Make an npm module

[jamesarosen]

This depends on https://github.com/jamesarosen/CLDR.js, which isn't an npm module. There's already a cldr project in npm, but it isn't browser-friendly because it requires a CLDR data XML file.

[jamesarosen]

@papandreou would it be at all possible to make https://github.com/papandreou/node-cldr browser-friendly? If not, I can register my CLDR.js as in npm as "cldr-browser" to avoid the conflict with yours.

[papandreou]

@jamesarosen: node-cldr is just the code for extracting stuff from CLDR, so it's not meant to be used in a browser. However, it's very useful if you're building a browser-compatible library. I'm building one myself: https://github.com/papandreou/inter -- but I chose to move all the non-opinionated, boring extraction code into a separate library so others could benefit from it.

It would be pretty easy to recreate your CLDR.js with node-cldr as a build-time dependency. It looks like you've hand-transcribed those pluralization rules?

[jamesarosen]

I sure did! That's actually not a bad idea: build a grunt task to generate a library from node-cldr. I wasn't able to find the pluralization rules in inter itself; perhaps that's a TODO. Let me know if you'd like to collaborate on any of this stuff. I'm all for reducing redundancy if the use cases match up well enough. My CLDR.js is meant to support Ember-I18n.

[papandreou]

I'd be happy to lend a hand! Something like this could generate the main part of your plurals.js (except for the enum):

var cldr = require('cldr')('/path/to/cldr'),
    localeIdsByPluralRuleFunctionSource = {};

cldr.localeIds.forEach(function (localeId) {
    var pluralRuleFunctionSource = cldr.extractPluralRuleFunction(localeId).toString().replace(/^function anonymous/, "function"); // Fix V8-ism
    (localeIdsByPluralRuleFunctionSource[pluralRuleFunctionSource] = localeIdsByPluralRuleFunctionSource[pluralRuleFunctionSource] || []).push(localeId);
});

Object.keys(localeIdsByPluralRuleFunctionSource).forEach(function (pluralRuleFunctionSource) {
    var localeIds = localeIdsByPluralRuleFunctionSource[pluralRuleFunctionSource];
    console.log("define([" + localeIds.map(function (localeId) {return "'" + localeId + "'";}).join(",") + '], ' + pluralRuleFunctionSource + ');');
});

Output:

