/// <exception cref="System.IO.IOException"/>
/// <exception cref="System.Exception"/>
internal virtual void CheckFormat(Job job, int expectedN, int lastN)
{
NLineInputFormat format = new NLineInputFormat();
IList <InputSplit> splits = format.GetSplits(job);
int count = 0;
for (int i = 0; i < splits.Count; i++)
{
NUnit.Framework.Assert.AreEqual("There are no split locations", 0, splits[i].GetLocations
().Length);
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
RecordReader <LongWritable, Text> reader = format.CreateRecordReader(splits[i], context
);
Type clazz = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is LineRecordReader.", typeof(LineRecordReader
), clazz);
MapContext <LongWritable, Text, LongWritable, Text> mcontext = new MapContextImpl <
LongWritable, Text, LongWritable, Text>(job.GetConfiguration(), context.GetTaskAttemptID
(), reader, null, null, MapReduceTestUtil.CreateDummyReporter(), splits[i]);
reader.Initialize(splits[i], mcontext);
try
{
count = 0;
while (reader.NextKeyValue())
{
count++;
}
}
finally
{
reader.Close();
}
if (i == splits.Count - 1)
{
NUnit.Framework.Assert.AreEqual("number of lines in split(" + i + ") is wrong", lastN
, count);
}
else
{
NUnit.Framework.Assert.AreEqual("number of lines in split(" + i + ") is wrong", expectedN
, count);
}
}
}
// 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.Exception"/>
public virtual void TestFormat()
{
Job job = Job.GetInstance(new Configuration(defaultConf));
Random random = new Random();
long seed = random.NextLong();
Log.Info("seed = " + seed);
random.SetSeed(seed);
localFs.Delete(workDir, true);
FileInputFormat.SetInputPaths(job, workDir);
int length = 10000;
int numFiles = 10;
// create files with various lengths
CreateFiles(length, numFiles, random);
// create a combined split for the files
CombineTextInputFormat format = new CombineTextInputFormat();
for (int i = 0; i < 3; i++)
{
int numSplits = random.Next(length / 20) + 1;
Log.Info("splitting: requesting = " + numSplits);
IList <InputSplit> splits = format.GetSplits(job);
Log.Info("splitting: got = " + splits.Count);
// we should have a single split as the length is comfortably smaller than
// the block size
NUnit.Framework.Assert.AreEqual("We got more than one splits!", 1, splits.Count);
InputSplit split = splits[0];
NUnit.Framework.Assert.AreEqual("It should be CombineFileSplit", typeof(CombineFileSplit
), split.GetType());
// check the split
BitSet bits = new BitSet(length);
Log.Debug("split= " + split);
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
RecordReader <LongWritable, Text> reader = format.CreateRecordReader(split, context
);
NUnit.Framework.Assert.AreEqual("reader class is CombineFileRecordReader.", typeof(
CombineFileRecordReader), reader.GetType());
MapContext <LongWritable, Text, LongWritable, Text> mcontext = new MapContextImpl <
LongWritable, Text, LongWritable, Text>(job.GetConfiguration(), context.GetTaskAttemptID
(), reader, null, null, MapReduceTestUtil.CreateDummyReporter(), split);
reader.Initialize(split, mcontext);
try
{
int count = 0;
while (reader.NextKeyValue())
{
LongWritable key = reader.GetCurrentKey();
NUnit.Framework.Assert.IsNotNull("Key should not be null.", key);
Text value = reader.GetCurrentValue();
int v = System.Convert.ToInt32(value.ToString());
Log.Debug("read " + v);
NUnit.Framework.Assert.IsFalse("Key in multiple partitions.", bits.Get(v));
bits.Set(v);
count++;
}
Log.Debug("split=" + split + " count=" + count);
}
finally
{
reader.Close();
}
NUnit.Framework.Assert.AreEqual("Some keys in no partition.", length, bits.Cardinality
());
}
}
/// <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
());
}
}
}
/// <exception cref="System.IO.IOException"/>
private void RunRandomTests(CompressionCodec codec)
{
StringBuilder fileName = new StringBuilder("testFormat.txt");
if (codec != null)
{
fileName.Append(".gz");
}
localFs.Delete(workDir, true);
Path file = new Path(workDir, fileName.ToString());
int seed = new Random().Next();
Log.Info("Seed = " + seed);
Random random = new Random(seed);
int MaxTests = 20;
LongWritable key = new LongWritable();
BytesWritable value = new BytesWritable();
for (int i = 0; i < MaxTests; i++)
{
Log.Info("----------------------------------------------------------");
// Maximum total records of 999
int totalRecords = random.Next(999) + 1;
// Test an empty file
if (i == 8)
{
totalRecords = 0;
}
// Maximum bytes in a record of 100K
int recordLength = random.Next(1024 * 100) + 1;
// For the 11th test, force a record length of 1
if (i == 10)
{
recordLength = 1;
}
// The total bytes in the test file
int fileSize = (totalRecords * recordLength);
Log.Info("totalRecords=" + totalRecords + " recordLength=" + recordLength);
// Create the job
JobConf job = new JobConf(defaultConf);
if (codec != null)
{
ReflectionUtils.SetConf(codec, job);
}
// Create the test file
AList <string> recordList = CreateFile(file, codec, recordLength, totalRecords);
NUnit.Framework.Assert.IsTrue(localFs.Exists(file));
//set the fixed length record length config property for the job
FixedLengthInputFormat.SetRecordLength(job, recordLength);
int numSplits = 1;
// Arbitrarily set number of splits.
