0

Good evening.

I have many .bib files going back 30 years. As a result, I have stuck to the old way of inputting accented letters. I'm just trying biblatex/biber with a small file and although compiling the bibliography works, when I next run pdflatex I get the error:

! Package inputenc Error: Unicode character ̂ (U+0302) (inputenc)
not set up for use with LaTeX.

See the inputenc package documentation for explanation. Type H for immediate help. ...

l.288 \printbibliography[heading=reading]

I tried adding \^\i directly into the main text and get the correct character, so it is only \^\i in the bibliography which is causing the problem. Here is a minimal (non) working example:

\documentclass[11pt,a4paper,onecolumn,twoside]{article}
\usepackage[style=authoryear,backend=biber]{biblatex}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}

\addbibresource{ARCLG021.bib}


\defbibheading{reading}{\subsubsection*{Reading}}


\begin{document}
\title{Test}
\author{Me}
\date{\today}
\maketitle


\newrefsection
\textcite{Lockyear:2000}
\printbibliography[heading=reading]


\end{document}

and the offending bibliography entry is:

@Incollection{Lockyear:2000,
  Title                    = {Experiments with {D}etrended {C}orrespondence {A}nalysis},
  Author                   = {Lockyear, K.},
  Booktitle                = {{C}omputer {A}pplications and {Q}uantitative {M}ethods in {A}rchaeology 1996},
  Publisher                = {British Archaeological Reports International Series 845},
  Year                     = {2000},
  Editor                   = {K. Lockyear and T. J. T. Sly and V. Mih{\u{a}}ilescu-B{\^\i}rliba},
  Pages                    = {9--17},
  Addendum                 = {\textsc{inst arch ak 20 qto com}},
  Date                     = {2000},
  Location                 = {Oxford},
  Timestamp                = {2017.01.09}
}

Any suggestions?

Thanks, Kris.

Kris Lockyear
  • 137
  • see e.g. https://tex.stackexchange.com/a/469556/2388 – Ulrike Fischer Jan 13 '22 at 21:25
  • @UlrikeFischer this really should be fixed in biber :( (even though it is by design) – David Carlisle Jan 13 '22 at 22:39
  • @DavidCarlisle I don't know the code and how complicated it would be to force a composed form if it exists. But the input \^i works, as does the unicode input, or you can use --output-safechars to avoid unicode output, so there are work arounds. – Ulrike Fischer Jan 13 '22 at 22:58
  • @UlrikeFischer my understanding of one of the gh threads is that \^\i used to work and be normalised to match the handling of \^i but was intentionally changed to produce the combining charcter pair. Given that biber is more or less only used with tex and combining characters are tricky in unicode tex and don't work at all in classic tex, this choice seems.. suboptimal. But anyway I'll mark this as duplicate – David Carlisle Jan 14 '22 at 08:25
  • 1
    @DavidCarlisle the problem (explained here https://tex.stackexchange.com/a/251381/2388) is that \i is the dotless i, and there seem to be no composed version for it with the accent. Which means that biber can't use simply some existing normalization library but would have to maintain a special lists for it. – Ulrike Fischer Jan 14 '22 at 08:33
  • @UlrikeFischer but that's a very weak argument as it is expanding \^ to use the combining accent so it is introducing the error. It should not do that but rather expand \^\i the way latex expands it. – David Carlisle Jan 14 '22 at 08:39
  • @DavidCarlisle I guess that perl code first converts the base to unicode, and then handles the accent, so it has to handle \^ı. But imho the step to unicode is not really the problem, the question is if biber could add a bit normalization before outputting the result similar to the way it converts to latex commands with --output-safechars. – Ulrike Fischer Jan 14 '22 at 09:05
  • The best solution is replace the LateX code by the unicode charactes in the .bib file (e.g, from Mih{\u{a}}ilescu to Mihăilescu) not only to avoid the compilation problems but also to make the source readable for humans. JabRef can make easily this conversion for a conflicting field, but for large files any good replacing tool as rpl will be faster. – Fran Jan 14 '22 at 10:38

0 Answers0