Sync Flask API from GitHub 782d7c9418536a8a930bf3f588b6f5d5b1fd593d

README.md

@@ -1,6 +1,5 @@
 ---
 title: Bogor Xplore API
-emoji: 🏞️
 colorFrom: green
 colorTo: blue
 sdk: docker

app.py

@@ -9,9 +9,9 @@ CORS(app)  # Enable CORS for Laravel
 BASE_DIR = os.path.dirname(os.path.abspath(__file__))
 
 # Initialize recommender
-print("🚀 Initializing Tourism Recommender...")
+print("Initializing Tourism Recommender...")
 recommender = TourismRecommender(data_path=os.path.join(BASE_DIR, 'data'))
-print("✅ Ready!")
+print("Ready!")
 
 @app.route('/')
 def index():
@@ -103,7 +103,7 @@ def get_recommendations():
         place_details = None
         top_n = data.get('top_n', 10)
         
-        print(f"🔍 REQUEST: {data}")
+        print(f"REQUEST: {data}")
 
         # 1. Try lookup by NAME first (Most reliable if IDs don't match)
         if 'place_name' in data and data['place_name']:
@@ -113,7 +113,7 @@ def get_recommendations():
                 place_idx = search_result['query_idx']
                 # Get details using this index
                 place_details = recommender.get_place_by_id(place_idx)
-                print(f"   ✅ Found by name: Index {place_idx} -> {place_details['nama']}")
+                print(f"   Found by name: Index {place_idx} -> {place_details['nama']}")
         
         # 2. Fallback to ID if provided and Name lookup failed
         if place_idx is None and 'place_id' in data:
@@ -125,14 +125,14 @@ def get_recommendations():
                 if 0 <= potential_idx < len(recommender.df):
                     place_idx = potential_idx
                     place_details = recommender.get_place_by_id(place_idx)
-                    print(f"   ✅ Using ID as Index: Index {place_idx} -> {place_details['nama']}")
+                    print(f"   Using ID as Index: Index {place_idx} -> {place_details['nama']}")
                 else:
-                    print(f"   ❌ ID {potential_idx} is out of bounds (0-{len(recommender.df)-1})")
+                    print(f"   ID {potential_idx} is out of bounds (0-{len(recommender.df)-1})")
             except ValueError:
                 pass
         
         if place_idx is None:
-            print("   ❌ Place not found")
+            print("   Place not found")
             return jsonify({
                 'status': 'error',
                 'message': 'Place not found. Provide valid place_name or place_id.'
@@ -151,7 +151,7 @@ def get_recommendations():
         })
     except Exception as e:
         import traceback
-        print(f"❌ API ERROR: {str(e)}")
+        print(f"API ERROR: {str(e)}")
         traceback.print_exc()
         return jsonify({
             'status': 'error',

data/check_ids.py

@@ -18,14 +18,14 @@ try:
         # Check alignment
         mismatch = df[df.index != df['id']]
         if not mismatch.empty:
-            print(f"\n⚠️ Mismatch found! {len(mismatch)} rows have Index != ID")
+            print(f"\nMismatch found! {len(mismatch)} rows have Index != ID")
             print("Example Mismatch:")
             print(mismatch[['id', 'nama']].head(3))
         else:
-            print("\n✅ Index matches ID exactly!")
+            print("\nIndex matches ID exactly!")
             
     else:
-        print("\n⚠️ 'id' column NOT FOUND in CSV. Pure index usage.")
+        print("\n'id' column NOT FOUND in CSV. Pure index usage.")
         
 except Exception as e:
     print(f"Error: {e}")

models/recommender.py

@@ -24,7 +24,7 @@ class TourismRecommender:
     """
     
     def __init__(self, data_path='data/'):
-        print("🚀 Initializing Tourism Recommender...")
+        print("Initializing Tourism Recommender...")
         
         self.data_path = data_path
         
@@ -37,7 +37,7 @@ class TourismRecommender:
         indobert_embeddings_path = os.path.join(data_path, 'indobert_embeddings.npy')
         
