Incorrect parsing of fenced code blocks inside quoted blocks when followed by another fenced code outside quoted block

[artyom]

When fenced code block is matched inside quote block that's followed later by some non-quoted fenced code block, it is incorrectly parsed, matching end of fenced code block greedily.

Try parsing and then rendering this text:

Bug demo:

> quoted block 1 start
>
> ```
> fenced pre block 1
> ```
>
> quoted block 2 end

Paragraph

```
fenced pre block 2
```

> quoted block 2 start
>
> ```
> fenced pre block 3
> ```
>
> quoted block 2 end

It's incorrectly parsed/rended to this html:

<p>Bug demo:</p>

<blockquote>
<p>quoted block 1 start</p>

<pre><code>&gt; fenced pre block 1
&gt; ```
&gt;
&gt; quoted block 2 end

Paragraph

</code></pre>

<p>fenced pre block 2</p>

<pre><code>
quoted block 2 start

</code></pre>

<p>fenced pre block 3
&ldquo;`</p>

<p>quoted block 2 end</p>
</blockquote>

Please see this gist with complete code reproducing this output (also notice how github parser/renderer handles this case as expected).

[artyom]

This patch to testdata/FencedCodeInsideBlockquotes.tests illustrates the issue:

bug126-test.zip

[miekg]

I think I tried fixing this a while back, but the fix failed to pass the test suite I have for mmarkdown/mmark

…

On Wed, 7 Aug 2019, 13:05 Artyom Pervukhin, ***@***.***> wrote: This patch to testdata/FencedCodeInsideBlockquotes.tests illustrates the issue: bug126-test.zip <https://github.com/gomarkdown/markdown/files/3476644/bug126-test.zip> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#126?email_source=notifications&email_token=AACWIW3TRM2TUUWDWSACWMDQDKT6TA5CNFSM4IJYQZH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD3YBGEQ#issuecomment-519050002>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AACWIW2RQBVD4APXEOA2VZDQDKT6TANCNFSM4IJYQZHQ> .

[miekg]

shorter example

Bug demo:

> quoted block 1 start
>
> ```
> fenced pre block 1
> ```
> quoted block 2 end

Paragraph

```
fenced pre block 2
```

[chmike]

Bug is still not corrected. Is this gomarkdown module supported ?

[kjk]

A minimized test case:

> ```
> fenced 1
> ```


```
fenced 2
```

We generate:

<blockquote>
<pre><code>&gt; fenced 1
&gt; ```

</code></pre>

<p>fenced 2
&ldquo;`</p>
</blockquote>

It should be more like:

<blockquote>
  <p>
    <code>
      fenced 1
    </code>
  </p>
</blockquote>
<p>
  <code>
    fenced 2
  </code>
</p>

Babelmark: https://babelmark.github.io/?text=%3E+%60%60%60%0A%3E+fenced+1%0A%3E+%60%60%60%0A%0A%60%60%60%0Afenced+2%0A%60%60%60%0A

The issue seems to be that we don't recognize code block ending inside > which messes the rest of parsing.

[kjk]

Seems like Parser.fencedCodeBlock would need to recognize end of fence line inside blockquote.

[kjk]

Tried to modify fencedCodeBlock to add insideBlockquote flag and added:

if insideBlockquote {
	// skip '>' prefix
	pre := p.quotePrefix(data[beg:])
	if pre > 0 {
		beg += pre
	}
}

but it's more complicated than that. Would have to strip > from code inside blockquote and properly detect termination of blockquote (p.terminateBlockquote() didn't hit)

More complicated than I thought.