package de.juplo.kafka;
-import org.apache.kafka.clients.consumer.ConsumerRecord;
+import org.apache.kafka.clients.consumer.Consumer;
import org.apache.kafka.clients.consumer.KafkaConsumer;
import org.apache.kafka.common.serialization.StringDeserializer;
import org.springframework.boot.context.properties.EnableConfigurationProperties;
import java.util.Properties;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
-import java.util.function.Consumer;
@Configuration
public class ApplicationConfiguration
{
@Bean
- public EndlessConsumer endlessConsumer(
+ public WordcountRecordHandler wordcountRecordHandler(
+ PartitionStatisticsRepository repository,
+ Consumer<String, String> consumer,
+ ApplicationProperties properties)
+ {
+ return new WordcountRecordHandler(
+ repository,
+ properties.getClientId(),
+ properties.getTopic(),
+ Clock.systemDefaultZone(),
+ properties.getCommitInterval(),
+ consumer);
+ }
+
+ @Bean
+ public EndlessConsumer<String, String> endlessConsumer(
KafkaConsumer<String, String> kafkaConsumer,
ExecutorService executor,
- PartitionStatisticsRepository repository,
+ WordcountRecordHandler wordcountRecordHandler,
ApplicationProperties properties)
{
return
- new EndlessConsumer(
+ new EndlessConsumer<>(
executor,
- repository,
properties.getClientId(),
properties.getTopic(),
- Clock.systemDefaultZone(),
- properties.getCommitInterval(),
- kafkaConsumer);
+ kafkaConsumer,
+ wordcountRecordHandler);
}
@Bean
@RequiredArgsConstructor
public class ApplicationHealthIndicator implements HealthIndicator
{
- private final EndlessConsumer consumer;
+ private final EndlessConsumer<String, String> consumer;
@Override
public class DriverController
{
private final EndlessConsumer consumer;
+ private final WordcountRecordHandler wordcount;
@PostMapping("start")
@GetMapping("seen")
public Map<Integer, Map<String, Map<String, Long>>> seen()
{
- return consumer.getSeen();
+ return wordcount.getSeen();
}
@GetMapping("seen/{user}")
public ResponseEntity<Map<String, Long>> seen(@PathVariable String user)
{
- for (Map<String, Map<String, Long>> users : consumer.getSeen().values())
+ for (Map<String, Map<String, Long>> users : wordcount.getSeen().values())
{
Map<String, Long> words = users.get(user);
if (words != null)
import org.apache.kafka.common.errors.WakeupException;
import javax.annotation.PreDestroy;
-import java.time.Clock;
import java.time.Duration;
-import java.time.Instant;
import java.util.*;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.locks.Condition;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;
-import java.util.regex.Pattern;
@Slf4j
@RequiredArgsConstructor
-public class EndlessConsumer implements ConsumerRebalanceListener, Runnable
+public class EndlessConsumer<K, V> implements ConsumerRebalanceListener, Runnable
{
- final static Pattern PATTERN = Pattern.compile("\\W+");
-
-
private final ExecutorService executor;
- private final PartitionStatisticsRepository repository;
private final String id;
private final String topic;
- private final Clock clock;
- private final Duration commitInterval;
- private final Consumer<String, String> consumer;
+ private final Consumer<K, V> consumer;
+ private final RecordHandler<K, V> handler;
private final Lock lock = new ReentrantLock();
private final Condition condition = lock.newCondition();
private Exception exception;
private long consumed = 0;
- private final Map<Integer, Map<String, Map<String, Long>>> seen = new HashMap<>();
-
@Override
public void onPartitionsRevoked(Collection<TopicPartition> partitions)
{
- partitions.forEach(tp ->
- {
- Integer partition = tp.partition();
- Long newOffset = consumer.position(tp);
- log.info(
- "{} - removing partition: {}, offset of next message {})",
- id,
- partition,
- newOffset);
- Map<String, Map<String, Long>> removed = seen.remove(partition);
- repository.save(new StatisticsDocument(partition, removed, consumer.position(tp)));
- });
+ partitions.forEach(tp -> handler.onPartitionRevoked(tp));
}
@Override
public void onPartitionsAssigned(Collection<TopicPartition> partitions)
{
- partitions.forEach(tp ->
- {
- Integer partition = tp.partition();
- Long offset = consumer.position(tp);
- log.info("{} - adding partition: {}, offset={}", id, partition, offset);
- StatisticsDocument document =
- repository
- .findById(Integer.toString(partition))
- .orElse(new StatisticsDocument(partition));
- if (document.offset >= 0)
- {
- // Only seek, if a stored offset was found
- // Otherwise: Use initial offset, generated by Kafka
- consumer.seek(tp, document.offset);
- }
- seen.put(partition, document.statistics);
- });
+ partitions.forEach(tp -> handler.onPartitionAssigned(tp));
}
log.info("{} - Subscribing to topic {}", id, topic);
consumer.subscribe(Arrays.asList(topic), this);
- Instant lastCommit = clock.instant();
-
while (true)
{
- ConsumerRecords<String, String> records =
+ ConsumerRecords<K, V> records =
consumer.poll(Duration.ofSeconds(1));
// Do something with the data...
