/// <summary>From each split sampled, take the first numSamples / numSplits records.</summary>
/// <exception cref="System.IO.IOException"/>
public virtual K[] GetSample(InputFormat <K, V> inf, JobConf job)
{
// ArrayList::toArray doesn't preserve type
InputSplit[] splits = inf.GetSplits(job, job.GetNumMapTasks());
AList <K> samples = new AList <K>(numSamples);
int splitsToSample = Math.Min(maxSplitsSampled, splits.Length);
int splitStep = splits.Length / splitsToSample;
int samplesPerSplit = numSamples / splitsToSample;
long records = 0;
for (int i = 0; i < splitsToSample; ++i)
{
RecordReader <K, V> reader = inf.GetRecordReader(splits[i * splitStep], job, Reporter
.Null);
K key = reader.CreateKey();
V value = reader.CreateValue();
while (reader.Next(key, value))
{
samples.AddItem(key);
key = reader.CreateKey();
++records;
if ((i + 1) * samplesPerSplit <= records)
{
break;
}
}
reader.Close();
}
return((K[])Sharpen.Collections.ToArray(samples));
}
/// <summary>
/// Randomize the split order, then take the specified number of keys from
/// each split sampled, where each key is selected with the specified
/// probability and possibly replaced by a subsequently selected key when
/// the quota of keys from that split is satisfied.
/// </summary>
/// <exception cref="System.IO.IOException"/>
public virtual K[] GetSample(InputFormat <K, V> inf, JobConf job)
{
// ArrayList::toArray doesn't preserve type
InputSplit[] splits = inf.GetSplits(job, job.GetNumMapTasks());
AList <K> samples = new AList <K>(numSamples);
int splitsToSample = Math.Min(maxSplitsSampled, splits.Length);
Random r = new Random();
long seed = r.NextLong();
r.SetSeed(seed);
Log.Debug("seed: " + seed);
// shuffle splits
for (int i = 0; i < splits.Length; ++i)
{
InputSplit tmp = splits[i];
int j = r.Next(splits.Length);
splits[i] = splits[j];
splits[j] = tmp;
}
// our target rate is in terms of the maximum number of sample splits,
// but we accept the possibility of sampling additional splits to hit
// the target sample keyset
for (int i_1 = 0; i_1 < splitsToSample || (i_1 < splits.Length && samples.Count <
numSamples); ++i_1)
{
RecordReader <K, V> reader = inf.GetRecordReader(splits[i_1], job, Reporter.Null);
K key = reader.CreateKey();
V value = reader.CreateValue();
while (reader.Next(key, value))
{
if (r.NextDouble() <= freq)
{
if (samples.Count < numSamples)
{
samples.AddItem(key);
}
else
{
// When exceeding the maximum number of samples, replace a
// random element with this one, then adjust the frequency
// to reflect the possibility of existing elements being
// pushed out
int ind = r.Next(numSamples);
if (ind != numSamples)
{
samples.Set(ind, key);
}
freq *= (numSamples - 1) / (double)numSamples;
}
key = reader.CreateKey();
}
}
reader.Close();
}
return((K[])Sharpen.Collections.ToArray(samples));
}
/// <exception cref="System.IO.IOException"/>
internal WrappedRecordReader(int id, RecordReader <K, U> rr, Type cmpcl, Configuration
conf)
{
// index at which values will be inserted in collector
// key at the top of this RR
// value assoc with khead
this.id = id;
this.rr = rr;
this.conf = (conf == null) ? new Configuration() : conf;
khead = rr.CreateKey();
vhead = rr.CreateValue();
try
{
cmp = (null == cmpcl) ? WritableComparator.Get(khead.GetType(), this.conf) : System.Activator.CreateInstance
(cmpcl);
}
catch (InstantiationException e)
{
throw (IOException)Sharpen.Extensions.InitCause(new IOException(), e);
}
catch (MemberAccessException e)
{
throw (IOException)Sharpen.Extensions.InitCause(new IOException(), e);
}
vjoin = new StreamBackedIterator <U>();
Next();
}
/// <exception cref="System.IO.IOException"/>
private static IList <Text> ReadSplit(KeyValueTextInputFormat format, InputSplit split
, JobConf job)
{
IList <Text> result = new AList <Text>();
