Extracting text elements with bounding box

[topcat30]

Could you tell me how to extract text chunks with coordinates?
I want to stich the chunks together to closely display the text like the original pdf. (With spacing etc)

I'm trying to replace itext but still using rectandtext location strategy

[topcat30]

Could you tell me how to extract text chunks with coordinates?
I want to stich the chunks together to closely display the text like the original pdf. (With spacing etc)

I'm trying to replace itext but still using rectandtext location strategy

I was able to do this. All fixed.

[EspressoWillie]

Would you share your code on how you did this? I am trying to reproduce a text report including white spaces and all new lines.
Thank you.

[topcat30]

#383

[topcat30]

Hi, this gets me most of the way. I usually use a GAP of .3 and OrigRow is true. The GAP is basically used to add space between each character, so you can play with it depending on the font size. Things get a bit messy if the fonts change size quite a lot.
Let me know if you find a better way.

oh and you can swap the output to be either spool or lines if you want an xml dump of the text data.

using System;
using System.Text;
using System.IO;
using Org.BouncyCastle.Cms;
using System.Collections.Generic;
using System.Xml;
using System.Xml.Linq;
using System.Linq;
using System.Collections;
using UglyToad.PdfPig;
using UglyToad.PdfPig.DocumentLayoutAnalysis.TextExtractor;
namespace PDFTools
{
public class Pdf2Text
{

public string Encoded(string EncDocin, double gap, bool OrigRow)
{
    StringBuilder text = new StringBuilder();


    string lines = "";
    string spool = "";
    byte[] str2 = Convert.FromBase64String(EncDocin);
    MemoryStream mstream = new MemoryStream(str2);

    using (var r = PdfDocument.Open(mstream))
    {
        string pageText;
        string pagetext2;
        XElement xmlElements = new XElement("Root");
        for (int i = 1; i <= r.NumberOfPages; i++)

        {
            //ContentOrderTextExtractor ct = new ContentOrderTextExtractor(GetText)

            pageText = "";
            pagetext2 = "";
            var page = r.GetPage(i);
          

            xmlElements.Add(new XElement("Page" + i));
            int groupby;
            if (OrigRow == true) groupby = 1; else groupby = 10;
            var sorted = page.Letters.OrderByDescending(x => Math.Round((x.StartBaseLine.Y) / groupby) * groupby).ThenBy(x => x.GlyphRectangle.Left).ToList();
            //var sorted = page.Letters.OrderByDescending(x => (x.StartBaseLine.Y) / groupby * groupby).ThenBy(x => x.GlyphRectangle.Left).ToList();



            //  var grouped = sorted.GroupBy(x => Math.Round((x.StartBaseLine.Y) / 10) * 10).ToList();
            var grouped = sorted.GroupBy(x => Math.Round((x.StartBaseLine.Y) / groupby) * groupby).ToList();
            //var grouped = sorted.GroupBy(x => x.StartBaseLine.Y*1).ToList();
            //var grouped = sorted.GroupBy(x => x.StartBaseLine.Y).ToList();

            int n = 1;
            string Row = "";
            foreach (var group in grouped)
            {
                //double gap = .3;
                Row = "";
                int LoopCount = 0;
                int LineCount = 0;
                double endloop = 0;
                xmlElements.Element("Page" + i).Add(new XElement("ROW" + n));
                string varText = "";
                double previousRight = 0;
                double previousLeft = 0;
                string previousText = "";
                int rowLength = 0;

                foreach (var p in group)
                {
                    endloop = 0;
                    LoopCount = 1;
                    // varText = p.Value.PadLeft((int)p.GlyphRectangle.Left - rowLength, ' ');
                    double chunkDif = (p.StartBaseLine.X) - previousRight;
                    //var diff = Math.Abs(LineCount - (p.GlyphRectangle.Left/.3));
                    xmlElements.Element("Page" + i).Element("ROW" + n).Add(new XElement("Chunk", new XElement("Value", p.Value), new XElement("leftPos", p.GlyphRectangle.Left), new XElement("bottomPos", p.GlyphRectangle.Bottom), new XElement("topPosOrig", p.GlyphRectangle.Top), new XElement("topPos", Math.Round((p.GlyphRectangle.Top) / 10) * 10), new XElement("rightPos", p.GlyphRectangle.Right), new XElement("Length", Math.Round(p.GlyphRectangle.Width, 0, MidpointRounding.AwayFromZero)), new XElement("Height", Math.Round(p.GlyphRectangle.Height)), new XElement("Type", p.GlyphRectangle.Rotation), new XElement("Diff", chunkDif)));

                    if (LineCount == 1)
                    {
                        endloop = (p.StartBaseLine.X * gap);
                    }
                    else
                    {
                        //endloop = (diff)/2;
                        if (LineCount <= (p.StartBaseLine.X * gap))
                        { endloop = (((p.StartBaseLine.X * gap)) - LineCount) + p.Value.Length; }
                        else { endloop = (p.StartBaseLine.X * gap) + p.Value.Length; }

                    }
                    if (chunkDif > gap)
                    {
                        //varText = p.Value;
                        varText = p.Value.PadLeft((int)endloop, ' ');
                    }
                    else
                    { varText = p.Value; }



                    Row = Row + varText.Replace("\r\n", "");
                    rowLength = Row.Length;
                    LineCount = Row.Length;//(LoopCount + p.Value.Replace("\r\n", "").Length) + LineCount;
                    previousRight = p.EndBaseLine.X;
                    previousLeft = (p.StartBaseLine.X);
                    previousText = varText;
                }
                previousRight = 0;
                //Console.Write(Row);
                // Console.ReadLine();
                if (n == 1)
                { pageText = pageText + Row; }
                else { pageText = pageText + "\r\n" + Row; }
                n++;
            }
            if (i == 1)
            {
                spool = spool + pageText;
            }
            else { spool = spool + "\f" + pageText; }

            page = null;
        }
        text.Append(xmlElements);
        lines = text.ToString();
        text.Length = 0;
        text.Capacity = 0;
    }

//    File.WriteAllText(@"C:\Temp\csc2.txt", spool);
//    File.WriteAllText(@"C:\Temp\csc.xml", lines);
    //Console.WriteLine(lines);
    //Console.ReadLine();
    return spool;
   
}

}

[An-Aviator]

Fixed my problem perfectly, Thank you!

[BobLd]

I'm not sure to understand exactly what you want to achieve, but did you have a look at the Layout Analysis tools available? Maybe have a look at the wiki here... I'd start with the NearestNeighbourWordExtractor

[topcat30]

The PDF chuncks do not contains spaces so extracting words doess not give the same representation of the original PDF in plain text, its just a list of words. My solution groups the chunks by row and column, then using a common font size we supply the maximum character spacing within a word and the program sitches the the characters together.