public virtual void CompareMultiLangImage()
{
String testName = "compareMultiLangImage";
String filename = "multilang";
String expectedPdfPath = TEST_DOCUMENTS_DIRECTORY + filename + "_" + testFileTypeName + ".pdf";
String resultPdfPath = GetTargetDirectory() + filename + "_" + testName + "_" + testFileTypeName + ".pdf";
try {
Tesseract4OcrEngineProperties properties = tesseractReader.GetTesseract4OcrEngineProperties();
properties.SetTextPositioning(TextPositioning.BY_WORDS);
properties.SetPathToTessData(GetTessDataDirectory());
properties.SetPageSegMode(3);
tesseractReader.SetTesseract4OcrEngineProperties(properties);
DoOcrAndSavePdfToPath(tesseractReader, TEST_IMAGES_DIRECTORY + filename + ".jpg", resultPdfPath, JavaUtil.ArraysAsList
("eng", "deu", "spa"), DeviceCmyk.BLACK);
NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(resultPdfPath, expectedPdfPath, TEST_DOCUMENTS_DIRECTORY
, "diff_"));
}
finally {
NUnit.Framework.Assert.AreEqual(TextPositioning.BY_WORDS, tesseractReader.GetTesseract4OcrEngineProperties
().GetTextPositioning());
NUnit.Framework.Assert.AreEqual(3, tesseractReader.GetTesseract4OcrEngineProperties().GetPageSegMode().Value
);
}
}
public virtual void CompareInvoiceFrontThaiImage()
{
String testName = "compareInvoiceFrontThaiImage";
String filename = "invoice_front_thai";
//Tesseract for Java and Tesseract for .NET give different output
//So we cannot use one reference pdf file for them
String expectedPdfPathJava = TEST_DOCUMENTS_DIRECTORY + filename + "_" + testFileTypeName + "_java.pdf";
String expectedPdfPathDotNet = TEST_DOCUMENTS_DIRECTORY + filename + "_" + testFileTypeName + "_dotnet.pdf";
String resultPdfPath = GetTargetDirectory() + filename + "_" + testName + "_" + testFileTypeName + ".pdf";
Tesseract4OcrEngineProperties properties = tesseractReader.GetTesseract4OcrEngineProperties();
properties.SetTextPositioning(TextPositioning.BY_WORDS_AND_LINES);
properties.SetPathToTessData(GetTessDataDirectory());
properties.SetLanguages(JavaUtil.ArraysAsList("tha", "eng"));
tesseractReader.SetTesseract4OcrEngineProperties(properties);
DoOcrAndSavePdfToPath(tesseractReader, TEST_IMAGES_DIRECTORY + filename + ".jpg", resultPdfPath, JavaUtil.ArraysAsList
("tha", "eng"), JavaUtil.ArraysAsList(NOTO_SANS_THAI_FONT_PATH, NOTO_SANS_FONT_PATH), DeviceRgb.RED);
bool javaTest = new CompareTool().CompareByContent(resultPdfPath, expectedPdfPathJava, TEST_DOCUMENTS_DIRECTORY
, "diff_") == null;
bool dotNetTest = new CompareTool().CompareByContent(resultPdfPath, expectedPdfPathDotNet, TEST_DOCUMENTS_DIRECTORY
, "diff_") == null;
NUnit.Framework.Assert.IsTrue(javaTest || dotNetTest);
}
public virtual void TestSpanishPNG()
{
String testName = "compareSpanishPNG";
String filename = "scanned_spa_01";
String expectedText1 = "¿Y SI ENSAYARA COMO ACTUAR?";
String expectedText2 = "¿Y SI ENSAYARA ACTUAR?";
String resultPdfPath = GetTargetDirectory() + filename + "_" + testName + "_" + testFileTypeName + ".pdf";
IList <String> languages = JavaUtil.ArraysAsList("spa", "spa_old");
Tesseract4OcrEngineProperties properties = tesseractReader.GetTesseract4OcrEngineProperties();
if (isExecutableReaderType)
{
properties.SetPreprocessingImages(false);
}
// locate text by words
properties.SetTextPositioning(TextPositioning.BY_WORDS);
properties.SetLanguages(languages);
tesseractReader.SetTesseract4OcrEngineProperties(properties);
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetTextColor(DeviceCmyk.BLACK);
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(tesseractReader, ocrPdfCreatorProperties);
using (PdfWriter pdfWriter = GetPdfWriter(resultPdfPath)) {
ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(new FileInfo(TEST_IMAGES_DIRECTORY + filename
+ ".png")), pdfWriter).Close();
}
try {
String result = GetTextFromPdfLayer(resultPdfPath, null, 1).Replace("\n", " ");
NUnit.Framework.Assert.IsTrue(result.Contains(expectedText1) || result.Contains(expectedText2));
}
finally {
NUnit.Framework.Assert.AreEqual(TextPositioning.BY_WORDS, tesseractReader.GetTesseract4OcrEngineProperties
().GetTextPositioning());
}
}
