adjusted namespaces and made separate data dir

Ocr/Ocr.Tesseract/Extensions/PageExtensions.cs

@@ -2,44 +2,43 @@
 using System.Collections.Generic;
 using Tesseract;
 
-namespace Ocr.Tesseract.Extensions
+namespace Ocr.Tesseract.Extensions;
+
+/// <summary>
+/// Extensions for the <see cref="Page"/> type
+/// </summary>
+public static class PageExtensions
 {
+
+
   /// <summary>
-  /// Extensions for the <see cref="Page"/> type
+  /// Retrieves the <see cref="Word"/>s in the given <paramref name="page"/>
   /// </summary>
-  public static class PageExtensions
+  /// <param name="page">The <see cref="Page"/> to extract words from</param>
+  /// <returns>The extracted <see cref="Word"/>s in the given <paramref name="page"/></returns>
+  public static IEnumerable<Word> GetWords(this Page page)
   {
+    using var iterator = page.GetIterator();
+    iterator.Begin();
 
-
-    /// <summary>
-    /// Retrieves the <see cref="Word"/>s in the given <paramref name="page"/>
-    /// </summary>
-    /// <param name="page">The <see cref="Page"/> to extract words from</param>
-    /// <returns>The extracted <see cref="Word"/>s in the given <paramref name="page"/></returns>
-    public static IEnumerable<Word> GetWords(this Page page)
+    do
     {
-      using var iterator = page.GetIterator();
-      iterator.Begin();
-
       do
       {
         do
         {
           do
           {
-            do
+            var word = Word.Parse(iterator);
+            if (string.IsNullOrEmpty(word.Text))
             {
-              var word = Word.Parse(iterator);
-              if (string.IsNullOrEmpty(word.Text))
-              {
-                continue;
-              }
+              continue;
+            }
 
-              yield return word;
-            } while (iterator.Next(PageIteratorLevel.TextLine, PageIteratorLevel.Word));
-          } while (iterator.Next(PageIteratorLevel.Para, PageIteratorLevel.TextLine));
-        } while (iterator.Next(PageIteratorLevel.Block, PageIteratorLevel.Para));
-      } while (iterator.Next(PageIteratorLevel.Block));
-    }
+            yield return word;
+          } while (iterator.Next(PageIteratorLevel.TextLine, PageIteratorLevel.Word));
+        } while (iterator.Next(PageIteratorLevel.Para, PageIteratorLevel.TextLine));
+      } while (iterator.Next(PageIteratorLevel.Block, PageIteratorLevel.Para));
+    } while (iterator.Next(PageIteratorLevel.Block));
   }
 }

Ocr/Ocr.Tesseract/TesseractProcessor.cs

@@ -7,61 +7,60 @@ using System.Collections.Generic;
 using System.Linq;
 using Tesseract;
 
-namespace Ocr.Tesseract
+namespace Ocr.Tesseract;
+
+/// <summary>
+/// Scans <see cref="MagickImage"/>s for <see cref="Word"/>s
+/// and maps the results to a <see cref="ScanResult"/>
+/// </summary>
+public class TesseractProcessor : Processor<MagickImage, ScanResult>
 {
-  /// <summary>
-  /// Scans <see cref="MagickImage"/>s for <see cref="Word"/>s
-  /// and maps the results to a <see cref="ScanResult"/>
-  /// </summary>
-  public class TesseractProcessor : Processor<MagickImage, ScanResult>
+  /// <inheritdoc cref="ITesseractConfiguration"/>
+  public ITesseractConfiguration Configuration { get; }
+
+  /// <inheritdoc />
+  public TesseractProcessor(ITesseractConfiguration config)
   {
-    /// <inheritdoc cref="ITesseractConfiguration"/>
-    public ITesseractConfiguration Configuration { get; }
+    Configuration = config;
+  }
 
-    /// <inheritdoc />
-    public TesseractProcessor(ITesseractConfiguration config)
-    {
-      Configuration = config;
-    }
-
-    /// <summary>
-    /// Scans the provided <paramref name="image"/> for <see cref="Word"/>s
-    /// </summary>
-    /// <param name="image">The <see cref="MagickImage"/> to scan</param>
-    /// <returns>
-    /// A list of <see cref="Word"/>s found
-    /// in the provided <paramref name="image"/>
-    /// </returns>
-    private IEnumerable<Word> Scan(MagickImage image)
-    {
-      // Convert image
-      using var pix = PixConverter.ToPix(image.ToBitmapWithDensity());
-      using var engine = new TesseractEngine(
-        Configuration.DataPath,
-        string.Join('+', Configuration.Languages),
-        EngineMode.Default,
-        Enumerable.Empty<string>(),
-        Configuration.Variables,
-        false
-      )
-      {
-        DefaultPageSegMode = PageSegMode.AutoOsd
-      };
-
-      // Scan
-      return engine
-        .Process(pix)
-        .GetWords()
-        .ToArray();
-    }
-
-    /// <inheritdoc />
-    public override IEnumerable<ScanResult> Process(
-      IEnumerable<MagickImage> inputs
+  /// <summary>
+  /// Scans the provided <paramref name="image"/> for <see cref="Word"/>s
+  /// </summary>
+  /// <param name="image">The <see cref="MagickImage"/> to scan</param>
+  /// <returns>
+  /// A list of <see cref="Word"/>s found
+  /// in the provided <paramref name="image"/>
+  /// </returns>
+  private IEnumerable<Word> Scan(MagickImage image)
+  {
+    // Convert image
+    using var pix = PixConverter.ToPix(image.ToBitmapWithDensity());
+    using var engine = new TesseractEngine(
+      Configuration.DataPath,
+      string.Join('+', Configuration.Languages),
+      EngineMode.Default,
+      Enumerable.Empty<string>(),
+      Configuration.Variables,
+      false
     )
     {
-      return inputs
-        .SelectMany(Scan, (input, word) => new ScanResult(word, input));
-    }
+      DefaultPageSegMode = PageSegMode.AutoOsd
+    };
+
+    // Scan
+    return engine
+      .Process(pix)
+      .GetWords()
+      .ToArray();
   }
-}
+
+  /// <inheritdoc />
+  public override IEnumerable<ScanResult> Process(
+    IEnumerable<MagickImage> inputs
+  )
+  {
+    return inputs
+      .SelectMany(Scan, (input, word) => new ScanResult(word, input));
+  }
+}