Apply PR review feedback

- Trim redundant 2nd comment line in TesseractOcrProvider (the +".txt"
  re-attach is self-evident from the code).
- Drop duplicate WHAT comment in TesseractOcrProviderTests; the test name
  and variable names already state the same fact.
- Convert ExtractTextFromImageAsync_WhenOutputPathHasNonTxtExtension to a
  Theory with 5 InlineData rows: out.log, out (no ext), foo.bar.log
  (multi-dot), out.TXT (uppercase), out. (trailing dot). Closes the
  edge-case gaps in path derivation.
- Add ExtractTextFromImageAsync_WithNonTxtInputPath_ReadsFromDerivedTxtPath
  to cover the happy-path File.ReadAllTextAsync(actualTxtPath, ...) line
  that was previously only reachable via the throw branch.
- Refactor OcrService.TryDeleteDirectory with an optional
  Action<string,bool> deleteImpl parameter (defaults to Directory.Delete)
  so the catch-when filter can be tested deterministically and
  cross-platform without chmod / FileShare tricks.
- Add two tests that inject throwing delegates to cover IOException and
  UnauthorizedAccessException paths of the catch-when filter.

Coverage gaps flagged by review are now closed; 253 tests pass on net8.0,
net9.0, net10.0.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: ANcpLua

CreatePdf.NET.Tests/OcrServiceTests.cs

@@ -85,4 +85,40 @@ public void TryDeleteDirectory_NonExistentDirectory_DoesNotThrow()
 
         act.Should().NotThrow();
     }
+
+    [Fact]
+    public void TryDeleteDirectory_WhenDeleteThrowsIOException_DoesNotPropagate()
+    {
+        var tempDir = Directory.CreateTempSubdirectory("createpdf-ocr-test-");
+        try
+        {
+            var act = () => OcrService.TryDeleteDirectory(
+                tempDir.FullName,
+                (_, _) => throw new IOException("simulated lock"));
+
+            act.Should().NotThrow();
+        }
+        finally
+        {
+            Directory.Delete(tempDir.FullName, recursive: true);
+        }
+    }
+
+    [Fact]
+    public void TryDeleteDirectory_WhenDeleteThrowsUnauthorizedAccess_DoesNotPropagate()
+    {
+        var tempDir = Directory.CreateTempSubdirectory("createpdf-ocr-test-");
+        try
+        {
+            var act = () => OcrService.TryDeleteDirectory(
+                tempDir.FullName,
+                (_, _) => throw new UnauthorizedAccessException("simulated permission denied"));
+
+            act.Should().NotThrow();
+        }
+        finally
+        {
+            Directory.Delete(tempDir.FullName, recursive: true);
+        }
+    }
 }

CreatePdf.NET.Tests/TesseractOcrProviderTests.cs

@@ -29,11 +29,17 @@ await act.Should().ThrowAsync<FileNotFoundException>()
         processRunner.StartInfos[0].Arguments.Should().Contain("input.png");
     }
 
-    [Fact]
-    public async Task ExtractTextFromImageAsync_WhenOutputPathHasNonTxtExtension_ChecksDerivedTxtPath()
+    [Theory]
+    [InlineData("out.log", "out.txt")]
+    [InlineData("out", "out.txt")]
+    [InlineData("foo.bar.log", "foo.bar.txt")]
+    [InlineData("out.TXT", "out.txt")]
+    [InlineData("out.", "out.txt")]
+    public async Task ExtractTextFromImageAsync_DerivesActualTxtPathFromBase(string inputName, string expectedName)
     {
-        // Tesseract always appends ".txt" to the base it is given. If the caller hands us
-        // a path like "out.log", the provider must read back "out.txt", not "out.log".
+        var tempDir = Path.GetTempPath();
+        var inputPath = Path.Combine(tempDir, inputName);
+        var expectedPath = Path.Combine(tempDir, expectedName);
         var checkedPaths = new List<string>();
         var processRunner = new FakeProcessRunner();
         var environment = new FakeSystemEnvironment
@@ -46,20 +52,41 @@ public async Task ExtractTextFromImageAsync_WhenOutputPathHasNonTxtExtension_Che
         };
         var engine = new TesseractOcrProvider(environment, processRunner);
 
-        var tempDir = Path.GetTempPath();
-        var nonTxtPath = Path.Combine(tempDir, "out.log");
-        var expectedDerivedPath = Path.Combine(tempDir, "out.txt");
-
         var act = () => engine.ExtractTextFromImageAsync(
             "input.png",
-            nonTxtPath,
+            inputPath,
             new OcrOptions { TesseractPath = "/bin/echo" });
 
         await act.Should().ThrowAsync<FileNotFoundException>()
-            .Where(e => e.FileName == expectedDerivedPath)
+            .Where(e => e.FileName == expectedPath)
             .ConfigureAwait(true);
 
-        checkedPaths.Should().ContainSingle().Which.Should().Be(expectedDerivedPath);
+        checkedPaths.Should().ContainSingle().Which.Should().Be(expectedPath);
+    }
+
+    [Fact]
+    public async Task ExtractTextFromImageAsync_WithNonTxtInputPath_ReadsFromDerivedTxtPath()
+    {
+        var actualTxtPath = Path.Combine(Path.GetTempPath(), $"createpdf-test-{Guid.NewGuid():N}.txt");
+        var nonTxtInputPath = Path.ChangeExtension(actualTxtPath, ".log");
+        await File.WriteAllTextAsync(actualTxtPath, "Hello\nWorld").ConfigureAwait(true);
+        try
+        {
+            var processRunner = new FakeProcessRunner();
+            var environment = new FakeSystemEnvironment { FileExistsImpl = File.Exists };
+            var engine = new TesseractOcrProvider(environment, processRunner);
+
+            var result = await engine.ExtractTextFromImageAsync(
+                "input.png",
+                nonTxtInputPath,
+                new OcrOptions { TesseractPath = "/bin/echo" }).ConfigureAwait(true);
+
+            result.Should().Be("Hello World");
+        }
+        finally
+        {
+            if (File.Exists(actualTxtPath)) File.Delete(actualTxtPath);
+        }
     }
 
     [Fact]

CreatePdf.NET/Internal/OcrService.cs

@@ -62,12 +62,12 @@ private async Task<string> OcrAsync(string pdfPath, string workDir, OcrOptions o
             .ConfigureAwait(false);
     }
 
-    internal static void TryDeleteDirectory(string path)
+    internal static void TryDeleteDirectory(string path, Action<string, bool>? deleteImpl = null)
     {
         try
         {
-            if (Directory.Exists(path))
-                Directory.Delete(path, recursive: true);
+            if (!Directory.Exists(path)) return;
+            (deleteImpl ?? Directory.Delete)(path, true);
         }
         catch (Exception ex) when (ex is IOException or UnauthorizedAccessException)
         {

CreatePdf.NET/Internal/TesseractOcrProvider.cs

@@ -50,8 +50,7 @@ await _processRunner.RunAsync(
     public async Task<string> ExtractTextFromImageAsync(string pngPath, string txtPath, OcrOptions options,
         CancellationToken cancellationToken = default)
     {
-        // Tesseract appends ".txt" itself, so strip the extension to give it the base path
-        // and then re-attach .txt for the file we actually read back.
+        // Tesseract appends ".txt" itself, so strip the extension to give it the base path.
         var outputBase = Path.Combine(
             Path.GetDirectoryName(txtPath) ?? string.Empty,
             Path.GetFileNameWithoutExtension(txtPath));