public void testStep0Mapper() throws Exception {
Random rng = RandomUtils.getRandom();
// create a dataset large enough to be split up
String descriptor = Utils.randomDescriptor(rng, numAttributes);
double[][] source = Utils.randomDoubles(rng, descriptor, numInstances);
String[] sData = Utils.double2String(source);
// write the data to a file
Path dataPath = Utils.writeDataToTestFile(sData);
JobConf job = new JobConf();
job.setNumMapTasks(numMaps);
FileInputFormat.setInputPaths(job, dataPath);
// retrieve the splits
TextInputFormat input = (TextInputFormat) job.getInputFormat();
InputSplit[] splits = input.getSplits(job, numMaps);
InputSplit[] sorted = Arrays.copyOf(splits, splits.length);
Builder.sortSplits(sorted);
Step0OutputCollector collector = new Step0OutputCollector(numMaps);
Reporter reporter = Reporter.NULL;
for (int p = 0; p < numMaps; p++) {
InputSplit split = sorted[p];
RecordReader<LongWritable, Text> reader = input.getRecordReader(split, job, reporter);
LongWritable key = reader.createKey();
Text value = reader.createValue();
Step0Mapper mapper = new Step0Mapper();
mapper.configure(p);
Long firstKey = null;
int size = 0;
while (reader.next(key, value)) {
if (firstKey == null) {
firstKey = key.get();
}
mapper.map(key, value, collector, reporter);
size++;
}
mapper.close();
// validate the mapper's output
assertEquals(p, collector.keys[p]);
assertEquals(firstKey.longValue(), collector.values[p].getFirstId());
assertEquals(size, collector.values[p].getSize());
}
}
public void testInputFormat() {
try {
JobConf conf = new JobConf();
String TMP_DIR = System.getProperty("test.build.data", "/tmp");
Path filename = new Path("file:///" + TMP_DIR + "/tmpSeqFile");
SequenceFile.Writer sfw =
SequenceFile.createWriter(
FileSystem.getLocal(conf),
conf,
filename,
ChukwaArchiveKey.class,
ChunkImpl.class,
SequenceFile.CompressionType.NONE,
Reporter.NULL);
StringBuilder buf = new StringBuilder();
int offsets[] = new int[lines.length];
for (int i = 0; i < lines.length; ++i) {
buf.append(lines[i]);
buf.append("\n");
offsets[i] = buf.length() - 1;
}
ChukwaArchiveKey key = new ChukwaArchiveKey(0, "datatype", "sname", 0);
ChunkImpl val = new ChunkImpl("datatype", "sname", 0, buf.toString().getBytes(), null);
val.setRecordOffsets(offsets);
sfw.append(key, val);
sfw.append(key, val); // write it twice
sfw.close();
long len = FileSystem.getLocal(conf).getFileStatus(filename).getLen();
InputSplit split = new FileSplit(filename, 0, len, (String[]) null);
ChukwaInputFormat in = new ChukwaInputFormat();
RecordReader<LongWritable, Text> r = in.getRecordReader(split, conf, Reporter.NULL);
LongWritable l = r.createKey();
Text line = r.createValue();
for (int i = 0; i < lines.length * 2; ++i) {
boolean succeeded = r.next(l, line);
assertTrue(succeeded);
assertEquals(i, l.get());
assertEquals(lines[i % lines.length], line.toString());
System.out.println("read line: " + l.get() + " " + line);
}
boolean succeeded = r.next(l, line);
assertFalse(succeeded);
} catch (IOException e) {
e.printStackTrace();
fail("IO exception " + e);
}
}
public void run(
RecordReader<Text, ArcFileItem> input,
OutputCollector<Text, CrawlURLMetadata> output,
Reporter reporter)
throws IOException {
int lastValidPos = 0;
try {
// allocate key & value instances that are re-used for all entries
Text key = input.createKey();
ArcFileItem value = input.createValue();
while (input.next(key, value)) {
lastValidPos = value.getArcFilePos();
// map pair to output
map(key, value, output, reporter);
}
} catch (IOException e) {
String errorMessage =
"Exception processing Split:"
+ _splitDetails
+ " Exception:"
+ StringUtils.stringifyException(e);
LOG.error(errorMessage);
if (_attemptID.getId() == 0
|| (lastValidPos == 0 && _attemptID.getId() != _maxAttemptsPerTask - 1)) {
throw new IOException(errorMessage);
}
// and just ignore the message
} catch (Throwable e) {
String errorMessage =
"Unknown Exception processing Split:"
+ _splitDetails
+ " Exception:"
+ StringUtils.stringifyException(e);
LOG.error(errorMessage);
// if attempt number is not max attempt number configured...
