@Test @Category(NeedsRunner.class) public void testGroupGlobally() { Collection<POJO> elements = ImmutableList.of( new POJO("key1", 1, "value1"), new POJO("key1", 1, "value2"), new POJO("key2", 2, "value3"), new POJO("key2", 2, "value4")); PCollection<Iterable<POJO>> grouped = pipeline.apply(Create.of(elements)).apply(Group.globally()); PAssert.that(grouped).satisfies(actual -> containsSingleIterable(elements, actual)); pipeline.run(); }
@Test @Category(NeedsRunner.class) public void testGlobalAggregation() { Collection<POJO> elements = ImmutableList.of( new POJO("key1", 1, "value1"), new POJO("key1", 1, "value2"), new POJO("key2", 2, "value3"), new POJO("key2", 2, "value4")); PCollection<Long> count = pipeline .apply(Create.of(elements)) .apply(Group.<POJO>globally().aggregate(Count.combineFn())); PAssert.that(count).containsInAnyOrder(4L); pipeline.run(); }
@Test @Category(NeedsRunner.class) public void testAggregateByMultipleFields() { Collection<AggregatePojos> elements = ImmutableList.of( new AggregatePojos(1, 1, 2), new AggregatePojos(2, 1, 3), new AggregatePojos(3, 2, 4), new AggregatePojos(4, 2, 5)); List<String> fieldNames = Lists.newArrayList("field1", "field2"); PCollection<Row> aggregate = pipeline .apply(Create.of(elements)) .apply( Group.<AggregatePojos>globally() .aggregateFields(fieldNames, new MultipleFieldCombineFn(), "field1+field2")); Schema outputSchema = Schema.builder().addInt64Field("field1+field2").build(); Row expectedRow = Row.withSchema(outputSchema).addValues(16L).build(); PAssert.that(aggregate).containsInAnyOrder(expectedRow); pipeline.run(); }
@Test @Category(NeedsRunner.class) public void testGloballyWithSchemaAggregateFn() { Collection<AggregatePojos> elements = ImmutableList.of( new AggregatePojos(1, 1, 2), new AggregatePojos(2, 1, 3), new AggregatePojos(3, 2, 4), new AggregatePojos(4, 2, 5)); PCollection<Row> aggregate = pipeline .apply(Create.of(elements)) .apply( Group.<AggregatePojos>globally() .aggregateField("field1", Sum.ofLongs(), "field1_sum") .aggregateField("field3", Sum.ofIntegers(), "field3_sum") .aggregateField("field1", Top.largestLongsFn(1), "field1_top")); Schema aggregateSchema = Schema.builder() .addInt64Field("field1_sum") .addInt32Field("field3_sum") .addArrayField("field1_top", FieldType.INT64) .build(); Row expectedRow = Row.withSchema(aggregateSchema).addValues(10L, 14).addArray(4L).build(); PAssert.that(aggregate).containsInAnyOrder(expectedRow); pipeline.run(); }
.apply(Create.of(elements)) .apply( Group.<OuterAggregate>globally() .aggregateFields(field1Selector, Sum.ofLongs(), "field1_sum") .aggregateFields(field3Selector, Sum.ofIntegers(), "field3_sum")