On 7 November 2017 at 20:38, Thomas A. Schmitz wrote:
When mkiv was in its infancy, Hans helped me in writing something like this for my Greek module. It basically applies a Lua string.gsub to the input to produce and typeset utf8 output.
This would be done with font features now. So it would in fact not be applied at input (which is more dangerous), but rather before the characters end up in PDF which makes a lot more sense anyway. The documentation is here, but there should be plenty more examples: http://pragma-ade.com/general/manuals/fonts-mkiv.pdf
I can send you the relevant code if you want, and you could adapt it to your case.
I'm pretty sure it's outdated if it was written at the infancy stage of mkiv.
But I would advise against it. In the long run, changing keyboards is less hassle than this sort of semi-solution to an obsolete problem.
One reason why I would not abandon it *immediately* for Serbian is that Serbian can actually be written in both scripts and there are straightforward rules for transliteration (it's not exactly one-to-one character, but those "dj"s should be easy enough to handle, in particular because there are also all the required digraphs in Unicode - or maybe more difficult exactly because of that). The fun part is that they cannot decide which script to use themselves :), so you end up with schoolmates using different scripts for their lecture notes. I'm still not arguing that this is the most brilliant idea, but I can totally imagine a Serbian professor wanting to "auto-generate" a Cyrillic version of his book on top of the Latin edition with close-to-zero extra effort. Greek, in contrast, hardly makes any sense when written in Latin alphabet. Mojca