static void Main(string[] args)
{
string[] inputFiles = new string[]
{
@".\encrypted1 (password is 'password').pdf",
@".\encrypted2 (password is 'password').pdf",
};
using (DocumentMerger merger = new DocumentMerger("demo", "demo"))
{
// Handle `PasswordRequired` event
merger.PasswordRequired += new PasswordEventHandler(merger_PasswordRequired);
// Ignore document permissions
merger.CheckPermissions = false;
merger.Merge(inputFiles, @".\result.pdf");
}
// Open result document in default associated application (for demo purpose)
ProcessStartInfo processStartInfo = new ProcessStartInfo("result.pdf");
processStartInfo.UseShellExecute = true;
Process.Start(processStartInfo);
}
static void Main(string[] args)
{
const string inputFile = "sample.pdf";
const string resultFile = "result.pdf";
int pageCount;
// Get document page count
using (var infoExtractor = new InfoExtractor("demo", "demo"))
{
infoExtractor.LoadDocumentFromFile(inputFile);
pageCount = infoExtractor.GetPageCount();
}
// Process the document by 10-page pieces
int numberOfThreads = pageCount / 10;
if (pageCount - numberOfThreads * 10 > 0)
{
numberOfThreads += 1;
}
WaitHandle[] doneEvents = new WaitHandle[numberOfThreads];
Stopwatch stopwatch = Stopwatch.StartNew();
int startPage, endPage;
string[] pieces = new string[numberOfThreads];
for (int i = 0; i < numberOfThreads; i++)
{
doneEvents[i] = new ManualResetEvent(false);
startPage = i * 10;
endPage = Math.Min(pageCount - 1, (i + 1) * 10 - 1);
pieces[i] = string.Format("temp-{0}-{1}.pdf", startPage, endPage);
ThreadPool.QueueUserWorkItem(new WaitCallback(ThreadProc),
new object[] { i, doneEvents[i], inputFile, pieces[i], startPage, endPage });
}
// Wait for all threads
WaitHandle.WaitAll(doneEvents);
// Merge pieces
using (DocumentMerger merger = new DocumentMerger("demo", "demo"))
merger.Merge(pieces, resultFile);
// Delete temp files
foreach (string tempFile in pieces)
{
File.Delete(tempFile);
}
Console.WriteLine("All done in {0}.", stopwatch.Elapsed);
Console.WriteLine();
Console.WriteLine("Press any key to exit...");
Console.ReadKey();
}
static void Main(string[] args)
{
// Create and setup Bytescout.PDFExtractor.TextExtractor instance
TextExtractor extractor = new TextExtractor("demo", "demo");
// Load PDF document
extractor.LoadDocumentFromFile(InputFile);
// List to keep non-empty page numbers
List <string> nonEmptyPages = new List <string>();
// Iterate through pages
for (int pageIndex = 0; pageIndex < extractor.GetPageCount(); pageIndex++)
{
// Extract page text
string pageText = extractor.GetTextFromPage(pageIndex);
// If extracted text is not empty keep the page number
if (pageText.Length > 0)
{
nonEmptyPages.Add((pageIndex + 1).ToString());
}
}
// Cleanup
extractor.Dispose();
// Form comma-separated list of page numbers to split("1,3,5")
string ranges = string.Join(",", nonEmptyPages);
// Create Bytescout.PDFExtractor.DocumentSplitter instance
DocumentSplitter splitter = new DocumentSplitter("demo", "demo");
splitter.OptimizeSplittedDocuments = true;
// Split document by non-empty in temp folder
string[] parts = splitter.Split(InputFile, ranges, TempFolder);
// Cleanup
splitter.Dispose();
// Create Bytescout.PDFExtractor.DocumentMerger instance
DocumentMerger merger = new DocumentMerger("demo", "demo");
// Merge parts
merger.Merge(parts, OutputFile);
// Cleanup
merger.Dispose();
// Delete temp folder
Directory.Delete(TempFolder, true);
// Open the result file in default PDF viewer (for demo purposes)
Process.Start(OutputFile);
}
static void Main(string[] args)
{
string[] inputFiles = new string[] { "sample1.pdf", "sample2.pdf", "sample3.pdf" };
using (DocumentMerger merger = new DocumentMerger("demo", "demo"))
{
merger.Merge(inputFiles, "result.pdf");
}
Process.Start("result.pdf");
}
/// <summary>
/// Merge the set of PDF documents.
/// </summary>
public static byte[] Merge(IReadOnlyList <byte[]> files, IReadOnlyList <IReadOnlyList <int> > pagesBundle = null)
{
if (files == null)
{
throw new ArgumentNullException(nameof(files));
}
const bool isLenientParsing = false;
var documentBuilder = new DocumentMerger();
foreach (var fileIndex in Enumerable.Range(0, files.Count))
{
var file = files[fileIndex];
IReadOnlyList <int> pages = null;
if (pagesBundle != null && fileIndex < pagesBundle.Count)
{
pages = pagesBundle[fileIndex];
}
var inputBytes = new ByteArrayInputBytes(file);
var coreScanner = new CoreTokenScanner(inputBytes);
var version = FileHeaderParser.Parse(coreScanner, isLenientParsing, Log);
var crossReferenceParser = new CrossReferenceParser(Log, new XrefOffsetValidator(Log),
new Parser.Parts.CrossReference.CrossReferenceStreamParser(FilterProvider));
CrossReferenceTable crossReference = null;
// ReSharper disable once AccessToModifiedClosure
var locationProvider = new ObjectLocationProvider(() => crossReference, inputBytes);
var pdfScanner = new PdfTokenScanner(inputBytes, locationProvider, FilterProvider, NoOpEncryptionHandler.Instance);
var crossReferenceOffset = FileTrailerParser.GetFirstCrossReferenceOffset(inputBytes, coreScanner, isLenientParsing);
crossReference = crossReferenceParser.Parse(inputBytes, isLenientParsing, crossReferenceOffset, version.OffsetInFile, pdfScanner, coreScanner);
var catalogDictionaryToken = ParseCatalog(crossReference, pdfScanner, out var encryptionDictionary);
if (encryptionDictionary != null)
{
throw new PdfDocumentEncryptedException("Unable to merge document with password");
}
var documentCatalog = CatalogFactory.Create(crossReference.Trailer.Root, catalogDictionaryToken, pdfScanner, isLenientParsing);
documentBuilder.AppendDocument(documentCatalog, version.Version, pdfScanner, pages);
}
return(documentBuilder.Build());
}
static void Main(string[] args)
{
string[] inputFiles = new string[] { "sample1.pdf", "sample2.pdf", "sample3.pdf" };
using (DocumentMerger merger = new DocumentMerger("demo", "demo"))
{
merger.Merge(inputFiles, "result.pdf");
}
// Open result document in default associated application (for demo purpose)
ProcessStartInfo processStartInfo = new ProcessStartInfo("result.pdf");
processStartInfo.UseShellExecute = true;
Process.Start(processStartInfo);
}
static void Main(string[] args)
{
var SourceFolder = "PDFDocs";
var DestinationFile = "result.pdf";
var IncludeSubFolder = true;
using (DocumentMerger merger = new DocumentMerger("demo", "demo"))
{
merger.MergeFolder(SourceFolder, DestinationFile, IncludeSubFolder);
}
// Open result document in default associated application (for demo purpose)
ProcessStartInfo processStartInfo = new ProcessStartInfo(DestinationFile);
processStartInfo.UseShellExecute = true;
Process.Start(processStartInfo);
}
