Tesseract5 support (Tentacule#37)

* added tesseract 5 support * added tesseract version log * set tesseract version 4 as default for non windows platform * fixed default tesseract version * updated doc
lyj0309 · Sep 4, 2023 · e6f6d88 · e6f6d88
1 parent d5de23e
commit e6f6d88
Show file tree

Hide file tree

Showing 6 changed files with 77 additions and 38 deletions.
diff --git a/README.md b/README.md
@@ -23,6 +23,7 @@ dotnet PgsToSrt.dll [parameters]
 | `--trackLanguage`     | Convert all tracks of the specified language (only works with `.mkv` input)                                                                      |
 | `--tesseractlanguage` | Tesseract language to use if multiple languages are available in the tesseract data directory.                                                   |
 | `--tesseractdata`     | Path of tesseract language data files, by default `tessdata` in the executable directory.                                                        |
+| `--tesseractversion`  | libtesseract version, support 4 and 5 (default: 4) (ignored on Windows platform)                                                                 |
 
 ## Example (Command Line)
 

diff --git a/src/PgsToSrt/CommandLineOptions.cs b/src/PgsToSrt/CommandLineOptions.cs
@@ -1,4 +1,4 @@
-using CommandLine;
+using CommandLine;
 
 namespace PgsToSrt
 {
@@ -21,5 +21,9 @@ internal class CommandLineOptions
 
         [Option(HelpText = "Path of tesseract language data files, by default 'tessdata' in the executable directory.")]
         public string TesseractData { get; set; }
+
+        [Option(HelpText = $"Tesseract version", Default = Runner.DefaultTesseractVersion)]
+        public string TesseractVersion { get; set; }
+
     }
 }
diff --git a/src/PgsToSrt/PgsOcr.cs b/src/PgsToSrt/PgsOcr.cs
@@ -1,4 +1,4 @@
-using Microsoft.Extensions.Logging;
+using Microsoft.Extensions.Logging;
 using Nikse.SubtitleEdit.Core;
 using Nikse.SubtitleEdit.Core.BluRaySup;
 using Nikse.SubtitleEdit.Core.SubtitleFormats;
@@ -16,14 +16,17 @@ public class PgsOcr
 {
     private readonly Microsoft.Extensions.Logging.ILogger _logger;
     private readonly Subtitle _subtitle = new Subtitle();
+    private readonly string _tesseractVersion;
 
     private List<BluRaySupParser.PcsData> _bluraySubtitles;
+
     public string TesseractDataPath { get; set; }
     public string TesseractLanguage { get; set; } = "eng";
 
-    public PgsOcr(Microsoft.Extensions.Logging.ILogger logger)
+    public PgsOcr(Microsoft.Extensions.Logging.ILogger logger,  string tesseractVersion)
     {
         _logger = logger;
+        _tesseractVersion = tesseractVersion;
     }
 
     public bool ToSrt(List<BluRaySupParser.PcsData> subtitles, string outputFileName)
@@ -57,8 +60,9 @@ private void Save(string outputFileName)
     private bool DoOcr()
     {
         _logger.LogInformation($"Starting OCR for {_bluraySubtitles.Count} items...");
+        _logger.LogInformation($"Tesseract verion {_tesseractVersion}");
 
-        var exception = TesseractApi.Initialize();
+        var exception = TesseractApi.Initialize(_tesseractVersion);
         if (exception != null)
         {
             _logger.LogError(exception, $"Failed: {exception.Message}");

diff --git a/src/PgsToSrt/PgsToSrt.csproj b/src/PgsToSrt/PgsToSrt.csproj
@@ -1,33 +1,37 @@
-<Project Sdk="Microsoft.NET.Sdk">
+<Project Sdk="Microsoft.NET.Sdk">
 
-  <PropertyGroup>
-    <OutputType>Exe</OutputType>
-    <TargetFrameworks>net5.0;net6.0</TargetFrameworks>
-    <AssemblyVersion>1.4.2.0</AssemblyVersion>
-    <FileVersion>1.4.2.0</FileVersion>
-    <Version>1.4.2</Version>
-  </PropertyGroup>
+    <PropertyGroup>
+        <OutputType>Exe</OutputType>
+        <TargetFrameworks>net5.0;net6.0</TargetFrameworks>
+        <AssemblyVersion>1.4.3.0</AssemblyVersion>
+        <FileVersion>1.4.3.0</FileVersion>
+        <Version>1.4.3</Version>
+    </PropertyGroup>
 
-  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
-    <DebugType>none</DebugType>
-    <DebugSymbols>false</DebugSymbols>
-  </PropertyGroup>
+    <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
+        <DebugType>none</DebugType>
+        <DebugSymbols>false</DebugSymbols>
+    </PropertyGroup>
 
- <ItemGroup>
-    <PackageReference Include="CommandLineParser" Version="2.9.0-preview1" />
-    <PackageReference Include="libse" Version="3.5.8" />
-    <PackageReference Include="Microsoft.Extensions.DependencyInjection" Version="5.0.0" />
-    <PackageReference Include="NLog" Version="4.7.5" />
-    <PackageReference Include="NLog.Extensions.Logging" Version="1.6.5" />
-    <PackageReference Include="SixLabors.ImageSharp.Drawing" Version="1.0.0-beta0010" />
-    <PackageReference Include="Tesseract" Version="4.1.1" />
-    <PackageReference Include="Tncl.NativeLoader" Version="1.0.6" />
-  </ItemGroup>
+    <PropertyGroup>
+        <LangVersion>preview</LangVersion>
+    </PropertyGroup>
 
