internal void Initialize(Subtitle subtitle, int firstSelectedIndex)
{
var watermark = ReadWaterMark(subtitle.GetAllTexts().Trim());
labelWatermark.Text = string.Format(_language.WatermarkX, watermark);
if (watermark.Length == 0)
{
buttonRemove.Enabled = false;
textBoxWatermark.Focus();
}
else
{
groupBoxGenerate.Enabled = false;
buttonOK.Focus();
}
_firstSelectedIndex = firstSelectedIndex;
var current = subtitle.GetParagraphOrDefault(_firstSelectedIndex);
if (current != null)
{
radioButtonCurrentLine.Text = string.Format(_language.CurrentLineOnlyX, current.Text.Replace(Environment.NewLine, Configuration.Settings.General.ListViewLineSeparatorString));
}
else
{
radioButtonCurrentLine.Text = string.Format(_language.CurrentLineOnlyX, string.Empty);
radioButtonCurrentLine.Enabled = false;
}
}
public static string AutoDetectGoogleLanguageOrNull(Subtitle subtitle)
{
var s = new Subtitle(subtitle);
s.RemoveEmptyLines();
string languageId = AutoDetectGoogleLanguage(s.GetAllTexts(), s.Paragraphs.Count / 14);
if (string.IsNullOrEmpty(languageId))
{
languageId = null;
}
return(languageId);
}
public static string AutoDetectLanguageName(string languageName, Subtitle subtitle)
{
if (string.IsNullOrEmpty(languageName))
languageName = "en_US";
int bestCount = subtitle.Paragraphs.Count / 14;
string text = subtitle.GetAllTexts();
List<string> dictionaryNames = Utilities.GetDictionaryLanguages();
bool containsEnGb = false;
bool containsEnUs = false;
bool containsHrHr = false;
bool containsSrLatn = false;
foreach (string name in dictionaryNames)
{
if (name.Contains("[en_GB]"))
containsEnGb = true;
if (name.Contains("[en_US]"))
containsEnUs = true;
if (name.Contains("[hr_HR]"))
containsHrHr = true;
if (name.Contains("[sr-Latn]"))
containsSrLatn = true;
}
foreach (string name in dictionaryNames)
{
string shortName = string.Empty;
int start = name.IndexOf('[');
int end = name.IndexOf(']');
if (start > 0 && end > start)
{
start++;
shortName = name.Substring(start, end - start);
}
int count;
switch (shortName)
{
case "da_DK":
count = GetCount(text, AutoDetectWordsDanish);
if (count > bestCount)
{
int norwegianCount = GetCount(text, "ut", "deg", "meg", "merkelig", "mye", "spørre");
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (norwegianCount < 2 && dutchCount < count)
languageName = shortName;
}
break;
case "nb_NO":
count = GetCount(text, AutoDetectWordsNorwegian);
if (count > bestCount)
{
int danishCount = GetCount(text, "siger", "dig", "mig", "mærkelig", "tilbage", "spørge");
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (danishCount < 2 && dutchCount < count)
languageName = shortName;
}
break;
case "sv_SE":
count = GetCount(text, AutoDetectWordsSwedish);
if (count > bestCount)
languageName = shortName;
break;
case "en_US":
count = GetCount(text, AutoDetectWordsEnglish);
if (count > bestCount)
{
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (dutchCount < count)
{
languageName = shortName;
if (containsEnGb)
{
int usCount = GetCount(text, "color", "flavor", "honor", "humor", "neighbor", "honor");
int gbCount = GetCount(text, "colour", "flavour", "honour", "humour", "neighbour", "honour");
if (gbCount > usCount)
languageName = "en_GB";
}
}
}
break;
case "en_GB":
count = GetCount(text, AutoDetectWordsEnglish);
if (count > bestCount)
{
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (dutchCount < count)
{
languageName = shortName;
if (containsEnUs)
{
int usCount = GetCount(text, "color", "flavor", "honor", "humor", "neighbor", "honor");
int gbCount = GetCount(text, "colour", "flavour", "honour", "humour", "neighbour", "honour");
if (gbCount < usCount)
languageName = "en_US";
}
}
}
break;
