@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) {
FlowFile flowFile = session.get();
if (flowFile == null) {
return;
}
final ProcessorLog logger = getLogger();
boolean encode = context.getProperty(MODE).getValue().equalsIgnoreCase(ENCODE_MODE);
String encoding = context.getProperty(ENCODING).getValue();
StreamCallback encoder = null;
// Select the encoder/decoder to use
if (encode) {
if (encoding.equalsIgnoreCase(BASE64_ENCODING)) {
encoder = new EncodeBase64();
} else if (encoding.equalsIgnoreCase(BASE32_ENCODING)) {
encoder = new EncodeBase32();
} else if (encoding.equalsIgnoreCase(HEX_ENCODING)) {
encoder = new EncodeHex();
}
} else {
if (encoding.equalsIgnoreCase(BASE64_ENCODING)) {
encoder = new DecodeBase64();
} else if (encoding.equalsIgnoreCase(BASE32_ENCODING)) {
encoder = new DecodeBase32();
} else if (encoding.equalsIgnoreCase(HEX_ENCODING)) {
encoder = new DecodeHex();
}
}
if (encoder == null) {
logger.warn(
"Unknown operation: {} {}", new Object[] {encode ? "encode" : "decode", encoding});
return;
}
try {
final StopWatch stopWatch = new StopWatch(true);
flowFile = session.write(flowFile, encoder);
logger.info("Successfully {} {}", new Object[] {encode ? "encoded" : "decoded", flowFile});
session
.getProvenanceReporter()
.modifyContent(flowFile, stopWatch.getElapsed(TimeUnit.MILLISECONDS));
session.transfer(flowFile, REL_SUCCESS);
} catch (Exception e) {
logger.error(
"Failed to {} {} due to {}", new Object[] {encode ? "encode" : "decode", flowFile, e});
session.transfer(flowFile, REL_FAILURE);
}
}
private DestinationAccepts getDestinationAcceptance(
final HttpClient client,
final String uri,
final ProcessorLog logger,
final String transactionId)
throws IOException {
final HttpHead head = new HttpHead(uri);
head.addHeader(TRANSACTION_ID_HEADER, transactionId);
final HttpResponse response = client.execute(head);
final int statusCode = response.getStatusLine().getStatusCode();
if (statusCode == Status.METHOD_NOT_ALLOWED.getStatusCode()) {
// we assume that the destination can support FlowFile v1 always.
return new DestinationAccepts(false, false, true, false, null);
} else if (statusCode == Status.OK.getStatusCode()) {
boolean acceptsFlowFileV3 = false;
boolean acceptsFlowFileV2 = false;
boolean acceptsFlowFileV1 = true;
boolean acceptsGzip = false;
Integer protocolVersion = null;
Header[] headers = response.getHeaders(ACCEPT);
if (headers != null) {
for (final Header header : headers) {
for (final String accepted : header.getValue().split(",")) {
final String trimmed = accepted.trim();
if (trimmed.equals(APPLICATION_FLOW_FILE_V3)) {
acceptsFlowFileV3 = true;
} else if (trimmed.equals(APPLICATION_FLOW_FILE_V2)) {
acceptsFlowFileV2 = true;
} else {
// we assume that the destination accepts FlowFile V1 because legacy versions
// of NiFi that accepted V1 did not use an Accept header to indicate it... or
// any other header. So the bets thing we can do is just assume that V1 is
// accepted, if we're going to send as FlowFile.
