fix: correct websocket_base_url derivation to preserve http in query params

[AliSharjeell]

Summary

• Converted websocket_base_url from a simple attribute to a property
• The property now correctly derives the websocket URL from base_url by only replacing the scheme prefix
• This fixes the bug where str.replace() was replacing ALL occurrences of http:// or https:// in the URL, including those in query parameters

Fix

When websocket_base_url is not explicitly set, it is now derived from base_url using httpx.URL's copy_with method, which only changes the scheme component rather than doing string replacement.

Test

Verified that URLs with http:// in query parameters are handled correctly.

Closes #3294

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a52297ef13

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Restore AsyncOpenAI.websocket_base_url attribute

Assigning only self._websocket_base_url here removes the public websocket_base_url instance attribute on AsyncOpenAI, but async websocket code still reads self.__client.websocket_base_url (for example in src/openai/resources/realtime/realtime.py and src/openai/resources/responses/responses.py). Because AsyncOpenAI has no corresponding property, those accesses now raise AttributeError at runtime for async clients, breaking realtime/response streaming flows.

Useful? React with 👍 / 👎.