///<summary>
/// Constructor.
///</summary>
///<param name="idsInFile"></param>
///<param name="replaceIdsInFile"></param>
///<param name="writingSystemRepository"></param>
public static void FindOrphans(
IEnumerable<string> idsInFile,
IdReplacementStrategy replaceIdsInFile,
IWritingSystemRepository writingSystemRepository
) {
var originalIds = new List<string>(idsInFile);
var updatedIds = new List<string>(idsInFile);
foreach (var wsId in originalIds)
{
// Check if it's in the repo
if (writingSystemRepository.Contains(wsId))
{
continue;
}
string newId = wsId;
if (writingSystemRepository.WritingSystemIdHasChanged(wsId))
{
newId = writingSystemRepository.WritingSystemIdHasChangedTo(wsId);
}
else
{
// It's an orphan
// Check for the writing system repository compatibility mode
if (writingSystemRepository.CompatibilityMode == WritingSystemCompatibility.Flex7V0Compatible)
{
if (!wsId.StartsWith("x-"))
{
// Clean it
var rfcTagCleaner = new Rfc5646TagCleaner(wsId);
rfcTagCleaner.Clean();
newId = rfcTagCleaner.GetCompleteTag();
}
}
else
{
// Clean it
var rfcTagCleaner = new Rfc5646TagCleaner(wsId);
rfcTagCleaner.Clean();
newId = rfcTagCleaner.GetCompleteTag();
}
}
var conformantWritingSystem = WritingSystemDefinition.Parse(newId);
// If it changed, then change
if (conformantWritingSystem.Bcp47Tag != wsId)
{
conformantWritingSystem = WritingSystemDefinition.CreateCopyWithUniqueId(conformantWritingSystem, updatedIds);
replaceIdsInFile(wsId, conformantWritingSystem.Bcp47Tag);
updatedIds.Remove(wsId);
updatedIds.Add(conformantWritingSystem.Bcp47Tag);
}
// Check if it's in the repo
if (writingSystemRepository.Contains(conformantWritingSystem.Bcp47Tag))
{
continue;
}
// It's not in the repo so set it
writingSystemRepository.Set(conformantWritingSystem);
}
writingSystemRepository.Save();
}
public void CompleteTagConstructor_QaaWithXDashBeforeValidLanguageName_NoChange()
{
var cleaner = new Rfc5646TagCleaner("qaa-x-th");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-x-th"));
Assert.That(cleaner.Language, Is.EqualTo("qaa"));
Assert.That(cleaner.PrivateUse, Is.EqualTo("th"));
}
public void CompleteTagConstructor_InvalidLanguageNameWithScript_QaaAdded()
{
var cleaner = new Rfc5646TagCleaner("wee-Latn");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-Latn-x-wee"));
// Also when initially "Latn" is properly in the Script field.
cleaner = new Rfc5646TagCleaner("wee", "Latn", "", "", "");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-Latn-x-wee"));
}
public void AllPrivateComponents_InsertsStandardPrivateCodes()
{
var cleaner = new Rfc5646TagCleaner("x-kal", "x-script", "x-RG", "fonipa-x-etic", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "qaa", "Qaaa", "QM", "fonipa", "qaa-Qaaa-QM-fonipa-x-kal-script-RG-etic");
}
public void ValidLanguageCodeMarkedPrivate_InsertsQaa()
{
var cleaner = new Rfc5646TagCleaner("x-de", "", "", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "qaa", "", "", "", "qaa-x-de");
}
public void ScriptEndingWithX_IsHandledCorrectly()
{
var cleaner = new Rfc5646TagCleaner("zh-Phnx-CN-fonipa-x-emic");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "Phnx", "CN", "fonipa", "zh-Phnx-CN-fonipa-x-emic");
}
public void RegionCodesThatMatchLanguageCodesNotMovedToPrivateUse()
{
var cleaner = new Rfc5646TagCleaner("rwr-IN");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "rwr", "", "IN", "", "rwr-IN");
}
public void NewTagWithPrivateLanguage_IsNotModified()
{
var cleaner = new Rfc5646TagCleaner("qaa-Qaaa-QM-x-kal-Mysc-YY");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "qaa", "Qaaa", "QM", "", "qaa-Qaaa-QM-x-kal-Mysc-YY");
}
public void CmnNoRegion_BecomesZhCN()
{
var cleaner = new Rfc5646TagCleaner("cmn", "", "", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "", "CN", "", "zh-CN");
}
public void ZhNoRegion_InsertsRegionCN()
{
var cleaner = new Rfc5646TagCleaner("zh", "", "", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "", "CN", "", "zh-CN");
}
public void MultiPartVariantWithoutX_InsertsX()
{
var cleaner = new Rfc5646TagCleaner("fr", "", "", "fonipa-etic", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fr", "", "", "fonipa", "fr-fonipa-x-etic");