acceptsFlowFileV1 = true;
}
}
}
}
final Header destinationVersion = response.getFirstHeader(PROTOCOL_VERSION_HEADER);
if (destinationVersion != null) {
try {
protocolVersion = Integer.valueOf(destinationVersion.getValue());
} catch (final NumberFormatException e) {
// nothing to do here really.... it's an invalid value, so treat the same as if not
// specified
}
}
if (acceptsFlowFileV3) {
logger.debug(
"Connection to URI "
+ uri
+ " will be using Content Type "
+ APPLICATION_FLOW_FILE_V3
+ " if sending data as FlowFile");
} else if (acceptsFlowFileV2) {
logger.debug(
"Connection to URI "
+ uri
+ " will be using Content Type "
+ APPLICATION_FLOW_FILE_V2
+ " if sending data as FlowFile");
} else if (acceptsFlowFileV1) {
logger.debug(
"Connection to URI "
+ uri
+ " will be using Content Type "
+ APPLICATION_FLOW_FILE_V1
+ " if sending data as FlowFile");
}
headers = response.getHeaders(ACCEPT_ENCODING);
if (headers != null) {
for (final Header header : headers) {
for (final String accepted : header.getValue().split(",")) {
if (accepted.equalsIgnoreCase("gzip")) {
acceptsGzip = true;
}
}
}
}
if (acceptsGzip) {
logger.debug(
"Connection to URI " + uri + " indicates that inline GZIP compression is supported");
} else {
logger.debug(
"Connection to URI "
+ uri
+ " indicates that it does NOT support inline GZIP compression");
}
return new DestinationAccepts(
acceptsFlowFileV3, acceptsFlowFileV2, acceptsFlowFileV1, acceptsGzip, protocolVersion);
} else {
logger.warn(
"Unable to communicate with destination; when attempting to perform an HTTP HEAD, got unexpected response code of "
+ statusCode
+ ": "
+ response.getStatusLine().getReasonPhrase());
return new DestinationAccepts(false, false, false, false, null);
}
}
@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) {
final boolean sendAsFlowFile = context.getProperty(SEND_AS_FLOWFILE).asBoolean();
final int compressionLevel = context.getProperty(COMPRESSION_LEVEL).asInteger();
final String userAgent = context.getProperty(USER_AGENT).getValue();
final RequestConfig.Builder requestConfigBuilder = RequestConfig.custom();
requestConfigBuilder.setConnectionRequestTimeout(
context.getProperty(DATA_TIMEOUT).asTimePeriod(TimeUnit.MILLISECONDS).intValue());
requestConfigBuilder.setConnectTimeout(
context.getProperty(CONNECTION_TIMEOUT).asTimePeriod(TimeUnit.MILLISECONDS).intValue());
requestConfigBuilder.setRedirectsEnabled(false);
requestConfigBuilder.setSocketTimeout(
context.getProperty(DATA_TIMEOUT).asTimePeriod(TimeUnit.MILLISECONDS).intValue());
final RequestConfig requestConfig = requestConfigBuilder.build();
final StreamThrottler throttler = throttlerRef.get();
final ProcessorLog logger = getLogger();
final Double maxBatchBytes = context.getProperty(MAX_BATCH_SIZE).asDataSize(DataUnit.B);
String lastUrl = null;
long bytesToSend = 0L;
final List<FlowFile> toSend = new ArrayList<>();
DestinationAccepts destinationAccepts = null;
CloseableHttpClient client = null;
final String transactionId = UUID.randomUUID().toString();
final ObjectHolder<String> dnHolder = new ObjectHolder<>("none");
while (true) {
FlowFile flowFile = session.get();
if (flowFile == null) {
break;
}
final String url = context.getProperty(URL).evaluateAttributeExpressions(flowFile).getValue();
try {
new java.net.URL(url);
} catch (final MalformedURLException e) {
logger.error(
"After substituting attribute values for {}, URL is {}; this is not a valid URL, so routing to failure",
new Object[] {flowFile, url});
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
continue;
}
// If this FlowFile doesn't have the same url, throw it back on the queue and stop grabbing
// FlowFiles
if (lastUrl != null && !lastUrl.equals(url)) {
session.transfer(flowFile);
break;
}
lastUrl = url;
toSend.add(flowFile);
if (client == null || destinationAccepts == null) {
final Config config = getConfig(url, context);
final HttpClientConnectionManager conMan = config.getConnectionManager();
final HttpClientBuilder clientBuilder = HttpClientBuilder.create();
clientBuilder.setConnectionManager(conMan);
