public void MostLikely()
{
Assert.AreEqual("en-Latn-US", LocaleIdentifier.Parse("en").MostLikelySubtags().ToString());
Assert.AreEqual("zh-Hans-CN", LocaleIdentifier.Parse("zh").MostLikelySubtags().ToString());
Assert.AreEqual("zh-Hant-TW", LocaleIdentifier.Parse("zh-TW").MostLikelySubtags().ToString());
Assert.AreEqual("zh-Hans-SG", LocaleIdentifier.Parse("ZH-ZZZZ-SG").MostLikelySubtags().ToString());
}
public void Parsing_Language_Variant()
{
var id = LocaleIdentifier.Parse("sl-nedis");
Assert.AreEqual("sl", id.Language);
Assert.IsTrue(id.Variants.Contains("nedis"));
}
public void Parsing_Normalisation()
{
Assert.AreEqual("en-US", LocaleIdentifier.Parse("eng-840").ToString());
Assert.AreEqual("en-NZ", LocaleIdentifier.Parse("eng-NZ").ToString());
Assert.AreEqual("en-GB", LocaleIdentifier.Parse("en-UK").ToString());
Assert.AreEqual("en-GB", LocaleIdentifier.Parse("eng-UK").ToString());
}
public void Parsing_Language()
{
var id = LocaleIdentifier.Parse("en");
Assert.AreEqual("en", id.Language);
Assert.IsFalse(LocaleIdentifier.TryParse("NotALanguageCode", out id));
}
public void Parsing_Underscore_Separator()
{
var id = LocaleIdentifier.Parse("zh_Hant_CN");
Assert.AreEqual("zh", id.Language);
Assert.AreEqual("Hant", id.Script);
Assert.AreEqual("CN", id.Region);
}
public void Parsing_Script()
{
var id = LocaleIdentifier.Parse("Cyrl");
Assert.AreEqual("Cyrl", id.Script);
Assert.IsFalse(LocaleIdentifier.TryParse("NotAScript", out id));
}
/// <summary>
/// Tries parsing the string representation of a locale identifier.
/// </summary>
/// <param name="s">
/// A case insensitive string containing a locale identifier, based on BCP47.
/// </param>
/// <param name="result">
/// A local identifier that refers to <paramref name="s"/> or <b>null</b> if the parsing
/// failed.
/// </param>
/// <param name="message">
/// The reason why the parsing failed.
/// </param>
/// <returns>
/// <b>true</b> if <paramref name="s"/> was parsed successfully; otherwise, <b>false</b>.
/// </returns>
/// <remarks>
/// A locale identifier that refers to <paramref name="s"/>.
/// </remarks>
public static bool TryParse(string s, out LocaleIdentifier result, out string message)
{
if (TryParseBcp47(s, out result, out message))
{
message = result.TransformFromBcp47();
}
return(message == null);
}
public void SearchChain_root()
{
var id = LocaleIdentifier.Parse("root");
var chain = id.SearchChain().ToArray();
Assert.AreEqual(1, chain.Length);
Assert.AreEqual("root", chain[0]);
}
public void Parsing_Language_Region_Variant()
{
var id = LocaleIdentifier.Parse("de-CH-1996");
Assert.AreEqual("de", id.Language);
Assert.AreEqual("CH", id.Region);
Assert.IsTrue(id.Variants.Contains("1996"));
}
/// <summary>
/// A new locale identifier with the all the empty subtags filled in
/// with a likely value.
/// </summary>
/// <returns>
/// A new locale identifier with all subtags filled in.
/// </returns>
public LocaleIdentifier MostLikelySubtags()
{
var result = (LocaleIdentifier)this.MemberwiseClone();
// Remove the script code 'Zzzz' and the region code 'ZZ' if they occur.
if (result.Script == "Zzzz")
{
result.Script = String.Empty;
}
if (result.Region == "ZZ")
{
result.Region = String.Empty;
}
// Short cut if all subtags have a value.
if (result.Language != "" && result.Script != "" && result.Region != "")
{
return(result);
}
// Find the language in likely subtags.
var likely = new[]
{
$"{result.Language}_{result.Script}_{result.Region}",
$"{result.Language}_{result.Region}",
$"{result.Language}_{result.Script}",
$"{result.Language}",
$"und_{result.Script}",
}
.Select(k => k.Replace("__", "_").Trim('_'))
.Where(k => k != String.Empty)
.Distinct()
.Select(k => Cldr.Instance
.GetDocuments("common/supplemental/likelySubtags.xml")
.FirstElementOrDefault($"supplementalData/likelySubtags/likelySubtag[@from='{k}']")
)
.FirstOrDefault(e => e != null);
if (likely != null)
{
var defaults = LocaleIdentifier.ParseBcp47(likely.GetAttribute("to", ""));
if (result.Language == "")
{
result.Language = defaults.Language;
}
if (result.Script == "")
{
result.Script = defaults.Script;
}
if (result.Region == "")
{
result.Region = defaults.Region;
}
}
return(result);
}
public void Parsing_Language_Region()
{
var id = LocaleIdentifier.Parse("es-419");
Assert.AreEqual("es", id.Language);
Assert.AreEqual("419", id.Region);
Assert.IsFalse(LocaleIdentifier.TryParse("es-12345", out id));
}
public void Parsing_Invalid_Unicode_Extension()
{
LocaleIdentifier id;
string message;
var ok = LocaleIdentifier.TryParse("th-u-xx-invalidtype", out id, out message);
Assert.IsFalse(ok);
Assert.IsNull(id);
}
public void Parsing_Language_Script()
{
var id = LocaleIdentifier.Parse("uz-Cyrl");
Assert.AreEqual("uz", id.Language);
Assert.AreEqual("Cyrl", id.Script);
Assert.IsFalse(LocaleIdentifier.TryParse("es-NotScript", out id));
}
public void Parsing_Language_Country()
{
var id = LocaleIdentifier.Parse("en-US");
Assert.AreEqual("en", id.Language);
Assert.AreEqual("US", id.Region);
Assert.IsFalse(LocaleIdentifier.TryParse("en-NotACountryCode", out id));
}
public void Parsing_Script_Region()
{
var id = LocaleIdentifier.Parse("Cyrl-RU");
Assert.AreEqual("Cyrl", id.Script);
Assert.AreEqual("RU", id.Region);
Assert.IsFalse(LocaleIdentifier.TryParse("Cyrl-NotACounty", out id));
}
/// <summary>
/// Creates or reuses a locale with the specified <see cref="LocaleIdentifier"/>.
/// </summary>
/// <param name="id">
/// A locale identifier.
/// </param>
/// <returns>
/// A locale for the specified <paramref name="id"/>.
/// </returns>
/// <remarks>
/// Uses the <see cref="LocaleIdentifier.CanonicalForm"/> of the
/// <paramref name="id"/>.
/// </remarks>
public static Locale Create(LocaleIdentifier id)
{
var cid = id.CanonicalForm();
if (log.IsDebugEnabled)
{
log.DebugFormat("Resolved locale '{0}' to '{1}'", id, cid);
}
return(LocaleCache.GetOrAdd(cid.ToString(), name => new Locale(cid)));
}
public void Parsing_Language_Script_Country()
{
var id = LocaleIdentifier.Parse("zh-Hant-CN");
Assert.AreEqual("zh", id.Language);
Assert.AreEqual("Hant", id.Script);
Assert.AreEqual("CN", id.Region);
Assert.IsFalse(LocaleIdentifier.TryParse("zh-Hant-NotACountry", out id));
}
public void Parsing_Messages()
{
string message;
LocaleIdentifier id;
var ok = LocaleIdentifier.TryParse("en-r-alpha-r-alpha", out id, out message);
Assert.AreEqual(false, ok);
Assert.AreEqual(null, id);
Assert.AreEqual("'en-r-alpha-r-alpha' is not a valid locale identifier because an extension is duplicated.", message);
}
public void SearchChain_Unicode_Variant_Extension()
{
var id = LocaleIdentifier.Parse("en-u-va-posix");
var chain = id.SearchChain().ToArray();
Assert.AreEqual(3, chain.Length);
Assert.AreEqual("en_POSIX", chain[0]);
Assert.AreEqual("en", chain[1]);
Assert.AreEqual("root", chain[2]);
}
public void Parsing_Extension()
{
var id = LocaleIdentifier.Parse("en-Latn-GB-r-extended-sequence-r-foo");
Assert.AreEqual("en", id.Language);
Assert.AreEqual("Latn", id.Script);
Assert.AreEqual("GB", id.Region);
Assert.IsTrue(id.Extensions.Contains("r-extended-sequence"));
Assert.IsTrue(id.Extensions.Contains("r-foo"));
}
/// <summary>
/// Tries parsing the string representation of a locale identifier.
/// </summary>
/// <param name="s">
/// A case insensitive string containing a locale identifier, based on BCP47.
/// </param>
/// <param name="result">
/// A BCP 47 language identifier that refers to <paramref name="s"/> or <b>null</b> if the parsing
/// failed.
/// </param>
/// <param name="message">
/// The reason why the parsing failed.
/// </param>
/// <returns>
/// <b>true</b> if <paramref name="s"/> was parsed successfully; otherwise, <b>false</b>.
/// </returns>
/// <remarks>
/// The transformation rules for converting a BCP 47 tag into a
/// Unicode Locale ID are <b>not</b> applied.
/// </remarks>
public static bool TryParseBcp47(string s, out LocaleIdentifier result, out string message)
{
result = null;
message = null;
var match = idRegex.Match(s.ToLowerInvariant());
if (!match.Success)
{
message = $"'{s}' is not a valid locale identifier.";
return(false);
}
// Variants cannot be repeated.
var variants = match.Groups["variants"]
.Value.Split('-', '_')
.Where(sv => !string.IsNullOrEmpty(sv))
.ToArray();
if (variants.Distinct().Count() != variants.Length)
{
message = $"'{s}' is not a valid locale identifier because a variant is duplicated.";
return(false);
}
// Extensions cannot be repeated.
var extensions = new List <string>();
foreach (Capture capture in match.Groups["ext"].Captures)
{
extensions.Add(capture.Value.Replace('_', '-'));
}
if (extensions.Distinct().Count() != extensions.Count)
{
message = $"'{s}' is not a valid locale identifier because an extension is duplicated.";
return(false);
}
var u = extensions
.Where(x => x.StartsWith("u-"))
.Select(x => LocaleExtension.Parse(x))
.FirstOrDefault();
result = new LocaleIdentifier
{
Language = match.Groups["lang"].Value,
Script = ToTitleCase(match.Groups["script"].Value),
Region = match.Groups["region"].Value.ToUpperInvariant(),
Variants = variants,
Extensions = extensions.ToArray(),
UnicodeExtension = u ?? LocaleExtension.Empty
};
return(true);
}
public void Parsing_Unicode_Extension()
{
var id = LocaleIdentifier.Parse("th-u-bar-foo-ca-buddhist-kk-nu-thai");
var extension = id.UnicodeExtension;
CollectionAssert.Contains(extension.Attributes.ToArray(), "foo");
CollectionAssert.Contains(extension.Attributes.ToArray(), "bar");
Assert.AreEqual("buddhist", extension.Keywords["ca"]);
Assert.AreEqual("true", extension.Keywords["kk"]);
Assert.AreEqual("thai", extension.Keywords["nu"]);
}
public void Parsing_Variants_Are_Not_Repeated()
{
string message;
LocaleIdentifier id;
var ok = LocaleIdentifier.TryParse("de-CH-1901-1901", out id, out message);
Assert.AreEqual(false, ok);
Assert.AreEqual(null, id);
Assert.AreEqual("'de-CH-1901-1901' is not a valid locale identifier because a variant is duplicated.", message);
ExceptionAssert.Throws <FormatException>(() => LocaleIdentifier.Parse("de-CH-1901-1901"));
}
public void Parsing_Transforms()
{
Assert.AreEqual("en-US", LocaleIdentifier.Parse("en-US").ToString());
Assert.AreEqual("root", LocaleIdentifier.Parse("und").ToString());
Assert.AreEqual("und-US", LocaleIdentifier.Parse("und-US").ToString());
Assert.AreEqual("root-u-cu-usd", LocaleIdentifier.Parse("und-u-cu-USD").ToString());
Assert.AreEqual("zh-TW", LocaleIdentifier.Parse("cmn-TW").ToString());
Assert.AreEqual("sr-RS", LocaleIdentifier.Parse("sr-CS").ToString());
Assert.AreEqual("sr-Latn", LocaleIdentifier.Parse("sh").ToString());
