fix(generators): serialize timeout and max_retries in OpenAIImageGenerator.to_dict

[Anai-Guo]

Related Issues

• No open issue; found while auditing to_dict() coverage across components.

Proposed Changes:

OpenAIImageGenerator.__init__ takes timeout and max_retries, stores both on the instance, and _client_kwargs() passes them to OpenAI/AsyncOpenAI. to_dict() never emitted them, so they were lost on the way through a serialized pipeline:

g = OpenAIImageGenerator(timeout=60.0, max_retries=10)
g2 = OpenAIImageGenerator.from_dict(g.to_dict())

g._client_kwargs()   # {'timeout': 60.0, 'max_retries': 10}
g2._client_kwargs()  # {'timeout': 30.0, 'max_retries': 5}   <- silently back to the defaults

The fallback is silent: _client_kwargs() reads OPENAI_TIMEOUT/OPENAI_MAX_RETRIES or defaults to 30s / 5 retries whenever the attribute is None, so a deserialized pipeline just runs with different client settings than the one that was saved.

This is the only OpenAI-backed component with the gap — OpenAIChatGenerator, OpenAIGenerator, AzureOpenAIChatGenerator, AzureOpenAIResponsesChatGenerator, OpenAIResponsesChatGenerator, OpenAI{Text,Document}Embedder, AzureOpenAI{Text,Document}Embedder and OpenAITokenCounter all serialize both fields already.

response_format is deliberately left out: it is documented as ignored and only kept for backward compatibility, and __init__ warns when it is set to anything other than "b64_json".

How did you test it?

hatch run test:unit -k openai_image_generator equivalent (pytest test/components/generators/test_openai_image_generator.py):

• before the fix: 3 failed, 20 passed, 2 skipped — the two updated to_dict expectations plus the new round-trip test
• after the fix: 23 passed, 2 skipped

Updated test_to_dict, test_to_dict_with_params and test_from_dict to cover the two fields (test_to_dict_with_params already passed timeout=60, max_retries=10 into the constructor but asserted they were dropped), and added test_to_dict_from_dict_roundtrip_keeps_client_kwargs, which asserts _client_kwargs() is identical across a to_dict()/from_dict() round trip.

ruff check and ruff format --check are clean on both touched files.

Notes for the reviewer

The branch is cut from an older main because my fork cannot be fast-forwarded (my token has no workflow scope and syncing touches .github/workflows/). The three hunks are pure additions and merge cleanly onto current main; happy to rebase if you prefer.

Checklist

• I have read the contributors guidelines and the code of conduct.
• I have updated the related issue with new insights and changes.
• I have added unit tests and updated the docstrings.
• I've used one of the conventional commit types for my PR title.
• I have documented my code.
• I have added a release note file, following the contributors guidelines.
• I have run pre-commit hooks and fixed any issue.

---

This PR was written with an AI assistant (Claude Code). I have reviewed the changes and run the relevant tests locally.

🤖 Generated with Claude Code

[vercel]

@Anai-Guo is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[julian-risch]

The changes look good to me @Anai-Guo ! Thank you for opening this PR. As soon as you agree to the CLA we can merge this PR.

[HaystackBot]

Hi @Anai-Guo, thanks a lot for your contribution! 🙏

We noticed that the Contributor License Agreement (CLA) check (license/cla) hasn't passed yet, so we've temporarily moved this PR to draft and paused the review assignment.

To get your PR reviewed, please sign the CLA via the link in the license/cla check below (or in the CLA bot comment). As soon as the check turns green, this PR will automatically be marked ready for review again and a reviewer will be re-assigned.