public RFC5646Tag(string language, string script, string region, string variant, string privateUse)
{
_language = language ?? "";
_script = script ?? "";
_region = region ?? "";
_variant.AddToSubtag(variant ?? "");
_privateUse.AddToSubtag(privateUse ?? "");
Validate();
}
public void KeepFirstAndMoveRemainderTo(SubTag tagToMoveTo)
{
for (int i = _subTagParts.Count - 1; i > 0; i--)
{
tagToMoveTo.AddToSubtag(_subTagParts[i]);
_subTagParts.RemoveAt(i);
}
}
private static void MoveTagsMatching(SubTag from, SubTag to, Predicate <string> moveAllMatching)
{
var list = new List <string>(from.AllParts.Where(part => moveAllMatching(part)));
foreach (var part in list)
{
to.AddToSubtag(part);
from.RemoveParts(part);
}
}
/// <summary>
/// This method should move all subtag parts in the 'from' subtag which match the moveAllMatching predicate into the 'to' subtag.
/// Because some parts of a subtag may match in more than one language tag area care must be taken to prevent emptying all parts of
/// one subtag into another so the first part that matches the keepFirstMatching predicate will not be moved.
/// i.e. if the languageTag is 'from' and the regionTag is 'to' and keepFirstMatching matches language codes and moveAllMatching
/// matches region codes, all region looking parts would be placed in 'to' with the possible exception of the first language looking
/// part.
/// </summary>
/// <param name="from">SubTag to move parts from</param>
/// <param name="to">SubTag to move matching parts to</param>
/// <param name="moveAllMatching">predicate matching parts to move</param>
/// <param name="keepFirstMatching">predicate matching part to keep</param>
private static void MoveTagsMatching(SubTag from, SubTag to, Predicate <string> moveAllMatching, Predicate <string> keepFirstMatching)
{
bool haveFirstMatching = false;
var allParts = new List <string>(from.AllParts);
foreach (var part in allParts)
{
if (!haveFirstMatching && keepFirstMatching(part))
{
haveFirstMatching = true;
continue;
}
if (!moveAllMatching(part))
{
continue;
}
to.AddToSubtag(part);
from.RemoveParts(part);
}
}
/// <summary>
/// Cleans the tag.
/// </summary>
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))
{
LanguageSubtag language;
if (StandardSubtags.TryGetLanguageFromIso3Code(part, out language) && language.Code != language.Iso3Code)
{
migrateFrom = part;
migrateTo = language.Code;
}
}
}
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, StandardSubtags.IsValidIso639LanguageCode, "qaa", true);
MoveFirstPartToPrivateUseIfNecessary(_scriptSubTag, StandardSubtags.IsValidIso15924ScriptCode, "Qaaa", false);
MoveFirstPartToPrivateUseIfNecessary(_regionSubTag, StandardSubtags.IsValidIso3166RegionCode, "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.AudioScript))
{
MoveTagsMatching(_variantSubTag, _scriptSubTag, tag => tag.Equals(WellKnownSubtags.AudioScript));
_privateUseSubTag.AddToSubtag(WellKnownSubtags.AudioPrivateUse);
}
// 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, StandardSubtags.IsValidIso15924ScriptCode, StandardSubtags.IsValidIso639LanguageCode);
MoveTagsMatching(_languageSubTag, _regionSubTag, StandardSubtags.IsValidIso3166RegionCode, StandardSubtags.IsValidIso639LanguageCode);
MoveTagsMatching(_languageSubTag, _variantSubTag, StandardSubtags.IsValidRegisteredVariantCode, StandardSubtags.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 => !StandardSubtags.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.IpaPhonemicPrivateUse.EndsWith(part) &&
!WellKnownSubtags.IpaPhoneticPrivateUse.EndsWith(part))
{
_scriptSubTag = new SubTag("Qaaa");
}
}
MoveTagsMatching(_scriptSubTag, _privateUseSubTag, tag => !StandardSubtags.IsValidIso15924ScriptCode(tag));
MoveTagsMatching(_regionSubTag, _privateUseSubTag, tag => !StandardSubtags.IsValidIso3166RegionCode(tag));
MoveTagsMatching(_variantSubTag, _privateUseSubTag, tag => !StandardSubtags.IsValidRegisteredVariantCode(tag));
_languageSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
_scriptSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
_regionSubTag.KeepFirstAndMoveRemainderTo(_privateUseSubTag);
if (_privateUseSubTag.Contains(WellKnownSubtags.AudioPrivateUse))
{
// Move every tag that's not a Zxxx to private use
if (!_scriptSubTag.IsEmpty && !_scriptSubTag.Contains(WellKnownSubtags.AudioScript))
{
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.AudioScript))
{
_scriptSubTag = new SubTag(WellKnownSubtags.AudioScript);
}
}
//These two methods may produce duplicates that will subsequently be removed. Do we care? - TA 29/3/2011
_privateUseSubTag.RemoveNonAlphaNumericCharacters();
_privateUseSubTag.TruncatePartsToNumCharacters(8);
_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 (string part in _privateUseSubTag.AllParts)
{
if (string.IsNullOrEmpty(Variant) &&
(part.Equals("etic", StringComparison.OrdinalIgnoreCase) || part.Equals("emic", StringComparison.OrdinalIgnoreCase)))
{
Variant = "fonipa";
}
}
}
/// <summary>
/// This method should move all subtag parts in the 'from' subtag which match the moveAllMatching predicate into the 'to' subtag.
