Testing Your Pipeline

The documentation on this page applies only to the Dataflow SDK 1.x for Java.

The Dataflow SDK 2.x for Java and the Dataflow SDK for Python are based on Apache Beam. See the documentation for those SDKs.

Testing your pipeline is a particularly important step in developing an effective data processing solution. The indirect nature of Cloud Dataflow's model, in which your user code constructs a pipeline graph to be executed remotely on Google Cloud Platform, can make debugging failed runs a non-trivial task. Often it is faster and simpler to perform local unit testing on your pipeline code than to debug a pipeline's remote execution.

Unit testing your pipeline code locally, before performing full runs with the Cloud Dataflow service, is often the best and most direct way to identify and fix bugs in your pipeline code. Unit testing your pipeline locally also allows you to use your familiar/favorite local debugging tools.

Note: When testing your code with the Dataflow service, consider limiting the number of worker instances for your pipeline to 1, or the minimum number appropriate for your test. Limiting the number of worker instances used during repeated test runs can provide significant time and cost savings.

You can limit the number of workers your pipeline uses during test runs by setting the --maxNumWorkers execution option when you run your test pipeline.

The Dataflow SDKs provide a number of ways to unit test your pipeline code, from the lowest to the highest levels. From the lowest to the highest level, these are:

• You can test the individual function objects, such as DoFns, inside your pipeline's core transforms.
• You can test an entire Composite Transform as a unit.
• You can perform an end-to-end test for an entire pipeline.

Java

To support unit testing, the Dataflow SDK for Java provides a number of test classes in the package com.google.cloud.dataflow.sdk.testing. In addition, the transforms included in the SDK have unit tests, and the example programs in the SDK also contain tests. You can use these tests as references and guides.

Testing Individual DoFn Objects

The code in your pipeline's DoFn functions runs often, and often across multiple Compute Engine instances. Unit-testing your DoFn objects before using them in a Dataflow run can save a great deal of debugging time and energy.

Java

The Dataflow SDK for Java provides a convenient way to test an individual DoFn called DoFnTester, which is included in the SDK Transforms package.

DoFnTesteruses the JUnit framework. To use DoFnTester, you'll need to do the following:

1. Create a DoFnTester. You'll need to pass an instance of the DoFn you want to test to the static factory method for DoFnTester.
2. Create one or more main test inputs of the appropriate type for your DoFn. If your DoFn takes side inputs and/or produces side outputs, you should also create the side inputs and the side output tags.
3. Call DoFnTester.processBatch to process the main inputs.
4. Use JUnit's Assert.assertThat method to ensure the test outputs returned from processBatch match your expected values.

Creating a DoFnTester

To create a DoFnTester, first create an instance of the DoFn you want to test. You then use that instance when you create a DoFnTester using the .of() static factory method:

  static class MyDoFn extends DoFn<String, Integer> { ... }
  MyDoFn myDoFn = ...;

  DoFnTester<String, Integer> fnTester = DoFnTester.of(myDoFn);

Creating Test Inputs

You'll need to create one or more test inputs for DoFnTester to send to your DoFn. To create test inputs, simply create one or more input variables of the same input type that your DoFn accepts. In the case above:

  static class MyDoFn extends DoFn<String, Integer> { ... }
  MyDoFn myDoFn = ...;
  DoFnTester<String, Integer> fnTester = DoFnTester.of(myDoFn);

  String testInput = "test1";

Side Inputs and Outputs

If your DoFn accepts side inputs, you can create those side inputs by using the method DoFnTester.setSideInputs.

  static class MyDoFn extends DoFn<String, Integer> { ... }
  MyDoFn myDoFn = ...;
  DoFnTester<String, Integer> fnTester = DoFnTester.of(myDoFn);

  PCollectionView<List<Integer>> sideInput = ...;
  Iterable<Integer> value = ...;
  fnTester.setSideInputInGlobalWindow(sideInput, value);

If your DoFn produces side outputs, you'll need to set the appropriate TupleTag objects that you'll use to access each output. A DoFn with side outputs produces a PCollectionTuple for each side output; you'll need to provide a TupleTagList that corresponds to each side output in that tuple.

