Merge pull request #3797 from BerriAI/litellm_fix_post_call_streaming_hooks

[Fix]  async_post_call_streaming_hook not triggered on proxy server

docs/my-website/docs/proxy/call_hooks.md

@@ -17,6 +17,8 @@ This function is called just before a litellm completion call is made, and allow
 ```python
 from litellm.integrations.custom_logger import CustomLogger
 import litellm
+from litellm.proxy.proxy_server import UserAPIKeyAuth, DualCache
+from typing import Optional, Literal
 
 # This file includes the custom callbacks for LiteLLM Proxy
 # Once defined, these can be passed in proxy_config.yaml
@@ -34,7 +36,7 @@ class MyCustomHandler(CustomLogger): # https://docs.litellm.ai/docs/observabilit
             "image_generation",
             "moderation",
             "audio_transcription",
-        ]) -> Optional[dict, str, Exception]: 
+        ]): 
         data["model"] = "my-new-model"
         return data 
 

litellm/proxy/custom_callbacks1.py

@@ -0,0 +1,64 @@
+from litellm.integrations.custom_logger import CustomLogger
+import litellm
+from litellm.proxy.proxy_server import UserAPIKeyAuth, DualCache
+from typing import Optional, Literal
+
+
+# This file includes the custom callbacks for LiteLLM Proxy
+# Once defined, these can be passed in proxy_config.yaml
+class MyCustomHandler(
+    CustomLogger
+):  # https://docs.litellm.ai/docs/observability/custom_callback#callback-class
+    # Class variables or attributes
+    def __init__(self):
+        pass
+
+    #### CALL HOOKS - proxy only ####
+
+    async def async_pre_call_hook(
+        self,
+        user_api_key_dict: UserAPIKeyAuth,
+        cache: DualCache,
+        data: dict,
+        call_type: Literal[
+            "completion",
+            "text_completion",
+            "embeddings",
+            "image_generation",
+            "moderation",
+            "audio_transcription",
+        ],
+    ):
+        return data
+
+    async def async_post_call_failure_hook(
+        self, original_exception: Exception, user_api_key_dict: UserAPIKeyAuth
+    ):
+        pass
+
+    async def async_post_call_success_hook(
+        self,
+        user_api_key_dict: UserAPIKeyAuth,
+        response,
+    ):
+        # print("in async_post_call_success_hook")
+        pass
+
+    async def async_moderation_hook(  # call made in parallel to llm api call
+        self,
+        data: dict,
+        user_api_key_dict: UserAPIKeyAuth,
+        call_type: Literal["completion", "embeddings", "image_generation"],
+    ):
+        pass
+
+    async def async_post_call_streaming_hook(
+        self,
+        user_api_key_dict: UserAPIKeyAuth,
+        response: str,
+    ):
+        # print("in async_post_call_streaming_hook")
+        pass
+
+
+proxy_handler_instance = MyCustomHandler()

litellm/proxy/proxy_config.yaml

@@ -20,16 +20,8 @@ model_list:
       api_base: https://exampleopenaiendpoint-production.up.railway.app/triton/embeddings
 
 general_settings:
-  store_model_in_db: true
   master_key: sk-1234
   alerting: ["slack"]
 
 litellm_settings:
-  success_callback: ["langfuse"]
-  failure_callback: ["langfuse"]
-  default_team_settings:
-    - team_id: 7bf09cd5-217a-40d4-8634-fc31d9b88bf4
-      success_callback: ["langfuse"]
-      failure_callback: ["langfuse"]
-      langfuse_public_key: "os.environ/LANGFUSE_DEV_PUBLIC_KEY"
-      langfuse_secret_key: "os.environ/LANGFUSE_DEV_SK_KEY"
+  callbacks: custom_callbacks1.proxy_handler_instance

litellm/proxy/proxy_server.py

@@ -3402,6 +3402,12 @@ async def async_data_generator(
     try:
         start_time = time.time()
         async for chunk in response:
+
+            ### CALL HOOKS ### - modify outgoing data
+            chunk = await proxy_logging_obj.async_post_call_streaming_hook(
+                user_api_key_dict=user_api_key_dict, response=chunk
+            )
+
             chunk = chunk.model_dump_json(exclude_none=True)
             try:
                 yield f"data: {chunk}\n\n"

litellm/proxy/utils.py

@@ -516,6 +516,27 @@ class ProxyLogging:
                 raise e
         return response
 
+    async def async_post_call_streaming_hook(
+        self,
+        response: Union[ModelResponse, EmbeddingResponse, ImageResponse],
+        user_api_key_dict: UserAPIKeyAuth,
+    ):
+        """
+        Allow user to modify outgoing streaming data -> per chunk
+
+        Covers:
+        1. /chat/completions
+        """
+        for callback in litellm.callbacks:
+            try:
+                if isinstance(callback, CustomLogger):
+                    await callback.async_post_call_streaming_hook(
+                        user_api_key_dict=user_api_key_dict, response=response
+                    )
+            except Exception as e:
+                raise e
+        return response
+
     async def post_call_streaming_hook(
         self,
         response: str,