PR: Harden Ollama Provider Config Parsing, Add Robust Logging & Diagnostics

backend/app/ai_providers/ollama.py

@@ -14,9 +14,23 @@ def provider_name(self) -> str:
         return "ollama"
 
     async def initialize(self, config: Dict[str, Any]) -> bool:
+        import logging
+        logger = logging.getLogger(__name__)
+
+        # Debug logging for server URL resolution
+        logger.info(f"🔧 Ollama provider initializing with config: {config}")
+
         self.server_url = config.get("server_url", "http://localhost:11434")
         self.api_key = config.get("api_key", "")
         self.server_name = config.get("server_name", "Default Ollama Server")
+
+        # Log what URL we're actually using
+        if self.server_url == "http://localhost:11434" and "server_url" not in config:
+            logger.warning(f"⚠️  Ollama provider defaulting to localhost! Config was: {config}")
+        else:
+            logger.info(f"✅ Ollama provider using server_url: {self.server_url}")
+
+        logger.info(f"🎯 Ollama provider initialized - server_name: {self.server_name}, server_url: {self.server_url}")
         return True
 
     async def get_models(self) -> List[Dict[str, Any]]:
@@ -41,13 +55,12 @@ async def generate_stream(self, prompt: str, model: str, params: Dict[str, Any])
             yield chunk
 
     async def chat_completion(self, messages: List[Dict[str, Any]], model: str, params: Dict[str, Any]) -> Dict[str, Any]:
-        print(f"Ollama chat_completion received {len(messages)} messages")
-        print(f"DETAILED MESSAGE INSPECTION:")
-        for i, msg in enumerate(messages):
-            print(f"  Message {i+1}: role={msg.get('role', 'unknown')}, content={msg.get('content', '')}")
-
+        print(f"🤖 OLLAMA CHAT_COMPLETION CALLED")
+        print(f"📊 Server URL: {self.server_url}")
+        print(f"📊 Server Name: {self.server_name}")
+        print(f"📊 Model: {model}")
+        print(f"📊 Messages: {len(messages)} messages")
         prompt = self._format_chat_messages(messages)
-        print(f"Formatted prompt (full):\n{prompt}")
 
         result = await self._call_ollama_api(prompt, model, params, is_streaming=False)
         if "error" not in result:
@@ -61,13 +74,7 @@ async def chat_completion(self, messages: List[Dict[str, Any]], model: str, para
         return result
 
     async def chat_completion_stream(self, messages: List[Dict[str, Any]], model: str, params: Dict[str, Any]) -> AsyncGenerator[Dict[str, Any], None]:
-        print(f"Ollama chat_completion_stream received {len(messages)} messages")
-        print(f"DETAILED MESSAGE INSPECTION (STREAMING):")
-        for i, msg in enumerate(messages):
-            print(f"  Message {i+1}: role={msg.get('role', 'unknown')}, content={msg.get('content', '')}")
-
         prompt = self._format_chat_messages(messages)
-        print(f"Formatted prompt (full):\n{prompt}")
 
         async for chunk in self._stream_ollama_api(prompt, model, params):
             if "error" not in chunk:
@@ -81,16 +88,23 @@ async def chat_completion_stream(self, messages: List[Dict[str, Any]], model: st
             yield chunk
 
     async def _call_ollama_api(self, prompt: str, model: str, params: Dict[str, Any], is_streaming: bool = False) -> Dict[str, Any]:
+        import logging
+        logger = logging.getLogger(__name__)
+
         payload_params = params.copy() if params else {}
         payload_params["stream"] = False
         payload = {"model": model, "prompt": prompt, **payload_params}
         headers = {'Content-Type': 'application/json'}
         if self.api_key:
             headers['Authorization'] = f'Bearer {self.api_key}'
 
