fix: native function calling system prompt duplication

backend/open_webui/routers/ollama.py

@@ -1256,6 +1256,7 @@ async def generate_chat_completion(
     url_idx: Optional[int] = None,
     user=Depends(get_verified_user),
     bypass_filter: Optional[bool] = False,
+    bypass_system_prompt: bool = False,
     db: Session = Depends(get_session),
 ):
     if BYPASS_MODEL_ACCESS_CONTROL:
@@ -1295,7 +1296,8 @@ async def generate_chat_completion(
             system = params.pop("system", None)
 
             payload = apply_model_params_to_body_ollama(params, payload)
-            payload = apply_system_prompt_to_body(system, payload, metadata, user)
+            if not bypass_system_prompt:
+                payload = apply_system_prompt_to_body(system, payload, metadata, user)
 
         # Check if user has access to the model
         if not bypass_filter and user.role == "user":

backend/open_webui/routers/openai.py

@@ -800,6 +800,7 @@ async def generate_chat_completion(
     form_data: dict,
     user=Depends(get_verified_user),
     bypass_filter: Optional[bool] = False,
+    bypass_system_prompt: bool = False,
     db: Session = Depends(get_session),
 ):
     if BYPASS_MODEL_ACCESS_CONTROL:
@@ -830,7 +831,8 @@ async def generate_chat_completion(
             system = params.pop("system", None)
 
             payload = apply_model_params_to_body_openai(params, payload)
-            payload = apply_system_prompt_to_body(system, payload, metadata, user)
+            if not bypass_system_prompt:
+                payload = apply_system_prompt_to_body(system, payload, metadata, user)
 
         # Check if user has access to the model
         if not bypass_filter and user.role == "user":

backend/open_webui/utils/chat.py

@@ -166,6 +166,7 @@ async def generate_chat_completion(
     form_data: dict,
     user: Any,
     bypass_filter: bool = False,
+    bypass_system_prompt: bool = False,
 ):
     log.debug(f"generate_chat_completion: {form_data}")
     if BYPASS_MODEL_ACCESS_CONTROL:
@@ -237,7 +238,8 @@ async def generate_chat_completion(
                         yield chunk
 
                 response = await generate_chat_completion(
-                    request, form_data, user, bypass_filter=True
+                    request, form_data, user, bypass_filter=True,
+                    bypass_system_prompt=bypass_system_prompt
                 )
                 return StreamingResponse(
                     stream_wrapper(response.body_iterator),
@@ -248,7 +250,8 @@ async def generate_chat_completion(
                 return {
                     **(
                         await generate_chat_completion(
-                            request, form_data, user, bypass_filter=True
+                            request, form_data, user, bypass_filter=True,
+                            bypass_system_prompt=bypass_system_prompt
                         )
                     ),
                     "selected_model_id": selected_model_id,
@@ -267,6 +270,7 @@ async def generate_chat_completion(
                 form_data=form_data,
                 user=user,
                 bypass_filter=bypass_filter,
+                bypass_system_prompt=bypass_system_prompt,
             )
             if form_data.get("stream"):
                 response.headers["content-type"] = "text/event-stream"
@@ -283,6 +287,7 @@ async def generate_chat_completion(
                 form_data=form_data,
                 user=user,
                 bypass_filter=bypass_filter,
+                bypass_system_prompt=bypass_system_prompt,
             )
 
 

backend/open_webui/utils/middleware.py

@@ -3410,6 +3410,7 @@ async def process_chat_response(
                             request,
                             new_form_data,
                             user,
+                            bypass_system_prompt=True,
                         )
 
                         if isinstance(res, StreamingResponse):
@@ -3589,6 +3590,7 @@ async def process_chat_response(
                                 request,
                                 new_form_data,
                                 user,
+                                bypass_system_prompt=True,
                             )
 
                             if isinstance(res, StreamingResponse):