log.info("{} - Received {} messages", id, records.count());
- for (ConsumerRecord<String, String> record : records)
+ for (ConsumerRecord<K, V> record : records)
{
log.info(
"{} - {}: {}/{} - {}={}",
record.value()
);
- consumed++;
-
- Integer partition = record.partition();
- String user = record.key();
- Map<String, Map<String, Long>> users = seen.get(partition);
-
- Map<String, Long> words = users.get(user);
- if (words == null)
- {
- words = new HashMap<>();
- users.put(user, words);
- }
+ handler.accept(record);
- for (String word : PATTERN.split(record.value()))
- {
- Long num = words.get(word);
- if (num == null)
- {
- num = 1l;
- }
- else
- {
- num++;
- }
- words.put(word, num);
- }
+ consumed++;
}
- if (lastCommit.plus(commitInterval).isBefore(clock.instant()))
- {
- log.debug("Storing data and offsets, last commit: {}", lastCommit);
- seen.forEach((partiton, statistics) -> repository.save(
- new StatisticsDocument(
- partiton,
- statistics,
- consumer.position(new TopicPartition(topic, partiton)))));
- lastCommit = clock.instant();
- }
+ handler.beforeNextPoll();
}
}
catch(WakeupException e)
}
}
- public Map<Integer, Map<String, Map<String, Long>>> getSeen()
- {
- return seen;
- }
-
public void start()
{
lock.lock();
--- /dev/null
+package de.juplo.kafka;
+
+import org.apache.kafka.clients.consumer.ConsumerRecord;
+import org.apache.kafka.common.TopicPartition;
+
+import java.util.function.Consumer;
+
+
+public interface RecordHandler<K, V> extends Consumer<ConsumerRecord<K,V>>
+{
+ default void beforeNextPoll() {}
+
+ default void onPartitionAssigned(TopicPartition tp) {}
+
+ default void onPartitionRevoked(TopicPartition tp) {}
+}
--- /dev/null
+package de.juplo.kafka;
+
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.apache.kafka.clients.consumer.Consumer;
+import org.apache.kafka.clients.consumer.ConsumerRecord;
+import org.apache.kafka.common.TopicPartition;
+
+import java.time.Clock;
+import java.time.Duration;
+import java.time.Instant;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.regex.Pattern;
+
+
+@RequiredArgsConstructor
+@Slf4j
+public class WordcountRecordHandler implements RecordHandler<String, String>
+{
+ final static Pattern PATTERN = Pattern.compile("\\W+");
+
+
+ private final PartitionStatisticsRepository repository;
+ private final String id;
+ private final String topic;
+ private final Clock clock;
+ private final Duration commitInterval;
+ private final Consumer<String, String> consumer;
+
+ private final Map<Integer, Map<String, Map<String, Long>>> seen = new HashMap<>();
+
+ private Instant lastCommit = Instant.EPOCH;
+
+
+ @Override
+ public void accept(ConsumerRecord<String, String> record)
+ {
+ Integer partition = record.partition();
+ String user = record.key();
+ Map<String, Map<String, Long>> users = seen.get(partition);
+
+ Map<String, Long> words = users.get(user);
+ if (words == null)
+ {
+ words = new HashMap<>();
+ users.put(user, words);
+ }
+
+ for (String word : PATTERN.split(record.value()))
+ {
+ Long num = words.get(word);
+ if (num == null)
+ {
+ num = 1l;
+ }
+ else
+ {
+ num++;
+ }
+ words.put(word, num);
+ }
+ }
+
+
+ @Override
+ public void beforeNextPoll()
+ {
+ if (lastCommit.plus(commitInterval).isBefore(clock.instant()))
+ {
+ log.debug("Storing data and offsets, last commit: {}", lastCommit);
+ seen.forEach((partiton, statistics) -> repository.save(
+ new StatisticsDocument(
+ partiton,
+ statistics,
+ consumer.position(new TopicPartition(topic, partiton)))));
+ lastCommit = clock.instant();
+ }
+ }
+
+ @Override
+ public void onPartitionAssigned(TopicPartition tp)
+ {
+ Integer partition = tp.partition();
