public virtual void TestLigatureBeforeLigature()
{
System.Console.Out.WriteLine(new FileInfo(sourceFolder).FullName);
PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "ligature.pdf"));
// build strategy
RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("ca");
// get locations
IList <IPdfTextLocation> locationList = new List <IPdfTextLocation>();
for (int x = 1; x <= pdfDocument.GetNumberOfPages(); x++)
{
new PdfCanvasProcessor(extractionStrategy).ProcessPageContent(pdfDocument.GetPage(x));
foreach (IPdfTextLocation location in extractionStrategy.GetResultantLocations())
{
if (location != null)
{
locationList.Add(location);
}
}
}
// compare
NUnit.Framework.Assert.AreEqual(1, locationList.Count);
IPdfTextLocation loc = locationList[0];
NUnit.Framework.Assert.AreEqual("ca", loc.GetText());
Rectangle rect = loc.GetRectangle();
NUnit.Framework.Assert.AreEqual(36, rect.GetX(), 0.0001);
NUnit.Framework.Assert.AreEqual(655.4600, rect.GetY(), 0.0001);
NUnit.Framework.Assert.AreEqual(25.1000, rect.GetWidth(), 0.0001);
NUnit.Framework.Assert.AreEqual(20, rect.GetHeight(), 0.0001);
pdfDocument.Close();
}
public virtual void Test01()
{
System.Console.Out.WriteLine(new FileInfo(sourceFolder).FullName);
PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "in01.pdf"));
// build strategy
RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy(iText.IO.Util.StringUtil.RegexCompile
("\\{\\{Signature\\}\\}"));
// get locations
IList <IPdfTextLocation> locationList = new List <IPdfTextLocation>();
for (int x = 1; x <= pdfDocument.GetNumberOfPages(); x++)
{
new PdfCanvasProcessor(extractionStrategy).ProcessPageContent(pdfDocument.GetPage(x));
foreach (IPdfTextLocation location in extractionStrategy.GetResultantLocations())
{
if (location != null)
{
locationList.Add(location);
}
}
}
// compare
NUnit.Framework.Assert.AreEqual(1, locationList.Count);
IPdfTextLocation loc = locationList[0];
NUnit.Framework.Assert.AreEqual("{{Signature}}", loc.GetText());
NUnit.Framework.Assert.AreEqual(23, (int)loc.GetRectangle().GetX());
NUnit.Framework.Assert.AreEqual(375, (int)loc.GetRectangle().GetY());
NUnit.Framework.Assert.AreEqual(55, (int)loc.GetRectangle().GetWidth());
NUnit.Framework.Assert.AreEqual(11, (int)loc.GetRectangle().GetHeight());
// close
pdfDocument.Close();
}
public virtual void TestRotatedText()
{
PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "rotatedText.pdf"));
// build strategy
RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("abc");
// get locations
IList <IPdfTextLocation> locationList = new List <IPdfTextLocation>();
for (int x = 1; x <= pdfDocument.GetNumberOfPages(); x++)
{
new PdfCanvasProcessor(extractionStrategy).ProcessPageContent(pdfDocument.GetPage(x));
foreach (IPdfTextLocation location in extractionStrategy.GetResultantLocations())
{
if (location != null)
{
locationList.Add(location);
}
}
}
// compare
NUnit.Framework.Assert.AreEqual(2, locationList.Count);
NUnit.Framework.Assert.IsTrue(locationList[0].GetRectangle().EqualsWithEpsilon(new Rectangle(188.512f, 450f
, 14.800003f, 25.791992f)));
NUnit.Framework.Assert.IsTrue(locationList[1].GetRectangle().EqualsWithEpsilon(new Rectangle(36f, 746.688f
, 25.792f, 14.799988f)));
pdfDocument.Close();
}
public virtual void RegexWithOnlyWhiteSpace()
{
PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "regexWithWhiteSpacesTest.pdf"));
RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy(" ");
new PdfCanvasProcessor(extractionStrategy).ProcessPageContent(pdfDocument.GetPage(1));
IList <IPdfTextLocation> locations = new List <IPdfTextLocation>(extractionStrategy.GetResultantLocations());
pdfDocument.Close();
NUnit.Framework.Assert.AreEqual(0, locations.Count);
}
public virtual void RegexWithNewLinesTest()
{
PdfDocument pdfDocument = new PdfDocument(new PdfReader(sourceFolder + "regexWithNewLinesTest.pdf"));
RegexBasedLocationExtractionStrategy extractionStrategy = new RegexBasedLocationExtractionStrategy("\\nstart\\n"
);
new PdfCanvasProcessor(extractionStrategy).ProcessPageContent(pdfDocument.GetPage(1));
IList <IPdfTextLocation> locations = new List <IPdfTextLocation>(extractionStrategy.GetResultantLocations());
pdfDocument.Close();
NUnit.Framework.Assert.AreEqual(1, locations.Count);
NUnit.Framework.Assert.AreEqual("\nstart\n", locations[0].GetText());
NUnit.Framework.Assert.IsTrue(new Rectangle(56.8f, 729.5970f, 20.6159f, 13.2839f).EqualsWithEpsilon(locations
[0].GetRectangle()));
}
