Index: pdfbox/src/main/java/org/apache/pdfbox/pdfparser/BaseParser.java =================================================================== --- developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/BaseParser.java (revision 1467796) +++ developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/BaseParser.java (working copy) @@ -54,6 +54,10 @@ public abstract class BaseParser { + private static final long OBJECT_NUMBER_THRESHOLD = 10000000000L; + + private static final long GENERATION_NUMBER_THRESHOLD = 65535; + /** * system property allowing to define size of push back buffer. */ @@ -1585,6 +1589,36 @@ } /** + * This will read a long from the Stream and throw an {@link IllegalArgumentException} if the long value + * has more than 10 digits (i.e. : bigger than {@link #OBJECT_NUMBER_THRESHOLD}) + * @return + * @throws IOException + */ + protected long readObjectNumber() throws IOException + { + long retval = readLong(); + if(Math.abs(retval) >= OBJECT_NUMBER_THRESHOLD) { // an object number may be negative + throw new IllegalArgumentException("Object Number '" + retval + "' has more than 10 digits"); + } + return retval; + } + + /** + * This will read a integer from the Stream and throw an {@link IllegalArgumentException} if the integer value + * has more than the maximum object revision (i.e. : bigger than {@link #GENERATION_NUMBER_THRESHOLD}) + * @return + * @throws IOException + */ + protected int readGenerationNumber() throws IOException + { + int retval = readInt(); + if(retval < 0 || retval >= GENERATION_NUMBER_THRESHOLD) { + throw new IllegalArgumentException("Generation Number '" + retval + "' has more than 5 digits"); + } + return retval; + } + + /** * This will read an integer from the stream. * * @return The integer that was read from the stream. @@ -1596,8 +1630,57 @@ skipSpaces(); int retval = 0; + StringBuffer intBuffer = readStringNumber(); + + try + { + retval = Integer.parseInt( intBuffer.toString() ); + } + catch( NumberFormatException e ) + { + pdfSource.unread(intBuffer.toString().getBytes("ISO-8859-1")); + throw new IOException( "Error: Expected an integer type, actual='" + intBuffer + "'" ); + } + return retval; + } + + + /** + * This will read an long from the stream. + * + * @return The long that was read from the stream. + * + * @throws IOException If there is an error reading from the stream. + */ + protected long readLong() throws IOException + { + skipSpaces(); + long retval = 0; + + StringBuffer longBuffer = readStringNumber(); + + try + { + retval = Long.parseLong( longBuffer.toString() ); + } + catch( NumberFormatException e ) + { + pdfSource.unread(longBuffer.toString().getBytes("ISO-8859-1")); + throw new IOException( "Error: Expected a long type, actual='" + longBuffer + "'" ); + } + return retval; + } + + /** + * This method is used to read a token by the {@linkplain #readInt()} method and the {@linkplain #readLong()} method. + * + * @return the token to parse as integer or long by the calling method. + * @throws IOException throws by the {@link #pdfSource} methods. + */ + private StringBuffer readStringNumber() throws IOException + { int lastByte = 0; - StringBuffer intBuffer = new StringBuffer(); + StringBuffer buffer = new StringBuffer(); while( (lastByte = pdfSource.read() ) != 32 && lastByte != 10 && lastByte != 13 && @@ -1605,22 +1688,13 @@ lastByte != 0 && //See sourceforge bug 853328 lastByte != -1 ) { - intBuffer.append( (char)lastByte ); + buffer.append( (char)lastByte ); } if( lastByte != -1 ) { pdfSource.unread( lastByte ); } - - try - { - retval = Integer.parseInt( intBuffer.toString() ); - } - catch( NumberFormatException e ) - { - pdfSource.unread(intBuffer.toString().getBytes("ISO-8859-1")); - throw new IOException( "Error: Expected an integer type, actual='" + intBuffer + "'" ); - } - return retval; + return buffer; } + } Index: pdfbox/src/main/java/org/apache/pdfbox/pdfparser/NonSequentialPDFParser.java =================================================================== --- developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/NonSequentialPDFParser.java (revision 1488059) +++ developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/NonSequentialPDFParser.java (working copy) @@ -455,8 +455,8 @@ private long parseXrefObjStream(long objByteOffset) throws IOException { // ---- parse indirect object head - readInt(); - readInt(); + readObjectNumber(); + readGenerationNumber(); readPattern(OBJ_MARKER); COSDictionary dict = parseCOSDictionary(); @@ -1180,8 +1180,8 @@ setPdfSource(offsetOrObjstmObNr); // ---- we must have an indirect object - final int readObjNr = readInt(); - final int readObjGen = readInt(); + final long readObjNr = readObjectNumber(); + final long readObjGen = readGenerationNumber(); readPattern(OBJ_MARKER); // ---- consistency check Index: pdfbox/src/main/java/org/apache/pdfbox/pdfparser/PDFObjectStreamParser.java =================================================================== --- developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/PDFObjectStreamParser.java (revision 1467796) +++ developpement/java/workspace/pdfbox-trunk/pdfbox/src/main/java/org/apache/pdfbox/pdfparser/PDFObjectStreamParser.java (working copy) @@ -44,7 +44,7 @@ LogFactory.getLog(PDFObjectStreamParser.class); private List streamObjects = null; - private List objectNumbers = null; + private List objectNumbers = null; private COSStream stream; /** @@ -92,13 +92,13 @@ { //need to first parse the header. int numberOfObjects = stream.getInt( "N" ); - objectNumbers = new ArrayList( numberOfObjects ); + objectNumbers = new ArrayList( numberOfObjects ); streamObjects = new ArrayList( numberOfObjects ); for( int i=0; i