Java Code Examples for org.apache.pig.data.BagFactory#getInstance()

@Test
public void testUseDefaultDelimiterBagToStringUDF() throws Exception {
	BagFactory bf = BagFactory.getInstance();
	TupleFactory tf = TupleFactory.getInstance();

	Tuple t1 = tf.newTuple(2);
	t1.set(0, "a");
	t1.set(1, 5);

	Tuple t2 = tf.newTuple(2);
	t2.set(0, "c");
	t2.set(1, 6);

	DataBag bag = bf.newDefaultBag();
	bag.add(t1);
	bag.add(t2);

	BagToString udf = new BagToString();
	Tuple udfInput = tf.newTuple(1);
	udfInput.set(0, bag);
	String result = udf.exec(udfInput);

	assertEquals("a_5_c_6", result);
}
 

@Test
public void testBasicBagToStringUDF() throws Exception {
	BagFactory bf = BagFactory.getInstance();
	TupleFactory tf = TupleFactory.getInstance();

	Tuple t1 = tf.newTuple(2);
	t1.set(0, "a");
	t1.set(1, 5);

	Tuple t2 = tf.newTuple(2);
	t2.set(0, "c");
	t2.set(1, 6);

	DataBag bag = bf.newDefaultBag();
	bag.add(t1);
	bag.add(t2);

	BagToString udf = new BagToString();
	Tuple udfInput = tf.newTuple(2);
	udfInput.set(0, bag);
	udfInput.set(1, "-");
	String result = udf.exec(udfInput);

	assertEquals("a-5-c-6", result);
}
 

@Test
public void tupleFromBagAccumulateTest() throws Exception
{
  TupleFactory tf = TupleFactory.getInstance();
  BagFactory bf = BagFactory.getInstance();
 
  TupleFromBag op = new TupleFromBag();
  
  Tuple defaultValue = tf.newTuple(1000);
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(4))), 0, defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(9))), 0, defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(16))), 0, defaultValue)));
  assertEquals(op.getValue(), tf.newTuple(4));
  op.cleanup();
  
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(11))), 1, defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(17))), 1, defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(5))), 1, defaultValue)));
  assertEquals(op.getValue(), tf.newTuple(17));
  op.cleanup();
  
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(), 2, defaultValue)));
  assertEquals(op.getValue(), defaultValue);
  op.cleanup();
}
 

@Test
public void firstTupleFromBagAccumulateTest() throws Exception
{
  TupleFactory tf = TupleFactory.getInstance();
  BagFactory bf = BagFactory.getInstance();
 
  FirstTupleFromBag op = new FirstTupleFromBag();
  
  Tuple defaultValue = tf.newTuple(1000);
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(4))), defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(9))), defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(16))), defaultValue)));
  assertEquals(op.getValue(), tf.newTuple(4));
  op.cleanup();
  
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(11))), defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(17))), defaultValue)));
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(Arrays.asList(tf.newTuple(5))), defaultValue)));
  assertEquals(op.getValue(), tf.newTuple(11));
  op.cleanup();
  
  op.accumulate(tf.newTuple(Arrays.asList(bf.newDefaultBag(), defaultValue)));
  assertEquals(op.getValue(), defaultValue);
  op.cleanup();
}
 

/**
 * Implementation for PIG_BAG functions. Builds a Pig DataBag from
 * the corresponding input
 *
 * @param elements Input that contains a bag
 * @return Pig Tuple
 */
public static Tuple buildBag(Object... elements) {
  final TupleFactory tupleFactory = TupleFactory.getInstance();
  final BagFactory bagFactory = BagFactory.getInstance();
  // Convert each row into a Tuple
  List<Tuple> tupleList = new ArrayList<>();
  if (elements != null) {
    // The first input contains a list of rows for the bag
    final List bag = (elements[0] instanceof List)
        ? (List) elements[0]
        : Collections.singletonList(elements[0]);
    for (Object row : bag) {
      tupleList.add(tupleFactory.newTuple(Arrays.asList(row)));
    }
  }

  // Then build a bag from the tuple list
  DataBag resultBag = bagFactory.newDefaultBag(tupleList);

  // The returned result is a new Tuple with the newly constructed DataBag
  // as the first item.
  List<Object> finalTuple = new ArrayList<>();
  finalTuple.add(resultBag);

  if (elements != null) {
    // Add the remaining elements from the input
    for (int i = 1; i < elements.length; i++) {
      finalTuple.add(elements[i]);
    }
  }

  return tupleFactory.newTuple(finalTuple);
}
 

@Override
public Map<String, Object> exec(Tuple input) throws IOException {

    TupleFactory tupleFactory = TupleFactory.getInstance();
    ArrayList<Object> objList = new ArrayList<Object>();
    objList.add(new Integer(1));
    objList.add(new Double(1.0));
    objList.add(new Float(1.0));
    objList.add(new String("World!"));
    Tuple tuple = tupleFactory.newTuple(objList);

