public virtual void TestFormatWithCustomSeparator()
{
JobConf job = new JobConf();
string separator = "\u0001";
job.Set("mapreduce.output.textoutputformat.separator", separator);
job.Set(JobContext.TaskAttemptId, attempt);
FileOutputFormat.SetOutputPath(job, workDir.GetParent().GetParent());
FileOutputFormat.SetWorkOutputPath(job, workDir);
FileSystem fs = workDir.GetFileSystem(job);
if (!fs.Mkdirs(workDir))
{
NUnit.Framework.Assert.Fail("Failed to create output directory");
}
string file = "test_custom.txt";
// A reporter that does nothing
Reporter reporter = Reporter.Null;
TextOutputFormat <object, object> theOutputFormat = new TextOutputFormat <object, object
>();
RecordWriter <object, object> theRecordWriter = theOutputFormat.GetRecordWriter(localFs
, job, file, reporter);
Org.Apache.Hadoop.IO.Text key1 = new Org.Apache.Hadoop.IO.Text("key1");
Org.Apache.Hadoop.IO.Text key2 = new Org.Apache.Hadoop.IO.Text("key2");
Org.Apache.Hadoop.IO.Text val1 = new Org.Apache.Hadoop.IO.Text("val1");
Org.Apache.Hadoop.IO.Text val2 = new Org.Apache.Hadoop.IO.Text("val2");
NullWritable nullWritable = NullWritable.Get();
try
{
theRecordWriter.Write(key1, val1);
theRecordWriter.Write(null, nullWritable);
theRecordWriter.Write(null, val1);
theRecordWriter.Write(nullWritable, val2);
theRecordWriter.Write(key2, nullWritable);
theRecordWriter.Write(key1, null);
theRecordWriter.Write(null, null);
theRecordWriter.Write(key2, val2);
}
finally
{
theRecordWriter.Close(reporter);
}
FilePath expectedFile = new FilePath(new Path(workDir, file).ToString());
StringBuilder expectedOutput = new StringBuilder();
expectedOutput.Append(key1).Append(separator).Append(val1).Append("\n");
expectedOutput.Append(val1).Append("\n");
expectedOutput.Append(val2).Append("\n");
expectedOutput.Append(key2).Append("\n");
expectedOutput.Append(key1).Append("\n");
expectedOutput.Append(key2).Append(separator).Append(val2).Append("\n");
string output = UtilsForTests.Slurp(expectedFile);
NUnit.Framework.Assert.AreEqual(expectedOutput.ToString(), output);
}
/// <exception cref="System.IO.IOException"/>
public virtual void TestcheckOutputSpecsForbidRecordCompression()
{
JobConf job = new JobConf();
FileSystem fs = FileSystem.GetLocal(job);
Path dir = new Path(Runtime.GetProperty("test.build.data", ".") + "/mapred");
Path outputdir = new Path(Runtime.GetProperty("test.build.data", ".") + "/output"
);
fs.Delete(dir, true);
fs.Delete(outputdir, true);
if (!fs.Mkdirs(dir))
{
Fail("Failed to create output directory");
}
FileOutputFormat.SetWorkOutputPath(job, dir);
// Without outputpath, FileOutputFormat.checkoutputspecs will throw
// InvalidJobConfException
FileOutputFormat.SetOutputPath(job, outputdir);
// SequenceFileAsBinaryOutputFormat doesn't support record compression
// It should throw an exception when checked by checkOutputSpecs
SequenceFileAsBinaryOutputFormat.SetCompressOutput(job, true);
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Block);
try
{
new SequenceFileAsBinaryOutputFormat().CheckOutputSpecs(fs, job);
}
catch (Exception e)
{
Fail("Block compression should be allowed for " + "SequenceFileAsBinaryOutputFormat:"
+ "Caught " + e.GetType().FullName);
}
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Record);
try
{
new SequenceFileAsBinaryOutputFormat().CheckOutputSpecs(fs, job);
Fail("Record compression should not be allowed for " + "SequenceFileAsBinaryOutputFormat"
);
}
catch (InvalidJobConfException)
{
}
catch (Exception e)
{
// expected
Fail("Expected " + typeof(InvalidJobConfException).FullName + "but caught " + e.GetType
().FullName);
}
}
/// <exception cref="System.Exception"/>
public virtual void TestCommitter()
{
JobConf job = new JobConf();
SetConfForFileOutputCommitter(job);
JobContext jContext = new JobContextImpl(job, ((JobID)taskID.GetJobID()));
TaskAttemptContext tContext = new TaskAttemptContextImpl(job, taskID);
FileOutputCommitter committer = new FileOutputCommitter();
