Unclosed HTML comments not being escaped with `safe_mode='escape'`

[michaelkedar]

Unclosed HTML comments not being escaped with `safe_mode='escape'`

[Crozzers]

Thanks for spotting this. I've opened a PR for the incomplete HTML comments.

It's also inconsistent whether it puts <em> tags around the 'commented' section. I believe it should be replacing it in both cases

The library currently regards the contents of any HTML tag as HTML and will not process it as markdown. We do have an extra called markdown-in-html that allows you to put markdown content inside HTML tags, but you need to add a markdown="1" attribute to the tag, which won't work for comments.

Looking on babelmark, most other MD processors leave the HTML comment intact, except for markdown-it which escapes the comment tag and then processes the contents.

It's feasible that we could escape the comment tag, sanitise the contents (to remove any commented out HTML) and then process it as markdown.

[michaelkedar]

Thanks for fixing this!

The library currently regards the contents of any HTML tag as HTML and will not process it as markdown.

Personally, I find having this behaviour when safe_mode='escape' unusual, since setting it is supposed to not treat the input as HTML. markdown-it seems to sanitize HTML by default, which seems to be why it processes the markdown in the comment tags.

To bring it back to this issue, I'm wondering if the unclosed comment fix will result in everything after the <!-- to remain unprocessed?

[Crozzers]

To bring it back to this issue, I'm wondering if the unclosed comment fix will result in everything after the <!-- to remain unprocessed?

From my testing, this doesn't happen. Incomplete tags are handled differently to normal HTML. Normal HTML is escaped and the contents are hashed to prevent further processing. Incomplete tags just have the tag escaped and the contents left in place.

Personally, I find having this behaviour when safe_mode='escape' unusual, since setting it is supposed to not treat the input as HTML

This makes sense, and the difference in treatment between complete and incomplete HTML tags is a bit confusing. I'll look into this