clientBuilder.setUserAgent(userAgent);
clientBuilder.addInterceptorFirst(
new HttpResponseInterceptor() {
@Override
public void process(final HttpResponse response, final HttpContext httpContext)
throws HttpException, IOException {
HttpCoreContext coreContext = HttpCoreContext.adapt(httpContext);
ManagedHttpClientConnection conn =
coreContext.getConnection(ManagedHttpClientConnection.class);
if (!conn.isOpen()) {
return;
}
SSLSession sslSession = conn.getSSLSession();
if (sslSession != null) {
final X509Certificate[] certChain = sslSession.getPeerCertificateChain();
if (certChain == null || certChain.length == 0) {
throw new SSLPeerUnverifiedException("No certificates found");
}
final X509Certificate cert = certChain[0];
dnHolder.set(cert.getSubjectDN().getName().trim());
}
}
});
clientBuilder.disableAutomaticRetries();
clientBuilder.disableContentCompression();
final String username = context.getProperty(USERNAME).getValue();
final String password = context.getProperty(PASSWORD).getValue();
// set the credentials if appropriate
if (username != null) {
final CredentialsProvider credentialsProvider = new BasicCredentialsProvider();
if (password == null) {
credentialsProvider.setCredentials(
AuthScope.ANY, new UsernamePasswordCredentials(username));
} else {
credentialsProvider.setCredentials(
AuthScope.ANY, new UsernamePasswordCredentials(username, password));
}
clientBuilder.setDefaultCredentialsProvider(credentialsProvider);
}
client = clientBuilder.build();
// determine whether or not destination accepts flowfile/gzip
destinationAccepts = config.getDestinationAccepts();
if (destinationAccepts == null) {
try {
if (sendAsFlowFile) {
destinationAccepts =
getDestinationAcceptance(client, url, getLogger(), transactionId);
} else {
destinationAccepts = new DestinationAccepts(false, false, false, false, null);
}
config.setDestinationAccepts(destinationAccepts);
} catch (IOException e) {
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
logger.error(
"Unable to communicate with destination {} to determine whether or not it can accept "
+ "flowfiles/gzip; routing {} to failure due to {}",
new Object[] {url, flowFile, e});
context.yield();
return;
}
}
}
// if we are not sending as flowfile, or if the destination doesn't accept V3 or V2
// (streaming) format,
// then only use a single FlowFile
if (!sendAsFlowFile
|| (!destinationAccepts.isFlowFileV3Accepted()
&& !destinationAccepts.isFlowFileV2Accepted())) {
break;
}
bytesToSend += flowFile.getSize();
if (bytesToSend > maxBatchBytes.longValue()) {
break;
}
}
if (toSend.isEmpty()) {
return;
}
final String url = lastUrl;
final HttpPost post = new HttpPost(url);
final List<FlowFile> flowFileList = toSend;
final DestinationAccepts accepts = destinationAccepts;
final boolean isDestinationLegacyNiFi = accepts.getProtocolVersion() == null;
final EntityTemplate entity =
new EntityTemplate(
new ContentProducer() {
@Override
public void writeTo(final OutputStream rawOut) throws IOException {
final OutputStream throttled =
(throttler == null) ? rawOut : throttler.newThrottledOutputStream(rawOut);
OutputStream wrappedOut = new BufferedOutputStream(throttled);
if (compressionLevel > 0 && accepts.isGzipAccepted()) {
wrappedOut = new GZIPOutputStream(wrappedOut, compressionLevel);
}
try (final OutputStream out = wrappedOut) {
for (final FlowFile flowFile : flowFileList) {
session.read(
flowFile,
new InputStreamCallback() {
@Override
public void process(final InputStream rawIn) throws IOException {
try (final InputStream in = new BufferedInputStream(rawIn)) {
FlowFilePackager packager = null;
if (!sendAsFlowFile) {
packager = null;
} else if (accepts.isFlowFileV3Accepted()) {
packager = new FlowFilePackagerV3();
} else if (accepts.isFlowFileV2Accepted()) {
packager = new FlowFilePackagerV2();
} else if (accepts.isFlowFileV1Accepted()) {
packager = new FlowFilePackagerV1();
}
// if none of the above conditions is met, we should never get here,
// because
// we will have already verified that at least 1 of the FlowFile
// packaging
// formats is acceptable if sending as FlowFile.
