XML minimal diff saving bug with text markup

python/CHANGELOG.md

@@ -12,6 +12,10 @@
 - Added `Name.is_valid()` to check the basic validity of first and last names.
 - `Name.case_normalize()` now repairs lowercase initials (for example, `C.s.` to `C.S.`); this is applied automatically by `create_paper()` and `create_volume()`.
 
+### Changed
+
+- Fixed a bug with `xml.ensure_minimal_diff()` that would sometimes break with markup-containing tags.
+
 ### Removed
 
 - Removed `NameSpecification.case_normalize()` in favor of `.normalize()`.

python/acl_anthology/utils/xml.py

@@ -23,6 +23,10 @@
 TAGS_WITH_MARKUP = {
     "b",
     "i",
+    "u",
+    "sc",
+    "tt",
+    "a",
     "fixed-case",
     "title",
     "abstract",
@@ -280,6 +284,10 @@ def ensure_minimal_diff(elem: etree._Element, reference: etree._Element) -> None
         for key, value in attribs:
             elem.set(key, value)
 
+    # If element contains markup, we do NOT recurse into its children and stop here
+    if elem.tag in TAGS_WITH_MARKUP:
+        return
+
     # Sort child elements to match order in reference, if element allows reordering
     if elem.tag in TAGS_WITH_UNORDERED_CHILDREN:
         # Follow tag order in reference, with keys new in elem coming last

python/tests/utils/xml_test.py

@@ -205,6 +205,11 @@ def test_xml_indent(inp, out):
         "<title><i>Italics</i><b>Bold</b></title>",
         False,
     ),
+    (  # Formatting markup differences are not logically equivalent
+        "<abstract><b>Important</b> message</abstract>",
+        "<abstract>Important message</abstract>",
+        False,
+    ),
 )
 
 
@@ -388,6 +393,23 @@ def test_xml_assert_equals(a, b, should_be_equal):
              <attachment type="software" hash="079d4f4b">2022.acl-long.48.software.zip</attachment>
            </paper>""",
     ),
+    # Added markup after <tex-math>
+    (
+        "<abstract><tex-math>a</tex-math> <b>xxx</b></abstract>",
+        "<abstract><tex-math>a</tex-math> xxx</abstract>",
+        "<abstract><tex-math>a</tex-math> <b>xxx</b></abstract>",
+    ),
+    # Test <par/>
+    (
+        "<abstract>a<par/><b>xxx</b></abstract>",
+        "<abstract>a<par/>xxx</abstract>",
+        "<abstract>a<par/><b>xxx</b></abstract>",
+    ),
+    (
+        "<abstract><i>a</i><par/><b>xxx</b></abstract>",
+        "<abstract><i>a</i>\n<b>xxx</b></abstract>",
+        "<abstract><i>a</i><par/><b>xxx</b></abstract>",
+    ),
 )