/// <summary>
/// Gets the subtags of the specified language tag.
/// </summary>
/// <param name="langTag">The language tag.</param>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public static bool GetSubtags(string langTag, out LanguageSubtag languageSubtag, out ScriptSubtag scriptSubtag,
out RegionSubtag regionSubtag, out VariantSubtag variantSubtag)
{
if (string.IsNullOrEmpty(langTag))
throw new ArgumentNullException("langTag");
languageSubtag = null;
scriptSubtag = null;
regionSubtag = null;
variantSubtag = null;
Match match = s_langTagPattern.Match(langTag);
if (!match.Success)
return false;
var parts = new[] { "language", "script", "region", "variant" };
Group privateUseGroup = match.Groups["privateuse"];
string[] privateUseSubTags = null;
int privateUseSubTagIndex = 0;
bool privateUsePrefix = false;
string privateUseSubTag = null;
int part = -1;
if (privateUseGroup.Success)
{
for (part = parts.Length - 1; part >= 0; part--)
{
Group group = match.Groups[parts[part]];
if (group.Success && privateUseGroup.Index > group.Index)
break;
}
part++;
privateUseSubTags = privateUseGroup.Value.Split('-');
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
string languageCode = match.Groups["language"].Value;
if (!string.IsNullOrEmpty(languageCode))
{
languageSubtag = GetLanguageSubtag(languageCode);
}
else if (privateUseSubTag != null && part <= 0)
{
languageSubtag = new LanguageSubtag(privateUseSubTag, null, true, null);
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
string scriptCode = match.Groups["script"].Value;
if (!string.IsNullOrEmpty(scriptCode))
{
scriptSubtag = GetScriptSubtag(scriptCode);
}
else if (privateUseSubTag != null && part <= 1 && s_scriptPattern.IsMatch(privateUseSubTag))
{
scriptSubtag = privateUsePrefix ? new ScriptSubtag(privateUseSubTag, null, true) : GetScriptSubtag(privateUseSubTag);
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
string regionCode = match.Groups["region"].Value;
if (!string.IsNullOrEmpty(regionCode))
{
regionSubtag = GetRegionSubtag(regionCode);
}
else if (privateUseSubTag != null && part <= 2 && s_regionPattern.IsMatch(privateUseSubTag))
{
regionSubtag = GetRegionSubtag(privateUseSubTag);
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
var variantSb = new StringBuilder();
bool variantPrivateUsePrefix = false;
string variantCode = match.Groups["variant"].Value;
if (!string.IsNullOrEmpty(variantCode))
{
variantSb.Append(variantCode);
}
// We would like to also check this subtag against the variant pattern
// to ensure that we have a legitimate variant code, but for loading legacy projects
// with poorly-formed codes, we have to do something with the private use subtag,
// so if it doesn't match any of the others we force it to be a variant even if
// it is not properly formed.
else if (privateUseSubTag != null && part <= 3)
{
variantSb.Append(privateUseSubTag);
variantPrivateUsePrefix = privateUsePrefix;
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
while (privateUseSubTag != null)
{
variantSb.Append("-");
if (privateUsePrefix)
variantSb.Append("x-");
variantSb.Append(privateUseSubTag);
privateUseSubTag = NextSubTag(privateUseSubTags, ref privateUseSubTagIndex, out privateUsePrefix);
}
variantCode = variantSb.ToString();
if (!string.IsNullOrEmpty(variantCode))
{
variantSubtag = variantPrivateUsePrefix ? new VariantSubtag(variantCode, null, true, null)
: GetVariantSubtag(variantCode);
}
return true;
}
/// <summary>
/// Generates a language tag from the specified subtags.