if (packager == null) {
StreamUtils.copy(in, out);
} else {
final Map<String, String> flowFileAttributes;
if (isDestinationLegacyNiFi) {
// Old versions of NiFi expect nf.file.name and nf.file.path to
// indicate filename & path;
// in order to maintain backward compatibility, we copy the
// filename & path to those attribute keys.
flowFileAttributes = new HashMap<>(flowFile.getAttributes());
flowFileAttributes.put(
"nf.file.name",
flowFile.getAttribute(CoreAttributes.FILENAME.key()));
flowFileAttributes.put(
"nf.file.path",
flowFile.getAttribute(CoreAttributes.PATH.key()));
} else {
flowFileAttributes = flowFile.getAttributes();
}
packager.packageFlowFile(
in, out, flowFileAttributes, flowFile.getSize());
}
}
}
});
}
out.flush();
}
}
});
entity.setChunked(context.getProperty(CHUNKED_ENCODING).asBoolean());
post.setEntity(entity);
post.setConfig(requestConfig);
final String contentType;
if (sendAsFlowFile) {
if (accepts.isFlowFileV3Accepted()) {
contentType = APPLICATION_FLOW_FILE_V3;
} else if (accepts.isFlowFileV2Accepted()) {
contentType = APPLICATION_FLOW_FILE_V2;
} else if (accepts.isFlowFileV1Accepted()) {
contentType = APPLICATION_FLOW_FILE_V1;
} else {
logger.error(
"Cannot send data to {} because the destination does not accept FlowFiles and this processor is "
+ "configured to deliver FlowFiles; rolling back session",
new Object[] {url});
session.rollback();
context.yield();
return;
}
} else {
final String attributeValue = toSend.get(0).getAttribute(CoreAttributes.MIME_TYPE.key());
contentType = (attributeValue == null) ? DEFAULT_CONTENT_TYPE : attributeValue;
}
final String attributeHeaderRegex = context.getProperty(ATTRIBUTES_AS_HEADERS_REGEX).getValue();
if (attributeHeaderRegex != null && !sendAsFlowFile && flowFileList.size() == 1) {
final Pattern pattern = Pattern.compile(attributeHeaderRegex);
final Map<String, String> attributes = flowFileList.get(0).getAttributes();
for (final Map.Entry<String, String> entry : attributes.entrySet()) {
final String key = entry.getKey();
if (pattern.matcher(key).matches()) {
post.setHeader(entry.getKey(), entry.getValue());
}
}
}
post.setHeader(CONTENT_TYPE, contentType);
post.setHeader(FLOWFILE_CONFIRMATION_HEADER, "true");
post.setHeader(PROTOCOL_VERSION_HEADER, PROTOCOL_VERSION);
post.setHeader(TRANSACTION_ID_HEADER, transactionId);
if (compressionLevel > 0 && accepts.isGzipAccepted()) {
post.setHeader(GZIPPED_HEADER, "true");
}
// Do the actual POST
final String flowFileDescription =
toSend.size() <= 10 ? toSend.toString() : toSend.size() + " FlowFiles";
final String uploadDataRate;
final long uploadMillis;
CloseableHttpResponse response = null;
try {
final StopWatch stopWatch = new StopWatch(true);
response = client.execute(post);
// consume input stream entirely, ignoring its contents. If we
// don't do this, the Connection will not be returned to the pool
EntityUtils.consume(response.getEntity());
stopWatch.stop();
uploadDataRate = stopWatch.calculateDataRate(bytesToSend);
uploadMillis = stopWatch.getDuration(TimeUnit.MILLISECONDS);
} catch (final IOException e) {
logger.error(
"Failed to Post {} due to {}; transferring to failure",
new Object[] {flowFileDescription, e});
context.yield();
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
}
return;
} finally {
