|
| 1 | +using System.Diagnostics; |
| 2 | +using CreatePdf.NET.Internal; |
| 3 | + |
| 4 | +namespace CreatePdf.NET.Tests; |
| 5 | + |
| 6 | +public class TesseractOcrEngineTests |
| 7 | +{ |
| 8 | + private const string AppleSiliconPath = "/opt/homebrew/bin/tesseract"; |
| 9 | + private const string IntelMacPath = "/usr/local/bin/tesseract"; |
| 10 | + |
| 11 | + [Fact] |
| 12 | + public async Task ExtractTextFromImageAsync_WhenOutputIsMissing_ThrowsFileNotFound() |
| 13 | + { |
| 14 | + var processRunner = new FakeProcessRunner(); |
| 15 | + var environment = new FakeSystemEnvironment { FileExistsImpl = _ => false }; |
| 16 | + var engine = new TesseractOcrEngine(environment, processRunner); |
| 17 | + |
| 18 | + var act = () => engine.ExtractTextFromImageAsync( |
| 19 | + "input.png", |
| 20 | + Path.Combine(Path.GetTempPath(), "missing-output.txt"), |
| 21 | + new OcrOptions { TesseractPath = "/bin/echo" }); |
| 22 | + |
| 23 | + await act.Should().ThrowAsync<FileNotFoundException>() |
| 24 | + .WithMessage("OCR output file not found. Tesseract execution failed*") |
| 25 | + .ConfigureAwait(true); |
| 26 | + |
| 27 | + processRunner.StartInfos.Should().HaveCount(1); |
| 28 | + processRunner.StartInfos[0].FileName.Should().Be("/bin/echo"); |
| 29 | + processRunner.StartInfos[0].Arguments.Should().Contain("input.png"); |
| 30 | + } |
| 31 | + |
| 32 | + [Fact] |
| 33 | + public void GetPdfRasterizerExecutable_UsesExplicitConverterPath() |
| 34 | + { |
| 35 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment(), new FakeProcessRunner()); |
| 36 | + var options = new OcrOptions { PdfConverterPath = "/custom/gs" }; |
| 37 | + |
| 38 | + engine.GetPdfRasterizerExecutable(options).Should().Be("/custom/gs"); |
| 39 | + } |
| 40 | + |
| 41 | + [Fact] |
| 42 | + public void GetPdfRasterizerExecutable_MacOs_ReturnsSips() |
| 43 | + { |
| 44 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment { IsMacOS = true }, new FakeProcessRunner()); |
| 45 | + |
| 46 | + engine.GetPdfRasterizerExecutable(new OcrOptions()).Should().Be("/usr/bin/sips"); |
| 47 | + } |
| 48 | + |
| 49 | + [Theory] |
| 50 | + [InlineData(true, "gswin64c")] |
| 51 | + [InlineData(false, "gswin32c")] |
| 52 | + public void GetPdfRasterizerExecutable_Windows_SelectsBitnessSpecificCommand(bool is64Bit, string expected) |
| 53 | + { |
| 54 | + var environment = new FakeSystemEnvironment { IsWindows = true, Is64BitOperatingSystem = is64Bit }; |
| 55 | + var engine = new TesseractOcrEngine(environment, new FakeProcessRunner()); |
| 56 | + |
| 57 | + engine.GetPdfRasterizerExecutable(new OcrOptions()).Should().Be(expected); |
| 58 | + } |
| 59 | + |
| 60 | + [Fact] |
| 61 | + public void GetPdfRasterizerExecutable_DefaultsToUnixCommand() |
| 62 | + { |
| 63 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment(), new FakeProcessRunner()); |
| 64 | + |
| 65 | + engine.GetPdfRasterizerExecutable(new OcrOptions()).Should().Be("gs"); |
| 66 | + } |
| 67 | + |
| 68 | + [Fact] |
| 69 | + public void GetTesseractExecutable_UsesProvidedPath() |
| 70 | + { |
| 71 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment(), new FakeProcessRunner()); |
| 72 | + var options = new OcrOptions { TesseractPath = "/custom/tesseract" }; |
| 73 | + |
| 74 | + engine.GetTesseractExecutable(options).Should().Be("/custom/tesseract"); |
