"Fossies" - the Fresh Open Source Software Archive  

Source code changes of the file "pdfbox/src/main/java/org/apache/pdfbox/pdfparser/BaseParser.java" between
pdfbox-2.0.23-src.zip and pdfbox-2.0.24-src.zip

About: Apache PDFBox is a Java PDF library tool that allows creation of new PDF documents, manipulation of existing documents and the ability to extract content from documents.

BaseParser.java  (pdfbox-2.0.23-src)	:	BaseParser.java  (pdfbox-2.0.24-src)
skipping to change at line 156		skipping to change at line 156
* @return The parsed Dictionary object.		* @return The parsed Dictionary object.
*		*
* @throws IOException If there is an error parsing the dictionary object.		* @throws IOException If there is an error parsing the dictionary object.
*/		*/
private COSBase parseCOSDictionaryValue() throws IOException		private COSBase parseCOSDictionaryValue() throws IOException
{		{
long numOffset = seqSource.getPosition();		long numOffset = seqSource.getPosition();
COSBase value = parseDirObject();		COSBase value = parseDirObject();
skipSpaces();		skipSpaces();
// proceed if the given object is a number and the following is a number as well		// proceed if the given object is a number and the following is a number as well
if (!(value instanceof COSNumber) || !isDigit())		if ((!(value instanceof COSNumber) || !isDigit()))
{		{
return value;		return value;
}		}
// read the remaining information of the object number		// read the remaining information of the object number
long genOffset = seqSource.getPosition();		long genOffset = seqSource.getPosition();
COSBase generationNumber = parseDirObject();		COSBase generationNumber = parseDirObject();
skipSpaces();		skipSpaces();
readExpectedChar('R');		readExpectedChar('R');
if (!(value instanceof COSInteger))		if (!(value instanceof COSInteger))
{		{
skipping to change at line 215		skipping to change at line 215
while (!done)		while (!done)
{		{
skipSpaces();		skipSpaces();
char c = (char) seqSource.peek();		char c = (char) seqSource.peek();
if (c == '>')		if (c == '>')
{		{
done = true;		done = true;
}		}
else if (c == '/')		else if (c == '/')
{		{
parseCOSDictionaryNameValuePair(obj);		// something went wrong, most likely the dictionary is corrupted
		// stop immediately and return everything read so far
		if (!parseCOSDictionaryNameValuePair(obj))
		{
		return obj;
		}
}		}
else		else
{		{
// invalid dictionary, we were expecting a /Name, read until the end or until we can recover		// invalid dictionary, we were expecting a /Name, read until the end or until we can recover
LOG.warn("Invalid dictionary, found: '" + c + "' but expected: ' /' at offset " + seqSource.getPosition());		LOG.warn("Invalid dictionary, found: '" + c + "' but expected: ' /' at offset " + seqSource.getPosition());
if (readUntilEndOfCOSDictionary())		if (readUntilEndOfCOSDictionary())
{		{
// we couldn't recover		// we couldn't recover
return obj;		return obj;
}		}
skipping to change at line 279		skipping to change at line 284
c = seqSource.read();		c = seqSource.read();
}		}
if (c == -1)		if (c == -1)
{		{
return true;		return true;
}		}
seqSource.unread(c);		seqSource.unread(c);
return false;		return false;
}		}
private void parseCOSDictionaryNameValuePair(COSDictionary obj) throws IOExc eption		private boolean parseCOSDictionaryNameValuePair(COSDictionary obj) throws IO Exception
{		{
COSName key = parseCOSName();		COSName key = parseCOSName();
COSBase value = parseCOSDictionaryValue();		COSBase value = parseCOSDictionaryValue();
skipSpaces();		skipSpaces();
if (((char) seqSource.peek()) == 'd')
{
// if the next string is 'def' then we are parsing a cmap stream
// and want to ignore it, otherwise throw an exception.
