fix(openai): preserve response output content index - #1437
Open
yaodong-shen wants to merge 1 commit into
Open
Conversation
Use the parent output item's completion index for nested message content so reasoning outputs cannot absorb the following message text. Add coverage with real OpenAI response models.
| for content in message.content: | ||
| if isinstance(content, ResponseOutputText): | ||
| attributes.update(get_response_output_text_attributes(content, i)) | ||
| attributes.update(get_response_output_text_attributes(content, index)) |
There was a problem hiding this comment.
[P2] Preserve every text block in a response message
A ResponseOutputMessage.content list can legally contain multiple ResponseOutputText blocks. Passing the same parent index for each block makes every helper call emit gen_ai.completion.{index}.content, so attributes.update() silently replaces all earlier text with the last block. The helper also restarts each block's annotation index at j=0, causing later annotations to overwrite earlier ones under the same completion. Please aggregate the text blocks in order and keep annotation indices unique across the whole message (with offsets consistent with the aggregated content), and add a regression containing two annotated output_text blocks.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
📥 Pull Request
📘 Description
Fixes #1436.
OpenAI Responses can return a reasoning output before the assistant message. The message metadata correctly used its index in
response.output, but nestedResponseOutputTextattributes used the index withinmessage.content. This placed message text under the preceding reasoning completion and left the actual message completion without content.This change preserves the parent output index while extracting nested message text. The regression test uses the real OpenAI response models and verifies that a reasoning item at index 0 cannot absorb message content from index 1.
🧪 Testing
.venv/bin/python -m pytest tests/unit/instrumentation/openai_core/test_response_attributes.py -q -o log_cli=false— 14 passed.venv/bin/python -m pytest tests/unit -q -o log_cli=false— 489 passed, 1 skippedgit diff --check— passed