@Override
public void tearDown() throws IOException {
if (writer != null) {
try {
writer.close(context);
} catch (InterruptedException e) {
throw new IOException(e);
}
writer = null;
}
if (outputCommitter.needsTaskCommit(context)) outputCommitter.commitTask(context);
HadoopShims.commitOrCleanup(outputCommitter, context);
}
/**
* Creates an lzo file with random data.
*
* @param outputDir Output directory.
* @param fs File system we're using.
* @param attemptContext Task attempt context, contains task id etc.
* @throws IOException
* @throws InterruptedException
*/
private byte[] createTestInput(
Path outputDir, FileSystem fs, TaskAttemptContext attemptContext, int charsToOutput)
throws IOException, InterruptedException {
TextOutputFormat<Text, Text> output = new TextOutputFormat<Text, Text>();
RecordWriter<Text, Text> rw = null;
md5_.reset();
try {
rw = output.getRecordWriter(attemptContext);
char[] chars = "abcdefghijklmnopqrstuvwxyz\u00E5\u00E4\u00F6".toCharArray();
Random r = new Random(System.currentTimeMillis());
Text key = new Text();
Text value = new Text();
int charsMax = chars.length - 1;
for (int i = 0; i < charsToOutput; ) {
i += fillText(chars, r, charsMax, key);
i += fillText(chars, r, charsMax, value);
rw.write(key, value);
md5_.update(key.getBytes(), 0, key.getLength());
// text output format writes tab between the key and value
md5_.update("\t".getBytes("UTF-8"));
md5_.update(value.getBytes(), 0, value.getLength());
}
} finally {
if (rw != null) {
rw.close(attemptContext);
OutputCommitter committer = output.getOutputCommitter(attemptContext);
committer.commitTask(attemptContext);
committer.cleanupJob(attemptContext);
}
}
byte[] result = md5_.digest();
md5_.reset();
return result;
}
public void testBinary() throws IOException, InterruptedException {
Configuration conf = new Configuration();
Job job = new Job(conf);
Path outdir = new Path(System.getProperty("test.build.data", "/tmp"), "outseq");
Random r = new Random();
long seed = r.nextLong();
r.setSeed(seed);
FileOutputFormat.setOutputPath(job, outdir);
SequenceFileAsBinaryOutputFormat.setSequenceFileOutputKeyClass(job, IntWritable.class);
SequenceFileAsBinaryOutputFormat.setSequenceFileOutputValueClass(job, DoubleWritable.class);
SequenceFileAsBinaryOutputFormat.setCompressOutput(job, true);
SequenceFileAsBinaryOutputFormat.setOutputCompressionType(job, CompressionType.BLOCK);
BytesWritable bkey = new BytesWritable();
BytesWritable bval = new BytesWritable();
TaskAttemptContext context =
MapReduceTestUtil.createDummyMapTaskAttemptContext(job.getConfiguration());
OutputFormat<BytesWritable, BytesWritable> outputFormat =
new SequenceFileAsBinaryOutputFormat();
OutputCommitter committer = outputFormat.getOutputCommitter(context);
committer.setupJob(job);
RecordWriter<BytesWritable, BytesWritable> writer = outputFormat.getRecordWriter(context);
IntWritable iwritable = new IntWritable();
DoubleWritable dwritable = new DoubleWritable();
DataOutputBuffer outbuf = new DataOutputBuffer();
LOG.info("Creating data by SequenceFileAsBinaryOutputFormat");
try {
for (int i = 0; i < RECORDS; ++i) {
iwritable = new IntWritable(r.nextInt());
iwritable.write(outbuf);
bkey.set(outbuf.getData(), 0, outbuf.getLength());
outbuf.reset();
dwritable = new DoubleWritable(r.nextDouble());
dwritable.write(outbuf);
bval.set(outbuf.getData(), 0, outbuf.getLength());
outbuf.reset();
writer.write(bkey, bval);
}
} finally {
writer.close(context);
}
committer.commitTask(context);
committer.commitJob(job);
InputFormat<IntWritable, DoubleWritable> iformat =
new SequenceFileInputFormat<IntWritable, DoubleWritable>();
int count = 0;
r.setSeed(seed);
SequenceFileInputFormat.setInputPaths(job, outdir);
LOG.info("Reading data by SequenceFileInputFormat");
for (InputSplit split : iformat.getSplits(job)) {
RecordReader<IntWritable, DoubleWritable> reader = iformat.createRecordReader(split, context);
MapContext<IntWritable, DoubleWritable, BytesWritable, BytesWritable> mcontext =
new MapContextImpl<IntWritable, DoubleWritable, BytesWritable, BytesWritable>(
job.getConfiguration(),
context.getTaskAttemptID(),
reader,
null,
null,
MapReduceTestUtil.createDummyReporter(),
split);
reader.initialize(split, mcontext);
try {
int sourceInt;
double sourceDouble;
while (reader.nextKeyValue()) {
sourceInt = r.nextInt();
sourceDouble = r.nextDouble();
iwritable = reader.getCurrentKey();
dwritable = reader.getCurrentValue();
assertEquals(
"Keys don't match: " + "*" + iwritable.get() + ":" + sourceInt + "*",
sourceInt,
iwritable.get());
assertTrue(
"Vals don't match: " + "*" + dwritable.get() + ":" + sourceDouble + "*",
Double.compare(dwritable.get(), sourceDouble) == 0);
++count;
}
} finally {
reader.close();
}
}
assertEquals("Some records not found", RECORDS, count);
}