-  <ItemGroup>
-    <None Update="NLog.config">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
-    </None>
-  </ItemGroup>
+    <ItemGroup>
+        <PackageReference Include="CommandLineParser" Version="2.9.0-preview1" />
+        <PackageReference Include="libse" Version="3.5.8" />
+        <PackageReference Include="Microsoft.Extensions.DependencyInjection" Version="5.0.0" />
+        <PackageReference Include="NLog" Version="4.7.5" />
+        <PackageReference Include="NLog.Extensions.Logging" Version="1.6.5" />
+        <PackageReference Include="SixLabors.ImageSharp.Drawing" Version="1.0.0-beta0010" />
+        <PackageReference Include="Tesseract" Version="5.2.0" />
+        <PackageReference Include="Tncl.NativeLoader" Version="1.0.6" />
+    </ItemGroup>
+
+    <ItemGroup>
+        <None Update="NLog.config">
+            <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+        </None>
+    </ItemGroup>
 
 </Project>
diff --git a/src/PgsToSrt/Runner.cs b/src/PgsToSrt/Runner.cs
@@ -1,18 +1,24 @@
-using CommandLine;
+using CommandLine;
 using Microsoft.Extensions.Logging;
 using PgsToSrt.Options;
 using System;
 using System.Collections.Generic;
 using System.IO;
+using System.Linq;
+using System.Runtime.InteropServices;
 
 namespace PgsToSrt
 {
     internal class Runner
     {
+        public const string DefaultTesseractVersion = "4";
+
+        private readonly string[] _tesseractSupportedVersions = new[] { "4", "5" };
         private readonly ILogger _logger;
 
         private string _tesseractData;
         private string _tesseractLanguage;
+        private string _tesseractVersion = DefaultTesseractVersion;
 
         public Runner(ILogger<Runner> logger)
         {
@@ -44,6 +50,26 @@ public void Run(Parsed<CommandLineOptions> values)
             var trackLanguage = values.Value.TrackLanguage;
             var track = values.Value.Track;
 
+            // Windows uses tesseract50.dll installed by nuget package, so always use v5
+            // Other systems can uses different libtesseract versions, keep v4 as default.
+            if (!RuntimeInformation.IsOSPlatform(OSPlatform.Windows))
+            {
+                if (values.Value.TesseractVersion != null)
+                {
+                    _tesseractVersion = values.Value.TesseractVersion;
+
+                    if (!_tesseractSupportedVersions.Contains(_tesseractVersion))
+                    {
+                        _logger.LogError($"Unsupported Tesseract version '{_tesseractVersion}' (Supported versions: 4, 5)");
+                        result = false;
+                    }
+                }
+            }
+            else
+            {
+                _tesseractVersion = "5";
+            }
+
             _tesseractData = !string.IsNullOrEmpty(values.Value.TesseractData)
                 ? values.Value.TesseractData
                 : Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "tessdata");
@@ -107,7 +133,7 @@ private bool ConvertPgs(string input, int? track, string output)
             if (subtitles is null)
                 return false;
 
-            var pgsOcr = new PgsOcr(_logger)
+            var pgsOcr = new PgsOcr(_logger, _tesseractVersion)
             {
                 TesseractDataPath = _tesseractData,
                 TesseractLanguage = _tesseractLanguage
@@ -116,6 +142,6 @@ private bool ConvertPgs(string input, int? track, string output)
             pgsOcr.ToSrt(subtitles, output);
 
             return true;
-        }             
+        }
     }
 }
diff --git a/src/PgsToSrt/TesseractApi.cs b/src/PgsToSrt/TesseractApi.cs
@@ -1,4 +1,4 @@
-using System;
+using System;
 using System.Collections.Generic;
 using System.Linq;
 using System.Reflection;
@@ -13,15 +13,15 @@ static class TesseractApi
     {
         private const string _assemblyName = "Assembly.Tesseract";
 
-        public static Exception Initialize()
+        public static Exception Initialize(string tesseractVersion)
         {
             Exception exception = null;
 
             var tessApiType = typeof(Tesseract.Page).Assembly.GetType("Tesseract.Interop.TessApi");
             var leptApiType = typeof(Tesseract.Page).Assembly.GetType("Tesseract.Interop.LeptonicaApi");
 
-            var tessApiCustomType = CreateInterfaceType<ITessApiSignatures>("tesseract41", "tesseract", "4");
-            var leptApiCustomType = CreateInterfaceType<ILeptonicaApiSignatures>("leptonica-1.80.0", "lept", "5");
+            var tessApiCustomType = CreateInterfaceType<ITessApiSignatures>("tesseract50", "tesseract", tesseractVersion);
+            var leptApiCustomType = CreateInterfaceType<ILeptonicaApiSignatures>("leptonica-1.82.0", "lept", "5");
 
             var loader = new NativeLoader();
             loader.WindowsOptions.UseSetDllDirectory = true;