protected void skipWhiteSpaces(IRandomAccessRead reader)
{
//PDF Ref 3.2.7 A stream must be followed by either
//a CRLF or LF but nothing else.
int whitespace = reader.Read();
//see brother_scan_cover.pdf, it adds whitespaces
//after the stream but before the start of the
//data, so just read those first
while (whitespace == ' ')
{
whitespace = reader.Read();
}
if (whitespace == ReadHelper.AsciiCarriageReturn)
{
whitespace = reader.Read();
if (whitespace != ReadHelper.AsciiLineFeed)
{
reader.Unread(whitespace);
//The spec says this is invalid but it happens in the real
//world so we must support it.
}
}
else if (whitespace != ReadHelper.AsciiLineFeed)
{
//we are in an error.
//but again we will do a lenient parsing and just assume that everything
//is fine
reader.Unread(whitespace);
}
}
public void Unread(int b)
{
if (Throw)
{
throw new InvalidOperationException();
}
reader.Unread(b);
}
private static StringBuilder ReadStringNumber(IRandomAccessRead reader)
{
int lastByte = 0;
StringBuilder buffer = new StringBuilder();
while ((lastByte = reader.Read()) != ' ' &&
lastByte != AsciiLineFeed &&
lastByte != AsciiCarriageReturn &&
lastByte != 60 && //see sourceforge bug 1714707
lastByte != '[' && // PDFBOX-1845
lastByte != '(' && // PDFBOX-2579
lastByte != 0 && //See sourceforge bug 853328
lastByte != -1)
{
buffer.Append((char)lastByte);
if (buffer.Length > long.MaxValue.ToString("D").Length)
{
throw new IOException("Number '" + buffer + "' is getting too long, stop reading at offset " + reader.GetPosition());
}
}
if (lastByte != -1)
{
reader.Unread(lastByte);
}
return(buffer);
}
private static int CheckForEndOfString(IRandomAccessRead reader, int bracesParameter)
{
int braces = bracesParameter;
byte[] nextThreeBytes = new byte[3];
int amountRead = reader.Read(nextThreeBytes);
// Check the next 3 bytes if available
// The following cases are valid indicators for the end of the string
// 1. Next line contains another COSObject: CR + LF + '/'
// 2. CosDictionary ends in the next line: CR + LF + '>'
// 3. Next line contains another COSObject: CR + '/'
// 4. CosDictionary ends in the next line: CR + '>'
if (amountRead == 3 && nextThreeBytes[0] == ReadHelper.AsciiCarriageReturn)
{
if (nextThreeBytes[1] == ReadHelper.AsciiLineFeed && nextThreeBytes[2] == '/' || nextThreeBytes[2] == '>' ||
nextThreeBytes[1] == '/' || nextThreeBytes[1] == '>')
{
braces = 0;
}
}
if (amountRead > 0)
{
reader.Unread(nextThreeBytes, 0, amountRead);
}
return(braces);
}
public static void SkipSpaces(IRandomAccessRead reader)
{
const int commentCharacter = 37;
int c = reader.Read();
while (IsWhitespace(c) || c == 37)
{
if (c == commentCharacter)
{
// skip past the comment section
c = reader.Read();
while (!IsEndOfLine(c) && c != -1)
{
c = reader.Read();
}
}
else
{
c = reader.Read();
}
}
if (c != -1)
{
reader.Unread(c);
}
}
private (CosName key, CosBase value) ParseCosDictionaryNameValuePair(IRandomAccessRead reader, IBaseParser baseParser, CosObjectPool pool)
{
var key = nameParser.Parse(reader);
var value = ParseValue(reader, baseParser, pool);
ReadHelper.SkipSpaces(reader);
if ((char)reader.Peek() == 'd')
{
// if the next string is 'def' then we are parsing a cmap stream
// and want to ignore it, otherwise throw an exception.
