Search in sources :

Example 1 with OperationProcessor

use of com.nextdoor.bender.operation.OperationProcessor in project bender by Nextdoor.

the class BaseHandler method processInternal.

/**
 * Method called by Handler implementations to process records.
 *
 * @param context Lambda invocation context.
 * @throws HandlerException
 */
private void processInternal(Context context) throws HandlerException {
    Stat runtime = new Stat("runtime.ns");
    runtime.start();
    Source source = this.getSource();
    DeserializerProcessor deser = source.getDeserProcessor();
    List<OperationProcessor> operations = source.getOperationProcessors();
    List<String> containsStrings = source.getContainsStrings();
    List<Pattern> regexPatterns = source.getRegexPatterns();
    this.getIpcService().setContext(context);
    Iterator<InternalEvent> events = this.getInternalEventIterator();
    /*
     * For logging purposes log when the function started running
     */
    this.monitor.invokeTimeNow();
    AtomicLong eventCount = new AtomicLong(0);
    AtomicLong oldestArrivalTime = new AtomicLong(System.currentTimeMillis());
    AtomicLong oldestOccurrenceTime = new AtomicLong(System.currentTimeMillis());
    /*
     * Process each record
     */
    int characteristics = Spliterator.IMMUTABLE;
    Spliterator<InternalEvent> spliterator = Spliterators.spliteratorUnknownSize(events, characteristics);
    Stream<InternalEvent> input = StreamSupport.stream(spliterator, false);
    /*
     * Filter out raw events
     */
    Stream<InternalEvent> filtered = input.filter(/*
         * Perform regex filter
         */
    ievent -> {
        eventCount.incrementAndGet();
        String eventStr = ievent.getEventString();
        /*
           * Apply String contains filters before deserialization
           */
        for (String containsString : containsStrings) {
            if (eventStr.contains(containsString)) {
                return false;
            }
        }
        /*
           * Apply regex patterns before deserialization
           */
        for (Pattern regexPattern : regexPatterns) {
            Matcher m = regexPattern.matcher(eventStr);
            if (m.find()) {
                return false;
            }
        }
        return true;
    });
    /*
     * Deserialize
     */
    Stream<InternalEvent> deserialized = filtered.map(ievent -> {
        DeserializedEvent data = deser.deserialize(ievent.getEventString());
        if (data == null || data.getPayload() == null) {
            logger.warn("Failed to deserialize: " + ievent.getEventString());
            return null;
        }
        ievent.setEventObj(data);
        return ievent;
    }).filter(Objects::nonNull);
    /*
     * Perform Operations
     */
    Stream<InternalEvent> operated = deserialized;
    for (OperationProcessor operation : operations) {
        operated = operation.perform(operated);
    }
    /*
     * Serialize
     */
    Stream<InternalEvent> serialized = operated.map(ievent -> {
        try {
            String raw = null;
            raw = this.ser.serialize(this.wrapper.getWrapped(ievent));
            ievent.setSerialized(raw);
            return ievent;
        } catch (SerializationException e) {
            return null;
        }
    }).filter(Objects::nonNull);
    /*
     * Transport
     */
    serialized.forEach(ievent -> {
        /*
       * Update times
       */
        updateOldest(oldestArrivalTime, ievent.getArrivalTime());
        updateOldest(oldestOccurrenceTime, ievent.getEventTime());
        try {
            this.getIpcService().add(ievent);
        } catch (TransportException e) {
            logger.warn("error adding event", e);
        }
    });
    /*
     * Wait for transporters to finish
     */
    try {
        this.getIpcService().shutdown();
    } catch (TransportException e) {
        throw new HandlerException("encounted TransportException while shutting down ipcService", e);
    } catch (InterruptedException e) {
        throw new HandlerException("thread was interruptedwhile shutting down ipcService", e);
    } finally {
        String evtSource = this.getSourceName();
        runtime.stop();
        if (!this.skipWriteStats) {
            writeStats(eventCount.get(), oldestArrivalTime.get(), oldestOccurrenceTime.get(), evtSource, runtime);
        }
        if (logger.isTraceEnabled()) {
            getGCStats();
        }
    }
}
Also used : Monitor(com.nextdoor.bender.monitoring.Monitor) Spliterators(java.util.Spliterators) Wrapper(com.nextdoor.bender.wrapper.Wrapper) Context(com.amazonaws.services.lambda.runtime.Context) Stat(com.nextdoor.bender.monitoring.Stat) InternalEvent(com.nextdoor.bender.InternalEvent) OperationProcessor(com.nextdoor.bender.operation.OperationProcessor) ArrayList(java.util.ArrayList) IpcSenderService(com.nextdoor.bender.ipc.IpcSenderService) Logger(org.apache.log4j.Logger) Matcher(java.util.regex.Matcher) GarbageCollectorMXBean(java.lang.management.GarbageCollectorMXBean) AmazonS3ClientFactory(com.nextdoor.bender.aws.AmazonS3ClientFactory) TransportException(com.nextdoor.bender.ipc.TransportException) BenderConfig(com.nextdoor.bender.config.BenderConfig) StreamSupport(java.util.stream.StreamSupport) ManagementFactory(java.lang.management.ManagementFactory) DeserializedEvent(com.nextdoor.bender.deserializer.DeserializedEvent) Iterator(java.util.Iterator) IOException(java.io.IOException) SerializerProcessor(com.nextdoor.bender.serializer.SerializerProcessor) ConfigurationException(com.nextdoor.bender.config.ConfigurationException) Objects(java.util.Objects) AtomicLong(java.util.concurrent.atomic.AtomicLong) List(java.util.List) Stream(java.util.stream.Stream) SerializationException(com.nextdoor.bender.serializer.SerializationException) BenderLayout(com.nextdoor.bender.logging.BenderLayout) Pattern(java.util.regex.Pattern) Source(com.nextdoor.bender.config.Source) Spliterator(java.util.Spliterator) DeserializerProcessor(com.nextdoor.bender.deserializer.DeserializerProcessor) AmazonS3URI(com.amazonaws.services.s3.AmazonS3URI) HandlerResources(com.nextdoor.bender.config.HandlerResources) Pattern(java.util.regex.Pattern) DeserializedEvent(com.nextdoor.bender.deserializer.DeserializedEvent) SerializationException(com.nextdoor.bender.serializer.SerializationException) Matcher(java.util.regex.Matcher) OperationProcessor(com.nextdoor.bender.operation.OperationProcessor) TransportException(com.nextdoor.bender.ipc.TransportException) Source(com.nextdoor.bender.config.Source) InternalEvent(com.nextdoor.bender.InternalEvent) AtomicLong(java.util.concurrent.atomic.AtomicLong) Stat(com.nextdoor.bender.monitoring.Stat) Objects(java.util.Objects) DeserializerProcessor(com.nextdoor.bender.deserializer.DeserializerProcessor)

Example 2 with OperationProcessor

use of com.nextdoor.bender.operation.OperationProcessor in project bender by Nextdoor.

the class BaseHandlerTest method testOperationException.

@Test
public void testOperationException() throws HandlerException {
    BaseHandler.CONFIG_FILE = "/config/handler_config.json";
    handler.skipWriteStats = true;
    List<DummyEvent> events = new ArrayList<DummyEvent>(1);
    events.add(new DummyEvent("foo", 0));
    TestContext context = new TestContext();
    context.setInvokedFunctionArn("arn:aws:lambda:us-east-1:123:function:test:tag");
    handler.init(context);
    List<OperationProcessor> operationProcessors = handler.sources.get(0).getOperationProcessors();
    for (OperationProcessor operationProcessor : operationProcessors) {
        BaseOperation operation = spy(operationProcessor.getOperation());
        doThrow(new OperationException("expected")).when(operation).perform(any());
        operationProcessor.setOperation(operation);
    }
    handler.handler(events, context);
    assertEquals(1, operationProcessors.get(0).getErrorCountStat().getValue());
}
Also used : TestContext(com.nextdoor.bender.aws.TestContext) ArrayList(java.util.ArrayList) BaseOperation(com.nextdoor.bender.operation.BaseOperation) OperationProcessor(com.nextdoor.bender.operation.OperationProcessor) OperationException(com.nextdoor.bender.operation.OperationException) Test(org.junit.Test)

Example 3 with OperationProcessor

use of com.nextdoor.bender.operation.OperationProcessor in project bender by Nextdoor.

the class BaseHandlerTest method testMultipleOperationsConfig.