/// </summary>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public static string ToLangTag(LanguageSubtag languageSubtag, ScriptSubtag scriptSubtag, RegionSubtag regionSubtag, VariantSubtag variantSubtag)
{
if (languageSubtag == null)
throw new ArgumentNullException("languageSubtag");
bool inPrivateUse = false;
var sb = new StringBuilder();
if (languageSubtag.IsPrivateUse)
{
sb.Append("x-");
inPrivateUse = true;
}
sb.Append(languageSubtag.Code);
if (scriptSubtag != null)
{
sb.Append("-");
if (!inPrivateUse && scriptSubtag.IsPrivateUse)
{
sb.Append("x-");
inPrivateUse = true;
}
sb.Append(scriptSubtag.Code);
}
if (regionSubtag != null)
{
sb.Append("-");
if (!inPrivateUse && regionSubtag.IsPrivateUse && !IsPrivateUseRegionCode(regionSubtag.Code))
{
sb.Append("x-");
inPrivateUse = true;
}
sb.Append(regionSubtag.Code);
}
else if (languageSubtag.Code == "zh" && languageSubtag.ISO3Code == "cmn")
{
sb.Append("-CN");
}
if (variantSubtag != null)
{
sb.Append("-");
if (!inPrivateUse && variantSubtag.IsPrivateUse)
sb.Append("x-");
sb.Append(variantSubtag.Code);
}
return sb.ToString();
}
/// <summary></summary>
public IWritingSystem Create(LanguageSubtag languageSubtag, ScriptSubtag scriptSubtag, RegionSubtag regionSubtag, VariantSubtag variantSubtag)
{
throw new NotImplementedException();
}
/// <summary>
/// Creates a new writing system.
/// </summary>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public IWritingSystem Create(LanguageSubtag languageSubtag, ScriptSubtag scriptSubtag, RegionSubtag regionSubtag, VariantSubtag variantSubtag)
{
if (!languageSubtag.IsValid)
throw new ArgumentException("Can not create a new writing system with an invalid language tag.");
PalasoWritingSystem ws;
lock (m_syncRoot)
ws = (PalasoWritingSystem)m_localStore.CreateNew();
ws.LanguageSubtag = languageSubtag;
ws.ScriptSubtag = scriptSubtag;
ws.RegionSubtag = regionSubtag;
ws.VariantSubtag = variantSubtag;
if (!string.IsNullOrEmpty(languageSubtag.Name))
ws.Abbreviation = languageSubtag.Name.Length > 3 ? languageSubtag.Name.Substring(0, 3) : languageSubtag.Name;
else
ws.Abbreviation = ws.Id;
CultureInfo ci = MiscUtils.GetCultureForWs(ws.Id);
if (ci != null)
{
ws.SortUsing = WritingSystemDefinition.SortRulesType.OtherLanguage;
ws.SortRules = ci.Name;
}
ws.Modified = false;
return ws;
}
/// <summary>
/// Gets the variant subtag with the specified code. If it is not a known (non-private-use) one,
/// make a private-use one with the specified values. Insert 'x' or move it earlier
/// if any leading parts are not standard.
/// </summary>
public static VariantSubtag GetVariantSubtag(string code, string name, IEnumerable<string> prefixes)
{
if (string.IsNullOrEmpty(code))
throw new ArgumentNullException("code");
var parts = code.Split('-').Where(part => part.Length > 0);
var fixedParts = new List<string>();
bool gotX = false;
foreach (var part in parts)
{
if (part.Equals("x", StringComparison.OrdinalIgnoreCase))
{
if (gotX)
continue; // no duplicate x.
fixedParts.Add(part);
gotX = true;
continue;
}
if (gotX)
{
fixedParts.Add(part); // copy the rest unchanged.
continue;
}
if (!StandardTags.IsValidRegisteredVariant(part))
{
fixedParts.Add("x");
gotX = true;
}
fixedParts.Add(part); // copy the rest unchanged.
}
var code2 = fixedParts.Aggregate((partialCode, part) => partialCode + "-" + part);
VariantSubtag subtag;
if (!s_variantSubtags.TryGetValue(code2, out subtag))
subtag = new VariantSubtag(code2, name, true, prefixes);
return subtag;
}
/// <summary>
/// Load the controls from the writing system, if it is not null. If it is null, clear all controls.
/// If the combo boxes are not populated, do nothing...the method will get called again
/// when the form loads.