case "es_ES":
count = GetCount(text, AutoDetectWordsSpanish);
if (count > bestCount)
{
int frenchCount = GetCount(text, "[Cc]'est", "pas", "vous", "pour", "suis", "Pourquoi", "maison", "souviens", "quelque"); // not spanish words
int portugueseCount = GetCount(text, "[NnCc]ão", "Então", "h?ouve", "pessoal", "rapariga", "tivesse", "fizeste",
"jantar", "conheço", "atenção", "foste", "milhões", "devias", "ganhar", "raios"); // not spanish words
if (frenchCount < 2 && portugueseCount < 2)
languageName = shortName;
}
break;
case "it_IT":
count = GetCount(text, AutoDetectWordsItalian);
if (count > bestCount)
{
int frenchCount = GetCount(text, "[Cc]'est", "pas", "vous", "pour", "suis", "Pourquoi", "maison", "souviens", "quelque"); // not italian words
int spanishCount = GetCount(text, "Hola", "nada", "Vamos", "pasa", "los", "como"); // not italian words
if (frenchCount < 2 && spanishCount < 2)
languageName = shortName;
}
break;
case "fr_FR":
count = GetCount(text, AutoDetectWordsFrench);
if (count > bestCount)
{
int romanianCount = GetCount(text, "[Vv]reau", "[Ss]înt", "[Aa]cum", "pentru", "domnule", "aici");
int spanishCount = GetCount(text, "Hola", "nada", "Vamos", "pasa", "los", "como"); // not french words
int italianCount = GetCount(text, AutoDetectWordsItalian);
if (romanianCount < 5 && spanishCount < 2 && italianCount < 2)
languageName = shortName;
}
break;
case "de_DE":
count = GetCount(text, AutoDetectWordsGerman);
if (count > bestCount)
languageName = shortName;
break;
case "nl_NL":
count = GetCount(text, AutoDetectWordsDutch);
if (count > bestCount)
languageName = shortName;
break;
case "pl_PL":
count = GetCount(text, AutoDetectWordsPolish);
if (count > bestCount)
languageName = shortName;
break;
case "el_GR":
count = GetCount(text, AutoDetectWordsGreek);
if (count > bestCount)
languageName = shortName;
break;
case "ru_RU":
count = GetCount(text, AutoDetectWordsRussian);
if (count > bestCount)
languageName = shortName;
break;
case "uk_UA":
count = GetCount(text, AutoDetectWordsUkrainian);
if (count > bestCount)
languageName = shortName;
break;
case "ro_RO":
count = GetCount(text, AutoDetectWordsRomanian1);
if (count <= bestCount)
count = GetCount(text, AutoDetectWordsRomanian2);
if (count > bestCount)
languageName = shortName;
break;
case "hr_HR": // Croatian
count = GetCount(text, AutoDetectWordsCroatianAndSerbian);
if (count > bestCount)
{
languageName = shortName;
if (containsSrLatn)
{
int croatianCount = GetCount(text, AutoDetectWordsCroatian);
int serbianCount = GetCount(text, AutoDetectWordsSerbian);
if (serbianCount > croatianCount)
languageName = "sr-Latn";
}
}
break;
case "sr-Latn": // Serbian (Latin)
count = GetCount(text, AutoDetectWordsCroatianAndSerbian);
if (count > bestCount)
{
languageName = shortName;
if (containsHrHr)
{
int croatianCount = GetCount(text, AutoDetectWordsCroatian);
int serbianCount = GetCount(text, AutoDetectWordsSerbian);
if (serbianCount < croatianCount)
languageName = "hr_HR";
}
}
break;
case "sr": // Serbian (Cyrillic)
count = GetCount(text, AutoDetectWordsSerbianCyrillic);
if (count > bestCount)
languageName = shortName;
break;
case "pt_PT": // Portuguese
count = GetCount(text, AutoDetectWordsPortuguese);
if (count > bestCount)
languageName = shortName;
break;
case "pt_BR": // Portuguese (Brasil)
count = GetCount(text, AutoDetectWordsPortuguese);
if (count > bestCount)
languageName = shortName;
break;
case "hu_HU": // Hungarian
count = GetCount(text, AutoDetectWordsHungarian);
if (count > bestCount)
languageName = shortName;
break;
case "cs_CZ": // Czech
count = GetCount(text, AutoDetectWordsCzech);
if (count > bestCount)
languageName = shortName;
