Debug: Add logging to diagnose api_data path issue on HF Spaces

server.py

@@ -85,13 +85,23 @@ if INITIAL_PROCESSED.exists() and not PERSISTENT_PROCESSED.exists():
 # Copy api_data (standings, games, ELO data) to persistent storage
 INITIAL_API_DATA = ROOT_DIR / "data" / "api_data"
 PERSISTENT_API_DATA = PERSISTENT_DIR / "api_data"
+logger.info(f"INITIAL_API_DATA path: {INITIAL_API_DATA}")
+logger.info(f"INITIAL_API_DATA exists: {INITIAL_API_DATA.exists()}")
+if INITIAL_API_DATA.exists():
+    logger.info(f"INITIAL_API_DATA files: {len(list(INITIAL_API_DATA.glob('*')))}")
+logger.info(f"PERSISTENT_API_DATA path: {PERSISTENT_API_DATA}")
+logger.info(f"PERSISTENT_API_DATA exists: {PERSISTENT_API_DATA.exists()}")
+
 if INITIAL_API_DATA.exists() and not PERSISTENT_API_DATA.exists():
     logger.info("Copying initial API data (standings, games) to persistent storage...")
     shutil.copytree(INITIAL_API_DATA, PERSISTENT_API_DATA)
+elif INITIAL_API_DATA.exists() and PERSISTENT_API_DATA.exists():
+    logger.info("API data already in persistent storage")
 
 # Update environment to use persistent paths
 os.environ["NBA_ML_DATA_DIR"] = str(PERSISTENT_DIR)
 os.environ["NBA_ML_MODELS_DIR"] = str(PERSISTENT_MODELS_DIR)
+logger.info(f"Set NBA_ML_DATA_DIR to: {PERSISTENT_DIR}")
 
 # Add project root to path
 sys.path.insert(0, str(ROOT_DIR))

src/prediction_pipeline.py

@@ -72,8 +72,13 @@ class PredictionPipeline:
             from src.config import API_CACHE_DIR
             
             games_path = API_CACHE_DIR / "all_games_summary.parquet"
+            logger.info(f"Looking for ELO data at: {games_path}")
+            logger.info(f"API_CACHE_DIR exists: {API_CACHE_DIR.exists()}")
+            if API_CACHE_DIR.exists():
+                logger.info(f"API_CACHE_DIR contents: {list(API_CACHE_DIR.glob('*.parquet'))[:5]}")
+            
             if not games_path.exists():
-                logger.warning("No historical game data found for ELO initialization")
+                logger.warning(f"No historical game data found for ELO initialization at {games_path}")
                 return
             
             games_df = pd.read_parquet(games_path)