fix: fall back to w:t when w:del lacks w:delText in revision parsing (#34)

* fix: fall back to w:t when w:del lacks w:delText in revision parsing

Nonconforming producers sometimes leave plain w:t inside w:del, which
made such deletions appear in list_revisions() with empty text. When a
w:del contains no w:delText elements, extract text from w:t instead.
The fallback only fires on an empty w:delText node list, so conforming
documents behave identically.

* docs: clarify scope of w:delText fallback comment

Multi-reviewer pass confirmed the logic but flagged that the mixed-
content behavior (w:delText present alongside stray w:t reads only
w:delText) and the deliberate-fallback nature were documented only in
tests. State both in the code comment. Skipped as out of scope: nested
w:ins filtering in the fallback (doubly-malformed input only).

Author: pablospe

docx_editor/track_changes.py

@@ -1644,6 +1644,11 @@ def _parse_revision(self, elem, rev_type: Literal["insertion", "deletion"]) -> R
             text_elems = elem.getElementsByTagName("w:t")
         else:
             text_elems = elem.getElementsByTagName("w:delText")
+            if not text_elems:
+                # Deliberate interop fallback: nonconforming producers may
+                # leave plain w:t inside w:del. Fires only when the w:del has
+                # no w:delText at all; mixed content reads only w:delText.
+                text_elems = elem.getElementsByTagName("w:t")
 
         text_parts = [self._get_node_text(t_elem) for t_elem in text_elems]
 

tests/test_track_changes.py

@@ -1647,6 +1647,59 @@ def test_restore_deletion_converts_rsiddel_to_rsidr(self):
         assert not r_elems[0].hasAttribute("w:rsidDel")
 
 
+class TestParseRevisionDelTextFallback:
+    """Tests for w:delText -> w:t fallback when reading deletion text."""
+
+    def test_deletion_with_plain_wt_falls_back(self):
+        """Test that deletion text falls back to w:t when w:delText is absent."""
+        import defusedxml.minidom
+
+        # Nonconforming producers may leave plain w:t inside w:del
+        xml = """<?xml version="1.0"?>
+        <w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+            <w:del w:id="1" w:author="Foreign">
+                <w:r>
+                    <w:t>lost text</w:t>
+                </w:r>
+            </w:del>
+        </w:document>"""
+
+        mock_editor = MagicMock()
+        mock_editor.dom = defusedxml.minidom.parseString(xml)
+
+        manager = RevisionManager(mock_editor)
+
+        revisions = manager.list_revisions()
+        assert len(revisions) == 1
+        assert revisions[0].type == "deletion"
+        assert revisions[0].text == "lost text"
+
+    def test_deletion_with_deltext_unchanged(self):
+        """Test that the fallback does not fire when w:delText exists."""
+        import defusedxml.minidom
+
+        xml = """<?xml version="1.0"?>
+        <w:document xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main">
+            <w:del w:id="1" w:author="Test">
+                <w:r>
+                    <w:delText>proper</w:delText>
+                </w:r>
+                <w:r>
+                    <w:t>stray</w:t>
+                </w:r>
+            </w:del>
+        </w:document>"""
+
+        mock_editor = MagicMock()
+        mock_editor.dom = defusedxml.minidom.parseString(xml)
+
+        manager = RevisionManager(mock_editor)
+
+        revisions = manager.list_revisions()
+        assert len(revisions) == 1
+        assert revisions[0].text == "proper"
+
+
 def _split_wt_text_nodes(doc, target_text, tag="w:t"):
     """Reach into the DOM and split an element's TEXT_NODE into multiple siblings.