if (_attemptID.getId() != _maxAttemptsPerTask - 1) {
// then bubble up exception
throw new IOException(errorMessage);
}
} finally {
close();
}
}
private void writeThenReadByRecordReader(
int intervalRecordCount,
int writeCount,
int splitNumber,
long minSplitSize,
CompressionCodec codec)
throws IOException {
Path testDir =
new Path(System.getProperty("test.data.dir", ".") + "/mapred/testsmallfirstsplit");
Path testFile = new Path(testDir, "test_rcfile");
fs.delete(testFile, true);
Configuration cloneConf = new Configuration(conf);
RCFileOutputFormat.setColumnNumber(cloneConf, bytesArray.length);
cloneConf.setInt(RCFile.RECORD_INTERVAL_CONF_STR, intervalRecordCount);
RCFile.Writer writer = new RCFile.Writer(fs, cloneConf, testFile, null, codec);
BytesRefArrayWritable bytes = new BytesRefArrayWritable(bytesArray.length);
for (int i = 0; i < bytesArray.length; i++) {
BytesRefWritable cu = null;
cu = new BytesRefWritable(bytesArray[i], 0, bytesArray[i].length);
bytes.set(i, cu);
}
for (int i = 0; i < writeCount; i++) {
if (i == intervalRecordCount) {
System.out.println("write position:" + writer.getLength());
}
writer.append(bytes);
}
writer.close();
RCFileInputFormat inputFormat = new RCFileInputFormat();
JobConf jonconf = new JobConf(cloneConf);
jonconf.set("mapred.input.dir", testDir.toString());
jonconf.setLong("mapred.min.split.size", minSplitSize);
InputSplit[] splits = inputFormat.getSplits(jonconf, splitNumber);
assertEquals("splits length should be " + splitNumber, splits.length, splitNumber);
int readCount = 0;
for (int i = 0; i < splits.length; i++) {
int previousReadCount = readCount;
RecordReader rr = inputFormat.getRecordReader(splits[i], jonconf, Reporter.NULL);
Object key = rr.createKey();
Object value = rr.createValue();
while (rr.next(key, value)) {
readCount++;
}
System.out.println("The " + i + "th split read " + (readCount - previousReadCount));
}
assertEquals("readCount should be equal to writeCount", readCount, writeCount);
}
/** For a given RecordReader rr, occupy position id in collector. */
WrappedRecordReader(int id, RecordReader<K, U> rr, Class<? extends WritableComparator> cmpcl)
throws IOException {
this.id = id;
this.rr = rr;
khead = rr.createKey();
vhead = rr.createValue();
try {
cmp = (null == cmpcl) ? WritableComparator.get(khead.getClass()) : cmpcl.newInstance();
} catch (InstantiationException e) {
throw (IOException) new IOException().initCause(e);
} catch (IllegalAccessException e) {
throw (IOException) new IOException().initCause(e);
}
vjoin = new StreamBackedIterator<U>();
next();
}
@Override
public void setFile(String file, long offset, long length) {
JobConf defaultConf = new JobConf();
this.split = new FileSplit(new Path(file), offset, length, defaultConf);
this.jobConf = defaultConf;
// this.split = split;
this.input_format = new TextInputFormat();
try {
this.reader = input_format.getRecordReader(this.split, this.jobConf, voidReporter);
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
this.key = reader.createKey();
}
@Override
public void initialize(InputSplit split, final TaskAttemptContext context)
throws IOException, InterruptedException {
org.apache.hadoop.mapred.InputSplit oldSplit;
if (split.getClass() == FileSplit.class) {
oldSplit =
new org.apache.hadoop.mapred.FileSplit(
((FileSplit) split).getPath(),
((FileSplit) split).getStart(),
((FileSplit) split).getLength(),
split.getLocations());
} else {
oldSplit = ((InputSplitWrapper) split).realSplit;
}
@SuppressWarnings("unchecked")
Reporter reporter = new Reporter() { // Reporter interface over ctx
final TaskInputOutputContext ioCtx =
