ByteScout PDF Suite – C# – Extract Text From Large PDF Document in Parallel Processing

Home
/
Articles
/
ByteScout PDF Suite – C# – Extract Text From Large PDF Document in Parallel Processing

printable version:
ByteScout-PDF-Suite-C-sharp-Extract-Text-From-Large-PDF-Document-in-Parallel-Processing.pdf

How to extract text from large PDF document in parallel processing in C# using ByteScout PDF Suite

Learning is essential in computer world and the tutorial below will demonstrate how to extract text from large PDF document in parallel processing in C#

Extract text from large PDF document in parallel processing is simple to apply in C# if you use these source codes below. ByteScout PDF Suite is the set that includes 6 SDK products to work with PDF from generating rich PDF reports to extracting data from PDF documents and converting them to HTML. This bundle includes PDF (Generator) SDK, PDF Renderer SDK, PDF Extractor SDK, PDF to HTML SDK, PDF Viewer SDK and PDF Generator SDK for Javascript. It can extract text from large PDF document in parallel processing in C#.

These C# code samples for C# guide developers to speed up coding of the application when using ByteScout PDF Suite. Simply copy and paste in your C# project or application you and then run your app! Further improvement of the code will make it more robust.

ByteScout provides the free trial version of ByteScout PDF Suite along with the documentation and source code samples.

On-demand (REST Web API) version:
Web API (on-demand version)

On-premise offline SDK for Windows:
60 Day Free Trial (on-premise)

Program.cs

      using System;
using System.Diagnostics;
using System.IO;
using System.Threading;
using Bytescout.PDFExtractor;

namespace MultithreadProcessing
{
    class Program
    {
        // Limit to 4 threads in queue.
        // Set this value to number of cores in your CPU for max performance.
        private static readonly Semaphore _threadLimiter = new Semaphore(4, 4);
        private static int _runningThreadsCounter;

        static void Main(string[] args)
        {
            const string inputFileName = "sample.pdf";
            const string resultFileName = "result.txt";
            int CHUNK_SIZE = 10;

            int pageCount;

            // Get document page count
            using (var infoExtractor = new InfoExtractor("demo", "demo"))
            {
                infoExtractor.LoadDocumentFromFile(inputFileName);
                pageCount = infoExtractor.GetPageCount();
            }

            Stopwatch stopwatch = Stopwatch.StartNew();

            int numberOfThreads = pageCount / CHUNK_SIZE;
            if (pageCount - numberOfThreads * CHUNK_SIZE > 0)
                numberOfThreads += 1;

            ManualResetEvent allFinishedEvent = new ManualResetEvent(false);
            _runningThreadsCounter = 0;
            string[] chunks = new string[numberOfThreads];

            for (int i = 0; i < numberOfThreads; i++)
            {
                // Wait for the queue
                _threadLimiter.WaitOne();

                var startPage = i * CHUNK_SIZE;
                var endPage = Math.Min(pageCount - 1, (i + 1) * CHUNK_SIZE - 1);

                // Prepare temp file name for the chunk
                chunks[i] = string.Format("temp-{0}-{1}.txt", startPage, endPage);

                // Increase the thread counter
                Interlocked.Increment(ref _runningThreadsCounter);

                ThreadPool.QueueUserWorkItem(new WaitCallback(ThreadProc),
                    new object[] { i, allFinishedEvent, inputFileName, chunks[i], startPage, endPage });
            }

            // Wait for all threads
            allFinishedEvent.WaitOne();

            // Merge pieces into a single text file
            using (Stream resultFileStream = File.Create(resultFileName))
            {
                foreach (string tempFile in chunks)
                    using (Stream srcStream = File.OpenRead(tempFile))
                        srcStream.CopyTo(resultFileStream);
            }

            // Delete temp files
            foreach (string tempFile in chunks)
                File.Delete(tempFile);


            Console.WriteLine("All done in {0}.", stopwatch.Elapsed);
            Console.WriteLine();

            Console.WriteLine("Press any key to exit...");
            Console.ReadKey();
        }

        private static void ThreadProc(object stateInfo)
        {
            int threadIndex = (int) ((object[]) stateInfo)[0];
            ManualResetEvent allFinishedEvent = (ManualResetEvent) ((object[]) stateInfo)[1];
            string inputFile = (string) ((object[]) stateInfo)[2];
            string outputFile = (string) ((object[]) stateInfo)[3];
            int startPage = (int) ((object[]) stateInfo)[4];
            int endPage = (int) ((object[]) stateInfo)[5];

            try
            {
                Console.WriteLine("Thread #{0} started with the page range from {1} to {2}.", threadIndex, startPage, endPage);

                Stopwatch stopwatch = Stopwatch.StartNew();

                // Process the piece
                using (TextExtractor textExtractor = new TextExtractor("demo", "demo"))
                {
                    // Set page separator. Default is '\f' (Form Feed)
                    textExtractor.PageSeparator = Environment.NewLine;
                    // Since we are only extracting text, disable the caching to reduce memory usage
                    textExtractor.PageDataCaching = PageDataCaching.None;

                    textExtractor.OCRMode = OCRMode.Auto;
                    textExtractor.OCRLanguageDataFolder = @"c:\Program Files\Bytescout PDF Extractor SDK\net4.00\tessdata\";
                    textExtractor.OCRLanguage = "eng";
                    // 300 DPI resolution is recommended. 
                    // Using of higher values will slow down the processing but does not guarantee the higher quality.
                    textExtractor.OCRResolution = 300;

                    textExtractor.LoadDocumentFromFile(inputFile);

                    textExtractor.SaveTextToFile(startPage, endPage, outputFile);
                }

                Console.WriteLine("Thread #{0} finished in {1}.", threadIndex, stopwatch.Elapsed);
            }
            finally
            {
                // If it was the last thread, signal the main thread about the finish.
                if (Interlocked.Decrement(ref _runningThreadsCounter) == 0)
                    allFinishedEvent.Set();

                // Release semaphore
                _threadLimiter.Release();
            }
        }
    }
}