#3 HTML sanitization

[Dave Brondsema]

This may land as part of [#2] but documenting it as a separate issue.

When upgrading Markdown to 3.0, its safe_mode param will be ignored. We don't have to upgrade Markdown to 3.0 soon though, 2.x versions seem to work fine on python 3.7 already.

We may need it soon for Textile. If https://github.com/textile/python-textile/issues/65 isn't resolved, we'll have to use a forked version of textile to get py3 support, or use latest textile with sanitization applied.

I think it'll probably be best to have sanitization rather than relying on built-in safety within each formatting library. It'd be nice to sanitize by default within Pypeline, but the application using Pypeline might have its own preferences for what to allow or not. Maybe a default list? https://forge-allura.apache.org/p/allura/tickets/8297/ discusses bleach vs html5lib sanitizer

[Kenton Taylor]

Merged