Functional spec for wpautop

[mrclay]

I've put together an algorithm-based ​functional spec that I hope could be useful in maintaining wpautop. Also might be handy is this combination of ​wpautop failure cases.

The algorithm is based on work I've started on a DOMDocument-based implementation (a project I contribute to needs an MIT-licensed version), and although it's incomplete and slow, it does a bang-up job so far.

DOMDocument may just be too slow or strict in its parsing to allow the variety of markup that wpautop needs to accept, but it's ace at the level of surgical precision needed to do this right.

[scribu]

Although not directly relevant, it's good to remember that the HTML runs through kses afterwards, which can alter it significantly.

[nacin]

This is cool. For anything not currently covered by tests, it'd be great if we added unit tests for those.

Maybe such a document could find its way into ​http://make.wordpress.org/core/handbook/ as an appendix of some sort.

[pento]

It's interesting to explore, but I'm not sure there's a practical way to replace wpautop() with a proper HTML parser.

I'm fine with reopening this ticket at a later date if someone has a working solution for it.