// look up 3 letter codes
private LanguageInfo Set3LetterCode(LanguageInfo langInfo)
{
if (!Force3LetterCodes)
{
return(langInfo);
}
if (langInfo.Code.Length == 3)
{
return(langInfo);
}
var found = ThreeToTwoLetter.Where(p => p.Value == langInfo.Code).Select(p => p.Key).FirstOrDefault();
if (!string.IsNullOrEmpty(found))
{
langInfo.Code = found;
}
return(langInfo);
}
private LanguageInfo GetOrCreateLanguageFromCode(string code, string country)
{
LanguageInfo language;
var countryName = CountryCodeToCountryName[country];
if (!CodeToLanguageIndex.TryGetValue(code, out language))
{
language = new LanguageInfo()
{
Code = code, Country = countryName
};
CodeToLanguageIndex.Add(code, language);
}
else
{
if (!language.Country.Contains(countryName))
{
language.Country += ", " + countryName;
}
}
return(language);
}
public EthnologueLookup()
{
foreach (var line in LanguageRegistryResources.TwoToThreeCodes.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries))
{
var items = line.Split('\t');
ThreeToTwoLetter.Add(items[1].Trim(), items[0].Trim());
}
foreach (var line in LanguageRegistryResources.CountryCodes.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries))
{
var items = line.Split('\t'); //id name area
CountryCodeToCountryName.Add(items[0].Trim(), items[1].Trim());
}
//LanguageIndex.txt Format: LangID CountryID NameType Name
//a language appears on one row for each of its alternative langauges
string[] entries = LanguageRegistryResources.LanguageIndex.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string entry in entries.Skip(1)) //skip the header
{
var items = entry.Split('\t');
if (items.Length != 4)
{
continue;
}
var code = items[0].Trim();
string TwoLetterCode;
if (ThreeToTwoLetter.TryGetValue(code, out TwoLetterCode))
{
code = TwoLetterCode;
}
LanguageInfo language = GetOrCreateLanguageFromCode(code, items[1].Trim());
var name = items[3].Trim();
if (items[2] == "L")
{
while (language.Names.Contains(name))
{
language.Names.Remove(name);
}
language.Names.Insert(0, name);
}
else
{
if (!language.Names.Contains(name))
{
language.Names.Add(name); //intentionally not lower-casing
}
}
}
foreach (var languageInfo in CodeToLanguageIndex.Values)
{
foreach (var name in languageInfo.Names)
{
GetOrCreateListFromName(name).Add(languageInfo);
}
}
}
public EthnologueLookup()
{
Force3LetterCodes = false;
foreach (var line in LanguageRegistryResources.TwoToThreeCodes.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries))
{
var items = line.Split('\t');
ThreeToTwoLetter.Add(items[1].Trim(), items[0].Trim());
}
foreach (var line in LanguageRegistryResources.CountryCodes.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries))
{
var items = line.Split('\t'); //id name area
CountryCodeToCountryName.Add(items[0].Trim(), items[1].Trim());
}
CountryCodeToCountryName.Add("?", "?"); //for unlisted language
//LanguageIndex.txt Format: LangID CountryID NameType Name
//a language appears on one row for each of its alternative langauges
List <string> entries = new List <string>(LanguageRegistryResources.LanguageIndex.Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries));
entries.Add("qaa\t?\tL\tUnlisted Language");
foreach (string entry in entries.Skip(1)) //skip the header
{
var items = entry.Split('\t');
if (items.Length != 4)
{
continue;
}
var code = items[0].Trim();
string TwoLetterCode;
if (ThreeToTwoLetter.TryGetValue(code, out TwoLetterCode))
{
code = TwoLetterCode;
}
LanguageInfo language = GetOrCreateLanguageFromCode(code, items[1].Trim());
var name = items[3].Trim();
if (items[2] == "L")
{
while (language.Names.Contains(name))
{
language.Names.Remove(name);
}
language.Names.Insert(0, name);
}
else
{
if (!language.Names.Contains(name))
{
language.Names.Add(name); //intentionally not lower-casing
}
}
}
//Why just this small set? Only out of convenience. Ideally we'd have a db of all languages as they write it in their literature.
this.CodeToLanguageIndex["fr"].LocalName = "français";
this.CodeToLanguageIndex["es"].LocalName = "español";
this.CodeToLanguageIndex["zho"].LocalName = "中文"; //chinese
this.CodeToLanguageIndex["hi"].LocalName = "हिन्दी"; //hindi
this.CodeToLanguageIndex["bn"].LocalName = "বাংলা"; //bengali
this.CodeToLanguageIndex["te"].LocalName = "తెలుగు"; //telugu
this.CodeToLanguageIndex["ta"].LocalName = "தமிழ்"; //tamil
this.CodeToLanguageIndex["ur"].LocalName = "اُردُو"; //urdu
this.CodeToLanguageIndex["ar"].LocalName = "العربية/عربي"; //arabic
this.CodeToLanguageIndex["th"].LocalName = "ภาษาไทย"; //thai
this.CodeToLanguageIndex["id"].LocalName = "Bahasa Indonesia"; //indonesian
foreach (var languageInfo in CodeToLanguageIndex.Values)
{
foreach (var name in languageInfo.Names)
{
GetOrCreateListFromName(name).Add(languageInfo);
}
if (!string.IsNullOrEmpty(languageInfo.LocalName))
{
GetOrCreateListFromName(languageInfo.LocalName).Add(languageInfo);
}
}
}
