Search in sources :

Example 11 with DummyInvokable

use of org.apache.flink.runtime.operators.testutils.DummyInvokable in project flink by apache.

the class MassiveStringSorting method testStringTuplesSorting.

@SuppressWarnings("unchecked")
public void testStringTuplesSorting() {
    final int NUM_STRINGS = 300000;
    File input = null;
    File sorted = null;
    try {
        // the source file
        input = generateFileWithStringTuples(NUM_STRINGS, "http://some-uri.com/that/is/a/common/prefix/to/all");
        // the sorted file
        sorted = File.createTempFile("sorted_strings", "txt");
        String[] command = { "/bin/bash", "-c", "export LC_ALL=\"C\" && cat \"" + input.getAbsolutePath() + "\" | sort > \"" + sorted.getAbsolutePath() + "\"" };
        Process p = null;
        try {
            p = Runtime.getRuntime().exec(command);
            int retCode = p.waitFor();
            if (retCode != 0) {
                throw new Exception("Command failed with return code " + retCode);
            }
            p = null;
        } finally {
            if (p != null) {
                p.destroy();
            }
        }
        // sort the data
        UnilateralSortMerger<Tuple2<String, String[]>> sorter = null;
        BufferedReader reader = null;
        BufferedReader verifyReader = null;
        try {
            MemoryManager mm = new MemoryManager(1024 * 1024, 1);
            IOManager ioMan = new IOManagerAsync();
            TupleTypeInfo<Tuple2<String, String[]>> typeInfo = (TupleTypeInfo<Tuple2<String, String[]>>) TypeInfoParser.<Tuple2<String, String[]>>parse("Tuple2<String, String[]>");
            TypeSerializer<Tuple2<String, String[]>> serializer = typeInfo.createSerializer(new ExecutionConfig());
            TypeComparator<Tuple2<String, String[]>> comparator = typeInfo.createComparator(new int[] { 0 }, new boolean[] { true }, 0, new ExecutionConfig());
            reader = new BufferedReader(new FileReader(input));
            MutableObjectIterator<Tuple2<String, String[]>> inputIterator = new StringTupleReaderMutableObjectIterator(reader);
            sorter = new UnilateralSortMerger<Tuple2<String, String[]>>(mm, ioMan, inputIterator, new DummyInvokable(), new RuntimeSerializerFactory<Tuple2<String, String[]>>(serializer, (Class<Tuple2<String, String[]>>) (Class<?>) Tuple2.class), comparator, 1.0, 4, 0.8f, true, /* use large record handler */
            false);
            // use this part to verify that all if good when sorting in memory
            //				List<MemorySegment> memory = mm.allocatePages(new DummyInvokable(), mm.computeNumberOfPages(1024*1024*1024));
            //				NormalizedKeySorter<Tuple2<String, String[]>> nks = new NormalizedKeySorter<Tuple2<String,String[]>>(serializer, comparator, memory);
            //
            //				{
            //					Tuple2<String, String[]> wi = new Tuple2<String, String[]>("", new String[0]);
            //					while ((wi = inputIterator.next(wi)) != null) {
            //						Assert.assertTrue(nks.write(wi));
            //					}
            //					
            //					new QuickSort().sort(nks);
            //				}
            //				
            //				MutableObjectIterator<Tuple2<String, String[]>> sortedData = nks.getIterator();
            MutableObjectIterator<Tuple2<String, String[]>> sortedData = sorter.getIterator();
            reader.close();
            // verify
            verifyReader = new BufferedReader(new FileReader(sorted));
            MutableObjectIterator<Tuple2<String, String[]>> verifyIterator = new StringTupleReaderMutableObjectIterator(verifyReader);
            Tuple2<String, String[]> next = new Tuple2<String, String[]>("", new String[0]);
            Tuple2<String, String[]> nextFromStratoSort = new Tuple2<String, String[]>("", new String[0]);
            int num = 0;
            while ((next = verifyIterator.next(next)) != null) {
                num++;
                nextFromStratoSort = sortedData.next(nextFromStratoSort);
                Assert.assertNotNull(nextFromStratoSort);
                Assert.assertEquals(next.f0, nextFromStratoSort.f0);
                Assert.assertArrayEquals(next.f1, nextFromStratoSort.f1);
            }
            Assert.assertNull(sortedData.next(nextFromStratoSort));
            Assert.assertEquals(NUM_STRINGS, num);
        } finally {
            if (reader != null) {
                reader.close();
            }
            if (verifyReader != null) {
                verifyReader.close();
            }
            if (sorter != null) {
                sorter.close();
            }
        }
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        Assert.fail(e.getMessage());
    } finally {
        if (input != null) {
            input.delete();
        }
        if (sorted != null) {
            sorted.delete();
        }
    }
}
Also used : IOManager(org.apache.flink.runtime.io.disk.iomanager.IOManager) RuntimeSerializerFactory(org.apache.flink.api.java.typeutils.runtime.RuntimeSerializerFactory) ExecutionConfig(org.apache.flink.api.common.ExecutionConfig) MemoryManager(org.apache.flink.runtime.memory.MemoryManager) IOException(java.io.IOException) TupleTypeInfo(org.apache.flink.api.java.typeutils.TupleTypeInfo) IOManagerAsync(org.apache.flink.runtime.io.disk.iomanager.IOManagerAsync) Tuple2(org.apache.flink.api.java.tuple.Tuple2) BufferedReader(java.io.BufferedReader) FileReader(java.io.FileReader) DummyInvokable(org.apache.flink.runtime.operators.testutils.DummyInvokable) File(java.io.File)

