public virtual void TestTextFromPdfLayers()
{
String testName = "testTextFromPdfLayers";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetImageLayerName("Image Layer");
properties.SetTextLayerName("Text Layer");
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(new CustomOcrEngine(), properties);
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
(pdfPath));
NUnit.Framework.Assert.IsNotNull(doc);
IList <PdfLayer> layers = doc.GetCatalog().GetOCProperties(true).GetLayers();
NUnit.Framework.Assert.AreEqual(2, layers.Count);
NUnit.Framework.Assert.AreEqual("Image Layer", layers[0].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[0].IsOn());
NUnit.Framework.Assert.AreEqual("Text Layer", layers[1].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[1].IsOn());
doc.Close();
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, PdfHelper.GetTextFromPdfLayer(pdfPath, "Text Layer"
));
NUnit.Framework.Assert.AreEqual("", PdfHelper.GetTextFromPdfLayer(pdfPath, "Image Layer"));
}
public virtual void TestReusingFontProvider()
{
String testName = "testReusingFontProvider";
String path = PdfHelper.GetDefaultImagePath();
String pdfPathA3u = PdfHelper.GetTargetDirectory() + testName + "_a3u.pdf";
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
FontProvider fontProvider = new FontProvider("FreeSans");
fontProvider.AddFont(PdfHelper.GetFreeSansFontPath());
PdfOcrFontProvider pdfOcrFontProvider = new PdfOcrFontProvider(fontProvider.GetFontSet(), "FreeSans");
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetFontProvider(pdfOcrFontProvider);
PdfHelper.CreatePdfA(pdfPathA3u, file, ocrPdfCreatorProperties, PdfHelper.GetCMYKPdfOutputIntent());
PdfHelper.CreatePdf(pdfPath, file, ocrPdfCreatorProperties);
ExtractionStrategy strategy = PdfHelper.GetExtractionStrategy(pdfPathA3u);
PdfFont font = strategy.GetPdfFont();
String fontName = font.GetFontProgram().GetFontNames().GetFontName();
NUnit.Framework.Assert.IsTrue(fontName.Contains("FreeSans"));
NUnit.Framework.Assert.IsTrue(font.IsEmbedded());
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, strategy.GetResultantText());
strategy = PdfHelper.GetExtractionStrategy(pdfPath);
font = strategy.GetPdfFont();
fontName = font.GetFontProgram().GetFontNames().GetFontName();
NUnit.Framework.Assert.IsTrue(fontName.Contains("FreeSans"));
NUnit.Framework.Assert.IsTrue(font.IsEmbedded());
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, strategy.GetResultantText());
}
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 TestScaleWidthMode()
{
String testName = "testScaleWidthMode";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
float pageWidthPt = 400f;
float pageHeightPt = 400f;
Rectangle pageSize = new Rectangle(pageWidthPt, pageHeightPt);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetScaleMode(ScaleMode.SCALE_WIDTH);
properties.SetPageSize(pageSize);
PdfHelper.CreatePdf(pdfPath, file, properties);
Rectangle rect = GetImageBBoxRectangleFromPdf(pdfPath);
ImageData originalImageData = ImageDataFactory.Create(file.FullName);
// page size should be equal to the result image size
// result image height should be equal to the value that
// was set as page height result image width should be scaled
// proportionally according to the provided image height
// and original image size
NUnit.Framework.Assert.AreEqual(pageHeightPt, rect.GetHeight(), DELTA);
NUnit.Framework.Assert.AreEqual(originalImageData.GetWidth() / originalImageData.GetHeight(), rect.GetWidth
() / rect.GetHeight(), DELTA);
}
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();
}
public virtual void TestPdfLayersWithTextLayerOnly()
{
String path = PdfHelper.GetDefaultImagePath();
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetTextLayerName("Text Layer");
OcrPdfCreator ocrPdfCreator = new OcrPdfCreator(new CustomOcrEngine(), properties);
PdfDocument doc = ocrPdfCreator.CreatePdf(JavaCollectionsUtil.SingletonList <FileInfo>(file), PdfHelper.GetPdfWriter
());
NUnit.Framework.Assert.IsNotNull(doc);
IList <PdfLayer> layers = doc.GetCatalog().GetOCProperties(true).GetLayers();
NUnit.Framework.Assert.AreEqual(1, layers.Count);
NUnit.Framework.Assert.AreEqual("Text Layer", layers[0].GetPdfObject().Get(PdfName.Name).ToString());
NUnit.Framework.Assert.IsTrue(layers[0].IsOn());
doc.Close();
}
