PDF exports seem to be untagged (no /StructTreeRoot, headings, or logical structure)

[Ovgodd]

I tagged @YousefED here, as recommended by @virgile-dev.

Describe the bug

When exporting a document to PDF through BlockNote’s default PDF pipeline (BlockNote → React-PDF → @react-pdf/pdfkit), the generated PDF seems to contain no Tagged PDF structure.

Based on my tests :

• “Tagged PDF” appears as No
• There is no /StructTreeRoot
• No /H1, /P, /Table, /Figure structure elements
• No /MCID entries
• No /RoleMap
• No logical reading order

This results in a PDF that screen readers cannot interpret and that cannot comply with WCAG / PDF/UA / RGAA accessibility requirements.

React-PDF’s own issue (#3179) also mentions that accessibility is currently not supported, even though @react-pdf/pdfkit internally contains PDFKit’s accessibility APIs.

For more detailed research, please refer to my document:
FR : EN :

---

To Reproduce

1. Create a BlockNote document containing headings, paragraphs, lists, and images.
2. Use the built-in PDF export.
3. Open the resulting PDF using
4. Observe that:
   • Tagged PDF = No
   • The structure tree is empty
   • Screen readers do not announce headings or regions
   • No alternative text is present on images

---

Expected behavior

When semantic elements (headings, paragraphs, lists, tables, images, etc.) are used in BlockNote, the exported PDF should ideally embed the corresponding structure:

• /StructTreeRoot
• /H1, /H2, /H3
• /P, /Span
• /Table, /TR, /TH, /TD
• /Figure with /Alt text
• /MCID entries for reading order
• A valid RoleMap mapping custom types to standard PDF roles

These elements are required for proper accessibility and PDF/UA compliance.