fix: parse list-of-parts content in ChatMessage.from_openai_dict_format

[Vedant-Agarwal]

Related Issues

• fixes ChatMessage.from_openai_dict_format wraps list-of-parts content in TextContent, breaking msg.text contract #12445

Proposed Changes:

ChatMessage.from_openai_dict_format passed list-of-parts content verbatim into a single TextContent for user, system, and developer messages (only the tool branch handled lists). This made msg.text return a list (violating its str | None contract) and made to_dict() / to_openai_dict_format() emit invalid content:

from haystack.dataclasses import ChatMessage

b64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mP8/x8AAwMCAO+ip1sAAAAASUVORK5CYII="
msg = ChatMessage.from_openai_dict_format({
    "role": "user",
    "content": [
        {"type": "text", "text": "What is in this image?"},
        {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{b64}"}},
    ],
})
type(msg.text)  # was <class 'list'>, now str
msg.images      # was [], now [ImageContent(...)]

This PR makes list parsing the exact inverse of to_openai_dict_format:

• user messages: text parts become TextContent; image_url parts with base64 data URLs become ImageContent (MIME type and base64 payload split from the data URL, detail preserved); file parts with inline file_data become FileContent (filename preserved). The message is built via ChatMessage.from_user(content_parts=...).
• system/developer messages: lists of text parts are accepted (joined with a newline, since from_system stores a single text); non-text parts raise ValueError.
• unsupported parts — non-data image URLs, files referenced by file_id, unknown part types — raise ValueError instead of silently corrupting the message.

String content behaves exactly as before.

How did you test it?

Added unit tests in test/dataclasses/test_chat_message.py covering: list of text parts, text + image data URL part (with detail), text + file part, unsupported parts (http image URL, file_id file, unknown part type), system message with text parts / with non-text parts, and a full multimodal round-trip from_openai_dict_format(msg.to_openai_dict_format()) == msg.

pytest test/dataclasses/test_chat_message.py  ->  103 passed
ruff check / ruff format --check              ->  clean
mypy haystack/dataclasses/chat_message.py     ->  no issues

Notes for the reviewer

• The data-URL parser and part converter are small private helpers (_parse_openai_data_url, _from_openai_content_parts) next to _validate_openai_message.
• Joining multiple system text parts with "\n" is a judgment call — happy to change to rejecting multiple parts 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: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test: and added ! in case the PR includes breaking changes.
• 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.

(AI-assisted: implemented with Claude Code.)

[vercel]

@Vedant-Agarwal is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
haystack/dataclasses
chat_message.py
Project Total

This report was generated by python-coverage-comment-action