fix(core): make normalize_text idempotent after mix-keyword removal

Author: stuchain

src/cuepoint/core/text_processing.py

@@ -106,6 +106,9 @@ def normalize_text(s: str) -> str:
         s,
     )
     s = re.sub(r"(?i)(originalmix|extendedmix|radioedit|clubmix)$", " ", s)
+    # Mix-keyword replacements can leave whitespace-only strings; strip so
+    # normalize(normalize(s)) == normalize(s) (e.g. "vip" alone → "" not " ").
+    s = re.sub(r"\s+", " ", s).strip()
     return s
 
 

src/tests/unit/core/test_text_processing.py

@@ -48,6 +48,12 @@ def test_normalize_text_empty(self):
         assert normalize_text("") == ""
         assert normalize_text(None) == ""
 
+    def test_normalize_text_idempotent_after_mix_keyword_strip(self) -> None:
+        """Regression (Design 3.16): lone token 'vip' must not normalize to a bare space."""
+        # NFKD + lower → "vip"; \\bvip\\b replacement leaves whitespace only → final strip → "".
+        once = normalize_text("vÌP")
+        assert normalize_text(once) == once
+
     @pytest.mark.unit
     @pytest.mark.skipif(
         sys.platform == "win32",