fix: updated caption references in add_node_items

docling_core/types/doc/document.py

@@ -4562,6 +4562,30 @@ def _append_item_copies(
         for item in node_items:
             item_copy = item.model_copy(deep=True)
 
+            # handle DocItem pointers (comments)
+            if isinstance(item, DocItem):
+                if item.comments:
+                    if isinstance(item_copy, DocItem):
+                        item_copy.comments = self._copy_and_reindex_refs(item.comments, doc=doc, parent_ref=parent_ref)
+
+            # handling new references for floating items
+            if isinstance(item, FloatingItem):
+                if item.captions:
+                    if isinstance(item_copy, FloatingItem):
+                        item_copy.captions = self._copy_and_reindex_refs(item.captions, doc=doc, parent_ref=parent_ref)
+
+                if item.footnotes:
+                    if isinstance(item_copy, FloatingItem):
+                        item_copy.footnotes = self._copy_and_reindex_refs(
+                            item.footnotes, doc=doc, parent_ref=parent_ref
+                        )
+
+                if item.references:
+                    if isinstance(item_copy, FloatingItem):
+                        item_copy.references = self._copy_and_reindex_refs(
+                            item.references, doc=doc, parent_ref=parent_ref
+                        )
+
             self._append_item(item=item_copy, parent_ref=parent_ref)
 
             if item_copy.children:
@@ -4578,6 +4602,31 @@ def _append_item_copies(
 
         return new_refs
 
+    def _copy_and_reindex_refs(self, ref_list: list[Any], doc: "DoclingDocument", parent_ref: RefItem) -> list[Any]:
+        """Helper to copy referenced items and return their new indices
+
+        :param ref_list: list[Any]: The list of references (e.g., captions, footnotes, comments) to be copied
+        :param doc: "DoclingDocument": The document from which the NodeItems are taken
+        :param parent_ref: RefItem: The reference of the parent item in the current document where copies will be appended to
+
+        :returns: list[Any]: A new list of references pointing to the newly appended items in the current document
+        """
+        if not ref_list:
+            return []
+
+        new_refs = []
+        for ref in ref_list:
+            resolved_item = ref.resolve(doc)
+            if resolved_item:
+                ref_copy = resolved_item.model_copy(deep=True)
+                self._append_item(item=ref_copy, parent_ref=parent_ref)
+
+                new_ref_pointer = ref.model_copy(deep=True)
+                new_ref_pointer.cref = ref_copy.get_ref().cref
+
+                new_refs.append(new_ref_pointer)
+        return new_refs
+
     def num_pages(self):
         """num_pages."""
         return len(self.pages.values())

test/test_docling_doc.py

@@ -1986,3 +1986,46 @@ def test_docitem_comments_delete_updates_refs():
     # The resolved comment should still work
     resolved = updated_para.comments[0].resolve(doc)
     assert resolved.text == "Comment on second paragraph."
+
+def test_add_node_items_updates_all_pointers():
+    """
+    Verifies that copying an item updates its references, footnotes, and comments to point to the newly assigned indices in the destination document.
+    """
+    # set source document
+    src_doc = DoclingDocument(name="source")
+
+    # create the items that will be pointed to and attach them
+    ref_text = src_doc.add_text(label=DocItemLabel.REFERENCE, text="[1] Source Reference")
+    foot_text = src_doc.add_text(label=DocItemLabel.FOOTNOTE, text="* Source Footnote")
+
+    table = src_doc.add_table(data=TableData(num_rows=1, num_cols=1))
+    table.references = [ref_text.get_ref()]
+    table.footnotes = [foot_text.get_ref()]
+    src_doc.add_comment(text="Source Comment", targets=[table])
+
+    dest_doc = DoclingDocument(name="dest")
+
+    # pad the destination so the indices are forced to shift
+    pad_text = dest_doc.add_text(label=DocItemLabel.TEXT, text="Padding Text")
+    dest_doc.add_comment(text="Padding Comment", targets=[pad_text])
+
+    dest_doc.add_node_items(node_items=[table], doc=src_doc)
+    new_table = dest_doc.tables[0]
+
+    # references
+    assert len(new_table.references) == 1
+    resolved_ref = new_table.references[0].resolve(dest_doc)
+    assert resolved_ref is not None, "Reference pointer is broken!"
+    assert resolved_ref.text == "[1] Source Reference"
+
+    # footnotes
+    assert len(new_table.footnotes) == 1
+    resolved_foot = new_table.footnotes[0].resolve(dest_doc)
+    assert resolved_foot is not None, "Footnote pointer is broken!"
+    assert resolved_foot.text == "* Source Footnote"
+
+    # comments
+    assert len(new_table.comments) == 1
+    resolved_comment = new_table.comments[0].resolve(dest_doc)
+    assert resolved_comment is not None, "Comment pointer is broken!"
+    assert resolved_comment.text == "Source Comment"