public void IgnorePixelDataLessThan(bool ignoreShortText)
{
var opts = new IsIdentifiableDicomFileOptions
{
ColumnReport = true,
TessDirectory = _tessDir.FullName,
};
// NOTE(rkm 2020-11-16) The test image should report 3 bits of text with lengths 123, 127, and 170.
// If we ignore less than 170 then only 1 bit of text should be reported.
opts.IgnoreTextLessThan = ignoreShortText ? 170 : 0U;
string fileName = Path.Combine(TestContext.CurrentContext.TestDirectory, nameof(DicomFileRunnerTest), "f1.dcm");
TestData.Create(new FileInfo(fileName), TestData.BURNED_IN_TEXT_IMG);
var runner = new DicomFileRunner(opts);
var fileSystem = new FileSystem();
IFileInfo fileInfo = fileSystem.FileInfo.FromFileName(fileName);
Assert.True(fileInfo.Exists);
var toMemory = new ToMemoryFailureReport();
runner.Reports.Add(toMemory);
runner.ValidateDicomFile(fileInfo);
List <Failure> failures = toMemory.Failures.ToList();
Assert.AreEqual(ignoreShortText ? 1 : 3, failures.Count);
}
private static int OnParse(GlobalOptions globals, object parsedOpts)
{
var opts = SmiCliInit.Verify <IsIdentifiableAbstractOptions>(parsedOpts);
return(opts switch
{
IsIdentifiableRelationalDatabaseOptions o => Run(o),
IsIdentifiableDicomFileOptions o => Run(o),
IsIdentifiableMongoOptions o => Run(o),
IsIdentifiableServiceOptions o => Run(o),
IsIdentifiableFileOptions o => Run(o),
_ => throw new NotImplementedException($"No case for '{opts.GetType()}'")
});
public DicomFileRunner(IsIdentifiableDicomFileOptions opts) : base(opts)
{
_opts = opts;
_ignoreTextLessThan = opts.IgnoreTextLessThan;
//if using Efferent.Native DICOM codecs
// (see https://github.com/Efferent-Health/Dicom-native)
//Dicom.Imaging.Codec.TranscoderManager.SetImplementation(new Efferent.Native.Codec.NativeTranscoderManager());
//OR if using fo-dicom.Native DICOM codecs
// (see https://github.com/fo-dicom/fo-dicom/issues/631)
ImageManager.SetImplementation(new WinFormsImageManager());
//if there is a value we are treating as a zero date
if (!string.IsNullOrWhiteSpace(_opts.ZeroDate))
{
_zeroDate = DateTime.Parse(_opts.ZeroDate);
}
//if the user wants to run text detection
if (!string.IsNullOrWhiteSpace(_opts.TessDirectory))
{
var dir = new DirectoryInfo(_opts.TessDirectory);
if (!dir.Exists)
{
throw new DirectoryNotFoundException("Could not find TESS directory '" + _opts.TessDirectory + "'");
}
//to work with Tesseract eng.traineddata has to be in a folder called tessdata
if (!dir.Name.Equals("tessdata"))
{
dir = dir.CreateSubdirectory("tessdata");
}
var languageFile = new FileInfo(Path.Combine(dir.FullName, "eng.traineddata"));
if (!languageFile.Exists)
{
throw new FileNotFoundException($"Could not find tesseract models file ('{languageFile.FullName}')", languageFile.FullName);
}
_tesseractEngine = new TesseractEngine(dir.FullName, "eng", EngineMode.Default);
_tesseractEngine.DefaultPageSegMode = PageSegMode.Auto;
_tesseractReport = new PixelTextFailureReport(_opts.GetTargetName());
Reports.Add(_tesseractReport);
}
}
//public TesseractStanfordDicomFileClassifier(DirectoryInfo dataDirectory) : base(dataDirectory)
public TesseractStanfordDicomFileClassifier(DirectoryInfo dataDirectory, IsIdentifiableServiceOptions isIdentifiableServiceOptions) : base(dataDirectory)
{
var fileOptions = new IsIdentifiableDicomFileOptions();
//need to pass this so that the runner doesn't get unhappy about there being no reports (even though we clear it below)
fileOptions.ColumnReport = true;
fileOptions.TessDirectory = dataDirectory.FullName;
fileOptions.IgnoreTextLessThan = isIdentifiableServiceOptions.IgnoreTextLessThan;
// The Rules directory is always called "IsIdentifiableRules"
DirectoryInfo[] subDirs = dataDirectory.GetDirectories("IsIdentifiableRules");
foreach (DirectoryInfo subDir in subDirs)
{
fileOptions.RulesDirectory = subDir.FullName;
}
_runner = new DicomFileRunner(fileOptions);
}