RecordReader <Text, Text> reader = null;
try
{
reader = format.GetRecordReader(split, job, voidReporter);
Text key = reader.CreateKey();
Text value = reader.CreateValue();
while (reader.Next(key, value))
{
result.AddItem(value);
value = (Text)reader.CreateValue();
}
}
finally
{
if (reader != null)
{
reader.Close();
}
}
return(result);
}
/// <summary>test DBInputFormat class.</summary>
/// <remarks>test DBInputFormat class. Class should split result for chunks</remarks>
/// <exception cref="System.Exception"/>
public virtual void TestDBInputFormat()
{
JobConf configuration = new JobConf();
SetupDriver(configuration);
DBInputFormat <DBInputFormat.NullDBWritable> format = new DBInputFormat <DBInputFormat.NullDBWritable
>();
format.SetConf(configuration);
format.SetConf(configuration);
DBInputFormat.DBInputSplit splitter = new DBInputFormat.DBInputSplit(1, 10);
Reporter reporter = Org.Mockito.Mockito.Mock <Reporter>();
RecordReader <LongWritable, DBInputFormat.NullDBWritable> reader = format.GetRecordReader
(splitter, configuration, reporter);
configuration.SetInt(MRJobConfig.NumMaps, 3);
InputSplit[] lSplits = format.GetSplits(configuration, 3);
NUnit.Framework.Assert.AreEqual(5, lSplits[0].GetLength());
NUnit.Framework.Assert.AreEqual(3, lSplits.Length);
// test reader .Some simple tests
NUnit.Framework.Assert.AreEqual(typeof(LongWritable), reader.CreateKey().GetType(
));
NUnit.Framework.Assert.AreEqual(0, reader.GetPos());
NUnit.Framework.Assert.AreEqual(0, reader.GetProgress(), 0.001);
reader.Close();
}
/// <exception cref="System.IO.IOException"/>
internal static long ReadBench(JobConf conf)
{
// InputFormat instantiation
InputFormat inf = conf.GetInputFormat();
string fn = conf.Get("test.filebench.name", string.Empty);
Path pin = new Path(FileInputFormat.GetInputPaths(conf)[0], fn);
FileStatus @in = pin.GetFileSystem(conf).GetFileStatus(pin);
RecordReader rr = inf.GetRecordReader(new FileSplit(pin, 0, @in.GetLen(), (string
[])null), conf, Reporter.Null);
try
{
object key = rr.CreateKey();
object val = rr.CreateValue();
DateTime start = new DateTime();
while (rr.Next(key, val))
{
}
DateTime end = new DateTime();
return(end.GetTime() - start.GetTime());
}
finally
{
rr.Close();
}
}
/// <summary>Run the map task.</summary>
/// <param name="input">the set of inputs</param>
/// <param name="output">the object to collect the outputs of the map</param>
/// <param name="reporter">the object to update with status</param>
/// <exception cref="System.IO.IOException"/>
public override void Run(RecordReader <K1, V1> input, OutputCollector <K2, V2> output
, Reporter reporter)
{
Application <K1, V1, K2, V2> application = null;
try
{
RecordReader <FloatWritable, NullWritable> fakeInput = (!Submitter.GetIsJavaRecordReader
(job) && !Submitter.GetIsJavaMapper(job)) ? (RecordReader <FloatWritable, NullWritable
>)input : null;
application = new Application <K1, V1, K2, V2>(job, fakeInput, output, reporter, (
Type)job.GetOutputKeyClass(), (Type)job.GetOutputValueClass());
}
catch (Exception ie)
{
throw new RuntimeException("interrupted", ie);
}
DownwardProtocol <K1, V1> downlink = application.GetDownlink();
bool isJavaInput = Submitter.GetIsJavaRecordReader(job);
downlink.RunMap(reporter.GetInputSplit(), job.GetNumReduceTasks(), isJavaInput);
bool skipping = job.GetBoolean(MRJobConfig.SkipRecords, false);
try
{
if (isJavaInput)
{
// allocate key & value instances that are re-used for all entries
K1 key = input.CreateKey();
V1 value = input.CreateValue();
downlink.SetInputTypes(key.GetType().FullName, value.GetType().FullName);
while (input.Next(key, value))
{
// map pair to output
downlink.MapItem(key, value);
if (skipping)
{
//flush the streams on every record input if running in skip mode
//so that we don't buffer other records surrounding a bad record.
