fix(app): Multimodal file storage

api/app/services/conversation_service.py

@@ -21,6 +21,7 @@ from app.models.conversation_model import ConversationDetail
 from app.models.prompt_optimizer_model import RoleType
 from app.repositories.conversation_repository import ConversationRepository, MessageRepository
 from app.schemas.conversation_schema import ConversationOut
+from app.schemas.model_schema import ModelInfo
 from app.services import workspace_service
 from app.services.model_service import ModelConfigService
 
@@ -269,10 +270,11 @@ class ConversationService:
 
         return messages
 
-    def get_conversation_history(
+    async def get_conversation_history(
             self,
             conversation_id: uuid.UUID,
-            max_history: Optional[int] = None
+            max_history: Optional[int] = None,
+            api_config: Optional[ModelInfo] = None
     ) -> List[dict]:
         """
         Retrieve historical conversation messages formatted as dictionaries.
@@ -280,6 +282,7 @@ class ConversationService:
         Args:
             conversation_id (uuid.UUID): Conversation UUID.
             max_history (Optional[int]): Maximum number of messages to retrieve.
+            api_config (Optional[ModelInfo]): Model API configuration for multimodal processing.
 
         Returns:
             List[dict]: List of message dictionaries with keys 'role' and 'content'.
@@ -290,13 +293,37 @@ class ConversationService:
         )
 
         # 转换为字典格式
-        history = [
-            {
+        history = []
+        for msg in messages:
+            content = [{"type": "text", "text": msg.content}]
+            
+            # 处理 meta_data 中的 files
+            if msg.meta_data and msg.meta_data.get("files"):
+                files = msg.meta_data.get("files", [])
+                if api_config:
+                    # 使用 MultimodalService 处理文件
+                    from app.services.multimodal_service import MultimodalService
+                    multimodal_service = MultimodalService(self.db, api_config=api_config)
+                    
+                    # 将 files 转换为 FileInput 格式
+                    file_inputs = []
+                    for file in files:
+                        from app.schemas.app_schema import FileInput, TransferMethod
+                        file_input = FileInput(
+                            type=file.get("type"),
+                            transfer_method=TransferMethod.REMOTE_URL,
+                            url=file.get("url")
+                        )
+                        file_inputs.append(file_input)
+
+                    processed_files = await multimodal_service.history_process_files(files=file_inputs)
+
+                    content.extend(processed_files)
+            
+            history.append({
                 "role": msg.role,
-                "content": [{"type": "text", "text": msg.content}] + (msg.meta_data.get("files", []) if msg.meta_data else [])
-            }
-            for msg in messages
-        ]
+                "content": content
+            })
 
         return history
 
@@ -524,9 +551,18 @@ class ConversationService:
             type=ModelType(model_type)
         )
 
-        conversation_messages = self.get_conversation_history(
+        conversation_messages = await self.get_conversation_history(
             conversation_id=conversation_id,
-            max_history=20
+            max_history=20,
+            api_config=ModelInfo(
+                model_name=model_name,
+                provider=provider,
+                api_key=api_key,
+                api_base=api_base,
+                capability=api_config.capability,
+                is_omni=api_config.is_omni,
+                model_type=model_type
+            )
         )
         if len(conversation_messages) == 0:
             return ConversationOut(