/// </summary>
private void LoadControlsFromWritingSystem()
{
m_enableLangTagSideEffects = false;
if (m_ws == null)
return; // Probably in design mode; can't populate.
m_origVariantSubtag = m_ws.VariantSubtag;
m_origRegionSubtag = m_ws.RegionSubtag;
m_origScriptSubtag = m_ws.ScriptSubtag;
m_scriptName.ClearItems();
m_scriptName.Items.AddRange(LangTagUtils.ScriptSubtags.ToArray());
ScriptSubtag = m_origScriptSubtag;
m_regionName.ClearItems();
m_regionName.Items.AddRange(LangTagUtils.RegionSubtags.ToArray());
RegionSubtag = m_origRegionSubtag;
PopulateVariantCombo(false);
VariantSubtag = m_origVariantSubtag;
m_enableLangTagSideEffects = true;
}
static LangTagUtils()
{
s_icuTagPattern = new Regex(IcuTagExpr, RegexOptions.ExplicitCapture);
s_langTagPattern = new Regex(LangTagExpr, RegexOptions.ExplicitCapture);
s_langPattern = new Regex("\\A(" + LanguageExpr + ")\\z", RegexOptions.ExplicitCapture);
s_scriptPattern = new Regex("\\A(" + ScriptExpr + ")\\z", RegexOptions.ExplicitCapture);
s_regionPattern = new Regex("\\A(" + RegionExpr + ")\\z", RegexOptions.ExplicitCapture);
s_variantPattern = new Regex("\\A(" + FuzzyVariantExpr + ")\\z", RegexOptions.ExplicitCapture);
s_languageSubtags = new Dictionary<string, LanguageSubtag>();
foreach (var langCode in StandardTags.ValidIso639LanguageCodes)
{
var code = langCode.Code;
switch (code)
{
// ISO3Code is now only set when it differs from Code.
case "cmn":
code = "zh";
langCode.ISO3Code = "cmn";
break;
case "pes":
code = "fa";
langCode.ISO3Code = "pes";
break;
case "arb":
code = "ar";
langCode.ISO3Code = "arb";
break;
case "zlm":
code = "ms";
langCode.ISO3Code = "zlm";
break;
}
var languageSubtag = new LanguageSubtag(code, langCode.Name, false, langCode.ISO3Code);
s_languageSubtags[languageSubtag.Code] = languageSubtag;
if (!string.IsNullOrEmpty(languageSubtag.ISO3Code))
s_languageSubtags[languageSubtag.ISO3Code] = languageSubtag;
}
s_scriptSubtags = new Dictionary<string, ScriptSubtag>();
foreach (var scriptCode in StandardTags.ValidIso15924Scripts)
s_scriptSubtags[scriptCode.Code] = new ScriptSubtag(scriptCode.Code, scriptCode.Label, false);
s_regionSubtags = new Dictionary<string, RegionSubtag>();
foreach (var regionCode in StandardTags.ValidIso3166Regions)
s_regionSubtags[regionCode.Subtag] = new RegionSubtag(regionCode.Subtag, regionCode.Description, false);
s_variantSubtags = new Dictionary<string, VariantSubtag>();
foreach (var variantCode in StandardTags.ValidRegisteredVariants)
s_variantSubtags[variantCode.Subtag] = new VariantSubtag(variantCode.Subtag, variantCode.Description,
false, variantCode.Prefixes);
// These ones are considered non-private in that the user can't edit the code, but they already contain needed X's.
s_variantSubtags["fonipa-x-etic"] = new VariantSubtag("fonipa-x-etic", "Phonetic", false, null);
s_variantSubtags["fonipa-x-emic"] = new VariantSubtag("fonipa-x-emic", "Phonemic", false, null);
s_variantSubtags["x-py"] = new VariantSubtag("x-py", "Pinyin", false, null);
s_variantSubtags["x-pyn"] = new VariantSubtag("x-pyn", "Pinyin Numbered", false, null);
s_variantSubtags["x-audio"] = new VariantSubtag("x-audio", "Audio", false, null);
}
/// <summary>
/// Initializes a new instance of the <see cref="T:VariantSubtag"/> class.
/// </summary>
/// <param name="subtag">The subtag.</param>
/// <param name="name">The name.</param>
public VariantSubtag(VariantSubtag subtag, string name)
: this(subtag.Code, name, subtag.IsPrivateUse, subtag.Prefixes)
{
}
/// <summary>
/// Gets the subtags of the specified language tag.
