Character replacement in unicode string

Question

I need a command to replace (erase) the characters · and | from a string.

Consider the following naive code:

\documentclass{article}

\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

\usepackage{stringstrings}

\newcommand\mycommand[1]{%
  \convertchar[q]{#1}        {·}{}%
  \convertchar[v]{\thestring}{|}{}}

\begin{document}
\mycommand{foo·bar|baz}
\end{document}

It gives a stack exceeded error; I think the reason is that the occurrences of · are passed as \IeC {\textperiodcentered } to \convertchar.

Also, the substitution of | is not working. I think that's due to the font encoding as suggested by the manual of stringstrings.

Fiddling with \unexpanded and \detokenize got me nowhere: I don't quite get what is happening, so I just made a mess.

What is the best way to write such a command? Why?

(An implementation using xstring is equally welcome; I tried, obtaining essentially the same results.)

Answer 1

The problem is, as far as I can see, that stringstrings tries expansion and that · is a two byte character.

I'd suggest expl3 and \tl_remove_all:Nn that doesn't attempt expansion.

\documentclass{article}

\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}

\usepackage{xparse}

\ExplSyntaxOn
\NewDocumentCommand\mycommand{m}
 {
  \tl_set:Nn \l_tmpa_tl { #1 }
  \tl_remove_all:Nn \l_tmpa_tl { · }
  \tl_remove_all:Nn \l_tmpa_tl { | }
  \tl_use:N \l_tmpa_tl
 }
\ExplSyntaxOff

\begin{document}
\mycommand{foo·bar|baz}
\end{document}