+ Long offset = consumer.position(tp);
+ log.info("{} - adding partition: {}, offset={}", id, partition, offset);
+ StatisticsDocument document =
+ repository
+ .findById(Integer.toString(partition))
+ .orElse(new StatisticsDocument(partition));
+ if (document.offset >= 0)
+ {
+ // Only seek, if a stored offset was found
+ // Otherwise: Use initial offset, generated by Kafka
+ consumer.seek(tp, document.offset);
+ }
+ seen.put(partition, document.statistics);
+ }
+
+ @Override
+ public void onPartitionRevoked(TopicPartition tp)
+ {
+ Integer partition = tp.partition();
+ Long newOffset = consumer.position(tp);
+ log.info(
+ "{} - removing partition: {}, offset of next message {})",
+ id,
+ partition,
+ newOffset);
+ Map<String, Map<String, Long>> removed = seen.remove(partition);
+ repository.save(new StatisticsDocument(partition, removed, consumer.position(tp)));
+ }
+
+
+ public Map<Integer, Map<String, Map<String, Long>>> getSeen()
+ {
+ return seen;
+ }
+}
import java.util.concurrent.ExecutionException;
import java.util.concurrent.ExecutorService;
import java.util.function.BiConsumer;
-import java.util.function.Consumer;
import java.util.function.Function;
import java.util.stream.Collectors;
import java.util.stream.IntStream;
ExecutorService executor;
@Autowired
PartitionStatisticsRepository repository;
+ @Autowired
+ WordcountRecordHandler wordcountRecordHandler;
- Consumer<ConsumerRecord<String, String>> testHandler;
- EndlessConsumer endlessConsumer;
+ EndlessConsumer<String, String> endlessConsumer;
Map<TopicPartition, Long> oldOffsets;
Map<TopicPartition, Long> newOffsets;
Set<ConsumerRecord<String, String>> receivedRecords;
@BeforeEach
public void init()
{
- testHandler = record -> {} ;
-
oldOffsets = new HashMap<>();
newOffsets = new HashMap<>();
receivedRecords = new HashSet<>();
newOffsets.put(tp, offset - 1);
});
- Consumer<ConsumerRecord<String, String>> captureOffsetAndExecuteTestHandler =
- record ->
- {
- newOffsets.put(
- new TopicPartition(record.topic(), record.partition()),
- record.offset());
- receivedRecords.add(record);
- testHandler.accept(record);
+ TestRecordHandler<String, String> captureOffsetAndExecuteTestHandler =
+ new TestRecordHandler<String, String>(wordcountRecordHandler) {
+ @Override
+ public void onNewRecord(ConsumerRecord<String, String> record)
+ {
+ newOffsets.put(
+ new TopicPartition(record.topic(), record.partition()),
+ record.offset());
+ receivedRecords.add(record);
+ }
};
endlessConsumer =
- new EndlessConsumer(
+ new EndlessConsumer<>(
executor,
- repository,
properties.getClientId(),
properties.getTopic(),
- Clock.systemDefaultZone(),
- properties.getCommitInterval(),
- kafkaConsumer);
+ kafkaConsumer,
+ captureOffsetAndExecuteTestHandler);
endlessConsumer.start();
}
--- /dev/null
+package de.juplo.kafka;
+
+import lombok.RequiredArgsConstructor;
+import org.apache.kafka.clients.consumer.ConsumerRecord;
+import org.apache.kafka.common.TopicPartition;
+
+
+@RequiredArgsConstructor
+public abstract class TestRecordHandler<K, V> implements RecordHandler<K, V>
+{
+ private final RecordHandler<K, V> handler;
+
+
+ public abstract void onNewRecord(ConsumerRecord<K, V> record);
+
+
+ @Override
+ public void accept(ConsumerRecord<K, V> record)
+ {
+ this.onNewRecord(record);
+ handler.accept(record);
+ }
+ @Override
+
+ public void beforeNextPoll()
+ {
+ handler.beforeNextPoll();
+ }
+
+ @Override
+ public void onPartitionAssigned(TopicPartition tp)
+ {
+ handler.onPartitionAssigned(tp);
+ }
+
+ @Override
+ public void onPartitionRevoked(TopicPartition tp)
+ {
+ handler.onPartitionRevoked(tp);
+ }
+}
projects / demos / kafka / training / commitdiff