Description

Fixes a bug where DOM serialization crashes with "Cannot read properties of null (reading 'sheet')" error when processing stylesheets on complex pages (e.g., Confluence).

Root Cause

In packages/dom/src/serialize-cssom.js, the code attempts to find cloned style elements using querySelector() but does not check if the element exists before passing it to styleSheetFromNode(). When the element is not found in the clone, cloneOwnerNode becomes null, causing the error.

This typically happens with:

• Style elements in shadow DOM that weren't properly cloned
• Dynamically inserted styles after initial cloning
• Complex DOM structures with late-loaded stylesheets
• Styles in ignored tags

Changes

Code Changes

• packages/dom/src/serialize-cssom.js: Added null check at line 64 to skip stylesheets when cloneOwnerNode is not found in the cloned DOM

Test Changes

• packages/dom/test/serialize-css.test.js: Added test case "skips stylesheet when cloneOwnerNode is not found in clone" to verify the fix handles missing elements gracefully

Testing

✅ All existing tests pass (212/212 on Chrome)
✅ New test case validates the fix
✅ Linting passes on modified files

Related

• Jira: PER-6656
• Customer: NVIDIA experiencing this error on Confluence pages
• Workaround: Use ignoreStyleSheetSerializationErrors: true option

Impact

• Low risk: Only adds a safety check to prevent crashes
• No breaking changes
• Improves reliability for complex DOM structures