String potentialDEF = readString();
if (!potentialDEF.equals(DEF))
{
seqSource.unread(potentialDEF.getBytes(ISO_8859_1));
}
else
{
skipSpaces();
}
}
if (value == null)		if (value == null)
{		{
LOG.warn("Bad dictionary declaration at offset " + seqSource.getPosi tion());		LOG.warn("Bad dictionary declaration at offset " + seqSource.getPosi tion());
		return false;
		}
		else if (value instanceof COSInteger && !((COSInteger)value).isValid())
		{
		LOG.warn("Skipped out of range number value at offset " + seqSource.
		getPosition());
}		}
else		else
{		{
// label this item as direct, to avoid signature problems.		// label this item as direct, to avoid signature problems.
value.setDirect(true);		value.setDirect(true);
obj.setItem(key, value);		obj.setItem(key, value);
}		}
		return true;
}		}
protected void skipWhiteSpaces() throws IOException		protected void skipWhiteSpaces() throws IOException
{		{
//PDF Ref 3.2.7 A stream must be followed by either		//PDF Ref 3.2.7 A stream must be followed by either
//a CRLF or LF but nothing else.		//a CRLF or LF but nothing else.
int whitespace = seqSource.read();		int whitespace = seqSource.read();
//see brother_scan_cover.pdf, it adds whitespaces		//see brother_scan_cover.pdf, it adds whitespaces
skipping to change at line 662		skipping to change at line 658
pbo = null;		pbo = null;
}		}
}		}
if( pbo != null )		if( pbo != null )
{		{
po.add( pbo );		po.add( pbo );
}		}
else		else
{		{
//it could be a bad object in the array which is just skipped		//it could be a bad object in the array which is just skipped
LOG.warn("Corrupt object reference at offset " +		LOG.warn("Corrupt array element at offset "
seqSource.getPosition() + ", start offset: " + startPosi		+ seqSource.getPosition() + ", start offset: " + startPo
tion);		sition);
// This could also be an "endobj" or "endstream" which means we
can assume that
// the array has ended.
String isThisTheEnd = readString();		String isThisTheEnd = readString();
		// return immediately if a corrupt element is followed by anothe
		r array
		// to avoid a possible infinite recursion as most likely the who
		le array is corrupted
		if (isThisTheEnd.isEmpty() && seqSource.peek() == '[')
		{
		return po;
		}
seqSource.unread(isThisTheEnd.getBytes(ISO_8859_1));		seqSource.unread(isThisTheEnd.getBytes(ISO_8859_1));
		// This could also be an "endobj" or "endstream" which means we
		can assume that
		// the array has ended.
if(ENDOBJ_STRING.equals(isThisTheEnd) || ENDSTREAM_STRING.equals (isThisTheEnd))		if(ENDOBJ_STRING.equals(isThisTheEnd) || ENDSTREAM_STRING.equals (isThisTheEnd))
{		{
return po;		return po;
}		}
}		}
skipSpaces();		skipSpaces();
}		}
// read ']'		// read ']'
seqSource.read();		seqSource.read();
skipSpaces();		skipSpaces();
skipping to change at line 692		skipping to change at line 693
/**		/**
* Determine if a character terminates a PDF name.		* Determine if a character terminates a PDF name.
*		*
* @param ch The character		* @param ch The character
* @return true if the character terminates a PDF name, otherwise false.		* @return true if the character terminates a PDF name, otherwise false.
*/		*/
protected boolean isEndOfName(int ch)		protected boolean isEndOfName(int ch)
{		{
return ch == ASCII_SPACE || ch == ASCII_CR || ch == ASCII_LF || ch == 9 || ch == '>' ||		return ch == ASCII_SPACE || ch == ASCII_CR || ch == ASCII_LF || ch == 9 || ch == '>' ||
ch == '<' || ch == '[' || ch =='/' || ch ==']' || ch ==')' || ch =='(' ||		ch == '<' || ch == '[' || ch =='/' || ch ==']' || ch ==')' || ch =='(' ||
ch == 0 || ch == '\f';		ch == 0 || ch == '\f' || ch == '%';
}		}
/**		/**
* This will parse a PDF name from the stream.		* This will parse a PDF name from the stream.
