Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-23 03:33:18 +01:00
Code Issues Projects Releases Wiki Activity

Fixed tessdata links + select last downloaded language

Browse Source
This commit is contained in:
Nikolaj Olsson 2017-04-27 21:48:05 +02:00
parent 5236831b7c
commit b2bbef1163
4 changed files with 131 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/Forms/GetTesseractDictionaries.cs
View File

@ -16,6 +16,7 @@ namespace Nikse.SubtitleEdit.Forms
private List<string> _descriptions = new List<string>();
private string _xmlName = null;
private string _dictionaryFileName = null;
internal string ChosenLanguage { get; private set; }
public GetTesseractDictionaries()
{
@ -101,6 +102,7 @@ namespace Nikse.SubtitleEdit.Forms
int index = comboBoxDictionaries.SelectedIndex;
string url = _dictionaryDownloadLinks[index];
ChosenLanguage = comboBoxDictionaries.Items[index].ToString();
var wc = new WebClient { Proxy = Utilities.GetProxy() };
if (url.EndsWith(".traineddata", StringComparison.OrdinalIgnoreCase) || url.EndsWith(".traineddata?raw=true", StringComparison.OrdinalIgnoreCase))

13
src/Forms/Ocr/VobSubOcr.cs
View File

@ -6878,7 +6878,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
}
}
private void InitializeTesseract()
private void InitializeTesseract(string chosenLanguage = null)
{
if (!Directory.Exists(Configuration.TesseractDirectory) && !Configuration.IsRunningOnLinux() && !Configuration.IsRunningOnMac())
{
@ -6924,8 +6924,15 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
{
for (int i = 0; i < comboBoxTesseractLanguages.Items.Count; i++)
{
if ((comboBoxTesseractLanguages.Items[i] as TesseractLanguage).Id == Configuration.Settings.VobSubOcr.TesseractLastLanguage)
if (chosenLanguage != null && chosenLanguage == (comboBoxTesseractLanguages.Items[i] as TesseractLanguage).Text)
{
comboBoxTesseractLanguages.SelectedIndex = i;
break;
}
if ((comboBoxTesseractLanguages.Items[i] as TesseractLanguage).Id == Configuration.Settings.VobSubOcr.TesseractLastLanguage)
{
comboBoxTesseractLanguages.SelectedIndex = i;
}
}
if (comboBoxTesseractLanguages.SelectedIndex == -1)
@ -8468,8 +8475,8 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
using (var form = new GetTesseractDictionaries())
{
form.ShowDialog(this);
InitializeTesseract(form.ChosenLanguage);
}
InitializeTesseract();
}
private void toolStripMenuItemInspectNOcrMatches_Click(object sender, EventArgs e)