         # Load data
-        print("   📂 Loading pre-computed data from dataset...")
+        print("   Loading pre-computed data from dataset...")
         self.df = pd.read_csv(data_csv_path)
 
         if 'deskripsi_clean' in self.df.columns:
@@ -62,8 +62,8 @@ class TourismRecommender:
         # Load similarity matrices
         self.ngram_sim = self._expand_square_matrix(np.load(ngram_similarity_path), 'ngram_similarity')
         self.indobert_sim = self._expand_square_matrix(np.load(indobert_similarity_path), 'indobert_similarity')
-        print(f"   ✅ Loaded ngram_similarity: {self.ngram_sim.shape}")
-        print(f"   ✅ Loaded indobert_similarity: {self.indobert_sim.shape}")
+        print(f"   Loaded ngram_similarity: {self.ngram_sim.shape}")
+        print(f"   Loaded indobert_similarity: {self.indobert_sim.shape}")
         
         # Load TF-IDF untuk query search (if needed)
         self.tfidf_matrix = self._expand_rows(np.load(tfidf_matrix_path), 'tfidf_matrix')
@@ -75,15 +75,15 @@ class TourismRecommender:
         self.indobert_embeddings = self._expand_rows(np.load(indobert_embeddings_path), 'indobert_embeddings')
         
         # Load IndoBERT model untuk query embedding
-        print("   📥 Loading IndoBERT model for query encoding...")
+        print("   Loading IndoBERT model for query encoding...")
         self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
         self.tokenizer = AutoTokenizer.from_pretrained("indobenchmark/indobert-base-p1")
         self.model = AutoModel.from_pretrained("indobenchmark/indobert-base-p1")
         self.model.eval()
         self.model.to(self.device)
-        print(f"   ✅ IndoBERT model loaded! Device: {self.device}")
+        print(f"   IndoBERT model loaded! Device: {self.device}")
         
-        print(f"\n✅ Recommender ready!")
+        print(f"\nRecommender ready!")
         print(f"   Total destinations: {len(self.df)}")
         print(f"   Method: IndoBERT (Search) & N-Gram (Detail Recommendations)")
 
@@ -99,7 +99,7 @@ class TourismRecommender:
             expanded = np.zeros((target_len, target_len), dtype=matrix.dtype)
             expanded[np.ix_(self.valid_text_indices, self.valid_text_indices)] = matrix
             np.fill_diagonal(expanded, 1.0)
-            print(f"   ↔️ Expanded {name}: {matrix.shape} -> {expanded.shape}")
+            print(f"   Expanded {name}: {matrix.shape} -> {expanded.shape}")
             return expanded
 
         raise ValueError(
@@ -120,7 +120,7 @@ class TourismRecommender:
             expanded_shape = (target_len, *values.shape[1:])
             expanded = np.zeros(expanded_shape, dtype=values.dtype)
             expanded[self.valid_text_indices] = values
-            print(f"   ↔️ Expanded {name}: {values.shape} -> {expanded.shape}")
+            print(f"   Expanded {name}: {values.shape} -> {expanded.shape}")
             return expanded
 
         raise ValueError(
@@ -183,7 +183,7 @@ class TourismRecommender:
         """
         # Validate index
         if place_idx is None or place_idx < 0 or place_idx >= len(self.df):
-            print(f"❌ Invalid Place Index: {place_idx}")
+            print(f"Invalid Place Index: {place_idx}")
             return []
         
         # Use pre-computed N-Gram similarity matrix directly
@@ -316,14 +316,14 @@ class TourismRecommender:
             
             if place_idx is not None:
                 # Use pre-computed IndoBERT similarity matrix (like notebook)
-                print(f"   📊 Using IndoBERT similarity matrix for place: {self.df.iloc[place_idx]['nama']}")
+                print(f"   Using IndoBERT similarity matrix for place: {self.df.iloc[place_idx]['nama']}")
                 sim_scores = self.indobert_sim[place_idx]
                 
                 # Get top N (excluding itself at index 0)
                 top_indices = sim_scores.argsort()[::-1][1:top_n+1]
             else:
                 # Fall back to on-the-fly query encoding
-                print(f"   🔍 Query not found as place name, using query encoding...")
+                print(f"   Query not found as place name, using query encoding...")
                 query_emb = self._get_query_embedding(query)
                 query_emb = query_emb.reshape(1, -1)
                 sim_scores = cosine_similarity(query_emb, self.indobert_embeddings)[0]
@@ -333,7 +333,7 @@ class TourismRecommender:
             for idx in top_indices:
                 # SAFEGUARD: Ignore indices that are out of bounds
                 if idx >= len(self.df):
-                    print(f"   ⚠️ Ignored out-of-bounds index: {idx}")
+                    print(f"   Ignored out-of-bounds index: {idx}")
                     continue
                     
                 place = self.df.iloc[idx]