3-letter language & country codes

[rwd]

I would like to implement this todo from the README:

include other language/country code formats (3-letter codes...) ?

In the process, I would also like to propose some refactoring. What I have in mind here is:

1. Use 3-letter language and country codes in filenames and for lookup keys in data caches, because all languages have a 3-letter code, but not all have a 2-letter code.
2. Use codes in the case they are specified in the relevant ISO standard, i.e. lower-case for language codes, upper-case for country codes.
3. Change the storage format of the cached data to JSON, e.g.
   Languages in English:

   {
     "eng": "English",
     "fra": "French",
     "spa": "Spanish; Castilian"
   }

4. Rename the cache files thus:
   • Countries: /cache/countries/eng.json, /cache/countries/fra.json, etc
   • Languages: /cache/languages/eng.json, /cache/languages/fra.json, etc
5. Make the live data provider add to the cache two JSON dictionaries of alternate codes (2-letter/bibliographic/numeric) mapping them to their 3-letter equivalent, e.g.
   Language code dictionary:

   {
     "en": "eng",
     "es": "spa",
     "fr": "fra",
     "chi": "zho"
   }

   Country code dictionary:

   {
     "GMB": "GM",
     "887": "YEM"
   }

6. When performing lookups by code, if the supplied code is unknown, i.e. has no matching cache file, lookup a 3-letter equivalent in the relevant dictionary and use that.

Before I get started working on a PR for the above, is there general support for some/any/all of this proposal?

[grosser]

Sounds good.

An alternative might be keeping 2-letter codes around and adding 3-letter files where necessary and then doing a 3-letter->2-letter lookup on missing, but that might get more complicated than jjust doing 3-letter lookup.

[grosser]

Make sure to keep the 2-3 lookup-table in memory so we don't load that file all the time ;)