4

In the MWE below, I show that I have a bibliographic entry, which has had comment characters % embedded in the abstract field; while this passes with bibtex, with biblatex/biber it makes a problem, since biber will compact all lines of the field into a single line in the .bbl file. So I use a biber.conf file to completely avoid having the abstract field entered in the .bbl file:

\documentclass{article}
\usepackage[
  bibencoding=utf8,
  backend=biber,
  natbib,
  style=ieee,
  isbn=true,
  url=true,
  defernumbers=true,
  sorting=nyt,
]{biblatex}
\usepackage{filecontents}
\usepackage{trace}
\usepackage{hyperref}

\begin{filecontents*}{\jobname.bib}
@article {jo-audm08,
  author = {Some Author},
  title = {{Testing the title of a paper}},
  url = {http://example.com/paper.pdf},
  journal={Proceedings of Some Conference},
  pages={123--127},
  publisher = {example.com},
  year={2008},
  mycustomfield = {whatever},
  lastaccessed = {Last Accessed: 13 April, 2009},
  lastmodified = {Last Modified: 10/28/2008 14:46:42},
  abstract = {%
In this paper, we would like to describe the employment of  .. %
  },
}
\end{filecontents*}

% http://tex.stackexchange.com/questions/32930/is-it-possible-to-suppress-a-specific-field-from-bibtex-bbl-in-biblatex/33309#33309
\begin{filecontents*}{biber.conf}
<config>
  <sourcemap>
    <maps datatype="bibtex" map_overwrite="1">
      <map>
        <map_step map_field_set="ABSTRACT" map_null="1"/>
      </map>
    </maps>
  </sourcemap>
</config>
\end{filecontents*}

\addbibresource{\jobname.bib}

\newif\ifDoingSourcemap
\DoingSourcemapfalse
% \DoingSourcemaptrue % uncomment this

\ifDoingSourcemap
  \DeclareSourcemap{
      \maps[datatype=bibtex]{
        \map{
          \step[fieldsource=mycustomfield]
          \step[fieldset=usera,origfieldval]
      }
    }
  }
  \DeclareFieldFormat{usera}{#1} % not mycustomfield
\fi

\begin{document}

Citing here: \cite{jo-audm08}

\printbibliography[sorting=none]

\end{document}

This MWE compiles fine with pdflatex test.tex && biber test && pdflatex test.tex && pdflatex test.tex.

Now, I want to use a custom field mycustomfield in my bibliographic entry, and for its handling, I've set up a \DeclareSourcemap which maps mycustomfield to usera of biblatex. So, I uncomment the \DoingSourcemaptrue line, and recompile; but the pdflatex call after the biber call then fails with:

Runaway argument?
{\field {journaltitle}{Proceedings of Some Conference} \field {title}\ETC.
! File ended while scanning use of \field.
<inserted text> 
                \par 
l.65 \begin{document}

... and one can afterwards confirm that:

$ grep abstract test.bbl
      \field{abstract}{% In this paper, we would like to describe the employment of .. %}
      \field{abstract}{% In this paper, we would like to describe the employment of .. %}

... with the enabling of the \DeclareSourcemap, the biber.conf is possibly ignored, and the abstract field ends up in the .bbl anyway, causing trouble.

So, how can I both use a \DeclareSourcemap in my tex file, along with the mapping already set in biber.conf (in this case, both have the abstract field ignored and unentered in the .bbl file - and use the mycustomfield as usera)?

sdaau
  • 17,079
  • That is an interesting problem. It is known that \DeclareSourcemap can only be used once (see also \DeclareSourcemap in biblatex). But I would have thought that Biber's biber.conf would still apply regardless. What I find odd is that Biber's documentations states that biber.conf takes precedence over .bcf while you seem to have experienced the opposite. – moewe May 30 '15 at 06:33
  • @moewe: While that's true of options and settings more generally, the precedence order is a bit different for sourcemap entries (see 3.1.2 of the Biber documentation or 4.5.2 of the BibLaTeX documentation) where \DeclareSourcemap takes precedence over the configuration file. Nevertheless, I would've thought that non-conflicting rules wouldn't, you know, conflict... – ig0774 May 30 '15 at 06:57
  • @ig0774 Ahhh that explains it, then. As I said two sourcemaps in a document are not possible (I seem to remember reading why some time ago, can't din it now, though), maybe that carries over to the .conf file as well. – moewe May 30 '15 at 07:29

2 Answers2

2

While I'd still appreciate a more erudite answer on the interference demonstrated in the OP (between biber.conf and \DeclareSourcemap) in general, I think I managed to fix this particular case.

After stumbling at biber/biblatex: \DeclareSourcemap conflicting with bcf file?, and realizing that biber.conf actually looks like what is written in \DeclareSourcemap - I wrote this, instead, in the OP MWE:

\ifDoingSourcemap
  \DeclareSourcemap{
      \maps[datatype=bibtex]{
        \map{
          \step[fieldsource=mycustomfield]
          \step[fieldset=usera,origfieldval]
      }
        \map[overwrite]{
          \step[fieldsource=abstract]
          \step[fieldset=abstract,null]
      }
    }
  }
  \DeclareFieldFormat{usera}{#1} % not mycustomfield
\fi

... and this compiles fine, and basically does what the biber.conf used to do (eliminate the abstract field from the .bbl) - so the biber.conf seems not necessary any more (in this case at least).

Community
  • 1
sdaau
  • 17,079
  • Even though this doesn't really explain why the problem arose in the first place. It is definitely the way to go. I really recommend having all of Biber's sourcemapping manipulations in one place, so you don't forget about one and end up with something you didn't want. – moewe May 30 '15 at 06:34
  • This is correct - the problem is that the \DeclareSourcemap and the .conf file write are at the same "level" of maps and the most specific settings (the particular document) take precedence. I could potentially split this as I did with the "style" level of maps but as @moewe says, it will probably encourage real confusing with maps all over the place. – PLK May 31 '15 at 11:51
  • Actually, this may be a bug, looking into it. – PLK May 31 '15 at 12:02
1

This was a bug. Please try biber 2.2 dev version now on Sourceforge. biber 2.2 can be used with biblatex 3.0.

PLK
  • 22,776