/// </summary>
/// <param name="langTag">The language tag.</param>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public static bool GetSubtags(string langTag, out LanguageSubtag languageSubtag, out ScriptSubtag scriptSubtag,
out RegionSubtag regionSubtag, out VariantSubtag variantSubtag)
{
if (string.IsNullOrEmpty(langTag))
throw new ArgumentNullException("langTag");
languageSubtag = null;
scriptSubtag = null;
regionSubtag = null;
variantSubtag = null;
if (langTag.Any(c => !Char.IsLetterOrDigit(c) && c != '-'))
{
return false;
}
var cleaner = new Palaso.WritingSystems.Migration.Rfc5646TagCleaner(langTag);
cleaner.Clean();
List<string> privateUseSubTags =
new List<string>(cleaner.PrivateUse.Split(new char[] {'-'}, StringSplitOptions.RemoveEmptyEntries));
int privateUseSubTagIndex = 0;
bool privateUsePrefix = false;
string privateUseSubTag = null;
int part = -1;
string languageCode = cleaner.Language;
if (string.IsNullOrEmpty(languageCode))
return false;
if (languageCode.Equals("qaa", StringComparison.OrdinalIgnoreCase))
{
// In our own WS dialog, we don't allow no language, but if it isn't a standard one, a language like xkal
// produces an identifier like qaa-x-kal, and we interepret the first thing after the x as a private
// language code (not allowed as the first three characters according to the standard).
// If it's NOT a valid language code (e.g., too many characters), probably came from some other
// program. Treating it as a language code will fail if we try to create such a writing system,
// since we will detect the invalid language code. So only interpret the first element
// after the x as a language code if it is a valid one. Otherwise, we just let qaa be the language.
if (privateUseSubTags.Count > 0 && IsLanguageCodeValid(privateUseSubTags[0]))
{
languageSubtag = new LanguageSubtag(privateUseSubTags[0], "", true, null);
privateUseSubTags.RemoveAt(0);
}
else
languageSubtag = GetLanguageSubtag("qaa"); // We do allow just plain qaa.
}
else
{
languageSubtag = GetLanguageSubtag(languageCode);
}
string scriptCode = cleaner.Script;
if (!string.IsNullOrEmpty(scriptCode))
{
if (scriptCode.Equals("Qaaa", StringComparison.OrdinalIgnoreCase) && privateUseSubTags.Count > 0)
{
scriptSubtag = new ScriptSubtag(privateUseSubTags[0], "", true);
privateUseSubTags.RemoveAt(0);
}
else
scriptSubtag = GetScriptSubtag(scriptCode);
}
string regionCode = cleaner.Region;
if (!string.IsNullOrEmpty(regionCode))
{
if (regionCode.Equals("QM", StringComparison.OrdinalIgnoreCase) && privateUseSubTags.Count > 0)
{
regionSubtag = new RegionSubtag(privateUseSubTags[0], "", true);
privateUseSubTags.RemoveAt(0);
}
else
regionSubtag = GetRegionSubtag(regionCode);
}
var variantSb = new StringBuilder();
string variantCode = cleaner.Variant;
if (!string.IsNullOrEmpty(variantCode))
{
variantSb.Append(variantCode);
}
if (privateUseSubTags.Count > 0)
{
if (variantSb.Length > 0)
variantSb.Append("-");
variantSb.Append("x");
foreach (var item in privateUseSubTags)
{
variantSb.Append("-");
variantSb.Append(item);
}
}
variantCode = variantSb.ToString();
if (!string.IsNullOrEmpty(variantCode))
{
variantSubtag = GetVariantSubtag(variantCode);
}
return true;
}
/// <summary>
/// Generates an ICU locale from the specified language tag subtags.
