Renamed Examples to Implementation

Implementation/ReportGenerator/Models/ImageStats.cs

@@ -0,0 +1,27 @@
+namespace Ocr.Report.Models;
+
+public readonly struct ImageStats
+{
+  public string ImageName { get; }
+
+  public ICollection<string> Reference { get; }
+
+  public ICollection<ProcessorStat> Processors { get; }
+
+  public ImageStats(
+    string imageName,
+    ICollection<string> taggedWords,
+    IEnumerable<ScanFileInfo> scanResult
+  )
+  {
+    Reference = taggedWords;
+    ImageName = imageName;
+    Processors = scanResult
+      .Select(t =>
+      {
+        var (elapsed, words) = t.GetData();
+        return new ProcessorStat(t.ProcessorName, elapsed, taggedWords, words);
+      })
+      .ToArray();
+  }
+}

Implementation/ReportGenerator/Models/ProcessorStat.cs

@@ -0,0 +1,85 @@
+using Common.Distance;
+
+namespace Ocr.Report.Models;
+
+public readonly struct ProcessorStat : IDistanceComparer<IEnumerable<string>>
+{
+  /// <summary>
+  /// The name of the processor
+  /// </summary>
+  public string Name { get; }
+
+  /// <summary>
+  /// The total milliseconds it took the processor to process the data
+  /// </summary>
+  public double ProcessingTime { get; }
+
+  /// <inheritdoc />
+  public IEnumerable<string> Reference { get; }
+
+  /// <inheritdoc />
+  public IEnumerable<string>? Hypothesis { get; }
+
+  /// <inheritdoc />
+  public double Distance { get; }
+
+  /// <summary>
+  /// Information about the processed values
+  /// </summary>
+  public ICollection<IDistanceComparer<string>> Words { get; }
+
+  public ProcessorStat(
+    string name,
+    double processingTime,
+    ICollection<string> reference,
+    ICollection<string> hypothesis
+  )
+  {
+    Name = name;
+    ProcessingTime = processingTime;
+
+    Reference = reference;
+    Hypothesis = hypothesis;
+
+    Distance = Calculator.GetDistance(
+      reference.OrderBy(s => s).ToArray(),
+      hypothesis.OrderBy(s => s).ToArray()
+    ) / reference.Count;
+
+    Words = GetDistanceInfos(reference, hypothesis).ToArray();
+    // Words = reference.Select(r => GetDistanceInfo(r, hypothesis)).ToArray();
+  }
+
+
+  /// <summary>
+  /// Compares the <paramref name="referenceCollection"/> with all given values in the
+  /// <paramref name="hypothesisCollection"/>, determining the <see cref="IDistanceComparer{T}"/>
+  /// with the lowest error
+  /// </summary>
+  private static IEnumerable<IDistanceComparer<string>> GetDistanceInfos(
+    ICollection<string> referenceCollection,
+    ICollection<string> hypothesisCollection
+  )
+  {
+    var results = new List<IDistanceComparer<string>>();
+
+    foreach (var reference in referenceCollection)
+    {
+      var tResults = hypothesisCollection
+        .Select(hypothesis => new DistanceComparer<string>(reference, hypothesis))
+        .Cast<IDistanceComparer<string>>()
+        .ToList();
+
+      results.AddRange(tResults);
+    }
+
+    var lookup = results
+      .OrderBy(result => result.Distance)
+      .ToLookup(result => result.Reference);
+
+    foreach (var reference in referenceCollection)
+    {
+      yield return lookup[reference].FirstOrDefault() ?? new DistanceComparer<string>(reference);
+    }
+  }
+}

Implementation/ReportGenerator/Models/ScanFileInfo.cs

@@ -0,0 +1,48 @@
+using System.Text.Json;
+using System.Text.RegularExpressions;
+
+namespace Ocr.Report.Models;
+
+public struct ScanFileInfo
+{
+  public string Path { get; private init; }
+
+  private static readonly Regex parseRegex = new(
+    @"(?'image'.+)\.(?'processor'.+)\..+",
+    RegexOptions.Compiled | RegexOptions.CultureInvariant | RegexOptions.IgnoreCase
+  );
+
+  public string ProcessorName { get; set; }
+
+  public string ImageName { get; set; }
+
+  public (double Elapsed, ICollection<string> Words) GetData()
+  {
+    using var file = File.OpenRead(Path);
+    var root = JsonDocument.Parse(file).RootElement;
+
+    var words = root
+      .GetProperty("Words")
+      .EnumerateArray()
+      .Select(e => e.GetProperty("Text").GetString() ?? string.Empty)
+      .ToArray();
+
+    var elapsed = root.GetProperty("Elapsed").GetDouble();
+
+    return (elapsed, words);
+  }
+
+  public static ScanFileInfo FromPath(string path)
+  {
+    var match = parseRegex.Match(System.IO.Path.GetFileName(path));
+    return new ScanFileInfo
+    {
+      Path = path,
+      ProcessorName = match.Groups["processor"].Value,
+      ImageName = match.Groups["image"].Value
+    };
+  }
+
+  /// <inheritdoc />
+  public override string ToString() => ImageName;
+}

Implementation/ReportGenerator/Models/TagFileInfo.cs

@@ -0,0 +1,31 @@
+using System.Text.Json;
+
+namespace Ocr.Report.Models;
+
+internal struct TagFileInfo
+{
+  public string Path { get; private init; }
+
+  public string ImageName { get; set; }
+
+  public ICollection<string> GetWords()
+  {
+    using var file = File.OpenRead(Path);
+    return JsonDocument
+      .Parse(file)
+      .RootElement
+      .GetProperty("words")
+      .EnumerateArray()
+      .Select(w => w.GetString() ?? throw new Exception("Cannot parse null words"))
+      .ToArray();
+  }
+
+  public static TagFileInfo FromPath(string path) => new()
+  {
+    Path = path,
+    ImageName = System.IO.Path.GetFileNameWithoutExtension(path),
+  };
+
+  /// <inheritdoc />
+  public override string ToString() => ImageName;
+}