autop: Correctly preserve attributed multi-line paragraphs in removep

[aduth]

Related: #9056 (comment)

This pull request seeks to resolve an issue where removep will wrongly remove a paragraph tag when it has attributes, if the paragraph spans multiple lines. As described in #9056 (comment), this occurs because the default behavior of . in JavaScript regular expressions will not match newline characters.

Before:

removep( '<p style="text-align: center">\nHello World\n</p>' );
// '<p style="text-align: center">\nHello World'

After:

removep( '<p style="text-align: center">\nHello World\n</p>' );
// '<p style="text-align: center">\nHello World</p>'

Testing Instructions:

Ensure unit tests pass.

Verify that content is not mangled when multi-line paragraph in a classic block (i.e. in non-block-demarcated content).

1. Navigate to Posts > Add New
2. Switch to Code Editor
3. In the text field, paste:

<p style="text-align: center">
Hello World
</p>

4. Blur the text field
5. Observe that text is not mangled
   • The trailing newline preceding the closing tag will be omitted. This had existed in the source implementation, and could likely be preserved, though it seemed incidental to the primary bug.

[ellatrix]

It's a bit scary that this is the only removep test for such a huge pile of regular expressions.

[aduth]

It's a bit scary that this is the only removep test for such a huge pile of regular expressions.

It is, yes. For historical context, the module originated as a direct port of the PHP implementation of autop, including its tests. There is no removep in PHP, so there are no tests. There was a previous removep implementation in JavaScript from which this implementation was adapted, but it had no tests to port over.

I can plan to create a task issue to follow-up on adding more tests here.

[aduth]

I can plan to create a task issue to follow-up on adding more tests here.

See #15152