static StandardSubtags()
{
// JohnT: can't find anywhere else to document this, so here goes: TwoToThreeMap is a file adapted from
// FieldWorks Ethnologue\Data\iso-639-3_20080804.tab, by discarding all but the first column (3-letter
// ethnologue codes) and the fourth (two-letter IANA codes), and all the rows where the fourth column is empty.
// I then swapped the columns. So, in this resource, the string before the tab in each line is a 2-letter
// Iana code, and the string after it is the one we want to return as the corresponding ISO3Code.
// The following block of code assembles these lines into a map we can use to fill this slot properly
// when building the main table.
var twoToThreeMap = new Dictionary<string, string>();
string[] encodingPairs = LanguageRegistryResources.TwoToThreeCodes.Replace("\r\n", "\n").Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries);
foreach (string pair in encodingPairs)
{
var items = pair.Split('\t');
if (items.Length != 2)
continue;
twoToThreeMap[items[0]] = items[1];
}
var languages = new List<LanguageSubtag>();
var scripts = new List<ScriptSubtag>();
var regions = new List<RegionSubtag>();
var variants = new List<VariantSubtag>();
string[] ianaSubtagsAsStrings = LanguageRegistryResources.ianaSubtagRegistry.Split(new[] { "%%" }, StringSplitOptions.None);
foreach (string ianaSubtagAsString in ianaSubtagsAsStrings)
{
string[] subTagComponents = ianaSubtagAsString.Replace("\r\n", "\n").Split(new[] { "\n" }, StringSplitOptions.RemoveEmptyEntries);
if (subTagComponents[0].Contains("File-Date"))
{
continue; //This is the first line of the file.
}
CheckIfIanaSubtagFromFileHasExpectedForm(subTagComponents);
string type = subTagComponents[0].Split(' ')[1];
string subtag = subTagComponents[1].Split(' ')[1];
string description = SubTagComponentDescription(subTagComponents[2]);
if (subtag.Contains("..")) // do not add private use subtags to the list
{
continue;
}
/* Note: currently we are only using the first "Description:" line in each entry.
* A few script entries contain multiple Description: lines, as in the example below:
*
* Type: script
* Subtag: Deva
* Description: Devanagari
* Description: Nagari
* Added: 2005-10-16
*
* In the future it may be necessary to build a separate iana script entry collection
* that contains duplicate script codes, for the purposes of including all possible
* script Descriptions.
*/
switch (type)
{
case "language":
string iso3Code;
if (!twoToThreeMap.TryGetValue(subtag, out iso3Code))
iso3Code = subtag;
languages.Add(new LanguageSubtag(subtag, description, false, iso3Code));
break;
case "script":
scripts.Add(new ScriptSubtag(subtag, description, false));
break;
case "region":
regions.Add(new RegionSubtag(subtag, description, false));
break;
case "variant":
variants.Add(new VariantSubtag(subtag, description, false, GetVariantPrefixes(subTagComponents)));
break;
}
}
IEnumerable<LanguageSubtag> sortedLanguages = languages.OrderBy(l => Regex.Replace(l.Name, @"[^\w]", ""))
.Concat(new[] {new LanguageSubtag(WellKnownSubtags.UnlistedLanguage, "Language Not Listed", true, string.Empty)});
RegisteredLanguages = new ReadOnlyKeyedCollection<string, LanguageSubtag>(new KeyedList<string, LanguageSubtag>(sortedLanguages, l => l.Code, StringComparer.InvariantCultureIgnoreCase));
RegisteredScripts = new ReadOnlyKeyedCollection<string, ScriptSubtag>(new KeyedList<string, ScriptSubtag>(scripts.OrderBy(s => s.Name), s => s.Code, StringComparer.InvariantCultureIgnoreCase));
RegisteredRegions = new ReadOnlyKeyedCollection<string, RegionSubtag>(new KeyedList<string, RegionSubtag>(regions.OrderBy(r => r.Name), r => r.Code, StringComparer.InvariantCultureIgnoreCase));
RegisteredVariants = new ReadOnlyKeyedCollection<string, VariantSubtag>(new KeyedList<string, VariantSubtag>(variants.OrderBy(v => v.Name), v => v.Code, StringComparer.InvariantCultureIgnoreCase));
CommonPrivateUseVariants = new ReadOnlyKeyedCollection<string, VariantSubtag>(new KeyedList<string, VariantSubtag>(new[]
{
new VariantSubtag(WellKnownSubtags.IpaPhoneticPrivateUse, "Phonetic"),
new VariantSubtag(WellKnownSubtags.IpaPhonemicPrivateUse, "Phonemic"),
new VariantSubtag(WellKnownSubtags.AudioPrivateUse, "Audio")
}, v => v.Code, StringComparer.InvariantCultureIgnoreCase));
Iso3Languages = RegisteredLanguages.Where(l => !string.IsNullOrEmpty(l.Iso3Code)).ToDictionary(l => l.Iso3Code, StringComparer.InvariantCultureIgnoreCase);
}