| 75 | + } |
| 76 | + |
| 77 | + [Fact] |
| 78 | + public void GetTesseractExecutable_MacOsPrefersAppleSiliconBinary() |
| 79 | + { |
| 80 | + var environment = new FakeSystemEnvironment |
| 81 | + { |
| 82 | + IsMacOS = true, |
| 83 | + FileExistsImpl = path => string.Equals(path, AppleSiliconPath, StringComparison.Ordinal) |
| 84 | + }; |
| 85 | + var engine = new TesseractOcrEngine(environment, new FakeProcessRunner()); |
| 86 | + |
| 87 | + engine.GetTesseractExecutable(new OcrOptions()).Should().Be(AppleSiliconPath); |
| 88 | + } |
| 89 | + |
| 90 | + [Fact] |
| 91 | + public void GetTesseractExecutable_MacOsFallsBackToIntelBinary() |
| 92 | + { |
| 93 | + var environment = new FakeSystemEnvironment |
| 94 | + { |
| 95 | + IsMacOS = true, |
| 96 | + FileExistsImpl = path => string.Equals(path, IntelMacPath, StringComparison.Ordinal) |
| 97 | + }; |
| 98 | + var engine = new TesseractOcrEngine(environment, new FakeProcessRunner()); |
| 99 | + |
| 100 | + engine.GetTesseractExecutable(new OcrOptions()).Should().Be(IntelMacPath); |
| 101 | + } |
| 102 | + |
| 103 | + [Fact] |
| 104 | + public void GetTesseractExecutable_UsesFallbackWhenNoMacBinaryFound() |
| 105 | + { |
| 106 | + var engine = new TesseractOcrEngine( |
| 107 | + new FakeSystemEnvironment { IsMacOS = true }, |
| 108 | + new FakeProcessRunner()); |
| 109 | + |
| 110 | + engine.GetTesseractExecutable(new OcrOptions()).Should().Be("tesseract"); |
| 111 | + } |
| 112 | + |
| 113 | + [Fact] |
| 114 | + public void GetRasterizationArguments_MacOs_UsesSipsFormat() |
| 115 | + { |
| 116 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment { IsMacOS = true }, new FakeProcessRunner()); |
| 117 | + |
| 118 | + engine.GetRasterizationArguments("file.pdf", "file.png", new OcrOptions { Dpi = 150 }) |
| 119 | + .Should() |
| 120 | + .Be("-s format png -s dpiHeight 150 -s dpiWidth 150 \"file.pdf\" --out \"file.png\""); |
| 121 | + } |
| 122 | + |
| 123 | + [Fact] |
| 124 | + public void GetRasterizationArguments_NonMac_UsesGhostscriptFormat() |
| 125 | + { |
| 126 | + var engine = new TesseractOcrEngine(new FakeSystemEnvironment(), new FakeProcessRunner()); |
| 127 | + |
| 128 | + engine.GetRasterizationArguments("file.pdf", "file.png", new OcrOptions { Dpi = 200 }) |
| 129 | + .Should() |
| 130 | + .Be("-dNOPAUSE -dBATCH -sDEVICE=pngalpha -dFirstPage=1 -dLastPage=1 -r200 -sOutputFile=\"file.png\" \"file.pdf\" "); |
| 131 | + } |
| 132 | + |
| 133 | + private sealed class FakeSystemEnvironment : ISystemEnvironment |
| 134 | + { |
| 135 | + public bool IsMacOS { get; set; } |
| 136 | + |
| 137 | + public bool IsWindows { get; set; } |
| 138 | + |
| 139 | + public bool Is64BitOperatingSystem { get; set; } |
| 140 | + |
| 141 | + public Func<string, bool>? FileExistsImpl { get; set; } |
| 142 | + |
| 143 | + public bool FileExists(string path) => (FileExistsImpl ?? (_ => false)).Invoke(path); |
| 144 | + } |
| 145 | + |
| 146 | + private sealed class FakeProcessRunner : IProcessRunner |
| 147 | + { |
| 148 | + public List<ProcessStartInfo> StartInfos { get; } = []; |
| 149 | + |
| 150 | + public Task RunAsync(ProcessStartInfo startInfo, CancellationToken cancellationToken) |
| 151 | + { |
| 152 | + StartInfos.Add(startInfo); |
| 153 | + return Task.CompletedTask; |
| 154 | + } |
| 155 | + } |
| 156 | +} |
0 commit comments