Update Tesseract from 5.2.0 to 5.3.0

Changelog.txt

@@ -10,6 +10,7 @@
    * Add new subtitle format - thx Chris
    * Add ASSA style "Replace" - thx Mishasama
    * Add "audio to text selected lines" shortcuts
+   * Add Vosk Korean/Uzbek models
 * IMPROVED:
    * Update Chinese translation - thx nkh0472
    * Update Korean translation - thx domddol
@@ -31,6 +32,7 @@
    * Remember "Merge lines with same text" settings - thx kamitsu2004
    * Update Whisper CPP to v1.0.4
    * Rename "Fix dialogs on one line" to "Break dialogs on one line"
+   * Update Tesseract 5.2.0 to 5.3.0
 * FIXED:
    * Fix italic in DCinema interop - thx Andrey
    * Fix Whisper model download - thx darnn

src/libse/Common/Configuration.cs

@@ -24,7 +24,7 @@ namespace Nikse.SubtitleEdit.Core.Common
         public static readonly string ShotChangesDirectory = DataDirectory + "ShotChanges" + Path.DirectorySeparatorChar;
         public static readonly string AutoBackupDirectory = DataDirectory + "AutoBackup" + Path.DirectorySeparatorChar;
         public static readonly string VobSubCompareDirectory = DataDirectory + "VobSub" + Path.DirectorySeparatorChar;
-        public static readonly string TesseractDirectory = DataDirectory + "Tesseract520" + Path.DirectorySeparatorChar;
+        public static readonly string TesseractDirectory = DataDirectory + "Tesseract530" + Path.DirectorySeparatorChar;
         public static readonly string Tesseract302Directory = DataDirectory + "Tesseract302" + Path.DirectorySeparatorChar;
         public static readonly string WaveformsDirectory = DataDirectory + "Waveforms" + Path.DirectorySeparatorChar;
         public static readonly string PluginsDirectory = DataDirectory + "Plugins";
@@ -227,6 +227,7 @@ namespace Nikse.SubtitleEdit.Core.Common
                     return "/usr/share/tessdata";
                 }
             }
+
             return Path.Combine(TesseractDirectory, "tessdata");
         }
 

src/ui/Forms/AudioToText/VoskModelDownload.cs

@@ -142,8 +142,6 @@ namespace Nikse.SubtitleEdit.Forms.AudioToText
             Cursor = Cursors.Default;
             labelPleaseWait.Text = string.Empty;
 
-            
-
             if (AutoClose)
             {
                 DialogResult = DialogResult.OK;

src/ui/Forms/Ocr/DownloadTesseract5.cs

@@ -2,7 +2,6 @@
 using Nikse.SubtitleEdit.Logic;
 using System;
 using System.IO;
-using System.IO.Compression;
 using System.Threading;
 using System.Windows.Forms;
 
@@ -10,7 +9,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
 {
     public sealed partial class DownloadTesseract5 : Form
     {
-        public const string TesseractDownloadUrl = "https://github.com/SubtitleEdit/support-files/raw/master/Tesseract520.tar.gz";
+        public const string TesseractDownloadUrl = "https://github.com/SubtitleEdit/support-files/raw/master/Tesseract530.zip";
         private readonly CancellationTokenSource _cancellationTokenSource;
 
         public DownloadTesseract5(string version)
@@ -77,35 +76,18 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                 Directory.CreateDirectory(dictionaryFolder);
             }
 
-            downloadStream.Position = 0;
+            using (var zip = ZipExtractor.Open(downloadStream))
-            var tempFileName = FileUtil.GetTempFileName(".tar");
-            using (var fs = new FileStream(tempFileName, FileMode.Create))
-            using (var zip = new GZipStream(downloadStream, CompressionMode.Decompress))
             {
-                byte[] buffer = new byte[1024];
+                var dir = zip.ReadCentralDir();
-                int nRead;
+                foreach (var entry in dir)
-                while ((nRead = zip.Read(buffer, 0, buffer.Length)) > 0)
                 {
-                    fs.Write(buffer, 0, nRead);
+                    var path = Path.Combine(dictionaryFolder, entry.FilenameInZip);
+                    zip.ExtractFile(entry, path);
                 }
             }
 
-            using (var tr = new TarReader(tempFileName))
+            Cursor = Cursors.Default;
-            {
+            labelPleaseWait.Text = string.Empty;
-                foreach (var th in tr.Files)
-                {
-                    var fn = Path.Combine(dictionaryFolder, th.FileName.Replace('/', Path.DirectorySeparatorChar));
-                    if (th.IsFolder)
-                    {
-                        Directory.CreateDirectory(Path.Combine(dictionaryFolder, th.FileName.Replace('/', Path.DirectorySeparatorChar)));
-                    }
-                    else if (th.FileSizeInBytes > 0)
-                    {
-                        th.WriteData(fn);
-                    }
-                }
-            }
-            File.Delete(tempFileName);
             Cursor = Cursors.Default;
             DialogResult = DialogResult.OK;
         }

src/ui/Forms/Ocr/VobSubOcr.cs

@@ -329,7 +329,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
         // Dictionaries/spellchecking/fixing
         private OcrFixEngine _ocrFixEngine;
         private int _tesseractOcrAutoFixes;
-        private string Tesseract5Version = "5.2.0";
+        private string Tesseract5Version = "5.3.0";
 
         private Subtitle _bdnXmlOriginal;
         private Subtitle _bdnXmlSubtitle;