public virtual void TestCompliantThaiPdfA()
{
String testName = "testCompliantThaiPdfA";
String path = PdfHelper.GetThaiImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetTextColor(DeviceRgb.BLACK);
FontProvider fontProvider = new FontProvider("Kanit");
fontProvider.AddFont(PdfHelper.GetKanitFontPath());
PdfOcrFontProvider pdfOcrFontProvider = new PdfOcrFontProvider(fontProvider.GetFontSet(), "Kanit");
ocrPdfCreatorProperties.SetFontProvider(pdfOcrFontProvider);
PdfHelper.CreatePdfA(pdfPath, new FileInfo(path), ocrPdfCreatorProperties, PdfHelper.GetRGBPdfOutputIntent
());
String resultWithActualText = PdfHelper.GetTextFromPdfLayerUseActualText(pdfPath, null);
NUnit.Framework.Assert.AreEqual(PdfHelper.THAI_TEXT, resultWithActualText);
String resultWithoutUseActualText = PdfHelper.GetTextFromPdfLayer(pdfPath, null);
NUnit.Framework.Assert.AreEqual(PdfHelper.THAI_TEXT, resultWithoutUseActualText);
NUnit.Framework.Assert.AreEqual(resultWithoutUseActualText, resultWithActualText);
ExtractionStrategy strategy = PdfHelper.GetExtractionStrategy(pdfPath);
PdfFont font = strategy.GetPdfFont();
String fontName = font.GetFontProgram().GetFontNames().GetFontName();
NUnit.Framework.Assert.IsTrue(fontName.Contains("Kanit"));
NUnit.Framework.Assert.IsTrue(font.IsEmbedded());
}
public virtual void TestIncompatibleOutputIntentAndFontColorSpaceException()
{
NUnit.Framework.Assert.That(() => {
String testName = "testIncompatibleOutputIntentAndFontColorSpaceException";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetTextColor(DeviceCmyk.BLACK);
PdfHelper.CreatePdfA(pdfPath, new FileInfo(path), ocrPdfCreatorProperties, PdfHelper.GetRGBPdfOutputIntent
());
}
, NUnit.Framework.Throws.InstanceOf <PdfException>().With.Message.EqualTo(PdfAConformanceException.DEVICECMYK_MAY_BE_USED_ONLY_IF_THE_FILE_HAS_A_CMYK_PDFA_OUTPUT_INTENT_OR_DEFAULTCMYK_IN_USAGE_CONTEXT))
;
}
public virtual void TestPdfA3uWithNullIntent()
{
String testName = "testPdfA3uWithNullIntent";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetTextColor(DeviceCmyk.BLACK);
properties.SetScaleMode(ScaleMode.SCALE_TO_FIT);
PdfHelper.CreatePdfA(pdfPath, new FileInfo(path), properties, null);
String result = PdfHelper.GetTextFromPdfLayer(pdfPath, null);
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, result);
NUnit.Framework.Assert.AreEqual(ScaleMode.SCALE_TO_FIT, properties.GetScaleMode());
}
public virtual void TestNonCompliantThaiPdfA()
{
NUnit.Framework.Assert.That(() => {
String testName = "testNonCompliantThaiPdfA";
String path = PdfHelper.GetThaiImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetTextColor(DeviceRgb.BLACK);
PdfHelper.CreatePdfA(pdfPath, new FileInfo(path), ocrPdfCreatorProperties, PdfHelper.GetRGBPdfOutputIntent
());
}
, NUnit.Framework.Throws.InstanceOf <OcrException>().With.Message.EqualTo(MessageFormatUtil.Format(OcrException.CANNOT_CREATE_PDF_DOCUMENT, MessageFormatUtil.Format(PdfOcrLogMessageConstant.COULD_NOT_FIND_CORRESPONDING_GLYPH_TO_UNICODE_CHARACTER, 3611))))
;
}
/// <summary>
/// Perform OCR using provided path to image (imgPath)
/// and save result PDF document to "pdfPath".
