HTMLRewriter Replace text issue with HTML entities and <script>
I have a Worker which replaces some text in a webpage. In general it's working fine, but I seem to be caught between a couple of Worker quirks:
* I had an issue that the replace would sometimes cause problems with HTML entities and end up with weird ASCII characters - so I followed https://community.cloudflare.com/t/how-to-unescape-html-entities-in-strings/255667 to use html-entites and decode them, it solved that issue
* I then started to have some problems replacing text in general in too many tags, and it was breaking scripts in the head - so I targetted the HTMLRewriter with theselector in the on() method to say what to process - this seemed to fix most issues for <head> scripts and scripts before the end of the </body> tag.
* I now have a site where there is a <script> tag inside a multiple div tags - and I do notwant to replace the text in the script, but because it is inside multiple divs, it is processed as a child element of div - a less than sign is becoming < which then cannot run as code...
So far, I've been trying to work out the perfect selector query to help target what I am doing, and maybe there is a way to do that, but maybe the possible solutions are:
* target the selector further (but I don't think I can handle the case above)
* if I didn't have the issue with HTML entities I would not have the problem at all, is there another way to prevent this ASCII character issue creeping in when doing a replace?
* as far as I understand it - I'm going through the text for an element and its children - is there some way to have more control of that?
1 Reply
When filling an element with data, you got text or HTML insertion, sometimes you want HTML support and only in those cases you could consider escaping for the client the arrows <>. This is if JS modifications.
Server at all times will handle serializing and de-serialization of the data in a safe manor and additionally for client if you intend on advanced features like WYSIWYG messages.
May not help question was a bit vague.