FileOutputFormat.SetWorkOutputPath(job, committer.GetTaskAttemptPath(tContext));
committer.SetupJob(jContext);
committer.SetupTask(tContext);
string file = "test.txt";
// A reporter that does nothing
Reporter reporter = Reporter.Null;
// write output
FileSystem localFs = FileSystem.GetLocal(job);
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(localFs, job, file
, reporter);
WriteOutput(theRecordWriter, reporter);
// do commit
committer.CommitTask(tContext);
committer.CommitJob(jContext);
// validate output
FilePath expectedFile = new FilePath(new Path(outDir, file).ToString());
StringBuilder expectedOutput = new StringBuilder();
expectedOutput.Append(key1).Append('\t').Append(val1).Append("\n");
expectedOutput.Append(val1).Append("\n");
expectedOutput.Append(val2).Append("\n");
expectedOutput.Append(key2).Append("\n");
expectedOutput.Append(key1).Append("\n");
expectedOutput.Append(key2).Append('\t').Append(val2).Append("\n");
string output = UtilsForTests.Slurp(expectedFile);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.IO.IOException"/>
public virtual void TestAbort()
{
JobConf job = new JobConf();
SetConfForFileOutputCommitter(job);
JobContext jContext = new JobContextImpl(job, ((JobID)taskID.GetJobID()));
TaskAttemptContext tContext = new TaskAttemptContextImpl(job, taskID);
FileOutputCommitter committer = new FileOutputCommitter();
FileOutputFormat.SetWorkOutputPath(job, committer.GetTaskAttemptPath(tContext));
// do setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
string file = "test.txt";
// A reporter that does nothing
Reporter reporter = Reporter.Null;
// write output
FileSystem localFs = FileSystem.GetLocal(job);
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(localFs, job, file
, reporter);
WriteOutput(theRecordWriter, reporter);
// do abort
committer.AbortTask(tContext);
FilePath expectedFile = new FilePath(new Path(committer.GetTaskAttemptPath(tContext
), file).ToString());
NUnit.Framework.Assert.IsFalse("task temp dir still exists", expectedFile.Exists(
));
committer.AbortJob(jContext, JobStatus.State.Failed);
expectedFile = new FilePath(new Path(outDir, FileOutputCommitter.TempDirName).ToString
());
NUnit.Framework.Assert.IsFalse("job temp dir " + expectedFile + " still exists",
expectedFile.Exists());
NUnit.Framework.Assert.AreEqual("Output directory not empty", 0, new FilePath(outDir
.ToString()).ListFiles().Length);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
// A random task attempt id for testing.
/// <exception cref="System.IO.IOException"/>
public virtual void TestBinary()
{
JobConf job = new JobConf();
FileSystem fs = FileSystem.GetLocal(job);
Path dir = new Path(new Path(new Path(Runtime.GetProperty("test.build.data", ".")
), FileOutputCommitter.TempDirName), "_" + attempt);
Path file = new Path(dir, "testbinary.seq");
Random r = new Random();
long seed = r.NextLong();
r.SetSeed(seed);
fs.Delete(dir, true);
if (!fs.Mkdirs(dir))
{
Fail("Failed to create output directory");
}
job.Set(JobContext.TaskAttemptId, attempt);
FileOutputFormat.SetOutputPath(job, dir.GetParent().GetParent());
FileOutputFormat.SetWorkOutputPath(job, dir);
SequenceFileAsBinaryOutputFormat.SetSequenceFileOutputKeyClass(job, typeof(IntWritable
));
SequenceFileAsBinaryOutputFormat.SetSequenceFileOutputValueClass(job, typeof(DoubleWritable
));
SequenceFileAsBinaryOutputFormat.SetCompressOutput(job, true);
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Block);
BytesWritable bkey = new BytesWritable();
BytesWritable bval = new BytesWritable();
RecordWriter <BytesWritable, BytesWritable> writer = new SequenceFileAsBinaryOutputFormat
().GetRecordWriter(fs, job, file.ToString(), Reporter.Null);
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.Next());
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(Reporter.Null);
}
InputFormat <IntWritable, DoubleWritable> iformat = new SequenceFileInputFormat <IntWritable
