public static PdfParser Parse(Stream InputStream)
{
PdfParser analyzer = new PdfParser();
analyzer.inputStream = InputStream;
analyzer.objectMemory = new Dictionary<int, PdfFileObject>();
analyzer.inputStream.Seek(0, SeekOrigin.Begin);
string ver = new StreamReader(analyzer.inputStream).ReadLine().Substring(5);
analyzer._version = double.Parse(ver, System.Globalization.CultureInfo.InvariantCulture);
analyzer.startxref = GetStartxref(analyzer.inputStream);
analyzer.objectMemory = analyzer.LoadObjects(analyzer.inputStream);
if (analyzer.trailer == null)
{
analyzer.trailer = analyzer.ParseTrailer(analyzer.startxref);
}
analyzer.root = analyzer.trailer.GetChildObject("Root");
analyzer.root.Type = PdfObjectTypes.Root;
analyzer.rootPages = analyzer.root.GetChildObject("Pages");
analyzer.isEncrypted = analyzer.trailer.GetChildObject("Encrypt") != null;
analyzer.pageCount = analyzer.rootPages.GetInt32Value("Count").Value;
return analyzer;
}
public void StartSplitter()
{
this.DialogResult = DialogResult.OK;
int n = 1;
foreach (PdfProjectElement ppe in this.Project.Elements)
{
if (!ppe.Enabled)
{
continue;
}
if (ppe is PdfProjectPdf)
{
foreach (int page in ((PdfProjectPdf)ppe).CalculatePageSelection(true))
{
using (Stream s = File.OpenRead(ppe.Path))
{
PdfParser parsed = PdfParser.Parse(s);
this.GenerateSplit(n, this.Project.Target, parsed, page);
n++;
}
}
}
if (ppe is PdfProjectImage)
{
this.GenerateSplit(n, this.Project.Target, PdfParser.Parse(ImageToPDFConverter.ConvertImageToPDF(ppe.Path)), 0);
n++;
}
}
}
public static PdfParser Parse(Stream InputStream)
{
PdfParser analyzer = new PdfParser();
analyzer.inputStream = InputStream;
analyzer.objectMemory = new Dictionary <int, PdfFileObject>();
analyzer.inputStream.Seek(0, SeekOrigin.Begin);
string ver = new StreamReader(analyzer.inputStream).ReadLine().Substring(5);
analyzer._version = double.Parse(ver, System.Globalization.CultureInfo.InvariantCulture);
analyzer.startxref = GetStartxref(analyzer.inputStream);
analyzer.objectMemory = analyzer.LoadObjects(analyzer.inputStream);
if (analyzer.trailer == null)
{
analyzer.trailer = analyzer.ParseTrailer(analyzer.startxref);
}
analyzer.root = analyzer.trailer.GetChildObject("Root");
analyzer.root.Type = PdfObjectTypes.Root;
analyzer.rootPages = analyzer.root.GetChildObject("Pages");
analyzer.isEncrypted = analyzer.trailer.GetChildObject("Encrypt") != null;
analyzer.pageCount = analyzer.rootPages.GetInt32Value("Count").Value;
return(analyzer);
}
public override void AddToMerger(PdfMerger pdfMerger)
{
using (Stream s = new FileStream(this.Path, FileMode.Open, FileAccess.Read, FileShare.Read))
{
PdfParser pa = PdfParser.Parse(s);
pdfMerger.Add(pa, this.CalculatePageSelection(true));
}
}
internal PdfFileObject CreateFileObject(PdfParser analyzer)
{
PdfFileObject pfo = new PdfFileObject(analyzer);
pfo.ObjID = this.ObjID;
pfo.OriginalObjectAbsolutePosition = this.Address;
pfo.OriginalObjectLength = this.Length;
pfo.StreamPosition = this.StreamPosition;
pfo.StreamLength = this.StreamLength;
return(pfo);
}
public void GenerateSplit(int n, string target, PdfParser parsedPdf, int page)
{
target = Path.Combine(Path.GetDirectoryName(target),
Path.GetFileNameWithoutExtension(target) + " - " + n.ToString("00000") + ".pdf");
using (Stream s = new FileStream(target, FileMode.Create, FileAccess.Write))
{
this.merger = new PdfMerger(s);
this.merger.PdfMergerProgress += new EventHandler <PdfMergerProgressEventArgs>(merger_PdfMergerProgress);
this.merger.Add(parsedPdf, new int[] { page });
this.merger.Finish();
}
}
public override void Analyze()
{
using (Stream s = new FileStream(this.Path, FileMode.Open, FileAccess.Read))
{
PdfParser pfa = PdfParser.Parse(s);
if (pfa.IsEncrypted)
{
throw new PdfEncryptedException(string.Format("{0} is an Encrypted PDF."
