Add enhanced debugging output for data setup

- Add verbose logging to track data loading process
- Try both relative and absolute paths for mock data
- Display current working directory and file listings
- Show which data source is being used (GitHub or mock)
- Verify copied files after setup

setup_data.py

@@ -82,11 +82,20 @@ def fetch_data_from_github():
 
 def copy_mock_data():
     """Copy mock data to the expected extraction directory."""
+    # Try both relative and absolute paths
     mock_source = Path("mock_results") / CONFIG_NAME
+    if not mock_source.exists():
+        # Try absolute path in case we're in a different working directory
+        mock_source = Path("/app/mock_results") / CONFIG_NAME
+    
     target_dir = Path(EXTRACTED_DATA_DIR) / CONFIG_NAME
     
+    print(f"Current working directory: {os.getcwd()}")
+    print(f"Looking for mock data at: {mock_source.absolute()}")
+    
     if not mock_source.exists():
-        print(f"Warning: Mock data directory {mock_source} not found")
+        print(f"ERROR: Mock data directory {mock_source} not found!")
+        print(f"Directory contents: {list(Path('.').glob('*'))}")
         return False
     
     # Create target directory
@@ -97,7 +106,11 @@ def copy_mock_data():
     if target_dir.exists():
         shutil.rmtree(target_dir)
     shutil.copytree(mock_source, target_dir)
-    print(f"Mock data copied successfully. Files: {list(target_dir.glob('*'))}")
+    
+    # Verify the copy
+    copied_files = list(target_dir.glob('*'))
+    print(f"Mock data copied successfully. Files: {copied_files}")
+    print(f"Target directory: {target_dir.absolute()}")
     return True
 
 def setup_mock_data():
@@ -105,26 +118,34 @@ def setup_mock_data():
     Setup data for the leaderboard.
     First tries to fetch from GitHub, falls back to mock data if unavailable.
     """
+    print("=" * 60)
+    print("STARTING DATA SETUP")
+    print("=" * 60)
+    
     target_dir = Path(EXTRACTED_DATA_DIR) / CONFIG_NAME
     
     # Check if data already exists
     if target_dir.exists() and any(target_dir.glob("*.jsonl")):
+        jsonl_files = list(target_dir.glob("*.jsonl"))
         print(f"Data already exists at {target_dir}")
+        print(f"Found {len(jsonl_files)} JSONL files: {[f.name for f in jsonl_files]}")
         return
     
     # Try to fetch from GitHub first
-    print("Checking for data from GitHub repository...")
+    print("\n--- Attempting to fetch from GitHub ---")
     if fetch_data_from_github():
-        print("Using data from GitHub repository")
+        print("✓ Successfully using data from GitHub repository")
         return
     
     # Fall back to mock data
-    print("GitHub data not available, falling back to mock data...")
+    print("\n--- GitHub data not available, falling back to mock data ---")
     if copy_mock_data():
-        print("Using mock data")
+        print("✓ Successfully using mock data")
         return
     
+    print("\n" + "!" * 60)
     print("ERROR: No data available! Neither GitHub nor mock data could be loaded.")
+    print("!" * 60)
 
 if __name__ == "__main__":
     setup_mock_data()