Keep page.parsed_page.textline_cells and page.cells in sync, including OCR

Signed-off-by: Christoph Auer <cau@zurich.ibm.com>

docling/models/base_ocr_model.py

@@ -130,19 +130,49 @@ class BaseOcrModel(BasePageModel, BaseModelWithOptions):
         ]
         return filtered_ocr_cells
 
-    def post_process_cells(self, ocr_cells, programmatic_cells):
+    def post_process_cells(self, ocr_cells, page):
         r"""
-        Post-process the ocr and programmatic cells and return the final list of of cells
+        Post-process the OCR cells and update the page object.
+        Treats page.parsed_page as authoritative when available, with page.cells for compatibility.
         """
-        if self.options.force_full_page_ocr:
+        # Get existing cells (prefer parsed_page, fallback to page.cells)
-            # If a full page OCR is forced, use only the OCR cells
+        existing_cells = self._get_existing_cells(page)
-            cells = ocr_cells
-            return cells
 
-        ## Remove OCR cells which overlap with programmatic cells.
+        # Combine existing and OCR cells with overlap filtering
-        filtered_ocr_cells = self._filter_ocr_cells(ocr_cells, programmatic_cells)
+        final_cells = self._combine_cells(existing_cells, ocr_cells)
-        programmatic_cells.extend(filtered_ocr_cells)
+
-        return programmatic_cells
+        # Update both structures efficiently
+        self._update_page_structures(page, final_cells)
+
+    def _get_existing_cells(self, page):
+        """Get existing cells, preferring parsed_page when available."""
+        return page.parsed_page.textline_cells if page.parsed_page else page.cells
+
+    def _combine_cells(self, existing_cells, ocr_cells):
+        """Combine existing and OCR cells with filtering and re-indexing."""
+        if self.options.force_full_page_ocr:
+            combined = ocr_cells
+        else:
+            filtered_ocr_cells = self._filter_ocr_cells(ocr_cells, existing_cells)
+            combined = list(existing_cells) + filtered_ocr_cells
+
+        # Re-index in-place
+        for i, cell in enumerate(combined):
+            cell.index = i
+
+        return combined
+
+    def _update_page_structures(self, page, final_cells):
+        """Update both page structures efficiently."""
+        if page.parsed_page:
+            # Update parsed_page as primary source
+            page.parsed_page.textline_cells = final_cells
+            page.parsed_page.has_lines = bool(final_cells)
+            # Sync to page.cells for compatibility
+            page.cells = final_cells
+        else:
+            # Legacy fallback: only page.cells available
+            page.cells = final_cells
 
     def draw_ocr_rects_and_cells(self, conv_res, page, ocr_rects, show: bool = False):
         image = copy.deepcopy(page.image)

docling/models/easyocr_model.py

@@ -177,7 +177,7 @@ class EasyOcrModel(BaseOcrModel):
                         all_ocr_cells.extend(cells)
 
                     # Post-process the cells
-                    page.cells = self.post_process_cells(all_ocr_cells, page.cells)
+                    self.post_process_cells(all_ocr_cells, page)
 
                 # DEBUG code:
                 if settings.debug.visualize_ocr:

docling/models/layout_model.py

@@ -176,9 +176,9 @@ class LayoutModel(BasePageModel):
                     # Apply postprocessing
 
                     processed_clusters, processed_cells = LayoutPostprocessor(
-                        page.cells, clusters, page.size
+                        page, clusters
                     ).postprocess()
-                    # processed_clusters, processed_cells = clusters, page.cells
+                    # Note: LayoutPostprocessor updates page.cells and page.parsed_page internally
 
                     with warnings.catch_warnings():
                         warnings.filterwarnings(
@@ -198,7 +198,7 @@ class LayoutModel(BasePageModel):
                             )
                         )
 
-                    page.cells = processed_cells
+                    # page.cells is already updated by LayoutPostprocessor
                     page.predictions.layout = LayoutPrediction(
                         clusters=processed_clusters
                     )

docling/models/ocr_mac_model.py

@@ -132,7 +132,7 @@ class OcrMacModel(BaseOcrModel):
                         all_ocr_cells.extend(cells)
 
                     # Post-process the cells
-                    page.cells = self.post_process_cells(all_ocr_cells, page.cells)
+                    self.post_process_cells(all_ocr_cells, page)
 
                 # DEBUG code:
                 if settings.debug.visualize_ocr:

docling/models/tesseract_ocr_cli_model.py

@@ -306,7 +306,7 @@ class TesseractOcrCliModel(BaseOcrModel):
                             all_ocr_cells.append(cell)
 
                     # Post-process the cells
-                    page.cells = self.post_process_cells(all_ocr_cells, page.cells)
+                    self.post_process_cells(all_ocr_cells, page)
 
                 # DEBUG code:
                 if settings.debug.visualize_ocr:

docling/models/tesseract_ocr_model.py

@@ -235,7 +235,7 @@ class TesseractOcrModel(BaseOcrModel):
                         all_ocr_cells.extend(cells)
 
                     # Post-process the cells
-                    page.cells = self.post_process_cells(all_ocr_cells, page.cells)
+                    self.post_process_cells(all_ocr_cells, page)
 
                 # DEBUG code:
                 if settings.debug.visualize_ocr:

docling/utils/layout_postprocessor.py

@@ -194,11 +194,12 @@ class LayoutPostprocessor:
         DocItemLabel.TITLE: DocItemLabel.SECTION_HEADER,
     }
 
-    def __init__(self, cells: List[TextCell], clusters: List[Cluster], page_size: Size):
+    def __init__(self, page, clusters: List[Cluster]):
-        """Initialize processor with cells and clusters."""
+        """Initialize processor with page and clusters."""
-        """Initialize processor with cells and spatial indices."""
+        # Get cells from best available source (prefer parsed_page)
-        self.cells = cells
+        self.cells = self._get_page_cells(page)
-        self.page_size = page_size
+        self.page = page
+        self.page_size = page.size
         self.all_clusters = clusters
         self.regular_clusters = [
             c for c in clusters if c.label not in self.SPECIAL_TYPES
@@ -214,6 +215,24 @@ class LayoutPostprocessor:
             [c for c in self.special_clusters if c.label in self.WRAPPER_TYPES]
         )
 
+    def _get_page_cells(self, page):
+        """Get cells from best available source (prefer parsed_page)."""
+        return (
+            page.parsed_page.textline_cells
+            if page.parsed_page is not None
+            else page.cells
+        )
+
+    def _update_page_structures(self, final_cells):
+        """Update both page structures efficiently."""
+        if self.page.parsed_page is not None:
+            # Update parsed_page as primary source
+            self.page.parsed_page.textline_cells = final_cells
+            self.page.parsed_page.has_lines = len(final_cells) > 0
+
+        # Legacy fallback: only page.cells available
+        self.page.cells = final_cells
+
     def postprocess(self) -> Tuple[List[Cluster], List[TextCell]]:
         """Main processing pipeline."""
         self.regular_clusters = self._process_regular_clusters()
@@ -240,6 +259,9 @@ class LayoutPostprocessor:
             for child in cluster.children:
                 child.cells = self._sort_cells(child.cells)
 
+        # Update page structures with processed cells
+        self._update_page_structures(self.cells)
+
         return final_clusters, self.cells
 
     def _process_regular_clusters(self) -> List[Cluster]: