fix(html): don't spread one run's formatting or link to a whole heading

docling/backend/html_backend.py

@@ -340,30 +340,23 @@ class AnnotatedText(BaseModel):
 
 class AnnotatedTextList(list):
     def to_single_text_element(self) -> AnnotatedText:
-        current_h = None
         current_text = ""
-        current_f = None
         current_code = False
         current_source_tag_id = None
+        # The single element can only keep a formatting or a hyperlink that
+        # applies to all of its text, e.g. not the link of one word in a heading.
+        text_runs = [at for at in self if at.text.strip()]
+        current_f = text_runs[0].formatting if text_runs else None
+        if any(at.formatting != current_f for at in text_runs):
+            current_f = None
+        current_h = text_runs[0].hyperlink if text_runs else None
+        if any(at.hyperlink != current_h for at in text_runs):
+            current_h = None
         for at in self:
             t = at.text
-            h = at.hyperlink
-            f = at.formatting
             c = at.code
             s = at.source_tag_id
             current_text += t.strip() + " "
-            if f is not None and current_f is None:
-                current_f = f
-            elif f is not None and current_f is not None and f != current_f:
-                _log.warning(
-                    f"Clashing formatting: '{f}' and '{current_f}'! Chose '{current_f}'"
-                )
-            if h is not None and current_h is None:
-                current_h = h
-            elif h is not None and current_h is not None and h != current_h:
-                _log.warning(
-                    f"Clashing hyperlinks: '{h}' and '{current_h}'! Chose '{current_h}'"
-                )
             if s is not None and current_source_tag_id is None:
                 current_source_tag_id = s
             elif (

tests/data/epub/groundtruth/epub_purvis_poetry.epub.json

@@ -1553,13 +1553,6 @@
       "prov": [],
       "orig": "To the Hibernia",
       "text": "To the Hibernia",
-      "formatting": {
-        "bold": false,
-        "italic": true,
-        "underline": false,
-        "strikethrough": false,
-        "script": "baseline"
-      },
       "level": 1
     },
     {

tests/data/epub/groundtruth/epub_purvis_poetry.epub.md

@@ -218,7 +218,7 @@ The idol of my soul-
 Then blame me not-this burst of grief  
 I cannot now control.
 
-## *To the Hibernia*
+## To the Hibernia
 
 O, speed thee! speed thee! gallant bark,  
 Across the bounding wave;  

tests/data/html/groundtruth/hyperlink_05.html.json

@@ -92,8 +92,7 @@
       "label": "caption",
       "prov": [],
       "orig": "This is an example caption for the image.",
-      "text": "This is an example caption for the image.",
-      "hyperlink": "#caption"
+      "text": "This is an example caption for the image."
     }
   ],
   "pictures": [

tests/data/html/groundtruth/wiki_duck.html.json

@@ -10830,8 +10830,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Pacific black duck displaying the characteristic upending \"duck\"",
-      "text": "Pacific black duck displaying the characteristic upending \"duck\"",
-      "hyperlink": "/wiki/Pacific_black_duck"
+      "text": "Pacific black duck displaying the characteristic upending \"duck\""
     },
     {
       "self_ref": "#/texts/308",
@@ -11582,8 +11581,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Male mallard .",
-      "text": "Male mallard .",
-      "hyperlink": "/wiki/Mallard"
+      "text": "Male mallard ."
     },
     {
       "self_ref": "#/texts/359",
@@ -11595,8 +11593,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Wood ducks .",
-      "text": "Wood ducks .",
-      "hyperlink": "/wiki/Wood_duck"
+      "text": "Wood ducks ."
     },
     {
       "self_ref": "#/texts/360",
@@ -12042,8 +12039,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Mallard landing in approach",
-      "text": "Mallard landing in approach",
-      "hyperlink": "/wiki/Mallard"
+      "text": "Mallard landing in approach"
     },
     {
       "self_ref": "#/texts/390",
@@ -12881,8 +12877,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Male Mandarin duck",
-      "text": "Male Mandarin duck",
-      "hyperlink": "/wiki/Mandarin_duck"
+      "text": "Male Mandarin duck"
     },
     {
       "self_ref": "#/texts/445",
@@ -13272,8 +13267,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Flying steamer ducks in Ushuaia , Argentina",
-      "text": "Flying steamer ducks in Ushuaia , Argentina",
-      "hyperlink": "/wiki/Flying_steamer_ducks"
+      "text": "Flying steamer ducks in Ushuaia , Argentina"
     },
     {
       "self_ref": "#/texts/475",
@@ -13658,8 +13652,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Female mallard in Cornwall , England",
-      "text": "Female mallard in Cornwall , England",
-      "hyperlink": "/wiki/Cornwall"
+      "text": "Female mallard in Cornwall , England"
     },
     {
       "self_ref": "#/texts/503",
@@ -13767,8 +13760,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Pecten along the bill",
-      "text": "Pecten along the bill",
-      "hyperlink": "/wiki/Pecten_(biology)"
+      "text": "Pecten along the bill"
     },
     {
       "self_ref": "#/texts/508",
@@ -13780,8 +13772,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Mallard duckling preening",
-      "text": "Mallard duckling preening",
-      "hyperlink": "/wiki/Mallard"
+      "text": "Mallard duckling preening"
     },
     {
       "self_ref": "#/texts/509",
@@ -14183,8 +14174,7 @@
       "label": "caption",
       "prov": [],
       "orig": "A Muscovy duckling",
-      "text": "A Muscovy duckling",
-      "hyperlink": "/wiki/Muscovy_duck"
+      "text": "A Muscovy duckling"
     },
     {
       "self_ref": "#/texts/539",
@@ -16109,8 +16099,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Indian Runner ducks , a common breed of domestic ducks",
-      "text": "Indian Runner ducks , a common breed of domestic ducks",
-      "hyperlink": "/wiki/Indian_Runner_duck"
+      "text": "Indian Runner ducks , a common breed of domestic ducks"
     },
     {
       "self_ref": "#/texts/675",
@@ -16420,15 +16409,7 @@
       "label": "caption",
       "prov": [],
       "orig": "Three black-colored ducks in the coat of arms of Maaninka [ 49 ]",
-      "text": "Three black-colored ducks in the coat of arms of Maaninka [ 49 ]",
-      "formatting": {
-        "bold": false,
-        "italic": false,
-        "underline": false,
-        "strikethrough": false,
-        "script": "super"
-      },
-      "hyperlink": "/wiki/Maaninka"
+      "text": "Three black-colored ducks in the coat of arms of Maaninka [ 49 ]"
     },
     {
       "self_ref": "#/texts/696",