Example 12 with DummyInvokable

use of org.apache.flink.runtime.operators.testutils.DummyInvokable in project flink by apache.

the class MassiveStringValueSorting method testStringValueTuplesSorting.

@SuppressWarnings("unchecked")
public void testStringValueTuplesSorting() {
    final int NUM_STRINGS = 300000;
    File input = null;
    File sorted = null;
    try {
        // the source file
        input = generateFileWithStringTuples(NUM_STRINGS, "http://some-uri.com/that/is/a/common/prefix/to/all");
        // the sorted file
        sorted = File.createTempFile("sorted_strings", "txt");
        String[] command = { "/bin/bash", "-c", "export LC_ALL=\"C\" && cat \"" + input.getAbsolutePath() + "\" | sort > \"" + sorted.getAbsolutePath() + "\"" };
        Process p = null;
        try {
            p = Runtime.getRuntime().exec(command);
            int retCode = p.waitFor();
            if (retCode != 0) {
                throw new Exception("Command failed with return code " + retCode);
            }
            p = null;
        } finally {
            if (p != null) {
                p.destroy();
            }
        }
        // sort the data
        UnilateralSortMerger<Tuple2<StringValue, StringValue[]>> sorter = null;
        BufferedReader reader = null;
        BufferedReader verifyReader = null;
        try {
            MemoryManager mm = new MemoryManager(1024 * 1024, 1);
            IOManager ioMan = new IOManagerAsync();
            TupleTypeInfo<Tuple2<StringValue, StringValue[]>> typeInfo = (TupleTypeInfo<Tuple2<StringValue, StringValue[]>>) TypeInfoParser.<Tuple2<StringValue, StringValue[]>>parse("Tuple2<org.apache.flink.types.StringValue, org.apache.flink.types.StringValue[]>");
            TypeSerializer<Tuple2<StringValue, StringValue[]>> serializer = typeInfo.createSerializer(new ExecutionConfig());
            TypeComparator<Tuple2<StringValue, StringValue[]>> comparator = typeInfo.createComparator(new int[] { 0 }, new boolean[] { true }, 0, new ExecutionConfig());
            reader = new BufferedReader(new FileReader(input));
            MutableObjectIterator<Tuple2<StringValue, StringValue[]>> inputIterator = new StringValueTupleReaderMutableObjectIterator(reader);
            sorter = new UnilateralSortMerger<Tuple2<StringValue, StringValue[]>>(mm, ioMan, inputIterator, new DummyInvokable(), new RuntimeSerializerFactory<Tuple2<StringValue, StringValue[]>>(serializer, (Class<Tuple2<StringValue, StringValue[]>>) (Class<?>) Tuple2.class), comparator, 1.0, 4, 0.8f, true, /* use large record handler */
            false);
            // use this part to verify that all if good when sorting in memory
            //				List<MemorySegment> memory = mm.allocatePages(new DummyInvokable(), mm.computeNumberOfPages(1024*1024*1024));
            //				NormalizedKeySorter<Tuple2<String, String[]>> nks = new NormalizedKeySorter<Tuple2<String,String[]>>(serializer, comparator, memory);
            //
            //				{
            //					Tuple2<String, String[]> wi = new Tuple2<String, String[]>("", new String[0]);
            //					while ((wi = inputIterator.next(wi)) != null) {
            //						Assert.assertTrue(nks.write(wi));
            //					}
            //					
            //					new QuickSort().sort(nks);
            //				}
            //				
            //				MutableObjectIterator<Tuple2<String, String[]>> sortedData = nks.getIterator();
            MutableObjectIterator<Tuple2<StringValue, StringValue[]>> sortedData = sorter.getIterator();
            reader.close();
            // verify
            verifyReader = new BufferedReader(new FileReader(sorted));
            MutableObjectIterator<Tuple2<StringValue, StringValue[]>> verifyIterator = new StringValueTupleReaderMutableObjectIterator(verifyReader);
            Tuple2<StringValue, StringValue[]> nextVerify = new Tuple2<StringValue, StringValue[]>(new StringValue(), new StringValue[0]);
            Tuple2<StringValue, StringValue[]> nextFromFlinkSort = new Tuple2<StringValue, StringValue[]>(new StringValue(), new StringValue[0]);
            int num = 0;
            while ((nextVerify = verifyIterator.next(nextVerify)) != null) {
                num++;
                nextFromFlinkSort = sortedData.next(nextFromFlinkSort);
                Assert.assertNotNull(nextFromFlinkSort);
                Assert.assertEquals(nextVerify.f0, nextFromFlinkSort.f0);
                Assert.assertArrayEquals(nextVerify.f1, nextFromFlinkSort.f1);
            }
            Assert.assertNull(sortedData.next(nextFromFlinkSort));
            Assert.assertEquals(NUM_STRINGS, num);
        } finally {
            if (reader != null) {
                reader.close();
            }
            if (verifyReader != null) {
                verifyReader.close();
            }
            if (sorter != null) {
                sorter.close();
            }
        }
    } catch (Exception e) {
        System.err.println(e.getMessage());
        e.printStackTrace();
        Assert.fail(e.getMessage());
    } finally {
        if (input != null) {
            //noinspection ResultOfMethodCallIgnored
            input.delete();
        }
        if (sorted != null) {
            //noinspection ResultOfMethodCallIgnored
            sorted.delete();
        }
    }
}
Also used : RuntimeSerializerFactory(org.apache.flink.api.java.typeutils.runtime.RuntimeSerializerFactory) ExecutionConfig(org.apache.flink.api.common.ExecutionConfig) IOManagerAsync(org.apache.flink.runtime.io.disk.iomanager.IOManagerAsync) FileReader(java.io.FileReader) DummyInvokable(org.apache.flink.runtime.operators.testutils.DummyInvokable) StringValue(org.apache.flink.types.StringValue) IOManager(org.apache.flink.runtime.io.disk.iomanager.IOManager) MemoryManager(org.apache.flink.runtime.memory.MemoryManager) IOException(java.io.IOException) TupleTypeInfo(org.apache.flink.api.java.typeutils.TupleTypeInfo) Tuple2(org.apache.flink.api.java.tuple.Tuple2) BufferedReader(java.io.BufferedReader) File(java.io.File)

