@SuppressWarnings("unchecked")
@Override
public void map(LongWritable positionToMap, LongWritable record, Context context) {
try {
S pos = game.hashToState(positionToMap.get());
game.longToRecord(pos, record.get(), rec);
rec.previousPosition();
int numParents = 0;
numParents = ((Undoable<S>) game).possibleParents(pos, parentStates);
for (int i = 0; i < numParents; i++) {
long parentHash = game.stateToHash(parentStates[i]);
RangeFile parentFile = rangeFiles.getFile(parentHash);
srp.state = parentHash;
srp.record = game.recordToLong(parentStates[i], rec);
context.write(parentFile, new StateRecordPair(srp.state, srp.record));
}
} catch (IOException e) {
throw new Error(e);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
/**
* Map method.
*
* @param offset samples starting from the (offset+1)th sample.
* @param size the number of samples for this map
* @param out output {ture->numInside, false->numOutside}
* @param reporter
*/
public void map(
LongWritable offset,
LongWritable size,
OutputCollector<BooleanWritable, LongWritable> out,
Reporter reporter)
throws IOException {
final HaltonSequence haltonsequence = new HaltonSequence(offset.get());
long numInside = 0L;
long numOutside = 0L;
for (long i = 0; i < size.get(); ) {
// generate points in a unit square
final double[] point = haltonsequence.nextPoint();
// count points inside/outside of the inscribed circle of the square
final double x = point[0] - 0.5;
final double y = point[1] - 0.5;
if (x * x + y * y > 0.25) {
numOutside++;
} else {
numInside++;
}
// report status
i++;
if (i % 1000 == 0) {
reporter.setStatus("Generated " + i + " samples.");
}
}
// output map results
out.collect(new BooleanWritable(true), new LongWritable(numInside));
out.collect(new BooleanWritable(false), new LongWritable(numOutside));
}
/**
* Result file contains hierarchy of workerID-resultvar(incl filename). We deduplicate on the
* workerID. Without JVM reuse each task refers to a unique workerID, so we will not find any
* duplicates. With JVM reuse, however, each slot refers to a workerID, and there are duplicate
* filenames due to partial aggregation and overwrite of fname (the RemoteParWorkerMapper ensures
* uniqueness of those files independent of the runtime implementation).
*
* @param job
* @param fname
* @return
* @throws DMLRuntimeException
*/
@SuppressWarnings("deprecation")
public static LocalVariableMap[] readResultFile(JobConf job, String fname)
throws DMLRuntimeException, IOException {
HashMap<Long, LocalVariableMap> tmp = new HashMap<Long, LocalVariableMap>();
FileSystem fs = FileSystem.get(job);
Path path = new Path(fname);
LongWritable key = new LongWritable(); // workerID
Text value = new Text(); // serialized var header (incl filename)
int countAll = 0;
for (Path lpath : MatrixReader.getSequenceFilePaths(fs, path)) {
SequenceFile.Reader reader = new SequenceFile.Reader(FileSystem.get(job), lpath, job);
try {
while (reader.next(key, value)) {
// System.out.println("key="+key.get()+", value="+value.toString());
if (!tmp.containsKey(key.get())) tmp.put(key.get(), new LocalVariableMap());
Object[] dat = ProgramConverter.parseDataObject(value.toString());
tmp.get(key.get()).put((String) dat[0], (Data) dat[1]);
countAll++;
}
} finally {
if (reader != null) reader.close();
}
}
LOG.debug("Num remote worker results (before deduplication): " + countAll);
LOG.debug("Num remote worker results: " + tmp.size());
// create return array
return tmp.values().toArray(new LocalVariableMap[0]);
}
public void reduce(IntWritable key, Iterable<LongWritable> values, Context context)
throws IOException, InterruptedException {
long max = 0;
for (LongWritable val : values) {
if (val.get() > max) max = val.get();
}
context.write(key, new LongWritable(max));
}
