Update app.py

app.py

@@ -280,7 +280,7 @@ def fused_emotion(t_scores: Dict[str, float], a_scores: Dict[str, float]) -> Tup
 # ============================================================
 # 추천
 # ============================================================
-def get_recommendations(user_text: str, emotion: str, top_n: int = 3) -> List[Dict]:
+def get_recommendations(user_input: str, emotion: str, top_n: int = 3) -> List[Dict]:
     with _data_lock:
         ready = _data_ready
         _df   = df
@@ -291,7 +291,7 @@ def get_recommendations(user_text: str, emotion: str, top_n: int = 3) -> List[Di
 
     try:
         session_w = _session.score_multiplier(emotion)
-        user_vec  = sbert_model.encode(user_text, convert_to_tensor=True, show_progress_bar=False)
+        user_vec  = sbert_model.encode(user_input, convert_to_tensor=True, show_progress_bar=False)
         cos_sims  = sbert_util.cos_sim(user_vec, _emb)[0]
         if torch.cuda.is_available():
             cos_sims = cos_sims.cpu()
@@ -324,15 +324,15 @@ def get_recommendations(user_text: str, emotion: str, top_n: int = 3) -> List[Di
 # ============================================================
 # 추천 결과 → JSON 렌더링
 # ============================================================
-def _render_books_json(user_text: str, emotion: str, combined: Dict[str, float], books: List[Dict]) -> str:
+def _render_books_json(user_input: str, emotion: str, combined: Dict[str, float], books: List[Dict]) -> str:
     if not books:
         return json.dumps({"error": "추천할 책을 찾지 못했습니다."}, ensure_ascii=False, indent=2)
 
     output = {
-        "user_text":     user_text,
+        "user_input":     user_input,
         "emotion":       emotion,
         "emotion_score": round(combined.get(emotion, 0.0), 3),
-        "recommendations": [
+        "recommendation_books": [
             {
                 "isbn":      b["isbn"],
                 "title":     b["title"],
@@ -428,17 +428,17 @@ def process_voice(audio_input):
             y = y / max_v
 
         stt_result = stt_model({"sampling_rate": sr, "raw": y})
-        user_text  = stt_result["text"].strip()
+        user_input  = stt_result["text"].strip()
 
-        if not user_text:
+        if not user_input:
             return json.dumps({"error": "❌ 음성이 인식되지 않았습니다."}, ensure_ascii=False, indent=2), [], ""
 
-        t_scores = text_emotion_scores(user_text)
+        t_scores = text_emotion_scores(user_input)
         a_scores = audio_emotion_scores(y, sr)
         top_label, combined = fused_emotion(t_scores, a_scores)
 
-        books      = get_recommendations(user_text, top_label, top_n=3)
-        books_json = _render_books_json(user_text, top_label, combined, books)
+        books      = get_recommendations(user_input, top_label, top_n=3)
+        books_json = _render_books_json(user_input, top_label, combined, books)
 
         return books_json, books, top_label