Status
Current state: Draft
Discussion thread:
JIRA:
Please keep the discussion on the mailing list rather than commenting on the wiki (wiki discussions get unwieldy fast).
Motivation
THIS IS STILL WORK IN PROGRESS.
This KIP is inspired by the Discussion in KIP-456: Helper classes to make it simpler to write test logic with TopologyTestDriver
The proposal in KIP-456 was to add alternate way to input and output topic, but this KIP enhance those classes and deprecate old functionality to make clear interface for test writer to use.
When using the old TopologyTestDriver you need to call ConsumerRecordFactory to create ConsumerRecord passed into pipeInput method to write to topic. Also when calling readOutput to consume from topic, you need to provide correct Deserializers each time.
You easily end up writing helper methods in your test classes, but this can be avoided when adding generic input and output topic classes to implement the needed functionality.
Also the logic of the old TopologyTestDriver is confusing, when you need to pipe ConsumerRecords to produce record to input topic and receive ProducerRecords when consuming from output topic.
Non-existing topic and no record in the queue scenarious are modified to throw Exception instead of returning null.
Public Interfaces
package org.apache.kafka.clients; //New Interface to make possible to use ProducereRecord in TopologyTestDriver //ConsumerRecord not modified to implement this because timestamp() return long, not Long public interface ClientRecord<K, V> { String topic(); Headers headers(); K key(); V value(); Long timestamp(); }
package org.apache.kafka.clients.producer; public class ProducerRecord<K, V> implements ClientRecord<K, V> { ...
package org.apache.kafka.streams; public class TopologyTestDriver { ... @Deprecate public void pipeInput(ConsumerRecord<byte[], byte[]> record); // can trigger event-time punctuation @Deprecate public void pipeInput(List<ConsumerRecord<byte[], byte[]>> records); // can trigger event-time punctuation // methods for result verification @Deprecate public ProducerRecord<byte[], byte[]> readOutput(String topic); @Deprecate public <K, V> ProducerRecord<K, V> readOutput(String topic, Deserializer<K> keyDeserializer, Deserializer<V> valueDeserializer); ... }
package org.apache.kafka.streams.test; public class TopologyTestDriver { public TopologyTestDriver(Topology topology, Properties config); // initialized WallClockMockTimeMs with System.currentTimeMillis() public TopologyTestDriver(Topology topology, Properties config, long initialWallClockTimeMs); // event-time is automatically advanced based on the provided input records, and thus event-time punctuation are triggered automatically // wall-clock time is mocked and not advanced automatically; user can advance wall-clock time manually, and thus, trigger wall-clock time punctuations manually public void advanceWallClockTime(long advanceMs); // can trigger wall-clock-time punctuation // methods for TestTopic object creation //TODO investigate, is there way to find out serde based on topology public final <K, V> TestInputTopic<K, V> createInputTopic(final String topicName, final Serde<K> keySerde, final Serde<V> valueSerde); public final <K, V> TestOutputTopic<K, V> createOutputTopic(final String topicName, final Serde<K> keySerde, final Serde<V> valueSerde); public Map<String, StateStore> getAllStateStores() public StateStore getStateStore(String name); public <K, V> KeyValueStore<K, V> getKeyValueStore(String name); public <K, V> WindowStore<K, V> getWindowStore(String name); public <K, V> SessionStore<K, V> getSessionStore(String name); public void close(); }
package org.apache.kafka.streams.test; public class TestInputTopic<K, V> { //Create by TopologyTestDriver, Constructors are package private //Timestamp handling //Record timestamp can be provided when piping input or use internally tracked time initialized configured with configureTiming: //startTimestampMs the initial timestamp for generated records, if not provided uses current system time as start timestamp. //autoAdvanceMs the time increment per generated record, if not provided auto-advance is disabled. void configureTiming(final long startTimestampMs, final long autoAdvanceMs); //Advances the internally tracked time. void advanceTimeMs(final long advanceMs); //Methods to pipe single record void pipeInput(final V value); void pipeInput(final K key, final V value); void pipeInput(final V value, final long timestampMs); void pipeInput(final K key, final V value, final long timestampMs); void pipeInput(final K key, final V value, final Headers headers); void pipeInput(final K key, final V value, final Headers headers, final long timestampMs); //Methods to pipe list of records void pipeKeyValueList(final List<KeyValue<K, V>> keyValues); void pipeValueList(final List<V> values); void pipeKeyValueList(final List<KeyValue<K, V>> keyValues, final long startTimestamp, final long advanceMs); void pipeValueList(final List<V> values, final long startTimestamp, final long advanceMs); }
package org.apache.kafka.streams.test; public class TestOutputTopic<K, V> { //Create by TopologyTestDriver, Constructors are package private //Methods to readOutput ClientRecord<K, V> readRecord(); KeyValue<K, V> readKeyValue(); V readValue(); //Output as collection Map<K, V> readKeyValuesToMap(); List<KeyValue<K, V>> readKeyValuesToList(); List<V> readValuesToList(); }
Proposed Changes
This improvement adds TestInputTopic class which wraps TopologyTestDriver and ConsumerRecordFactory methods as one class to be used to write to Input Topics and TestOutputTopic class which collects TopologyTestDriver reading methods and provide typesafe read methods.
public class SimpleTopicTest { private TopologyTestDriver testDriver; private TestInputTopic<String, String> inputTopic; private TestOutputTopic<String, String> outputTopic; @Before public void setup() { testDriver = new TopologyTestDriver(TestStream.getTopology(), TestStream.getConfig()); inputTopic = new TestInputTopic<>(testDriver, TestStream.INPUT_TOPIC, new Serdes.StringSerde(), new Serdes.StringSerde()); outputTopic = new TestOutputTopic<>(testDriver, TestStream.OUTPUT_TOPIC, new Serdes.StringSerde(), new Serdes.StringSerde()); } @After public void tearDown() { testDriver.close(); } @Test public void testOneWord() { //Feed word "Hello" to inputTopic and no kafka key, timestamp is irrelevant in this case inputTopic.pipeInput("Hello"); assertThat(outputTopic.readValue()).isEqualTo("Hello"); //No more output in topic assertThat(outputTopic.readRecord()).isNull(); } }
- Examples in Testing Kafka Streams https://kafka.apache.org/22/documentation/streams/developer-guide/testing.html updated to use TestInputTopic and TestOutputTopic
Compatibility, Deprecation, and Migration Plan
This is only adding new classes. The tests utilizing directly TopologyTestDriver can still be used. There are no compatibility issues.
Rejected Alternatives
- It was considered to add methods to to return Iterable like in KIP-451: Make TopologyTestDriver output iterable , but it seems to be redundant with these List based methods