protected void initStream(InputSplit inSplit) throws IOException {
file = ((FileSplit) inSplit).getPath();
FSDataInputStream fileIn = fs.open(file);
String codecString =
conf.get(ConfigConstants.CONF_INPUT_COMPRESSION_CODEC, CompressionCodec.ZIP.toString())
.toUpperCase();
codec = CompressionCodec.valueOf(codecString);
switch (codec) {
case ZIP:
zipIn = new ZipInputStream(fileIn);
break;
case GZIP:
zipIn = new GZIPInputStream(fileIn);
String uri = makeURIFromPath(file);
if (uri == null) {
key = null;
} else {
if (uri.toLowerCase().endsWith(".gz") || uri.toLowerCase().endsWith(".gzip")) {
uri = uri.substring(0, uri.lastIndexOf('.'));
}
setKey(uri);
}
break;
default:
String error = "Unsupported codec: " + codec.name();
LOG.error(error, new UnsupportedOperationException(error));
}
}
@Override
public boolean nextKeyValue() throws IOException, InterruptedException {
if (zipIn == null) {
hasNext = false;
return false;
}
if (codec == CompressionCodec.ZIP) {
ZipEntry zipEntry;
ZipInputStream zis = (ZipInputStream) zipIn;
while ((zipEntry = zis.getNextEntry()) != null) {
if (zipEntry != null && zipEntry.getSize() != 0) {
String uri = makeURIForZipEntry(file, zipEntry.getName());
if (uri != null) {
setKey(uri);
setValue(zipEntry.getSize());
} else {
key = null;
}
return true;
}
}
} else if (codec == CompressionCodec.GZIP) {
setValue(0);
zipIn.close();
zipIn = null;
hasNext = false;
return true;
} else {
throw new UnsupportedOperationException("Unsupported codec: " + codec.name());
}
if (iterator != null && iterator.hasNext()) {
close();
initStream(iterator.next());
return nextKeyValue();
} else {
hasNext = false;
return false;
}
}
