private static IKeyedCollection <string, SldrLanguageTagInfo> DeriveTagsFromJsonEntries(List <AllTagEntry> rootObject)
{
var tags = new KeyedList <string, SldrLanguageTagInfo>(info => info.LanguageTag, StringComparer.InvariantCultureIgnoreCase);
foreach (AllTagEntry entry in rootObject)
{
// tags starting with x- have undefined structure so ignoring them
// tags starting with _ showed up in buggy data so we'll drop them also
if (!entry.tag.StartsWith("x-") && !entry.tag.StartsWith("_"))
{
LanguageSubtag languageTag;
if (!entry.deprecated && !StandardSubtags.RegisteredLanguages.TryGet(entry.tag.Split('-')[0], out languageTag))
{
if (entry.iso639_3 == null)
{
StandardSubtags.AddLanguage(entry.tag.Split('-')[0], entry.name, false, entry.tag.Split('-')[0]);
}
else if (!StandardSubtags.RegisteredLanguages.TryGet(entry.iso639_3, out languageTag))
{
StandardSubtags.AddLanguage(entry.iso639_3, entry.name, false, entry.iso639_3);
}
}
string implicitStringCode = null;
// the script is always in the full tag
string scriptCode = entry.full.Split('-')[1];
if (scriptCode.Length == 4)
{
var tagComponents = entry.tag.Split('-');
ScriptSubtag scriptTag;
if (!StandardSubtags.RegisteredScripts.TryGet(scriptCode, out scriptTag))
{
StandardSubtags.AddScript(scriptCode, scriptCode);
}
// if the script is also in the tag then it is explicit not implicit
if (tagComponents.Length == 1 || tagComponents[1] != scriptCode)
{
implicitStringCode = scriptCode;
}
}
tags.Add(new SldrLanguageTagInfo(entry.tag, implicitStringCode, entry.tag, entry.sldr));
}
}
return(tags);
}
internal static IKeyedCollection <string, SldrLanguageTagInfo> ParseAllTagsJson(string allTagsContent)
{
// read in the json file
/* {
* "full": "aa-Latn-ET",
* "name": "Afar",
* "names": [
* "Adal",
* ...
* ],
* "sldr": true,
* "tag": "aa",
* "tags": [
* "aa-ET",
* "aa-Latn"
* ]
* },*/
// for each entry
// tag -> langtag which is same as sldrtag
// sldr -> isAvailable
// tags -> process to find implicitScript
var tags = new KeyedList <string, SldrLanguageTagInfo>(info => info.LanguageTag, StringComparer.InvariantCultureIgnoreCase);
List <AllTagEntry> rootObject = JsonConvert.DeserializeObject <List <AllTagEntry> >(allTagsContent);
foreach (AllTagEntry entry in rootObject)
{
if (!entry.tag.StartsWith("x-")) // tags starting with x- have undefined structure so ignoring them
{
LanguageSubtag languageTag;
if (!entry.deprecated && !StandardSubtags.RegisteredLanguages.TryGet(entry.tag.Split('-')[0], out languageTag))
{
if (entry.iso639_3 == null)
{
StandardSubtags.AddLanguage(entry.tag.Split('-')[0], entry.name, false, entry.tag.Split('-')[0]);
}
else if (!StandardSubtags.RegisteredLanguages.TryGet(entry.iso639_3, out languageTag))
{
StandardSubtags.AddLanguage(entry.iso639_3, entry.name, false, entry.iso639_3);
}
}
string implicitStringCode = null;
// the script is always in the full tag
string scriptCode = entry.full.Split('-')[1];
if (scriptCode.Length == 4)
{
var tagComponents = entry.tag.Split('-');
ScriptSubtag scriptTag;
if (!StandardSubtags.RegisteredScripts.TryGet(scriptCode, out scriptTag))
{
StandardSubtags.AddScript(scriptCode, scriptCode);
}
// if the script is also in the tag then it is explicit not implicit
if (tagComponents.Length == 1 || tagComponents[1] != scriptCode)
{
implicitStringCode = scriptCode;
}
}
tags.Add(new SldrLanguageTagInfo(entry.tag, implicitStringCode, entry.tag, entry.sldr));
}
}
return(tags);
}
private bool AddLanguage(string code, string threelettercode, string full = null,
string name = null, string localName = null, string region = null, List <string> names = null, string regions = null, List <string> tags = null)
{
string primarycountry;
if (region == null)
{
primarycountry = "";
}
else if (StandardSubtags.IsValidIso3166RegionCode(region))
{
if (StandardSubtags.IsPrivateUseRegionCode(region))
{
if (region == "XK")
{
primarycountry = "Kosovo";
}
else
{
primarycountry = "Unknown private use";
}
}
else
{
primarycountry = StandardSubtags.RegisteredRegions[region].Name; // convert to full region name
}
}
else
{
primarycountry = "Invalid region";
}
LanguageInfo language = new LanguageInfo
{
LanguageTag = code,
ThreeLetterTag = threelettercode,
// DesiredName defaults to Names[0], which is set below.
PrimaryCountry = primarycountry
};
language.Countries.Add(primarycountry);
if (regions != null)
{
string[] countries = regions.Split();
foreach (string country in countries)
{
if (!country.Contains('?') && country != "")
{
language.Countries.Add(StandardSubtags.RegisteredRegions[country].Name);
}
}
}
// For sorting, it is better to store name first instead of localName, which may be in a local script.
// Names[0] is used in several ways in sorting languages in the list of possible matches: 1) bring
// to the top of the list languages where Names[0] matches what the user typed, 2) order by the
// "typing distance" of Names[0] from what the user typed, and 3) order by comparing the Names[0]
// value of the two languages if neither matches the search string and their typing distances from
// the search string are the same.
// Names[1] (if it exists) is used to move the language toward the top of the list if it exactly
// matches the search string. It is not used otherwise in the sorting heuristics. No other
// values in the Names list are involved in the sorting process.
if (name != null)
{
language.Names.Add(name.Trim());
}
if (localName != null && localName != name)
{
language.Names.Add(localName.Trim());
}
if (names != null)
{
foreach (string langname in names)
{
if (!language.Names.Contains(langname))
{
language.Names.Add(langname.Trim());
}
}
}
// If we end up needing to add the language code, that reflects a deficiency in the data. But
// having a bogus name value probably hurts less that not having any name at all. The sort
// process mentioned above using the language tag as well as the first two items in the Names list.
Debug.Assert(language.Names.Count > 0);
if (language.Names.Count == 0)
{
language.Names.Add(code);
}
// add language to _codeToLanguageIndex and _nameToLanguageIndex
// if 2 letter code then add both 2 and 3 letter codes to _codeToLanguageIndex
_codeToLanguageIndex[code] = language;
if (full != null && !string.Equals(full, code))
{
_codeToLanguageIndex[full] = language; // add the full expanded tag
}
if (threelettercode != null && !string.Equals(code, threelettercode))
{
_codeToLanguageIndex[threelettercode] = language;
}
if (tags != null)
{
foreach (string langtag in tags)
{
_codeToLanguageIndex[langtag] = language;
}
}
foreach (string langname in language.Names)
{
GetOrCreateListFromName(langname).Add(language);
}
// add to _countryToLanguageIndex
foreach (var country in language.Countries)
{
if (!string.IsNullOrEmpty(country))
{
List <LanguageInfo> list;
if (!_countryToLanguageIndex.TryGetValue(country, out list))
{
list = new List <LanguageInfo>();
_countryToLanguageIndex[country] = list;
}
list.Add(language);
}
}
return(true);
}
