@Test
public void testMoreThanEightProps() {
List<Field> fields = new ArrayList<Field>(30);
for (int i = 0; i < 30; i++) {
fields.add(Field.create("field" + i, Type.STRING));
}
Schema schema = new Schema("bigmammaschema", fields);
ITuple tuple = new Tuple(schema);
for (int i = 0; i < 30; i++) {
// Fill in with some default values
tuple.set("field" + i, "defaultValue" + i);
}
NullableTuple nullableTuple = new NullableTuple(tuple);
for (int i = 0; i < 30; i++) {
// Assert the default values
assertEquals("defaultValue" + i, nullableTuple.get("field" + i));
}
// Set fields to null one by one and assert that things go well
for (int i = 0; i < 30; i++) {
nullableTuple.set("field" + i, null);
assertEquals(null, nullableTuple.getNullable("field" + i));
nullableTuple.set("field" + i, "defaultValue" + i);
assertEquals("defaultValue" + i, nullableTuple.getNullable("field" + i));
}
}
protected static void fillObject(boolean isRandom, ITuple tuple, Field field, int index) {
Object instance = ReflectionUtils.newInstance(field.getObjectClass(), null);
if (instance instanceof A) {
A a = (A) instance;
a.setId(isRandom ? random.nextInt() + "" : "");
a.setUrl(isRandom ? random.nextLong() + "" : "");
}
tuple.set(index, instance);
}
protected static void fillString(boolean isRandom, ITuple tuple, int index) {
if (isRandom) {
switch (random.nextInt(4)) {
case 0:
tuple.set(index, "");
break;
case 1:
tuple.set(index, random.nextLong() + "");
break;
case 2:
tuple.set(index, new Utf8(random.nextLong() + ""));
break;
case 3:
tuple.set(index, new Text(random.nextLong() + ""));
break;
}
} else {
tuple.set(index, "");
}
}
@Test
public void testNoNullableWrapper() {
Schema schema =
new Schema(
"testSchema", Fields.parse("a:string, b:int, c:double, d:float, e:boolean, f:long"));
ITuple tuple = new Tuple(schema);
tuple.set(0, "foo");
tuple.set(1, 10);
tuple.set(2, 20d);
tuple.set(3, 30f);
tuple.set(4, false);
tuple.set(5, 40l);
NullableTuple nullableTuple = new NullableTuple(tuple);
assertEquals("foo", nullableTuple.get(0).toString());
assertEquals(10, nullableTuple.get(1));
assertEquals(20d, nullableTuple.get(2));
assertEquals(30f, nullableTuple.get(3));
assertEquals(false, nullableTuple.get(4));
assertEquals(40l, nullableTuple.get(5));
}
@Override
public void reduce(
ITuple group, Iterable<ITuple> tuples, TupleMRContext context, Collector collector)
throws IOException, InterruptedException, TupleMRException {
int count = 0;
ITuple outputTuple = null;
for (ITuple tuple : tuples) {
outputTuple = tuple;
count += (Integer) tuple.get(1);
}
outputTuple.set(1, count);
collector.write(outputTuple, NullWritable.get());
}
/** Fills the fields specified by the range (minIndex, maxIndex) with random data. */
protected static void fillTuple(boolean isRandom, ITuple tuple, int minIndex, int maxIndex) {
try {
for (int i = minIndex; i <= maxIndex; i++) {
Field field = tuple.getSchema().getField(i);
switch (field.getType()) {
case INT:
tuple.set(i, isRandom ? random.nextInt() : 0);
break;
case LONG:
tuple.set(i, isRandom ? random.nextLong() : 0);
break;
case BOOLEAN:
tuple.set(i, isRandom ? random.nextBoolean() : false);
break;
case DOUBLE:
tuple.set(i, isRandom ? random.nextDouble() : 0.0);
break;
case FLOAT:
tuple.set(i, isRandom ? random.nextFloat() : 0f);
break;
case STRING:
fillString(isRandom, tuple, i);
break;
case ENUM:
fillEnum(isRandom, field, tuple, i);
break;
case OBJECT:
fillObject(isRandom, tuple, field, i);
break;
default:
throw new IllegalArgumentException("Not supported type " + field.getType());
}
}
} catch (Exception e) {
throw new RuntimeException(e);
}
}
protected static void fillEnum(boolean isRandom, Field field, ITuple tuple, int index)
throws Exception {
Method method = field.getObjectClass().getMethod("values", (Class[]) null);
Enum[] values = (Enum[]) method.invoke(null);
tuple.set(index, values[isRandom ? random.nextInt(values.length) : 0]);
}
public void createInput(String input, Configuration conf)
throws IOException, InterruptedException {
Path inPath = new Path(input);
FileSystem fs = FileSystem.get(inPath.toUri(), conf);
TupleFile.Writer writer = new TupleFile.Writer(fs, conf, inPath, TopicalWordCount.getSchema());
// Topic 1, words: { a, 10 } { b, 1 } , { c, 5 }
// Top 2 words = a(10), c(5)
ITuple tuple = new Tuple(TopicalWordCount.getSchema());
tuple.set("word", "a");
tuple.set("topic", 1);
tuple.set("count", 10);
writer.append(tuple);
tuple.set("word", "b");
tuple.set("topic", 1);
tuple.set("count", 1);
writer.append(tuple);
tuple.set("word", "c");
tuple.set("topic", 1);
tuple.set("count", 5);
writer.append(tuple);
// Topic 2, words: { a, 10 } { b, 9 } , { c, 5 }
// Top 2 words = a(10), b(9)
tuple.set("word", "a");
tuple.set("topic", 2);
tuple.set("count", 10);
writer.append(tuple);
tuple.set("word", "b");
tuple.set("topic", 2);
tuple.set("count", 9);
writer.append(tuple);
tuple.set("word", "c");
tuple.set("topic", 2);
tuple.set("count", 5);
writer.append(tuple);
writer.close();
}
