Just to add a little more justification:
The need has been stated eloquently by others. For example: We need DOM APIs in Workers - Modern Web Development with Chrome by Paul Kinlan.
Cloudflare has a strong interest in supporting the most helpful api it can provide within the tight constraints it needs to impose as far as memory usage / cpu utlization. This is what they came up with. I think the same findings would likely hold in a browser setting, especially on a resource-limited mobile phone.
I’ve faced scenarios where I wanted to implement functionality (such as providing link preview functionality) in workers, but abandoned it due to the lack of a parser.
Another use case – integrating data that is in indexedDB or other worker accessible locations into a server-rendered stream.
Being that we are encouraged to take processing off the main thread, lack of some way of parsing HTML has been a significant barrier for me.
From my explorations of the HTMLRewriter api, I believe it does provide sufficient hooks to implement a DOM Parser of sorts (maybe without css query capabilities) in a fairly straightforward way, with a fairly small download footprint (at the expense of the performance toll on memory this would entail, which Cloudflare steers clear from). Perhaps if the need for such a DOM Parser is really required, it could build on the HTMLRewriter down the road, once a good implementation is proven out.
