Search in sources :

Example 6 with ProvenanceEventBuilder

use of org.apache.nifi.provenance.ProvenanceEventBuilder in project nifi by apache.

the class StandardProcessSession method updateProvenanceRepo.

private void updateProvenanceRepo(final Checkpoint checkpoint) {
    // Update Provenance Repository
    final ProvenanceEventRepository provenanceRepo = context.getProvenanceRepository();
    // We need to de-dupe the events that we've created and those reported to the provenance reporter,
    // in case the Processor developer submitted the same events to the reporter. So we use a LinkedHashSet
    // for this, so that we are able to ensure that the events are submitted in the proper order.
    final Set<ProvenanceEventRecord> recordsToSubmit = new LinkedHashSet<>();
    final Map<String, Set<ProvenanceEventType>> eventTypesPerFlowFileId = new HashMap<>();
    final Set<ProvenanceEventRecord> processorGenerated = checkpoint.reportedEvents;
    // by the Processor contains any of the FORK events that we generated
    for (final Map.Entry<FlowFile, ProvenanceEventBuilder> entry : checkpoint.forkEventBuilders.entrySet()) {
        final ProvenanceEventBuilder builder = entry.getValue();
        final FlowFile flowFile = entry.getKey();
        updateEventContentClaims(builder, flowFile, checkpoint.records.get(flowFile));
        final ProvenanceEventRecord event = builder.build();
        if (!event.getChildUuids().isEmpty() && !isSpuriousForkEvent(event, checkpoint.removedFlowFiles)) {
            // If framework generated the event, add it to the 'recordsToSubmit' Set.
            if (!processorGenerated.contains(event)) {
                recordsToSubmit.add(event);
            }
            // Register the FORK event for each child and each parent.
            for (final String childUuid : event.getChildUuids()) {
                addEventType(eventTypesPerFlowFileId, childUuid, event.getEventType());
            }
            for (final String parentUuid : event.getParentUuids()) {
                addEventType(eventTypesPerFlowFileId, parentUuid, event.getEventType());
            }
        }
    }
    // Now add any Processor-reported events.
    for (final ProvenanceEventRecord event : processorGenerated) {
        if (isSpuriousForkEvent(event, checkpoint.removedFlowFiles)) {
            continue;
        }
        // connection from which it was pulled (and only this connection). If so, discard the event.
        if (isSpuriousRouteEvent(event, checkpoint.records)) {
            continue;
        }
        recordsToSubmit.add(event);
        addEventType(eventTypesPerFlowFileId, event.getFlowFileUuid(), event.getEventType());
    }
    // Finally, add any other events that we may have generated.
    for (final List<ProvenanceEventRecord> eventList : checkpoint.generatedProvenanceEvents.values()) {
        for (final ProvenanceEventRecord event : eventList) {
            if (isSpuriousForkEvent(event, checkpoint.removedFlowFiles)) {
                continue;
            }
            recordsToSubmit.add(event);
            addEventType(eventTypesPerFlowFileId, event.getFlowFileUuid(), event.getEventType());
        }
    }
    // Check if content or attributes changed. If so, register the appropriate events.
    for (final StandardRepositoryRecord repoRecord : checkpoint.records.values()) {
        final ContentClaim original = repoRecord.getOriginalClaim();
        final ContentClaim current = repoRecord.getCurrentClaim();
        boolean contentChanged = false;
        if (original == null && current != null) {
            contentChanged = true;
        }
        if (original != null && current == null) {
            contentChanged = true;
        }
        if (original != null && current != null && !original.equals(current)) {
            contentChanged = true;
        }
        final FlowFileRecord curFlowFile = repoRecord.getCurrent();
        final String flowFileId = curFlowFile.getAttribute(CoreAttributes.UUID.key());
        boolean eventAdded = false;
        if (checkpoint.removedFlowFiles.contains(flowFileId)) {
            continue;
        }
        final boolean newFlowFile = repoRecord.getOriginal() == null;
        if (contentChanged && !newFlowFile) {
            recordsToSubmit.add(provenanceReporter.build(curFlowFile, ProvenanceEventType.CONTENT_MODIFIED).build());
            addEventType(eventTypesPerFlowFileId, flowFileId, ProvenanceEventType.CONTENT_MODIFIED);
            eventAdded = true;
        }
        if (checkpoint.createdFlowFiles.contains(flowFileId)) {
            final Set<ProvenanceEventType> registeredTypes = eventTypesPerFlowFileId.get(flowFileId);
            boolean creationEventRegistered = false;
            if (registeredTypes != null) {
                if (registeredTypes.contains(ProvenanceEventType.CREATE) || registeredTypes.contains(ProvenanceEventType.FORK) || registeredTypes.contains(ProvenanceEventType.JOIN) || registeredTypes.contains(ProvenanceEventType.RECEIVE) || registeredTypes.contains(ProvenanceEventType.FETCH)) {
                    creationEventRegistered = true;
                }
            }
            if (!creationEventRegistered) {
                recordsToSubmit.add(provenanceReporter.build(curFlowFile, ProvenanceEventType.CREATE).build());
                eventAdded = true;
            }
        }
        if (!eventAdded && !repoRecord.getUpdatedAttributes().isEmpty()) {
            // event is redundant if another already exists.
            if (!eventTypesPerFlowFileId.containsKey(flowFileId)) {
                recordsToSubmit.add(provenanceReporter.build(curFlowFile, ProvenanceEventType.ATTRIBUTES_MODIFIED).build());
                addEventType(eventTypesPerFlowFileId, flowFileId, ProvenanceEventType.ATTRIBUTES_MODIFIED);
            }
        }
    }
    // We want to submit the 'recordsToSubmit' collection, followed by the auto-terminated events to the Provenance Repository.
    // We want to do this with a single call to ProvenanceEventRepository#registerEvents because it may be much more efficient
    // to do so.
    // However, we want to modify the events in 'recordsToSubmit' to obtain the data from the most recent version of the FlowFiles
    // (except for SEND events); see note below as to why this is
    // Therefore, we create an Iterable that can iterate over each of these events, modifying them as needed, and returning them
    // in the appropriate order. This prevents an unnecessary step of creating an intermediate List and adding all of those values
    // to the List.
    // This is done in a similar veign to how Java 8's streams work, iterating over the events and returning a processed version
    // one-at-a-time as opposed to iterating over the entire Collection and putting the results in another Collection. However,
    // we don't want to change the Framework to require Java 8 at this time, because it's not yet as prevalent as we would desire
    final Map<String, FlowFileRecord> flowFileRecordMap = new HashMap<>();
    for (final StandardRepositoryRecord repoRecord : checkpoint.records.values()) {
        final FlowFileRecord flowFile = repoRecord.getCurrent();
        flowFileRecordMap.put(flowFile.getAttribute(CoreAttributes.UUID.key()), flowFile);
    }
    final List<ProvenanceEventRecord> autoTermEvents = checkpoint.autoTerminatedEvents;
    final Iterable<ProvenanceEventRecord> iterable = new Iterable<ProvenanceEventRecord>() {