break;
case "sk_SK": // Slovak
count = GetCount(text, AutoDetectWordsSlovak);
if (count > bestCount)
languageName = shortName;
break;
}
}
return languageName;
}
public static string AutoDetectGoogleLanguageOrNull(Subtitle subtitle)
{
string languageId = AutoDetectGoogleLanguage(subtitle.GetAllTexts(), subtitle.Paragraphs.Count / 14);
if (string.IsNullOrEmpty(languageId))
languageId = null;
return languageId;
}
public static string AutoDetectLanguageName(string languageName, Subtitle subtitle)
{
if (string.IsNullOrEmpty(languageName))
{
languageName = "en_US";
}
int bestCount = subtitle.Paragraphs.Count / 14;
string text = subtitle.GetAllTexts();
List <string> dictionaryNames = Utilities.GetDictionaryLanguages();
bool containsEnGb = false;
bool containsEnUs = false;
bool containsHrHr = false;
bool containsSrLatn = false;
foreach (string name in dictionaryNames)
{
if (name.Contains("[en_GB]"))
{
containsEnGb = true;
}
if (name.Contains("[en_US]"))
{
containsEnUs = true;
}
if (name.Contains("[hr_HR]"))
{
containsHrHr = true;
}
if (name.Contains("[sr_Latn]"))
{
containsSrLatn = true;
}
}
foreach (string name in dictionaryNames)
{
string shortName = string.Empty;
int start = name.IndexOf('[');
int end = name.IndexOf(']');
if (start > 0 && end > start)
{
start++;
shortName = name.Substring(start, end - start);
}
int count;
switch (shortName)
{
case "da_DK":
count = GetCount(text, AutoDetectWordsDanish);
if (count > bestCount)
{
int norwegianCount = GetCount(text, "ut", "deg", "meg", "merkelig", "mye", "spørre");
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (norwegianCount < 2 && dutchCount < count)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "nb_NO":
count = GetCount(text, AutoDetectWordsNorwegian);
if (count > bestCount)
{
int danishCount = GetCount(text, "siger", "dig", "mig", "mærkelig", "tilbage", "spørge");
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (danishCount < 2 && dutchCount < count)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "sv_SE":
count = GetCount(text, AutoDetectWordsSwedish);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "en_US":
count = GetCount(text, AutoDetectWordsEnglish);
if (count > bestCount)
{
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (dutchCount < count)
{
languageName = shortName;
bestCount = count;
if (containsEnGb)
{
int usCount = GetCount(text, "color", "flavor", "honor", "humor", "neighbor", "honor");
int gbCount = GetCount(text, "colour", "flavour", "honour", "humour", "neighbour", "honour");
if (gbCount > usCount)
{
languageName = "en_GB";
}
}
}
}
break;
case "en_GB":
count = GetCount(text, AutoDetectWordsEnglish);
if (count > bestCount)
{
int dutchCount = GetCount(text, AutoDetectWordsDutch);
if (dutchCount < count)
{
languageName = shortName;
bestCount = count;
if (containsEnUs)
{
int usCount = GetCount(text, "color", "flavor", "honor", "humor", "neighbor", "honor");
int gbCount = GetCount(text, "colour", "flavour", "honour", "humour", "neighbour", "honour");
if (gbCount < usCount)
{
languageName = "en_US";
}
}
}
}
break;
case "es_ES":
count = GetCount(text, AutoDetectWordsSpanish);
if (count > bestCount)
{
int frenchCount = GetCount(text, "[Cc]'est", "pas", "vous", "pour", "suis", "Pourquoi", "maison", "souviens", "quelque"); // not spanish words
int portugueseCount = GetCount(text, "[NnCc]ão", "Então", "h?ouve", "pessoal", "rapariga", "tivesse", "fizeste",
"jantar", "conheço", "atenção", "foste", "milhões", "devias", "ganhar", "raios"); // not spanish words
if (frenchCount < 2 && portugueseCount < 2)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "it_IT":
count = GetCount(text, AutoDetectWordsItalian);