@Test
public void testMultipleOperationsConfig() throws HandlerException {
    BaseHandler.CONFIG_FILE = "/config/handler_config_two_operations.json";
    List<DummyEvent> events = new ArrayList<DummyEvent>(1);
    events.add(new DummyEvent("foo", 0));
    TestContext context = new TestContext();
    context.setInvokedFunctionArn("arn:aws:lambda:us-east-1:123:function:test:tag");
    handler.init(context);
    List<OperationProcessor> operationProcessores = handler.sources.get(0).getOperationProcessors();
    for (int i = 0; i < operationProcessores.size(); i++) {
        OperationProcessor operationProcessor = spy(operationProcessores.get(i));
        operationProcessores.set(i, operationProcessor);
    }
    handler.handler(events, context);
    /*
     * 2 operations specified in the config file
     */
    verify(operationProcessores.get(0), times(1)).perform(any());
    verify(operationProcessores.get(1), times(1)).perform(any());
}
Also used : TestContext(com.nextdoor.bender.aws.TestContext) ArrayList(java.util.ArrayList) OperationProcessor(com.nextdoor.bender.operation.OperationProcessor) Test(org.junit.Test)

Example 4 with OperationProcessor

use of com.nextdoor.bender.operation.OperationProcessor in project bender by Nextdoor.

the class PartitionOperationTest method testOperationThroughProcessor.

@Test
public void testOperationThroughProcessor() {
    List<PartitionSpec> partitionSpecs = new ArrayList<PartitionSpec>(1);
    List<String> sources = Arrays.asList("foo");
    PartitionSpec spec = new PartitionSpec("foo", sources, PartitionSpec.Interpreter.STRING);
    partitionSpecs.add(spec);
    PartitionOperation op = new PartitionOperation(partitionSpecs);
    InternalEvent ievent = new InternalEvent("foo", null, 1);
    DummyDeserializedEvent devent = spy(new DummyDeserializedEvent(""));
    ievent.setEventObj(devent);
    doReturn("baz").when(devent).getField("foo");
    DummyOperationFactory opFact = new DummyOperationFactory(op);
    OperationProcessor opProc = new OperationProcessor(opFact);
    opProc.perform(Stream.of(ievent)).count();
    LinkedHashMap<String, String> actual = ievent.getPartitions();
    LinkedHashMap<String, String> expected = new LinkedHashMap<String, String>(1);
    expected.put("foo", "baz");
    assertEquals(expected, actual);
}
Also used : DummyDeserializedEvent(com.nextdoor.bender.testutils.DummyDeserializerHelper.DummyDeserializedEvent) ArrayList(java.util.ArrayList) DummyOperationFactory(com.nextdoor.bender.testutils.DummyOperationHelper.DummyOperationFactory) OperationProcessor(com.nextdoor.bender.operation.OperationProcessor) InternalEvent(com.nextdoor.bender.InternalEvent) LinkedHashMap(java.util.LinkedHashMap) Test(org.junit.Test)

Aggregations

OperationProcessor (com.nextdoor.bender.operation.OperationProcessor)4 ArrayList (java.util.ArrayList)4 Test (org.junit.Test)3 InternalEvent (com.nextdoor.bender.InternalEvent)2 TestContext (com.nextdoor.bender.aws.TestContext)2 Context (com.amazonaws.services.lambda.runtime.Context)1 AmazonS3URI (com.amazonaws.services.s3.AmazonS3URI)1 AmazonS3ClientFactory (com.nextdoor.bender.aws.AmazonS3ClientFactory)1 BenderConfig (com.nextdoor.bender.config.BenderConfig)1 ConfigurationException (com.nextdoor.bender.config.ConfigurationException)1 HandlerResources (com.nextdoor.bender.config.HandlerResources)1 Source (com.nextdoor.bender.config.Source)1 DeserializedEvent (com.nextdoor.bender.deserializer.DeserializedEvent)1 DeserializerProcessor (com.nextdoor.bender.deserializer.DeserializerProcessor)1 IpcSenderService (com.nextdoor.bender.ipc.IpcSenderService)1 TransportException (com.nextdoor.bender.ipc.TransportException)1 BenderLayout (com.nextdoor.bender.logging.BenderLayout)1 Monitor (com.nextdoor.bender.monitoring.Monitor)1 Stat (com.nextdoor.bender.monitoring.Stat)1 BaseOperation (com.nextdoor.bender.operation.BaseOperation)1