feat(mentions): normalize plus-sign variants in token matching

Author: sonwr

backend/src/null_engine/services/mention_extractor.py

@@ -25,7 +25,7 @@
 
 def _normalize(text: str) -> str:
     normalized = text.lower().strip()
-    for token in ("-", "_", "/", ".", ",", "!", "?", ":", ";", "(", ")", "[", "]", "{", "}", "\"", "'", "&", "＆"):
+    for token in ("-", "_", "/", ".", ",", "!", "?", ":", ";", "(", ")", "[", "]", "{", "}", "\"", "'", "&", "＆", "+", "＋"):
         normalized = normalized.replace(token, " ")
     return " ".join(normalized.split())
 

backend/tests/test_mention_extractor_unit.py

@@ -5,6 +5,7 @@ def test_normalize_collapses_punctuation_and_spaces() -> None:
     assert _normalize("  Roman-Empire!!  ") == "roman empire"
     assert _normalize("Dune/Prophecy") == "dune prophecy"
     assert _normalize("Law & Order") == "law order"
+    assert _normalize("Chaos + Order") == "chaos order"
 
 
 def test_fuzzy_match_handles_punctuation_variants() -> None: