/**
* Parse command line, create hive input
*
* @param args Command line arguments
* @return Parsed arguments
*/
private static Optional<BenchmarkArgs> handleCommandLine(String[] args) {
BenchmarkArgs parsedArgs = new BenchmarkArgs();
try {
Args.parse(parsedArgs, args);
} catch (IllegalArgumentException e) {
System.err.println("ERROR: " + e);
Args.usage(parsedArgs);
return Optional.absent();
}
if (parsedArgs.isHelp()) {
Args.usage(parsedArgs);
return Optional.absent();
}
return Optional.of(parsedArgs);
}
private static void run(BenchmarkArgs parsedArgs)
throws TTransportException, IOException, InterruptedException {
HiveInputDescription input = new HiveInputDescription();
input.setDbName(parsedArgs.getDatabase());
input.setTableName(parsedArgs.getTable());
input.setPartitionFilter(parsedArgs.getPartitionFilter());
HiveConf hiveConf = new HiveConf(InputBenchmark.class);
ThriftHiveMetastore.Iface client =
HiveMetastores.create(parsedArgs.getHiveHost(), parsedArgs.getHivePort());
System.err.println("Initialize profile with input data");
HiveApiInputFormat.setProfileInputDesc(hiveConf, input, HiveApiInputFormat.DEFAULT_PROFILE_ID);
HiveApiInputFormat defaultInputFormat = new HiveApiInputFormat();
if (parsedArgs.isTrackMetrics()) {
defaultInputFormat.setObserver(
new MetricsObserver("default", parsedArgs.getRecordPrintPeriod()));
}
List<InputSplit> splits = defaultInputFormat.getSplits(hiveConf, client);
System.err.println("getSplits returned " + splits.size() + " splits");
long numRows = 0;
for (int i = 0; i < splits.size(); ++i) {
InputSplit split = splits.get(i);
TaskAttemptID taskID = new TaskAttemptID();
TaskAttemptContext taskContext = new TaskAttemptContext(hiveConf, taskID);
if (i % parsedArgs.getSplitPrintPeriod() == 0) {
System.err.println("Handling split " + i + " of " + splits.size());
}
RecordReader<WritableComparable, HiveReadableRecord> reader =
defaultInputFormat.createRecordReader(split, taskContext);
reader.initialize(split, taskContext);
numRows += readFully(reader);
}
System.err.println("Parsed " + numRows + " rows");
}