public bool MoveNext()
{
// Read until we find object-number generation obj, e.g. "69 420 obj".
int tokensRead = 0;
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
continue;
}
tokensRead++;
previousTokens[0] = previousTokens[1];
previousTokenPositions[0] = previousTokenPositions[1];
previousTokens[1] = coreTokenScanner.CurrentToken;
previousTokenPositions[1] = coreTokenScanner.CurrentTokenStart;
}
// We only read partial tokens.
if (tokensRead < 2)
{
return(false);
}
var startPosition = previousTokenPositions[0];
var objectNumber = previousTokens[0] as NumericToken;
var generation = previousTokens[1] as NumericToken;
if (objectNumber == null || generation == null)
{
throw new PdfDocumentFormatException("The obj operator (start object) was not preceded by a 2 numbers." +
$"Instead got: {previousTokens[0]} {previousTokens[1]} obj");
}
// Read all tokens between obj and endobj.
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
continue;
}
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
// This should never happen.
Debug.Assert(false, "Encountered a start object 'obj' operator before the end of the previous object.");
return(false);
}
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartStream))
{
// Read stream: special case.
if (TryReadStream(coreTokenScanner.CurrentTokenStart, out var stream))
{
readTokens.Clear();
readTokens.Add(stream);
}
}
else
{
readTokens.Add(coreTokenScanner.CurrentToken);
}
previousTokens[0] = previousTokens[1];
previousTokenPositions[0] = previousTokenPositions[1];
previousTokens[1] = coreTokenScanner.CurrentToken;
previousTokenPositions[1] = coreTokenScanner.CurrentPosition;
}
if (!ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
readTokens.Clear();
return(false);
}
var reference = new IndirectReference(objectNumber.Long, generation.Int);
IToken token;
if (readTokens.Count == 3 && readTokens[0] is NumericToken objNum &&
readTokens[1] is NumericToken genNum &&
ReferenceEquals(readTokens[2], OperatorToken.R))
{
// I have no idea if this can ever happen.
token = new IndirectReferenceToken(new IndirectReference(objNum.Long, genNum.Int));
}
public bool MoveNext()
{
if (isDisposed)
{
throw new ObjectDisposedException(nameof(PdfTokenScanner));
}
// Read until we find object-number generation obj, e.g. "69 420 obj".
int tokensRead = 0;
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
continue;
}
tokensRead++;
previousTokens[0] = previousTokens[1];
previousTokenPositions[0] = previousTokenPositions[1];
previousTokens[1] = coreTokenScanner.CurrentToken;
previousTokenPositions[1] = coreTokenScanner.CurrentTokenStart;
}
// We only read partial tokens.
if (tokensRead < 2)
{
return(false);
}
var startPosition = previousTokenPositions[0];
var objectNumber = previousTokens[0] as NumericToken;
var generation = previousTokens[1] as NumericToken;
if (objectNumber == null || generation == null)
{
// Handle case where the scanner correctly reads most of an object token but includes too much of the first token
// specifically %%EOF1 0 obj where scanning starts from 'F'.
if (generation != null && previousTokens[0] is OperatorToken op)
{
var match = EndsWithNumberRegex.Match(op.Data);
if (match.Success && int.TryParse(match.Value, NumberStyles.Any, CultureInfo.InvariantCulture, out var number))
{
startPosition = previousTokenPositions[0] + match.Index;
objectNumber = new NumericToken(number);
}
else
{
return(false);
}
}
else
{
return(false);
}
}
// Read all tokens between obj and endobj.
while (coreTokenScanner.MoveNext() && !Equals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
if (coreTokenScanner.CurrentToken is CommentToken)
{
continue;
}
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartObject))
{
// This should never happen.
Debug.Assert(false, "Encountered a start object 'obj' operator before the end of the previous object.");
return(false);
}
if (ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.StartStream))
{
var streamIdentifier = new IndirectReference(objectNumber.Long, generation.Int);
// Prevent an infinite loop where a stream's length references the stream or the stream's offset.
var getLengthFromFile = !(callingObject.HasValue && callingObject.Value.Equals(streamIdentifier));
var outerCallingObject = callingObject;
try
{
callingObject = streamIdentifier;
// Read stream: special case.
if (TryReadStream(coreTokenScanner.CurrentTokenStart, getLengthFromFile, out var stream))
{
readTokens.Clear();
readTokens.Add(stream);
}
}
finally
{
callingObject = outerCallingObject;
}
}
else
{
readTokens.Add(coreTokenScanner.CurrentToken);
}
previousTokens[0] = previousTokens[1];
previousTokenPositions[0] = previousTokenPositions[1];
previousTokens[1] = coreTokenScanner.CurrentToken;
previousTokenPositions[1] = coreTokenScanner.CurrentPosition;
}
if (!ReferenceEquals(coreTokenScanner.CurrentToken, OperatorToken.EndObject))
{
readTokens.Clear();
return(false);
}
var reference = new IndirectReference(objectNumber.Long, generation.Int);
IToken token;
if (readTokens.Count == 3 && readTokens[0] is NumericToken objNum &&
readTokens[1] is NumericToken genNum &&
ReferenceEquals(readTokens[2], OperatorToken.R))
{
// I have no idea if this can ever happen.
token = new IndirectReferenceToken(new IndirectReference(objNum.Long, genNum.Int));
}