private void compareToUDFUnixTimeStampDate(long t, long y) {
TimestampWritable tsw = toTimestampWritable(t);
LongWritable res = getLongWritable(tsw);
if (res.get() != y) {
System.out.printf(
"%d vs %d for %d, %d\n", res.get(), y, t, tsw.getTimestamp().getTime() / 1000);
}
Assert.assertEquals(res.get(), y);
}
@Override
public LongWritable evaluate(LongWritable a, LongWritable b) {
// LOG.info("Get input " + a.getClass() + ":" + a + " " + b.getClass() + ":"
// + b);
if ((a == null) || (b == null)) {
return null;
}
longWritable.set(a.get() + b.get());
return longWritable;
}
public void reduce(Text key, Iterable<LongWritable> values, Context context)
throws IOException, InterruptedException {
StringBuilder out = new StringBuilder();
long max = Long.MIN_VALUE;
for (LongWritable id : values) {
if (id.get() > max) {
max = id.get();
}
}
context.write(key, new LongWritable(max));
}
/** Compares in the descending order of the keys. */
@Override
public int compare(WritableComparable a, WritableComparable b) {
LongWritable o1 = (LongWritable) a;
LongWritable o2 = (LongWritable) b;
if (o1.get() < o2.get()) {
return 1;
} else if (o1.get() > o2.get()) {
return -1;
} else {
return 0;
}
}
/**
* Accumulate number of points inside/outside results from the mappers.
*
* @param isInside Is the points inside?
* @param values An iterator to a list of point counts
* @param context dummy, not used here.
*/
public void reduce(BooleanWritable isInside, Iterable<LongWritable> values, Context context)
throws IOException, InterruptedException {
if (isInside.get()) {
for (LongWritable val : values) {
numInside += val.get();
}
} else {
for (LongWritable val : values) {
numOutside += val.get();
}
}
}
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);
}
}
/**
* Produce splits such that each is no greater than the quotient of the total size and the
* number of splits requested.
*
* @param job The handle to the JobConf object
* @param numSplits Number of splits requested
*/
public InputSplit[] getSplits(JobConf job, int numSplits) throws IOException {
int cnfiles = job.getInt(SRC_COUNT_LABEL, -1);
long cbsize = job.getLong(TOTAL_SIZE_LABEL, -1);
String srcfilelist = job.get(SRC_LIST_LABEL, "");
if (cnfiles < 0 || cbsize < 0 || "".equals(srcfilelist)) {
throw new RuntimeException(
"Invalid metadata: #files("
+ cnfiles
+ ") total_size("
+ cbsize
+ ") listuri("
+ srcfilelist
+ ")");
}
Path src = new Path(srcfilelist);
FileSystem fs = src.getFileSystem(job);
FileStatus srcst = fs.getFileStatus(src);
ArrayList<FileSplit> splits = new ArrayList<FileSplit>(numSplits);
LongWritable key = new LongWritable();
FilePair value = new FilePair();
final long targetsize = cbsize / numSplits;
long pos = 0L;
long last = 0L;
long acc = 0L;
long cbrem = srcst.getLen();
SequenceFile.Reader sl = null;
try {
sl = new SequenceFile.Reader(fs, src, job);
for (; sl.next(key, value); last = sl.getPosition()) {
// if adding this split would put this split past the target size,
// cut the last split and put this next file in the next split.
if (acc + key.get() > targetsize && acc != 0) {
long splitsize = last - pos;
splits.add(new FileSplit(src, pos, splitsize, (String[]) null));
cbrem -= splitsize;
pos = last;
acc = 0L;
}
acc += key.get();
}
} finally {
checkAndClose(sl);
}
if (cbrem != 0) {
splits.add(new FileSplit(src, pos, cbrem, (String[]) null));
}
return splits.toArray(new FileSplit[splits.size()]);
}
public void reduce(
LongWritable userid,
Iterator<LongWritable> tstamps,
OutputCollector<Text, Session> output,
Reporter reporter)
throws IOException {
// Copy the iterator to an array.
ArrayList<LongWritable> tsarray = new ArrayList<LongWritable>();
while (tstamps.hasNext()) tsarray.add(new LongWritable(tstamps.next().get()));
// Sort the timestamps.