*		*
* @return The parsed PDF name.		* @return The parsed PDF name.
* @throws IOException If there is an error reading from the stream.		* @throws IOException If there is an error reading from the stream.
*/		*/
protected COSName parseCOSName() throws IOException		protected COSName parseCOSName() throws IOException
{		{
skipping to change at line 855		skipping to change at line 856
skipSpaces();		skipSpaces();
char c = (char)seqSource.peek();		char c = (char)seqSource.peek();
switch(c)		switch(c)
{		{
case '<':		case '<':
// pull off first left bracket		// pull off first left bracket
int leftBracket = seqSource.read();		int leftBracket = seqSource.read();
// check for second left bracket		// check for second left bracket
c = (char) seqSource.peek();		c = (char) seqSource.peek();
seqSource.unread(leftBracket);		seqSource.unread(leftBracket);
if(c == '<')		return c == '<' ? parseCOSDictionary() : parseCOSString();
{
COSDictionary retval = parseCOSDictionary();
skipSpaces();
return retval;
}
else
{
return parseCOSString();
}
case '[':		case '[':
// array		// array
return parseCOSArray();		return parseCOSArray();
case '(':		case '(':
return parseCOSString();		return parseCOSString();
case '/':		case '/':
// name		// name
return parseCOSName();		return parseCOSName();
case 'n':		case 'n':
// null		// null
skipping to change at line 910		skipping to change at line 901
case 'R':		case 'R':
seqSource.read();		seqSource.read();
return new COSObject(null);		return new COSObject(null);
case (char)-1:		case (char)-1:
return null;		return null;
default:		default:
if( Character.isDigit(c) || c == '-' || c == '+' || c == '.')		if( Character.isDigit(c) || c == '-' || c == '+' || c == '.')
{		{
return parseCOSNumber();		return parseCOSNumber();
}		}
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
		long startOffset = seqSource.getPosition();
		String badString = readString();
		if (badString.isEmpty())
{		{
//This is not suppose to happen, but we will allow for it		int peek = seqSource.peek();
//so we are more compatible with POS writers that don't		// we can end up in an infinite loop otherwise
//follow the spec		throw new IOException(
String badString = readString();		"Unknown dir object c='" + c + "' cInt=" + (int) c + " p
if (badString.isEmpty())		eek='" + (char) peek
{		+ "' peekInt=" + peek + " at offset " + seqSource.getPos
int peek = seqSource.peek();		ition()
// we can end up in an infinite loop otherwise		+ " (start offset: " + startOffset + ")");
throw new IOException( "Unknown dir object c='" + c +		}
"' cInt=" + (int)c + " peek='" + (char)peek
+ "' peekInt=" + peek + " at offset " + seqSource.ge
tPosition() );
}
// if it's an endstream/endobj, we want to put it back so the ca		// if it's an endstream/endobj, we want to put it back so the caller
ller will see it		will see it
if(ENDOBJ_STRING.equals(badString) || ENDSTREAM_STRING.equals(ba		if (ENDOBJ_STRING.equals(badString) || ENDSTREAM_STRING.equals(badSt
dString))		ring))
{		{
seqSource.unread(badString.getBytes(ISO_8859_1));		seqSource.unread(badString.getBytes(ISO_8859_1));
}		}
else		else
{		{
LOG.warn("Skipped unexpected dir object = '" + badString + "		LOG.warn("Skipped unexpected dir object = '" + badString + "' at
' at offset "		offset "
+ seqSource.getPosition());		+ seqSource.getPosition() + " (start offset: " + startOf
}		fset + ")");
}		}
}		}
return null;		return null;
}		}
private COSNumber parseCOSNumber() throws IOException		private COSNumber parseCOSNumber() throws IOException
{		{
StringBuilder buf = new StringBuilder();		StringBuilder buf = new StringBuilder();
int ic = seqSource.read();		int ic = seqSource.read();
char c = (char) ic;		char c = (char) ic;
End of changes. 14 change blocks.
64 lines changed or deleted		59 lines changed or added

---