Link text inline content fails to parse image references

The CommonMark spec says that link text may itself contain inline content. Example 525:

[![moon](moon.jpg)](/uri)

yields:

<p><a href="/uri"><img src="moon.jpg" alt="moon" /></a></p>

In FlexMark, however, inline content that consists of an image reference fails to parse and is passed straight through as the link text. Example test case:

[![][moon]](/uri)

[moon]: moon.jpg

yields:

<p><a href=\"/uri\">![][moon]</a></p>

The expected output is:

<p><a href="/uri"><img src="moon.jpg" alt="moon" /></a></p>

Verified with ParserEmulationProfile.COMMONMARK_0_28 and ParserEmulationProfile.MULTI_MARKDOWN.

Additional context: My application uses ParserEmulationProfile.MULTI_MARKDOWN. MultiMarkdown uses image references to encode width/height data, which I am reliant on. So this is a hard blocker for me that I will need to work around in one way or another.

[vsch]

@jingibus, btw, the expected output in your case would be:

<p><a href="/uri"><img src="moon.jpg" alt="" /></a></p>

the alt is empty because you are using syntax [![alt text][moon]](/uri), which overrides the default alt with one which is empty. Use [![moon]](/uri) or [![moon][]](/uri) to use the reference id as alt text.

[jingibus]

Unbelievable. Thank you so much!

I’ve got a followup that’s specific to MultiMarkdown coming, but I’ve confirmed the issue reported here is fixed on my end.