[18170] | 1 | package org.greenstone.gatherer.util;
|
---|
| 2 |
|
---|
| 3 | import java.io.Reader;
|
---|
| 4 | import java.io.IOException;
|
---|
| 5 |
|
---|
| 6 | public class RemoveContentBeforeRootElementXMLReader extends Reader {
|
---|
| 7 |
|
---|
| 8 | static final byte[] xmlIndicator = "<?xml".getBytes();
|
---|
| 9 |
|
---|
| 10 | Reader ur;
|
---|
| 11 | int foundBytesReturned = 0;
|
---|
| 12 | boolean allFoundBytesReturned = false; //redundant, but may help performance
|
---|
| 13 |
|
---|
| 14 |
|
---|
| 15 | public RemoveContentBeforeRootElementXMLReader( Reader ur ) {
|
---|
| 16 |
|
---|
| 17 | this.ur = ur;
|
---|
| 18 |
|
---|
| 19 | //read up to the xml indicator
|
---|
[18195] | 20 | int foundBytes = 0;
|
---|
| 21 | int c = 0;
|
---|
[18170] | 22 | while ( c != -1 ) {
|
---|
| 23 | try {
|
---|
[18195] | 24 | c = ur.read();
|
---|
[18170] | 25 | } catch( Exception e ) {
|
---|
| 26 | System.err.println( "Exception while reading underlying Reader in RemoveContentBeforeRootElementXMLReader" );
|
---|
| 27 | }
|
---|
[18195] | 28 |
|
---|
| 29 | if ( c == (int)xmlIndicator[foundBytes] ) {
|
---|
[18170] | 30 | foundBytes++;
|
---|
[18195] | 31 | } else {
|
---|
| 32 | if ( foundBytes != 0 ) {
|
---|
| 33 | for ( int i=0; i<foundBytes; i++ ) {
|
---|
| 34 | System.out.print( (char)xmlIndicator[i] );
|
---|
| 35 | }
|
---|
[18170] | 36 | foundBytes = 0;
|
---|
[18195] | 37 | }
|
---|
| 38 | if ( c != -1 ) {
|
---|
| 39 | System.out.print( (char)c );
|
---|
| 40 | }
|
---|
[18170] | 41 | }
|
---|
| 42 |
|
---|
| 43 | if ( foundBytes == xmlIndicator.length ) {
|
---|
[18195] | 44 | return;
|
---|
[18170] | 45 | }
|
---|
| 46 |
|
---|
| 47 | }
|
---|
[18195] | 48 |
|
---|
[18170] | 49 | System.err.println( "RemoveContentBeforeRootElementXMLReader:\n" +
|
---|
[18195] | 50 | "The XML being loaded did not contain the '<?xml' string as expected" );
|
---|
[18170] | 51 | }
|
---|
| 52 |
|
---|
| 53 | public int read( char[] cbuf, int off, int len ) throws IOException {
|
---|
| 54 |
|
---|
| 55 | for ( int i=off; i<off+len && i<cbuf.length; i++ ) {
|
---|
| 56 |
|
---|
| 57 | //read from underlying reader
|
---|
| 58 | int c = read();
|
---|
| 59 |
|
---|
| 60 | //catch end of stream
|
---|
| 61 | if ( c == -1 ) {
|
---|
| 62 | if ( i == off ) {
|
---|
| 63 | return -1;
|
---|
| 64 | }
|
---|
| 65 | return i - off;
|
---|
| 66 | }
|
---|
| 67 |
|
---|
| 68 | //insert character into the array
|
---|
| 69 | cbuf[i] = (char)c;
|
---|
| 70 | }
|
---|
| 71 | return len;
|
---|
| 72 |
|
---|
| 73 | }
|
---|
| 74 |
|
---|
| 75 | public int read() throws IOException {
|
---|
| 76 |
|
---|
| 77 | if ( allFoundBytesReturned ) {
|
---|
| 78 | return ur.read();
|
---|
| 79 | }
|
---|
| 80 |
|
---|
| 81 | if ( foundBytesReturned < xmlIndicator.length ) {
|
---|
| 82 | return xmlIndicator[foundBytesReturned++];
|
---|
| 83 | }
|
---|
| 84 |
|
---|
| 85 | allFoundBytesReturned = true;
|
---|
| 86 | return ur.read();
|
---|
| 87 | }
|
---|
| 88 |
|
---|
| 89 | public void close() throws IOException {
|
---|
| 90 | ur.close();
|
---|
| 91 | }
|
---|
| 92 | }
|
---|