public static void Parse(Dictionary <string, LangMatrixRow> res)
{
var googleLocsCodes = File.ReadAllLines(Directory.GetCurrentDirectory() + @"\google\googleTrans.txt").
Select(l => l.Split('\t')).
Select(p => p[1].Split(' ')[0].Replace("**", "")).
ToArray();
var googleLocs = googleLocsCodes.
Select(w => LocaleIdentifier.Parse(w).MostLikelySubtags()).
ToArray();
var oks = googleLocs.
Select(loc => Langs.fullNameToMeta.TryGetValue(loc.ToString(), out Langs.CldrLang cl) ? cl : null).
NotNulls().
ToArray();
var wrongs = googleLocs.
Select(loc => Langs.fullNameToMeta.TryGetValue(loc.ToString(), out Langs.CldrLang cl) ? null : loc.ToString()).
NotNulls().
ToArray();
if (googleLocsCodes.Length != oks.Length)
{
throw new Exception();
}
oks.ForEach((item, idx) => {
var row = LangsDesignLib.adjustNewfulltextDataRow(res, item.Id.ToString());
row.row[7] = googleLocsCodes[idx];
});
}
public static void Build()
{
var langs = Corpus.DownloadWikies.getUrls().Where(u => u.size > /*1000000*/ 0).Select(u => u.name.Split(new string[] { "wi" }, StringSplitOptions.RemoveEmptyEntries)[0]).Distinct().ToArray();
var lmLangs = Langs.meta.Select(l => l.Lang).Distinct().ToArray();
var notInWiki = lmLangs.Except(langs).ToArray();
var validLangs = langs.Where(l => LocaleIdentifier.TryParse(l, out LocaleIdentifier li)).ToArray();
var wikiLocs = validLangs.Select(l => LocaleIdentifier.Parse(l).MostLikelySubtags().ToString()).ToArray();
var oks = wikiLocs.
Select(loc => Langs.fullNameToMeta.TryGetValue(loc.ToString(), out Langs.CldrLang cl) ? cl : null).
NotNulls().
ToArray();
var wrongs = wikiLocs.
Select(loc => Langs.fullNameToMeta.TryGetValue(loc.ToString(), out Langs.CldrLang cl) ? null : loc).
NotNulls().
ToArray();
//ALPHAs
// from clibs\utils\unicode\unicodeBlocks.json
//Armi (http://zuga.net/articles/unicode/script/imperial-aramaic/) and Goth (? https://en.wikipedia.org/wiki/Gothic_alphabet) missing
var alphas = new HashSet <String> {
"Latn", "Zyyy", "Grek", "Copt", "Cyrl", "Armn", "Hebr", "Arab", "Syrc", "Thaa", "Nkoo", "Samr", "Mand", "Deva", "Beng", "Guru", "Gujr", "Orya", "Taml", "Telu", "Knda", "Mlym", "Sinh", "Thai", "Laoo", "Tibt", "Mymr", "Geor", "Hang", "Ethi", "Cher", "Cans", "Ogam", "Runr", "Tglg", "Hano", "Buhd", "Tagb", "Khmr", "Mong", "Limb", "Tale", "Talu", "Bugi", "Lana", "Bali", "Sund", "Batk", "Lepc", "Olck", "Glag", "Tfng", "Hira", "Kana", "Bopo", "Hani", "Yiii", "Lisu", "Vaii", "Bamu", "Sylo", "Phag", "Saur", "Kali", "Rjng", "Java", "Cham", "Tavt", "Mtei"
};
var wrongAlphas = wrongs.Where(l => !alphas.Contains(LocaleIdentifier.Parse(l).Script)).ToArray();
//var path = LangsDesignDirs.cldrRepo;
}
public static IEnumerable <LangMatrixRow> fromNetCultureInfos(LocaleIdentifier[] cldrSpecifics)
{
// get NON cldr culture data
var cldrs = new HashSet <string>(cldrSpecifics.Select(c => c.Language));
return(CultureInfo.GetCultures(CultureTypes.AllCultures).
Select(cu => {
LocaleIdentifier lid = null;
{ try { lid = LocaleIdentifier.Parse(cu.Name); } catch { } }
if (
lid == null ||
string.IsNullOrEmpty(lid.Region) ||
char.IsDigit(lid.Region[0]) ||
cldrs.Contains(lid.Language)
)
{
return null;
}
var res = CldrUtils.getNetRowData(cu, lid.Language, lid.Region, out LocaleIdentifier locId);
return new LangMatrixRow {
lang = locId.ToString(),
row = res,
};
}).