downlink.Flush();
}
}
downlink.EndOfInput();
}
application.WaitForFinish();
}
catch (Exception t)
{
application.Abort(t);
}
finally
{
application.Cleanup();
}
}
/// <exception cref="System.IO.IOException"/>
private static IList <Text> ReadSplit(TextInputFormat format, InputSplit split, JobConf
jobConf)
{
IList <Text> result = new AList <Text>();
RecordReader <LongWritable, Text> reader = format.GetRecordReader(split, jobConf,
voidReporter);
LongWritable key = reader.CreateKey();
Text value = reader.CreateValue();
while (reader.Next(key, value))
{
result.AddItem(value);
value = reader.CreateValue();
}
reader.Close();
return(result);
}
// A reporter that does nothing
/// <exception cref="System.IO.IOException"/>
internal virtual void CheckFormat(JobConf job, int expectedN)
{
NLineInputFormat format = new NLineInputFormat();
format.Configure(job);
int ignoredNumSplits = 1;
InputSplit[] splits = format.GetSplits(job, ignoredNumSplits);
// check all splits except last one
int count = 0;
for (int j = 0; j < splits.Length - 1; j++)
{
NUnit.Framework.Assert.AreEqual("There are no split locations", 0, splits[j].GetLocations
().Length);
RecordReader <LongWritable, Text> reader = format.GetRecordReader(splits[j], job,
voidReporter);
Type readerClass = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is LineRecordReader.", typeof(LineRecordReader
), readerClass);
LongWritable key = reader.CreateKey();
Type keyClass = key.GetType();
NUnit.Framework.Assert.AreEqual("Key class is LongWritable.", typeof(LongWritable
), keyClass);
Text value = reader.CreateValue();
Type valueClass = value.GetType();
NUnit.Framework.Assert.AreEqual("Value class is Text.", typeof(Text), valueClass);
try
{
count = 0;
while (reader.Next(key, value))
{
count++;
}
}
finally
{
reader.Close();
}
NUnit.Framework.Assert.AreEqual("number of lines in split is " + expectedN, expectedN
, count);
}
}
/// <exception cref="System.IO.IOException"/>
private static IList <string> ReadSplit(FixedLengthInputFormat format, InputSplit
split, JobConf job)
{
IList <string> result = new AList <string>();
RecordReader <LongWritable, BytesWritable> reader = format.GetRecordReader(split,
job, voidReporter);
LongWritable key = reader.CreateKey();
BytesWritable value = reader.CreateValue();
try
{
while (reader.Next(key, value))
{
result.AddItem(Sharpen.Runtime.GetStringForBytes(value.GetBytes(), 0, value.GetLength
()));
}
}
finally
{
reader.Close();
}
return(result);
}
/// <exception cref="System.IO.IOException"/>
public virtual void Run(RecordReader <K1, V1> input, OutputCollector <K2, V2> output
, Reporter reporter)
{
try
{
// allocate key & value instances that are re-used for all entries
K1 key = input.CreateKey();
V1 value = input.CreateValue();
while (input.Next(key, value))
{
// map pair to output
mapper.Map(key, value, output, reporter);
if (incrProcCount)
{
reporter.IncrCounter(SkipBadRecords.CounterGroup, SkipBadRecords.CounterMapProcessedRecords
, 1);
}
}
}
finally
{
mapper.Close();
}
}
/// <exception cref="System.Exception"/>
public virtual void TestFormat()
{
JobConf job = new JobConf(conf);
FileSystem fs = FileSystem.GetLocal(conf);
Path dir = new Path(Runtime.GetProperty("test.build.data", ".") + "/mapred");
Path file = new Path(dir, "test.seq");
