Update config.py

config.py

@@ -80,7 +80,7 @@ class Config:
             "dir": self.directories["data"],
             "name": "legal_documents.db",
             "path": os.path.join(self.directories["data"], "legal_documents.db"),
-            "backup_interval": 3600 if self.is_hf_spaces else 86400,  # More frequent in HF Spaces
+            "backup_interval": 3600 if self.is_hf_spaces else 86400,
         }
     
     @property
@@ -91,7 +91,7 @@ class Config:
             "algorithm": "HS256",
             "access_token_expire_minutes": int(os.getenv("ACCESS_TOKEN_EXPIRE_MINUTES", "30")),
             "refresh_token_expire_days": int(os.getenv("REFRESH_TOKEN_EXPIRE_DAYS", "7")),
-            "bcrypt_rounds": 12 if not self.is_hf_spaces else 10,  # Lighter for HF Spaces
+            "bcrypt_rounds": 12 if not self.is_hf_spaces else 10,
         }
     
     @property
@@ -103,16 +103,16 @@ class Config:
             "workers": 1 if self.is_hf_spaces else int(os.getenv("WORKERS", "4")),
             "reload": self.is_development,
             "log_level": os.getenv("LOG_LEVEL", "info").lower(),
-            "access_log": not self.is_hf_spaces,  # Disable access log in HF Spaces
+            "access_log": not self.is_hf_spaces,
         }
     
     @property
     def ai_config(self) -> Dict[str, Any]:
-        """AI/ML configuration"""
+        """AI/ML configuration - Updated for stability"""
         return {
             "cache_dir": self.cache_dir,
-            "model_name": "microsoft/trocr-small-stage1" if self.is_hf_spaces else "microsoft/trocr-base-stage1",
-            "device": "cpu",  # Force CPU for compatibility
+            "ocr_model_name": "microsoft/trocr-base-printed",  # Changed to 'base-printed' as it's more robust
+            "device": "cpu",
             "max_workers": 1 if self.is_hf_spaces else 2,
             "batch_size": 1 if self.is_hf_spaces else 4,
             "timeout": 30 if self.is_hf_spaces else 60,
@@ -129,8 +129,7 @@ class Config:
             "socket_timeout": 5,
             "decode_responses": True,
             "retry_on_timeout": True,
-            "health_check_interval": 30,
-            "fallback_to_memory": True,  # Always fallback if Redis unavailable
+            "fallback_to_memory": True,
         }
     
     @property
@@ -140,7 +139,7 @@ class Config:
             "level": logging.INFO if not self.is_development else logging.DEBUG,
             "format": "%(asctime)s - %(name)s - %(levelname)s - %(message)s",
             "file": os.path.join(self.directories["logs"], "legal_dashboard.log") if not self.is_hf_spaces else None,
-            "max_bytes": 10 * 1024 * 1024,  # 10MB
+            "max_bytes": 10 * 1024 * 1024,
             "backup_count": 5,
         }
     
@@ -152,13 +151,14 @@ class Config:
             "DATABASE_PATH": self.database_config["path"],
             "PYTHONPATH": self.base_dir,
             
-            # AI/ML
+            # AI/ML - Unified cache paths
             "HF_HOME": self.cache_dir,
-            "TRANSFORMERS_CACHE": self.cache_dir,  # For backward compatibility
+            "TRANSFORMERS_CACHE": self.cache_dir,
             "HF_HUB_CACHE": self.cache_dir,
             "TORCH_HOME": self.cache_dir,
+            "HF_TOKEN": os.getenv("HF_TOKEN", ""),  # Added for authentication
             "TOKENIZERS_PARALLELISM": "false",
-            "CUDA_VISIBLE_DEVICES": "",  # Force CPU
+            "CUDA_VISIBLE_DEVICES": "",
             
             # Performance
             "OMP_NUM_THREADS": "1" if self.is_hf_spaces else "4",
@@ -181,7 +181,7 @@ class Config:
         
         for key, value in env_vars.items():
             os.environ[key] = value
-            if not key.startswith(("JWT_", "SECRET")):  # Don't log sensitive data
+            if not key.startswith(("JWT_", "SECRET", "HF_TOKEN")):
                 self.logger.info(f"🔧 {key}={value}")
             else:
                 self.logger.info(f"🔧 {key}=***")