Merge pull request #5 from ANcpLua/fix/capture-tesseract-stderr

Capture Tesseract stderr in OCR-failure exception message

Author: ANcpLua

CreatePdf.NET.Tests/TesseractOcrProviderTests.cs

@@ -11,7 +11,7 @@ public class TesseractOcrProviderTests
     [Fact]
     public async Task ExtractTextFromImageAsync_WhenOutputIsMissing_ThrowsFileNotFound()
     {
-        var processRunner = new FakeProcessRunner();
+        var processRunner = new FakeProcessRunner { NextResult = new ProcessResult(1, "", "") };
         var environment = new FakeSystemEnvironment { FileExistsImpl = _ => false };
         var engine = new TesseractOcrProvider(environment, processRunner);
 
@@ -21,14 +21,36 @@ public async Task ExtractTextFromImageAsync_WhenOutputIsMissing_ThrowsFileNotFou
             new OcrOptions { TesseractPath = "/bin/echo" });
 
         await act.Should().ThrowAsync<FileNotFoundException>()
-            .WithMessage("OCR output file not found. Tesseract execution failed*")
+            .WithMessage("OCR output file not found (Tesseract exited with code 1)*")
             .ConfigureAwait(true);
 
         processRunner.StartInfos.Should().HaveCount(1);
         processRunner.StartInfos[0].FileName.Should().Be("/bin/echo");
         processRunner.StartInfos[0].Arguments.Should().Contain("input.png");
     }
 
+    [Fact]
+    public async Task ExtractTextFromImageAsync_WhenOutputIsMissing_IncludesStderrInExceptionMessage()
+    {
+        const string tesseractStderr = "Error opening data file /usr/share/tessdata/eng.traineddata";
+        var processRunner = new FakeProcessRunner
+        {
+            NextResult = new ProcessResult(ExitCode: 1, StandardOutput: "", StandardError: tesseractStderr)
+        };
+        var environment = new FakeSystemEnvironment { FileExistsImpl = _ => false };
+        var engine = new TesseractOcrProvider(environment, processRunner);
+
+        var act = () => engine.ExtractTextFromImageAsync(
+            "input.png",
+            Path.Combine(Path.GetTempPath(), "missing-output.txt"),
+            new OcrOptions { TesseractPath = "/bin/echo" });
+
+        await act.Should().ThrowAsync<FileNotFoundException>()
+            .Where(e => e.Message.Contains(tesseractStderr, StringComparison.Ordinal)
+                        && e.Message.Contains("exited with code 1", StringComparison.Ordinal))
+            .ConfigureAwait(true);
+    }
+
     [Theory]
     [InlineData("out.log", "out.txt")]
     [InlineData("out", "out.txt")]
@@ -207,10 +229,12 @@ private sealed class FakeProcessRunner : IProcessRunner
     {
         public List<ProcessStartInfo> StartInfos { get; } = [];
 
-        public Task RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken)
+        public ProcessResult NextResult { get; set; } = new(ExitCode: 0, StandardOutput: "", StandardError: "");
+
+        public Task<ProcessResult> RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken)
         {
             StartInfos.Add(startInfo);
-            return Task.CompletedTask;
+            return Task.FromResult(NextResult);
         }
     }
 }

CreatePdf.NET/Internal/IProcessRunner.cs

@@ -4,5 +4,5 @@ namespace CreatePdf.NET.Internal;
 
 internal interface IProcessRunner
 {
-    Task RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken);
+    Task<ProcessResult> RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken);
 }

CreatePdf.NET/Internal/ProcessResult.cs

@@ -0,0 +1,3 @@
+namespace CreatePdf.NET.Internal;
+
+internal readonly record struct ProcessResult(int ExitCode, string StandardOutput, string StandardError);

CreatePdf.NET/Internal/ProcessRunner.cs

@@ -6,13 +6,22 @@ internal sealed class ProcessRunner : IProcessRunner
 {
     public static ProcessRunner Instance { get; } = new();
 
-    public async Task RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken)
+    public async Task<ProcessResult> RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken)
     {
         ArgumentNullException.ThrowIfNull(startInfo);
 
         using var process = Process.Start(startInfo)
             ?? throw new InvalidOperationException(
                 $"Process.Start returned null for '{startInfo.FileName}' — no new process was created.");
+
+        // Read both streams concurrently to avoid the classic "child process blocks on a full pipe" deadlock.
+        var stdoutTask = process.StandardOutput.ReadToEndAsync(cancellationToken);
+        var stderrTask = process.StandardError.ReadToEndAsync(cancellationToken);
+
         await process.WaitForExitAsync(cancellationToken).ConfigureAwait(false);
+        var stdout = await stdoutTask.ConfigureAwait(false);
+        var stderr = await stderrTask.ConfigureAwait(false);
+
+        return new ProcessResult(process.ExitCode, stdout, stderr);
     }
 }

CreatePdf.NET/Internal/TesseractOcrProvider.cs

@@ -56,15 +56,21 @@ public async Task<string> ExtractTextFromImageAsync(string pngPath, string txtPa
             Path.GetFileNameWithoutExtension(txtPath));
         var actualTxtPath = outputBase + ".txt";
 
-        await _processRunner.RunAsync(
+        var result = await _processRunner.RunAsync(
                 CreateProcessInfo(
                     GetTesseractExecutable(options),
                     GetOcrArguments(pngPath, outputBase, options)),
                 cancellationToken)
             .ConfigureAwait(false);
 
         if (!_systemEnvironment.FileExists(actualTxtPath))
-            throw new FileNotFoundException("OCR output file not found. Tesseract execution failed.", actualTxtPath);
+        {
+            var exitCode = result.ExitCode.ToString(CultureInfo.InvariantCulture);
+            var message = string.IsNullOrWhiteSpace(result.StandardError)
+                ? $"OCR output file not found (Tesseract exited with code {exitCode})."
+                : $"OCR output file not found (Tesseract exited with code {exitCode}): {result.StandardError.Trim()}";
+            throw new FileNotFoundException(message, actualTxtPath);
+        }
 
         var text = await File.ReadAllTextAsync(actualTxtPath, cancellationToken).ConfigureAwait(false);
         return text.Trim().Replace("\n", " ").Replace("\r", " ");