if (response != null) {
try {
response.close();
} catch (IOException e) {
getLogger().warn("Failed to close HTTP Response due to {}", new Object[] {e});
}
}
}
// If we get a 'SEE OTHER' status code and an HTTP header that indicates that the intent
// of the Location URI is a flowfile hold, we will store this holdUri. This prevents us
// from posting to some other webservice and then attempting to delete some resource to which
// we are redirected
final int responseCode = response.getStatusLine().getStatusCode();
final String responseReason = response.getStatusLine().getReasonPhrase();
String holdUri = null;
if (responseCode == HttpServletResponse.SC_SEE_OTHER) {
final Header locationUriHeader = response.getFirstHeader(LOCATION_URI_INTENT_NAME);
if (locationUriHeader != null) {
if (LOCATION_URI_INTENT_VALUE.equals(locationUriHeader.getValue())) {
final Header holdUriHeader = response.getFirstHeader(LOCATION_HEADER_NAME);
if (holdUriHeader != null) {
holdUri = holdUriHeader.getValue();
}
}
}
if (holdUri == null) {
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
logger.error(
"Failed to Post {} to {}: sent content and received status code {}:{} but no Hold URI",
new Object[] {flowFile, url, responseCode, responseReason});
session.transfer(flowFile, REL_FAILURE);
}
return;
}
}
if (holdUri == null) {
if (responseCode == HttpServletResponse.SC_SERVICE_UNAVAILABLE) {
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
logger.error(
"Failed to Post {} to {}: response code was {}:{}; will yield processing, "
+ "since the destination is temporarily unavailable",
new Object[] {flowFile, url, responseCode, responseReason});
session.transfer(flowFile, REL_FAILURE);
}
context.yield();
return;
}
if (responseCode >= 300) {
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
logger.error(
"Failed to Post {} to {}: response code was {}:{}",
new Object[] {flowFile, url, responseCode, responseReason});
session.transfer(flowFile, REL_FAILURE);
}
return;
}
logger.info(
"Successfully Posted {} to {} in {} at a rate of {}",
new Object[] {
flowFileDescription,
url,
FormatUtils.formatMinutesSeconds(uploadMillis, TimeUnit.MILLISECONDS),
uploadDataRate
});
for (final FlowFile flowFile : toSend) {
session
.getProvenanceReporter()
.send(flowFile, url, "Remote DN=" + dnHolder.get(), uploadMillis, true);
session.transfer(flowFile, REL_SUCCESS);
}
return;
}
//
// the response indicated a Hold URI; delete the Hold.
//
// determine the full URI of the Flow File's Hold; Unfortunately, the responses that are
// returned have
// changed over the past, so we have to take into account a few different possibilities.
String fullHoldUri = holdUri;
if (holdUri.startsWith("/contentListener")) {
// If the Hold URI that we get starts with /contentListener, it may not really be
// /contentListener,
// as this really indicates that it should be whatever we posted to -- if posting directly to
// the
// ListenHTTP component, it will be /contentListener, but if posting to a proxy/load balancer,
// we may
// be posting to some other URL.
fullHoldUri = url + holdUri.substring(16);
} else if (holdUri.startsWith("/")) {
// URL indicates the full path but not hostname or port; use the same hostname & port that we
// posted
// to but use the full path indicated by the response.