Collections.sort(tsarray);
// Scan the array looking for session boundaries.
long t0 = 0;
long session_start = 0;
long session_end = 0;
int session_hits = 0;
for (LongWritable tstamp : tsarray) {
long tt = tstamp.get();
// How long since the prior hit?
long delta = tt - t0;
// Is this a new session?
if (delta > SESSION_GAP_MSEC) {
// Is there a prior session?
if (session_start != 0)
collect_session(userid.get(), session_start, session_end, session_hits, output);
// Reset for the new session.
session_start = tt;
session_hits = 0;
}
// Extend the current session.
session_hits += 1;
session_end = tt;
// On to the next hit ...
t0 = tt;
}
// Write out the last session.
if (session_start != 0)
collect_session(userid.get(), session_start, session_end, session_hits, output);
}
private void getOffsets(LogFilePath logFilePath, Set<Long> offsets) throws Exception {
String path = logFilePath.getLogFilePath();
Path fsPath = new Path(path);
FileSystem fileSystem = FileUtil.getFileSystem(path);
SequenceFile.Reader reader = new SequenceFile.Reader(fileSystem, fsPath, new Configuration());
LongWritable key = (LongWritable) reader.getKeyClass().newInstance();
BytesWritable value = (BytesWritable) reader.getValueClass().newInstance();
while (reader.next(key, value)) {
if (!offsets.add(key.get())) {
throw new RuntimeException(
"duplicate key " + key.get() + " found in file " + logFilePath.getLogFilePath());
}
}
reader.close();
}
@Override
public void map(
LongWritable rawKey,
Text rawValue,
OutputCollector<TaggedFirstSecondIndexes, CSVReblockMR.BlockRow> out,
Reporter reporter)
throws IOException {
if (_first) {
rowOffset = offsetMap.get(rawKey.get());
_reporter = reporter;
_first = false;
}
// output the header line
if (rawKey.get() == 0 && _partFileWithHeader) {
tfmapper.processHeaderLine();
if (tfmapper.hasHeader()) return;
}
// parse the input line and apply transformation
String[] words = tfmapper.getWords(rawValue);
if (!tfmapper.omit(words)) {
words = tfmapper.apply(words);
try {
tfmapper.check(words);
// Perform CSV Reblock
CSVReblockInstruction ins = csv_reblock_instructions.get(0).get(0);
idxRow =
CSVReblockMapper.processRow(
idxRow,
words,
rowOffset,
num,
ins.output,
ins.brlen,
ins.bclen,
ins.fill,
ins.fillValue,
out);
} catch (DMLRuntimeException e) {
throw new RuntimeException(e.getMessage() + ":" + rawValue.toString());
}
num++;
}
}
public static double getAsConstDouble(@Nonnull final ObjectInspector numberOI)
throws UDFArgumentException {
final String typeName = numberOI.getTypeName();
if (DOUBLE_TYPE_NAME.equals(typeName)) {
DoubleWritable v = getConstValue(numberOI);
return v.get();
} else if (FLOAT_TYPE_NAME.equals(typeName)) {
FloatWritable v = getConstValue(numberOI);
return v.get();
} else if (INT_TYPE_NAME.equals(typeName)) {
IntWritable v = getConstValue(numberOI);
return v.get();
} else if (BIGINT_TYPE_NAME.equals(typeName)) {
LongWritable v = getConstValue(numberOI);
return v.get();
} else if (SMALLINT_TYPE_NAME.equals(typeName)) {
ShortWritable v = getConstValue(numberOI);
return v.get();
} else if (TINYINT_TYPE_NAME.equals(typeName)) {
ByteWritable v = getConstValue(numberOI);
return v.get();
}
throw new UDFArgumentException(
"Unexpected argument type to cast as double: "
+ TypeInfoUtils.getTypeInfoFromObjectInspector(numberOI));
}
@Override
protected void reduce(Text key, Iterable<LongWritable> values, Context context)
throws IOException, InterruptedException {
/*
* Added by Xudong Zhang
*/
if (parseWeibo) {
String word = key.toString();
if (!ChineseUtils.allChinese(word)) {
context.getCounter("MyCounter", "NumWordsFilteredByWeiboParser").increment(1);
return;
}
}
long sum = 0;
for (LongWritable value : values) {
sum += value.get();
}
if (sum >= minSupport) {
context.write(key, new LongWritable(sum));
} else {
context.getCounter("MyCounter", "NumWordsLessThanMinSupport").increment(1);
}
}
public void process(Writable value) throws HiveException {
// A mapper can span multiple files/partitions.
