Fixed the Batch Processing assistant with an output folder above the input folder (#1098)

app/Tests/Assistants/BatchProcessingOutputFolderTests.cs

@@ -0,0 +1,165 @@
+using AIStudio.Assistants.BatchProcessing;
+
+namespace AIStudio.Tests.Assistants;
+
+/// <summary>
+/// Checks which files the search of the batch processing assistant leaves out because a batch run
+/// wrote them.
+/// </summary>
+/// <remarks>
+/// A run writes everything directly into its output folder. The search comes across any of it only
+/// when that folder lies below the input folder or is the input folder itself. An output folder
+/// above the input folder once made the search leave out every document.
+/// </remarks>
+[TestFixture]
+public sealed class BatchProcessingOutputFolderTests
+{
+    private const string RESULTS_FILE_NAME = "batch-results.csv";
+
+    private static readonly string DESKTOP = Path.Combine(Path.GetTempPath(), "ai-studio-batch-output-folder", "Desktop");
+    private static readonly string PROJECT_FOLDER = Path.Combine(DESKTOP, "Batch");
+    private static readonly string INPUT_FOLDER = Path.Combine(PROJECT_FOLDER, "Documents");
+
+    [Test]
+    public void AnOutputFolderAboveTheInputFolderLeavesOutNoDocument()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, PROJECT_FOLDER, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.pdf")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "notes", "minutes.md")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "log.csv")), Is.False, "The log of a run lies in the output folder, so a file of this name in the input folder is a document.");
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "notes")), Is.True);
+        });
+    }
+
+    [Test]
+    public void AnOutputFolderSeveralLevelsAboveTheInputFolderLeavesOutNoDocument()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, DESKTOP, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.pdf")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "notes", "minutes.md")), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "notes")), Is.True);
+        });
+    }
+
+    [Test]
+    public void TheDefaultOutputFolderBelowTheInputFolderIsLeftOut()
+    {
+        var resultsFolder = Path.Combine(INPUT_FOLDER, "ai-results");
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, resultsFolder, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(resultsFolder, "report_result.md")), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.pdf")), Is.False);
+            Assert.That(outputFolder.Enters(resultsFolder), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "notes")), Is.True);
+        });
+    }
+
+    [Test]
+    public void AnOutputFolderDeepBelowTheInputFolderIsLeftOutAlone()
+    {
+        var parentFolder = Path.Combine(INPUT_FOLDER, "2026", "october");
+        var resultsFolder = Path.Combine(parentFolder, "results");
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, resultsFolder, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(resultsFolder, "report_result.md")), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(parentFolder, "report.pdf")), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "2026")), Is.True);
+            Assert.That(outputFolder.Enters(parentFolder), Is.True);
+            Assert.That(outputFolder.Enters(resultsFolder), Is.False);
+        });
+    }
+
+    [Test]
+    public void ADocumentWhoseNameBeginsLikeTheOutputFolderIsNoOutput()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, Path.Combine(INPUT_FOLDER, "out"), RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "output-notes.md")), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "output")), Is.True);
+        });
+    }
+
+    [Test]
+    public void AnOutputFolderNextToTheInputFolderWithTheSameBeginningLeavesOutNothing()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, $"{INPUT_FOLDER}-results", RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.pdf")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "log.csv")), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "notes")), Is.True);
+        });
+    }
+
+    [TestCase(false)]
+    [TestCase(true)]
+    public void TheInputFolderAsOutputFolderLeavesOutTheLogAndTheResultsTable(bool withTrailingSeparator)
+    {
+        var outputDirectory = withTrailingSeparator ? INPUT_FOLDER + Path.DirectorySeparatorChar : INPUT_FOLDER;
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, outputDirectory, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "log.csv")), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, RESULTS_FILE_NAME)), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.pdf")), Is.False);
+            Assert.That(outputFolder.Enters(Path.Combine(INPUT_FOLDER, "notes")), Is.True, "Leaving out the output folder would leave out the input folder, too.");
+        });
+    }
+
+    [Test]
+    public void TheInputFolderAsOutputFolderLeavesOutTheResultFilesOfEveryFormat()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, INPUT_FOLDER, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report_result.md")), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report_result_2.docx")), Is.True, "A second result of the same name gets a counter.");
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "Report_Result.HTML")), Is.True);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.v2_result.md")), Is.True, "Only the last extension of the document makes way for the suffix.");
+        });
+    }
+
+    [Test]
+    public void TheInputFolderAsOutputFolderKeepsDocumentsWithSimilarNames()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, INPUT_FOLDER, RESULTS_FILE_NAME);
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report.md")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "result.md")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "my_results.md")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report_result_draft.md")), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(INPUT_FOLDER, "report_result_.md")), Is.False);
+        });
+    }
+
+    [Test]
+    public void TheInputFolderAsOutputFolderKeepsFilesInSubfolders()
+    {
+        var outputFolder = new BatchProcessingOutputFolder(INPUT_FOLDER, INPUT_FOLDER, RESULTS_FILE_NAME);
+        var subfolder = Path.Combine(INPUT_FOLDER, "notes");
+
+        Assert.Multiple(() =>
+        {
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(subfolder, "log.csv")), Is.False, "A run writes into the output folder itself, never into one of its subfolders.");
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(subfolder, RESULTS_FILE_NAME)), Is.False);
+            Assert.That(outputFolder.IsOutputOfRun(Path.Combine(subfolder, "report_result.md")), Is.False);
+        });
+    }
+}