Handle numbering levels defined without an index

Author: mwilliamson

NEWS

@@ -6,6 +6,8 @@
 
 * Support disabling external file accesses using the external_file_access argument.
 
+* Handle numbering levels defined without an index.
+
 # 1.10.0
 
 * Add "Heading" and "Body" styles, as found in documents created by Apple Pages,

mammoth/docx/body_xml.py

@@ -288,6 +288,12 @@ def _read_numbering_properties(paragraph_style_id, element):
             if level is not None:
                 return level
 
+        # Some malformed documents define numbering levels without an index, and
+        # reference the numbering using a w:numPr element without a w:ilvl child.
+        # To handle such cases, we assume a level of 0 as a fallback.
+        if num_id is not None:
+            return numbering.find_level(num_id, "0")
+
         return None
 
     def _read_paragraph_indent(element):

mammoth/docx/numbering_xml.py

@@ -36,15 +36,29 @@ class _AbstractNumLevel(object):
 
 
 def _read_abstract_num_levels(element):
-    levels = map(_read_abstract_num_level, element.find_children("w:lvl"))
-    return dict(
-        (level.level_index, level)
-        for level in levels
-    )
+    levels = {}
+
+    # Some malformed documents define numbering levels without an index, and
+    # reference the numbering using a w:numPr element without a w:ilvl child.
+    # To handle such cases, we assume a level of 0 as a fallback.
+    level_without_index = None
+
+    for level_element in element.find_children("w:lvl"):
+        level = _read_abstract_num_level(level_element)
+        if level.level_index is None:
+            level.level_index = "0"
+            level_without_index = level
+        else:
+            levels[level.level_index] = level
+
+    if "0" not in levels and level_without_index is not None:
+        levels[level_without_index.level_index] = level_without_index
+
+    return levels
 
 
 def _read_abstract_num_level(element):
-    level_index = element.attributes["w:ilvl"]
+    level_index = element.attributes.get("w:ilvl")
     num_fmt = element.find_child_or_null("w:numFmt").attributes.get("w:val")
     is_ordered = num_fmt != "bullet"
     paragraph_style_id = element.find_child_or_null("w:pStyle").attributes.get("w:val")

tests/docx/body_xml_tests.py

@@ -138,15 +138,15 @@ def test_numbering_properties_in_paragraph_properties_takes_precedence_over_numb
 
         assert_equal("1", paragraph.numbering.level_index)
 
-    def test_numbering_properties_are_ignored_if_lvl_is_missing(self):
+    def test_when_numbering_properties_are_missing_level_then_level_of_0_is_assumed(self):
         paragraph_xml = self._paragraph_with_numbering_properties([
             xml_element("w:numId", {"w:val": "42"}),
         ])
 
-        numbering = _NumberingMap({"42": {"1": documents.numbering_level("1", True)}})
+        numbering = _NumberingMap({"42": {"0": documents.numbering_level("0", True)}})
         paragraph = _read_and_get_document_xml_element(paragraph_xml, numbering=numbering)
 
-        assert_equal(None, paragraph.numbering)
+        assert_equal(documents.numbering_level("0", True), paragraph.numbering)
 
     def test_numbering_properties_are_ignored_if_num_id_is_missing(self):
         paragraph_xml = self._paragraph_with_numbering_properties([

tests/docx/numbering_xml_tests.py

@@ -72,6 +72,63 @@ def test_num_referencing_non_existent_abstract_num_is_ignored():
     assert_equal(None, numbering.find_level("47", "0"))
 
 
+def test_given_no_other_levels_with_index_of_0_when_level_is_missing_ilvl_then_level_index_is_0():
+    element = xml_element("w:numbering", {}, [
+        xml_element("w:abstractNum", {"w:abstractNumId": "42"}, [
+            xml_element("w:lvl", {}, [
+                xml_element("w:numFmt", {"w:val": "decimal"}),
+            ]),
+        ]),
+        xml_element("w:num", {"w:numId": "47"}, [
+            xml_element("w:abstractNumId", {"w:val": "42"})
+        ])
+    ])
+
+    numbering = _read_numbering_xml_element(element)
+
+    assert_equal(True, numbering.find_level("47", "0").is_ordered)
+
+
+def test_given_previous_other_level_with_index_of_0_when_level_is_missing_ilvl_then_level_is_ignored():
+    element = xml_element("w:numbering", {}, [
+        xml_element("w:abstractNum", {"w:abstractNumId": "42"}, [
+            xml_element("w:lvl", {"w:ilvl": "0"}, [
+                xml_element("w:numFmt", {"w:val": "bullet"}),
+            ]),
+            xml_element("w:lvl", {}, [
+                xml_element("w:numFmt", {"w:val": "decimal"}),
+            ]),
+        ]),
+        xml_element("w:num", {"w:numId": "47"}, [
+            xml_element("w:abstractNumId", {"w:val": "42"})
+        ])
+    ])
+
+    numbering = _read_numbering_xml_element(element)
+
+    assert_equal(False, numbering.find_level("47", "0").is_ordered)
+
+
+def test_given_subsequent_other_level_with_index_of_0_when_level_is_missing_ilvl_then_level_is_ignored():
+    element = xml_element("w:numbering", {}, [
+        xml_element("w:abstractNum", {"w:abstractNumId": "42"}, [
+            xml_element("w:lvl", {}, [
+                xml_element("w:numFmt", {"w:val": "decimal"}),
+            ]),
+            xml_element("w:lvl", {"w:ilvl": "0"}, [
+                xml_element("w:numFmt", {"w:val": "bullet"}),
+            ]),
+        ]),
+        xml_element("w:num", {"w:numId": "47"}, [
+            xml_element("w:abstractNumId", {"w:val": "42"})
+        ])
+    ])
+
+    numbering = _read_numbering_xml_element(element)
+
+    assert_equal(False, numbering.find_level("47", "0").is_ordered)
+
+
 def test_when_abstract_num_has_num_style_link_then_style_is_used_to_find_num():
     numbering = _read_numbering_xml_element(
         xml_element("w:numbering", {}, [