Reporter reporter = Reporter.Null;
int seed = new Random().Next();
//LOG.info("seed = "+seed);
Random random = new Random(seed);
fs.Delete(dir, true);
FileInputFormat.SetInputPaths(job, dir);
// for a variety of lengths
for (int length = 0; length < MaxLength; length += random.Next(MaxLength / 10) +
1)
{
//LOG.info("creating; entries = " + length);
// create a file with length entries
SequenceFile.Writer writer = SequenceFile.CreateWriter(fs, conf, file, typeof(IntWritable
), typeof(LongWritable));
try
{
for (int i = 0; i < length; i++)
{
IntWritable key = new IntWritable(i);
LongWritable value = new LongWritable(10 * i);
writer.Append(key, value);
}
}
finally
{
writer.Close();
}
// try splitting the file in a variety of sizes
InputFormat <Text, Text> format = new SequenceFileAsTextInputFormat();
for (int i_1 = 0; i_1 < 3; i_1++)
{
int numSplits = random.Next(MaxLength / (SequenceFile.SyncInterval / 20)) + 1;
//LOG.info("splitting: requesting = " + numSplits);
InputSplit[] splits = format.GetSplits(job, numSplits);
//LOG.info("splitting: got = " + splits.length);
// check each split
BitSet bits = new BitSet(length);
for (int j = 0; j < splits.Length; j++)
{
RecordReader <Text, Text> reader = format.GetRecordReader(splits[j], job, reporter
);
Type readerClass = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is SequenceFileAsTextRecordReader."
, typeof(SequenceFileAsTextRecordReader), readerClass);
Text value = reader.CreateValue();
Text key = reader.CreateKey();
try
{
int count = 0;
while (reader.Next(key, value))
{
// if (bits.get(key.get())) {
// LOG.info("splits["+j+"]="+splits[j]+" : " + key.get());
// LOG.info("@"+reader.getPos());
// }
int keyInt = System.Convert.ToInt32(key.ToString());
NUnit.Framework.Assert.IsFalse("Key in multiple partitions.", bits.Get(keyInt));
bits.Set(keyInt);
count++;
}
}
finally
{
//LOG.info("splits["+j+"]="+splits[j]+" count=" + count);
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some keys in no partition.", length, bits.Cardinality
());
}
}
}
/// <exception cref="System.IO.IOException"/>
public virtual void Run(RecordReader <K1, V1> input, OutputCollector <K2, V2> output
, Reporter reporter)
{
try
{
// allocate key & value instances these objects will not be reused
// because execution of Mapper.map is not serialized.
K1 key = input.CreateKey();
V1 value = input.CreateValue();
while (input.Next(key, value))
{
executorService.Execute(new MultithreadedMapRunner.MapperInvokeRunable(this, key,
value, output, reporter));
CheckForExceptionsFromProcessingThreads();
// Allocate new key & value instances as mapper is running in parallel
key = input.CreateKey();
value = input.CreateValue();
}
if (Log.IsDebugEnabled())
{
Log.Debug("Finished dispatching all Mappper.map calls, job " + job.GetJobName());
}
// Graceful shutdown of the Threadpool, it will let all scheduled
// Runnables to end.
executorService.Shutdown();
try
{
// Now waiting for all Runnables to end.
while (!executorService.AwaitTermination(100, TimeUnit.Milliseconds))
{
if (Log.IsDebugEnabled())
{
Log.Debug("Awaiting all running Mappper.map calls to finish, job " + job.GetJobName
());
}
// NOTE: while Mapper.map dispatching has concluded there are still
// map calls in progress and exceptions would be thrown.