var potentialDef = ReadHelper.ReadString(reader);
if (!potentialDef.Equals("def"))
{
reader.Unread(OtherEncodings.StringAsLatin1Bytes(potentialDef));
}
else
{
ReadHelper.SkipSpaces(reader);
}
}
if (value == null)
{
log?.Warn("Bad Dictionary Declaration " + ReadHelper.ReadString(reader));
return(null, null);
}
// label this item as direct, to avoid signature problems.
value.Direct = true;
return(key, value);
}
public static string ReadString(IRandomAccessRead reader)
{
SkipSpaces(reader);
StringBuilder buffer = new StringBuilder();
int c = reader.Read();
while (!IsEndOfName((char)c) && c != -1)
{
buffer.Append((char)c);
c = reader.Read();
}
if (c != -1)
{
reader.Unread(c);
}
return(buffer.ToString());
}
public static long ReadLong(IRandomAccessRead reader)
{
SkipSpaces(reader);
long retval;
StringBuilder longBuffer = ReadStringNumber(reader);
try
{
retval = long.Parse(longBuffer.ToString());
}
catch (FormatException e)
{
var bytesToReverse = OtherEncodings.StringAsLatin1Bytes(longBuffer.ToString());
reader.Unread(bytesToReverse);
throw new InvalidOperationException($"Error: Expected a long type at offset {reader.GetPosition()}, instead got \'{longBuffer}\'", e);
}
return(retval);
}
private static bool ReadUntilEnd(IRandomAccessRead reader)
{
var c = reader.Read();
while (c != -1 && c != '/' && c != '>')
{
// in addition to stopping when we find / or >, we also want
// to stop when we find endstream or endobj.
if (c == E)
{
c = reader.Read();
if (c == N)
{
c = reader.Read();
if (c == D)
{
c = reader.Read();
var isStream = c == S && reader.Read() == T && reader.Read() == R &&
reader.Read() == E && reader.Read() == A && reader.Read() == M;
var isObj = !isStream && c == O && reader.Read() == B && reader.Read() == J;
if (isStream || isObj)
{
// we're done reading this object!
return(true);
}
}
}
}
c = reader.Read();
}
if (c == -1)
{
return(true);
}
reader.Unread(c);
return(false);
}
public static int ReadInt(IRandomAccessRead reader)
{
if (reader == null)
{
throw new ArgumentNullException(nameof(reader));
}
SkipSpaces(reader);
int result;
var intBuffer = ReadStringNumber(reader);
try
{
result = int.Parse(intBuffer.ToString());
}
catch (Exception e)
{
reader.Unread(OtherEncodings.StringAsLatin1Bytes(intBuffer.ToString()));
throw new IOException("Error: Expected an integer type at offset " + reader.GetPosition(), e);
}
return(result);
}
public CosName Parse([NotNull] IRandomAccessRead reader)
{
if (reader == null)
{
throw new ArgumentNullException(nameof(reader));
}
ReadHelper.ReadExpectedChar(reader, '/');
using (var memoryStream = new MemoryStream())
using (var writer = new BinaryWriter(memoryStream))
{
int c = reader.Read();
while (c != -1)
{
byte ch = (byte)c;
if (ch == '#')
{
int ch1 = reader.Read();
int ch2 = reader.Read();
// Prior to PDF v1.2, the # was not a special character. Also,
// it has been observed that various PDF tools do not follow the
// spec with respect to the # escape, even though they report
// PDF versions of 1.2 or later. The solution here is that we
// interpret the # as an escape only when it is followed by two
// valid hex digits.