        final Iterator<ProvenanceEventRecord> recordsToSubmitIterator = recordsToSubmit.iterator();

        final Iterator<ProvenanceEventRecord> autoTermIterator = autoTermEvents == null ? null : autoTermEvents.iterator();

        @Override
        public Iterator<ProvenanceEventRecord> iterator() {
            return new Iterator<ProvenanceEventRecord>() {

                @Override
                public boolean hasNext() {
                    return recordsToSubmitIterator.hasNext() || autoTermIterator != null && autoTermIterator.hasNext();
                }

                @Override
                public ProvenanceEventRecord next() {
                    if (recordsToSubmitIterator.hasNext()) {
                        final ProvenanceEventRecord rawEvent = recordsToSubmitIterator.next();
                        // exposed.
                        return enrich(rawEvent, flowFileRecordMap, checkpoint.records, rawEvent.getEventType() != ProvenanceEventType.SEND);
                    } else if (autoTermIterator != null && autoTermIterator.hasNext()) {
                        return enrich(autoTermIterator.next(), flowFileRecordMap, checkpoint.records, true);
                    }
                    throw new NoSuchElementException();
                }

                @Override
                public void remove() {
                    throw new UnsupportedOperationException();
                }
            };
        }
    };
    provenanceRepo.registerEvents(iterable);
}
Also used : LinkedHashSet(java.util.LinkedHashSet) Set(java.util.Set) HashSet(java.util.HashSet) LinkedHashSet(java.util.LinkedHashSet) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) StandardProvenanceEventRecord(org.apache.nifi.provenance.StandardProvenanceEventRecord) ProvenanceEventRecord(org.apache.nifi.provenance.ProvenanceEventRecord) Iterator(java.util.Iterator) ProvenanceEventType(org.apache.nifi.provenance.ProvenanceEventType) ProvenanceEventBuilder(org.apache.nifi.provenance.ProvenanceEventBuilder) FlowFile(org.apache.nifi.flowfile.FlowFile) ProvenanceEventRepository(org.apache.nifi.provenance.ProvenanceEventRepository) ContentClaim(org.apache.nifi.controller.repository.claim.ContentClaim) Map(java.util.Map) ConcurrentHashMap(java.util.concurrent.ConcurrentHashMap) HashMap(java.util.HashMap) NoSuchElementException(java.util.NoSuchElementException)

Example 7 with ProvenanceEventBuilder

use of org.apache.nifi.provenance.ProvenanceEventBuilder in project nifi by apache.

the class StandardProvenanceReporter method drop.

