Support info strings on fenced code blocks

[sawyerh]

Currently this Markdown:

loses the info string when it's converted to WYSIWYM. According to the CommonMark Spec (see example 100), the code HTML element should receive a class of language-ruby.

[marijnh]

Yes, such an attribute will be part of a Markdown-specific schema that I will at some point in the future work on. There's a bunch of other Markdown quirks, such as 'tight' lists, that the default schema doesn't implement.

[marijnh]

The new schema in https://github.com/ProseMirror/prosemirror-markdown has a params attribute for this purpose.

[KyleAMathews]

@marijnh are we supposed to use the default param for setting the code block language? https://github.com/ProseMirror/prosemirror-markdown/blob/master/src/schema.js#L43

In my testing a code block like:

```javascript
var i;
```

gets parsed and then serialized back to:

var i;

Losing the language part. I'm writing my own parser and I'm passing in the language as the default param but that's not helping.

Is the language being stored inside Prosemirror's doc structure just not being serialized back to markdown correctly?

[marijnh]

are we supposed to use the default param for setting the code block language?

The attribute is called params, and its default value is the empty string. So you're supposed to use the params attribute to store the string after the triple backticks.

The Markdown parser was apparently not reading these anymore. Attached patch should help.

[KyleAMathews]

Oh! Just using params made things work! Thanks!

I'm writing a custom parser based on Remark so easy fix.