Include response information in ContentFilterFinishReasonError

[FloTeu]

Confirm this is a feature request for the Python library and not the underlying OpenAI API.

• This is a feature request for the Python library

Describe the feature or improvement you're requesting

Related Issue: #1700 (solved, but only for LengthFinishReasonError)

Situation

When calling AsyncOpenAI(...).beta.chat.completions.parse(..., response_format=SomePydanticModel), the OpenAI library raises ContentFilterFinishReasonError when finish_reason == "content_filter", without providing any information as to what the response contained.
Most of the times a BadRequestError error is raised if a content filter was applied. In our case an image with a self harm text written on a post it triggers the ContentFilterFinishReasonError.
Assumption: the image itself overcomes the filter, but afterwards the openai llm text triggers the content filter.

Complication

Because there is no way to retrieve any information about the response, I cannot programmatically save information about the context. For example, I cannot access and track information from the usage object in the chat completion response or get detailed information about which content filter was triggered (self_harm, sexuality etc.).
I also need to treat BadRequestError fundamentally different from ContentFilterFinishReasonError exception, even though i like to treat both cases equally.

Desired behavior

As a library user, I always want to know details about responses from LLM calls. For a ContentFilterFinishReasonError i would assume, to get the information what exact filter was triggered.

I see three potential solutions:

• Include the content filter details in the ContentFilterFinishReasonError class
• Return the response as an attribute in the exception object similar to BadRequestError
• Raise BadRequestError instead of ContentFilterFinishReasonError (If its a llm answer issue and not a request issue, this might be misleading)

OpenAI SDK Version: 1.109.1

Code location

• File and line: openai.lib._parsing._completions.py on line 103
• Function: parse_chat_completion

Additional context

No response

[nomiveritas]

Feature Request: Add response details to ContentFilterFinishReasonError.
Right now, when finish_reason == "content_filter", the SDK raises this error without exposing the response or usage info.
It would help a lot if the exception included a response attribute (like BadRequestError) for better debugging and logging.