/// Because some parts of a subtag may match in more than one language tag area care must be taken to prevent emptying all parts of
/// one subtag into another so the first part that matches the keepFirstMatching predicate will not be moved.
/// i.e. if the languageTag is 'from' and the regionTag is 'to' and keepFirstMatching matches language codes and moveAllMatching
/// matches region codes, all region looking parts would be placed in 'to' with the possible exception of the first language looking
/// part.
/// </summary>
/// <param name="from">SubTag to move parts from</param>
/// <param name="to">SubTag to move matching parts to</param>
/// <param name="moveAllMatching">predicate matching parts to move</param>
/// <param name="keepFirstMatching">predicate matching part to keep</param>
private static void MoveTagsMatching(SubTag from, SubTag to, Predicate<string> moveAllMatching, Predicate<string> keepFirstMatching)
{
bool haveFirstMatching = false;
var allParts = new List<string>(from.AllParts);
foreach (var part in allParts)
{
if (!haveFirstMatching && keepFirstMatching(part))
{
haveFirstMatching = true;
continue;
}
if (!moveAllMatching(part))
continue;
to.AddToSubtag(part);
from.RemoveParts(part);
}
}
private static void MoveTagsMatching(SubTag from, SubTag to, Predicate<string> moveAllMatching)
{
var list = new List<string>(from.AllParts.Where(part => moveAllMatching(part)));
foreach (var part in list)
{
to.AddToSubtag(part);
from.RemoveParts(part);
}
}
public void KeepFirstAndMoveRemainderTo(SubTag tagToMoveTo)
{
for (int i = _subTagParts.Count - 1; i > 0; i--)
{
tagToMoveTo.AddToSubtag(_subTagParts[i]);
_subTagParts.RemoveAt(i);
}
}
public void Migrate(string sourceFilePath, string destinationFilePath)
{
var migrationInfo = new List <LdmlVersion0MigrationStrategy.MigrationInfo>();
if (!Directory.Exists(destinationFilePath))
{
Directory.CreateDirectory(destinationFilePath);
}
var wesayWsCollection = new WritingSystemCollection_V1();
wesayWsCollection.LoadFromLegacyWeSayFile(sourceFilePath);
foreach (var writingSystem in wesayWsCollection.Values)
{
var currentMigrationInfo = new LdmlVersion0MigrationStrategy.MigrationInfo {
RfcTagBeforeMigration = writingSystem.ISO
};
var wsDef = new WritingSystemDefinitionV0();
if (writingSystem.IsAudio)
{
wsDef.Script = WellKnownSubTags.Audio.Script;
wsDef.Variant = WellKnownSubTags.Audio.PrivateUseSubtag;
}
else
{
var subtag = new SubTag();
subtag.AddToSubtag(writingSystem.ISO);
subtag.RemoveAllParts("audio");
writingSystem.ISO = subtag.CompleteTag;
}
wsDef.ISO639 = writingSystem.ISO;
wsDef.DefaultFontName = writingSystem.FontName;
wsDef.DefaultFontSize = writingSystem.FontSize;
wsDef.Abbreviation = writingSystem.Abbreviation;
wsDef.IsLegacyEncoded = !writingSystem.IsUnicode;
wsDef.Keyboard = writingSystem.KeyboardName;
wsDef.RightToLeftScript = writingSystem.RightToLeft;
wsDef.SpellCheckingId = writingSystem.SpellCheckingId;
wsDef.DateModified = DateTime.Now;
if (writingSystem.SortUsing.Equals("CustomICU"))
{
wsDef.SortRules = writingSystem.CustomSortRules;
wsDef.SortUsing = WritingSystemDefinitionV0.SortRulesType.CustomICU;
}
else if (writingSystem.SortUsing.Equals("CustomSimple"))
{
wsDef.SortRules = writingSystem.CustomSortRules;
wsDef.SortUsing = WritingSystemDefinitionV0.SortRulesType.CustomSimple;
}
else if (!String.IsNullOrEmpty(writingSystem.SortUsing))
{
//when sorting like other language
wsDef.SortRules = writingSystem.SortUsing;
wsDef.SortUsing = WritingSystemDefinitionV0.SortRulesType.OtherLanguage;
}
//wsDef.VerboseDescription //not written out by ldmladaptor - flex?
//wsDef.StoreID = ??? //what to do?
//wsDef.NativeName //not written out by ldmladaptor - flex?
string pathForNewLdmlFile = Path.Combine(destinationFilePath, wsDef.Rfc5646 + ".ldml");
new LdmlAdaptorV0().Write(pathForNewLdmlFile, wsDef, null);
currentMigrationInfo.RfcTagAfterMigration = wsDef.Rfc5646;
if (currentMigrationInfo.RfcTagBeforeMigration != currentMigrationInfo.RfcTagAfterMigration)
{
_changeLog.LogChange(currentMigrationInfo.RfcTagBeforeMigration, currentMigrationInfo.RfcTagAfterMigration);
}
migrationInfo.Add(currentMigrationInfo);
}
_onMigrationCallback(migrationInfo);
}