define(['fil','ti','am_et','nso','ak','ln_cf','ak_gh','ln_cd','fil_ph','ti_et','hi','mg','hi_in','nso_za','ln_cg','ln','ln_ao','mg_mg','am','ti_er'], function(n) {
if(n>=0&&n<=1)return"one";return"other"
});
define(['en_ag','seh_mz','en_ug','rwk_tz','en_ie','es_gt','pt_st','fur_it','pa_guru_in','nr','en_bm','seh','ml_in','tn_za','sq_mk','it_sm','rof','ee_tg','en_to','os_ge','jgo_cm','el_gr','ta_in','en_pg','es_bo','nn','fur','kl','mgo','fo_fo','ss_sz','wae','en_bs','en_gu','gl_es','xog','es_cr','es_do','sn_zw','en_ky','ur','es_cl','en_us_posix','so_so','en_tz','ca_ad','ky','en_pr','en_ke','en_bb','nl_aw','ky_kg','om_ke','gsw','de_lu','mas_ke','nl_cw','as','en_zm','ksb','en_ws','en_us','en_gy','bn_bd','saq','en_jm','bg_bg','is_is','mas','rm_ch','te','kk','ee_gh','ha_latn_ng','ks','bez','en','es_cu','af','es_ea','en_ls','en_za','bem','en_je','da','en_be','tn_bw','pa','kkj_cm','nyn','sq','ps','tn','es_pe','en_mg','en_as','nr_za','nl_sx','vo','ha_latn','en_zw','mr_in','ta_my','ta','wae_ch','as_in','es_ni','ee','cgg_ug','en_gg','st_ls','ks_arab','ssy_er','pt_mz','brx_in','en_vg','nnh','xh_za','de_ch','ssy','el','haw','nd','os_ru','en_in','de','pt_cv','pt_br','en_vc','nyn_ug','sw_tz','so','pt_pt','bn','en_pk','zu','es_uy','ha_latn_ne','en_tc','kk_cyrl_kz','zu_za','chr_us','ha_latn_gh','nn_no','pa_guru','haw_us','is','gu_in','pt_tl','en_au','pa_arab','ml','tig','es_ve','en_dsrt_us','gu','es_ar','en_mh','en_ph','rwk','ss','es_mx','vun_tz','sv_se','pa_arab_pk','nl_be','en_kn','or','en_hk','en_pw','gsw_ch','bg','en_gd','fo','ks_arab_in','es_419','en_sg','st','ksb_tz','teo_ke','asa_tz','mr','de_at','sv','lg','da_dk','asa','pt_mo','ha','en_sc','os','or_in','en_gh','en_lr','cgg','en_tt','kkj','xh','ca_es','es_pr','sw','sw_ke','es_gq','saq_ke','en_150','en_gm','en_na','en_gi','vun','en_lc','nb','so_dj','xog_ug','nd_zw','af_za','ve_za','en_sz','en_ca','so_ke','te_in','es_co','bn_in','el_cy','ast','ur_in','ne_np','pt_ao','es_pa','de_li','ts_za','en_vu','sv_ax','de_de','sn','de_be','nl_sr','eu_es','teo','rof_tz','ta_sg','es_us','lg_ug','sw_ug','en_bz','rm','eu','gl','en_mt','ur_pk','en_ki','bem_zm','af_na','en_bw','en_mw','ast_es','fi','sv_fi','jgo','eo','en_im','jmc_tz','ne_in','es','en_dsrt','en_ss','en_fm','pt','es_hn','en_sl','en_mp','en_vi','es_ec','it_ch','bez_tz','brx','nnh_cm','en_nz','om_et','fi_fi','nl_nl','es_py','es_ph','om','et_ee','en_cm','nb_no','en_mu','mas_tz','et','es_ic','en_gb','kk_cyrl','it','chr','sq_al','ve','ta_lk','ts','mgo_cm','so_et','teo_ug','ps_af','ss_za','es_es','en_dm','nl','st_za','es_sv','en_fj','ne','en_um','kl_gl','tig_er','ca','jmc','en_ng','it_it','pt_gw','en_sb'], function(n) {
if(n===1)return"one";return"other"
});

... And so on.

I wasn't able to find the pluralization rules in inter itself; perhaps that's a TODO

It's a little weird right now, and the docs are outdated. My plans is to clean it up and AMD-ify it properly, but I need to fix some things in my build system to make it work first.

[jamesarosen]

@papandreou one of the things we're considering is a npm library that has node-closure as a dependency and writes out I18n-related modules (plurals, number_formats, etc.). Closure Compiler has all that logic built-in -- no need to parse it out of XML files.

[papandreou]

@jamesarosen That'd be another fine solution, although it would be one step further from the real data source (Closure also gets all that data from CLDR). You'd be reliant on both node-closure and Closure staying alive and releasing new versions of their code in a timely manner.

Sorry I can't promise any time frame with inter. I'm mostly struggling with coming up with a nice API that works well in both node.js and the browser so that things are still split up enough that you only pay for the features that you use. Things like localized time zone lists need to be split up per locale because they're so huge, whereas pluralization rules are compact enough that a single lib can contain the data for all locales. My current approach is to use buildInter --bundle to build one huge library with all the features and locales you need for your project, and then use uglifyjs --define LOCALEID="en_us" etc. to split it into trimmed-down libraries for production that only contain one locale each. Not exactly self-explanatory.

[jamesarosen]

Great info, thanks :)

We'll try building a smaller library from both closure and node-cldr and see how they go.