Example 13 with DummyInvokable

use of org.apache.flink.runtime.operators.testutils.DummyInvokable in project flink by apache.

the class NonReusingBlockResettableIteratorTest method testTwelveFoldBufferedBlockResettableIterator.

@Test
public void testTwelveFoldBufferedBlockResettableIterator() throws Exception {
    final AbstractInvokable memOwner = new DummyInvokable();
    // create the resettable Iterator
    final NonReusingBlockResettableIterator<Record> iterator = new NonReusingBlockResettableIterator<Record>(this.memman, this.reader, this.serializer, 12, memOwner);
    // open the iterator
    iterator.open();
    // now test walking through the iterator
    int lower = 0;
    int upper = 0;
    do {
        lower = upper;
        upper = lower;
        // find the upper bound
        while (iterator.hasNext()) {
            Record target = iterator.next();
            int val = target.getField(0, IntValue.class).getValue();
            Assert.assertEquals(upper++, val);
        }
        // now reset the buffer a few times
        for (int i = 0; i < 5; ++i) {
            iterator.reset();
            int count = 0;
            while (iterator.hasNext()) {
                Record target = iterator.next();
                int val = target.getField(0, IntValue.class).getValue();
                Assert.assertEquals(lower + (count++), val);
            }
            Assert.assertEquals(upper - lower, count);
        }
    } while (iterator.nextBlock());
    Assert.assertEquals(NUM_VALUES, upper);
    // close the iterator
    iterator.close();
}
Also used : DummyInvokable(org.apache.flink.runtime.operators.testutils.DummyInvokable) Record(org.apache.flink.types.Record) AbstractInvokable(org.apache.flink.runtime.jobgraph.tasks.AbstractInvokable) IntValue(org.apache.flink.types.IntValue) Test(org.junit.Test)

Example 14 with DummyInvokable

use of org.apache.flink.runtime.operators.testutils.DummyInvokable in project flink by apache.

the class NonReusingBlockResettableIteratorTest method testSerialBlockResettableIterator.

