Unicode characters changing order in listings (XeTeX) [duplicate]

Question

I am trying to input code listings with unicode characters and everything seems to work (more or less) except some unicode characters being displayed out of order. For example instead of

służąca do obsługi

I get

łżąsuca do łobsugi

How can I fix that?

EDIT: I was asked for providing some code, but unfortunately I doubt it will be any helpful. Basically code like this:

\documentclass[11pt,polish]{article}

\usepackage{fontspec}
\usepackage{listings}
\usepackage{xunicode}
\usepackage{xltxtra}
\usepackage{babel}

\setmainfont{Charis SIL} % Some font with good unicode support

\begin{document}
\begin{lstlisting}
służący
\end{lstlisting}
\end{document}

will output this:

łżą

sucy

(yes there is a line break, I don't know where it came from, when I input code in form of a file there are characters out of order, but no line breaks).

Answer 1

Package listings is not aware of characters with character code greater than 8-bit. Internally the package redefines each character as active character. The following example allows the support of additional Unicode characters by adding them to the list of extended chars.

Supporting all Unicode characters this way is difficult:

• High memory usage.
• Bad run time behavior.
• There are different classes of characters (\lst@ProcessLetter, \lst@ProcessOther, …)

Example file:

\documentclass[11pt,polish]{article}

\usepackage{fontspec}
\usepackage{listings}
\lstset{extendedchars}

\begingroup
  \catcode0=12 %
  \makeatletter
  \g@addto@macro\lst@DefEC{%
    \lst@CCECUse\lst@ProcessLetter
    łżą% *** add Unicode characters ***
    ^^00% end marker
  }%
\endgroup

\begin{document}
\begin{lstlisting}
służący
\end{lstlisting}
\end{document}

Answer 2

Here's another solution, if listings isn't a requirement. My PythonTeX package provides syntax highlighting for most languages, via the Python highlighting library Pygments. The package offers full Unicode support, under pdflatex, lualatex, and xelatex. It works under Windows, Linux, and OS X.

PythonTeX provides a pygments environment for typesetting blocks of code, a \pygment command for inline use, and an \inputpygments command for bringing in external files. All of these take optional fancyvrb commands, so you can easily get features like line numbering.

Here's a slight modification of your minimal example:

\documentclass[11pt,polish]{article}

\usepackage{fontspec}
\usepackage{pythontex}
\usepackage{xunicode}
\usepackage{xltxtra}
\usepackage{babel}

\setmainfont{Charis SIL} % Some font with good unicode support

\begin{document}

% Just change `text' to the correct language for your code
\begin{pygments}{text}
służący
służąca do obsługi
łżąsuca do łobsugi
\end{pygments}

\end{document}

If your source code uses any tabs, you should use xelatex with the -8bit option, so that tabs can be correctly written to file by xelatex (see this for details).