, DoubleWritable>();
int count = 0;
r.SetSeed(seed);
DataInputBuffer buf = new DataInputBuffer();
int NumSplits = 3;
SequenceFileInputFormat.AddInputPath(job, file);
Log.Info("Reading data by SequenceFileInputFormat");
foreach (InputSplit split in iformat.GetSplits(job, NumSplits))
{
RecordReader <IntWritable, DoubleWritable> reader = iformat.GetRecordReader(split,
job, Reporter.Null);
try
{
int sourceInt;
double sourceDouble;
while (reader.Next(iwritable, dwritable))
{
sourceInt = r.Next();
sourceDouble = r.NextDouble();
NUnit.Framework.Assert.AreEqual("Keys don't match: " + "*" + iwritable.Get() + ":"
+ sourceInt + "*", sourceInt, iwritable.Get());
NUnit.Framework.Assert.IsTrue("Vals don't match: " + "*" + dwritable.Get() + ":"
+ sourceDouble + "*", double.Compare(dwritable.Get(), sourceDouble) == 0);
++count;
}
}
finally
{
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some records not found", Records, count);
}
/// <exception cref="System.Exception"/>
public virtual void TestFormat()
{
JobConf job = new JobConf();
job.Set(JobContext.TaskAttemptId, attempt);
FileOutputFormat.SetOutputPath(job, workDir.GetParent().GetParent());
FileOutputFormat.SetWorkOutputPath(job, workDir);
FileSystem fs = workDir.GetFileSystem(job);
if (!fs.Mkdirs(workDir))
{
Fail("Failed to create output directory");
}
//System.out.printf("workdir: %s\n", workDir.toString());
TestMultipleTextOutputFormat.Test1(job);
TestMultipleTextOutputFormat.Test2(job);
string file_11 = "1-part-00000";
FilePath expectedFile_11 = new FilePath(new Path(workDir, file_11).ToString());
//System.out.printf("expectedFile_11: %s\n", new Path(workDir, file_11).toString());
StringBuilder expectedOutput = new StringBuilder();
for (int i = 10; i < 20; i++)
{
expectedOutput.Append(string.Empty + i).Append('\t').Append(string.Empty + i).Append
("\n");
}
string output = UtilsForTests.Slurp(expectedFile_11);
//System.out.printf("File_2 output: %s\n", output);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
string file_12 = "2-part-00000";
FilePath expectedFile_12 = new FilePath(new Path(workDir, file_12).ToString());
//System.out.printf("expectedFile_12: %s\n", new Path(workDir, file_12).toString());
expectedOutput = new StringBuilder();
for (int i_1 = 20; i_1 < 30; i_1++)
{
expectedOutput.Append(string.Empty + i_1).Append('\t').Append(string.Empty + i_1)
.Append("\n");
}
output = UtilsForTests.Slurp(expectedFile_12);
//System.out.printf("File_2 output: %s\n", output);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
string file_13 = "3-part-00000";
FilePath expectedFile_13 = new FilePath(new Path(workDir, file_13).ToString());
//System.out.printf("expectedFile_13: %s\n", new Path(workDir, file_13).toString());
expectedOutput = new StringBuilder();
for (int i_2 = 30; i_2 < 40; i_2++)
{
expectedOutput.Append(string.Empty + i_2).Append('\t').Append(string.Empty + i_2)
.Append("\n");
}
output = UtilsForTests.Slurp(expectedFile_13);
//System.out.printf("File_2 output: %s\n", output);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
string file_2 = "2/3";
FilePath expectedFile_2 = new FilePath(new Path(workDir, file_2).ToString());
//System.out.printf("expectedFile_2: %s\n", new Path(workDir, file_2).toString());
expectedOutput = new StringBuilder();
for (int i_3 = 10; i_3 < 40; i_3++)
{
expectedOutput.Append(string.Empty + i_3).Append('\t').Append(string.Empty + i_3)
.Append("\n");
}
output = UtilsForTests.Slurp(expectedFile_2);
//System.out.printf("File_2 output: %s\n", output);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
}
public virtual void TestCompress()
{
JobConf job = new JobConf();
job.Set(JobContext.TaskAttemptId, attempt);
job.Set(FileOutputFormat.Compress, "true");
FileOutputFormat.SetOutputPath(job, workDir.GetParent().GetParent());
FileOutputFormat.SetWorkOutputPath(job, workDir);
FileSystem fs = workDir.GetFileSystem(job);
if (!fs.Mkdirs(workDir))
{
