Verified behavior

Notion’s web editor preserved accented Latin characters through a page reload. The current rich-text API represents text in the text.content JSON string; it does not require stripping accents or replacing them with HTML entities.

Test setup

We entered two synthetic strings on a private Notion page:

  • Café déjà vu — élève, façade, São Paulo
  • Crème brûlée à côté — naïve coöperate

After reloading the page, both complete strings remained present with the same accents and punctuation.

Tested steps

  1. Enter the first synthetic Unicode string.
  2. Enter the second, independently varied string.
  3. Reload the page.
  4. Confirm both exact strings remain in the loaded page.
  5. Compare the result with Notion’s current JSON and rich-text documentation.

Send ordinary Unicode strings in the documented rich-text structure and let a JSON library serialize the request. Before sending, log a safe synthetic value at the application boundary and inspect the encoded request body. Check for ASCII-only conversions, incorrect byte decoding, manual escaping, database-column loss, or double encoding earlier in the pipeline.

Evidence boundary

The UI round trip shows that Notion can retain these characters, while the documentation identifies the API representation. It does not prove how a particular SDK or integration transmits them. Reproduce with a synthetic authenticated API request before assigning the fault to Notion or to a specific client library.