natural-pdf 0.2.16__py3-none-any.whl → 0.2.18__py3-none-any.whl

natural_pdf/elements/element_collection.py

@@ -207,22 +207,82 @@ class ElementCollection(
         if not self._elements:
             return []
 
-        # Group elements by page
-        elements_by_page = {}
+        # Check for FlowRegions which need special handling
+        from natural_pdf.flows.region import FlowRegion
+
+        flow_regions = []
+        regular_elements = []
+
         for elem in self._elements:
+            if isinstance(elem, FlowRegion):
+                flow_regions.append(elem)
+            else:
+                regular_elements.append(elem)
+
+        # Start with specs from FlowRegions (they handle their own multi-page rendering)
+        all_specs = []
+        specs_by_page = {}  # Track specs by page for merging
+
+        for flow_region in flow_regions:
+            # FlowRegions have their own _get_render_specs method
+            flow_specs = flow_region._get_render_specs(
+                mode=mode,
+                color=color,
+                highlights=highlights,
+                crop=crop,
+                crop_bbox=crop_bbox,
+                **kwargs,
+            )
+            for spec in flow_specs:
+                # Check if we already have a spec for this page
+                if spec.page in specs_by_page:
+                    # Merge highlights into existing spec
+                    existing_spec = specs_by_page[spec.page]
+                    # Add all highlights from this spec to the existing one
+                    existing_spec.highlights.extend(spec.highlights)
+                    # Merge crop bbox if needed
+                    if spec.crop_bbox and not existing_spec.crop_bbox:
+                        existing_spec.crop_bbox = spec.crop_bbox
+                    elif spec.crop_bbox and existing_spec.crop_bbox:
+                        # Expand crop bbox to include both
+                        x0 = min(spec.crop_bbox[0], existing_spec.crop_bbox[0])
+                        y0 = min(spec.crop_bbox[1], existing_spec.crop_bbox[1])
+                        x1 = max(spec.crop_bbox[2], existing_spec.crop_bbox[2])
+                        y1 = max(spec.crop_bbox[3], existing_spec.crop_bbox[3])
+                        existing_spec.crop_bbox = (x0, y0, x1, y1)
+                else:
+                    # First spec for this page
+                    all_specs.append(spec)
+                    specs_by_page[spec.page] = spec
+
+        # Group regular elements by page
+        elements_by_page = {}
+        for elem in regular_elements:
             if hasattr(elem, "page"):
                 page = elem.page
                 if page not in elements_by_page:
                     elements_by_page[page] = []
                 elements_by_page[page].append(elem)
 
-        if not elements_by_page:
+        if not elements_by_page and not flow_regions:
             return []
 
-        # Create RenderSpec for each page
-        specs = []
+        # Create or update RenderSpec for each page with regular elements
         for page, page_elements in elements_by_page.items():
-            spec = RenderSpec(page=page)
+            # Check if we already have a spec for this page from FlowRegions
+            existing_spec = None
+            for spec in all_specs:
+                if spec.page == page:
+                    existing_spec = spec
+                    break
+
+            if existing_spec:
+                # We'll add to the existing spec
+                spec = existing_spec
+            else:
+                # Create new spec for this page
+                spec = RenderSpec(page=page)
+                all_specs.append(spec)
 
             # Handle cropping
             if crop_bbox:
@@ -390,9 +450,7 @@ class ElementCollection(
                                     element=elem, color=group_color, label=group_label
                                 )
 
-            specs.append(spec)
-
-        return specs
+        return all_specs
 
     def _get_highlighter(self):
         """Get the highlighting service for rendering.
@@ -889,7 +947,22 @@ class ElementCollection(
         return self
 
     def exclude(self):
-        self.page.add_exclusion(self)
+        """
+        Excludes all elements in the collection from their respective pages.
+
+        Since a collection can span multiple pages, this method iterates through
+        all elements and calls exclude() on each one individually.
+
+        Each element type is handled appropriately:
+        - Region elements exclude everything within their bounds
+        - Text/other elements exclude only the specific element, not the area
+
+        Returns:
+            Self for method chaining
+        """
+        for element in self._elements:
+            element.exclude()
+        return self
 
     def highlight(
         self,