if (ReadHelper.IsHexDigit((char)ch1) && ReadHelper.IsHexDigit((char)ch2))
{
string hex = "" + (char)ch1 + (char)ch2;
try
{
var byteToWrite = (byte)Convert.ToInt32(hex, 16);
writer.Write(byteToWrite);
}
catch (FormatException e)
{
throw new IOException("Error: expected hex digit, actual='" + hex + "'", e);
}
c = reader.Read();
}
else
{
// check for premature EOF
if (ch2 == -1 || ch1 == -1)
{
//LOG.error("Premature EOF in BaseParser#parseCosName");
c = -1;
break;
}
reader.Unread(ch2);
c = ch1;
writer.Write(ch);
}
}
else if (ReadHelper.IsEndOfName(ch))
{
break;
}
else
{
writer.Write(ch);
c = reader.Read();
}
}
if (c != -1)
{
reader.Unread(c);
}
byte[] bytes = memoryStream.ToArray();
var str = ReadHelper.IsValidUtf8(bytes) ? Encoding.UTF8.GetString(memoryStream.ToArray()) : Encoding.GetEncoding("windows-1252").GetString(memoryStream.ToArray());
return(CosName.Create(str));
}
}
public CosString Parse(IRandomAccessRead seqSource)
{
char nextChar = (char)seqSource.Read();
if (nextChar == '<')
{
return(ParseHexString(seqSource));
}
if (nextChar != '(')
{
throw new IOException("parseCOSstring string should start with '(' or '<' and not '" +
nextChar + "' " + seqSource);
}
using (var memoryStream = new MemoryStream())
using (var writer = new StreamWriter(memoryStream))
{
// This is the number of braces read
int braces = 1;
int c = seqSource.Read();
while (braces > 0 && c != -1)
{
char ch = (char)c;
int nextc = -2; // not yet read
if (ch == ')')
{
braces--;
braces = CheckForEndOfString(seqSource, braces);
if (braces != 0)
{
writer.Write(ch);
}
}
else if (ch == '(')
{
braces++;
writer.Write(ch);
}
else if (ch == '\\')
{
//patched by ram
char next = (char)seqSource.Read();
switch (next)
{
case 'n':
writer.Write('\n');
break;
case 'r':
writer.Write('\r');
break;
case 't':
writer.Write('\t');
break;
case 'b':
writer.Write('\b');
break;
case 'f':
writer.Write('\f');
break;
case ')':
// PDFBox 276 /Title (c:\)
braces = CheckForEndOfString(seqSource, braces);
if (braces != 0)
{
writer.Write(next);
}
else
{
writer.Write('\\');
}
break;
case '(':
case '\\':
writer.Write(next);
break;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
{
var octal = new StringBuilder();
octal.Append(next);
c = seqSource.Read();
char digit = (char)c;
if (digit >= '0' && digit <= '7')
{
octal.Append(digit);
c = seqSource.Read();
digit = (char)c;
if (digit >= '0' && digit <= '7')
{
octal.Append(digit);
}
else
{
nextc = c;
}
}
else
{
nextc = c;
}
int character;
try
{
character = Convert.ToInt32(octal.ToString(), 8);
}
catch (FormatException e)
{
throw new IOException("Error: Expected octal character, actual='" + octal + "'", e);
}
writer.Write(character);
break;
}
default:
if (c == ReadHelper.AsciiCarriageReturn || c == ReadHelper.AsciiLineFeed)
{
// this is a break in the line so ignore it and the newline and continue
c = seqSource.Read();
while (ReadHelper.IsEndOfLine(c) && c != -1)
{
c = seqSource.Read();
}
nextc = c;
break;
}
// dropping the backslash
// see 7.3.4.2 Literal strings for further information
writer.Write(next);
break;
}
}
else
{
writer.Write(ch);
}
if (nextc != -2)
{
c = nextc;
}
else
{
c = seqSource.Read();
}
}
if (c != -1)
{
seqSource.Unread(c);
}
writer.Flush();
return(new CosString(memoryStream.ToArray()));
}
}
public COSArray Parse(IRandomAccessRead reader, CosBaseParser baseParser, CosObjectPool pool)
{
ReadHelper.ReadExpectedChar(reader, '[');
var po = new COSArray();
CosBase pbo;
ReadHelper.SkipSpaces(reader);
int i;
while (((i = reader.Peek()) > 0) && ((char)i != ']'))
{
pbo = baseParser.Parse(reader, pool);
if (pbo is CosObject)
{
// We have to check if the expected values are there or not PDFBOX-385
if (po.get(po.size() - 1) is CosInt)
{
var genNumber = (CosInt)po.remove(po.size() - 1);
if (po.get(po.size() - 1) is CosInt)
{
var number = (CosInt)po.remove(po.size() - 1);
CosObjectKey key = new CosObjectKey(number.AsLong(), genNumber.AsInt());
pbo = pool.Get(key);
}
else
{
// the object reference is somehow wrong
pbo = null;
}
}
else
{
pbo = null;
}
}
if (pbo != null)
{
po.add(pbo);
}
else
{
//it could be a bad object in the array which is just skipped
// LOG.warn("Corrupt object reference at offset " + seqSource.getPosition());
// This could also be an "endobj" or "endstream" which means we can assume that
// the array has ended.
