Trouble with special characters in index titles

Question

I am a French student writing with LaTeX. I have a use for multiple index with different names for which I use splitidx. I wanted my indices to appear in the TOC, so I used tocbibind. Everything worked fine but with an accent in one of my index title I get a error message relating to encoding. Here is a MWE (well not working might be more accurate !). Have you got any idea of how I could politely ask tocbibind to expand correctly the indices title in the toc file ?

Thanks a lot.

\documentclass[10pt,a4paper]{book}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage[makeindex]{splitidx}
\usepackage{tocbibind}
\newindex[Index des journaux et séries]{jo}
\newindex[Index des noms de personnes]{np}

\begin{document}

\tableofcontents

Ici du texte avec des accents qui fonctionne : é, à ç, è. Une indexation pour jo\sindex[jo]{phenix@Phenix} et une pour np\sindex[np]{phenix@Phenix}.

\printindex[jo]
\printindex[np] 

\end{document}

I think I see where the problem lies. If I try a chapter with an accent it get written \IeC {\'e} in the toc file whereas my index title always keeps its é even if I put the \IeC {\'e} in the text.

Answer 1

The problem is the usage of \xdef in \newindex, which should be \protected@xdef; with \xdef almost any accented character in the index title will produce the error. The tocbibind package is not involved in the issue.

It's a bug that should be fixed by the package author.

\documentclass[10pt,a4paper]{book}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage[french]{babel}
\usepackage[makeindex]{splitidx}
\usepackage{xpatch}
% let's use \protected@xdef instead of \xdef in \newindex
\makeatletter
% there are four occurrences
\xpatchcmd{\newindex}{\xdef}{\protected@xdef}{}{}
\xpatchcmd{\newindex}{\xdef}{\protected@xdef}{}{}
\xpatchcmd{\newindex}{\xdef}{\protected@xdef}{}{}
\xpatchcmd{\newindex}{\xdef}{\protected@xdef}{}{}
\makeatother

\usepackage{tocbibind}
\newindex[Index des journaux et séries]{jo}
\newindex[Index des noms de personnes]{np}

\begin{document}

\tableofcontents

Ici du texte avec des accents qui fonctionne : é, à ç, è. Une indexation pour 
jo\sindex[jo]{phenix@Phenix} et une pour np\sindex[np]{phenix@Phenix}.

\printindex[jo]
\printindex[np]

\end{document}

Answer 2

If you don't stick to splitindex try imakeidx instead, the syntax is similar and there are no errors about special characters, at least not in this example.

\documentclass[10pt,a4paper]{book}
\usepackage[french]{babel}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
%\usepackage[makeindex]{splitindex}
\usepackage[splitindex]{imakeidx}%
\usepackage{tocbibind}
%\newindex[Index des journaux et séries]{jo}
%\newindex[Index des noms de personnes]{np}

\makeindex[name=jo,title={Index des journaux et séries}]
\makeindex[name=np,title={Index des noms de personnes}]
\begin{document}

\tableofcontents

Ici du texte avec des accents qui fonctionne : é, à ç, è. Une indexation pour jo\index[jo]{phenix@Phenix} et une pour np\index[np]{phenix@Phenix}.

\index[jo]{é@letter é}


\printindex[jo]
\printindex[np] 

\end{document}

I randomly put some index entry there, without direct conjunction to the right place ;-)