Update api/utils.py

api/utils.py

@@ -55,7 +55,6 @@ def strip_model_prefix(content: str, model_prefix: Optional[str] = None) -> str:
 
 # Process streaming response with headers from config.py
 async def process_streaming_response(request: ChatRequest):
-    # Generate a unique ID for this request
     request_id = f"chatcmpl-{uuid.uuid4()}"
     logger.info(f"Processing request with ID: {request_id} - Model: {request.model}")
 
@@ -63,24 +62,18 @@ async def process_streaming_response(request: ChatRequest):
     trending_agent_mode = TRENDING_AGENT_MODE.get(request.model, {})
     model_prefix = MODEL_PREFIXES.get(request.model, "")
 
-    # Adjust headers_api_chat since referer_url is removed
     headers_api_chat = get_headers_api_chat(BASE_URL)
 
+    # Delay for 'o1-preview' model if necessary
     if request.model == 'o1-preview':
         delay_seconds = random.randint(1, 60)
-        logger.info(
-            f"Introducing a delay of {delay_seconds} seconds for model 'o1-preview' "
-            f"(Request ID: {request_id})"
-        )
+        logger.info(f"Introducing a delay of {delay_seconds} seconds for model 'o1-preview' (Request ID: {request_id})")
         await asyncio.sleep(delay_seconds)
 
-    # Fetch the h-value for the 'validated' field
     h_value = await getHid()
     if not h_value:
         logger.error("Failed to retrieve h-value for validation.")
-        raise HTTPException(
-            status_code=500, detail="Validation failed due to missing h-value."
-        )
+        raise HTTPException(status_code=500, detail="Validation failed due to missing h-value.")
 
     json_data = {
         "agentMode": agent_mode,
@@ -89,13 +82,11 @@ async def process_streaming_response(request: ChatRequest):
         "clickedForceWebSearch": False,
         "codeModelMode": True,
         "githubToken": None,
-        "id": None,  # Using request_id instead of chat_id
+        "id": request_id,
         "isChromeExt": False,
         "isMicMode": False,
         "maxTokens": request.max_tokens,
-        "messages": [
-            message_to_dict(msg, model_prefix=model_prefix) for msg in request.messages
-        ],
+        "messages": [message_to_dict(msg, model_prefix=model_prefix) for msg in request.messages],
         "mobileClient": False,
         "playgroundTemperature": request.temperature,
         "playgroundTopP": request.top_p,
@@ -104,24 +95,17 @@ async def process_streaming_response(request: ChatRequest):
         "userId": None,
         "userSelectedModel": MODEL_MAPPING.get(request.model, request.model),
         "userSystemPrompt": None,
-        "validated": h_value,  # Dynamically set the validated field
+        "validated": h_value,
         "visitFromDelta": False,
         "webSearchModePrompt": False,
-        "imageGenerationMode": False,  # Added this line
+        "imageGenerationMode": False,
     }
 
-    response_content = ""  # Variable to hold the full response content
-    advertisement_added = False  # Track if advertisement is added
+    advertisement_added = False  # Track if advertisement is already added
 
     async with httpx.AsyncClient() as client:
         try:
-            async with client.stream(
-                "POST",
-                f"{BASE_URL}/api/chat",
-                headers=headers_api_chat,
-                json=json_data,
-                timeout=100,
-            ) as response:
+            async with client.stream("POST", f"{BASE_URL}/api/chat", headers=headers_api_chat, json=json_data, timeout=100) as response:
                 response.raise_for_status()
 
                 timestamp = int(datetime.now().timestamp())
@@ -131,39 +115,33 @@ async def process_streaming_response(request: ChatRequest):
                         if content.startswith("$@$v=undefined-rv1$@$"):
                             content = content[21:]  # Remove unwanted prefix
 
-                        # Remove blocked message if present
                         if BLOCKED_MESSAGE in content:
                             logger.info(f"Blocked message detected in response for Request ID {request_id}.")
                             content = content.replace(BLOCKED_MESSAGE, '').strip()
 
                         if not content:
-                            continue  # Skip if content is empty after removal
+                            continue  # Skip empty content
 
-                        # Clean up the content
                         cleaned_content = strip_model_prefix(content, model_prefix)
 
-                        # Add the chunk to the full response content
-                        response_content += cleaned_content
-
-                        # Yield the cleaned chunk as part of the stream
+                        # Yield cleaned chunk as part of the stream
                         yield f"data: {json.dumps(create_chat_completion_data(cleaned_content, request.model, timestamp))}\n\n"
 
-                # After processing all chunks, add the advertisement once (at the end)
+                # Append advertisement as a separate chunk with a line break, if applicable
                 if ADVERTISEMENT_TEXT and not advertisement_added:
-                    response_content += "\n\n" + ADVERTISEMENT_TEXT  # Add line break here
+                    advertisement_with_line_break = "\n" + ADVERTISEMENT_TEXT
+                    yield f"data: {json.dumps(create_chat_completion_data(advertisement_with_line_break, request.model, timestamp))}\n\n"
                     advertisement_added = True
 
-                # Yield the final chunk with the advertisement text appended at the end (this is the final response)
-                yield f"data: {json.dumps(create_chat_completion_data(response_content, request.model, timestamp, 'stop'))}\n\n"
-
-                # Add the final "done" marker
+                # Yield final chunk indicating the end of the stream
+                yield f"data: {json.dumps(create_chat_completion_data('', request.model, timestamp, 'stop'))}\n\n"
                 yield "data: [DONE]\n\n"
 
         except httpx.HTTPStatusError as e:
             logger.error(f"HTTP error occurred for Request ID {request_id}: {e}")
             raise HTTPException(status_code=e.response.status_code, detail=str(e))
         except httpx.RequestError as e:
-            logger.error(f"Error occurred during request for Request ID {request_id}: {e}")
+            logger.error(f"Request error occurred for Request ID {request_id}: {e}")
             raise HTTPException(status_code=500, detail=str(e))
 
 # Process non-streaming response with headers from config.py