Suppose your DoFn produces side outputs of type String and Integer. You create TupleTag objects for each, and bundle them into a TupleTagList, then set it for the DoFnTester as follows:

  static class MyDoFn extends DoFn<String, Integer> { ... }
  MyDoFn myDoFn = ...;
  DoFnTester<String, Integer> fnTester = DoFnTester.of(myDoFn);

  TupleTag<String> tag1 = ...;
  TupleTag<Integer> tag2 = ...;
  TupleTagList tags = TupleTagList.of(tag1).and(tag2);

  fnTester.setSideOutputTags(tags);

See the ParDo documentation on side inputs for more information.

Processing Test Inputs and Checking Results

To process the inputs (and thus run the test on your DoFn), you call the method DoFnTester.processBatch. When you call processBatch, you pass one or more main test input values for your DoFn. If you set side inputs, the side inputs are available to each batch of main inputs that you provide.

DoFnTester.processBatch returns a List of outputs—that is, objects of the same type as the DoFn's specified output type. For a DoFn<String, Integer>, processBatch returns a List<Integer>:

  static class MyDoFn extends DoFn<String, Integer> { ... }
  MyDoFn myDoFn = ...;
  DoFnTester<String, Integer> fnTester = DoFnTester.of(myDoFn);

  String testInput = "test1";
  List<Integer> testOutputs = fnTester.processBatch(testInput);

To check the results of processBatch, you use JUnit's Assert.assertThat method to test if the List of outputs contains the values you expect:

  String testInput = "test1";
  List<Integer> testOutputs = fnTester.processBatch(testInput);

  Assert.assertThat(testOutputs, Matchers.hasItems(...));

  // Process a larger batch in a single step.
  Assert.assertThat(fnTester.processBatch("input1", "input2", "input3"), Matchers.hasItems(...));

Testing Composite Transforms

To test a composite transform you've created, you can use the following pattern:

• Create a TestPipeline.
• Create some static, known test input data.
• Use the Create transform to create a PCollection of your input data.
• Apply your composite transform to the input PCollection and save the resulting output PCollection.
• Use DataflowAssert and its subclasses to verify that the output PCollection contains the elements that you expect.

Java

Using the SDK Test Classes

TestPipeline and DataflowAssert are classes included in the Cloud Dataflow Java SDK specifically for testing transforms. TestPipeline and DataflowAssert work with tests configured to run both locally or against the remote Cloud Dataflow service.

TestPipeline

For tests, use TestPipeline in place of Pipeline when you create the pipeline object. Unlike Pipeline.create, TestPipeline.create handles setting PipelineOptions interally.

You create a TestPipeline as follows:

  Pipeline p = TestPipeline.create();

DataflowAssert

DataflowAssert is an assertion on the contents of a PCollection. You can use DataflowAssert to verify that a PCollection contains a specific set of expected elements.

For a given PCollection, you can use to DataflowAssert to verify the contents as follows:

  PCollection<String> output = ...;

  // Check whether a PCollection contains some elements in any order.
  DataflowAssert.that(output)
    .containsInAnyOrder(
      "elem1",
      "elem3",
      "elem2");

Any code that uses DataflowAssert must link in JUnit and Hamcrest. If you're using Maven, you can link in Hamcrest by adding the following dependency to your project's pom.xml file:

  <dependency>
    <groupId>org.hamcrest</groupId>
    <artifactId>hamcrest-all</artifactId>
    <version>1.3</version>
    <scope>test</scope>
  </dependency>

For more information on how these classes work, see the com.google.cloud.dataflow.sdk.testing package documentation.

Using the Create Transform

You can use the Create transform to create a PCollection out of a standard in-memory collection class, such as Java List. See Creating a PCollection for more information.