if (i > 0)
{
if (i == (MaxTests - 1))
{
// Test a split size that is less than record len
numSplits = (int)(fileSize / Math.Floor(recordLength / 2));
}
else
{
if (MaxTests % i == 0)
{
// Let us create a split size that is forced to be
// smaller than the end file itself, (ensures 1+ splits)
numSplits = fileSize / (fileSize - random.Next(fileSize));
}
else
{
// Just pick a random split size with no upper bound
numSplits = Math.Max(1, fileSize / random.Next(int.MaxValue));
}
}
Log.Info("Number of splits set to: " + numSplits);
}
// Setup the input path
FileInputFormat.SetInputPaths(job, workDir);
// Try splitting the file in a variety of sizes
FixedLengthInputFormat format = new FixedLengthInputFormat();
format.Configure(job);
InputSplit[] splits = format.GetSplits(job, numSplits);
Log.Info("Actual number of splits = " + splits.Length);
// Test combined split lengths = total file size
long recordOffset = 0;
int recordNumber = 0;
foreach (InputSplit split in splits)
{
RecordReader <LongWritable, BytesWritable> reader = format.GetRecordReader(split,
job, voidReporter);
Type clazz = reader.GetType();
NUnit.Framework.Assert.AreEqual("RecordReader class should be FixedLengthRecordReader:"
, typeof(FixedLengthRecordReader), clazz);
// Plow through the records in this split
while (reader.Next(key, value))
{
NUnit.Framework.Assert.AreEqual("Checking key", (long)(recordNumber * recordLength
), key.Get());
string valueString = Sharpen.Runtime.GetStringForBytes(value.GetBytes(), 0, value
.GetLength());
NUnit.Framework.Assert.AreEqual("Checking record length:", recordLength, value.GetLength
());
NUnit.Framework.Assert.IsTrue("Checking for more records than expected:", recordNumber
< totalRecords);
string origRecord = recordList[recordNumber];
NUnit.Framework.Assert.AreEqual("Checking record content:", origRecord, valueString
);
recordNumber++;
}
reader.Close();
}
NUnit.Framework.Assert.AreEqual("Total original records should be total read records:"
, recordList.Count, recordNumber);
}
}
public virtual void TestFormat()
{
Job job = Job.GetInstance(new Configuration(defaultConf));
Path file = new Path(workDir, "test.txt");
int seed = new Random().Next();
Log.Info("seed = " + seed);
Random random = new Random(seed);
localFs.Delete(workDir, true);
FileInputFormat.SetInputPaths(job, workDir);
int MaxLength = 10000;
// 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();
for (int i_1 = 0; i_1 < 3; i_1++)
{
int numSplits = random.Next(MaxLength / 20) + 1;
Log.Debug("splitting: requesting = " + numSplits);
IList <InputSplit> splits = format.GetSplits(job);
Log.Debug("splitting: got = " + splits.Count);
// check each split
BitSet bits = new BitSet(length);
for (int j = 0; j < splits.Count; j++)
{
Log.Debug("split[" + j + "]= " + splits[j]);
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
RecordReader <Text, Text> reader = format.CreateRecordReader(splits[j], context);
Type clazz = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is KeyValueLineRecordReader.", typeof(
KeyValueLineRecordReader), clazz);
MapContext <Text, Text, Text, Text> mcontext = new MapContextImpl <Text, Text, Text
, Text>(job.GetConfiguration(), context.GetTaskAttemptID(), reader, null, null,
MapReduceTestUtil.CreateDummyReporter(), splits[j]);
reader.Initialize(splits[j], mcontext);
Text key = null;
Text value = null;
try
{
int count = 0;
while (reader.NextKeyValue())
{
key = reader.GetCurrentKey();
clazz = key.GetType();
NUnit.Framework.Assert.AreEqual("Key class is Text.", typeof(Text), clazz);
value = reader.GetCurrentValue();
clazz = value.GetType();
NUnit.Framework.Assert.AreEqual("Value class is Text.", typeof(Text), clazz);
int k = System.Convert.ToInt32(key.ToString());
int v = System.Convert.ToInt32(value.ToString());
NUnit.Framework.Assert.AreEqual("Bad key", 0, k % 2);
NUnit.Framework.Assert.AreEqual("Mismatched key/value", k / 2, v);
Log.Debug("read " + v);
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 void TestSplitableCodecs()
{
Job job = Job.GetInstance(defaultConf);
Configuration conf = job.GetConfiguration();
// Create the codec
CompressionCodec codec = null;
try
{
codec = (CompressionCodec)ReflectionUtils.NewInstance(conf.GetClassByName("org.apache.hadoop.io.compress.BZip2Codec"
