Update app.py

app.py

@@ -24,29 +24,32 @@ client = OpenAI(
 DATASET_PATH = "data"
 JSON_PATH = f"{DATASET_PATH}/sign_language_data.json"
 
-
 if os.path.exists(JSON_PATH):
     with open(JSON_PATH, "r") as f:
         dataset = json.load(f)
 
     for item in dataset:
+        
         category = item["category"].lower().replace(" ", "_")
         
-        # Convert absolute paths to relative ones
-        if item["video_clip_path"].startswith("/content/sign_language_dataset/"):
-            item["video_clip_path"] = item["video_clip_path"].replace("/content/sign_language_dataset/", f"{DATASET_PATH}/")
+        
+        video_filename = os.path.basename(item["video_clip_path"])
+        item["video_clip_path"] = f"{DATASET_PATH}/clips/{category}/{video_filename}"
+        
+        
+        frame_filename = os.path.basename(item["frame_path"])
+        item["frame_path"] = f"{DATASET_PATH}/all_signs/{frame_filename}"
+
 
-        if item["frame_path"].startswith("/content/sign_language_dataset/"):
-            item["frame_path"] = item["frame_path"].replace("/content/sign_language_dataset/", f"{DATASET_PATH}/")
 else:
     
     dataset = []
     print(f"Warning: {JSON_PATH} does not exist. Using empty dataset.")
 
-# Configure logging
+
 logging.getLogger("sentence_transformers").setLevel(logging.ERROR)
 
-# Load embedding model
+
 print("Loading sentence transformer model...")
 embed_model = SentenceTransformer("all-MiniLM-L6-v2")
 
@@ -56,7 +59,7 @@ index = faiss.IndexFlatL2(dimension)
 text_to_video = {}
 idx_to_text = []
 
-# Add data to index
+
 for item in dataset:
     phrases = [item["text"]] + item.get("semantic_meaning", [])
 
@@ -76,7 +79,7 @@ def list_available_phrases():
 
 
 def preprocess_text(text):
-    # Remove emojis and special characters
+    
     emoji_pattern = re.compile("["
         u"\U0001F600-\U0001F64F"  
         u"\U0001F300-\U0001F5FF"  
@@ -149,7 +152,7 @@ def refine_sentence_with_deepseek(text):
 
 
 def retrieve_video(text, debug=False, similarity_threshold=0.7):
-    # Check for empty input
+    
     if not text or text.isspace():
         return None