}
public void PrivateRegionMultiPartVariant_InsertsPrivateRegionCode()
{
var cleaner = new Rfc5646TagCleaner("fr", "", "x-ZY", "fonipa-x-etic", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fr", "", "QM", "fonipa", "fr-QM-fonipa-x-ZY-etic");
}
public void PrivateRegionKnownLanguageAndScript_InsertsPrivateRegionCode()
{
var cleaner = new Rfc5646TagCleaner("fr", "Latn", "x-ZY", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fr", "Latn", "QM", "", "fr-Latn-QM-x-ZY");
}
public void CompleteTagConstructor_HasInvalidLanguageName_MovedToPrivateUse()
{
var cleaner = new Rfc5646TagCleaner("234");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-x-234"));
}
public void PrivateScriptKnownLanguageAndRegion_InsertsPrivateScriptCode()
{
var cleaner = new Rfc5646TagCleaner("fr", "x-script", "NO", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fr", "Qaaa", "NO", "", "fr-Qaaa-NO-x-script");
}
public void PrivateUseVariantLanguageCode_IsNotShortened()
{
var cleaner = new Rfc5646TagCleaner("qaa", "", "", "", "x-kal");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "qaa", "", "", "", "qaa-x-kal");
}
public void LanguageCodeAfterX_IsNotShortened()
{
var cleaner = new Rfc5646TagCleaner("qaa-x-kal");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "qaa", "", "", "", "qaa-x-kal");
}
public void Pes_BecomesFa()
{
var cleaner = new Rfc5646TagCleaner("pes", "Latn", "", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fa", "Latn", "", "", "fa-Latn");
}
public void NewTag_IsNotModified()
{
var cleaner = new Rfc5646TagCleaner("fr-Qaaa-QM-x-Mysc-YY");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "fr", "Qaaa", "QM", "", "fr-Qaaa-QM-x-Mysc-YY");
}
public void Arb_BecomesAr()
{
var cleaner = new Rfc5646TagCleaner("arb", "", "x-ZG", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "ar", "", "QM", "", "ar-QM-x-ZG");
}
public void CleanMarksCustomScriptMovedToPrivateUse()
{
var cleaner = new Rfc5646TagCleaner("en-Zyxw");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "en", "Qaaa", "", "", "en-Qaaa-x-Zyxw");
}
public void ZhRegion_NoChange()
{
var cleaner = new Rfc5646TagCleaner("zh", "", "NO", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "", "NO", "", "zh-NO");
cleaner = new Rfc5646TagCleaner("zh", "", "x-ZK", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "", "QM", "", "zh-QM-x-ZK");
}
public void CompleteTagConstructor_LanguageNameWithAudio_GetZxxxAdded()
{
var cleaner = new Rfc5646TagCleaner("aaa-x-audio");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("aaa-Zxxx-x-audio"));
}
public void LanguageSubtagContainsMultipleValidLanguageSubtagsAsWellAsDataThatIsNotValidLanguageScriptRegionOrVariant_AllSubtagsButFirstValidLanguageSubtagAreMovedToPrivateUse()
{
var cleaner = new Rfc5646TagCleaner("bogus-en-audio-tpi-bogus2-x-", "", "", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "en", "Zxxx", "", "", "en-Zxxx-x-bogus-audio-bogus2-tpi");
}
public void CompleteTagConstructor_XDashBeforeValidLanguageNameInVariant_NoChange()
{
var cleaner = new Rfc5646TagCleaner("", "", "", "x-de", "");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("x-de"));
}
public void CompleteTagConstructor_HasLanguageNameAndOtherName_OtherNameMovedToPrivateUse()
{
var cleaner = new Rfc5646TagCleaner("abc-123");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("abc-x-123"));
}
public void Language_XDashBeforeString_AddsQaa()
{
var cleaner = new Rfc5646TagCleaner("x-blah");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-x-blah"));
}
public void CmnRegion_BecomesZh()
{
var cleaner = new Rfc5646TagCleaner("cmn", "", "NO", "", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "zh", "", "NO", "", "zh-NO");
}
public void CompleteTagConstructor_TagContainsOnlyPrivateUseWithAdditionalXDash_RedundantXDashRemoved()
{
var cleaner = new Rfc5646TagCleaner("x-some-x-whatever");
cleaner.Clean();
Assert.That(cleaner.GetCompleteTag(), Is.EqualTo("qaa-x-some-whatever"));
}
public void EmicWithoutFonipa_AddsFonipa()
{
var cleaner = new Rfc5646TagCleaner("en", "", "", "x-emic", "");
cleaner.Clean();
VerifyRfcCleaner(cleaner, "en", "", "", "fonipa", "en-fonipa-x-emic");
}
