Script for separating groups and kerning based on language.
Groups with base language (such as Latin, identified by the first glyph in the group) will remain with names like public.kern1.A, and for other languages will be formed classes like public.kern1.A_cyrillic or public.kern1.A_greek with the language name suffix through an underscore _, kerning will be transferred to all related classes accordingly, and cross-language pairs are removed.
The script removes cross-language pairs.
Pairs are checked for the first level of compatibility - Latin/Cyrillic, Latin/Greek, Georgian/Armenian, etc. It is also possible to check and delete pairs according to the second level of compatibility, pairs within the same script - for example, in Latin: French / German, in Cyrillic: Abkhaz / Chuvash. But this is rarely applied in practice.
The script collects groups with language suffixes like _cyrillic _greek and so on, will be combined with similar Latin ones.
This script is currently under development. Main problem - a lot of exceptions are thrown.