refactor: consolidate watch_and_export implementations

[peter-gy]

📝 Summary

Consolidate watch_and_export_bytes and watch_and_export implementations & represent ExportResult contents as bytes while still allowing string-based access.

Follow-up of #7997

🔍 Description of Changes

• Main refactor is making ExportResult represent results as bytes instead of strings to support cases like PDF export
• Implemented "contents is bytes, text is string" convention familiar from httpx, requests, etc.
• killed watch_and_export_bytes instead of watch_and_export so that export_callback can keep returning ExportResult objects to (1) avoid duplicating encode / decode logic, (2) have a place for cross-cutting helpers like .text / .from_text and (3) keep the door open for additional metadata like e.g. warnings or execution stats

📋 Checklist

• I have read the contributor guidelines.
• For large changes, or changes that affect the public API: this change was discussed or approved through an issue, on Discord, or the community discussions (Please provide a link if applicable).
• Tests have been added for the changes made.
• Documentation has been updated where applicable, including docstrings for API changes.
• Pull request title is a good summary of the changes - it will be used in the release notes.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Review	Updated (UTC)
marimo-docs	Ready	Preview, Comment	Jan 27, 2026 6:34pm

[peter-gy]

@mscolnick I deviated from #7997 (comment) in this refactor and kept ExportResult instead of using plain bytes as otherwise we would find ourselves in .encode / .decode duplication hell + no easy way to associate further metadata with results down the line. what do you think?

[mscolnick]

there might be a small inefficiency over converting from text to bytes and back. maybe we could do:

contents: bytes | str

with a @property (or @cached_property) for bytes and text that will check the contents type and convert or not convert

[peter-gy]

Great point.

To make sure we don't shadow the built-in bytes via a def bytes(self) ...: how about contents: bytes | str and either as_bytes() / as_text() or via cached_property data_bytes / data_text ?

or just bytez? (:

[mscolnick]

yes, great call

[github-actions]

🚀 Development release published. You may be able to view the changes at https://marimo.app?v=0.19.7-dev23