scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-07 08:26:10 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 89

fix(streaming_handler.py): raise stop iteration post-finish reason

Browse source
This commit is contained in:
Krrish Dholakia 2025-03-16 20:40:41 -07:00
parent bde9ae8a95
commit a99251a4ab
1 changed files with 9 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

13
litellm/litellm_core_utils/streaming_handler.py
View file

@ -898,7 +898,9 @@ class CustomStreamWrapper:
return model_response
# Default - return StopIteration
return model_response
if hasattr(model_response, "usage"):
self.chunks.append(model_response)
raise StopIteration
# flush any remaining holding chunk
if len(self.holding_chunk) > 0:
if model_response.choices[0].delta.content is None:
@ -937,7 +939,11 @@ class CustomStreamWrapper:
and model_response.choices[0].delta.audio is not None
):
return model_response
return
else:
if hasattr(model_response, "usage"):
self.chunks.append(model_response)
return
def _optional_combine_thinking_block_in_choices(
self, model_response: ModelResponseStream
@ -1538,7 +1544,7 @@ class CustomStreamWrapper:
else:
chunk = next(self.completion_stream)
if chunk is not None and chunk != b"":
verbose_logger.debug(
print_verbose(
f"PROCESSED CHUNK PRE CHUNK CREATOR: {chunk}; custom_llm_provider: {self.custom_llm_provider}"
)
response: Optional[ModelResponseStream] = self.chunk_creator(
@ -1611,7 +1617,6 @@ class CustomStreamWrapper:
None,
cache_hit,
)
if self.sent_stream_usage is False and self.send_stream_usage is True:
self.sent_stream_usage = True
return response