/** Make sure the lzo index class works as described. */
@Test
public void testLzoIndex() {
LzoIndex index = new LzoIndex();
assertTrue(index.isEmpty());
index = new LzoIndex(4);
index.set(0, 0);
index.set(1, 5);
index.set(2, 10);
index.set(3, 15);
assertFalse(index.isEmpty());
assertEquals(0, index.findNextPosition(-1));
assertEquals(5, index.findNextPosition(1));
assertEquals(5, index.findNextPosition(5));
assertEquals(15, index.findNextPosition(11));
assertEquals(15, index.findNextPosition(15));
assertEquals(-1, index.findNextPosition(16));
assertEquals(5, index.alignSliceStartToIndex(3, 20));
assertEquals(15, index.alignSliceStartToIndex(15, 20));
assertEquals(10, index.alignSliceEndToIndex(8, 30));
assertEquals(10, index.alignSliceEndToIndex(10, 30));
assertEquals(30, index.alignSliceEndToIndex(17, 30));
assertEquals(LzoIndex.NOT_FOUND, index.alignSliceStartToIndex(16, 20));
}
@Override
public InputSplit[] getSplits(JobConf conf, int numSplits) throws IOException {
FileSplit[] splits = (FileSplit[]) super.getSplits(conf, numSplits);
// Find new starts/ends of the filesplit that align with the LZO blocks.
List<FileSplit> result = new ArrayList<FileSplit>();
for (FileSplit fileSplit : splits) {
Path file = fileSplit.getPath();
FileSystem fs = file.getFileSystem(conf);
LzoIndex index = indexes.get(file);
if (index == null) {
throw new IOException("Index not found for " + file);
}
if (index.isEmpty()) {
// Empty index, keep it as is.
result.add(fileSplit);
continue;
}
long start = fileSplit.getStart();
long end = start + fileSplit.getLength();
long lzoStart = index.alignSliceStartToIndex(start, end);
long lzoEnd = index.alignSliceEndToIndex(end, fs.getFileStatus(file).getLen());
if (lzoStart != LzoIndex.NOT_FOUND && lzoEnd != LzoIndex.NOT_FOUND) {
result.add(new FileSplit(file, lzoStart, lzoEnd - lzoStart, fileSplit.getLocations()));
}
}
return result.toArray(new FileSplit[result.size()]);
}