@Test
public void testSerialBlockResettableIterator() throws Exception {
    final AbstractInvokable memOwner = new DummyInvokable();
    // create the resettable Iterator
    final NonReusingBlockResettableIterator<Record> iterator = new NonReusingBlockResettableIterator<Record>(this.memman, this.reader, this.serializer, 1, memOwner);
    // open the iterator
    iterator.open();
    // now test walking through the iterator
    int lower = 0;
    int upper = 0;
    do {
        lower = upper;
        upper = lower;
        // find the upper bound
        while (iterator.hasNext()) {
            Record target = iterator.next();
            int val = target.getField(0, IntValue.class).getValue();
            Assert.assertEquals(upper++, val);
        }
        // now reset the buffer a few times
        for (int i = 0; i < 5; ++i) {
            iterator.reset();
            int count = 0;
            while (iterator.hasNext()) {
                Record target = iterator.next();
                int val = target.getField(0, IntValue.class).getValue();
                Assert.assertEquals(lower + (count++), val);
            }
            Assert.assertEquals(upper - lower, count);
        }
    } while (iterator.nextBlock());
    Assert.assertEquals(NUM_VALUES, upper);
    // close the iterator
    iterator.close();
}
Also used : DummyInvokable(org.apache.flink.runtime.operators.testutils.DummyInvokable) Record(org.apache.flink.types.Record) AbstractInvokable(org.apache.flink.runtime.jobgraph.tasks.AbstractInvokable) IntValue(org.apache.flink.types.IntValue) Test(org.junit.Test)

Example 15 with DummyInvokable

use of org.apache.flink.runtime.operators.testutils.DummyInvokable in project flink by apache.

the class NonReusingBlockResettableIteratorTest method testDoubleBufferedBlockResettableIterator.

@Test
public void testDoubleBufferedBlockResettableIterator() throws Exception {
    final AbstractInvokable memOwner = new DummyInvokable();
    // create the resettable Iterator
    final NonReusingBlockResettableIterator<Record> iterator = new NonReusingBlockResettableIterator<Record>(this.memman, this.reader, this.serializer, 2, memOwner);
    // open the iterator
    iterator.open();
    // now test walking through the iterator
    int lower = 0;
    int upper = 0;
    do {
        lower = upper;
        upper = lower;
        // find the upper bound
        while (iterator.hasNext()) {
            Record target = iterator.next();
            int val = target.getField(0, IntValue.class).getValue();
            Assert.assertEquals(upper++, val);
        }
        // now reset the buffer a few times
        for (int i = 0; i < 5; ++i) {
            iterator.reset();
            int count = 0;
            while (iterator.hasNext()) {
                Record target = iterator.next();
                int val = target.getField(0, IntValue.class).getValue();
                Assert.assertEquals(lower + (count++), val);
            }
            Assert.assertEquals(upper - lower, count);
        }
    } while (iterator.nextBlock());
    Assert.assertEquals(NUM_VALUES, upper);
    // close the iterator
    iterator.close();
}
Also used : DummyInvokable(org.apache.flink.runtime.operators.testutils.DummyInvokable) Record(org.apache.flink.types.Record) AbstractInvokable(org.apache.flink.runtime.jobgraph.tasks.AbstractInvokable) IntValue(org.apache.flink.types.IntValue) Test(org.junit.Test)

Aggregations

DummyInvokable (org.apache.flink.runtime.operators.testutils.DummyInvokable)51 Test (org.junit.Test)46 MemorySegment (org.apache.flink.core.memory.MemorySegment)38 AbstractInvokable (org.apache.flink.runtime.jobgraph.tasks.AbstractInvokable)22 Tuple2 (org.apache.flink.api.java.tuple.Tuple2)12 IOManager (org.apache.flink.runtime.io.disk.iomanager.IOManager)12 IOManagerAsync (org.apache.flink.runtime.io.disk.iomanager.IOManagerAsync)12 MemoryManager (org.apache.flink.runtime.memory.MemoryManager)12 FileIOChannel (org.apache.flink.runtime.io.disk.iomanager.FileIOChannel)11 ArrayList (java.util.ArrayList)10 Random (java.util.Random)8 TestData (org.apache.flink.runtime.operators.testutils.TestData)8 IntPair (org.apache.flink.runtime.operators.testutils.types.IntPair)8 Record (org.apache.flink.types.Record)8 EOFException (java.io.EOFException)7 File (java.io.File)7 ExecutionConfig (org.apache.flink.api.common.ExecutionConfig)7 TupleTypeInfo (org.apache.flink.api.java.typeutils.TupleTypeInfo)7 IOException (java.io.IOException)6 IntValue (org.apache.flink.types.IntValue)6