Added transcription of audio and video files to local data sources (#1081)

Co-authored-by: Paul Koudelka <106623909+PaulKoudelka@users.noreply.github.com>

app/Tests/Tools/Media/TranscriptCleanupTests.cs

@@ -0,0 +1,110 @@
+using AIStudio.Tools.Media;
+using AIStudio.Tools.Services;
+using AIStudio.Tools.Services.Indexing;
+
+using Microsoft.Extensions.Logging.Abstractions;
+
+namespace AIStudio.Tests.Tools.Media;
+
+/// <summary>
+/// Checks that the transcript of a recording leaves the data source together with the recording.
+/// </summary>
+/// <remarks>
+/// A transcript holds what was said. Once its recording is gone from the data source, keeping it
+/// below the data directory would keep that text, unseen by the user.
+/// </remarks>
+[TestFixture]
+public sealed class TranscriptCleanupTests
+{
+    private static readonly string KEY_A = new('a', 64);
+
+    private static readonly string KEY_B = new('b', 64);
+
+    private static readonly string KEY_C = new('c', 64);
+
+    private string dataDirectory = string.Empty;
+
+    private DataSourceTranscriptStore store = null!;
+
+    [SetUp]
+    public async Task CreateTranscripts()
+    {
+        this.dataDirectory = Path.Combine(Path.GetTempPath(), $"ai-studio-transcript-cleanup-{Guid.NewGuid():N}");
+        this.store = new DataSourceTranscriptStore(this.dataDirectory, Guid.NewGuid().ToString());
+        foreach (var key in new[] { KEY_A, KEY_B, KEY_C })
+        {
+            var path = this.store.GetPath(key);
+            Directory.CreateDirectory(Path.GetDirectoryName(path)!);
+            await File.WriteAllTextAsync(path, $"Transcript {key[0]}");
+        }
+    }
+
+    [TearDown]
+    public void DeleteTranscripts()
+    {
+        if (Directory.Exists(this.dataDirectory))
+            Directory.Delete(this.dataDirectory, true);
+    }
+
+    [Test]
+    public void TheTranscriptOfARemovedRecordingIsDeleted()
+    {
+        var manifest = Manifest(("/data/kept.mp3", KEY_A));
+
+        FileSourceIndexer.ForgetTranscript(this.store, manifest, KEY_B, NullLogger.Instance);
+
+        Assert.That(this.store.ListKeys(), Is.EquivalentTo(new[] { KEY_A, KEY_C }));
+    }
+
+    [Test]
+    public void ATranscriptWhichAnotherRecordingSharesIsKept()
+    {
+        var manifest = Manifest(("/data/copy-of-the-removed-recording.mp3", KEY_B));
+
+        FileSourceIndexer.ForgetTranscript(this.store, manifest, KEY_B, NullLogger.Instance);
+
+        Assert.That(this.store.ListKeys(), Does.Contain(KEY_B), "Two recordings with the same content share one transcript.");
+    }
+
+    [Test]
+    public void AFileWithoutTranscriptDeletesNothing()
+    {
+        FileSourceIndexer.ForgetTranscript(this.store, new DataSourceEmbeddingManifest(), null, NullLogger.Instance);
+        Assert.That(this.store.ListKeys(), Has.Count.EqualTo(3));
+    }
+
+    [Test]
+    public async Task ACompleteRunDeletesWhatNoFileUses()
+    {
+        var temporaryPath = $"{this.store.GetPath(KEY_A)}.{Guid.NewGuid():N}.tmp";
+        await File.WriteAllTextAsync(temporaryPath, "Interrupted");
+        var manifest = Manifest(("/data/meeting.mp3", KEY_A), ("/data/report.pdf", null));
+
+        FileSourceIndexer.DeleteUnusedTranscripts(this.store, manifest, NullLogger.Instance);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(this.store.ListKeys(), Is.EqualTo(new[] { KEY_A }));
+            Assert.That(File.Exists(temporaryPath), Is.False, "A temporary file of an interrupted write is no transcript of anything.");
+        });
+    }
+
+    [Test]
+    public void WithoutAStoreNothingHappens()
+    {
+        Assert.DoesNotThrow(() =>
+        {
+            FileSourceIndexer.DeleteUnusedTranscripts(null, new DataSourceEmbeddingManifest(), NullLogger.Instance);
+            FileSourceIndexer.ForgetTranscript(null, new DataSourceEmbeddingManifest(), KEY_A, NullLogger.Instance);
+        });
+    }
+
+    private static DataSourceEmbeddingManifest Manifest(params (string Path, string? TranscriptKey)[] files)
+    {
+        var manifest = new DataSourceEmbeddingManifest();
+        foreach (var (path, transcriptKey) in files)
+            manifest.Files[path] = new EmbeddedFileRecord("fingerprint", 16, DateTimeOffset.UtcNow, DateTimeOffset.UtcNow, 1, transcriptKey);
+
+        return manifest;
+    }
+}