zhipeng93 commented on code in PR #230:
URL: https://github.com/apache/flink-ml/pull/230#discussion_r1163672488
##########
flink-ml-core/src/test/java/org/apache/flink/ml/common/datastream/DataStreamUtilsTest.java:
##########
@@ -59,6 +64,54 @@ public void testMapPartition() throws Exception {
new int[] {5, 5, 5, 5},
counts.stream().mapToInt(Integer::intValue).toArray());
}
+ @Test
+ public void testCoGroup() throws Exception {
+ DataStream<Tuple2<Integer, Integer>> data1 =
+ env.fromCollection(
+ Arrays.asList(Tuple2.of(1, 1), Tuple2.of(2, 2),
Tuple2.of(3, 3)));
+ DataStream<Tuple2<Integer, Double>> data2 =
+ env.fromCollection(
+ Arrays.asList(
+ Tuple2.of(1, 1.5),
+ Tuple2.of(5, 5.5),
+ Tuple2.of(3, 3.5),
+ Tuple2.of(1, 2.5)));
+ DataStream<Double> result =
+ DataStreamUtils.coGroup(
+ data1,
+ data2,
+ (KeySelector<Tuple2<Integer, Integer>, Integer>) tuple
-> tuple.f0,
+ (KeySelector<Tuple2<Integer, Double>, Integer>) tuple
-> tuple.f0,
+ BasicTypeInfo.DOUBLE_TYPE_INFO,
+ new CoGroupFunction<
+ Tuple2<Integer, Integer>, Tuple2<Integer,
Double>, Double>() {
+ @Override
+ public void coGroup(
+ Iterable<Tuple2<Integer, Integer>>
iterableA,
+ Iterable<Tuple2<Integer, Double>>
iterableB,
+ Collector<Double> collector) {
+ List<Tuple2<Integer, Integer>> valuesA =
+
IteratorUtils.toList(iterableA.iterator());
+ List<Tuple2<Integer, Double>> valuesB =
+
IteratorUtils.toList(iterableB.iterator());
+
+ double sum = 0;
+ for (Tuple2<Integer, Integer> value : valuesA)
{
+ sum += value.f1;
+ }
+ for (Tuple2<Integer, Double> value : valuesB) {
+ sum += value.f1;
+ }
+ collector.collect(sum);
+ }
+ });
+
+ List<Double> resultValues =
IteratorUtils.toList(result.executeAndCollect());
+ double[] resultPrimitiveValues =
+
resultValues.stream().mapToDouble(Double::doubleValue).toArray();
+ assertArrayEquals(new double[] {5.0, 2.0, 6.5, 5.5},
resultPrimitiveValues, 1e-5);
Review Comment:
Thanks for the explanation.
BTW, do you think we need a unit test with paralellism not one?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
