public PdfDocument()
: base(null)
{
Version = new PdfVersion(this, 0, 0);
IndirectObjects = new PdfIndirectObjects(this);
DecryptHandler = new PdfDecryptNone(this);
}
public void Visit(PdfVersion version)
{
Append($"{version}");
}
public void Load(Stream stream, bool immediate = false, byte[] bytes = null)
{
if (_open)
{
throw new ApplicationException("Document already has a stream open.");
}
_stream = stream;
_parser = new Parser(_stream);
_parser.ResolveReference += Parser_ResolveReference;
// PDF file should have a well known marker at top of file
_parser.ParseHeader(out int versionMajor, out int versionMinor);
Version = new PdfVersion(this, versionMajor, versionMinor);
// Find stream position of the last cross-reference table
long xRefPosition = _parser.ParseXRefOffset();
bool lastHeader = true;
do
{
// Get the aggregated set of entries from all the cross-reference table sections
List <TokenXRefEntry> xrefs = _parser.ParseXRef(xRefPosition);
// Should always be positioned at the trailer after parsing cross-table references
PdfDictionary trailer = new PdfDictionary(this, _parser.ParseTrailer());
PdfInteger size = trailer.MandatoryValue <PdfInteger>("Size");
foreach (TokenXRefEntry xref in xrefs)
{
// Ignore unused entries and entries smaller than the defined size from the trailer dictionary
if (xref.Used && (xref.Id < size.Value))
{
IndirectObjects.AddXRef(xref);
}
}
if (lastHeader)
{
// Replace the default decryption handler with one from the document settings
DecryptHandler = PdfDecrypt.CreateDecrypt(this, trailer);
// We only care about the latest defined catalog and information dictionary
_refCatalog = trailer.MandatoryValue <PdfObjectReference>("Root");
_refInfo = trailer.OptionalValue <PdfObjectReference>("Info");
}
// If there is a previous cross-reference table, then we want to process that as well
PdfInteger prev = trailer.OptionalValue <PdfInteger>("Prev");
if (prev != null)
{
xRefPosition = prev.Value;
}
else
{
xRefPosition = 0;
}
lastHeader = false;
} while (xRefPosition > 0);
_open = true;
// Must load all objects immediately so the stream can then be closed
if (immediate)
{
// Is there enough work to justify using multiple threads
if ((bytes != null) && (IndirectObjects.Count > BACKGROUND_TRIGGER))
{
// Setup the synchronization event so we wait until all work is completed
_backgroundCount = NUM_BACKGROUND_ITEMS;
_backgroundEvent = new ManualResetEvent(false);
List <int> ids = IndirectObjects.Ids.ToList();
int idCount = ids.Count;
int batchSize = idCount / NUM_BACKGROUND_ITEMS;
for (int i = 0, index = 0; i < NUM_BACKGROUND_ITEMS; i++, index += batchSize)
{
// Create a parser per unit of work, so they can work in parallel
MemoryStream memoryStream = new MemoryStream(bytes);
Parser parser = new Parser(memoryStream);
// Make sure the last batch includes all the remaining Ids
ThreadPool.QueueUserWorkItem(BackgroundResolveReference, new BackgroundArgs()
{
Parser = parser,
Ids = ids,
Index = index,
Count = (i == (NUM_BACKGROUND_ITEMS - 1)) ? idCount - index : batchSize
});
}
_backgroundEvent.WaitOne();
_backgroundEvent.Dispose();
_backgroundEvent = null;
}
else
{
IndirectObjects.ResolveAllReferences(this);
}
Close();
}
}