CheckForExceptionsFromProcessingThreads();
}
// NOTE: it could be that a map call has had an exception after the
// call for awaitTermination() returing true. And edge case but it
// could happen.
CheckForExceptionsFromProcessingThreads();
}
catch (IOException ioEx)
{
// Forcing a shutdown of all thread of the threadpool and rethrowing
// the IOException
executorService.ShutdownNow();
throw;
}
catch (Exception iEx)
{
throw new RuntimeException(iEx);
}
}
finally
{
mapper.Close();
}
}
/// <exception cref="System.Exception"/>
public virtual void TestFormat()
{
JobConf job = new JobConf();
Path file = new Path(workDir, "test.txt");
// A reporter that does nothing
Reporter reporter = Reporter.Null;
int seed = new Random().Next();
Log.Info("seed = " + seed);
Random random = new Random(seed);
localFs.Delete(workDir, true);
FileInputFormat.SetInputPaths(job, workDir);
// for a variety of lengths
for (int length = 0; length < MaxLength; length += random.Next(MaxLength / 10) +
1)
{
Log.Debug("creating; entries = " + length);
// create a file with length entries
TextWriter writer = new OutputStreamWriter(localFs.Create(file));
try
{
for (int i = 0; i < length; i++)
{
writer.Write(Sharpen.Extensions.ToString(i * 2));
writer.Write("\t");
writer.Write(Sharpen.Extensions.ToString(i));
writer.Write("\n");
}
}
finally
{
writer.Close();
}
// try splitting the file in a variety of sizes
KeyValueTextInputFormat format = new KeyValueTextInputFormat();
format.Configure(job);
for (int i_1 = 0; i_1 < 3; i_1++)
{
int numSplits = random.Next(MaxLength / 20) + 1;
Log.Debug("splitting: requesting = " + numSplits);
InputSplit[] splits = format.GetSplits(job, numSplits);
Log.Debug("splitting: got = " + splits.Length);
// check each split
BitSet bits = new BitSet(length);
for (int j = 0; j < splits.Length; j++)
{
Log.Debug("split[" + j + "]= " + splits[j]);
RecordReader <Text, Text> reader = format.GetRecordReader(splits[j], job, reporter
);
Type readerClass = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is KeyValueLineRecordReader.", typeof(
KeyValueLineRecordReader), readerClass);
Text key = reader.CreateKey();
Type keyClass = key.GetType();
Text value = reader.CreateValue();
Type valueClass = value.GetType();
NUnit.Framework.Assert.AreEqual("Key class is Text.", typeof(Text), keyClass);
NUnit.Framework.Assert.AreEqual("Value class is Text.", typeof(Text), valueClass);
try
{
int count = 0;
while (reader.Next(key, value))
{
int v = System.Convert.ToInt32(value.ToString());
Log.Debug("read " + v);
if (bits.Get(v))
{
Log.Warn("conflict with " + v + " in split " + j + " at position " + reader.GetPos
());
}
NUnit.Framework.Assert.IsFalse("Key in multiple partitions.", bits.Get(v));
bits.Set(v);
count++;
}
Log.Debug("splits[" + j + "]=" + splits[j] + " count=" + count);
}
finally
{
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some keys in no partition.", length, bits.Cardinality
());
}
}
}
public virtual K CreateKey()
{
return(delegate_.CreateKey());
}
public virtual K CreateKey()
{
return(curReader.CreateKey());
}
/// <summary>Request new key from proxied RR.</summary>
public virtual K CreateKey()
{
return(rr.CreateKey());
}