), conf);
}
catch (TypeLoadException)
{
throw new IOException("Illegal codec!");
}
Path file = new Path(workDir, "test" + codec.GetDefaultExtension());
int seed = new Random().Next();
Log.Info("seed = " + seed);
Random random = new Random(seed);
localFs.Delete(workDir, true);
FileInputFormat.SetInputPaths(job, workDir);
int MaxLength = 500000;
FileInputFormat.SetMaxInputSplitSize(job, MaxLength / 20);
// for a variety of lengths
for (int length = 0; length < MaxLength; length += random.Next(MaxLength / 4) + 1)
{
Log.Info("creating; entries = " + length);
// create a file with length entries
TextWriter writer = new OutputStreamWriter(codec.CreateOutputStream(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();
NUnit.Framework.Assert.IsTrue("KVTIF claims not splittable", format.IsSplitable(job
, file));
for (int i_1 = 0; i_1 < 3; i_1++)
{
int numSplits = random.Next(MaxLength / 2000) + 1;
Log.Info("splitting: requesting = " + numSplits);
IList <InputSplit> splits = format.GetSplits(job);
Log.Info("splitting: got = " + splits.Count);
// check each split
BitSet bits = new BitSet(length);
for (int j = 0; j < splits.Count; j++)
{
Log.Debug("split[" + j + "]= " + splits[j]);
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
RecordReader <Text, Text> reader = format.CreateRecordReader(splits[j], context);
Type clazz = reader.GetType();
MapContext <Text, Text, Text, Text> mcontext = new MapContextImpl <Text, Text, Text
, Text>(job.GetConfiguration(), context.GetTaskAttemptID(), reader, null, null,
MapReduceTestUtil.CreateDummyReporter(), splits[j]);
reader.Initialize(splits[j], mcontext);
Text key = null;
Text value = null;
try
{
int count = 0;
while (reader.NextKeyValue())
{
key = reader.GetCurrentKey();
value = reader.GetCurrentValue();
int k = System.Convert.ToInt32(key.ToString());
int v = System.Convert.ToInt32(value.ToString());
NUnit.Framework.Assert.AreEqual("Bad key", 0, k % 2);
NUnit.Framework.Assert.AreEqual("Mismatched key/value", k / 2, v);
Log.Debug("read " + k + "," + v);
NUnit.Framework.Assert.IsFalse(k + "," + v + " in multiple partitions.", bits.Get
(v));
bits.Set(v);
count++;
}
if (count > 0)
{
Log.Info("splits[" + j + "]=" + splits[j] + " count=" + count);
}
else
{
Log.Debug("splits[" + j + "]=" + splits[j] + " count=" + count);
}
}
finally
{
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some keys in no partition.", length, bits.Cardinality
());
}
}
}
/// <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.Exception"/>
public virtual void TestFormat()
{
Job job = Job.GetInstance(conf);
FileSystem fs = FileSystem.GetLocal(conf);
Path dir = new Path(Runtime.GetProperty("test.build.data", ".") + "/mapred");
Path file = new Path(dir, "test.seq");
int seed = new Random().Next();
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)
{
// 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();
}
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
// 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++)
{
// check each split
BitSet bits = new BitSet(length);
int numSplits = random.Next(MaxLength / (SequenceFile.SyncInterval / 20)) + 1;
FileInputFormat.SetMaxInputSplitSize(job, fs.GetFileStatus(file).GetLen() / numSplits
);
foreach (InputSplit split in format.GetSplits(job))
{
RecordReader <Text, Text> reader = format.CreateRecordReader(split, context);
MapContext <Text, Text, Text, Text> mcontext = new MapContextImpl <Text, Text, Text
, Text>(job.GetConfiguration(), context.GetTaskAttemptID(), reader, null, null,
MapReduceTestUtil.CreateDummyReporter(), split);
reader.Initialize(split, mcontext);
Type readerClass = reader.GetType();
NUnit.Framework.Assert.AreEqual("reader class is SequenceFileAsTextRecordReader."
, typeof(SequenceFileAsTextRecordReader), readerClass);
Text key;
try
{
int count = 0;
while (reader.NextKeyValue())
{
key = reader.GetCurrentKey();
int keyInt = System.Convert.ToInt32(key.ToString());
NUnit.Framework.Assert.IsFalse("Key in multiple partitions.", bits.Get(keyInt));
bits.Set(keyInt);
count++;
}
}
finally
{
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some keys in no partition.", length, bits.Cardinality
());
}
}
}