string isThisTheEnd = ReadHelper.ReadString(reader);
reader.Unread(OtherEncodings.StringAsLatin1Bytes(isThisTheEnd));
if (string.Equals(isThisTheEnd, "endobj") || string.Equals(isThisTheEnd, "endstream"))
{
return(po);
}
}
ReadHelper.SkipSpaces(reader);
}
// read ']'
reader.Read();
ReadHelper.SkipSpaces(reader);
return(po);
}
public CosBase Parse(IRandomAccessRead reader, CosObjectPool pool)
{
CosBase retval = null;
ReadHelper.SkipSpaces(reader);
int nextByte = reader.Peek();
if (nextByte == -1)
{
return(null);
}
char c = (char)nextByte;
switch (c)
{
case '<':
{
// pull off first left bracket
int leftBracket = reader.Read();
// check for second left bracket
c = (char)reader.Peek();
reader.Unread(leftBracket);
if (c == '<')
{
retval = dictionaryParser.Parse(reader, this, pool);
ReadHelper.SkipSpaces(reader);
}
else
{
retval = stringParser.Parse(reader);
}
break;
}
case '[':
{
// array
retval = arrayParser.Parse(reader, this, pool);
break;
}
case '(':
retval = stringParser.Parse(reader);
break;
case '/':
// name
retval = nameParser.Parse(reader);
break;
case 'n':
{
// null
ReadHelper.ReadExpectedString(reader, "null");
retval = CosNull.Null;
break;
}
case 't':
{
string truestring = OtherEncodings.BytesAsLatin1String(reader.ReadFully(4));
if (truestring.Equals("true"))
{
retval = PdfBoolean.True;
}
else
{
throw new IOException("expected true actual='" + truestring + "' " + reader +
"' at offset " + reader.GetPosition());
}
break;
}
case 'f':
{
string falsestring = OtherEncodings.BytesAsLatin1String(reader.ReadFully(5));
if (falsestring.Equals("false"))
{
retval = PdfBoolean.False;
}
else
{
throw new IOException("expected false actual='" + falsestring + "' " + reader +
"' at offset " + reader.GetPosition());
}
break;
}
case 'R':
reader.Read();
retval = new CosObject(null);
break;
default:
if (char.IsDigit(c) || c == '-' || c == '+' || c == '.')
{
StringBuilder buf = new StringBuilder();
int ic = reader.Read();
c = (char)ic;
while (char.IsDigit(c) ||
c == '-' ||
c == '+' ||
c == '.' ||
c == 'E' ||
c == 'e')
{
buf.Append(c);
ic = reader.Read();
c = (char)ic;
}
if (ic != -1)
{
reader.Unread(ic);
}
retval = CosNumberFactory.get(buf.ToString()) as CosBase;
}
else
{
//This is not suppose to happen, but we will allow for it
//so we are more compatible with POS writers that don't
//follow the spec
string badstring = ReadHelper.ReadString(reader);
if (badstring == string.Empty)
{
int peek = reader.Peek();
// we can end up in an infinite loop otherwise
throw new IOException("Unknown dir object c='" + c +
"' cInt=" + (int)c + " peek='" + (char)peek
+ "' peekInt=" + peek + " at offset " + reader.GetPosition());
}
// if it's an endstream/endobj, we want to put it back so the caller will see it
if (string.Equals("endobj", badstring) || string.Equals("endstream", badstring))
{
reader.Unread(OtherEncodings.StringAsLatin1Bytes(badstring));
}
}
break;
}
return(retval);
}