Where(lt => lt != null));
}
public void checkTexts(Dictionary <string, Dictionary <string, string> > protocol)
{
var locId = LocaleIdentifier.Parse(lang);
var wrongs = UnicodeBlocks.checkBlockNames(row, locId.Script);
if (wrongs == null)
{
return;
}
protocol[lang] = wrongs;
}
static string[] getNetRowData(CultureInfo lc, string Language, string Region, out LocaleIdentifier locId)
{
var values = new string[count];
var fmt = lc.DateTimeFormat;
fmt.MonthNames.Take(12).ToArray(values, monthsIdx);
fmt.MonthGenitiveNames.Take(12).ToArray(values, smonthsIdx);
fmt.DayNames.Take(7).ToArray(values, daysIdx);
var script = UnicodeBlocks.getBlockNames(values).Select(kv => kv.Key).Single();
locId = LocaleIdentifier.Parse(string.Format("{0}-{1}-{2}", Language, script, Region));
return(values);
}
public static Langs.CldrLang[] getMissingLangs()
{
var wrongs = new string[] { "ceb", "ht", "hmn", "la", "ny", "sm", "su", Langs.invariantId.Split('-')[0] };
var wrongsEx = wrongs.Select(w => LocaleIdentifier.Parse(w).MostLikelySubtags()).ToArray();
var newLangs = wrongsEx.
Select(l => new Langs.CldrLang {
Id = string.Format("{0}-{1}", l.Language, l.Region),
Lang = l.Language,
ScriptId = l.Script,
DefaultRegion = l.Region,
Regions = new string[] { l.Region }
}).
ToArray();
//var newLangsStr = Json.SerializeStr(newLangs);
return(newLangs);
}
public static void Build()
{
var cldr = Json.Deserialize <Langs.CldrLang[]>(LangsDirs.dirCldrTexts);
var roots = cldr.SelectMany(c => c.Regions.Select(r => LocaleIdentifier.Parse(string.Format("{0}-{1}-{2}", c.Lang, c.ScriptId, r)))).ToArray();
var locs = roots.Select(root => new Locale(root)).ToArray();
// ALPHABETS
var rx = new Regex("[{} ]");
var alphas = locs.Select(loc => {
var localePattern = loc.Find("//characters");
var data = localePattern.SelectChildren(XPathNodeType.Element).OfType <XPathNavigator>().Where(al => al.Name == "exemplarCharacters").Select(al => {
var value = CldrUtils.decodeUnicodeLiteral(al.Value.Normalize().Trim('[', ']'));
value = rx.Replace(value, "");
var keyNode = al.SelectSingleNode("./@type");
var key = keyNode == null ? "root" : keyNode.Value;
//if (key == "numbers" || key == "punctuation" || string.IsNullOrEmpty(value)) return null;
return(new { key, value });
}).Where(n => n != null).ToArray();
return(new {
lang = loc.Id.ToString(),
data
});
}).Where(d => d.data.Length > 0).OrderBy(l => l.lang).ToArray();
Func <string, IEnumerable <LangMatrixRow> > alphasRes = (string key) => alphas.SelectMany(a => a.data.Where(aa => aa.key == key).Select(aa =>
new LangMatrixRow {
lang = a.lang, columnNames = Linq.Items(aa.key).ToArray(), row = Linq.Items(aa.value).ToArray()
}));
var alphaRoot = new LangMatrix(alphasRes("root"), null, true);
var alphaAuxlity = new LangMatrix(alphasRes("auxiliary"), null, true);
var alphaIndex = new LangMatrix(alphasRes("index"), null, true);
var alphaNumbers = new LangMatrix(alphasRes("numbers"), null, true);
var alphaPunctuation = new LangMatrix(alphasRes("punctuation"), null, true);
var patterns = new LangMatrix(locs.Select(loc => {
var localePattern = loc.FindOrDefault("//localeDisplayNames/localeDisplayPattern/localePattern").ToString();
var localeSeparator = loc.FindOrDefault("//localeDisplayNames/localeDisplayPattern/localeSeparator").ToString();
return(new LangMatrixRow {
lang = loc.Id.ToString(),
row = new string[] { localePattern, localeSeparator },
columnNames = new string[] { "pattern", "separator" },
});
}), null, true);
Dictionary <string, Dictionary <string, string> > langsProtocol = new Dictionary <string, Dictionary <string, string> >();
var langs = new LangMatrix(
locs.Select(loc => fromCldr(loc, "//localeDisplayNames/languages")),
langsProtocol, true
);
Dictionary <string, Dictionary <string, string> > scriptsProtocol = new Dictionary <string, Dictionary <string, string> >();
var scripts = new LangMatrix(
locs.Select(loc => fromCldr(loc, "//localeDisplayNames/scripts")),
scriptsProtocol, true
);
Dictionary <string, Dictionary <string, string> > regionsProtocol = new Dictionary <string, Dictionary <string, string> >();
var regions = new LangMatrix(
locs.Select(loc => fromCldr(loc, "//localeDisplayNames/territories")),
