Filter Replacement Does Not Allow Escaped Characters

In my use case, the goal is to replace the plain old space and avoid introducing HTML entities.

Other use cases could be:

• replace multiple spaces with a single space
• replace X number of spaces with a tab character. (Text within an HTML PRE element)
• replace tab character with spaces

In the admin UI, the filter Replacement value for all of these cases all look the same. And could be confused with being empty like a filter that removes matches.

Here are the Search Expression and Replacement value in action (via regex101) https://regex101.com/r/rthrOB/1

Thank you, @star-szr. This has helped me identify a quirk of the DEV server in use on a project. Replacement with an HTML entity is working fine on my LOCAL stack (DDEV) and the client's STAGING server.

And with some digging into the code a little:

web/modules/contrib/ckeditor5_paste_filter/js/ckeditor5_plugins/pasteFilter/src/pastefilter.js
lines 34-37
html = html.replace(
new RegExp(filter.search, 'gimsu'),
filter.replace,
);


According to Mozilla's reference document on String.prototype.replace(), the use of escaped characters is supported among the special replacement patterns when the replacement parameter is a string type. But this is not the behaviour we are observing.

But using HTML element(s) as the replacement string (field value), a later line of code kicks in.

line 51
data.content = this.editor.data.htmlProcessor.toView(html);


From the CKEditor 5 API, the HTML data processor method toView() parses and converts the pasted string into a `ViewDocumentFragment`. The parsing does replace some of the base HTML elements with UFT8 characters. In the case here   is replace with the space character (U+0020) . And as we can all guess now, the parsing ignores HTML elements like No-Break Space  .

Hope the info helps someone else down the road.

Closing the issue.