source: gli/branches/rtl-gli/src/org/greenstone/gatherer/util/RemoveContentBeforeRootElementXMLReader.java@ 18353

package org.greenstone.gatherer.util;

import java.io.Reader;
import java.io.IOException;

public class RemoveContentBeforeRootElementXMLReader extends Reader {

    static final byte[] xmlIndicator = "<?xml".getBytes();

    Reader ur;
    int foundBytesReturned = 0;
    boolean allFoundBytesReturned = false; //redundant, but may help performance


    public RemoveContentBeforeRootElementXMLReader( Reader ur ) {

        this.ur = ur;

        //read up to the xml indicator
        int foundBytes = 0;
        int c = 0;
        while ( c != -1 ) {
            try {
                c = ur.read();
            } catch( Exception e ) {
                System.err.println( "Exception while reading underlying Reader in RemoveContentBeforeRootElementXMLReader" );
            }
            
            if ( c == (int)xmlIndicator[foundBytes] ) {
                foundBytes++;
            } else {                
                if ( foundBytes != 0 ) {
                for ( int i=0; i<foundBytes; i++ ) {
                    System.out.print( (char)xmlIndicator[i] );
                }
                foundBytes = 0;
                }
                if ( c != -1 ) {
                System.out.print( (char)c );
                }
            }

            if ( foundBytes == xmlIndicator.length ) {
                return;
            }

        }
        
        System.err.println( "RemoveContentBeforeRootElementXMLReader:\n" +
            "The XML being loaded did not contain the '<?xml' string as expected" );
    }

    public int read( char[] cbuf, int off, int len ) throws IOException {

        for ( int i=off; i<off+len && i<cbuf.length; i++ ) {

            //read from underlying reader
            int c = read();

            //catch end of stream
            if ( c == -1 ) {
                if ( i == off ) {
                    return -1;
                }
                return i - off;
            }

            //insert character into the array
            cbuf[i] = (char)c;
        }
        return len;

    }

    public int read() throws IOException {

        if ( allFoundBytesReturned ) {
            return ur.read();
        }

        if ( foundBytesReturned < xmlIndicator.length ) {
            return xmlIndicator[foundBytesReturned++];
        }

        allFoundBytesReturned = true;
        return ur.read();
    }

    public void close() throws IOException {
        ur.close();
    }
}