scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-11-12 08:42:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 207

fix(caching): keep stream-replay class tuple free of rebinding for the type-discipline gate

Browse source
This commit is contained in:
Lakshya Dhariwal 2026-10-18 02:55:10 +05:30 • committed by GitHub
parent f00bc628a6
commit 24fd51b582
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 3 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
litellm/caching/caching_handler.py
View file

@ -184,7 +184,7 @@ def _should_defer_streaming_cache_hit_callbacks(*, cached_result: object) -> boo
"""
from litellm.responses.streaming_iterator import BaseResponsesAPIStreamingIterator
stream_replay_classes: tuple[type, ...] = (CustomStreamWrapper, BaseResponsesAPIStreamingIterator)
base_replay_classes: Final = (CustomStreamWrapper, BaseResponsesAPIStreamingIterator)
try:
from litellm.llms.anthropic.pass_through.messages.response_cache import (
CachedAnthropicMessagesStreamIterator,
@ -193,11 +193,8 @@ def _should_defer_streaming_cache_hit_callbacks(*, cached_result: object) -> boo
# The Anthropic Messages replay is only reachable through proxy-only
# extras (fastapi). A base SDK install can never hold one, so the
# isinstance check can safely go ahead without it.
pass
else:
stream_replay_classes = (*stream_replay_classes, CachedAnthropicMessagesStreamIterator)
return isinstance(cached_result, stream_replay_classes)
return isinstance(cached_result, base_replay_classes)
return isinstance(cached_result, (*base_replay_classes, CachedAnthropicMessagesStreamIterator))
def _prompt_tokens_details_as_mapping(details: "PromptTokensDetailsWrapper") -> Mapping[str, object]: