How can I get the initial at the beginning of the paragraph?

Question

I'm trying to extract text from a PDF document using iTextSharp. The text I'm interested in appears beneath the "Introduction" header in the example below:

I have several hundred PDF documents that contain this "Introduction" page, normally on page five or six of the document. The paragraph always begins with an initial, such as the large P in "Physical" in the example.

In the following code, I scan the document for a page that begins with the text "Introduction," then I extract the text until the next heading ("Chapter 1"):

private static string GetIntroductionText( string filePath )
{
    using ( var reader = new PdfReader( filePath ) )
    {
        var appending = false;
        var introText = new StringBuilder();

        for ( var i = 1; i <= reader.NumberOfPages; i++ )
        {
            var pageText = PdfTextExtractor.GetTextFromPage( reader, i );

            if ( pageText.Trim().StartsWith( "Introduction" ) )
            {
                appending = true;
            }

            if ( pageText.Trim().StartsWith( "Chapter" ) )
            {
                break;
            }

            if ( appending )
            {
                introText.Append( pageText );
            }
        }

        return introText.ToString();
    }
}

The problem is that it doesn't extract the initial, i.e. the P in "Physical". So the text is:

hysical reality is consistent with universal laws. Where the laws do not operate, there is no reality. All of this...is unreal.

How do I get the initial at the beginning of the text?

I thought it might involve using the LocationTextExtractionStrategy like so:

var pageText = PdfTextExtractor.GetTextFromPage( reader, i, new LocationTextExtractionStrategy() );

Unfortunately this produced the same result.

Answer 1

For the record, here's how I solved this after looking at the iText source code (specifically the LocationTextExtractionStrategy class). Keep in mind that the (0, 0) coordinate is at the bottom-left of the page, not the top-left.

public class ChunkExtractionStrategy : ITextExtractionStrategy
{
    public List<Chunk> Chunks = new List<Chunk>();

    public void BeginTextBlock()
    {}

    public void EndTextBlock()
    {}

    public string GetResultantText()
    {
        var text = new StringBuilder();

        Chunks.Sort();

        Chunk prevChunk = null;

        foreach ( var chunk in Chunks )
        {
            if ( prevChunk == null && string.IsNullOrWhiteSpace( chunk.Text ) )
            {
                // blank space at beginning of page
                continue;
            }

            if ( prevChunk != null && !chunk.SameLine( prevChunk, 20 ) )
            {
                text.Append( "\n\n" );
            }

            text.Append( chunk.Text );

            prevChunk = chunk;
        }

        return text.ToString();
    }

    public void RenderImage( ImageRenderInfo renderInfo )
    {}

    public void RenderText( TextRenderInfo renderInfo )
    {
        Chunks.Add( new Chunk
                        {
                            TopLeft = renderInfo.GetAscentLine().GetStartPoint(),
                            BottomRight = renderInfo.GetDescentLine().GetEndPoint(),
                            Text = renderInfo.GetText(),
                        } );
    }

    public class Chunk : IComparable<Chunk>
    {
        public Vector TopLeft { get; set; }

        public Vector BottomRight { get; set; }

        public string Text { get; set; }

        public int CompareTo( Chunk other )
        {
            var y1 = (int)Math.Round( TopLeft[1] );
            var y2 = (int)Math.Round( other.TopLeft[1] );

            if ( y1 < y2 )
            {
                return 1;
            }

            if ( y1 > y2 )
            {
                return -1;
            }

            var x1 = (int)Math.Round( TopLeft[0] );
            var x2 = (int)Math.Round( other.TopLeft[0] );

            if ( x1 < x2 )
            {
                return -1;
            }

            if ( x1 > x2 )
            {
                return 1;
            }

            return 0;
        }

        public bool SameLine( Chunk other, int maxDiff = 0 )
        {
            var diff = Math.Abs( TopLeft[1] - other.TopLeft[1] );

            return diff <= maxDiff;
        }
    }
}

At first, I tried something similar to this answer. But then I found myself overriding everything in the class, so it made more sense to create a new implementation.