Update app.py

app.py

@@ -6,7 +6,7 @@ import os
 
 os.environ["HF_HOME"] = "/tmp/hf"
 
-app = FastAPI(title="Telegram UID Multi-Search API")
+app = FastAPI(title="Telegram UID Multi-Search")
 
 app.add_middleware(
     CORSMiddleware,
@@ -17,14 +17,7 @@ app.add_middleware(
 
 DATASET = "CodeXDevloper/MergedTgDataset"
 
-# Folders ki priority - pehle V1, phir V2, phir baaki
-FOLDER_PRIORITY = [
-    "TG_DATA_PARTS",       # V1
-    "Telegram_V2",         # V2
-    None,                  # None = baaki sab (root files)
-]
-
-def get_files_for_folder(folder):
+def get_files(folder=None):
     """Kisi folder ki saari parquet files nikalo"""
     api = HfApi()
     files = api.list_repo_files(DATASET, repo_type="dataset")
@@ -32,47 +25,49 @@ def get_files_for_folder(folder):
     if folder:
         return [f for f in files if f.startswith(folder + "/") and f.endswith(".parquet")]
     else:
-        # Root ki files (jo kisi folder mein nahi hain)
         return [f for f in files if "/" not in f and f.endswith(".parquet")]
 
 @app.get("/")
 def root():
     return {
         "status": "running",
-        "dataset": DATASET,
         "usage": "/search?uid=385167100",
-        "search_order": ["TG_DATA_PARTS (V1)", "Telegram_V2", "Root files"]
+        "search_order": "Telegram_V2 → TG_DATA_PARTS → Root"
     }
 
 @app.get("/search")
-def search_uid(
-    uid: str = Query(..., description="Telegram UID")
-):
+def search_uid(uid: str = Query(..., description="Telegram UID")):
     """
-    AUTOMATIC MULTI-FOLDER SEARCH:
-    1. Pehle V1 (TG_DATA_PARTS) mein dhoondho
-    2. Nahi mila toh V2 (Telegram_V2) mein
-    3. Nahi mila toh root files mein
+    AUTO MULTI-FOLDER SEARCH:
+    1. Telegram_V2 mein dhoondho
+    2. Nahi mila toh TG_DATA_PARTS mein
+    3. Nahi mila toh Root files mein
     """
     try:
-        results = []  # Har folder ka result track karo
+        search_log = []
         
-        for folder in FOLDER_PRIORITY:
-            folder_name = folder if folder else "Root"
-            print(f"🔍 Searching in: {folder_name}")
-            
+        # Priority order
+        folders = [
+            ("Telegram_V2", "Telegram_V2"),
+            ("TG_DATA_PARTS", "TG_DATA_PARTS"),
+            (None, "Root")
+        ]
+        
+        for folder_key, folder_name in folders:
             # Folder ki files nikalo
-            files = get_files_for_folder(folder)
+            files = get_files(folder_key)
             
             if not files:
-                results.append({
+                search_log.append({
                     "folder": folder_name,
                     "status": "no_files",
                     "scanned": 0
                 })
                 continue
             
-            # Dataset load karo (streaming)
+            # Sirf 20 files per folder (RAM safe)
+            files = files[:20]
+            
             try:
                 ds = load_dataset(
                     DATASET,
@@ -81,19 +76,20 @@ def search_uid(
                     streaming=True
                 )
             except Exception as e:
-                results.append({
+                search_log.append({
                     "folder": folder_name,
                     "status": "load_error",
-                    "error": str(e)
+                    "error": str(e)[:100]
                 })
                 continue
             
-            # Is folder mein dhoondho
+            # Is folder mein scan karo
             scanned = 0
-            max_scan = 200000  # har folder mein 2 lakh rows
+            max_scan = 100000  # 1 lakh rows per folder
             
             for row in ds:
                 scanned += 1
+                
                 if uid == str(row.get("user_id", "")).strip():
                     # ✅ MIL GAYA
                     return {
@@ -101,7 +97,7 @@ def search_uid(
                         "uid": uid,
                         "found_in": folder_name,
                         "scanned_in_folder": scanned,
-                        "search_log": results + [{
+                        "search_log": search_log + [{
                             "folder": folder_name,
                             "status": "FOUND",
                             "scanned": scanned
@@ -123,8 +119,7 @@ def search_uid(
                 if scanned >= max_scan:
                     break
             
-            # Is folder mein nahi mila
-            results.append({
+            search_log.append({
                 "folder": folder_name,
                 "status": "not_found",
                 "scanned": scanned
@@ -134,8 +129,8 @@ def search_uid(
         return {
             "found": False,
             "uid": uid,
-            "search_log": results,
-            "message": "Poore dataset mein nahi mila"
+            "search_log": search_log,
+            "message": "Telegram_V2, TG_DATA_PARTS, aur Root - teeno mein nahi mila"
         }
     
     except Exception as e: