public void SubTagComponentDescription_HasBeginningParenthesis_RemovesParens()
{
Assert.AreEqual(
"Hiragana + Katakana",
StandardTags.SubTagComponentDescription("Description: (Hiragana + Katakana)")
);
}
public void SubTagComponentDescription_HasForwardSlash_ReplacesWithPipe()
{
Assert.AreEqual(
"=|Kx'au||'ein",
StandardTags.SubTagComponentDescription("Description: =/Kx'au//'ein")
);
}
public void SubTagComponentDescription_HasAliasFor_RemovesAliasFor()
{
Assert.AreEqual(
"Japanese (Han + Hiragana + Katakana)",
StandardTags.SubTagComponentDescription("Description: Japanese (alias for Han + Hiragana + Katakana)")
);
}
public static string ValidVariantString(string unknownString)
{
// var1-var2-var3
// var1-privateUse1-x-privateUse2
unknownString = unknownString.Trim();
unknownString = Regex.Replace(unknownString, @"[ ,.]", "-");
unknownString = Regex.Replace(unknownString, @"-+", "-");
var tokens = unknownString.Split('-');
var variants = new List <string>();
var privateUse = new List <string>();
bool haveSeenX = false;
foreach (var token in tokens)
{
if (token == "x")
{
haveSeenX = true;
continue;
}
if (!haveSeenX && StandardTags.IsValidRegisteredVariant(token))
{
variants.Add(token);
}
else
{
privateUse.Add(token);
}
}
string combinedVariant = String.Join("-", variants.ToArray());
string combinedPrivateUse = String.Join("-", privateUse.ToArray());
string variantString = combinedVariant;
if (!String.IsNullOrEmpty(combinedPrivateUse))
{
variantString = "x-" + combinedPrivateUse;
if (!String.IsNullOrEmpty(combinedVariant))
{
variantString = combinedVariant + "-" + variantString;
}
}
return(variantString);
}
public void IsValidIso3166Region_QM_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsValidIso3166Region("QM"));
Assert.IsTrue(StandardTags.IsValidIso3166Region("qm"));
}
public void IsValidIso639LanguageCode_two_ReturnTrue()
{
// Yes it's true
Assert.IsTrue(StandardTags.IsValidIso639LanguageCode("two"));
}
public void IsValidIso639LanguageCode_fonipa_ReturnFalse()
{
Assert.IsFalse(StandardTags.IsValidIso639LanguageCode("fonipa"));
}
public void IsValidRegisteredVariant_en_ReturnsFalse()
{
Assert.IsFalse(StandardTags.IsValidRegisteredVariant("en"));
}
public void IsValidIso639LanguageCode_qaa_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsValidIso639LanguageCode("qaa"));
}
public void IsValidIso3166Region_fonipa_ReturnsFalse()
{
Assert.IsFalse(StandardTags.IsValidIso3166Region("fonipa"));
}
public void IsValidRegisteredVariant_fonipa_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsValidRegisteredVariant("fonipa"));
}
public void Clean()
{
// Migrate legacy ISO3 language codes to IANA 2 letter language codes, if there's a match.
// Do this before we look for valid codes, otherwise the 3-letter ones come up as invalid and
// get moved to private use. However, only do this to languages not identified as private-use.
if (!Language.StartsWith("x-", StringComparison.OrdinalIgnoreCase))
{
string migrateFrom = "";
string migrateTo = "";
foreach (string part in _languageSubTag.AllParts)
{
if (part.Equals("x", StringComparison.OrdinalIgnoreCase))
{
break; // don't migrate language code parts already explicitly marked private-use.
}
if (String.IsNullOrEmpty(migrateFrom))
{
foreach (var code in StandardTags.ValidIso639LanguageCodes)
{
if (code.ISO3Code.Equals(part))
{
migrateFrom = part;
migrateTo = code.Code;
break;
}
}
}
}
if (!String.IsNullOrEmpty(migrateFrom))
{
_languageSubTag.RemoveParts(migrateFrom);
_languageSubTag.AddToSubtag(migrateTo);
}
}
// The very next thing, before anything else gets moved to private use, is to move the parts whose position we
// care about to the appropriate position in the private use section.
// In the process we may remove anything non-alphanumeric, since otherwise we may move a marker that later
// disappears (pathologically).
MoveFirstPartToPrivateUseIfNecessary(_languageSubTag, StandardTags.IsValidIso639LanguageCode, "qaa", true);
MoveFirstPartToPrivateUseIfNecessary(_scriptSubTag, StandardTags.IsValidIso15924ScriptCode, "Qaaa", false);
MoveFirstPartToPrivateUseIfNecessary(_regionSubTag, StandardTags.IsValidIso3166Region, "QM", false);
//This fixes a bug where the LdmlAdaptorV1 was writing out Zxxx as part of the variant to mark an audio writing system
if (_variantSubTag.Contains(WellKnownSubTags.Audio.Script))
{
MoveTagsMatching(_variantSubTag, _scriptSubTag, tag => tag.Equals(WellKnownSubTags.Audio.Script));
_privateUseSubTag.AddToSubtag(WellKnownSubTags.Audio.PrivateUseSubtag);
}
// Fixes various legacy problems.
if (Language.Equals("cmn", StringComparison.OrdinalIgnoreCase))
{
Language = "zh";
}
if (Language.Equals("pes", StringComparison.OrdinalIgnoreCase))
{
Language = "fa";
}
if (Language.Equals("arb", StringComparison.OrdinalIgnoreCase))
{
Language = "ar";
}
if (Language.Equals("zh", StringComparison.OrdinalIgnoreCase) && String.IsNullOrEmpty(Region))
{
Region = "CN";
}
// If the language tag contains an x- , then move the string behind the x- to private use
MovePartsToPrivateUseIfNecessary(_languageSubTag);
// Move script, region, and variant present in the langauge tag to their proper subtag.
MoveTagsMatching(_languageSubTag, _scriptSubTag, StandardTags.IsValidIso15924ScriptCode, StandardTags.IsValidIso639LanguageCode);
MoveTagsMatching(_languageSubTag, _regionSubTag, StandardTags.IsValidIso3166Region, StandardTags.IsValidIso639LanguageCode);
MoveTagsMatching(_languageSubTag, _variantSubTag, StandardTags.IsValidRegisteredVariant, StandardTags.IsValidIso639LanguageCode);
// Move all other tags that don't belong to the private use subtag.
//keep track of everything that we moved
var tempSubTag = new SubTag();
MoveTagsMatching(
_languageSubTag, tempSubTag, tag => !StandardTags.IsValidIso639LanguageCode(tag)
);
//place all the moved parts in private use.
foreach (var part in tempSubTag.AllParts)
{
_privateUseSubTag.AddToSubtag(part);
//if it looks like we moved a custom script set the subtag to mark that we've moved it
if (_scriptSubTag.IsEmpty &&
part.Length == 4 && //potential custom script tag
!WellKnownSubTags.Ipa.PhonemicPrivateUseSubtag.EndsWith(part) &&
!WellKnownSubTags.Ipa.PhoneticPrivateUseSubtag.EndsWith(part))
{
_scriptSubTag = new SubTag("Qaaa");
}
}
MoveTagsMatching(
_scriptSubTag, _privateUseSubTag, tag => !StandardTags.IsValidIso15924ScriptCode(tag)
);
MoveTagsMatching(
_regionSubTag, _privateUseSubTag, tag => !StandardTags.IsValidIso3166Region(tag)
);
MoveTagsMatching(
_variantSubTag, _privateUseSubTag, tag => !StandardTags.IsValidRegisteredVariant(tag)
);
_languageSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
_scriptSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
_regionSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
if (_privateUseSubTag.Contains(WellKnownSubTags.Audio.PrivateUseSubtag))
{
// Move every tag that's not a Zxxx to private use
if (!_scriptSubTag.IsEmpty && !_scriptSubTag.Contains(WellKnownSubTags.Audio.Script))
{
MoveTagsMatching(_scriptSubTag, _privateUseSubTag, tag => !_privateUseSubTag.Contains(tag));
}
// If we don't have a Zxxx already, set it. This protects tags already present, but with unusual case
if (!_scriptSubTag.Contains(WellKnownSubTags.Audio.Script))
{
_scriptSubTag = new SubTag(WellKnownSubTags.Audio.Script);
}
}
//These two methods may produce duplicates that will subsequently be removed. Do we care? - TA 29/3/2011
_privateUseSubTag.TruncatePartsToNumCharacters(8);
_privateUseSubTag.RemoveNonAlphaNumericCharacters();
_variantSubTag.RemoveDuplicates();
_privateUseSubTag.RemoveDuplicates();
// Any 'x' in the other tags will have arrived in the privateUse tag, so remove them.
_privateUseSubTag.RemoveParts("x");
// if language is empty, we need to add qaa, unless only a privateUse is present (e.g. x-blah is a valid rfc5646 tag)
if ((_languageSubTag.IsEmpty && (!_scriptSubTag.IsEmpty || !_regionSubTag.IsEmpty || !_variantSubTag.IsEmpty)) ||
(_languageSubTag.IsEmpty && _scriptSubTag.IsEmpty && _regionSubTag.IsEmpty && _variantSubTag.IsEmpty && _privateUseSubTag.IsEmpty))
{
_languageSubTag.AddToSubtag("qaa");
}
// Two more legacy problems. We don't allow -etic or -emic without fonipa, so insert if needed.
// If it has some other standard variant we won't be able to fix it...not sure what the right answer would be.
// At least we catch the more common case.
foreach (var part in _privateUseSubTag.AllParts)
{
if (string.IsNullOrEmpty(Variant) &&
(part.Equals("etic", StringComparison.OrdinalIgnoreCase) || part.Equals("emic", StringComparison.OrdinalIgnoreCase)))
{
Variant = "fonipa";
}
}
}
public void IsStandardIso3166Region_QM_ReturnsFalse()
{
Assert.IsFalse(StandardTags.IsStandardIso3166Region("QM"));
}
public void IsStandardIso3166Region_US_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsStandardIso3166Region("US"));
}
public void IsValidIso15924ScriptCode_fonipa_ReturnsFalse()
{
Assert.IsFalse(StandardTags.IsValidIso15924ScriptCode("fonipa"));
}
public void IsStandardIso15924ScriptCode_Latn_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsStandardIso15924ScriptCode("Latn"));
}
public void IsStandardIso15924ScriptCode_Qaaa_ReturnsFalse()
{
Assert.IsFalse(StandardTags.IsStandardIso15924ScriptCode("Qaaa"));
}
public void IsValidIso15924ScriptCode_Qaaa_ReturnsTrue()
{
Assert.IsTrue(StandardTags.IsValidIso15924ScriptCode("Qaaa"));
}