if (count > bestCount)
{
int frenchCount = GetCount(text, "[Cc]'est", "pas", "vous", "pour", "suis", "Pourquoi", "maison", "souviens", "quelque"); // not italian words
int spanishCount = GetCount(text, "Hola", "nada", "Vamos", "pasa", "los", "como"); // not italian words
if (frenchCount < 2 && spanishCount < 2)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "fr_FR":
count = GetCount(text, AutoDetectWordsFrench);
if (count > bestCount)
{
int romanianCount = GetCount(text, "[Vv]reau", "[Ss]înt", "[Aa]cum", "pentru", "domnule", "aici");
int spanishCount = GetCount(text, "Hola", "nada", "Vamos", "pasa", "los", "como"); // not french words
int italianCount = GetCount(text, AutoDetectWordsItalian);
if (romanianCount < 5 && spanishCount < 2 && italianCount < 2)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "de_DE":
count = GetCount(text, AutoDetectWordsGerman);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "nl_NL":
count = GetCount(text, AutoDetectWordsDutch);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "pl_PL":
count = GetCount(text, AutoDetectWordsPolish);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "el_GR":
count = GetCount(text, AutoDetectWordsGreek);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "ru_RU":
count = GetCount(text, AutoDetectWordsRussian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "uk_UA":
count = GetCount(text, AutoDetectWordsUkrainian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "ro_RO":
count = GetCount(text, AutoDetectWordsRomanian1);
if (count <= bestCount)
{
count = GetCount(text, AutoDetectWordsRomanian2);
}
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "hr_HR": // Croatian
count = GetCount(text, AutoDetectWordsCroatianAndSerbian);
if (count > bestCount)
{
bestCount = count;
languageName = shortName;
if (containsSrLatn)
{
int croatianCount = GetCount(text, AutoDetectWordsCroatian);
int serbianCount = GetCount(text, AutoDetectWordsSerbian);
if (serbianCount > croatianCount)
{
languageName = "sr-Latn";
}
}
}
break;
case "sr_Latn": // Serbian (Latin)
count = GetCount(text, AutoDetectWordsCroatianAndSerbian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
if (containsHrHr)
{
int croatianCount = GetCount(text, AutoDetectWordsCroatian);
int serbianCount = GetCount(text, AutoDetectWordsSerbian);
if (serbianCount < croatianCount)
{
languageName = "hr_HR";
}
}
}
break;
case "sr": // Serbian (Cyrillic)
count = GetCount(text, AutoDetectWordsSerbianCyrillic);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "pt_PT": // Portuguese Portugal
case "pt_BR": // Portuguese Brazil
count = GetCount(text, AutoDetectWordsPortuguese);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "hu_HU": // Hungarian
count = GetCount(text, AutoDetectWordsHungarian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "cs_CZ": // Czech
count = GetCount(text, AutoDetectWordsCzech);
if (count > bestCount)
{
var lithuanianCount = GetCount(text, AutoDetectWordsLithuanian);
if (count > lithuanianCount)
{
languageName = shortName;
bestCount = count;
}
}
break;
case "sk_SK": // Slovak
count = GetCount(text, AutoDetectWordsSlovak);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "lv_LV": // Latvian
count = GetCount(text, AutoDetectWordsLatvian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "lt_LT": // Lithuanian
case "lt": // Lithuanian (Neutral)
count = GetCount(text, AutoDetectWordsLithuanian);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
case "hi_IN": // Hindi
case "hi":
count = GetCount(text, AutoDetectWordsHindi);
if (count > bestCount)
{
languageName = shortName;
bestCount = count;
}
break;
}
}
return(languageName);
}