int firstSlash = url.indexOf("/", 8);
if (firstSlash < 0) {
firstSlash = url.length();
}
final String beforeSlash = url.substring(0, firstSlash);
fullHoldUri = beforeSlash + holdUri;
} else if (!holdUri.startsWith("http")) {
// Absolute URL
fullHoldUri = url + (url.endsWith("/") ? "" : "/") + holdUri;
}
final HttpDelete delete = new HttpDelete(fullHoldUri);
delete.setHeader(TRANSACTION_ID_HEADER, transactionId);
while (true) {
try {
final HttpResponse holdResponse = client.execute(delete);
EntityUtils.consume(holdResponse.getEntity());
final int holdStatusCode = holdResponse.getStatusLine().getStatusCode();
final String holdReason = holdResponse.getStatusLine().getReasonPhrase();
if (holdStatusCode >= 300) {
logger.error(
"Failed to delete Hold that destination placed on {}: got response code {}:{}; routing to failure",
new Object[] {flowFileDescription, holdStatusCode, holdReason});
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
}
return;
}
logger.info(
"Successfully Posted {} to {} in {} milliseconds at a rate of {}",
new Object[] {flowFileDescription, url, uploadMillis, uploadDataRate});
for (FlowFile flowFile : toSend) {
session.getProvenanceReporter().send(flowFile, url);
session.transfer(flowFile, REL_SUCCESS);
}
return;
} catch (final IOException e) {
logger.warn(
"Failed to delete Hold that destination placed on {} due to {}",
new Object[] {flowFileDescription, e});
}
if (!isScheduled()) {
context.yield();
logger.warn(
"Failed to delete Hold that destination placed on {}; Processor has been stopped so routing FlowFile(s) to failure",
new Object[] {flowFileDescription});
for (FlowFile flowFile : toSend) {
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
}
return;
}
}
}
@Override
public void onTrigger(final ProcessContext context, final ProcessSession session)
throws ProcessException {
FlowFile flowFile = session.get();
if (flowFile == null) {
return;
}
final ProcessorLog logger = getLogger();
// cache key is computed from attribute 'CACHE_ENTRY_IDENTIFIER' with expression language
// support
final String cacheKey =
context
.getProperty(CACHE_ENTRY_IDENTIFIER)
.evaluateAttributeExpressions(flowFile)
.getValue();
// if the computed value is null, or empty, we transfer the flow file to failure relationship
if (StringUtils.isBlank(cacheKey)) {
logger.error(
"FlowFile {} has no attribute for given Cache Entry Identifier", new Object[] {flowFile});
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
return;
}
// the cache client used to interact with the distributed cache
final DistributedMapCacheClient cache =
context
.getProperty(DISTRIBUTED_CACHE_SERVICE)
.asControllerService(DistributedMapCacheClient.class);
try {
final long maxCacheEntrySize =
context.getProperty(CACHE_ENTRY_MAX_BYTES).asDataSize(DataUnit.B).longValue();
long flowFileSize = flowFile.getSize();
// too big flow file
if (flowFileSize > maxCacheEntrySize) {
logger.warn(
"Flow file {} size {} exceeds the max cache entry size ({} B).",
new Object[] {flowFile, flowFileSize, maxCacheEntrySize});
session.transfer(flowFile, REL_FAILURE);
return;
}
if (flowFileSize == 0) {
logger.warn("Flow file {} is empty, there is nothing to cache.", new Object[] {flowFile});
session.transfer(flowFile, REL_FAILURE);
return;
}
// get flow file content
final ByteArrayOutputStream byteStream = new ByteArrayOutputStream();
session.exportTo(flowFile, byteStream);
byte[] cacheValue = byteStream.toByteArray();
final String updateStrategy = context.getProperty(CACHE_UPDATE_STRATEGY).getValue();
boolean cached = false;
if (updateStrategy.equals(CACHE_UPDATE_REPLACE.getValue())) {
cache.put(cacheKey, cacheValue, keySerializer, valueSerializer);
cached = true;
} else if (updateStrategy.equals(CACHE_UPDATE_KEEP_ORIGINAL.getValue())) {
final byte[] oldValue =
cache.getAndPutIfAbsent(
cacheKey, cacheValue, keySerializer, valueSerializer, valueDeserializer);
if (oldValue == null) {
cached = true;
}
}
// set 'cached' attribute
flowFile = session.putAttribute(flowFile, CACHED_ATTRIBUTE_NAME, String.valueOf(cached));
if (cached) {
session.transfer(flowFile, REL_SUCCESS);
} else {
session.transfer(flowFile, REL_FAILURE);
}
} catch (final IOException e) {
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
logger.error(
"Unable to communicate with cache when processing {} due to {}",
new Object[] {flowFile, e});
}
}