    BagFactory bagFactory = BagFactory.getInstance();
    DataBag bag = bagFactory.newDefaultBag();
    bag.add(tuple);

    Map<String, Object> mapInMap = new HashMap<String, Object>();
    mapInMap.put("int", new Integer(10));
    mapInMap.put("float", new Float(10.0));

    Map<String, Object> myMap = new HashMap<String, Object>();
    myMap.put("string", new String("Hello"));
    myMap.put("int", new Integer(1));
    myMap.put("long", new Long(1));
    myMap.put("float", new Float(1.0));
    myMap.put("double", new Double(1.0));
    myMap.put("dba", new DataByteArray(new String("bytes").getBytes()));
    myMap.put("map", mapInMap);
    myMap.put("tuple", tuple);
    myMap.put("bag", bag);
    return myMap; 
}
 

@Test
public void testMap() throws Exception {
    String size = "100";
    String numHash = "3";
    String hashFunc = "JENKINS";
    TupleFactory tf = TupleFactory.getInstance();
    BagFactory bf = BagFactory.getInstance();

    Tuple t = tf.newTuple(1);
    t.set(0, 1);
    DataBag b = bf.newDefaultBag();
    b.add(t);
    Tuple input = tf.newTuple(b);

    BuildBloom.Initial map =
            new BuildBloom.Initial(hashFunc, "fixed", size, numHash);
    t = map.exec(input);

    Bloom bloom = new Bloom("bla");
    bloom.setFilter((DataByteArray)t.get(0));

    // Test that everything we put in passes.
    Tuple t1 = tf.newTuple(1);
    t1.set(0, 1);
    assertTrue(bloom.exec(t1));

    // A few that don't pass
    for (int i = 100; i < 10; i++) {
        Tuple t2 = tf.newTuple(1);
        t2.set(0, i);
        assertFalse(bloom.exec(t2));
    }
}
 

@Test
public void testNestedTupleForBagToStringUDF() throws Exception {
	BagFactory bf = BagFactory.getInstance();
	TupleFactory tf = TupleFactory.getInstance();

	Tuple t1 = tf.newTuple(2);
	t1.set(0, "a");
	t1.set(1, 5);

	Tuple nestedTuple = tf.newTuple(2);
	nestedTuple.set(0, "d");
	nestedTuple.set(1, 7);

	Tuple t2 = tf.newTuple(3);
	t2.set(0, "c");
	t2.set(1, 6);
	t2.set(2, nestedTuple);

	DataBag inputBag = bf.newDefaultBag();
	inputBag.add(t1);
	inputBag.add(t2);

	BagToString udf = new BagToString();
	Tuple udfInput = tf.newTuple(2);
	udfInput.set(0, inputBag);
	udfInput.set(1, "_");
	String result = udf.exec(udfInput);

	assertEquals("a_5_c_6_(d,7)", result);
}
 

@Override
public Map<String, Object> exec(Tuple input) throws IOException {

    TupleFactory tupleFactory = TupleFactory.getInstance();
    ArrayList<Object> objList = new ArrayList<Object>();
    objList.add(new Integer(1));
    objList.add(new Double(1.0));
    objList.add(new Float(1.0));
    objList.add(new String("World!"));
    Tuple tuple = tupleFactory.newTuple(objList);

    BagFactory bagFactory = BagFactory.getInstance();
    DataBag bag = bagFactory.newDefaultBag();
    bag.add(tuple);

    Map<String, Object> mapInMap = new HashMap<String, Object>();
    mapInMap.put("int", new Integer(10));
    mapInMap.put("float", new Float(10.0));

    Map<String, Object> myMap = new HashMap<String, Object>();
    myMap.put("string", new String("Hello"));
    myMap.put("int", new Integer(1));
    myMap.put("long", new Long(1));
    myMap.put("float", new Float(1.0));
    myMap.put("double", new Double(1.0));
    myMap.put("dba", new DataByteArray(new String("bytes").getBytes()));
    myMap.put("map", mapInMap);
    myMap.put("tuple", tuple);
    myMap.put("bag", bag);
    return myMap;
}
 

static DataBag createDataBag() {
    BagFactory bagFactory = BagFactory.getInstance();
    return bagFactory.newDefaultBag();
}
 

public NullableBag() {
    if (mFactory == null) {
        mFactory = BagFactory.getInstance();
    }
    mValue = mFactory.newDefaultBag();
}
 

public PigPerformanceLoader() {
    // Assume ^A as a delimiter
    super("
");
    bagFactory = BagFactory.getInstance();
    tupleFactory = TupleFactory.getInstance();
}