/// <exception cref="System.Exception"/>
public virtual void TestEmptyJoin()
{
JobConf job = new JobConf();
Path @base = cluster.GetFileSystem().MakeQualified(new Path("/empty"));
Path[] src = new Path[] { new Path(@base, "i0"), new Path("i1"), new Path("i2") };
job.Set("mapreduce.join.expr", CompositeInputFormat.Compose("outer", typeof(TestDatamerge.Fake_IF
), src));
job.SetInputFormat(typeof(CompositeInputFormat));
FileOutputFormat.SetOutputPath(job, new Path(@base, "out"));
job.SetMapperClass(typeof(IdentityMapper));
job.SetReducerClass(typeof(IdentityReducer));
job.SetOutputKeyClass(typeof(IncomparableKey));
job.SetOutputValueClass(typeof(NullWritable));
JobClient.RunJob(job);
@base.GetFileSystem(job).Delete(@base, true);
}
/// <exception cref="System.Exception"/>
private static void JoinAs(string jointype, Type c)
{
int srcs = 4;
Configuration conf = new Configuration();
JobConf job = new JobConf(conf, c);
Path @base = cluster.GetFileSystem().MakeQualified(new Path("/" + jointype));
Path[] src = WriteSimpleSrc(@base, conf, srcs);
job.Set("mapreduce.join.expr", CompositeInputFormat.Compose(jointype, typeof(SequenceFileInputFormat
), src));
job.SetInt("testdatamerge.sources", srcs);
job.SetInputFormat(typeof(CompositeInputFormat));
FileOutputFormat.SetOutputPath(job, new Path(@base, "out"));
job.SetMapperClass(c);
job.SetReducerClass(c);
job.SetOutputKeyClass(typeof(IntWritable));
job.SetOutputValueClass(typeof(IntWritable));
JobClient.RunJob(job);
@base.GetFileSystem(job).Delete(@base, true);
}
/// <summary>
/// Tests the class loader set by
/// <see cref="Org.Apache.Hadoop.Conf.Configuration.SetClassLoader(Sharpen.ClassLoader)
/// "/>
/// is inherited by any
/// <see cref="WrappedRecordReader{K, U}"/>
/// s created by
/// <see cref="CompositeRecordReader{K, V, X}"/>
/// </summary>
/// <exception cref="System.Exception"/>
public virtual void TestClassLoader()
{
JobConf job = new JobConf();
TestWrappedRecordReaderClassloader.Fake_ClassLoader classLoader = new TestWrappedRecordReaderClassloader.Fake_ClassLoader
();
job.SetClassLoader(classLoader);
NUnit.Framework.Assert.IsTrue(job.GetClassLoader() is TestWrappedRecordReaderClassloader.Fake_ClassLoader
);
FileSystem fs = FileSystem.Get(job);
Path testdir = new Path(Runtime.GetProperty("test.build.data", "/tmp")).MakeQualified
(fs);
Path @base = new Path(testdir, "/empty");
Path[] src = new Path[] { new Path(@base, "i0"), new Path("i1"), new Path("i2") };
job.Set("mapreduce.join.expr", CompositeInputFormat.Compose("outer", typeof(TestWrappedRecordReaderClassloader.IF_ClassLoaderChecker
), src));
CompositeInputFormat <NullWritable> inputFormat = new CompositeInputFormat <NullWritable
>();
inputFormat.GetRecordReader(inputFormat.GetSplits(job, 1)[0], job, Reporter.Null);
}
/// <exception cref="System.Exception"/>
public virtual void TestNestedJoin()
{
// outer(inner(S1,...,Sn),outer(S1,...Sn))
int Sources = 3;
int Items = (Sources + 1) * (Sources + 1);
JobConf job = new JobConf();
Path @base = cluster.GetFileSystem().MakeQualified(new Path("/nested"));
int[][] source = new int[Sources][];
for (int i = 0; i < Sources; ++i)
{
source[i] = new int[Items];
for (int j = 0; j < Items; ++j)
{
source[i][j] = (i + 2) * (j + 1);
}
}
Path[] src = new Path[Sources];
SequenceFile.Writer[] @out = CreateWriters(@base, job, Sources, src);
IntWritable k = new IntWritable();
for (int i_1 = 0; i_1 < Sources; ++i_1)
{
IntWritable v = new IntWritable();
v.Set(i_1);
for (int j = 0; j < Items; ++j)
{
k.Set(source[i_1][j]);
@out[i_1].Append(k, v);
}
@out[i_1].Close();
}
@out = null;
StringBuilder sb = new StringBuilder();
sb.Append("outer(inner(");
for (int i_2 = 0; i_2 < Sources; ++i_2)
{
sb.Append(CompositeInputFormat.Compose(typeof(SequenceFileInputFormat), src[i_2].