public virtual void TestInvalidFontWithInvalidDefaultFontFamily()
{
NUnit.Framework.Assert.That(() => {
String testName = "testInvalidFontWithInvalidDefaultFontFamily";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
FontProvider pdfOcrFontProvider = new FontProvider("Font");
pdfOcrFontProvider.GetFontSet().AddFont("font.ttf", PdfEncodings.IDENTITY_H, "Font");
properties.SetFontProvider(pdfOcrFontProvider, "Font");
properties.SetScaleMode(ScaleMode.SCALE_TO_FIT);
PdfHelper.CreatePdf(pdfPath, file, properties);
String result = PdfHelper.GetTextFromPdfLayer(pdfPath, null);
NUnit.Framework.Assert.AreEqual(PdfHelper.DEFAULT_TEXT, result);
NUnit.Framework.Assert.AreEqual(ScaleMode.SCALE_TO_FIT, properties.GetScaleMode());
}
, NUnit.Framework.Throws.InstanceOf <OcrException>().With.Message.EqualTo(MessageFormatUtil.Format(OcrException.CANNOT_CREATE_PDF_DOCUMENT, OcrException.CANNOT_RESOLVE_PROVIDED_FONTS)))
;
}
public virtual void TestCustomFontInPdf()
{
String testName = "testDefaultFontInPdf";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
FontProvider fontProvider = new FontProvider("FreeSans");
fontProvider.GetFontSet().AddFont(PdfHelper.GetFreeSansFontPath(), PdfEncodings.IDENTITY_H, "FreeSans");
OcrPdfCreatorProperties ocrPdfCreatorProperties = new OcrPdfCreatorProperties();
ocrPdfCreatorProperties.SetPdfLang("en-US");
ocrPdfCreatorProperties.SetFontProvider(fontProvider, "FreeSans");
PdfHelper.CreatePdfA(pdfPath, file, ocrPdfCreatorProperties, PdfHelper.GetCMYKPdfOutputIntent());
ExtractionStrategy strategy = PdfHelper.GetExtractionStrategy(pdfPath);
PdfFont font = strategy.GetPdfFont();
String fontName = font.GetFontProgram().GetFontNames().GetFontName();
NUnit.Framework.Assert.IsTrue(fontName.Contains("FreeSans"));
NUnit.Framework.Assert.IsTrue(font.IsEmbedded());
}
public virtual void TestScaleHeightMode()
{
String testName = "testScaleHeightMode";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
float pageWidthPt = 400f;
float pageHeightPt = 400f;
Rectangle pageSize = new Rectangle(pageWidthPt, pageHeightPt);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
properties.SetScaleMode(ScaleMode.SCALE_HEIGHT);
properties.SetPageSize(pageSize);
PdfHelper.CreatePdf(pdfPath, file, properties);
Rectangle rect = GetImageBBoxRectangleFromPdf(pdfPath);
ImageData originalImageData = ImageDataFactory.Create(file.FullName);
NUnit.Framework.Assert.AreEqual(pageWidthPt, rect.GetWidth(), DELTA);
NUnit.Framework.Assert.AreEqual(originalImageData.GetWidth() / originalImageData.GetHeight(), rect.GetWidth
() / rect.GetHeight(), DELTA);
}
public virtual void TestPdfCustomMetadata()
{
String testName = "testPdfCustomMetadata";
String path = PdfHelper.GetDefaultImagePath();
String pdfPath = PdfHelper.GetTargetDirectory() + testName + ".pdf";
FileInfo file = new FileInfo(path);
OcrPdfCreatorProperties properties = new OcrPdfCreatorProperties();
String locale = "nl-BE";
properties.SetPdfLang(locale);
String title = "Title";
properties.SetTitle(title);
PdfHelper.CreatePdfA(pdfPath, file, new OcrPdfCreatorProperties(properties), PdfHelper.GetCMYKPdfOutputIntent
());
PdfDocument pdfDocument = new PdfDocument(new PdfReader(pdfPath));
NUnit.Framework.Assert.AreEqual(locale, pdfDocument.GetCatalog().GetLang().ToString());
NUnit.Framework.Assert.AreEqual(title, pdfDocument.GetDocumentInfo().GetTitle());
NUnit.Framework.Assert.AreEqual(PdfAConformanceLevel.PDF_A_3U, pdfDocument.GetReader().GetPdfAConformanceLevel
());
pdfDocument.Close();
}
/// <summary>
/// Sets properties for
/// <see cref="OcrPdfCreator"/>.
/// </summary>
/// <param name="ocrPdfCreatorProperties">
/// set of properties
/// <see cref="OcrPdfCreatorProperties"/>
/// for
/// <see cref="OcrPdfCreator"/>
/// </param>
public void SetOcrPdfCreatorProperties(OcrPdfCreatorProperties ocrPdfCreatorProperties)
{
this.ocrPdfCreatorProperties = ocrPdfCreatorProperties;
}
/// <summary>
/// Creates a new
/// <see cref="OcrPdfCreator"/>
/// instance.
/// </summary>
/// <param name="ocrEngine">
/// selected OCR Reader
/// <see cref="IOcrEngine"/>
/// </param>
/// <param name="ocrPdfCreatorProperties">
/// set of properties for
/// <see cref="OcrPdfCreator"/>
/// </param>
public OcrPdfCreator(IOcrEngine ocrEngine, OcrPdfCreatorProperties ocrPdfCreatorProperties)
{
SetOcrEngine(ocrEngine);
SetOcrPdfCreatorProperties(ocrPdfCreatorProperties);
}