An Example Test for a Composite Transform

Java

The following code shows a complete test for a composite transform. The test applies the Count transform to an input PCollection of String elements. The test uses the Create transform to create the input PCollection from a Java List<String>.

  @RunWith(JUnit4.class)
  public class CountTest {

  // Our static input data, which will make up the initial PCollection.
  static final String[] WORDS_ARRAY = new String[] {
    "hi", "there", "hi", "hi", "sue", "bob",
    "hi", "sue", "", "", "ZOW", "bob", ""};

  static final List<String> WORDS = Arrays.asList(WORDS_ARRAY);

    @Test
    public void testCount() {
      // Create a test pipeline.
      Pipeline p = TestPipeline.create();

      // Create an input PCollection.
      PCollection<String> input = p.apply(Create.of(WORDS)).setCoder(StringUtf8Coder.of());

      // Apply the Count transform under test.
      PCollection<KV<String, Long>> output =
        input.apply(Count.<String>perElement());

      // Assert on the results.
      DataflowAssert.that(output)
        .containsInAnyOrder(
            KV.of("hi", 4L),
            KV.of("there", 1L),
            KV.of("sue", 2L),
            KV.of("bob", 2L),
            KV.of("", 3L),
            KV.of("ZOW", 1L));

      // Run the pipeline.
      p.run();
  }

Testing a Pipeline End-to-End

You can use the test classes in the Dataflow SDKs (such as TestPipeline and DataflowAssert in the Dataflow SDK for Java) to test an entire pipeline end-to-end. Typically, to test an entire pipeline, you do the following:

• For every source of input data to your pipeline, create some known static test input data.
• Create some static test output data that matches what you expect in your pipeline's final output PCollection(s).
• Create a TestPipeline in place of the standard Pipeline.create.
• In place of your pipeline's Read transform(s), use the Create transform to create one or more PCollections from your static input data.
• Apply your pipeline's transforms.
• In place of your pipeline's Write transform(s), use DataflowAssert to verify that the contents of the final PCollections your pipeline produces match the expected values in your static output data.

Testing the WordCount Pipeline

Java

The following example code shows how one might test the WordCount example pipeline. WordCount usually reads lines from a text file for input data; instead, the test creates a Java List<String> containing some text lines and uses a Create transform to create an initial PCollection.

WordCount's final transform (from the composite transform CountWords) produces a PCollection<String> of formatted word counts suitable for printing. Rather than write that PCollection to an output text file, our test pipeline uses DataflowAssert to verify that the elements of the PCollection match those of a static String array containing our expected output data.

  @RunWith(JUnit4.class)
  public class WordCountTest {

    // Our static input data, which will comprise the initial PCollection.
    static final String[] WORDS_ARRAY = new String[] {
      "hi there", "hi", "hi sue bob",
      "hi sue", "", "bob hi"};

    static final List<String> WORDS = Arrays.asList(WORDS_ARRAY);

    // Our static output data, which is the expected data that the final PCollection must match.
    static final String[] COUNTS_ARRAY = new String[] {
        "hi: 5", "there: 1", "sue: 2", "bob: 2"};

    // Example test that tests the pipeline's transforms.
    @Test
    @Category(com.google.cloud.dataflow.sdk.testing.RunnableOnService.class)
    public void testCountWords() throws Exception {
      Pipeline p = TestPipeline.create();

      // Create a PCollection from the WORDS static input data.
      PCollection<String> input = p.apply(Create.of(WORDS)).setCoder(StringUtf8Coder.of());

      // Run ALL the pipeline's transforms (in this case, the CountWords composite transform).
      PCollection<String> output = input.apply(new CountWords());

      // Assert that the output PCollection matches the COUNTS_ARRAY known static output data.
      DataflowAssert.that(output).containsInAnyOrder(COUNTS_ARRAY);

      // Run the pipeline.
      p.run();
    }
  }