linguistic sorting of country names

[nkhine]

I am using this library to localize country names to be displayed as an alpha sorted list for the http://www.zmgc.net website - TZM Network tab

If I change the language to Slovakian for example Čína is listed at the end of the list.

My list comes from https://github.com/TZM/tzm-blade/blob/master/data/chapters.json file and is being sorted by the following code, note i am using https://github.com/bminer/node-blade template engine which is similar to jade, here is a link to the actual file https://github.com/TZM/tzm-blade/blob/master/views/footer.blade:

var sortByKey = function(field, reverse, primer){
    var key = function (x) {return primer ? primer(x[field]) : x[field]};
    return function (a,b) {
        var A = key(a), B = key(b);
        return ((A < B) ? -1 : (A > B) ? +1 : 0) * [-1,1][+!!reverse];                  
    }
}
var countries = [];
var chapters = chapterJSON
    for(var i in chapters)
        var row = chapters[i].desc
        var code = row["LOCALES"].split("-")[1]
        countries.push({link:row["WEBSITE"],contact:row["CONTACT"],country: allCountries[code]})
var guide = locals.settings.translation.guide
countries.sort(sortByKey('country', true))

The countries list comes from https://github.com/TZM/tzm-blade/blob/master/app/config/apps.coffee#L171

Any advise on how to best improve the code to take into account the linguistic sorting use case would be much appreciated.

[papandreou]

CLDR seems to have all the collations you would ever dream of (https://github.com/papandreou/node-cldr/tree/master/3rdparty/cldr/common/collation) so I suppose it'd be possible to build an Array.prototype.sort-compatible comparator function for each locale from that.

I never looked into actually doing it because it seemed like String.prototype.localeCompare would do the trick. Unfortunately that seems to be crippled in node.js: https://groups.google.com/forum/#!topic/nodejs/edVSlqwM3qM

I guess that makes it worth doing, but there's quite a bit of spec to read up on before even understanding the data: http://www.unicode.org/reports/tr35/tr35-collation.html

Patches very welcome :)

[nkhine]

thanks for the reply, will look at the links and see if i can do something.

i just tested the String.prototype.localeCompare and this has been fixed

☺  locale | grep LC_COLLATE
LC_COLLATE="en_GB.UTF-8"
☺  node
> var x='�', y='�'
undefined
> x.localeCompare(y)
0