extraction fix: ValueError in table spans (#685)

* extraction fix: ValueError in table spans * review code
adbar · Aug 30, 2024 · ba43965 · ba43965
1 parent b3aea4a
commit ba43965
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 1 deletion.
diff --git a/tests/unit_tests.py b/tests/unit_tests.py
@@ -1123,6 +1123,10 @@ def test_table_processing():
     result = extract(htmlstring, no_fallback=True, output_format='txt', config=ZERO_CONFIG, include_tables=True)
     assert "a | b | |" in result
 
+    htmlstring = '<html><body><article><table><tr><td span="2.1">a</td><td>b</td></tr><tr><td>c</td><td>d</td><td>e</td></tr></table></article></body></html>'
+    result = extract(htmlstring, no_fallback=True, output_format='txt', config=ZERO_CONFIG, include_tables=True)
+    assert "a | b | |" in result
+
     # MemoryError: https://github.com/adbar/trafilatura/issues/657
     htmlstring = '<html><body><article><table><tr><td colspan="9007199254740991">a</td><td>b</td></tr><tr><td>c</td><td>d</td><td>e</td></tr></table></article></body></html>'
     result = extract(htmlstring, no_fallback=True, output_format='txt', config=ZERO_CONFIG, include_tables=True)

diff --git a/trafilatura/xml.py b/trafilatura/xml.py
@@ -51,6 +51,8 @@
 
 HI_FORMATTING = {'#b': '**', '#i': '*', '#u': '__', '#t': '`'}
 
+MAX_TABLE_WIDTH = 1000
+
 
 # https://github.com/lxml/lxml/blob/master/src/lxml/html/__init__.py
 def delete_element(element: _Element, keep_tail: bool = True) -> None:
@@ -318,7 +320,11 @@ def process_element(element: _Element, returnlist: List[str], include_formatting
             if element.tag == "row":
                 cell_count = len(element.xpath(".//cell"))
                 # restrict columns to a maximum of 1000
-                max_span = min(int(element.get("colspan") or element.get("span", 1)), 1000)
+                span_info = element.get("colspan") or element.get("span")
+                if not span_info or not span_info.isdigit():
+                    max_span = 1
+                else:
+                    max_span = min(int(span_info), MAX_TABLE_WIDTH)
                 # row ended so draw extra empty cells to match max_span
                 if cell_count < max_span:
                     returnlist.append(f'{"|" * (max_span - cell_count)}\n')