[Bugfix][Responses] Use standard reasoning content-part events (#59652)
Signed-off-by: ai-jz <ai-jz@users.noreply.github.com> Co-authored-by: ai-jz <ai-jz@users.noreply.github.com> Co-authored-by: Flora Feng <4florafeng@gmail.com>
This commit is contained in:
co-authored by
ai-jz
Flora Feng
parent
d848c4ed3d
commit
2a52715ef7
@@ -42,9 +42,6 @@ from openai.types.responses import (
|
||||
ResponseInProgressEvent as OpenAIResponseInProgressEvent,
|
||||
)
|
||||
from openai.types.responses.response import IncompleteDetails, ToolChoice
|
||||
from openai.types.responses.response_reasoning_item import (
|
||||
Content as ResponseReasoningTextContent,
|
||||
)
|
||||
from openai.types.responses.tool import Tool
|
||||
from openai.types.shared import Metadata, Reasoning
|
||||
from openai_harmony import Message as OpenAIHarmonyMessage
|
||||
@@ -845,50 +842,6 @@ class ResponsesResponse(OpenAIBaseModel):
|
||||
)
|
||||
|
||||
|
||||
# TODO: this code can be removed once
|
||||
# https://github.com/openai/openai-python/issues/2634 has been resolved
|
||||
class ResponseReasoningPartDoneEvent(OpenAIBaseModel):
|
||||
content_index: int
|
||||
"""The index of the content part that is done."""
|
||||
|
||||
item_id: str
|
||||
"""The ID of the output item that the content part was added to."""
|
||||
|
||||
output_index: int
|
||||
"""The index of the output item that the content part was added to."""
|
||||
|
||||
part: ResponseReasoningTextContent
|
||||
"""The content part that is done."""
|
||||
|
||||
sequence_number: int
|
||||
"""The sequence number of this event."""
|
||||
|
||||
type: Literal["response.reasoning_part.done"]
|
||||
"""The type of the event. Always `response.reasoning_part.done`."""
|
||||
|
||||
|
||||
# TODO: this code can be removed once
|
||||
# https://github.com/openai/openai-python/issues/2634 has been resolved
|
||||
class ResponseReasoningPartAddedEvent(OpenAIBaseModel):
|
||||
content_index: int
|
||||
"""The index of the content part that is done."""
|
||||
|
||||
item_id: str
|
||||
"""The ID of the output item that the content part was added to."""
|
||||
|
||||
output_index: int
|
||||
"""The index of the output item that the content part was added to."""
|
||||
|
||||
part: ResponseReasoningTextContent
|
||||
"""The content part that is done."""
|
||||
|
||||
sequence_number: int
|
||||
"""The sequence number of this event."""
|
||||
|
||||
type: Literal["response.reasoning_part.added"]
|
||||
"""The type of the event. Always `response.reasoning_part.added`."""
|
||||
|
||||
|
||||
# vLLM Streaming Events
|
||||
# Note: we override the response type with the vLLM ResponsesResponse type
|
||||
class ResponseCompletedEvent(OpenAIResponseCompletedEvent):
|
||||
@@ -913,8 +866,6 @@ StreamingResponsesResponse: TypeAlias = (
|
||||
| ResponseContentPartDoneEvent
|
||||
| ResponseReasoningTextDeltaEvent
|
||||
| ResponseReasoningTextDoneEvent
|
||||
| ResponseReasoningPartAddedEvent
|
||||
| ResponseReasoningPartDoneEvent
|
||||
| ResponseCodeInterpreterCallInProgressEvent
|
||||
| ResponseCodeInterpreterCallCodeDeltaEvent
|
||||
| ResponseWebSearchCallInProgressEvent
|
||||
|
||||
@@ -70,8 +70,6 @@ from vllm.entrypoints.openai.parser.harmony_utils import (
|
||||
is_function_recipient,
|
||||
)
|
||||
from vllm.entrypoints.openai.responses.protocol import (
|
||||
ResponseReasoningPartAddedEvent,
|
||||
ResponseReasoningPartDoneEvent,
|
||||
StreamingResponsesResponse,
|
||||
)
|
||||
from vllm.entrypoints.openai.responses.utils import (
|
||||
@@ -220,16 +218,13 @@ def emit_reasoning_delta_events(
|
||||
)
|
||||
state.current_content_index += 1
|
||||
events.append(
|
||||
ResponseReasoningPartAddedEvent(
|
||||
type="response.reasoning_part.added",
|
||||
ResponseContentPartAddedEvent(
|
||||
type="response.content_part.added",
|
||||
sequence_number=-1,
|
||||
output_index=state.current_output_index,
|
||||
item_id=state.current_item_id,
|
||||
content_index=state.current_content_index,
|
||||
part=ResponseReasoningTextContent(
|
||||
text="",
|
||||
type="reasoning_text",
|
||||
),
|
||||
part={"text": "", "type": "reasoning_text"},
|
||||
)
|
||||
)
|
||||
events.append(
|
||||
@@ -457,13 +452,13 @@ def emit_reasoning_done_events(
|
||||
)
|
||||
)
|
||||
events.append(
|
||||
ResponseReasoningPartDoneEvent(
|
||||
type="response.reasoning_part.done",
|
||||
ResponseContentPartDoneEvent(
|
||||
type="response.content_part.done",
|
||||
sequence_number=-1,
|
||||
item_id=state.current_item_id,
|
||||
output_index=state.current_output_index,
|
||||
content_index=state.current_content_index,
|
||||
part=content,
|
||||
part=content.model_dump(),
|
||||
)
|
||||
)
|
||||
events.append(
|
||||
@@ -969,16 +964,13 @@ def emit_simple_reasoning_open(
|
||||
status="in_progress",
|
||||
),
|
||||
),
|
||||
ResponseReasoningPartAddedEvent(
|
||||
type="response.reasoning_part.added",
|
||||
ResponseContentPartAddedEvent(
|
||||
type="response.content_part.added",
|
||||
sequence_number=-1,
|
||||
output_index=state.output_index,
|
||||
item_id=state.current_item_id,
|
||||
content_index=state.content_index,
|
||||
part=ResponseReasoningTextContent(
|
||||
text="",
|
||||
type="reasoning_text",
|
||||
),
|
||||
part={"text": "", "type": "reasoning_text"},
|
||||
),
|
||||
]
|
||||
|
||||
@@ -1016,13 +1008,13 @@ def emit_simple_reasoning_done(
|
||||
content_index=state.content_index,
|
||||
text=state.accumulated_text,
|
||||
),
|
||||
ResponseReasoningPartDoneEvent(
|
||||
type="response.reasoning_part.done",
|
||||
ResponseContentPartDoneEvent(
|
||||
type="response.content_part.done",
|
||||
sequence_number=-1,
|
||||
item_id=state.current_item_id,
|
||||
output_index=state.output_index,
|
||||
content_index=state.content_index,
|
||||
part=part,
|
||||
part=part.model_dump(),
|
||||
),
|
||||
ResponseOutputItemDoneEvent(
|
||||
type="response.output_item.done",
|
||||
|
||||
Reference in New Issue
Block a user