public TrailerPart ParseTrailer(PdfStreamReader streamReader, IPdf pdf)
{
var header = streamReader.ReadLine();
if (header != "trailer")
{
throw new ArgumentOutOfRangeException(nameof(streamReader), "First line in Trailer Content is not \"trailer\"");
}
var trailerDictionary = this._pdfReader.ReadObject(streamReader, pdf);
if (!(trailerDictionary is PdfDictionary))
{
throw new InvalidDataException("Trailer dictionary content was not a dictionary.");
}
var line = streamReader.ReadLine();
var foundCrossReference = false;
if (line == "startxref")
{
foundCrossReference = true;
}
if (!foundCrossReference)
{
throw new InvalidDataException("Missing startxref line");
}
var crossReferenceLine = streamReader.ReadLine();
if (!long.TryParse(crossReferenceLine, out var crossReferenceOffset))
{
throw new InvalidDataException("Cross Reference Offset value is not a parseable number");
}
line = streamReader.ReadLine();
if (line == null ||
line == "%%EOF")
{
var result = new TrailerPart((PdfDictionary)trailerDictionary, crossReferenceOffset);
return(result);
}
throw new InvalidDataException("Trailer content is invalid, the last 2 lines should be startxref and a number and optionally a 3rd line containing %%EOF");
}
public VersionPart ReadPart(PdfStreamReader streamReader, IPdf pdf)
{
var line = streamReader.ReadLine();
if (line == null)
{
throw new ArgumentOutOfRangeException(nameof(streamReader), "Unexpected end of stream");
}
else if (line.StartsWith(PdfVersionPrefix))
{
var result = this.GetVersionPart(line);
return(result);
}
else
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Expected content at current position was not a version");
}
}
public CrossReferencePart ReadPart(PdfStreamReader streamReader, IPdf pdf)
{
if (pdf == null)
{
throw new ArgumentNullException(nameof(pdf));
}
if (pdf.Trailer == null ||
pdf.Trailer.StartCrossReference == 0)
{
return(null);
}
var originalPosition = streamReader.Position;
try
{
streamReader.Seek(pdf.Trailer.StartCrossReference, SeekOrigin.Begin);
var line = streamReader.ReadLine();
if (line != "xref")
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid Cross Reference Header");
}
var subsections = new List <CrossReferencePartSubSection>();
while (line == "xref")
{
subsections.Add(_subSectionReader.ReadPart(streamReader, pdf));
line = streamReader.ReadLine();
}
var result = new CrossReferencePart
{
Sections = subsections.ToArray()
};
return(result);
}
finally
{
streamReader.Seek(originalPosition, SeekOrigin.Begin);
}
}
public CrossReferencePartSubSectionEntry ReadPart(PdfStreamReader streamReader, IPdf pdf)
{
var line = streamReader.ReadLine();
if (line.Length != 18)
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Line is not 18 characters long.");
}
var fields = line.Split(' ');
if (fields.Length != 3)
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Line does not have 3 fields.");
}
if (!long.TryParse(fields[0], out var offset))
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid offset field. Not a valid number.");
}
if (!int.TryParse(fields[1], out var generation))
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid generation field. Not a valid number.");
}
if (fields[2] != "f" &&
fields[2] != "n")
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid in use flag. Not 'n' or 'f'.");
}
var inUseFlag = (CrossReferenceEntryFlag)fields[2][0];
var result = new CrossReferencePartSubSectionEntry
{
Generation = generation,
InUse = inUseFlag,
Offset = offset
};
return(result);
}
public CrossReferencePartSubSection ReadPart(PdfStreamReader streamReader, IPdf pdf)
{
var line = streamReader.ReadLine();
var fields = line.Split(' ');
if (fields.Length != 2)