context instanceof TaskInputOutputContext ? (TaskInputOutputContext) context : null;
public void progress() {
HadoopCompat.progress(context);
}
// @Override
public float getProgress() {
return (ioCtx != null) ? ioCtx.getProgress() : 0;
}
public void setStatus(String status) {
if (ioCtx != null) HadoopCompat.setStatus(ioCtx, status);
}
public void incrCounter(String group, String counter, long amount) {
if (ioCtx != null)
HadoopCompat.incrementCounter(ioCtx.getCounter(group, counter), amount);
}
@SuppressWarnings("unchecked")
public void incrCounter(Enum<?> key, long amount) {
if (ioCtx != null) HadoopCompat.incrementCounter(ioCtx.getCounter(key), amount);
}
public org.apache.hadoop.mapred.InputSplit getInputSplit()
throws UnsupportedOperationException {
throw new UnsupportedOperationException();
}
public Counter getCounter(String group, String name) {
return ioCtx != null ? (Counter) HadoopCompat.getCounter(ioCtx, group, name) : null;
}
@SuppressWarnings("unchecked")
public Counter getCounter(Enum<?> name) {
return ioCtx != null ? (Counter) ioCtx.getCounter(name) : null;
}
};
realReader =
realInputFormat.getRecordReader(
oldSplit, (JobConf) HadoopCompat.getConfiguration(context), reporter);
keyObj = realReader.createKey();
valueObj = realReader.createValue();
}
public GenericHiveRecordCursor(
RecordReader<K, V> recordReader,
long totalBytes,
Properties splitSchema,
List<HivePartitionKey> partitionKeys,
List<HiveColumnHandle> columns,
DateTimeZone hiveStorageTimeZone,
DateTimeZone sessionTimeZone) {
checkNotNull(recordReader, "recordReader is null");
checkArgument(totalBytes >= 0, "totalBytes is negative");
checkNotNull(splitSchema, "splitSchema is null");
checkNotNull(partitionKeys, "partitionKeys is null");
checkNotNull(columns, "columns is null");
checkArgument(!columns.isEmpty(), "columns is empty");
checkNotNull(hiveStorageTimeZone, "hiveStorageTimeZone is null");
checkNotNull(sessionTimeZone, "sessionTimeZone is null");
this.recordReader = recordReader;
this.totalBytes = totalBytes;
this.key = recordReader.createKey();
this.value = recordReader.createValue();
this.hiveStorageTimeZone = hiveStorageTimeZone;
this.sessionTimeZone = sessionTimeZone;
this.deserializer = getDeserializer(splitSchema);
this.rowInspector = getTableObjectInspector(deserializer);
int size = columns.size();
String[] names = new String[size];
this.types = new Type[size];
this.hiveTypes = new HiveType[size];
this.structFields = new StructField[size];
this.fieldInspectors = new ObjectInspector[size];
this.isPartitionColumn = new boolean[size];
this.loaded = new boolean[size];
this.booleans = new boolean[size];
this.longs = new long[size];
this.doubles = new double[size];
this.slices = new Slice[size];
this.nulls = new boolean[size];
// initialize data columns
for (int i = 0; i < columns.size(); i++) {
HiveColumnHandle column = columns.get(i);
names[i] = column.getName();
types[i] = column.getType();
hiveTypes[i] = column.getHiveType();
if (!column.isPartitionKey()) {
StructField field = rowInspector.getStructFieldRef(column.getName());
structFields[i] = field;
fieldInspectors[i] = field.getFieldObjectInspector();
}
isPartitionColumn[i] = column.isPartitionKey();
}
// parse requested partition columns
Map<String, HivePartitionKey> partitionKeysByName =
uniqueIndex(partitionKeys, HivePartitionKey.nameGetter());
for (int columnIndex = 0; columnIndex < columns.size(); columnIndex++) {
HiveColumnHandle column = columns.get(columnIndex);
if (column.isPartitionKey()) {
HivePartitionKey partitionKey = partitionKeysByName.get(column.getName());
checkArgument(partitionKey != null, "Unknown partition key %s", column.getName());
byte[] bytes = partitionKey.getValue().getBytes(Charsets.UTF_8);
Type type = types[columnIndex];
if (BOOLEAN.equals(type)) {
if (isTrue(bytes, 0, bytes.length)) {
booleans[columnIndex] = true;
} else if (isFalse(bytes, 0, bytes.length)) {
booleans[columnIndex] = false;
} else {
String valueString = new String(bytes, Charsets.UTF_8);
throw new IllegalArgumentException(
String.format(
"Invalid partition value '%s' for BOOLEAN partition key %s",
valueString, names[columnIndex]));
}
} else if (BIGINT.equals(type)) {
if (bytes.length == 0) {
throw new IllegalArgumentException(
String.format(
"Invalid partition value '' for BIGINT partition key %s", names[columnIndex]));
}
longs[columnIndex] = parseLong(bytes, 0, bytes.length);
} else if (DOUBLE.equals(type)) {
if (bytes.length == 0) {
throw new IllegalArgumentException(
String.format(
"Invalid partition value '' for DOUBLE partition key %s", names[columnIndex]));
}
doubles[columnIndex] = parseDouble(bytes, 0, bytes.length);
} else if (VARCHAR.equals(type)) {
slices[columnIndex] = Slices.wrappedBuffer(Arrays.copyOf(bytes, bytes.length));
} else {
throw new UnsupportedOperationException("Unsupported column type: " + type);
}
}
}
}
public K createKey() {
return (K) recordReader.createKey();
}
public GenericHiveRecordCursor(
RecordReader<K, V> recordReader,
long totalBytes,
Properties splitSchema,
List<HivePartitionKey> partitionKeys,
List<HiveColumnHandle> columns,
DateTimeZone hiveStorageTimeZone,
TypeManager typeManager) {
requireNonNull(recordReader, "recordReader is null");
checkArgument(totalBytes >= 0, "totalBytes is negative");
requireNonNull(splitSchema, "splitSchema is null");
requireNonNull(partitionKeys, "partitionKeys is null");
requireNonNull(columns, "columns is null");
requireNonNull(hiveStorageTimeZone, "hiveStorageTimeZone is null");
this.recordReader = recordReader;
this.totalBytes = totalBytes;
this.key = recordReader.createKey();
this.value = recordReader.createValue();
this.hiveStorageTimeZone = hiveStorageTimeZone;
this.deserializer = getDeserializer(splitSchema);
this.rowInspector = getTableObjectInspector(deserializer);
int size = columns.size();
String[] names = new String[size];
this.types = new Type[size];
this.hiveTypes = new HiveType[size];
this.structFields = new StructField[size];
this.fieldInspectors = new ObjectInspector[size];
this.isPartitionColumn = new boolean[size];
this.loaded = new boolean[size];
this.booleans = new boolean[size];
this.longs = new long[size];
this.doubles = new double[size];
this.slices = new Slice[size];
this.objects = new Object[size];
this.nulls = new boolean[size];
// initialize data columns
for (int i = 0; i < columns.size(); i++) {
HiveColumnHandle column = columns.get(i);
names[i] = column.getName();
types[i] = typeManager.getType(column.getTypeSignature());
hiveTypes[i] = column.getHiveType();
if (!column.isPartitionKey()) {
StructField field = rowInspector.getStructFieldRef(column.getName());
structFields[i] = field;
fieldInspectors[i] = field.getFieldObjectInspector();
}
isPartitionColumn[i] = column.isPartitionKey();
}
// parse requested partition columns
Map<String, HivePartitionKey> partitionKeysByName =
uniqueIndex(partitionKeys, HivePartitionKey::getName);
for (int columnIndex = 0; columnIndex < columns.size(); columnIndex++) {
HiveColumnHandle column = columns.get(columnIndex);
if (column.isPartitionKey()) {
HivePartitionKey partitionKey = partitionKeysByName.get(column.getName());
checkArgument(partitionKey != null, "Unknown partition key %s", column.getName());
byte[] bytes = partitionKey.getValue().getBytes(UTF_8);
