Update main.py

main.py

@@ -23,12 +23,11 @@ POLLING_INTERVAL_SECONDS = 1  # How often to poll for updates
 # --- FastAPI App Initialization ---
 app = FastAPI(
     title="Replicate to OpenAI Compatibility Layer",
-    version="1.1.0 (Polling Strategy)",
+    version="1.1.1 (SyntaxError Fixed)",
 )
 
-# --- Pydantic Models for OpenAI Compatibility (No Changes) ---
+# --- Pydantic Models for OpenAI Compatibility ---
 
-# /v1/models endpoint
 class ModelCard(BaseModel):
     id: str
     object: str = "model"
@@ -39,7 +38,6 @@ class ModelList(BaseModel):
     object: str = "list"
     data: List[ModelCard] = []
 
-# /v1/chat/completions endpoint
 class ChatMessage(BaseModel):
     role: Literal["system", "user", "assistant", "tool"]
     content: Union[str, List[Dict[str, Any]]]
@@ -76,8 +74,10 @@ def format_tools_for_prompt(tools: List[Tool]) -> str:
     """Converts OpenAI tools to a string for the system prompt."""
     if not tools:
         return ""
+    
     prompt = "You have access to the following tools. To use a tool, respond with a JSON object in the following format:\n"
-    prompt += '{"type": "tool_call", "name": "tool_name", "arguments": {"arg_name": "value"}}\n\n"
+    # *** THIS IS THE CORRECTED LINE ***
+    prompt += '{"type": "tool_call", "name": "tool_name", "arguments": {"arg_name": "value"}}\n\n'
     prompt += "Available tools:\n"
     for tool in tools:
         prompt += json.dumps(tool.function.dict(), indent=2) + "\n"
@@ -128,7 +128,6 @@ def prepare_replicate_input(request: OpenAIChatCompletionRequest) -> Dict[str, A
 async def stream_replicate_with_polling(model_id: str, payload: dict):
     """
     Creates a prediction and then polls the 'get' URL to stream back results.
-    This is a reliable alternative to Replicate's native SSE stream.
     """
     url = f"https://api.replicate.com/v1/models/{model_id}/predictions"
     headers = {"Authorization": f"Bearer {REPLICATE_API_TOKEN}", "Content-Type": "application/json"}
@@ -181,7 +180,6 @@ async def stream_replicate_with_polling(model_id: str, payload: dict):
                         previous_output = current_output
             
             except httpx.HTTPStatusError as e:
-                # Don't stop polling on temporary network errors
                 print(f"Warning: Polling failed with status {e.response.status_code}, retrying...")
             except Exception as e:
                 yield f"data: {json.dumps({'error': f'Polling error: {str(e)}'})}\n\n"
@@ -218,10 +216,9 @@ async def create_chat_completion(request: OpenAIChatCompletionRequest):
     replicate_input = prepare_replicate_input(request)
 
     if request.stream:
-        # Use the new reliable polling-based streamer
         return EventSourceResponse(stream_replicate_with_polling(replicate_model_id, replicate_input))
     
-    # Synchronous request (no changes needed here)
+    # Synchronous request
     url = f"https://api.replicate.com/v1/models/{replicate_model_id}/predictions"
     headers = {"Authorization": f"Bearer {REPLICATE_API_TOKEN}", "Content-Type": "application/json", "Prefer": "wait=120"}