, System.IO.Path.GetFileName(this.Path)));
}
//if (pfa.Version > 1.4)
//{
// throw new PdfDamagedException("Not supported v1.5");
//}
//if (pfa.MissingRoot)
//{
// throw new PdfDamagedException("Unable to find root node");
//}
this._pagecount = pfa.PageCount;
}
}
public void GenerateSplit(int n, string target, PdfParser parsedPdf, int page)
{
target = Path.Combine(Path.GetDirectoryName(target),
Path.GetFileNameWithoutExtension(target) + " - " + n.ToString("00000") + ".pdf");
using (Stream s = new FileStream(target, FileMode.Create, FileAccess.Write))
{
this.merger = new PdfMerger(s);
this.merger.PdfMergerProgress += new EventHandler<PdfMergerProgressEventArgs>(merger_PdfMergerProgress);
this.merger.Add(parsedPdf, new int[] { page });
this.merger.Finish();
}
}
internal PdfFileObject(PdfParser pdfFileParser)
{
this.pdfFileParser = pdfFileParser;
}
public override void AddToMerger(PdfMerger pdfMerger)
{
pdfMerger.Add(PdfParser.Parse(ImageToPDFConverter.ConvertImageToPDF(this.Path)));
}
internal PdfFileObject CreateFileObject(PdfParser analyzer)
{
PdfFileObject pfo = new PdfFileObject(analyzer);
pfo.ObjID = this.ObjID;
pfo.OriginalObjectAbsolutePosition = this.Address;
pfo.OriginalObjectLength = this.Length;
pfo.StreamPosition = this.StreamPosition;
pfo.StreamLength = this.StreamLength;
return pfo;
}
public void Add(PdfParser pdfParser, int[] PageNumbers)
{
if (this.CancelPending)
return;
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
0, 0, "Analyzing PDF Structure"));
PdfFileObject[] pages = pdfParser.GetAllPages();
ArrayList selectedPages = new ArrayList();
#region gets needed objects
Hashtable relatedObjects = new Hashtable();
if (PageNumbers == null)
{
List<int> ps = new List<int>();
for (int p = 0; p < pages.Length; p++)
{
ps.Add(p);
}
PageNumbers = ps.ToArray();
}
int currentPageIndex = 1, pageCount = PageNumbers.Length;
int step = Math.Max(pageCount / 20, 10);
foreach (int pageNumber in PageNumbers)
{
if (this.CancelPending)
return;
if (currentPageIndex % step == 0 || currentPageIndex == pageCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
currentPageIndex, pageCount, "Analyzing Page {0} of {1}"));
}
PdfFileObject selectedPage = pages[pageNumber];
selectedPages.Add(selectedPage);
selectedPage.PopulateRelatedObjects(relatedObjects);
currentPageIndex++;
}
ArrayList neededObjects = new ArrayList();
neededObjects.AddRange(relatedObjects.Values);
neededObjects.Sort(new PdfFileObjectNumberComparer());
#endregion
#region creates IDs transformation table
int objectIndex = 1;
int objectCount = neededObjects.Count;
step = Math.Max(objectCount / 20, 50);
Hashtable transformationTable = new Hashtable();
foreach (PdfFileObject pfo in neededObjects)
{
if (this.CancelPending)
return;
if (objectIndex % step == 0 || objectIndex == objectCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
objectIndex, objectCount, "Rebuilding Indexes for object {0} of {1}"));
}
string hash = OptimizeStreams?pfo.Hash:null;
if (hash != null && this.alreadyUsedObjects.ContainsKey(hash))
{
pfo.excludedByHashComparison = true;
transformationTable.Add(pfo.ObjID, (int)this.alreadyUsedObjects[hash]);
}
else
{
number++;
transformationTable.Add(pfo.ObjID, number);
if (hash != null)
{
this.alreadyUsedObjects.Add(hash, number);
}
}
objectIndex++;
}
#endregion
objectIndex = 1;
foreach (PdfFileObject pfo in neededObjects)
{
if (this.CancelPending)
return;
if (objectIndex % step == 0 || objectIndex==objectCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
objectIndex, objectCount, "Writing object {0} of {1}"));
}
pfo.Transform(transformationTable);
if (!pfo.excludedByHashComparison)
{
this.xrefs.Add(pos);
this.pos += pfo.WriteToStream(this.target);
}
pfo.TransformatedTextPart = null;
objectIndex++;
}
foreach (PdfFileObject selectedPage in selectedPages)
{
this.pageNumbers.Add(selectedPage.TransformatedObjID);
}
this.elementCount++;
}
public void Add(PdfParser pdfParser)
{
this.Add(pdfParser, null);
}
public void Add(PdfParser pdfParser, int[] PageNumbers)
{
if (this.CancelPending)
{
return;
}
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
0, 0, "Analyzing PDF Structure"));
PdfFileObject[] pages = pdfParser.GetAllPages();
ArrayList selectedPages = new ArrayList();
#region gets needed objects
Hashtable relatedObjects = new Hashtable();
if (PageNumbers == null)
{
List <int> ps = new List <int>();
for (int p = 0; p < pages.Length; p++)
{
ps.Add(p);
}
PageNumbers = ps.ToArray();
}
