fix parser

Author: luukunn

fastdeploy/entrypoints/openai/tool_parsers/ernie_x1_tool_parser.py

@@ -58,18 +58,14 @@ def __init__(self, tokenizer):
         self.tool_call_start_token_id = self.vocab.get(self.tool_call_start_token)
         self.tool_call_end_token_id = self.vocab.get(self.tool_call_end_token)
         if self.tool_call_start_token_id is None or self.tool_call_end_token_id is None:
-            raise RuntimeError(
-                "Hermes 2 Pro Tool parser could not locate tool call start/end " "tokens in the tokenizer!"
-            )
+            raise RuntimeError("Ernie x1 Tool parser could not locate tool call start/end " "tokens in the tokenizer!")
 
         if not self.model_tokenizer:
             raise ValueError(
                 "The model tokenizer must be passed to the ToolCallParser constructor during construction."
             )
 
-    def extract_tool_calls(
-        self, model_output: str, request: ChatCompletionRequest, model_status: str
-    ) -> ExtractedToolCallInformation:
+    def extract_tool_calls(self, model_output: str, request: ChatCompletionRequest) -> ExtractedToolCallInformation:
         """
         Extract the tool calls from a complete model response.
         Supports XML-style formats with newlines:
@@ -81,13 +77,10 @@ def extract_tool_calls(
         3. Only name and arguments field without content: {"name": "get_weather", "argume
         """
 
-        extract_content = model_output
-        if model_status == "tool_call_start":
-            extract_content = "<tool_call>" + model_output
         try:
-            if self.tool_call_start_token not in extract_content:
+            if self.tool_call_start_token not in model_output:
                 return ExtractedToolCallInformation(tools_called=False, tool_calls=[], content=model_output)
-            function_call_tuples = self.tool_call_regex.findall(extract_content)
+            function_call_tuples = self.tool_call_regex.findall(model_output)
 
             raw_function_calls = [json.loads(match[0] if match[0] else match[1]) for match in function_call_tuples]
 
@@ -116,7 +109,6 @@ def extract_tool_calls_streaming(
         current_token_ids: Sequence[int],
         delta_token_ids: Sequence[int],
         request: dict,
-        model_status: str,
     ) -> Union[DeltaMessage, None]:
 
         if self.tool_call_start_token_id not in current_token_ids:

fastdeploy/input/ernie4_5_processor.py

@@ -323,10 +323,10 @@ def process_response_dict_normal(self, response_dict, **kwargs):
                 response_dict["outputs"]["reasoning_content"] = reasoning_content
             if self.tool_parser_obj:
                 tool_parser = self.tool_parser_obj(self.tokenizer)
-                tool_call_info = tool_parser.extract_tool_calls(full_text, response_dict, model_status)
+                tool_call_info = tool_parser.extract_tool_calls(full_text, response_dict)
                 if tool_call_info.tools_called:
                     response_dict["outputs"]["tool_call"] = tool_call_info.tool_calls
-                response_dict["outputs"]["text"] = tool_call_info.content
+                    response_dict["outputs"]["text"] = tool_call_info.content
             response_dict["outputs"]["raw_prediction"] = full_text
             data_processor_logger.info(f"req_id:{req_id}, decode_status: {self.decode_status[req_id]}")
             del self.decode_status[req_id]
@@ -378,7 +378,6 @@ def process_response_dict_streaming(self, response_dict, **kwargs):
                 previous_token_ids + token_ids,
                 token_ids,
                 response_dict,
-                model_status,
             )
             if tool_call_delta_message is None or tool_call_delta_message.tool_calls:
                 response_dict["outputs"]["delta_message"] = tool_call_delta_message