Ampersand in Biber Sourcemap

Question

I use Mendeley to create my bib-files and then \usepackage[style=authoryear,backend=biber]{biblatex} to create my citations. As I have some URLs, I also use DeclareSourcemap (as suggested in this answer) to replace special characters. This works until I encountered a new case where an ampersand is in the URL. The entry that creates the issue is (in mybib.bib):

@article{EP2014,
    author = {{European Parliament} and {European Council}},
    journal = {Official Journal of the European Union},
    number = {L173},
    pages = {349--496},
    title = {{Directive 2014/65/EU on markets in financial instruments and amending Directive 2002/92/EC and Directive 2011/61/EU}},
    url = {https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0065{\&}from=EN},
    volume = {57},
    year = {2014}
}

which has a URL containing the escaped ampersand {\&} which needs to be &.

I tried using the DeclaresourceMap like so, but it doesnt work (biber says main.bcf is malformed)

\DeclareSourcemap{
  \maps{
    \map{ % Replaces '{\&}' with just '&'
      \step[fieldsource=url,
        match=\regexp{\{\\\&\}},
        replace=\regexp{\&}]
    }
  }
}

When I look in main.bcf it gets parsed as

<bcf:map_step map_field_source="url" map_match="\{\\\&\}" map_replace="\&"/>

which my text editor marks as an error (due to the ampersand not being properly escaped? thats at least by best guess).

MWE:

main.tex

\documentclass{article}

\usepackage[style=authoryear,backend=biber]{biblatex}
\addbibresource{mybib.bib}

\DeclareSourcemap{
  \maps{
    \map{ % Replaces '{\&}' with just '&'
      \step[fieldsource=url,
        match=\regexp{\{\\\&\}},
        replace=\regexp{\&}]
    }
  }
}

\begin{document}
\textcite{EP2014}
\printbibliography
\end{document}

which I then run using either latexmk main or pdflatex main and then biber main.

Any idea how to solve the error and properly escape the ampersand?

Note:

As I use Mendeley for the creation of my bib-file, I cannot change the bib-file directly, and changing the escaping of special characters when exporting from Mendeley messes up other entries, therefore I think I am bound to using DeclareSourcemap. If you have another option I am open to that as well.

Answer 1

The .bcf file that biblatex uses to communicate the sourcemaps to Biber is essentially an XML file. As far as I can see XML files don't really like &s (https://www.w3.org/TR/xml/#syntax), so we can replace & with its Unicode code point \x{26}

\documentclass{article}

\usepackage[style=authoryear,backend=biber]{biblatex}

\DeclareSourcemap{
  \maps{
    \map{ % Replaces '{\&}' with just '&'
      \step[fieldsource=url,
        match=\regexp{\{\\\x{26}\}},
        replace=\regexp{\x{26}}]
    }
  }
}

\usepackage{filecontents}
\begin{filecontents}{\jobname.bib}
@article{EP2014,
    author = {{European Parliament} and {European Council}},
    journal = {Official Journal of the European Union},
    number = {L173},
    pages = {349--496},
    title = {{Directive 2014/65/EU on markets in financial instruments and amending Directive 2002/92/EC and Directive 2011/61/EU}},
    url = {https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0065{\&}from=EN},
    volume = {57},
    year = {2014}
}
\end{filecontents}
\addbibresource{\jobname.bib}

\begin{document}
\textcite{EP2014}
\printbibliography
\end{document}

But really you should complain to the Mendeley people, the URL field is a verbatim field where characters should not be escaped at all.

Answer 2

When using a XML configuration file for the biber tool, the answer provided by moewe can be implemented by:

<config>
  <sourcemap>
        <!-- Replace {\&} with & in URL field -->
        <map>
          <map_step map_field_source="url" map_match="\{\\\x{26}\}" map_replace="\x{26}"/>
        </map>
    </maps>
  </sourcemap>
</config>