fix(integrations): openai/openai-agents: convert input message format #5248
Conversation
…he schema we expect for the `gen_ai.request.messages`
constantinius
changed the title
…ntent and update message handling
constantinius
changed the title
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
![sentry[bot]](https://avatars.githubusercontent.com/u/1396951?s=60&v=4){kind=small}
…nAI message handling
Semver Impact of This PR🟢 Patch (bug fixes) 📋 Changelog PreviewThis is how your changes will appear in the changelog. New Features ✨
Bug Fixes 🐛
Documentation 📚
Internal Changes 🔧Release
Other
🤖 This preview updates automatically when you update the PR. |
sentrivana
left a comment
sentrivana
left a comment
There was a problem hiding this comment.
Looks ok to me, two things:
- Can we add a check to the tests that we're not modifying the user's messages? Either as a new test or just adding an assert to the tests added in this PR
- I assume there is no way to dedupe some of the trimming logic between OpenAI agents and OpenAI because the format is different?
| if item.get("type") == "image_url": | ||
| image_url = item.get("image_url") or {} | ||
| url = image_url.get("url", "") |
There was a problem hiding this comment.
Bug: The code will raise an AttributeError if item.get("image_url") returns a string, because the or {} fallback is not triggered and .get() is called on a string.
Severity: HIGH
Suggested Fix
Add a check to ensure image_url is a dictionary before calling .get() on it. A similar pattern is used elsewhere in the codebase: url = image_url.get("url", "") if isinstance(image_url, dict) else str(image_url). This will handle both dictionary and string formats gracefully.
Prompt for AI Agent
Review the code at the location below. A potential bug has been identified by an AI
agent.
Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not
valid.
Location: sentry_sdk/integrations/openai.py#L226-L228
Potential issue: In the `_convert_message_parts` function, the code processes message
parts to extract an `image_url`. The line `image_url = item.get("image_url") or {}` does
not correctly handle cases where the value of `image_url` is a string instead of a
dictionary. If a string is provided (e.g., `{"type": "image_url", "image_url":
"https://..."}`), the subsequent call to `image_url.get("url", "")` will raise an
`AttributeError`, as strings do not have a `.get()` method. This causes an unhandled
exception within the Sentry integration, preventing the span from being processed
correctly.
Did we get this right? 👍 / 👎 to inform future reviews.
Done
Looking into that. Cursor says no. But I'm not sure tbh |
| content = _transform_openai_agents_message_content(original_input) | ||
| if not isinstance(content, list): | ||
| content = [{"text": str(content), "type": "text"}] | ||
| messages.append({"content": content, "role": "user"}) |
There was a problem hiding this comment.
Non-dict content items produce invalid message structure
Low Severity
When original_input is a list containing non-dict items (like strings or numbers), _transform_openai_agents_message_content returns them unchanged. The calling code only wraps the result in text format when it's NOT a list, so lists with non-dict items like ["hello", "world"] become invalid content structures instead of proper [{"text": "hello", "type": "text"}, ...] format. The old code used safe_serialize() to handle any input type safely, producing valid message content for all cases.
Description
Convert messages to common
gen_ai.request.messagesstructureIssues
Closes https://linear.app/getsentry/issue/TET-1633/redact-images-openai-openai-agents