languageToBCP47 function is wrong #795
Description
languageToBCP47 function is defined in src/helpers/index.js
The comment says: loleaflet expects a BCP47 language tag syntax
However, the function does not provide this syntax.
The first problem is that region code is converted to lower case. It should be upper case, because the LOKit expects it to be in upper case. So for example, if you pass en-gb to LOKit, it will not understand it, and will use en-US instead of the intended en-GB.
The second problem is that this function creatively mixes language codes and locale codes. For example when the user selects language "English (US)" and locale "German (Switzerland)", the function generates the "en-ch" language code, which makes little sense, at least LOKit does not understand it. Even in theory it is not possible to get the right information from this string, because "ch" is ambiguous. What number format, or weekday names etc. would that imply? German? French? Italian?
Third problem is that it only matches 2 letter language codes. For example Asturian (ast) will turn into Assamese (as). Quite a difference. :)
Probably loleaflet should accept an optional "locale" parameter besides "lang". And LOKit API should be extended. Currently it has only one "Language" parameter and it is not possible set a different UI language and locale (and default currency), as it is possible in desktop LibreOffice.
As a first step, languageToBCP47 should return standard BCP47 language tags, preferably only such that loleaflet and LOKit can deal with.
These are:
ab ar be bo ca da dz eo fa ga gug hsb it kab kmr-Latn ks lt ml my nn om pt rw sd sl ss szl th ts ur vi zu
af as bg br ca-valencia de el es fi gd he hu ja kk kn ky lv mn nb nr or pt-BR sah si sq st ta ti tt uz xh
am ast bn brx cs dgo en-GB et fr gl hi id jv kl ko lb mai mni ne nso pa-IN ro sa-IN sid sr sv te tn ug ve zh-CN
an az bn-IN bs cy dsb en-ZA eu fy gu hr is ka km kok lo mk mr nl oc pl ru sat sk sr-Latn sw-TZ tg tr uk vec zh-TW