fix(uspto): parse proportional and missing CALS colwidth

docling/backend/xml/uspto_backend.py

@@ -1525,6 +1525,14 @@ def __init__(self, input: str) -> None:
         self.empty_text = ""
         self._soup = BeautifulSoup(input, features="xml")
 
+    @staticmethod
+    def _parse_colwidth(colwidth: str | None) -> float:
+        """Parse CALS colwidth magnitude from fixed ("36pt"), proportional ("2*"), or absent values."""
+        if not colwidth:
+            return 1.0
+        match = re.match(r"\s*([0-9]*\.?[0-9]+)", colwidth)
+        return float(match.group(1)) if match else 1.0
+
     def _create_tg_range(self, tgs: list[ColInfo]) -> dict[int, ColInfoType]:
         """Create a unified range along the table groups.
 
@@ -1548,13 +1556,7 @@ def _create_tg_range(self, tgs: list[ColInfo]) -> dict[int, ColInfoType]:
             }
             offst = 0
             for info in tg["colinfo"]:
-                cw = info["colwidth"]
-                cw = re.sub("pt", "", cw, flags=re.I)
-                cw = re.sub("mm", "", cw, flags=re.I)
-                try:
-                    cw = int(cw)
-                except BaseException:
-                    cw = float(cw)
+                cw = self._parse_colwidth(info["colwidth"])
                 colinfo[itg]["colwidth"].append(cw)
                 colinfo[itg]["offset"].append(offst)
                 offst += cw

tests/test_backend_patent_uspto.py

@@ -143,6 +143,29 @@ def test_tables(tables):
     assert len(file_table.table_cells) == 130
 
 
+@pytest.mark.parametrize(
+    "colspecs",
+    [
+        '<colspec colname="1" colwidth="1*"/><colspec colname="2" colwidth="2*"/>',
+        '<colspec colname="1"/><colspec colname="2"/>',
+        '<colspec colname="1" colwidth="50pt"/><colspec colname="2" colwidth="60pt"/>',
+    ],
+)
+def test_table_colwidth_variants(colspecs):
+    """CALS colwidth may be proportional ("1*") or absent; both must parse, not drop the table."""
+    xml = (
+        f'<table><tgroup cols="2">{colspecs}'
+        "<tbody><row><entry>a</entry><entry>b</entry></row></tbody>"
+        "</tgroup></table>"
+    )
+    table = XmlTable(xml).parse()
+    assert table is not None
+    assert table.num_rows == 1
+    assert table.num_cols == 2
+    assert [cell.text for cell in table.table_cells] == ["a", "b"]
+    assert [cell.start_col_offset_idx for cell in table.table_cells] == [0, 1]
+
+
 def test_table_out_of_range_namest_does_not_crash():
     """An entry whose numeric namest points past the declared columns must degrade, not crash."""
     xml = (