Update app.py

app.py

@@ -170,11 +170,8 @@ def build_categorical_features(profiles):
 
     for col in CATEGORICAL_COLS:
         try:
-            # Get all values and sort for deterministic ordering
             unique_vals = sorted(set(p[col] for p in profiles if p[col]))
-            # Create value-to-index mapping (0-based, 0 reserved for unknown)
             val_to_idx = {val: idx+1 for idx, val in enumerate(unique_vals)}
-            # Encode each profile's value as integer (0 for unknown/missing)
             encoded = np.array([val_to_idx.get(p[col], 0) for p in profiles])
             cat_features.append(encoded.reshape(-1, 1))
         except Exception as e:
@@ -196,7 +193,6 @@ def build_text_features(profiles, vectorizers=None):
         try:
             corpus = [p[col] for p in profiles]
 
-            # Always create fresh vectorizer to match current corpus
             tfidf = TfidfVectorizer(
                 max_features=30,
                 stop_words='english',
@@ -235,7 +231,7 @@ def build_feature_matrix(profiles, vectorizers=None):
         features.append(ages_norm)
         logger.debug(f"Age features: {ages_norm.shape}")
         
-        # 2. Categorical features (one-hot encoded)
+        # 2. Categorical features
         logger.debug("Building categorical features...")
         cat_feat = build_categorical_features(profiles)
         if cat_feat.size > 0:
@@ -266,7 +262,6 @@ async def refresh_cache():
     try:
         logger.info("🔄 Refreshing profile cache...")
         
-        # Fetch profiles
         raw_profiles = await fetch_profiles_paginated()
         
         if not raw_profiles or len(raw_profiles) < 5:
@@ -274,7 +269,6 @@ async def refresh_cache():
             cache.error_message = f"Only {len(raw_profiles)} profiles, need at least 5"
             return False
         
-        # Preprocess
         logger.info("Preprocessing profiles...")
         profiles = []
         for p in raw_profiles:
@@ -292,7 +286,6 @@ async def refresh_cache():
         
         logger.info(f"Preprocessed {len(profiles)} profiles")
         
-        # Build features
         logger.info("Building feature matrix...")
         X, vectorizers, scaler = build_feature_matrix(profiles, cache.tfidf_vectorizers)
         
@@ -301,7 +294,6 @@ async def refresh_cache():
             cache.error_message = "Feature matrix is empty"
             return False
         
-        # Update cache
         cache.profiles = profiles
         cache.feature_matrix = X
         cache.profile_ids = [p['id'] for p in profiles]
@@ -397,7 +389,6 @@ async def recommend_query(
     Returns raw list: [{user_id, score, reasons}, ...]
     """
     matches = await _compute_matches(user_id, top_n, min_score)
-    # Transform to AiMatchService format
     return [
         {
             "user_id": m['id'],
@@ -480,7 +471,7 @@ async def _compute_matches(user_id: str, top_n: int, min_score: float):
     # Top N
     top_positions = np.argsort(sims)[::-1][:top_n]
 
-    # Build result list with full candidate data + score
+    # Build result list
     result = []
     for pos in top_positions:
         score = float(sims[pos])
@@ -498,7 +489,20 @@ async def _compute_matches(user_id: str, top_n: int, min_score: float):
             'score': score
         })
 
+    # ── Display recommendations in terminal ──
     logger.info(f"✓ Generated {len(result)} recommendations for {user_id}")
+    logger.info(f"┌{'─'*60}")
+    logger.info(f"│ Requesting user : {user['full_name']} ({user['gender']}, {user['age']})")
+    logger.info(f"├{'─'*60}")
+    for rank, r in enumerate(result, 1):
+        logger.info(
+            f"│ #{rank:<2} {r['full_name']:<25} "
+            f"Age: {r['age']:<4} "
+            f"City: {r['city']:<15} "
+            f"Score: {r['score']*100:.1f}%"
+        )
+    logger.info(f"└{'─'*60}")
+
     return result