ToString()));
if (i_2 + 1 != Sources)
{
sb.Append(",");
}
}
sb.Append("),outer(");
sb.Append(CompositeInputFormat.Compose(typeof(TestDatamerge.Fake_IF), "foobar"));
sb.Append(",");
for (int i_3 = 0; i_3 < Sources; ++i_3)
{
sb.Append(CompositeInputFormat.Compose(typeof(SequenceFileInputFormat), src[i_3].
ToString()));
sb.Append(",");
}
sb.Append(CompositeInputFormat.Compose(typeof(TestDatamerge.Fake_IF), "raboof") +
"))");
job.Set("mapreduce.join.expr", sb.ToString());
job.SetInputFormat(typeof(CompositeInputFormat));
Path outf = new Path(@base, "out");
FileOutputFormat.SetOutputPath(job, outf);
TestDatamerge.Fake_IF.SetKeyClass(job, typeof(IntWritable));
TestDatamerge.Fake_IF.SetValClass(job, typeof(IntWritable));
job.SetMapperClass(typeof(IdentityMapper));
job.SetReducerClass(typeof(IdentityReducer));
job.SetNumReduceTasks(0);
job.SetOutputKeyClass(typeof(IntWritable));
job.SetOutputValueClass(typeof(TupleWritable));
job.SetOutputFormat(typeof(SequenceFileOutputFormat));
JobClient.RunJob(job);
FileStatus[] outlist = cluster.GetFileSystem().ListStatus(outf, new Utils.OutputFileUtils.OutputFilesFilter
());
NUnit.Framework.Assert.AreEqual(1, outlist.Length);
NUnit.Framework.Assert.IsTrue(0 < outlist[0].GetLen());
SequenceFile.Reader r = new SequenceFile.Reader(cluster.GetFileSystem(), outlist[
0].GetPath(), job);
TupleWritable v_1 = new TupleWritable();
while (r.Next(k, v_1))
{
NUnit.Framework.Assert.IsFalse(((TupleWritable)v_1.Get(1)).Has(0));
NUnit.Framework.Assert.IsFalse(((TupleWritable)v_1.Get(1)).Has(Sources + 1));
bool chk = true;
int ki = k.Get();
for (int i_4 = 2; i_4 < Sources + 2; ++i_4)
{
if ((ki % i_4) == 0 && ki <= i_4 * Items)
{
NUnit.Framework.Assert.AreEqual(i_4 - 2, ((IntWritable)((TupleWritable)v_1.Get(1)
).Get((i_4 - 1))).Get());
}
else
{
chk = false;
}
}
if (chk)
{
// present in all sources; chk inner
NUnit.Framework.Assert.IsTrue(v_1.Has(0));
for (int i_5 = 0; i_5 < Sources; ++i_5)
{
NUnit.Framework.Assert.IsTrue(((TupleWritable)v_1.Get(0)).Has(i_5));
}
}
else
{
// should not be present in inner join
NUnit.Framework.Assert.IsFalse(v_1.Has(0));
}
}
r.Close();
@base.GetFileSystem(job).Delete(@base, true);
}