ProvenanceEventRecord drop(final FlowFile flowFile, final String reason) {
    try {
        final ProvenanceEventBuilder builder = build(flowFile, ProvenanceEventType.DROP);
        if (reason != null) {
            builder.setDetails("Discard reason: " + reason);
        }
        final ProvenanceEventRecord record = builder.build();
        events.add(record);
        return record;
    } catch (final Exception e) {
        logger.error("Failed to generate Provenance Event due to " + e);
        if (logger.isDebugEnabled()) {
            logger.error("", e);
        }
        return null;
    }
}
Also used : ProvenanceEventRecord(org.apache.nifi.provenance.ProvenanceEventRecord) FlowFileHandlingException(org.apache.nifi.processor.exception.FlowFileHandlingException) ProvenanceEventBuilder(org.apache.nifi.provenance.ProvenanceEventBuilder)

Example 8 with ProvenanceEventBuilder

use of org.apache.nifi.provenance.ProvenanceEventBuilder in project nifi by apache.

the class StandardProvenanceReporter method fork.

@Override
public void fork(final FlowFile parent, final Collection<FlowFile> children, final String details, final long forkDuration) {
    verifyFlowFileKnown(parent);
    try {
        final ProvenanceEventBuilder eventBuilder = build(parent, ProvenanceEventType.FORK);
        eventBuilder.addParentFlowFile(parent);
        for (final FlowFile child : children) {
            eventBuilder.addChildFlowFile(child);
        }
        if (forkDuration > -1L) {
            eventBuilder.setEventDuration(forkDuration);
        }
        if (details != null) {
            eventBuilder.setDetails(details);
        }
        events.add(eventBuilder.build());
    } catch (final Exception e) {
        logger.error("Failed to generate Provenance Event due to " + e);
        if (logger.isDebugEnabled()) {
            logger.error("", e);
        }
    }
}
Also used : FlowFile(org.apache.nifi.flowfile.FlowFile) FlowFileHandlingException(org.apache.nifi.processor.exception.FlowFileHandlingException) ProvenanceEventBuilder(org.apache.nifi.provenance.ProvenanceEventBuilder)

Example 9 with ProvenanceEventBuilder

use of org.apache.nifi.provenance.ProvenanceEventBuilder in project nifi by apache.

the class StandardProvenanceReporter method build.

ProvenanceEventBuilder build(final FlowFile flowFile, final ProvenanceEventType eventType) {
    final ProvenanceEventBuilder builder = repository.eventBuilder();
    builder.setEventType(eventType);
    builder.fromFlowFile(flowFile);
    builder.setLineageStartDate(flowFile.getLineageStartDate());
    builder.setComponentId(processorId);
    builder.setComponentType(processorType);
    return builder;
}
Also used : ProvenanceEventBuilder(org.apache.nifi.provenance.ProvenanceEventBuilder)

Example 10 with ProvenanceEventBuilder

use of org.apache.nifi.provenance.ProvenanceEventBuilder in project nifi by apache.

the class StandardProvenanceReporter method join.

@Override
public void join(final Collection<FlowFile> parents, final FlowFile child, final String details, final long joinDuration) {
    verifyFlowFileKnown(child);
    try {
        final ProvenanceEventBuilder eventBuilder = build(child, ProvenanceEventType.JOIN);
        eventBuilder.addChildFlowFile(child);
        eventBuilder.setDetails(details);
        for (final FlowFile parent : parents) {
            eventBuilder.addParentFlowFile(parent);
        }
        events.add(eventBuilder.build());
    } catch (final Exception e) {
        logger.error("Failed to generate Provenance Event due to " + e);
        if (logger.isDebugEnabled()) {
            logger.error("", e);
        }
    }
}
Also used : FlowFile(org.apache.nifi.flowfile.FlowFile) FlowFileHandlingException(org.apache.nifi.processor.exception.FlowFileHandlingException) ProvenanceEventBuilder(org.apache.nifi.provenance.ProvenanceEventBuilder)

Aggregations

ProvenanceEventBuilder (org.apache.nifi.provenance.ProvenanceEventBuilder)17 FlowFileHandlingException (org.apache.nifi.processor.exception.FlowFileHandlingException)9 FlowFile (org.apache.nifi.flowfile.FlowFile)8 ProvenanceEventRecord (org.apache.nifi.provenance.ProvenanceEventRecord)4 HashMap (java.util.HashMap)3 ContentClaim (org.apache.nifi.controller.repository.claim.ContentClaim)3 StandardProvenanceEventRecord (org.apache.nifi.provenance.StandardProvenanceEventRecord)3 HashSet (java.util.HashSet)2 Iterator (java.util.Iterator)2 LinkedHashSet (java.util.LinkedHashSet)2 Map (java.util.Map)2 NoSuchElementException (java.util.NoSuchElementException)2 Set (java.util.Set)2 ConcurrentHashMap (java.util.concurrent.ConcurrentHashMap)2 Connectable (org.apache.nifi.connectable.Connectable)2 ProvenanceEventRepository (org.apache.nifi.provenance.ProvenanceEventRepository)2 ProvenanceEventType (org.apache.nifi.provenance.ProvenanceEventType)2 BufferedOutputStream (java.io.BufferedOutputStream)1 ByteArrayInputStream (java.io.ByteArrayInputStream)1 Closeable (java.io.Closeable)1