scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-10-13 03:59:12 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 66

refactor(responses): share streamed annotation conversion

Browse source
This commit is contained in:
jinwukong 2026-10-06 13:13:03 +08:00
parent 2551344a8a
commit dd16ab98ab
1 changed files with 14 additions and 25 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

39
litellm/responses/litellm_completion_transformation/streaming_iterator.py
View file

@ -28,6 +28,7 @@ from litellm.responses.utils import ResponsesAPIRequestUtils
from litellm.types.llms.openai import (
PART_UNION_TYPES,
BaseLiteLLMOpenAIResponseObject,
ChatCompletionAnnotation,
ContentPartAddedEvent,
ContentPartDoneEvent,
ContentPartDonePartOutputText,
@ -823,11 +824,7 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
text: Final = getattr(litellm_complete_object.choices[0].message, "content", "") or ""
annotations: Final = getattr(litellm_complete_object.choices[0].message, "annotations", None)
response_annotations: Final = self._message_annotations(
LiteLLMCompletionResponsesConfig.transform_chat_completion_annotations_to_response_output_annotations(
annotations=annotations
)
)
response_annotations: Final = self._message_annotations(annotations)
part: Final[PART_UNION_TYPES] = ContentPartDonePartOutputText(
type="output_text",
text=text,
@ -850,11 +847,7 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
text: Final = litellm_complete_object.choices[0].message.content or ""
annotations = getattr(litellm_complete_object.choices[0].message, "annotations", None)
response_annotations: Final = self._message_annotations(
LiteLLMCompletionResponsesConfig.transform_chat_completion_annotations_to_response_output_annotations(
annotations=annotations
)
)
response_annotations: Final = self._message_annotations(annotations)
return OutputItemDoneEvent(
type=ResponsesAPIStreamEvents.OUTPUT_ITEM_DONE,
output_index=self._message_output_index,
@ -880,10 +873,11 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
def _message_annotations(
self,
annotations: list[GenericResponseOutputItemContentAnnotation],
annotations: list[ChatCompletionAnnotation] | None,
*,
start: int | None = None,
end: int | None = None,
positioned_only: bool = False,
) -> list[GenericResponseOutputItemContentAnnotation]:
offset: Final = self._closed_message_text_length if start is None else start
return [
@ -894,8 +888,10 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
if value is not None
}
)
for annotation in annotations
if (annotation.start_index is None and end is None)
for annotation in LiteLLMCompletionResponsesConfig.transform_chat_completion_annotations_to_response_output_annotations(
annotations=annotations
)
if (annotation.start_index is None and end is None and not positioned_only)
or (
annotation.start_index is not None
and annotation.start_index >= offset
@ -922,13 +918,9 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
def _messages_with_final_annotations(
self, response: ModelResponse
) -> tuple[tuple[int, GenericResponseOutputItem], ...]:
annotations: Final = [
annotation
for annotation in LiteLLMCompletionResponsesConfig._transform_chat_completion_annotations_to_response_output_annotations(
annotations=getattr(response.choices[0].message, "annotations", None)
)
if annotation.start_index is not None
]
annotations: Final[list[ChatCompletionAnnotation] | None] = getattr(
response.choices[0].message, "annotations", None
)
offsets: Final = tuple(
accumulate((len(item.content[0].text or "") for _, item in self._completed_message_items), initial=0)
)
@ -946,6 +938,7 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
end=offsets[position + 1]
if position + 1 < len(self._completed_message_items)
else None,
positioned_only=True,
)
+ self._unpositioned_message_annotations(position)
}
@ -1411,11 +1404,7 @@ class LiteLLMCompletionStreamingIterator(ResponsesAPIStreamingIterator):
annotations: Final = chunk.choices[0].delta.annotations
if annotations and self.sent_annotation_events is False and not allow_reasoning_resumption:
# Store annotation events to emit them one by one
response_annotations = self._message_annotations(
LiteLLMCompletionResponsesConfig.transform_chat_completion_annotations_to_response_output_annotations(
annotations=annotations
)
)
response_annotations = self._message_annotations(annotations)
self.sent_annotation_events = bool(response_annotations)
self._pending_annotation_events = []
for idx, annotation in enumerate(response_annotations):