修改流处理逻辑，修改 /api/tags错误

lightrag/api/lightrag_ollama.py

@@ -26,7 +26,7 @@ load_dotenv()
 # Constants for model information
 LIGHTRAG_NAME = "lightrag"
 LIGHTRAG_TAG = "latest"
-LIGHTRAG_MODEL = "{LIGHTRAG_NAME}:{LIGHTRAG_TAG}"
+LIGHTRAG_MODEL = "lightrag:latest"
 LIGHTRAG_SIZE = 7365960935
 LIGHTRAG_CREATED_AT = "2024-01-15T00:00:00Z"
 LIGHTRAG_DIGEST = "sha256:lightrag"
@@ -247,8 +247,25 @@ class OllamaChatResponse(BaseModel):
 class OllamaVersionResponse(BaseModel):
     version: str
 
+class OllamaModelDetails(BaseModel):
+    parent_model: str
+    format: str
+    family: str
+    families: List[str]
+    parameter_size: str
+    quantization_level: str
+
+class OllamaModel(BaseModel):
+    name: str
+    model: str
+    tag: str
+    size: int
+    digest: str
+    modified_at: str
+    details: OllamaModelDetails
+
 class OllamaTagResponse(BaseModel):
-    models: List[Dict[str, str]]
+    models: List[OllamaModel]
 
 # Original LightRAG models
 class QueryRequest(BaseModel):
@@ -632,26 +649,46 @@ def create_app(args):
     async def chat(request: OllamaChatRequest):
         """Handle chat completion requests"""
         try:
-            # Convert chat format to query
+            # 获取所有消息内容
-            query = request.messages[-1].content if request.messages else ""
+            messages = request.messages
+            if not messages:
+                raise HTTPException(status_code=400, detail="No messages provided")
             
-            # Parse query mode and clean query
+            # 获取最后一条消息作为查询
+            query = messages[-1].content
+            
+            # 解析查询模式
             cleaned_query, mode = parse_query_mode(query)
             
-            # Call RAG with determined mode
+            # 构建系统提示词（如果有历史消息）
-            response = await rag.aquery(
+            system_prompt = None
-                cleaned_query,
+            history_messages = []
-                param=QueryParam(
+            if len(messages) > 1:
+                # 如果第一条消息是系统消息，提取为system_prompt
+                if messages[0].role == "system":
+                    system_prompt = messages[0].content
+                    messages = messages[1:]
+                
+                # 收集历史消息（除了最后一条）
+                history_messages = [(msg.role, msg.content) for msg in messages[:-1]]
+            
+            # 调用RAG进行查询
+            kwargs = {
+                "param": QueryParam(
                     mode=mode,
-                    stream=request.stream
+                    stream=request.stream,
                 )
-            )
+            }
+            if system_prompt is not None:
+                kwargs["system_prompt"] = system_prompt
+            if history_messages:
+                kwargs["history_messages"] = history_messages
+                
+            response = await rag.aquery(cleaned_query, **kwargs)
 
             if request.stream:
                 async def stream_generator():
-                    result = ""
                     async for chunk in response:
-                        result += chunk
                         yield OllamaChatResponse(
                             model=LIGHTRAG_MODEL,
                             created_at=LIGHTRAG_CREATED_AT,
@@ -661,13 +698,13 @@ def create_app(args):
                             ),
                             done=False
                         )
-                    # Send final message
+                    # 发送一个空的完成消息
                     yield OllamaChatResponse(
                         model=LIGHTRAG_MODEL,
                         created_at=LIGHTRAG_CREATED_AT,
                         message=OllamaMessage(
                             role="assistant",
-                            content=result
+                            content=""
                         ),
                         done=True
                     )

start-server.sh

@@ -1 +1 @@
-lightrag-server --llm-binding openai --llm-model deepseek-chat --embedding-model "bge-m3:latest" --embedding-dim 1024
+lightrag-ollama --llm-binding openai --llm-model deepseek-chat --embedding-model "bge-m3:latest" --embedding-dim 1024