Bad Character Code Error in Math Mode for Unicode Character

Question

I used this solution by a user Cicada from one of the questions on how to type Unicode characters in math mode:

\documentclass{article}
\usepackage{amsmath,amssymb}
\usepackage{polyglossia}
\setmainlanguage{arabic}
% Credits: Cicada's answer from https://bit.ly/3dpDg9t
\setmainfont[Script=Arabic]{Amiri}
\DeclareSymbolFont{arabicletters}{\encodingdefault}{Amiri(0)}{m}{n}
\newcommand{\makearabicmathletter}[1]{%
    \begingroup\lccodea=#1\lowercase{\endgroup \Umathcodea}="0 \csname symarabicletters\endcsname\space #1
}
% Normal Arabic Letters
\count255="0605
\loop\ifnum\count255<"064B
  \makearabicmathletter{\count255}
  \advance\count255 by 1
\repeat
% Arabic Mathematical Symbols!!!!!!
\count255="1EE00
\loop\ifnum\count255<"1EEBB
  \makearabicmathletter{\count255}
  \advance\count255 by 1
\repeat
\begin{document}
  [ 
    \frac{م}{2؈} = ط
  ]
[\char126555 + م]
\end{document}

The first loop seems to work well. I can directly type Arabic letters. However, since the latter loop deals with glyphs which don't exist on usual keyboards, my only chance of using them was to directly type in the character code. However it gave me an error:

! Bad character code (126555).
l.27 \char126555

Overleaf went as far as telling me that I can't use any character number above 65535. What should I do? And why did it work for Cicada, when they used Egyptian hieroglyphics, and not now?

EDIT: Credit where credit's due, the user egreg discovered the lccode trick.

It may be an Overleaf problem. Unrelated: the lcc trick is @egreg 's. — Cicada, Apr 18 '21 at 06:10

score 2 · Accepted Answer · answered Apr 18 '21 at 07:59

Apparently, XeTeX doesn't like \char in math mode. You can use \Umathchar instead, hidden in a macro for ease of use.

\documentclass{article}
\usepackage{amsmath,amssymb}
\usepackage{polyglossia}
\setmainlanguage{arabic}
% Credits: Cicada's answer from https://bit.ly/3dpDg9t
\setmainfont[Script=Arabic]{Amiri}
\DeclareSymbolFont{arabicletters}{\encodingdefault}{Amiri(0)}{m}{n}
\newcommand{\makearabicmathletter}[1]{%
    \begingroup\lccodea=#1\lowercase{\endgroup \Umathcodea}="0 \symarabicletters\space #1\relax
}
% Normal Arabic Letters
\count255="0605
\loop\ifnum\count255<"064B
  \makearabicmathletter{\count255}
  \advance\count255 by 1
\repeat
% Arabic Mathematical Symbols!!!!!!
\count255="1EE00
\loop\ifnum\count255<"1EEBB
  \makearabicmathletter{\count255}
  \advance\count255 by 1
\repeat
\newcommand{\achar}[1]{\Umathchar"0 \symarabicletters#1\relax}
\begin{document}
  [ 
    \frac{م}{2؈} = ط
  ]
  [\achar{126555} + م]
\end{document}

Is there more documentation over this stuff, I find it kind of hidden and I have no idea from where it comes from. — Bored Comedy, Apr 18 '21 at 17:40
@BoredComedy \Umathcode is decribed in the XeTeX manual, texdoc xetex — egreg, Apr 18 '21 at 19:28

score 0 · Answer 2 · answered Apr 18 '21 at 06:11

0

Code compiles OK in Lualatex under TexLive2020.

What is the output supposed to look like?

answered Apr 18 '21 at 06:11

Cicada

10,129

That's weird, yours seems ok. Perhaps it's a XeTeX problem? – Bored Comedy Apr 18 '21 at 16:50
In Xelatex, use the glyph directly, ^^^^^1ee5b, instead of via the \char command. – Cicada Apr 19 '21 at 02:45
For convenience, you can define named macros for the symbols, \newcommand\mghain{^^^^^1ee5b} and then use the name: \[\mghain + م\], if swapping over to Lualatex is not available as a choice. – Cicada Apr 19 '21 at 02:53

Bad Character Code Error in Math Mode for Unicode Character

2 Answers2

Linked