// The serializers need to be reset if the input file changed
ExecMapperContext context = getExecContext();
if (context != null && context.inputFileChanged()) {
// The child operators cleanup if input file has changed
cleanUpInputFileChanged();
}
int childrenDone = 0;
for (MapOpCtx current : currentCtxs) {
Object row = null;
try {
row = current.readRow(value, context);
if (!current.forward(row)) {
childrenDone++;
}
} catch (Exception e) {
// TODO: policy on deserialization errors
String message = toErrorMessage(value, row, current.rowObjectInspector);
if (row == null) {
deserialize_error_count.set(deserialize_error_count.get() + 1);
throw new HiveException("Hive Runtime Error while processing writable " + message, e);
}
throw new HiveException("Hive Runtime Error while processing row " + message, e);
}
}
rowsForwarded(childrenDone, 1);
}
public void reduce(LongWritable key, Iterable<Text> values, Context context)
throws IOException, InterruptedException {
double sumOfRankValues = 0.0;
String targetUrlsList = "";
int sourceUrl = (int) key.get();
int numUrls = context.getConfiguration().getInt("numUrls", 1);
// hints each tuple may include: rank value tuple or link relation tuple
for (Text value : values) {
String[] strArray = value.toString().split("#");
// sourceUrl = Integer.parseInt(strArray[0].split("\t")[1]);
if (strArray.length == 1) {
sumOfRankValues += Double.parseDouble(strArray[0]);
} else {
StringBuffer sb = new StringBuffer();
for (int i = 1; i < strArray.length; i++) {
sb.append("#" + strArray[i]);
}
targetUrlsList = sb.toString();
}
}
sumOfRankValues = 0.85 * sumOfRankValues + 0.15 * (1.0) / (double) numUrls;
context.write(key, new Text(sumOfRankValues + targetUrlsList));
}
/**
* write.
*
* @param out output stream
* @throws IOException -
*/
@Override
public void write(final DataOutput out) throws IOException {
out.writeInt(this.relatedUsers.size());
for (LongWritable item : this.relatedUsers) {
out.writeLong(item.get());
}
}
@Override
public void reduce(Text key, Iterable<LongWritable> values, Context context)
throws IOException, InterruptedException {
long s = 0;
for (LongWritable lw : values) s += lw.get();
context.write(key, new LongWritable(s));
}
@Override
public void map(LongWritable row, NullWritable ignored, Context context)
throws IOException, InterruptedException {
context.setStatus("Entering");
long rowId = row.get();
if (rand == null) {
// we use 3 random numbers per a row
rand = new RandomGenerator(rowId * 3);
}
addKey();
value.clear();
// addRowId(rowId);
addFiller(rowId);
// New
Mutation m = new Mutation(key);
m.put(
new Text("c"), // column family
getRowIdString(rowId), // column qual
new Value(value.toString().getBytes())); // data
context.setStatus("About to add to accumulo");
context.write(tableName, m);
context.setStatus("Added to accumulo " + key.toString());
}
public void reduce(
LongWritable key,
Iterator<Text> values,
OutputCollector<Text, Text> output,
Reporter reporter)
throws IOException {
// String line = "";
StringBuilder lineBuilder = new StringBuilder();
String tempStr = null;
boolean first = true;
if (key.get() != -1) {
while (values.hasNext()) {
if (first) {
tempStr = values.next().toString().trim();
lineBuilder.append(tempStr);
first = false;
if (tempStr.equals("-1")) {
lineBuilder = new StringBuilder("");
}
} else {
lineBuilder.append(" ");
lineBuilder.append(values.next().toString());
}
}
output.collect(null, new Text(lineBuilder.toString()));
}
}
/**
* Convert from long to an integer. This is called for CAST(... AS INT)
*
* @param i The long value to convert
* @return IntWritable
*/
public IntWritable evaluate(LongWritable i) {