+        # Log the actual URL being called
+        api_url = f"{self.server_url}/api/generate"
+        logger.debug(f"Making Ollama API call to: {api_url}")
+
         try:
             async with httpx.AsyncClient(timeout=60.0) as client:
-                response = await client.post(f"{self.server_url}/api/generate", json=payload, headers=headers)
+                response = await client.post(api_url, json=payload, headers=headers)
                 response.raise_for_status()
                 result = response.json()
                 return {
@@ -100,6 +114,36 @@ async def _call_ollama_api(self, prompt: str, model: str, params: Dict[str, Any]
                     "metadata": result,
                     "finish_reason": result.get("done") and "stop" or None
                 }
+        except httpx.ConnectError as e:
+            logger.error(f"❌ Cannot connect to Ollama server at {api_url}")
+            return {
+                "error": f"Cannot connect to Ollama server at {self.server_url}. "
+                        f"Please check if the server is running and accessible.",
+                "provider": "ollama",
+                "model": model,
+                "server_name": self.server_name,
+                "server_url": self.server_url
+            }
+        except httpx.HTTPStatusError as e:
+            if e.response.status_code == 404:
+                logger.error(f"❌ Model '{model}' not found on server {self.server_name}")
+                return {
+                    "error": f"Model '{model}' not found on Ollama server '{self.server_name}'. "
+                            f"Please check if the model is installed or use a different model.",
+                    "provider": "ollama",
+                    "model": model,
+                    "server_name": self.server_name,
+                    "server_url": self.server_url
+                }
+            else:
+                logger.error(f"❌ HTTP error {e.response.status_code} from server {self.server_name}")
+                return {
+                    "error": f"HTTP {e.response.status_code} error from Ollama server '{self.server_name}': {e.response.text}",
+                    "provider": "ollama",
+                    "model": model,
+                    "server_name": self.server_name,
+                    "server_url": self.server_url
+                }
         except Exception as e:
             return self._format_error(e, model)
 

backend/app/ai_providers/registry.py

@@ -35,6 +35,10 @@ def register_provider(self, name: str, provider_class: Type[AIProvider]) -> None
 
     async def get_provider(self, name: str, instance_id: str, config: Dict[str, Any]) -> AIProvider:
         """Get or create a provider instance."""
+        import logging
+        logger = logging.getLogger(__name__)
+        logger.debug(f"Provider registry: get_provider called - {name}, instance: {instance_id}")
+
         if name not in self._providers:
             raise ValueError(f"Provider '{name}' not registered")
 
@@ -43,9 +47,11 @@ async def get_provider(self, name: str, instance_id: str, config: Dict[str, Any]
 
         # Check if instance exists
         if instance_key in self._instances[name]:
+            logger.debug(f"Using existing provider instance: {instance_key}")
             return self._instances[name][instance_key]
 
         # Create new instance
+        logger.debug(f"Creating new provider instance: {instance_key}")
         provider = self._providers[name]()
         await provider.initialize(config)
         self._instances[name][instance_key] = provider

backend/app/api/v1/endpoints/ai_providers.py

@@ -18,6 +18,7 @@
 from app.models.user import User
 from app.ai_providers.registry import provider_registry
 from app.ai_providers.ollama import OllamaProvider
+from app.utils.json_parsing import safe_encrypted_json_parse, validate_ollama_settings_format, create_default_ollama_settings
 from app.schemas.ai_providers import (
     TextGenerationRequest,
     ChatCompletionRequest,
@@ -41,6 +42,12 @@ async def get_provider_instance_from_request(request, db):
         user_id = user_id.replace("-", "")
 
     logger = logging.getLogger(__name__)
+    print(f"🚀 PROVIDER REQUEST RECEIVED")
+    print(f"📊 Provider: {request.provider}")
+    print(f"📊 Settings ID: {request.settings_id}")
+    print(f"📊 Server ID: {request.server_id}")
+    print(f"📊 Model: {getattr(request, 'model', 'N/A')}")
+    print(f"📊 User ID: {user_id}")
     logger.info(f"Getting provider instance for: settings_id={request.settings_id}, user_id={user_id}")
     logger.info(f"Original user_id from request: {request.user_id}")
 
@@ -94,43 +101,63 @@ async def get_provider_instance_from_request(request, db):
 
