Fix Shortcode—Markdown conflicts

[mcsf]

Spawned from discussion in #4944.

Issue Overview

By their respective syntax, Markdown and WordPress shortcodes are prone to conflict with one another. This is particularly visible in the examples below.

Links

When pasting the following:

This contains an [awesome] link to something. This here is even [better][awesomer].

[awesomer]: http://example.org

the ideal reconciliation would be as follows:

1. [awesomer] is the identifier for a reference-style link
2. [better] is the wrapped text for said link
3. thus, let Markdown handle and convert the pair [better][awesomer] to a link (to https://example.org)
4. [awesome] is a standalone tag with no link attached to it
5. thus, let Gutenberg treat it as a shortcode — which for standalone shortcodes means converting to a shortcode block, and for inline shortcodes means leaving it untouched

Escaping shortcode contents

When pasting the following:

[my_shortcode]hello, _there_.[/my_shortcode]
**oh**, [another_shortcode]hello, _there_.

the ideal reconciliation would be as follows:

1. [my_shortcode] is recognized as a shortcode with opening and closing tags and containing the string hello, _there_.
2. thus, don't perform any Markdown conversion on the tag itself and on its contents
3. [another_shortcode] is recognized as being a shortcode and not a link, but it is assumed self-closing
4. thus, no escaping is due, and the fragments **oh**, and hello, _there_. will be rendered as <strong>oh</strong>, and hello, <em>there</em>., respectively.

Possible Solution

This comment of mine #4944 (comment) points to a Gist offering a working solution to get started.

Screenshots / Video

Related Issues and/or PRs

Todos

• Tests
• Documentation

[mcsf]

/cc @iseulde, since we chatted about this in #4944.

[youknowriad]

A lot of things changed since this issue was opened. Is it still valid?

[mcsf]

The only thing I could see as an enhancement would be paste-time support for referenced links, as opposed to inline links. For example:

Some text with [a link][1] and
another [link][2].

[1]: http://example.com/ "Title"
[2]: http://example.org/ "Title"

Aside from that, overall support seems pretty good, defaults are sane, and I don't think the issue surfaces at all as a priority. Will close.