[23:58:30] Found some interesting work by the WordPress Performance team. They introduced a way of transforming HTML from hooks, such as to strip or modify an element, without a full Dom tree, and without regexes, with low constant memory cost. It assumes pre-normalised input I think. https://developer.wordpress.org/reference/classes/wp_html_tag_processor/ [23:59:47] Could be interesting if we end up changing a ton more things to happen at runtime, eg mobile frontend transform overhead, lazy img loading, etc. For now we seem to do okay, but appealing approach I must say.