hyperref.sty.ltxml throws error for \url containing underscores in \footnote

[siegfriedzoetzsche]

LaTeXML throws an

Error:unexpected:_ Script _ can only appear in math mode

for

    \documentclass{article}
    \usepackage{hyperref}
    \begin{document}
    \footnote{\url{url.with.under_score}}
    \end{document}

but does not complain about

    \documentclass{article}
    \usepackage{hyperref}
    \begin{document}
    \url{url.with.under_score}
    \footnote{\url{url.without.underscore}}
    \end{document}

or

    \documentclass{article}
    \usepackage{url}
    \begin{document}
    \footnote{\url{url.with.under_score}}
    \end{document}

Despite the error message, LaTeXML seems to produce correct output.

[dginev]

Indeed, it seems that the hyperref binding keeps underscores as is? @brucemiller may know more, the relevant line is:

my @toks = map { (((($cc = $_->getCatcode) == CC_CS) || ($cc == CC_SUB)) ? $_ : T_OTHER(ToString($_))) } $url->unlist;

Thanks for the report and the pointed examples!

[dginev]

I have recently seen an increase with these

Error:unexpected:_ Script _ can only appear in math mode

errors while testing over recent batches of papers from arXiv, so a fix here may be an easy win for improving the statistics there.

[brucemiller]

Curious... that's part of it, but url.sty and hyperref.sty differ slightly, and the above doesn't quite capture the difference. I'll have to create some good tests (and testcases!)

[brucemiller]

Boy, once I started digging and comparing url & hyperref, and trying out tests, it got a bit messier. hyperref expands, url doesn't, which carries with it a few peculiarities. I think I've got it sorted out and added a couple of more thorough test cases. Thanks for the report!

(and hopefully it'll help arXiv as well :> )