On Thu, Jan 06, 2011 at 02:42:50PM +0100, Philipp Stephani wrote: > Am 06.01.2011 um 13:12 schrieb Khaled Hosny: > > > May be you can use "process_input_buffer" callback, but this gets the > > raw input before luatex even processing it so you have to be careful > > with control sequences and so. > > It is next to impossible to find out from the input where math mode is > active, given the millions of possibilities to start math input.
That what I meant, he either be careful about what to replace or things get messed up. > > Or you can do the processing on the node list, but then you can't used > > regular text processing functions (gsub, lpeg, etc.) > > And before you arrive at a node list, TeX will already have complained about > ^ outside math mode. Probably a token_filter callback would work better. I was not thinking about this particular case, one can use tex-friendly transliteration scheme but still need to escape non-textual material which makes node list processing appealing. I would have suggested token_filter but tex tokens are beyond my understanding. Regards, Khaled -- Khaled Hosny Arabic localiser and member of Arabeyes.org team Free font developer
