TopologyTestDriver runs a complete Kafka Streams topology in memory, letting tests exercise operators and state stores without a live Kafka broker. Use it alongside focused unit tests for individual mappers, aggregators, filters, and punctuators. section 14.1.3 section 14.2
A typical test follows this sequence:
- Create the application and obtain the same Topology used by the running application. section 14.1.3
- Construct a TopologyTestDriver, usually with try-with-resources, then create TestInputTopic and TestOutputTopic instances for the topology's topics. section 14.1.3
- Pipe values, key-value pairs, TestRecord instances, or timestamped records into the input topic; read values, records, or key-value results from the output topic. section 14.1.3
- Assert both emitted results and, for stateful topologies, the contents of named state stores retrieved through getKeyValueStore or related accessors. section 14.1.4
Testing a stateful score reducer
Create input and output topics, pipe scores for Anna, Matthias, and Neil, and assert that three output records preserve input order. Retrieve the named KeyValueStore and verify each score. After more inputs, compare the final output map with the store's latest values. For windowed or punctuated processing, supply Instants or advance the input topic's event time so stream-time behavior occurs deterministically. section 14.1.4 section 14.1.4
Listing 14.11 Building the TopologyTestDriver and input and output topics
try (TopologyTestDriver driver =
new TopologyTestDriver(yellingTopology)) { #1
TestInputTopic<String, String> inputTopic = #2
driver.createInputTopic("src-topic",
stringSerializer,
stringSerializer);
TestOutputTopic<String, String> outputTopic = #3
driver.createOutputTopic("out-topic",
stringDeserializer,
stringDeserializer);
Inspect the compact setup showing driver construction and creation of serialized input and deserialized output topics. section 14.1.3
