/// <summary>Preprocess given image if it is needed.</summary>
/// <param name="inputImage">
/// original input image
/// <see cref="System.IO.FileInfo"/>
/// </param>
/// <param name="pageNumber">number of page to be OCRed</param>
/// <returns>
/// path to output image as
/// <see cref="System.String"/>
/// </returns>
private String PreprocessImage(FileInfo inputImage, int pageNumber)
{
String tmpFileName = TesseractOcrUtil.GetTempFilePath(Guid.NewGuid().ToString(), GetExtension(inputImage));
String path = inputImage.FullName;
try {
if (GetTesseract4OcrEngineProperties().IsPreprocessingImages())
{
Pix pix = ImagePreprocessingUtil.PreprocessImage(inputImage, pageNumber);
TesseractOcrUtil.SavePixToTempPngFile(tmpFileName, pix);
if (!File.Exists(System.IO.Path.Combine(tmpFileName)))
{
System.Drawing.Bitmap img = TesseractOcrUtil.ConvertPixToImage(pix);
if (img != null)
{
TesseractOcrUtil.SaveImageToTempPngFile(tmpFileName, img);
}
}
}
if (!GetTesseract4OcrEngineProperties().IsPreprocessingImages() || !File.Exists(System.IO.Path.Combine(tmpFileName
)))
{
TesseractOcrUtil.CreateTempFileCopy(path, tmpFileName);
}
if (File.Exists(System.IO.Path.Combine(tmpFileName)))
{
path = tmpFileName;
}
}
catch (System.IO.IOException e) {
LogManager.GetLogger(GetType()).Error(MessageFormatUtil.Format(Tesseract4LogMessageConstant.CANNOT_READ_INPUT_IMAGE
, e.Message));
}
return(path);
}
public virtual void TestNullSavingAsPng()
{
String tmpFileName = TesseractOcrUtil.GetTempFilePath(GetTargetDirectory() + "/testNullSavingAsPng", ".png"
);
TesseractOcrUtil.SaveImageToTempPngFile(tmpFileName, null);
NUnit.Framework.Assert.IsFalse(File.Exists(System.IO.Path.Combine(tmpFileName)));
TesseractOcrUtil.SavePixToTempPngFile(tmpFileName, null);
NUnit.Framework.Assert.IsFalse(File.Exists(System.IO.Path.Combine(tmpFileName)));
}
public virtual void TestDoTesseractOcrForNonAsciiPathForExecutable()
{
String path = TEST_IMAGES_DIRECTORY + "tèst/noisy_01.png";
FileInfo imgFile = new FileInfo(path);
FileInfo outputFile = new FileInfo(TesseractOcrUtil.GetTempFilePath("test", ".hocr"));
Tesseract4OcrEngineProperties properties = new Tesseract4OcrEngineProperties();
properties.SetPathToTessData(GetTessDataDirectory());
properties.SetPreprocessingImages(false);
Tesseract4ExecutableOcrEngine engine = new Tesseract4ExecutableOcrEngine(GetTesseractDirectory(), properties
);
engine.DoTesseractOcr(imgFile, outputFile, OutputFormat.HOCR);
NUnit.Framework.Assert.IsTrue(File.Exists(System.IO.Path.Combine(outputFile.FullName)));
TesseractHelper.DeleteFile(outputFile.FullName);
NUnit.Framework.Assert.IsFalse(File.Exists(System.IO.Path.Combine(outputFile.FullName)));
}
/// <summary>
/// Using provided input stream there will be created
/// temporary file (with name 'language.user-words')
/// containing words (one per line) which ends with
/// a new line character.
/// </summary>
/// <remarks>
/// Using provided input stream there will be created
/// temporary file (with name 'language.user-words')
/// containing words (one per line) which ends with
/// a new line character. Train data for provided language
/// should exist in specified tess data directory.
/// NOTE:
/// User words dictionary doesn't work properly in tesseract4
/// and hidden for public usage until fix is available
/// </remarks>
/// <param name="language">
/// language as
/// <see cref="System.String"/>
/// , tessdata for
/// this languages has to exist in tess data directory
/// </param>
/// <param name="inputStream">
/// custom user words as
/// <see cref="System.IO.Stream"/>
/// </param>
/// <returns>
/// the
/// <see cref="Tesseract4OcrEngineProperties"/>
/// instance
/// </returns>
internal virtual iText.Pdfocr.Tesseract4.Tesseract4OcrEngineProperties SetUserWords(String language, Stream
inputStream)
{
SetPathToUserWordsFile(null);
if (!GetLanguages().Contains(language))
{
if (DEFAULT_LANGUAGE.Equals(language.ToLowerInvariant()))
{
IList <String> languagesList = GetLanguages();
languagesList.Add(language);
SetLanguages(languagesList);
}
else
{
throw new Tesseract4OcrException(Tesseract4OcrException.LANGUAGE_IS_NOT_IN_THE_LIST).SetMessageParams(language
);
}
}
String userWordsFileName = TesseractOcrUtil.GetTempFilePath(language, "." + DEFAULT_USER_WORDS_SUFFIX);
try {
using (StreamWriter writer = new StreamWriter(userWordsFileName)) {
TextReader reader = new StreamReader(inputStream, System.Text.Encoding.UTF8);
int data;
while ((data = reader.Read()) != -1)
{
writer.Write(data);
}
writer.Write(Environment.NewLine);
SetPathToUserWordsFile(userWordsFileName, true);
}
}
catch (System.IO.IOException e) {
SetPathToUserWordsFile(null);
LogManager.GetLogger(GetType()).Warn(MessageFormatUtil.Format(Tesseract4LogMessageConstant.CANNOT_USE_USER_WORDS
, e.Message));
}
return(this);
}
/// <summary>Creates a temporary file with given extension.</summary>
/// <param name="extension">
/// file extension for a new file
/// <see cref="System.String"/>
/// </param>
/// <returns>
/// a new created
/// <see cref="System.IO.FileInfo"/>
/// instance
/// </returns>
private FileInfo CreateTempFile(String extension)
{
String tmpFileName = TesseractOcrUtil.GetTempFilePath(Guid.NewGuid().ToString(), extension);
return(new FileInfo(tmpFileName));
}
