public static void main(String[] args) throws Exception {
sourcePhoto = "/home/hduser/workspace/images/source.jpg";
sourceFingerprint = SimilarImageSearch.produceFingerPrint(sourcePhoto);
final Configuration conf = new Configuration();
MongoConfigUtil.setInputURI(conf, "mongodb://localhost/photo.fingerprint");
MongoConfigUtil.setOutputURI(conf, "mongodb://localhost/photo.handsomeOut");
System.out.println("Conf: " + conf);
final Job job = new Job(conf, "similar photo");
job.setJarByClass(MdbSimilarPhoto.class);
// Mapper,Reduce and Combiner type definition
job.setMapperClass(PhotoMapper.class);
job.setCombinerClass(SimilarityReducer.class);
job.setReducerClass(SimilarityReducer.class);
// output key/value type definition
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(DoubleWritable.class);
// InputFormat and OutputFormat type definition
job.setInputFormatClass(MongoInputFormat.class);
job.setOutputFormatClass(MongoOutputFormat.class);
System.exit(job.waitForCompletion(true) ? 0 : 1);
}
/**
* @param process
* @param tap
* @param conf
*/
@Override
public void sourceConfInit(
FlowProcess<JobConf> process, Tap<JobConf, RecordReader, OutputCollector> tap, JobConf conf) {
MongoConfigUtil.setReadSplitsFromShards(conf, true);
MongoConfigUtil.setInputURI(conf, this.mongoUri);
FileInputFormat.setInputPaths(conf, this.getIdentifier());
conf.setInputFormat(MongoInputFormat.class);
// TODO: MongoConfigUtil.setFields(conf, fieldsBson);
// if (!this.query.isEmpty())
MongoConfigUtil.setQuery(conf, this.query);
// TODO: MongoConfigUtil.setFields(conf, fields);
}
/**
* @param process
* @param tap
* @param conf
*/
@Override
public void sinkConfInit(
FlowProcess<JobConf> process, Tap<JobConf, RecordReader, OutputCollector> tap, JobConf conf) {
conf.setOutputFormat(MongoOutputFormat.class);
MongoConfigUtil.setOutputURI(conf, this.mongoUri);
FileOutputFormat.setOutputPath(conf, getPath());
}
public EnronMail() {
JobConf conf = new JobConf(new Configuration());
if (MongoTool.isMapRedV1()) {
MapredMongoConfigUtil.setInputFormat(
conf, com.mongodb.hadoop.mapred.BSONFileInputFormat.class);
MapredMongoConfigUtil.setOutputFormat(
conf, com.mongodb.hadoop.mapred.MongoOutputFormat.class);
} else {
MongoConfigUtil.setInputFormat(conf, BSONFileInputFormat.class);
MongoConfigUtil.setOutputFormat(conf, MongoOutputFormat.class);
}
FileInputFormat.addInputPath(conf, new Path("/messages"));
MongoConfig config = new MongoConfig(conf);
config.setInputKey("headers.From");
config.setMapper(EnronMailMapper.class);
config.setReducer(EnronMailReducer.class);
config.setMapperOutputKey(MailPair.class);
config.setMapperOutputValue(IntWritable.class);
config.setOutputKey(MailPair.class);
config.setOutputValue(IntWritable.class);
config.setOutputURI("mongodb://localhost:27017/mongo_hadoop.message_pairs");
setConf(conf);
}
public HadoopTest() throws UnknownHostException {
setConf(new Configuration());
if (MongoTool.isMapRedV1()) {
MapredMongoConfigUtil.setInputFormat(
getConf(), com.mongodb.hadoop.mapred.MongoInputFormat.class);
MapredMongoConfigUtil.setOutputFormat(
getConf(), com.mongodb.hadoop.mapred.MongoOutputFormat.class);
} else {
MongoConfigUtil.setInputFormat(getConf(), MongoInputFormat.class);
MongoConfigUtil.setOutputFormat(getConf(), MongoOutputFormat.class);
}
MongoConfigUtil.setInputURI(getConf(), "mongodb://localhost:27017/lolmaster.playerMatch");
MongoConfigUtil.setOutputURI(getConf(), "mongodb://localhost:27017/lolmaster.out");
MongoConfigUtil.setMapper(getConf(), HadoopMapper.class);
MongoConfigUtil.setReducer(getConf(), HadoopReducer.class);
MongoConfigUtil.setMapperOutputKey(getConf(), Text.class);
MongoConfigUtil.setMapperOutputValue(getConf(), Text.class);
MongoConfigUtil.setOutputKey(getConf(), IntWritable.class);
MongoConfigUtil.setOutputValue(getConf(), BSONWritable.class);
}
@Override
public RecordWriter<K, V> getRecordWriter(final TaskAttemptContext context) throws IOException {
// Open data output stream
Path outPath = getDefaultWorkFile(context, ".bson");
LOG.info("output going into " + outPath);
FileSystem fs = outPath.getFileSystem(context.getConfiguration());
FSDataOutputStream outFile = fs.create(outPath);
FSDataOutputStream splitFile = null;
if (MongoConfigUtil.getBSONOutputBuildSplits(context.getConfiguration())) {
Path splitPath = new Path(outPath.getParent(), "." + outPath.getName() + ".splits");
splitFile = fs.create(splitPath);
}
long splitSize = BSONSplitter.getSplitSize(context.getConfiguration(), null);
return new BSONFileRecordWriter<K, V>(outFile, splitFile, splitSize);
}