/// </summary>
/// <remarks>
/// Perform OCR using provided path to image (imgPath)
/// and save result PDF document to "pdfPath".
/// (Method is used for compare tool)
/// </remarks>
protected internal virtual void DoOcrAndSavePdfToPath(AbstractTesseract4OcrEngine tesseractReader, String
imgPath, String pdfPath, IList <String> languages, IList <String> fonts, Color color)
{
if (languages != null)
{
Tesseract4OcrEngineProperties properties = tesseractReader.GetTesseract4OcrEngineProperties();
properties.SetLanguages(languages);
tesseractReader.SetTesseract4OcrEngineProperties(properties);
}
OcrPdfCreatorProperties properties_1 = new OcrPdfCreatorProperties();
properties_1.SetPdfLang("en-US");
properties_1.SetTitle("");
if (fonts != null && fonts.Count > 0)
{
FontProvider fontProvider = new FontProvider();
foreach (String fontPath in fonts)
{
String name = FONT_PATH_TO_FONT_NAME_MAP.Get(fontPath);
fontProvider.GetFontSet().AddFont(fontPath, PdfEncodings.IDENTITY_H, name);
}
properties_1.SetFontProvider(fontProvider);
}
if (color != null)
{
properties_1.SetTextColor(color);
}
if (languages != null)
{
NUnit.Framework.Assert.AreEqual(languages.Count, tesseractReader.GetTesseract4OcrEngineProperties().GetLanguages
().Count);
}
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(tesseractReader, properties_1);
try {
using (PdfWriter pdfWriter = GetPdfWriter(pdfPath)) {
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(new FileInfo(imgPath
)), pdfWriter);
NUnit.Framework.Assert.IsNotNull(doc);
doc.Close();
}
}
catch (System.IO.IOException e) {
LOGGER.Error(e.Message);
}
}
public virtual void TestDefaultFontInPdfARgb()
{
String testName = "testDefaultFontInPdf";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetTextColor(DeviceRgb.BLACK);
PdfHelper.CreatePdfA(pdfPath, file, ocrPdfCreatorProperties, PdfHelper.GetRGBPdfOutputIntent());
ExtractionStrategy strategy = PdfHelper.GetExtractionStrategy(pdfPath);
PdfFont font = strategy.GetPdfFont();
String fontName = font.GetFontProgram().GetFontNames().GetFontName();
NUnit.Framework.Assert.IsTrue(fontName.Contains("LiberationSans"));
NUnit.Framework.Assert.IsTrue(font.IsEmbedded());
}
public virtual void TestFontColor()
{
String testName = "testFontColor";
String path = PdfHelper.GetImagesTestDirectory() + "multipage.tiff";
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetScaleMode(ScaleMode.SCALE_TO_FIT);
properties.SetTextLayerName("Text1");
Color color = DeviceCmyk.CYAN;
properties.SetTextColor(color);
PdfHelper.CreatePdf(pdfPath, file, properties);
ExtractionStrategy strategy = PdfHelper.GetExtractionStrategy(pdfPath, "Text1");
Color fillColor = strategy.GetFillColor();
NUnit.Framework.Assert.AreEqual(color, fillColor);
}
public virtual void TestPdfA3DefaultMetadata()
{
String testName = "testPdfDefaultMetadata";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetTextColor(DeviceRgb.BLACK);
PdfHelper.CreatePdfA(pdfPath, file, ocrPdfCreatorProperties, PdfHelper.GetRGBPdfOutputIntent());
PdfDocument pdfDocument = new PdfDocument(new PdfReader(pdfPath));
NUnit.Framework.Assert.AreEqual("en-US", pdfDocument.GetCatalog().GetLang().ToString());
NUnit.Framework.Assert.AreEqual(null, pdfDocument.GetDocumentInfo().GetTitle());
NUnit.Framework.Assert.AreEqual(PdfAConformanceLevel.PDF_A_3U, pdfDocument.GetReader().GetPdfAConformanceLevel
());
pdfDocument.Close();
}