         # Use the first setting found
         setting = settings[0]
-        logger.info(f"Using setting with ID: {setting['id'] if isinstance(setting, dict) else setting.id}")
+        logger.debug(f"Using setting with ID: {setting['id'] if isinstance(setting, dict) else setting.id}")
 
-        # Extract configuration from settings value
-        # Parse the JSON string if value is a string
+        # Extract configuration from settings value using robust parsing
         setting_value = setting['value'] if isinstance(setting, dict) else setting.value
+        setting_id = setting['id'] if isinstance(setting, dict) else setting.id
 
-        if isinstance(setting_value, str):
-            try:
-                # First parse
-                logger.info("Parsing settings value as JSON string")
-                value_dict = json.loads(setting_value)
-                logger.debug(f"Parsed JSON string into: {value_dict}")
-
-                # Check if the result is still a string (double-encoded JSON)
-                if isinstance(value_dict, str):
-                    logger.info("Value is double-encoded JSON, parsing again")
-                    value_dict = json.loads(value_dict)
-                    logger.debug(f"Parsed double-encoded JSON into: {value_dict}")
-
-                # Now value_dict should be a dictionary
-                if not isinstance(value_dict, dict):
-                    logger.error(f"Error: value_dict is not a dictionary: {type(value_dict)}")
+        # Use our robust JSON parsing utility that handles encryption issues
+        try:
+            value_dict = safe_encrypted_json_parse(
+                setting_value,
+                context=f"settings_id={request.settings_id}, user_id={user_id}",
+                setting_id=setting_id,
+                definition_id=request.settings_id
+            )
+
+            # Ensure we have a dictionary
+            if not isinstance(value_dict, dict):
+                logger.error(f"Parsed value is not a dictionary: {type(value_dict)}")
+                # For Ollama settings, provide a default structure
+                if 'ollama' in request.settings_id.lower():
+                    logger.warning("Creating default Ollama settings structure")
+                    value_dict = create_default_ollama_settings()
+                else:
                     raise HTTPException(
                         status_code=500,
-                        detail=f"Error parsing settings value: expected dict, got {type(value_dict)}. "
-                               f"Please check the format of your settings."
+                        detail=f"Settings value must be a dictionary, got {type(value_dict)}. "
+                               f"Setting ID: {setting_id}"
                     )
-            except json.JSONDecodeError as e:
-                logger.error(f"Error parsing JSON string: {e}")
+
+            logger.debug(f"Successfully parsed settings value for {request.settings_id}")
+
+        except ValueError as e:
+            logger.error(f"Failed to parse encrypted settings: {e}")
+            # For Ollama settings, provide helpful error message and fallback
+            if 'ollama' in request.settings_id.lower():
+                logger.info("Ollama settings parsing failed, using fallback configuration")
+                value_dict = create_default_ollama_settings()
+            else:
                 raise HTTPException(
                     status_code=500,
-                    detail=f"Error parsing settings value: {str(e)}. "
-                           f"Please ensure the settings value is valid JSON."
+                    detail=str(e)
                 )
-        else:
-            logger.info("Settings value is already a dictionary")
-            value_dict = setting_value
+        except Exception as e:
+            logger.error(f"Unexpected error parsing settings: {e}")
+            raise HTTPException(
+                status_code=500,
+                detail=f"Unexpected error parsing settings: {str(e)}. Setting ID: {setting_id}"
+            )
+
+        # Add specific validation for Ollama settings
+        if 'ollama' in request.settings_id.lower():
+            logger.info("Validating Ollama settings format")
+            if not validate_ollama_settings_format(value_dict):
+                logger.warning("Ollama settings format validation failed, using default structure")
+                value_dict = create_default_ollama_settings()
+            else:
+                logger.info("Ollama settings format validation passed")
 