NUnit.Framework.Assert.Fail("Failed to create output directory");
}
string file = "test_compress.txt";
// A reporter that does nothing
Reporter reporter = Reporter.Null;
TextOutputFormat <object, object> theOutputFormat = new TextOutputFormat <object, object
>();
RecordWriter <object, object> theRecordWriter = theOutputFormat.GetRecordWriter(localFs
, job, file, reporter);
Org.Apache.Hadoop.IO.Text key1 = new Org.Apache.Hadoop.IO.Text("key1");
Org.Apache.Hadoop.IO.Text key2 = new Org.Apache.Hadoop.IO.Text("key2");
Org.Apache.Hadoop.IO.Text val1 = new Org.Apache.Hadoop.IO.Text("val1");
Org.Apache.Hadoop.IO.Text val2 = new Org.Apache.Hadoop.IO.Text("val2");
NullWritable nullWritable = NullWritable.Get();
try
{
theRecordWriter.Write(key1, val1);
theRecordWriter.Write(null, nullWritable);
theRecordWriter.Write(null, val1);
theRecordWriter.Write(nullWritable, val2);
theRecordWriter.Write(key2, nullWritable);
theRecordWriter.Write(key1, null);
theRecordWriter.Write(null, null);
theRecordWriter.Write(key2, val2);
}
finally
{
theRecordWriter.Close(reporter);
}
StringBuilder expectedOutput = new StringBuilder();
expectedOutput.Append(key1).Append("\t").Append(val1).Append("\n");
expectedOutput.Append(val1).Append("\n");
expectedOutput.Append(val2).Append("\n");
expectedOutput.Append(key2).Append("\n");
expectedOutput.Append(key1).Append("\n");
expectedOutput.Append(key2).Append("\t").Append(val2).Append("\n");
DefaultCodec codec = new DefaultCodec();
codec.SetConf(job);
Path expectedFile = new Path(workDir, file + codec.GetDefaultExtension());
FileInputStream istream = new FileInputStream(expectedFile.ToString());
CompressionInputStream cistream = codec.CreateInputStream(istream);
LineReader reader = new LineReader(cistream);
string output = string.Empty;
Org.Apache.Hadoop.IO.Text @out = new Org.Apache.Hadoop.IO.Text();
while (reader.ReadLine(@out) > 0)
{
output += @out;
output += "\n";
}
reader.Close();
NUnit.Framework.Assert.AreEqual(expectedOutput.ToString(), output);
}
/// <exception cref="System.IO.IOException"/>
public virtual void TestFailAbort()
{
JobConf job = new JobConf();
job.Set(FileSystem.FsDefaultNameKey, "faildel:///");
job.SetClass("fs.faildel.impl", typeof(TestMRCJCFileOutputCommitter.FakeFileSystem
), typeof(FileSystem));
SetConfForFileOutputCommitter(job);
JobContext jContext = new JobContextImpl(job, ((JobID)taskID.GetJobID()));
TaskAttemptContext tContext = new TaskAttemptContextImpl(job, taskID);
FileOutputCommitter committer = new FileOutputCommitter();
FileOutputFormat.SetWorkOutputPath(job, committer.GetTaskAttemptPath(tContext));
// do setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
string file = "test.txt";
FilePath jobTmpDir = new FilePath(committer.GetJobAttemptPath(jContext).ToUri().GetPath
());
FilePath taskTmpDir = new FilePath(committer.GetTaskAttemptPath(tContext).ToUri()
.GetPath());
FilePath expectedFile = new FilePath(taskTmpDir, file);
// A reporter that does nothing
Reporter reporter = Reporter.Null;
// write output
FileSystem localFs = new TestMRCJCFileOutputCommitter.FakeFileSystem();
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(localFs, job, expectedFile
.GetAbsolutePath(), reporter);
WriteOutput(theRecordWriter, reporter);
// do abort
Exception th = null;
try
{
committer.AbortTask(tContext);
}
catch (IOException ie)
{
th = ie;
}
NUnit.Framework.Assert.IsNotNull(th);
NUnit.Framework.Assert.IsTrue(th is IOException);
NUnit.Framework.Assert.IsTrue(th.Message.Contains("fake delete failed"));
NUnit.Framework.Assert.IsTrue(expectedFile + " does not exists", expectedFile.Exists
());
th = null;
try
{
committer.AbortJob(jContext, JobStatus.State.Failed);
}
catch (IOException ie)
{
th = ie;
}
NUnit.Framework.Assert.IsNotNull(th);
NUnit.Framework.Assert.IsTrue(th is IOException);
NUnit.Framework.Assert.IsTrue(th.Message.Contains("fake delete failed"));
NUnit.Framework.Assert.IsTrue("job temp dir does not exists", jobTmpDir.Exists());
}