if (i == null) {
return null;
} else {
intWritable.set((int) i.get());
return intWritable;
}
}
@Override
protected void map(LongWritable key, Text value, Context context)
throws IOException, InterruptedException {
id += increment;
context.write(
new LongWritable(id), new Text(String.format("%d, %s", key.get(), value.toString())));
}
public static long getConstLong(@Nonnull final ObjectInspector oi) throws UDFArgumentException {
if (!isBigIntOI(oi)) {
throw new UDFArgumentException(
"argument must be a BigInt value: " + TypeInfoUtils.getTypeInfoFromObjectInspector(oi));
}
LongWritable v = getConstValue(oi);
return v.get();
}
@Override
protected void reduce(Text token, Iterable<LongWritable> counts, Context context)
throws IOException, InterruptedException {
long n = 0;
for (LongWritable count : counts) n += count.get();
total.set(n);
context.write(token, total);
}
/** Reads the next key-value pair. */
public boolean next(LongWritable key, TrecDocument value) throws IOException {
if (reader.next(pos, text) == false) {
return false;
}
key.set(pos.get());
TrecDocument.readDocument(value, text.toString());
return true;
}
/**
* Propagates the smallest vertex id to all neighbors. Will always choose to halt and only
* reactivate if a smaller id has been sent to it.
*
* @param messages Iterator of messages from the previous superstep.
* @throws java.io.IOException
*/
@Override
public void compute(
Vertex<LongWritable, LongWritable, NullWritable> vertex, Iterable<LongWritable> messages)
throws IOException {
long currentComponent = vertex.getValue().get();
// First superstep is special, because we can simply look at the neighbors
if (getSuperstep() == 0) {
for (Edge<LongWritable, NullWritable> edge : vertex.getEdges()) {
long neighbor = edge.getTargetVertexId().get();
if (neighbor < currentComponent) {
currentComponent = neighbor;
}
}
// Only need to send value if it is not the own id
if (currentComponent != vertex.getValue().get()) {
vertex.setValue(new LongWritable(currentComponent));
for (Edge<LongWritable, NullWritable> edge : vertex.getEdges()) {
LongWritable neighbor = edge.getTargetVertexId();
if (neighbor.get() > currentComponent) {
sendMessage(neighbor, vertex.getValue());
}
}
}
vertex.voteToHalt();
return;
}
boolean changed = false;
// did we get a smaller id ?
for (LongWritable message : messages) {
long candidateComponent = message.get();
if (candidateComponent < currentComponent) {
currentComponent = candidateComponent;
changed = true;
}
}
// propagate new component id to the neighbors
if (changed) {
vertex.setValue(new LongWritable(currentComponent));
sendMessageToAllEdges(vertex, vertex.getValue());
}
vertex.voteToHalt();
}
@Override
protected void map(LongWritable key, BytesWritable value, Context context)
throws IOException, InterruptedException {
if (Long.compareUnsigned(key.get(), start) >= 0
&& Long.compareUnsigned(key.get(), end) <= 0) {
// It is assumed that the passed BytesWritable value is always a *single* PacketInfo object.
// Passing more than 1
// object will result in the whole set being passed through if any pass the filter. We
// cannot serialize PacketInfo
// objects back to byte arrays, otherwise we could support more than one packet.
// Note: short-circuit findAny() func on stream
boolean send = filteredPacketInfo(value).findAny().isPresent();
if (send) {
context.write(key, value);
}
}
}
@Override
protected void reduce(Text category, Iterable<LongWritable> ones, Context context)
throws IOException, InterruptedException {
Long sum = 0L;
for (LongWritable one : ones) {
sum += one.get();
}
context.write(category, new LongWritable(sum));
}
