Summary

Pages migrated from the Legacy Editor to the Cloud Editor may retain legacy formatting attributes (e.g., data-table-width, inline width styles). These unconverted attributes can cause Word export to fail. The system should sanitize or normalize these legacy attributes to prevent export failures.

Description

Currently, Confluence Cloud does not automatically sanitize or convert some types of legacy editor formatting metadata when a page is transitioned to the Cloud Editor.
During investigation, we found that legacy attributes such as:

• data-table-width
• style="width: …"
• Other legacy table/formatting markers

remain embedded in the storage format.
When a page containing these attributes is exported to Word, the export pipeline attempts to transform the content and fails, returning:

“Something went wrong. We're moving mountains to get it sorted.”

This behavior was observed on a customer site, and Support had to manually modify the storage format to resolve the issue.
The customer did not intentionally add these styles; they are remnants from the Legacy Editor.

This indicates a structural issue in the Cloud Editor where legacy formatting data should be automatically cleaned up, normalized, or safely ignored by the export pipeline.

Below is a minimal example extracted from the affected customer page (all identifiable content has been removed):

Problematic storage format (before fix):

<table data-table-width="700">
  <colgroup>
    <col style="width: 312.0px;"/>
    <col style="width: 388.0px;"/>
  </colgroup>
  <tbody>
    <tr>
      <td><p>…</p></td>
      <td><p>…</p></td>
    </tr>
  </tbody>
</table> 

Corrected storage format (after fix):
(legacy width attributes removed or normalized)

<table>
  <tbody>
    <tr>
      <td><p>…</p></td>
      <td><p>…</p></td>
    </tr>
  </tbody>
</table> 

Once the legacy width metadata was removed, Word export succeeded normally.

This issue can occur on any long-lived Confluence instance with pages created before the Cloud Editor existed.
Since customers have no visibility into storage format and no official method to clean corrupted legacy metadata, this becomes a Support-only fix.

Suggested features

1. Legacy editor metadata is automatically sanitized or removed when the page is converted to the Cloud Editor

1. The Word/PDF export pipeline includes a safe fallback mechanism so that unrecognized formatting attributes do not cause the export to fail

1. A background process (or save-time logic) normalizes table widths, inline styles, and deprecated attributes

Expected outcome

If implemented, customers—especially those with older content—will benefit from:

• Reliable Word/PDF export functionality
• No unexpected failures caused by invisible legacy attributes
• Reduced reliance on Support intervention
• A cleaner long-term migration path away from the Legacy Editor

This improvement would significantly reduce export-related issues for customers with older Confluence spaces and improve Cloud Editor reliability overall.