internal Teritory(EnglishNames names, string id, Dictionary <string, string[]> dir)
{
this.id = id;
name = names.Teritories[id];
var values = dir[id];
foreach (var v in values)
{
if (char.IsDigit(v[0]))
{
if (childs == null)
{
childs = new List <Teritory>();
}
childs.Add(new Teritory(names, v, dir));
}
else
{
if (regions == null)
{
regions = new List <Region>();
}
regions.Add(new Region {
id = v, name = names.Teritories[v]
});
}
}
}
static EnglishNames englishNames()
{
// missing trans from comments in supplements, e.g.
// <languagePopulation type="wbp" populationPercent="0.011"/> <!--Warlpiri-->
var missing = allLangs.Except(langTrans.Keys).ToHashSet();
var els = supplements.
Descendants("languagePopulation").
Select(el => new { el, lang = el.Attribute("type").Value }).Where(el => missing.Contains(el.lang)).Select(el => new {
el.lang, name = (el.el.NextNode as XComment).Value
}).
ToArray();
foreach (var kv in els)
{
langTrans[kv.lang] = kv.name;
}
var english = new EnglishNames {
Teritories = trans("territories"),
Langs = langTrans,
Scripts = trans("scripts"),
};
Json.Serialize(LangsDesignDirs.root + @"patches\cldrEnglish.json", english);
return(english);
}
internal static void createTeritoryTree(EnglishNames names)
{
var ters = supplements.
Descendants("territoryContainment").
Single().Elements("group").
Where(g => g.Attribute("status") == null && g.Attribute("grouping") == null).
ToDictionary(el => el.Attribute("type").Value, el => el.Attribute("contains").Value.Split(' '));
// testUniq is empty => every region is unique in tree
var testUniq = ters.SelectMany(kv => kv.Value.Select(Value => new { kv.Key, Value })).GroupBy(v => v.Value).Where(g => g.Count() > 1);
if (testUniq.Count() > 0)
{
throw new Exception();
}
// build tree
var root = new Teritory(names, "001", ters);
Json.Serialize(LangsDesignDirs.root + @"patches\cldrTeritoryTree.json", root);
}
static void computeLangRegionEnglish(EnglishNames english)
{
var scripts = extractScript().ToDictionary(
s => s.lang,
s => s.script + (s.other == null ? "" : "|" + string.Join(",", s.other)));
// **************************
// ***** REGION-LANG table: <region,lang>,<population,isOfficial,script>
var regionLang = supplements.Descendants("territoryInfo").
Single().
Elements("territory").
SelectMany(el => {
var allPop = int.Parse(el.Attribute("population").Value);
var region = el.Attribute("type").Value;
return(el.Elements("languagePopulation").
Select(ee => {
string lang = ee.Attribute("type").Value, key1 = $"{lang}|{region}", key2 = $"{lang}|";
return new regLang {
region = region,
lang = lang,
population = (int)Math.Round((double)allPop * float.Parse(ee.Attribute("populationPercent")?.Value ?? "0", CultureInfo.InvariantCulture) / 100),
isOfficial = ee.Attribute("officialStatus") != null,
};
}));
}).
ToArray();
// **************************
// ***** population grouping
// group and sum by <region> or <lang>
void groupAndSum(regLang[] data, bool byReg, bool noOffOnly = false)
{
string friendlyName(regLang d, bool isKey, bool sh = false)
{
var actId = byReg == isKey ? d.region : d.lang;
if (sh)
{
return(actId);
}
var actNames = byReg == isKey ? english.Teritories : english.Langs;
return(actNames[actId]);
}
string stringDetail(IGrouping <string, regLang> g, bool officialOnly)
{
var sd = string.Join(", ", g.Where(l => officialOnly == l.isOfficial && l.population > 0).
OrderByDescending(l => l.population).
Select(l => $"{friendlyName(l, false, false)}:{friendlyName(l, false, true)}:{l.population}"));
return(sd == "" ? null : sd);
}
var by = byReg ? data.GroupBy(d => d.region) : data.GroupBy(d => d.lang);
var res = by.Select(g => new LangOrRegion {
name = friendlyName(g.First(), true),
id = g.Key,
offSum = g.Where(l => l.isOfficial).Sum(l => l.population),
nonOffSum = g.Where(l => !l.isOfficial).Sum(l => l.population),
off = stringDetail(g, true),
nonOff = stringDetail(g, false),
likely = byReg ? null : LocaleIdentifier.Parse(g.Key).MostLikelySubtags().ToString(),
scripts = byReg ? null : scripts[g.Key],
}).
Where(r => r.id != "und" && (noOffOnly ? r.offSum == 0 : r.offSum + r.nonOffSum > 0)).
OrderByDescending(r => r.offSum + r.nonOffSum).
ToArray();
var nm = byReg ? "Regions" : (noOffOnly ? "LangsNotOfficial" : "Langs");
Json.Serialize(LangsDesignDirs.root + $"patches\\cldr{nm}.json", res);
}
groupAndSum(regionLang, true);
groupAndSum(regionLang, false, true);
groupAndSum(regionLang, false);
}