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid field count for subsection counts.");
}
if (!int.TryParse(fields[0], out var startingObject))
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid starting object field. Not a valid number.");
}
if (!int.TryParse(fields[1], out var objectCount))
{
throw new ArgumentOutOfRangeException(nameof(streamReader), line, "Invalid object count. Not a valid number.");
}
var entries = new CrossReferencePartSubSectionEntry[objectCount];
for (var i = 0; i < entries.Length; i++)
{
var entry = this._partReader.ReadPart(streamReader, pdf);
entry.ObjectNumber = startingObject + i;
entries[i] = entry;
}
var result = new CrossReferencePartSubSection
{
Entries = entries,
StartingObject = startingObject
};
return(result);
}
public IPdf Create(Stream stream, PdfOptions pdfOptions)
{
if (stream == null)
{
throw new ArgumentNullException(nameof(stream));
}
if (!stream.CanSeek)
{
throw new InvalidDataException("The stream must be seekable");
}
if (stream.Length == 0)
{
throw new InvalidDataException("Stream is 0 bytes.");
}
stream.Seek(0, SeekOrigin.Begin);
var streamReader = new PdfStreamReader(stream, pdfOptions.ReadBlockSize);
var pdf = new DefaultPdf(this._pdfReader, streamReader)
{
Options = pdfOptions
};
pdf.Version = this._versionPartReader.ReadPart(streamReader, pdf);
pdf.Trailer = this._trailerPartReader.ReadPart(streamReader, pdf);
if (pdf.Version >= new Version(1, 4))
{
//TODO: spec section 7.5.2 states that if the version in the first line is >= 1.4, use the version in the trailer if it exists
}
pdf.CrossReferenceTable = this._crossReferencePartReader.ReadPart(streamReader, pdf);
if (pdf.Trailer.Encrypt != null)
{
PdfDictionary encryptDictionary;
if (pdf.Trailer.Encrypt is PdfReference o)
{
var readObject = pdf.ReadIndirectObject(o);
if (readObject == null)
{
throw new InvalidDataException("Missing encrypt object");
}
if (!(readObject.InnerObject is PdfDictionary))
{
throw new InvalidDataException("Encrypt object is not a pdf dictionary");
}
encryptDictionary = (PdfDictionary)readObject.InnerObject;
}
else if (pdf.Trailer.Encrypt is PdfDictionary)
{
encryptDictionary = (PdfDictionary)pdf.Trailer.Encrypt;
}
else
{
throw new InvalidDataException("Invalid encrypt dictionary entry");
}
pdf.EncryptionDictionary = new EncryptionDictionary(encryptDictionary);
}
return(pdf);
}
public TrailerPart ReadPart(PdfStreamReader streamReader, IPdf pdf)
{
var currentPosition = streamReader.Position;
try
{
var trailerBlockSize = 512;
var block = new byte[trailerBlockSize];
int trailerStart;
//we'll read x number of bytes at a time with a 7 byte overlap to account for the word trailer.
//length = 1500
streamReader.Seek(-trailerBlockSize, SeekOrigin.End);
string content;
var foundTrailer = false;
do
{
var loopPosition = streamReader.Position;
streamReader.Read(block, 0, trailerBlockSize);
content = Encoding.ASCII.GetString(block);
if ((trailerStart = content.IndexOf("trailer")) >= 0)
{
streamReader.Seek(loopPosition + trailerStart, SeekOrigin.Begin);
foundTrailer = true;
break;
}
content = null;
if (loopPosition == 0)
{
//we were at the beginning of the stream meaning we can't go back any more
break;
}
if (streamReader.Position > trailerBlockSize)
{
streamReader.Seek((trailerBlockSize * -2) + 7, SeekOrigin.Current); //7 byte overlap
}
else
{
streamReader.Seek(0, SeekOrigin.Begin);
}
} while (streamReader.Position - trailerBlockSize > 0); //make sure we don't go before the beginning of the stream for files that have no trailer
if (foundTrailer)
{
var result = ParseTrailer(streamReader, pdf);
return(result);
}
return(null);
}
finally
{
// go back to where we were.
streamReader.Seek(currentPosition, SeekOrigin.Begin);
}
}