fix: escape pipes in HTML-derived table cells

packages/markitdown/src/markitdown/converters/_markdownify.py

@@ -6,6 +6,16 @@
 
 _PERCENT_ENCODED_OCTET = re.compile(r"%[0-9A-Fa-f]{2}")
 
+# Matches a pipe together with the (possibly empty) run of backslashes in front
+# of it, so that run can be doubled before the pipe is escaped. Mirrors the
+# escaping the CSV converter applies to its table cells.
+_TABLE_PIPE_RE = re.compile(r"(?<!\\)(\\*)\|")
+
+
+def _escape_table_pipes(text: str) -> str:
+    """Escape pipes so cell text cannot split a Markdown table row."""
+    return _TABLE_PIPE_RE.sub(lambda m: m.group(1) * 2 + r"\|", text)
+
 
 def _quote_path_preserving_percent_encoded_octets(path: str) -> str:
     """Quote a URL path while preserving existing %HH byte encodings."""
@@ -174,5 +184,13 @@ def convert_strike(self, el: Any, text: str, *args, **kwargs) -> str:
         """Obsolete <strike> is still in the wild; treat it like <s>/<del>."""
         return self.convert_s(el, text, *args, **kwargs)  # type: ignore
 
+    def convert_td(self, el: Any, text: str, *args, **kwargs) -> str:
+        """Same as usual, but escape pipes so they do not add columns."""
+        return super().convert_td(el, _escape_table_pipes(text), *args, **kwargs)  # type: ignore
+
+    def convert_th(self, el: Any, text: str, *args, **kwargs) -> str:
+        """Same as usual, but escape pipes so they do not add columns."""
+        return super().convert_th(el, _escape_table_pipes(text), *args, **kwargs)  # type: ignore
+
     def convert_soup(self, soup: Any) -> str:
         return super().convert_soup(soup)  # type: ignore

packages/markitdown/tests/test_html.py

@@ -346,3 +346,23 @@ def test_deeply_nested_html_fallback() -> None:
 
 if __name__ == "__main__":
     raise SystemExit(pytest.main([__file__]))
+
+
+@pytest.mark.parametrize(
+    "cell, expected",
+    [
+        ("a|b", r"a\|b"),
+        (r"a\|b", r"a\\\|b"),
+        ("|", r"\|"),
+        ("no pipe", "no pipe"),
+    ],
+)
+def test_table_cell_pipes_are_escaped(cell: str, expected: str) -> None:
+    markdown = _convert_html(
+        f"<table><tr><th>{cell}</th><th>h2</th></tr>"
+        f"<tr><td>{cell}</td><td>x</td></tr></table>"
+    )
+
+    rows = markdown.strip().splitlines()
+    assert rows[0] == f"| {expected} | h2 |"
+    assert rows[2] == f"| {expected} | x |"

packages/markitdown/tests/test_xlsx.py

@@ -542,3 +542,17 @@ def test_xlsx_show_zeroes_rename_is_scoped_to_sheet_view_tags() -> None:
 
 if __name__ == "__main__":
     raise SystemExit(pytest.main([__file__]))
+
+
+def test_xlsx_cell_pipes_do_not_add_columns() -> None:
+    workbook = openpyxl.Workbook()
+    sheet = workbook.active
+    sheet.append(["name", "value"])
+    sheet.append(["a|b", "c"])
+    stream = io.BytesIO()
+    workbook.save(stream)
+    stream.seek(0)
+
+    markdown = MarkItDown().convert_stream(stream, file_extension=".xlsx").markdown
+
+    assert r"| a\|b | c |" in markdown