Added oneshot apps

2024-01-10 23:33:58 +01:00
parent 946eea2347
commit 6eda25fce5
24 changed files with 192 additions and 0 deletions
@@ -0,0 +1,15 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+    <PropertyGroup>
+        <OutputType>Exe</OutputType>
+        <TargetFramework>net7.0</TargetFramework>
+        <ImplicitUsings>enable</ImplicitUsings>
+        <Nullable>enable</Nullable>
+    </PropertyGroup>
+
+    <ItemGroup>
+      <ProjectReference Include="..\..\Ocr\Ocr.Tesseract.Screenshots\Ocr.Tesseract.Screenshots.csproj" />
+      <ProjectReference Include="..\..\Ocr\Ocr.Tesseract\Ocr.Tesseract.csproj" />
+    </ItemGroup>
+
+</Project>
@@ -0,0 +1,51 @@
+// See https://aka.ms/new-console-template for more information
+
+using System.Text.RegularExpressions;
+using ImageMagick;
+using Ocr.Tesseract;
+using Ocr.Tesseract.Models;
+using Ocr.Tesseract.Screenshots;
+using Ocr.Tesseract.Screenshots.Configuration;
+using Ocr.Tesseract.Screenshots.Threshold;
+using Process.Abstract.Configuration;
+using Process.Interface;
+
+
+Console.WriteLine("Hello, World!");
+
+IProcessorChain<MagickImage, ScanResult> MakeProcessor()
+{
+    var wordRegex = new Regex(
+        @"[\w'\-]{2,}",
+        RegexOptions.Compiled | RegexOptions.Multiline | RegexOptions.IgnoreCase
+    );
+    
+    var tesseractConfig = new TesseractScreenshotConfiguration
+    {
+        DataPath = "tessdata",
+        Languages = new[] { "eng", "deu" }
+    };
+    
+    var preprocessing = new ProcessorChainConfiguration<MagickImage, MagickImage>()
+        .Use(new CloneImageProcessor())
+        .Use(new ResizeProcessor(FilterType.Lanczos2Sharp, PixelInterpolateMethod.Mesh))
+        .Use(new NormalizeProcessor())
+        .Use(new ThresholdAdaptiveProcessor(15, 15))
+        .Use(new AddBorderProcessor(10))
+        .Use(new BinarizeProcessor())
+        .Complete(new NegateCloneProcessor());
+
+    var postprocessing = new ProcessorChainConfiguration<ScanResult, ScanResult>()
+        .Use(new ConfidenceFilter(50))
+        // todo insert processing events and write to json files
+        .Use(new ToLowerProcessor())
+        .Use(new DuplicateFilter())
+        .Complete(new RegexFilter(wordRegex));
+
+    var scan = new TesseractProcessor(tesseractConfig);
+
+    return new ProcessorChainConfiguration<MagickImage, ScanResult>()
+        .Use(preprocessing)
+        .Use(scan)
+        .Complete(postprocessing);
+}