grammar for inline codes is not good

[icecola2018]

Expectation

``foo

foo``

`foo``

``foo`

should be

<p>foo``</p>
<p>``foo</p>
<p>`foo``</p>
<p>``foo`</p>

Result

marked returns

<p><code></code>foo</p>
<p>foo<code></code></p>
<p><code>foo`</code></p>
<p><code></code>foo`</p>

What was attempted

I edited marked.js:

1. line 471, replace code: /^(`+)(\s*)([\s\S]*?[^`]?)\2\1(?!`)/ with code: /^(`+(?!`))([\s\S]*?[^`])\1(?!`)/
2. line 694, replace cap[3] with cap[2]

then get

<p>foo``</p>
<p>``foo</p>
<p>`foo``</p>
<p>`<code>foo</code></p>

However it doesn't work well on the last one.

[Feder1co5oave]

Thanks for reporting, this was partially my mistake in #1013.

Babelmark for reference.

The fix you suggest brings us back before #1013 so it is not right.
I have an easy partial fix for this which would produce the following output:

<p>``foo</p>
<p>foo``</p>
<p>`foo``</p>
<p><code>`foo</code></p>

I'm adding this to the consolidating issue hoping that we'll conform with commonmark on codespans too.

[icecola2018]

Reply to @Feder1co5oave

I see the problem in #1030 .

Actually my suggestion

/^(`+(?!`))([\s\S]*?[^`])\1(?!`)/

is different from the one before #1030

/^(`+)([\s\S]*?[^`])\1(?!`)/

---

I found that back references (such as \1) perform strangely when they refer to groups ending with + or *. In fact \1 has chance to match any quantities of `.

For example, if (`+) matched `` and it was not closed, then \1 would match ` or ``` or ```` or ...
That's why /^(`+)([\s\S]*?[^`])\1(?!`)/ works well on `` ` `` but wrong on `` ` ```.

---

Then I added (?!`) after + to the first group, which seemed no sense. Actually it worked well, \1 will only match the same string that the group matches. You may test it to be sure.

---

I'm not very good at English, so I will try my best to explain if you don't understand what I want to say.

[Feder1co5oave]

Thanks, you've been very clear and detailed, I'll look into this!

[UziTech]

this seems to be fixed in the latest version of marked demo