         # Handle different provider configurations
         if request.provider == "openai":
@@ -207,32 +234,45 @@ async def get_provider_instance_from_request(request, db):
             logger.info(f"Created Groq config with API key")
         else:
             # Other providers (like Ollama) use servers array
-            logger.info("Processing server-based provider configuration")
+            logger.debug("Processing server-based provider configuration")
             servers = value_dict.get("servers", [])
-            logger.info(f"Found {len(servers)} servers in settings")
+            logger.debug(f"Found {len(servers)} servers in settings")
+
+            logger.debug("Processing server-based provider configuration")
+            servers = value_dict.get("servers", [])
+            logger.debug(f"Found {len(servers)} servers in settings")
 
             # Find the specific server by ID
-            logger.info(f"Looking for server with ID: {request.server_id}")
+            logger.debug(f"Looking for server with ID: '{request.server_id}'")
             server = next((s for s in servers if s.get("id") == request.server_id), None)
-            if not server and servers:
-                # If the requested server ID is not found but there are servers available,
-                # use the first server as a fallback
-                logger.warning(f"Server with ID {request.server_id} not found, using first available server as fallback")
-                server = servers[0]
-                logger.info(f"Using fallback server: {server.get('serverName')} ({server.get('id')})")
 
             if not server:
-                logger.error(f"No server found with ID: {request.server_id} and no fallback available")
-                raise HTTPException(
-                    status_code=404,
-                    detail=f"Server not found with ID: {request.server_id}. "
-                           f"Please check your server configuration or use a different server ID."
-                )
+                # Provide detailed error message about available servers
+                if servers:
+                    available_servers = [f"{s.get('serverName', 'Unknown')} (ID: {s.get('id', 'Unknown')})" for s in servers]
+                    available_list = ", ".join(available_servers)
+                    logger.error(f"❌ Server with ID '{request.server_id}' not found")
+                    logger.error(f"📋 Available servers: {available_list}")
+                    raise HTTPException(
+                        status_code=404,
+                        detail=f"Ollama server '{request.server_id}' not found. "
+                               f"Available servers: {available_list}. "
+                               f"Please select a valid server from your Ollama settings."
+                    )
+                else:
+                    logger.error(f"❌ No Ollama servers configured")
+                    raise HTTPException(
+                        status_code=404,
+                        detail="No Ollama servers are configured. "
+                               "Please add at least one Ollama server in your settings before using this provider."
+                    )
 
-            logger.info(f"Found server: {server.get('serverName')}")
+            logger.debug(f"Found server: {server.get('serverName')} (ID: {server.get('id')})")
 
             # Create provider configuration from server details
             server_url = server.get("serverAddress")
+            logger.debug(f"Server URL from settings: '{server_url}'")
+
             if not server_url:
                 logger.error(f"Server URL is missing for server: {server.get('id')}")
                 raise HTTPException(
@@ -246,11 +286,11 @@ async def get_provider_instance_from_request(request, db):
                 "api_key": server.get("apiKey", ""),
                 "server_name": server.get("serverName", "Unknown Server")
             }
-
-        logger.info(f"Created config with server_url: {config.get('server_url', 'N/A')}")
+            
+            logger.debug(f"Created server config: {config.get('server_name')} -> {config.get('server_url')}")
 
         # Get provider instance
-        logger.info(f"Getting provider instance for: {request.provider}, {request.server_id}")
+        logger.debug(f"Getting provider instance for: {request.provider}, {request.server_id}")
         provider_instance = await provider_registry.get_provider(
             request.provider,
             request.server_id,
@@ -716,6 +756,13 @@ async def chat_completion(request: ChatCompletionRequest, db: AsyncSession = Dep
     """
     logger = logging.getLogger(__name__)
     try:
+        print(f"🎯 CHAT COMPLETION ENDPOINT CALLED")
+        print(f"📊 Provider: {request.provider}")
+        print(f"📊 Settings ID: {request.settings_id}")
+        print(f"📊 Server ID: {request.server_id}")
+        print(f"📊 Model: {request.model}")
+        print(f"📊 User ID: {request.user_id}")
+        print(f"📊 Stream: {request.stream}")
         logger.info(f"Production chat endpoint called with: provider={request.provider}, settings_id={request.settings_id}, server_id={request.server_id}, model={request.model}")
         logger.debug(f"Messages: {request.messages}")
         logger.debug(f"Params: {request.params}")