fix: properly handle missing page image case for export_to_html (#166)

* fix: properly handle missing image case Signed-off-by: Yusik Kim <[email protected]> * chore: pre-commit changes Signed-off-by: Yusik Kim <[email protected]> * fix: simplify test and address broken test Signed-off-by: Yusik Kim <[email protected]> --------- Signed-off-by: Yusik Kim <[email protected]>
DS4SD · Feb 25, 2025 · 4708f93 · 4708f93
1 parent 28bd65c
commit 4708f93
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 4 deletions.
diff --git a/docling_core/types/doc/document.py b/docling_core/types/doc/document.py
@@ -2764,14 +2764,17 @@ def _image_fallback(item: TextItem):
                             "</figure>"
                         )
 
+                img_fallback = _image_fallback(item)
+
                 # If the formula is not processed correcty, use its image
                 if (
                     item.text == ""
                     and item.orig != ""
                     and image_mode == ImageRefMode.EMBEDDED
                     and len(item.prov) > 0
+                    and img_fallback is not None
                 ):
-                    text = _image_fallback(item)
+                    text = img_fallback
 
                 # Building a math equation in MathML format
                 # ref https://www.w3.org/TR/wai-aria-1.1/#math
@@ -2791,9 +2794,13 @@ def _image_fallback(item: TextItem):
                             "Malformed formula cannot be rendered. "
                             f"Error {err.__class__.__name__}, formula={math_formula}"
                         )
-                        if image_mode == ImageRefMode.EMBEDDED and len(item.prov) > 0:
-                            text = _image_fallback(item)
-                        else:
+                        if (
+                            image_mode == ImageRefMode.EMBEDDED
+                            and len(item.prov) > 0
+                            and img_fallback is not None
+                        ):
+                            text = img_fallback
+                        elif len(math_formula) > 0:
                             text = f"<pre>{math_formula}</pre>"
 
                 elif math_formula != "":

diff --git a/test/test_docling_doc.py b/test/test_docling_doc.py
@@ -775,6 +775,25 @@ def test_formula_mathml():
     assert doc_html == gt_html
 
 
+def test_formula_with_missing_fallback():
+    doc = DoclingDocument(name="Dummy")
+    bbox = BoundingBox.from_tuple((1, 2, 3, 4), origin=CoordOrigin.BOTTOMLEFT)
+    prov = ProvenanceItem(page_no=1, bbox=bbox, charspan=(0, 2))
+    doc.add_text(label=DocItemLabel.FORMULA, text="", orig="(II.24) 2 Imar", prov=prov)
+
+    actual = doc.export_to_html(
+        formula_to_mathml=True, html_head="", image_mode=ImageRefMode.EMBEDDED
+    )
+
+    expected = """<!DOCTYPE html>
+<html lang="en">
+
+<div class="formula-not-decoded">Formula not decoded</div>
+</html>"""
+
+    assert actual == expected
+
+
 def test_docitem_get_image():
     # Prepare the document
     doc = DoclingDocument(name="Dummy")