Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-22 19:22:53 +01:00
Code Issues Projects Releases Wiki Activity

Added Lithuanian language auto-detection

Browse Source
This commit is contained in:
Nikolaj Olsson 2016-11-19 22:30:47 +01:00
parent 5511d2df9e
commit e7b5e6e66f
1 changed files with 28 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
libse/LanguageAutoDetect.cs
View File

@ -135,6 +135,7 @@ namespace Nikse.SubtitleEdit.Core
"[Aa]?[Ll]ebo", "[Pp]ri", "[Pp]re", "([Íí]sť|[Ii](?:dem|de|deme|dú))", "[Pp]red.*", "[Mm]edzi", "[Ee]šte", "[Čč]lovek", "[Pp]odľa", "[Ďď]alš(í|ia|ie)" }; "[Aa]?[Ll]ebo", "[Pp]ri", "[Pp]re", "([Íí]sť|[Ii](?:dem|de|deme|dú))", "[Pp]red.*", "[Mm]edzi", "[Ee]šte", "[Čč]lovek", "[Pp]odľa", "[Ďď]alš(í|ia|ie)" };
private static readonly string[] AutoDetectWordsLatvian = { "Paldies", "neesmu ", "nezinu", "viòð", "viņš", "viņu", "kungs", "esmu", "Viņš", "Velns", "viņa", "dievs", "Pagaidi", "varonis", "agrāk", "varbūt" }; private static readonly string[] AutoDetectWordsLatvian = { "Paldies", "neesmu ", "nezinu", "viòð", "viņš", "viņu", "kungs", "esmu", "Viņš", "Velns", "viņa", "dievs", "Pagaidi", "varonis", "agrāk", "varbūt" };
private static readonly string[] AutoDetectWordsLithuanian = { "tavęs", "veidai", "apie", "jums", "Veidai", "Kaip", "kaip", "reikia", "Šūdas", "frensis", "Ačiū", "vilsonai", "Palauk", "Veidas", "viskas", "Tikrai", "manęs", "Tačiau", "žmogau", "Flagai", "Prašau", "Džiune", "Nakties", "šviesybe" };
private static string AutoDetectGoogleLanguage(string text, int bestCount) private static string AutoDetectGoogleLanguage(string text, int bestCount)
{ {
@ -298,6 +299,9 @@ namespace Nikse.SubtitleEdit.Core
count = GetCount(text, AutoDetectWordsCzechAndSlovak); count = GetCount(text, AutoDetectWordsCzechAndSlovak);
if (count > bestCount) if (count > bestCount)
{
var lithuanianCount = GetCount(text, AutoDetectWordsLithuanian);
if (lithuanianCount <= count)
{ {
int czechWordsCount = GetCount(text, AutoDetectWordsCzech); int czechWordsCount = GetCount(text, AutoDetectWordsCzech);
int slovakWordsCount = GetCount(text, AutoDetectWordsSlovak); int slovakWordsCount = GetCount(text, AutoDetectWordsSlovak);
@ -305,11 +309,16 @@ namespace Nikse.SubtitleEdit.Core
return "cs"; // Czech return "cs"; // Czech
return "sk"; // Slovak return "sk"; // Slovak
} }
}
count = GetCount(text, AutoDetectWordsLatvian); count = GetCount(text, AutoDetectWordsLatvian);
if (count > bestCount * 1.2) if (count > bestCount * 1.2)
return "lv"; return "lv";
count = GetCount(text, AutoDetectWordsLithuanian);
if (count > bestCount)
return "lt";
return string.Empty; return string.Empty;
} }
@ -551,7 +560,13 @@ namespace Nikse.SubtitleEdit.Core
case "cs_CZ": // Czech case "cs_CZ": // Czech
count = GetCount(text, AutoDetectWordsCzech); count = GetCount(text, AutoDetectWordsCzech);
if (count > bestCount) if (count > bestCount)
{
var lithuanianCount = GetCount(text, AutoDetectWordsLithuanian);
if (count > lithuanianCount)
{
languageName = shortName; languageName = shortName;
}
}
break; break;
case "sk_SK": // Slovak case "sk_SK": // Slovak
count = GetCount(text, AutoDetectWordsSlovak); count = GetCount(text, AutoDetectWordsSlovak);
@ -563,6 +578,12 @@ namespace Nikse.SubtitleEdit.Core
if (count > bestCount) if (count > bestCount)
languageName = shortName; languageName = shortName;
break; break;
case "lt_LT": // Lithuanian
case "lt": // Lithuanian
count = GetCount(text, AutoDetectWordsLithuanian);
if (count > bestCount)
languageName = shortName;
break;
} }
} }
return languageName; return languageName;