int currentPageIndex = 1, pageCount = PageNumbers.Length;
int step = Math.Max(pageCount / 20, 10);
foreach (int pageNumber in PageNumbers)
{
if (this.CancelPending)
{
return;
}
if (currentPageIndex % step == 0 || currentPageIndex == pageCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
currentPageIndex, pageCount, "Analyzing Page {0} of {1}"));
}
PdfFileObject selectedPage = pages[pageNumber];
selectedPages.Add(selectedPage);
selectedPage.PopulateRelatedObjects(relatedObjects);
currentPageIndex++;
}
ArrayList neededObjects = new ArrayList();
neededObjects.AddRange(relatedObjects.Values);
neededObjects.Sort(new PdfFileObjectNumberComparer());
#endregion
#region creates IDs transformation table
int objectIndex = 1;
int objectCount = neededObjects.Count;
step = Math.Max(objectCount / 20, 50);
Hashtable transformationTable = new Hashtable();
foreach (PdfFileObject pfo in neededObjects)
{
if (this.CancelPending)
{
return;
}
if (objectIndex % step == 0 || objectIndex == objectCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
objectIndex, objectCount, "Rebuilding Indexes for object {0} of {1}"));
}
string hash = OptimizeStreams?pfo.Hash:null;
if (hash != null && this.alreadyUsedObjects.ContainsKey(hash))
{
pfo.excludedByHashComparison = true;
transformationTable.Add(pfo.ObjID, (int)this.alreadyUsedObjects[hash]);
}
else
{
number++;
transformationTable.Add(pfo.ObjID, number);
if (hash != null)
{
this.alreadyUsedObjects.Add(hash, number);
}
}
objectIndex++;
}
#endregion
objectIndex = 1;
foreach (PdfFileObject pfo in neededObjects)
{
if (this.CancelPending)
{
return;
}
if (objectIndex % step == 0 || objectIndex == objectCount)
{
this.PdfMergerProgress(this, new PdfMergerProgressEventArgs(this.elementCount,
objectIndex, objectCount, "Writing object {0} of {1}"));
}
pfo.Transform(transformationTable);
if (!pfo.excludedByHashComparison)
{
this.xrefs.Add(pos);
this.pos += pfo.WriteToStream(this.target);
}
pfo.TransformatedTextPart = null;
objectIndex++;
}
foreach (PdfFileObject selectedPage in selectedPages)
{
this.pageNumbers.Add(selectedPage.TransformatedObjID);
}
this.elementCount++;
}
public void Add(PdfParser pdfParser)
{
this.Add(pdfParser, null);
}
internal PdfFileObject(PdfParser pdfFileParser)
{
this.pdfFileParser = pdfFileParser;
}
static void Main(string[] args)
{
Console.WriteLine("GIOS PDF Splitter and Merger - Version "
+ Assembly.GetEntryAssembly().GetName().Version);
Console.WriteLine("Copyright © 2009 Paolo Gios - www.paologios.com");
Console.WriteLine();
Console.WriteLine("Merges PDF and JPG files into a single PDF file.");
Console.WriteLine();
try
{
if (args.Length == 0)
{
Help();
#if DEBUG
Console.ReadLine();
#endif
return;
}
string output = "";
for (int index = 0; index < args.Length - 1; index++)
{
string f = args[index].ToLower();
if (f == "output")
{
output = args[index + 1];
break;
}
}
if (output == null)
{
return;
}
using (FileStream outputStream = new FileStream(output, FileMode.Create, FileAccess.Write))
{
Stopwatch sw = new Stopwatch();
sw.Start();
Console.WriteLine("Creating " + output + " ...");
PdfMerger merger = new PdfMerger(outputStream);
merger.PdfMergerProgress += new EventHandler <PdfMergerProgressEventArgs>(merger_PdfMergerProgress);
for (int index = 0; index < args.Length; index++)
{
string f = args[index].ToLower();
if (f == "output")
{
index++;
continue;
}
string directory = Path.GetDirectoryName(f);
if (string.IsNullOrEmpty(directory))
{
directory = Environment.CurrentDirectory;
}
string[] files = Directory.GetFiles(directory, Path.GetFileName(f), SearchOption.TopDirectoryOnly);
Array.Sort(files);
foreach (string file in files)
{
if (string.Compare(file, outputStream.Name) == 0)
{
continue;
}
string extension = Path.GetExtension(file).ToLower();
Stream fs = null;
switch (extension)
{
case ".pdf":
fs = new FileStream(file, FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
break;
case ".jpg":
fs = ImageToPDFConverter.ConvertImageToPDF(file);
break;
default:
continue;
}
using (fs)
{
Console.WriteLine();
Console.WriteLine("Adding " + file + "... ");
PdfParser pa = PdfParser.Parse(fs);
merger.Add(pa, null);
Console.SetCursorPosition(0, Console.CursorTop);
Console.WriteLine("Done. ");
}
}
}
Console.WriteLine();
Console.Write("Finishing ... ");
merger.Finish();
sw.Stop();
Console.WriteLine("Finished in " + sw.ElapsedMilliseconds + " ms.");
}
}
catch (Exception e)
{
Console.WriteLine(e.Message);
}
#if DEBUG
Console.ReadLine();
#endif
}
