public List <TokenXRefEntry> ParseXRef()
{
Tokenizer.IgnoreComments = true;
TokenObject t = Tokenizer.GetToken();
TokenKeyword keyword = t as TokenKeyword;
if ((keyword == null) || (keyword.Value != ParseKeyword.XRef))
{
// Scan entire source creating XRef entries for each indirect object
return(IndirectObjectsToXRef());
}
else
{
List <TokenXRefEntry> entries = new List <TokenXRefEntry>();
ParseXRefSections(entries);
return(entries);
}
}
public List <TokenXRefEntry> IndirectObjectsToXRef()
{
List <TokenXRefEntry> entries = new List <TokenXRefEntry>();
// Start scanning from beginning of the source
Tokenizer.IgnoreComments = false;
Tokenizer.Position = 0;
long lastTrailer = -1;
do
{
long t1Position = Tokenizer.Position;
TokenObject t1 = Tokenizer.GetToken();
if (t1 is TokenInteger)
{
TokenInteger t2 = Tokenizer.GetToken() as TokenInteger;
if (t2 != null)
{
TokenKeyword t3 = Tokenizer.GetToken() as TokenKeyword;
if ((t3 != null) && (t3.Value == ParseKeyword.Obj))
{
TokenInteger id = (TokenInteger)t1;
entries.Add(new TokenXRefEntry(id.Value, t2.Value, t1Position, true));
}
}
}
else if ((t1 is TokenKeyword) && ((TokenKeyword)t1).Value == ParseKeyword.Trailer)
{
lastTrailer = t1Position;
}
} while (Tokenizer.GotoNextLine());
// Leave with the position on the last 'trailer' as caller will then parse it
if (lastTrailer >= 0)
{
Tokenizer.Position = lastTrailer;
}
return(entries);
}
private TokenObject GetKeywordOrIdentifier()
{
long position = _position + _index;
// Scan looking for the end of the keyword characters
int key = _index;
while ((key < _length) && _lookupKeyword[_line[key]])
{
key++;
}
string text = Encoding.ASCII.GetString(_line, _index, key - _index);
TokenKeyword token = TokenKeyword.GetToken(text);
if (token != null)
{
_index = key;
return(token);
}
if (AllowIdentifiers)
{
// Scan looking for the end of the identifier
key = _index;
while ((key < _length) && !_lookupDelimiterWhitespace[_line[key]])
{
key++;
}
text = Encoding.ASCII.GetString(_line, _index, key - _index);
_index = key;
return(new TokenIdentifier(text));
}
else
{
return(new TokenError(position, $"Cannot parse '{text}' as a keyword."));
}
}
public ParseIndirectObject ParseIndirectObject()
{
Tokenizer.IgnoreComments = true;
TokenObject t = Tokenizer.GetToken();
ThrowOnEmptyOrError(t);
// Indirect object starts with an integer, the object identifier
if (!(t is TokenInteger))
{
Tokenizer.PushToken(t);
return(null);
}
// Second is another integer, the generation number
TokenObject u = Tokenizer.GetToken();
ThrowOnEmptyOrError(u);
if (!(u is TokenInteger))
{
Tokenizer.PushToken(t);
Tokenizer.PushToken(u);
return(null);
}
// This is the keyword 'obj'
TokenObject v = Tokenizer.GetToken();
ThrowOnEmptyOrError(v);
if (!(v is TokenKeyword) || ((v as TokenKeyword).Value != ParseKeyword.Obj))
{
Tokenizer.PushToken(t);
Tokenizer.PushToken(u);
Tokenizer.PushToken(v);
return(null);
}
// Get actual object that is the content
ParseObjectBase obj = ParseObject();
if (obj == null)
{
throw new ApplicationException($"Indirect object has missing content.");
}
// Must be followed by either 'endobj' or 'stream'
v = Tokenizer.GetToken();
ThrowOnEmptyOrError(v);
TokenKeyword keyword = v as TokenKeyword;
if (keyword == null)
{
// PH: FIXME:
//throw new ApplicationException($"Indirect object has missing 'endobj or 'stream'.");
return(new ParseIndirectObject(t as TokenInteger, u as TokenInteger, new ParseInteger(0)));
}
if (keyword.Value == ParseKeyword.EndObj)
{
return(new ParseIndirectObject(t as TokenInteger, u as TokenInteger, obj));
}
else if (keyword.Value == ParseKeyword.Stream)
{
ParseDictionary dictionary = obj as ParseDictionary;
if (dictionary == null)
{
throw new ApplicationException($"Stream must be preceded by a dictionary.");
}
if (!dictionary.ContainsName("Length"))
{
throw new ApplicationException($"Stream dictionary must contain a 'Length' entry.");
}
ParseObjectBase lengthObj = dictionary["Length"];
// Resolve any object reference
ParseObjectReference reference = lengthObj as ParseObjectReference;
if (reference != null)
{
lengthObj = OnResolveReference(reference);
}
ParseInteger length = lengthObj as ParseInteger;
if (length == null)
{
throw new ApplicationException($"Stream dictionary has a 'Length' entry that is not an integer entry.");
}
if (length.Value < 0)
{
throw new ApplicationException($"Stream dictionary has a 'Length' less than 0.");
}
byte[] bytes = Tokenizer.GetBytes(length.Value);
if (bytes == null)
{
throw new ApplicationException($"Cannot read in expected {length.Value} bytes from stream.");
}
// Stream contents must be followed by 'endstream'
v = Tokenizer.GetToken();
ThrowOnEmptyOrError(v);
keyword = v as TokenKeyword;
if (keyword == null)
{
throw new ApplicationException($"Stream has missing 'endstream' after content.");
}
if (keyword.Value != ParseKeyword.EndStream)
{
throw new ApplicationException($"Stream has unexpected keyword {keyword.Value} instead of 'endstream'.");
}
// Stream contents must be followed by 'endobj'
v = Tokenizer.GetToken();
ThrowOnEmptyOrError(v);
keyword = v as TokenKeyword;
if (keyword == null)
{
throw new ApplicationException($"Indirect object has missing 'endobj'.");
}
if (keyword.Value != ParseKeyword.EndObj)
{
throw new ApplicationException($"Indirect object has unexpected keyword {keyword.Value} instead of 'endobj'.");
}
return(new ParseIndirectObject(t as TokenInteger, u as TokenInteger, new ParseStream(dictionary, bytes)));
}
else
{
throw new ApplicationException($"Indirect object has unexpected keyword {keyword.Value}.");
}
}