cmapparser.java

源码包含生成 PDF 和 HTML 的类库

                int theNextByte = is.read();
                if( theNextByte == '<' )
                {
                    Map result = new HashMap();
                    //we are reading a dictionary
                    Object key = parseNextToken( is ); 
                    while( key instanceof LiteralName && key != MARK_END_OF_DICTIONARY )
                    {
                        Object value = parseNextToken( is );
                        result.put( ((LiteralName)key).name, value );
                        key = parseNextToken( is );
                    }
                    retval = result;
                }
                else
                {
                    //won't read more than 512 bytes
                    
                    int multiplyer = 16;
                    int bufferIndex = -1;
                    while( theNextByte != -1 && theNextByte != '>' )
                    {
                        int intValue = 0;
                        if( theNextByte >= '0' && theNextByte <= '9' )
                        {
                            intValue = theNextByte - '0';
                        }
                        else if( theNextByte >= 'A' && theNextByte <= 'F' )
                        {
                            intValue = 10 + theNextByte - 'A';
                        }
                        else if( theNextByte >= 'a' && theNextByte <= 'f' )
                        {
                            intValue = 10 + theNextByte - 'a';
                        }
                        else
                        {
                            throw new IOException( "Error: expected hex character and not " + 
                                (char)theNextByte + ":" + theNextByte );
                        }
                        intValue *= multiplyer;
                        if( multiplyer == 16 )
                        {
                            bufferIndex++;
                            tokenParserByteBuffer[bufferIndex] = 0;
                            multiplyer = 1;
                        }
                        else
                        {
                            multiplyer = 16;
                        }
                        tokenParserByteBuffer[bufferIndex]+= intValue;
                        theNextByte = is.read();
                    }
                    byte[] finalResult = new byte[bufferIndex+1];
                    System.arraycopy(tokenParserByteBuffer,0,finalResult, 0, bufferIndex+1);
                    retval = finalResult;
                }
                break;
            }
            case '/':
            {
                StringBuffer buffer = new StringBuffer();
                int stringByte = is.read();
                
                while( !isWhitespaceOrEOF( stringByte ) )
                {
                    buffer.append( (char)stringByte );
                    stringByte = is.read();
                }
                retval = new LiteralName( buffer.toString() );
                break;
            }
            case -1:
            {
                //EOF return null;
                break;
            }
            case '0':
            case '1':
            case '2':
            case '3':
            case '4':
            case '5':
            case '6':
            case '7':
            case '8':
            case '9':
            {
                StringBuffer buffer = new StringBuffer();
                buffer.append( (char)nextByte );
                nextByte = is.read();
                
                while( !isWhitespaceOrEOF( nextByte ) &&
                        (Character.isDigit( (char)nextByte )||
                         nextByte == '.' ) )
                {
                    buffer.append( (char)nextByte );
                    nextByte = is.read();
                }
                is.unread( nextByte );
                String value = buffer.toString();
                if( value.indexOf( '.' ) >=0 )
                {
                    retval = new Double( value );
                }
                else
                {
                    retval = new Integer( buffer.toString() );
                }
                break;
            }
            default:
            {
                StringBuffer buffer = new StringBuffer();
                buffer.append( (char)nextByte );
                nextByte = is.read();
                
                while( !isWhitespaceOrEOF( nextByte ) )
                {
                    buffer.append( (char)nextByte );
                    nextByte = is.read();
                }
                retval = new Operator( buffer.toString() );                        
                
                break;
            }
        }
        return retval;
    }
    
    private void readUntilEndOfLine( InputStream is, StringBuffer buf ) throws IOException
    {
        int nextByte = is.read();
        while( nextByte != -1 && nextByte != 0x0D && nextByte != 0x0D )
        {
            buf.append( (char)nextByte );
            nextByte = is.read();
        }
    }
    
    private boolean isWhitespaceOrEOF( int aByte )
    {
        return aByte == -1 || aByte == 0x20 || aByte == 0x0D || aByte == 0x0A; 
    }
    

    private void increment( byte[] data )
    {
        increment( data, data.length-1 );
    }

    private void increment( byte[] data, int position )
    {
        if( position > 0 && (data[position]+256)%256 == 255 )
        {
            data[position]=0;
            increment( data, position-1);
        }
        else
        {
            data[position] = (byte)(data[position]+1);
        }
    }
    
    private String createStringFromBytes( byte[] bytes ) throws IOException
    {
        String retval = null;
        if( bytes.length == 1 )
        {
            retval = new String( bytes );
        }
        else
        {
            retval = new String( bytes, "UTF-16BE" );
        }
        return retval;
    }

    private int compare( byte[] first, byte[] second )
    {
        int retval = 1;
        boolean done = false;
        for( int i=0; i<first.length && !done; i++ )
        {
            if( first[i] == second[i] )
            {
                //move to next position
            }
            else if( ((first[i]+256)%256) < ((second[i]+256)%256) )
            {
                done = true;
                retval = -1;
            }
            else
            {
                done = true;
                retval = 1;
            }
        }
        return retval;
    }
    
    /**
     * Internal class.
     */
    private class LiteralName
    {
        private String name;
        private LiteralName( String theName )
        {
            name = theName;
        }
    }
    
    /**
     * Internal class.
     */
    private class Operator
    {
        private String op;
        private Operator( String theOp )
        {
            op = theOp;
        }
    }
    
    /**
     * A simple class to test parsing of cmap files.
     * 
     * @param args Some command line arguments.
     * 
     * @throws Exception If there is an error parsing the file.
     */
    public static void main( String[] args ) throws Exception
    {
        if( args.length != 1 )
        {
            System.err.println( "usage: java org.pdfbox.cmapparser.CMapParser <CMAP File>" );
            System.exit( -1 );
        }
        CMapParser parser = new CMapParser(  );
        CMap result = parser.parse( new FileInputStream( args[0] ) );
        System.out.println( "Result:" + result );
    }
}