String name = names[columnIndex];
Type type = types[columnIndex];
if (HiveUtil.isHiveNull(bytes)) {
nulls[columnIndex] = true;
} else if (BOOLEAN.equals(type)) {
booleans[columnIndex] = booleanPartitionKey(partitionKey.getValue(), name);
} else if (BIGINT.equals(type)) {
longs[columnIndex] = bigintPartitionKey(partitionKey.getValue(), name);
} else if (INTEGER.equals(type)) {
longs[columnIndex] = integerPartitionKey(partitionKey.getValue(), name);
} else if (SMALLINT.equals(type)) {
longs[columnIndex] = smallintPartitionKey(partitionKey.getValue(), name);
} else if (TINYINT.equals(type)) {
longs[columnIndex] = tinyintPartitionKey(partitionKey.getValue(), name);
} else if (DOUBLE.equals(type)) {
doubles[columnIndex] = doublePartitionKey(partitionKey.getValue(), name);
} else if (isVarcharType(type)) {
slices[columnIndex] = varcharPartitionKey(partitionKey.getValue(), name, type);
} else if (DATE.equals(type)) {
longs[columnIndex] = datePartitionKey(partitionKey.getValue(), name);
} else if (TIMESTAMP.equals(type)) {
longs[columnIndex] =
timestampPartitionKey(partitionKey.getValue(), hiveStorageTimeZone, name);
} else if (isShortDecimal(type)) {
longs[columnIndex] =
shortDecimalPartitionKey(partitionKey.getValue(), (DecimalType) type, name);
} else if (isLongDecimal(type)) {
slices[columnIndex] =
longDecimalPartitionKey(partitionKey.getValue(), (DecimalType) type, name);
} else {
throw new PrestoException(
NOT_SUPPORTED,
format(
"Unsupported column type %s for partition key: %s", type.getDisplayName(), name));
}
}
}
}
/** Request new key from proxied RR. */
public K createKey() {
return rr.createKey();
}
public void testProcessOutput() throws Exception {
Random rng = RandomUtils.getRandom();
// create a dataset large enough to be split up
String descriptor = Utils.randomDescriptor(rng, numAttributes);
double[][] source = Utils.randomDoubles(rng, descriptor, numInstances);
// each instance label is its index in the dataset
int labelId = Utils.findLabel(descriptor);
for (int index = 0; index < numInstances; index++) {
source[index][labelId] = index;
}
String[] sData = Utils.double2String(source);
// write the data to a file
Path dataPath = Utils.writeDataToTestFile(sData);
// prepare a data converter
Dataset dataset = DataLoader.generateDataset(descriptor, sData);
DataConverter converter = new DataConverter(dataset);
JobConf job = new JobConf();
job.setNumMapTasks(numMaps);
FileInputFormat.setInputPaths(job, dataPath);
// retrieve the splits
TextInputFormat input = (TextInputFormat) job.getInputFormat();
InputSplit[] splits = input.getSplits(job, numMaps);
InputSplit[] sorted = Arrays.copyOf(splits, splits.length);
Builder.sortSplits(sorted);
Reporter reporter = Reporter.NULL;
int[] keys = new int[numMaps];
Step0Output[] values = new Step0Output[numMaps];
int[] expectedIds = new int[numMaps];
for (int p = 0; p < numMaps; p++) {
InputSplit split = sorted[p];
RecordReader<LongWritable, Text> reader = input.getRecordReader(split, job, reporter);
LongWritable key = reader.createKey();
Text value = reader.createValue();
Long firstKey = null;
int size = 0;
while (reader.next(key, value)) {
if (firstKey == null) {
firstKey = key.get();
expectedIds[p] = converter.convert(0, value.toString()).label;
}
size++;
}
keys[p] = p;
values[p] = new Step0Output(firstKey, size);
}
Step0Output[] partitions = Step0Job.processOutput(keys, values);
int[] actualIds = Step0Output.extractFirstIds(partitions);
assertTrue(
"Expected: " + Arrays.toString(expectedIds) + " But was: " + Arrays.toString(actualIds),
Arrays.equals(expectedIds, actualIds));
}