tests/data/html/groundtruth/wiki_duck.html.md

@@ -321,7 +321,7 @@ Pecten along the bill
 
 <!-- image -->
 
-[Mallard duckling preening](/wiki/Mallard)
+Mallard duckling preening
 
 Ducks eat food sources such as [grasses](/wiki/Poaceae) , aquatic plants, fish, insects, small amphibians, worms, and small [molluscs](/wiki/Mollusc) .
 

tests/test_backend_html.py

@@ -168,6 +168,58 @@ def test_heading_levels():
     assert found_lvl_1 and found_lvl_2
 
 
+@pytest.mark.parametrize(
+    ("heading", "bold", "hyperlink"),
+    [
+        # Formatting or a link on one word does not apply to the whole heading
+        ("<h2>Section <strong>bold</strong> heading</h2>", False, None),
+        ("<h2><a href='https://example.com/'>Docs</a> and more</h2>", False, None),
+        # Formatting or a link on all of the text is kept
+        ("<h2><strong>All</strong> <strong>bold</strong></h2>", True, None),
+        (
+            "<h2><a href='https://example.com/'>Docs</a></h2>",
+            False,
+            "https://example.com/",
+        ),
+    ],
+)
+def test_heading_keeps_only_formatting_of_all_its_text(
+    heading: str, bold: bool, hyperlink: str | None
+):
+    html = f"<html><body>{heading}<p>text</p></body></html>".encode()
+    in_doc = InputDocument(
+        path_or_stream=BytesIO(html),
+        format=InputFormat.HTML,
+        backend=HTMLDocumentBackend,
+        filename="test",
+    )
+    doc = in_doc._backend.convert()
+    headers = [item for item in doc.texts if isinstance(item, SectionHeaderItem)]
+    assert len(headers) == 1
+    header = headers[0]
+    assert bool(header.formatting and header.formatting.bold) is bold
+    assert (str(header.hyperlink) if header.hyperlink else None) == hyperlink
+
+
+def test_caption_with_partial_link_has_no_hyperlink():
+    html = (
+        b"<html><body><figure><img src='a.png' alt=''>"
+        b"<figcaption>Female mallard in <a href='/wiki/Cornwall'>Cornwall</a>,"
+        b" England</figcaption></figure></body></html>"
+    )
+    in_doc = InputDocument(
+        path_or_stream=BytesIO(html),
+        format=InputFormat.HTML,
+        backend=HTMLDocumentBackend,
+        filename="test",
+    )
+    doc = in_doc._backend.convert()
+    captions = [item for item in doc.texts if item.label == DocItemLabel.CAPTION]
+    assert len(captions) == 1
+    assert captions[0].text.startswith("Female mallard in Cornwall")
+    assert captions[0].hyperlink is None
+
+
 def test_table_header_rowspan_without_body_does_not_crash():
     # A table whose only row is a `th` with rowspan (no body rows to span into)
     # used to raise IndexError: get_html_table_row_col counts no rows for an