asciidoctor-pdf returns an error when encountering a footnote that contains a reference to a bibliography item.
Example:
Test footnote:[test <<TEST>> test]
[[references]]
[bibliography]
== References
* [[[TEST]]] test.
Error message:
asciidoctor: ERROR: failed to parse formatted text: Test <a name="_footnoteref_1"> </a><sup>[<a anchor="_footnotedef_1">1</a>]</sup></a> test]
asciidoctor: ERROR: failed to parse formatted text: <a name="_footnotedef_1"> </a>[<a anchor="_footnoteref_1">1</a>] test <a anchor="TEST">[TEST
Used version:
Asciidoctor PDF 1.5.0.alpha.17 using Asciidoctor 2.0.9 [https://asciidoctor.org]
This is not a use case that is supported in AsciiDoc in general. Try it with HTML output and you will see it doesn't work there either.
I am not sure to understand what you mean.
The HTML generated with antora seems to work well in this regard (href in the footnote working well and displayed as expected in chrome)
With the atom asciidoctor-preview plugin I get:

<body>
<div id="wrapper">
<div class="article">
<div id="__asciidoctor-preview-1__" class="paragraph">
<p>Test <sup class="footnote">[<a id="_footnoteref_1" class="footnote" href="#_footnotedef_1" title="View footnote.">1</a>]</sup></p>
</div>
<div class="sect1">
<h2 id="references">References</h2>
<div class="sectionbody">
<div id="__asciidoctor-preview-2__" class="ulist bibliography">
<ul class="bibliography">
<li>
<p><a id="TEST"></a>[TEST] test.</p>
</li>
</ul>
</div>
</div>
</div>
<div id="footnotes">
<hr>
<div class="footnote" id="_footnotedef_1">
<a href="#_footnoteref_1">1</a>. test <a href="#TEST">[TEST]</a> test
</div>
</div>
</div>
</div>
</body>
Of course I can easily find a workaround for this wrt the PDF generation.
I thought, its always better to report issues. even minor ones :-)
I was surprised by the error message and the PDF output (see PDF output of the footnote part for a similar example below), given that I did not detect any issue in the antora output.

I figured out how to solve this in the PDF converter. PR on the way.
The HTML generated with antora seems to work well in this regard
Not in Asciidoctor 2.
I thought, its always better to report issues. even minor ones :-)
I'm glad you reported it. You have uncovered a difference between the converters that will need to be addressed. We can start with PDF. HTML is going to be another challenge.
(Right now, not all the platforms are using the same version of Asciidoctor core, which is why we can't always assume they will produce the same result).
That was quick ! Thanks
I'll have a new release out soon to test.
I just tested it with v1.5.0.alpha.18
It works great :-)
Thanks a lot!
Thanks for confirming the fix!