Assert.AreEqual("sr-Cyrl", LocaleIdentifier.Parse("sh-Cyrl").ToString());
Assert.AreEqual("hy-AM", LocaleIdentifier.Parse("hy-SU").ToString());
}
public void Formatting()
{
var id = LocaleIdentifier.Parse("EN_LATN_GB_R_EXTENDED_SEQUENCE_R-FOO");
Assert.AreEqual("en-Latn-GB-r-extended-sequence-r-foo", id.ToString());
id = LocaleIdentifier.Parse("EN_nz");
Assert.AreEqual("en-NZ", id.ToString());
id = LocaleIdentifier.Parse("EN");
Assert.AreEqual("en", id.ToString());
}
public void Empty_Tags()
{
var id = LocaleIdentifier.Parse("Cyrl");
Assert.AreEqual("", id.Language);
Assert.AreEqual("Cyrl", id.Script);
Assert.AreEqual("", id.Region);
Assert.AreEqual(0, id.Extensions.Count());
Assert.AreEqual(0, id.Variants.Count());
id = LocaleIdentifier.Parse("en");
Assert.AreEqual("en", id.Language);
Assert.AreEqual("", id.Script);
Assert.AreEqual("", id.Region);
Assert.AreEqual(0, id.Extensions.Count());
Assert.AreEqual(0, id.Variants.Count());
}
static IEnumerable <string> SearchChainRecursive(LocaleIdentifier id)
{
foreach (var locale in id.SearchChain())
{
yield return(locale);
if (ParentLocales.Value.ContainsKey(locale))
{
var parent = LocaleIdentifier.Parse(ParentLocales.Value[locale]);
foreach (var p in SearchChainRecursive(parent))
{
yield return(p);
}
yield break;
}
}
}
/// <summary>
/// A new locale idenyifer with empty subtags that <see cref="MostLikelySubtags"/> would fill.
/// </summary>
/// <returns>
/// A new locale identifier.
/// </returns>
public LocaleIdentifier RemoveMostLikelySubtags()
{
var max = this.MostLikelySubtags();
max.Variants = new string[0];
var trials = new[] { max.Language, $"{max.Language}-{max.Region}", $"{max.Language}-{max.Script}" };
foreach (var trial in trials)
{
var id = LocaleIdentifier.ParseBcp47(trial);
if (max.ToUnicodeLanguage() == id.MostLikelySubtags().ToUnicodeLanguage())
{
id.Variants = this.Variants;
return(id);
}
}
max.Variants = this.Variants;
return(max);
}
public void UnicodeLangugeTag()
{
var id = LocaleIdentifier.Parse("EN_LATN_GB_R_EXTENDED_SEQUENCE_R-FOO");
Assert.AreEqual("en_Latn_GB", id.ToUnicodeLanguage());
id = LocaleIdentifier.Parse("EN_nz");
Assert.AreEqual("en_NZ", id.ToUnicodeLanguage());
id = LocaleIdentifier.Parse("EN");
Assert.AreEqual("en", id.ToUnicodeLanguage());
id = LocaleIdentifier.Parse("DE-ch-1996");
Assert.AreEqual("de_CH_1996", id.ToUnicodeLanguage());
id = LocaleIdentifier.Parse("sl-nedis");
Assert.AreEqual("sl_nedis", id.ToUnicodeLanguage());
id = LocaleIdentifier.Parse("sl-nedis");
Assert.AreEqual("sl_NEDIS", id.ToUnicodeLanguage(true));
}
public void SearchChain()
{
var id = LocaleIdentifier.Parse("sl-SI-nedis");
var chain = id.SearchChain().ToArray();
Assert.AreEqual(5, chain.Length);
Assert.AreEqual("sl_SI_NEDIS", chain[0]);
Assert.AreEqual("sl_NEDIS", chain[1]);
Assert.AreEqual("sl_SI", chain[2]);
Assert.AreEqual("sl", chain[3]);
Assert.AreEqual("root", chain[4]);
id = LocaleIdentifier.Parse("en-Latn-US-basiceng");
chain = id.SearchChain().ToArray();
Assert.AreEqual("en_Latn_US_BASICENG", chain[0]);
Assert.AreEqual("en_US_BASICENG", chain[1]);
Assert.AreEqual("en_Latn_BASICENG", chain[2]);
Assert.AreEqual("en_BASICENG", chain[3]);
Assert.AreEqual("en_Latn_US", chain[4]);
Assert.AreEqual("en_US", chain[5]);
Assert.AreEqual("en_Latn", chain[6]);
Assert.AreEqual("en", chain[7]);
Assert.AreEqual("root", chain[8]);
}