fix(search): extract non-ASCII capitalized names in _extract_entities

_extract_entities used the ASCII-only class [A-Z][a-zA-Z]+ to pull name
entities from a query, so non-ASCII names were dropped ("İstanbul",
"Zürich" yielded nothing) or shredded ("São Paulo" -> only "Paulo"),
degrading query enhancement for non-English/accented searches. Match
Unicode words and keep the alphabetic, uppercase-initial ones; ASCII
behaviour (the word boundary already excludes camelCase mid-word
capitals) is unchanged.

services/search/query.py

@@ -34,8 +34,14 @@ def _extract_entities(query: str) -> Dict[str, List[str]]:
     cleaned = query
     if qtype:
         cleaned = re.sub(rf"^{qtype}\b", "", cleaned, flags=re.I).strip()
-    for token in re.findall(r"\b[A-Z][a-zA-Z]+\b", cleaned):
-        entities["names"].append(token)
+    # Unicode-aware capitalized-word (name) detection. The old [A-Z][a-zA-Z]+
+    # class missed non-ASCII names like "İstanbul"/"Zürich" (dropped) and
+    # "São" (shredded). Keep the ASCII behaviour — the word boundary already
+    # excludes camelCase mid-word capitals — by requiring an all-alphabetic
+    # token of length > 1 whose first character is uppercase.
+    for token in re.findall(r"\b\w+\b", cleaned):
+        if len(token) > 1 and token[0].isupper() and token.isalpha():
+            entities["names"].append(token)
     for year in re.findall(r"\b(?:19|20)\d{2}\b", cleaned):
         entities["dates"].append(year)
     month_day_year = re.findall(

tests/test_search_query_unicode_names.py

@@ -0,0 +1,31 @@
+"""Regression: _extract_entities must find non-ASCII capitalized names.
+
+The name extractor used the ASCII-only class [A-Z][a-zA-Z]+, so a query like
+"İstanbul weather" or "Zürich hotels" yielded no name entities at all, and
+"São Paulo" lost "São" — non-English/accented place and proper names were
+silently dropped from query enhancement. Detection is now Unicode-aware;
+ASCII behaviour (including camelCase mid-word capitals not counting as names)
+is preserved.
+"""
+from services.search.query import _extract_entities
+
+
+def _names(q):
+    return _extract_entities(q)["names"]
+
+
+def test_non_ascii_names_are_extracted():
+    assert "İstanbul" in _names("İstanbul weather")
+    assert "Zürich" in _names("Zürich hotels")
+    assert set(_names("trip to São Paulo")) >= {"São", "Paulo"}
+
+
+def test_ascii_names_unchanged():
+    assert _names("What did Alice do in 2024") == ["Alice"]
+    assert _names("news about OpenAI and Google") == ["OpenAI", "Google"]
+
+
+def test_lowercase_camelcase_and_numbers_are_not_names():
+    assert _names("the iphone price") == []
+    assert _names("iPhone price") == []       # mid-word capital is not a name
+    assert _names("top 50 albums") == []