFactory:
     """
     A stateless factory that returns the appropriate DocumentProcessor
@@ -25,7 +39,7 @@ def __init__(
         # Default processors
         self._processors: Dict[str, DocumentProcessor] = {
             # PDF files
-            "pdf": DoclingPDFProcessor() if HAS_DOCLING else PDFProcessor(),
+            "pdf": _build_default_pdf_processor(),
             # Code files
             "py": CodeProcessor(),
             "js": CodeProcessor(),

tests/tests_document_processing/test_document_processor_factory.py

@@ -0,0 +1,56 @@
+import unittest
+from unittest.mock import MagicMock, patch
+
+from raglight.document_processing.document_processor_factory import (
+    DocumentProcessorFactory,
+)
+from raglight.document_processing.pdf_processor import PDFProcessor
+
+
+class TestDocumentProcessorFactory(unittest.TestCase):
+    def test_uses_docling_processor_when_available(self):
+        docling_instance = MagicMock(name="docling_processor")
+
+        with (
+            patch(
+                "raglight.document_processing.document_processor_factory.HAS_DOCLING",
+                True,
+            ),
+            patch(
+                "raglight.document_processing.document_processor_factory.DoclingPDFProcessor",
+                return_value=docling_instance,
+                create=True,
+            ) as mock_docling,
+        ):
+            factory = DocumentProcessorFactory()
+
+        self.assertIs(factory.get_processor("sample.pdf"), docling_instance)
+        mock_docling.assert_called_once()
+
+    def test_falls_back_to_pdf_processor_when_docling_init_fails(self):
+        with (
+            patch(
+                "raglight.document_processing.document_processor_factory.HAS_DOCLING",
+                True,
+            ),
+            patch(
+                "raglight.document_processing.document_processor_factory.DoclingPDFProcessor",
+                side_effect=ValueError("unsupported docling field"),
+                create=True,
+            ),
+        ):
+            factory = DocumentProcessorFactory()
+
+        self.assertIsInstance(factory.get_processor("sample.pdf"), PDFProcessor)
+
+    def test_falls_back_to_pdf_processor_when_docling_is_unavailable(self):
+        with patch(
+            "raglight.document_processing.document_processor_factory.HAS_DOCLING", False
+        ):
+            factory = DocumentProcessorFactory()
+
+        self.assertIsInstance(factory.get_processor("sample.pdf"), PDFProcessor)
+
+
+if __name__ == "__main__":
+    unittest.main()