LateXToUnicode Formatter does not handle commands correctly

[stefan-kolb]

that are a substring of a special LaTex command like \c ~ '...

A big problem is to support all kinds of special circumstances in LaTeX code.
The only chance to get the real output is to use the value of a real LaTeX run for stripping LaTeX commands and stuff. Can we somehow write a wrapper for this? Then we could transfer all the hard work to the LaTeX interpreter.

\documentclass{article}
\begin{document}
$\chi$ some other italic \textit{text}
\end{document}

https://github.com/mcmtroffaes/latexcodec/

[simonharrer]

@oscargus how much effort would this be to implement a solution? You are our expert for encodings. :)

[oscargus]

Regarding the encodings in general there is a quite substantial list. The weird thing, which I do not get, is that sometimes the Unicode characters ends up wrong. In the particular case \chi is not converted correctly (Unicode 967) while both \phi (966) and \psi (968) works without any problems... There is a related issue #1264 with another character \v{S} (352) behaving the same way. No obvious reason why exactly those characters should misbehave... In the latter case \v{S} is converted to a square, which can be converted back to \v{S}...

[oscargus]

Implemented in #1581

[matthiasgeiger]

Can delete the branch? @stefan-kolb @oscargus