How does TextRenderInfo work in iTextSharp? - asp.net

Question

I have got some codes from online and they are providing me the font sizes. I did not understand how the TextRenderInfo is reading text. I tried with renderInfo.GetText()) which is giving random number of characters, sometimes 3 characters, sometimes 2 characters or more or less. I need to know how the renderInfo is reading data ?
My intention is to separate every lines and paragraphs from pdf and also read their properties individually such as font size, font style etc. If you have any suggestion, please mention them.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using iTextSharp.text.pdf.parser;
using iTextSharp.text.pdf;
namespace FontSizeDig1
{
class Program
{
static void Main(string[] args)
{
// reader ==> http://itextsupport.com/apidocs/itext5/5.5.9/com/itextpdf/text/pdf/PdfReader.html#pdfVersion
PdfReader reader = new PdfReader(System.IO.Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "document.pdf"));
TextWithFontExtractionStategy S = new TextWithFontExtractionStategy();//strategy==> http://itextsupport.com/apidocs/itext5/5.5.9/com/itextpdf/text/pdf/parser/TextExtractionStrategy.html
// for (int i = 1; i <= reader.NumberOfPages; i++)
// {
string F = iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(reader, 1/*i*/, S);
// PdfTextExtractor.GetTextFromPage(reader, 6, S) ==>> http://itextsupport.com/apidocs/itext5/5.5.9/com/itextpdf/text/pdf/parser/PdfTextExtractor.html
Console.WriteLine(F);
// }
Console.ReadKey();
//this.Close();
}
}
public class TextWithFontExtractionStategy : iTextSharp.text.pdf.parser.ITextExtractionStrategy
{
//HTML buffer
private StringBuilder result = new StringBuilder();
//Store last used properties
private Vector lastBaseLine;
private string lastFont;
private float lastFontSize;
//http://api.itextpdf.com/itext/com/itextpdf/text/pdf/parser/TextRenderInfo.html
private enum TextRenderMode
{
FillText = 0,
StrokeText = 1,
FillThenStrokeText = 2,
Invisible = 3,
FillTextAndAddToPathForClipping = 4,
StrokeTextAndAddToPathForClipping = 5,
FillThenStrokeTextAndAddToPathForClipping = 6,
AddTextToPaddForClipping = 7
}
public void RenderText(iTextSharp.text.pdf.parser.TextRenderInfo renderInfo)
{
string curFont = renderInfo.GetFont().PostscriptFontName; // http://itextsupport.com/apidocs/itext5/5.5.9/com/itextpdf/text/pdf/parser/TextRenderInfo.html#getFont--
//Check if faux bold is used
if ((renderInfo.GetTextRenderMode() == 2/*(int)TextRenderMode.FillThenStrokeText*/))
{
curFont += "-Bold";
}
//This code assumes that if the baseline changes then we're on a newline
Vector curBaseline = renderInfo.GetBaseline().GetStartPoint();
Vector topRight = renderInfo.GetAscentLine().GetEndPoint();
iTextSharp.text.Rectangle rect = new iTextSharp.text.Rectangle(curBaseline[Vector.I1], curBaseline[Vector.I2], topRight[Vector.I1], topRight[Vector.I2]);
Single curFontSize = rect.Height;
//See if something has changed, either the baseline, the font or the font size
if ((this.lastBaseLine == null) || (curBaseline[Vector.I2] != lastBaseLine[Vector.I2]) || (curFontSize != lastFontSize) || (curFont != lastFont))
{
//if we've put down at least one span tag close it
if ((this.lastBaseLine != null))
{
this.result.AppendLine("</span>");
}
//If the baseline has changed then insert a line break
if ((this.lastBaseLine != null) && curBaseline[Vector.I2] != lastBaseLine[Vector.I2])
{
this.result.AppendLine("<br />");
}
//Create an HTML tag with appropriate styles
this.result.AppendFormat("<span style=\"font-family:{0};font-size:{1}\">", curFont, curFontSize);
}
//Append the current text
this.result.Append(renderInfo.GetText());
Console.WriteLine("me=" + renderInfo.GetText());//by imtiaj
//Set currently used properties
this.lastBaseLine = curBaseline;
this.lastFontSize = curFontSize;
this.lastFont = curFont;
}
public string GetResultantText()
{
//If we wrote anything then we'll always have a missing closing tag so close it here
if (result.Length > 0)
{
result.Append("</span>");
}
return result.ToString();
}
//Not needed
public void BeginTextBlock() { }
public void EndTextBlock() { }
public void RenderImage(ImageRenderInfo renderInfo) { }
}
}

Answer 1

Take a look at this PDF:
What do you see?
I see:
Hello World
Hello People
Now, let's parse this file? What do you expect?
You probably expect:
Hello World
Hello People
I don't.
That's where you and I differ, and that difference explains why you ask this question.
What do I expect?
Well, I'll start by looking inside the PDF, more specifically at the content stream of the first page:
I see 4 strings in the content stream: ld, Wor, llo, and He (in that order). I also see coordinates. Using those coordinates, I can compose what is shown:
Hello World
I don't immediately see "Hello People" anywhere, but I do see a reference to a Form XObject named /Xf1, so let's examine that Form XObject:
Woohoo! I'm in luck, "Hello People" is stored in the document as a single string value. I don't need to look at the coordinates to compose the actual text that I can see with my human eyes.
Now for your question. You say "I need to know how the renderInfo is reading data" and now you know: by default, iText will read all the strings from a page in the order they occur: ld, Wor, llo, He, and Hello People.
Depending on how the PDF is created, you can have output that is easy to read (Hello People), or output that is hard to read (ld, Wor, llo, He). iText comes with "strategies" that reorder all those snippets so that [ld, Wor, llo, He] is presented as [He, llo, Wor, ld], but detecting which of those parts belong to the same line, and which lines belong to the same paragraph, is something you will have to do.
NOTE: at iText Group, we already have plenty of closed source code that could save you plenty of time. Since we are the copyright owner of the iText library, we can ask money for that closed source code. That's something you typically can't do if you're using iText for free (because of the AGPL). However, if you are a customer of iText, we can probably disclose more source code. Do not expect us to give that code for free, as that code has too much commercial value.