/// </summary>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public static string ToIcuLocale(LanguageSubtag languageSubtag, ScriptSubtag scriptSubtag, RegionSubtag regionSubtag, VariantSubtag variantSubtag)
{
if (languageSubtag == null)
throw new ArgumentNullException("languageSubtag");
var sb = new StringBuilder();
//start with the LanguageCode
if (languageSubtag.IsPrivateUse)
sb.Append("x");
sb.Append(languageSubtag.Code);
//now add the Script if it exists
if (scriptSubtag != null)
sb.AppendFormat("_{0}", scriptSubtag.Code);
//now add the Region if it exists
if (regionSubtag != null)
sb.AppendFormat("_{0}", regionSubtag.Code);
// if variantCode is notNullofEmpty then add it
// and if CountryCode is empty add two underscores instead of one.
if (variantSubtag != null)
{
string icuVariant = null;
// convert language tag variants to known ICU variants
// TODO: are there any more ICU variants?
switch (variantSubtag.Code)
{
case "fonipa":
icuVariant = "IPA";
break;
case "fonipa-x-etic":
icuVariant = "X_ETIC";
break;
case "fonipa-x-emic":
icuVariant = "X_EMIC";
break;
case "pinyin":
icuVariant = "X_PY";
break;
}
if (!string.IsNullOrEmpty(icuVariant))
sb.AppendFormat(regionSubtag == null ? "__{0}" : "_{0}", icuVariant);
}
return sb.ToString();
}
/// <summary>
/// Generates a language tag from the specified subtags.
/// </summary>
/// <param name="languageSubtag">The language subtag.</param>
/// <param name="scriptSubtag">The script subtag.</param>
/// <param name="regionSubtag">The region subtag.</param>
/// <param name="variantSubtag">The variant subtag.</param>
/// <returns></returns>
public static string ToLangTag(LanguageSubtag languageSubtag, ScriptSubtag scriptSubtag, RegionSubtag regionSubtag, VariantSubtag variantSubtag)
{
if (languageSubtag == null)
throw new ArgumentNullException("languageSubtag");
var sb = new StringBuilder();
// Insert non-custom language, script, region into main part of code.
if (languageSubtag.IsPrivateUse)
{
sb.Append("qaa");
}
else
{
sb.Append(languageSubtag.Code);
}
var isCustomScript = false;
if (scriptSubtag != null)
{
sb.Append("-");
// Qaaa is our flag to expect a script in private-use. If the actual value is Qaaa, we need to treat it as custom,
// so we don't confuse some other private-use tag with a custom script.
isCustomScript = TreatAsCustomScript(scriptSubtag.Code);
if (isCustomScript)
sb.Append("Qaaa");
else
sb.Append(scriptSubtag.Code);
}
var isCustomRegion = false;
if (regionSubtag != null)
{
sb.Append("-");
// QM is our flag to expect a region in private-use. If the actual value is QM, we need to treat it as custom,
// so we don't confuse some other private-use tag with a custom region.
isCustomRegion = TreatAsCustomRegion(regionSubtag.Code);
if (isCustomRegion)
sb.Append("QM");
else
sb.Append(regionSubtag.Code);
}
string standardVariant = "";
string privateUse = "";
if (variantSubtag != null)
privateUse = GetPrivateUseAndStandardVariant(variantSubtag.Code, out standardVariant);
if (standardVariant != "")
{
sb.Append("-");
sb.Append(standardVariant);
}
// Insert custom language, script, or variant into private=use.
bool inPrivateUse = false;
if (languageSubtag.IsPrivateUse)
{
sb.Append("-");
if (!inPrivateUse)
{
inPrivateUse = true;
sb.Append("x-");
}
sb.Append(languageSubtag.Code);
}
if (isCustomScript)
{
sb.Append("-");
if (!inPrivateUse)
{
inPrivateUse = true;
sb.Append("x-");
}
sb.Append(scriptSubtag.Code);
}
if (isCustomRegion)
{
sb.Append("-");
if (!inPrivateUse)
{
inPrivateUse = true;
sb.Append("x-");
}
sb.Append(regionSubtag.Code);
}
else if (languageSubtag.Code == "zh" && languageSubtag.ISO3Code == "cmn" &&
regionSubtag == null)
{
sb.Append("-CN");
}
if (privateUse != "")
{
sb.Append("-");
if (!inPrivateUse)
sb.Append("x-");
sb.Append(privateUse);
}
return sb.ToString();
}