regionsProtocol, true
);
langs.save(LangsDesignDirs.cldr + "cldrNameLangs.csv", true);
scripts.save(LangsDesignDirs.cldr + "cldrNameScripts.csv", true);
regions.save(LangsDesignDirs.cldr + "cldrNameRegions.csv", true);
patterns.save(LangsDesignDirs.cldr + "cldrNamePatterns.csv", true);
alphaRoot.save(LangsDesignDirs.cldr + "alphaRoot.csv", true);
alphaAuxlity.save(LangsDesignDirs.cldr + "alphaAuxlity.csv", true);
alphaIndex.save(LangsDesignDirs.cldr + "alphaIndex.csv", true);
alphaNumbers.save(LangsDesignDirs.cldr + "alphaNumbers.csv", true);
alphaPunctuation.save(LangsDesignDirs.cldr + "alphaPunctuation.csv", true);
// save to DART messages
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\cldrNameLangs.msg", Protobuf.ToBytes(matrixToDart(langs)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\cldrNameScripts.msg", Protobuf.ToBytes(matrixToDart(scripts)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\cldrNameRegions.msg", Protobuf.ToBytes(matrixToDart(regions)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\cldrNamePatterns.msg", Protobuf.ToBytes(matrixToDart(patterns)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\alphaRoot.msg", Protobuf.ToBytes(matrixToDart(alphaRoot)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\alphaAuxlity.msg", Protobuf.ToBytes(matrixToDart(alphaAuxlity)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\alphaIndex.msg", Protobuf.ToBytes(matrixToDart(alphaIndex)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\alphaNumbers.msg", Protobuf.ToBytes(matrixToDart(alphaNumbers)));
File.WriteAllBytes(LangsDesignDirs.data + @"langsDesign\alphaPunctuation.msg", Protobuf.ToBytes(matrixToDart(alphaPunctuation)));
// var localePattern = loc.FindOrDefault("//localeDisplayNames/localeDisplayPattern/localePattern").ToString();
// var localeSeparator = loc.FindOrDefault("//localeDisplayNames/localeDisplayPattern/localeSeparator").ToString();
// fromCldr(loc, "//localeDisplayNames/languages");
// fromCldr(loc, "//localeDisplayNames/scripts");
// fromCldr(loc, "//localeDisplayNames/territories");
//});
}
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);
}
static List <LangScripts> extractScript()
{
var validScrits = UnicodeBlocks.ISO15924.ToHashSet();
validScrits.Add("Jpan"); validScrits.Add("Kore"); validScrits.Add("Hans"); validScrits.Add("Hant");
var res = new List <LangScripts>();
foreach (var row in XElement.Load(Directory.GetCurrentDirectory() + @"\cldr\langScripts.xml").Descendants("tr"))
{
// last TD with value.length==4 is SCRIPT
var script = row.Elements().Select(el => el.Value.Trim()).Where(s => s.Length == 4).Last();
// first TR with seven TD child defines LANG
if (row.Elements().Count() == 7)
{
var s = new LangScripts {
lang = row.Descendants("a").Single().Attribute("name").Value
};
res.Add(s);
}
// obsolete script? (info in last node)
var lastNodeValue = (row.LastNode as XElement).Value.Trim();
var obsolete = !validScrits.Contains(script) || lastNodeValue == "N" || lastNodeValue == "n/a";
//
if (!obsolete)
{
res.Last().other.Add(script);
}
}
res = res.Where(l => allLangs.Contains(l.lang)).ToList();
foreach (var s in res)
{
// no script found => add "2?" + likely
var likely = LocaleIdentifier.Parse(s.lang).MostLikelySubtags();
if (s.other.Count == 0)
{
s.script = "2?" + likely.Script;
s.other = null;
continue;
}
// likely is not in found script => add "1?" + likely
s.script = s.other.Contains(likely.Script) ? likely.Script : "1?" + likely.Script;
s.other.Remove(likely.Script);
if (s.other.Count == 0)
{
s.other = null;
}
}
// add missing langs: add script from lang (e.g. Arab for uz_Arab) OR "3?" + likely
foreach (var lang in allLangs.Except(res.Select(l => l.lang)))
{
res.Add(new LangScripts {
lang = lang, script = lang.Contains('_') ? lang.Split('_')[1] : "3?" + LocaleIdentifier.Parse(lang).MostLikelySubtags().Script, other = null
});
}
foreach (var lang in res)
{
lang.name = langTrans[lang.lang];
}
// sort, write and return
res = res.OrderByDescending(s => s.other == null ? 0 : s.other.Count).ThenBy(s => s.lang).ToList();
Json.Serialize(LangsDesignDirs.root + $"patches\\cldrScript.json", res);
return(res);
}