163
src/Resources/TesseractDictionaries.xml
View File

@ -1,28 +1,53 @@
<!-- Alternately from https://sourceforge.net/projects/tesseract-ocr-alt/files/ -->
<TesseractDictionaries>
<!--<Dictionary>
<Dictionary>
<EnglishName>Afrikaans</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.afr.tar.gz</DownloadLink>
<Description>Afrikaans language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Albanian</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.sqi.tar.gz</DownloadLink>
<Description>Albanian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Arabic</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ara.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ara.tar.gz</DownloadLink>
<Description>Arabic language data for Tesseract 3.02</Description>
</Dictionary>-->
</Dictionary>
<Dictionary>
<EnglishName>Azerbaijani</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.aze.tar.gz</DownloadLink>
<Description>Azerbaijani language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Basque</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/eus.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.eus.tar.gz</DownloadLink>
<Description>Basque language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Belarusian</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.bel.tar.gz</DownloadLink>
<Description>Belarusian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Bengali</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ben.tar.gz</DownloadLink>
<Description>Bengali language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Bulgarian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/bul.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.bul.tar.gz</DownloadLink>
<Description>Bulgarian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Catalan</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/cat.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.cat.tar.gz</DownloadLink>
<Description>Catalan language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Chinese (Simplified)</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/chi_sim.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.chi_sim.tar.gz</DownloadLink>
<Description>Chinese (Simplified) language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
@ -30,179 +55,229 @@
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/chi_tra.traineddata</DownloadLink>
<Description>Chinese (Traditional) language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Cherokee</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.chr.tar.gz</DownloadLink>
<Description>Cherokee language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Croatian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/hrv.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.hvr.tar.gz</DownloadLink>
<Description>Croatian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Czech</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ces.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ces.tar.gz</DownloadLink>
<Description>Czech language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Danish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/dan.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.dan.tar.gz</DownloadLink>
<Description>Danish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Dutch</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/nld.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.nld.tar.gz</DownloadLink>
<Description>Dutch language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>English</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/eng.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.eng.tar.gz</DownloadLink>
<Description>English language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Estonian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/est.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.est.tar.gz</DownloadLink>
<Description>Estonian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Finnish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/fin.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.fin.tar.gz</DownloadLink>
<Description>Finnish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>French</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/fra.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.fra.tar.gz</DownloadLink>
<Description>French language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Galician</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.glg.tar.gz</DownloadLink>
<Description>Galician language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>German</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/deu.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.deu.tar.gz</DownloadLink>
<Description>German language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Greek</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ell.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ell.tar.gz</DownloadLink>
<Description>Greek language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Hebrew</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/heb.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.heb.tar.gz</DownloadLink>
<Description>Hebrew language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Hindi</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/blob/master/tessdata/hin.tar.gz?raw=true</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.hin.tar.gz</DownloadLink>
<Description>Hindi language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Hungarian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/hun.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.hun.tar.gz</DownloadLink>
<Description>Hungarian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Icelandic</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/isl.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.isl.tar.gz</DownloadLink>
<Description>Icelandic language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Indonesian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ind.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ind.tar.gz</DownloadLink>
<Description>Indonesian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Italian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ita.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ita.tar.gz</DownloadLink>
<Description>Italian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Japanese</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/jpn.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.jpn.tar.gz</DownloadLink>
<Description>Japanese language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Kannada</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.kan.tar.gz</DownloadLink>
<Description>Kannada language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Korean</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/kor.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.kor.tar.gz</DownloadLink>
<Description>Korean language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Latvian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/lav.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.lav.tar.gz</DownloadLink>
<Description>Latvian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Lithuanian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/lit.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.lit.tar.gz</DownloadLink>
<Description>Lithuanian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Macedonian</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.mkd.tar.gz</DownloadLink>
<Description>Macedonian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Malay</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/msa.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.msa.tar.gz</DownloadLink>
<Description>Malay language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Malayalam</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.mal.tar.gz</DownloadLink>
<Description>Malayalam language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Maltese</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.mlt.tar.gz</DownloadLink>
<Description>Maltese language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Norwegian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/nor.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.nor.tar.gz</DownloadLink>
<Description>Norwegian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Polish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/pol.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.pol.tar.gz</DownloadLink>
<Description>Polish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Portuguese</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/por.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.por.tar.gz</DownloadLink>
<Description>Portuguese language data for Tesseract 3.02 </Description>
</Dictionary>
<Dictionary>
<EnglishName>Romanian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ron.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ron.tar.gz</DownloadLink>
<Description>Romanian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Russian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/rus.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.rus.tar.gz</DownloadLink>
<Description>Russian Language Data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Serbian (Latin)</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/srp.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.srp.tar.gz</DownloadLink>
<Description>Serbian (Latin) language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Slovak</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/074c37215b01ab8cc47a0e06ff7356383883d775/slk.traineddata</DownloadLink>
<Description>Slovak language data for Tesseract 3.04</Description>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.slk.tar.gz</DownloadLink>
<Description>Slovak language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Slovenian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/blob/bf82613055ebc6e63d9e3b438a5c234bfd638c93/slv.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.slv.tar.gz</DownloadLink>
<Description>Slovenian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Spanish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/spa.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.spa.tar.gz</DownloadLink>
<Description>Spanish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Swahili</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.swa.tar.gz</DownloadLink>
<Description>Swahili language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Swedish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/swe.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.swe.tar.gz</DownloadLink>
<Description>Swedish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Tagalog</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.tgl.tar.gz</DownloadLink>
<Description>Tagalog language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Tamil</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.tam.tar.gz</DownloadLink>
<Description>Tamil language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Thai</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/tha.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.tha.tar.gz</DownloadLink>
<Description>Thai language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Telugu</EnglishName>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.tel.tar.gz</DownloadLink>
<Description>Telugu language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Turkish</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/tur.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.tur.tar.gz</DownloadLink>
<Description>Turkish language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Ukrainian</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/ukr.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.ukr.tar.gz</DownloadLink>
<Description>Ukrainian language data for Tesseract 3.02</Description>
</Dictionary>
<Dictionary>
<EnglishName>Vietnamese</EnglishName>
<DownloadLink>https://github.com/tesseract-ocr/tessdata/raw/bf82613055ebc6e63d9e3b438a5c234bfd638c93/vie.traineddata</DownloadLink>
<DownloadLink>https://github.com/SubtitleEdit/support-files/raw/master/tessdata/tesseract-ocr-3.02.vie.tar.gz</DownloadLink>
<Description>Vietnamese Language Data for Tesseract 3.02</Description>
</Dictionary>
</TesseractDictionaries>
</TesseractDictionaries>

BIN
src/Resources/TesseractDictionaries.xml.gz
View File

Binary file not shown.