feat: Vllm url support v1

Author: zhanweizhang7

apps/chat/serializers/chat.py

@@ -158,9 +158,12 @@ def generate_prompt(self, instance: dict, with_valid=True):
         q = prompt.replace("{userInput}", message)
         messages[-1]['content'] = q
 
-        model_exist = QuerySet(Model).filter(workspace_id=workspace_id, id=model_id).exists()
+        model_exist = QuerySet(Model).filter(workspace_id=workspace_id,
+                                             id=model_id,
+                                             model_type = "LLM"
+                                             ).exists()
         if not model_exist:
-            raise Exception(_("model does not exists"))
+            raise Exception(_("Model does not exists or is not an LLM model"))
 
         def process():
             model = get_model_instance_by_model_workspace_id(model_id=model_id, workspace_id=workspace_id)

apps/models_provider/impl/vllm_model_provider/model/reranker.py

@@ -28,10 +28,11 @@ def is_cache_model():
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
+        r_url = model_credential.get('api_url')[:-3] if model_credential.get('api_url').endswith('/v1') else model_credential.get('api_url')
         return VllmBgeReranker(
             model=model_name,
             api_key=model_credential.get('api_key'),
-            api_url=model_credential.get('api_url'),
+            api_url=r_url,
             params=model_kwargs,
             **model_kwargs
         )

apps/models_provider/impl/vllm_model_provider/model/whisper_sst.py

@@ -44,7 +44,9 @@ def check_auth(self):
             self.speech_to_text(audio_file)
 
     def speech_to_text(self, audio_file):
-        base_url = f"{self.api_url}/v1"
+
+        base_url = self.api_url if self.api_url.endswith('v1') else f"{self.api_url}/v1"
+
         try:
             client = OpenAI(
                 api_key=self.api_key,