@Override
public void operate(FlowProcess flowProcess, BufferCall<NullContext> bufferCall) {
if (_useFiledsNone) {
int count = 0;
Iterator<TupleEntry> iter = bufferCall.getArgumentsIterator();
while (iter.hasNext()) {
iter.next();
count += 1;
}
bufferCall.getOutputCollector().add(new Tuple(count));
} else {
bufferCall.getOutputCollector().add(bufferCall.getGroup());
}
}
@Override
public void operate(FlowProcess flowProcess, BufferCall<NullContext> bufferCall) {
TupleEntry group = bufferCall.getGroup();
String protocolAndDomain = group.getString(0);
LOGGER.info("Processing tuple group: " + group);
DiskQueue<GroupedUrlDatum> urls = new DiskQueue<GroupedUrlDatum>(MAX_URLS_IN_MEMORY);
Iterator<TupleEntry> values = bufferCall.getArgumentsIterator();
while (values.hasNext()) {
urls.add(new GroupedUrlDatum(new TupleEntry(values.next())));
}
try {
Runnable doRobots =
new ProcessRobotsTask(
protocolAndDomain,
_scorer,
urls,
_fetcher,
_parser,
bufferCall.getOutputCollector(),
_flowProcess);
_executor.execute(doRobots);
} catch (RejectedExecutionException e) {
// should never happen.
LOGGER.error("Robots handling pool rejected our request for " + protocolAndDomain);
_flowProcess.increment(FetchCounters.DOMAINS_REJECTED, 1);
_flowProcess.increment(FetchCounters.URLS_REJECTED, urls.size());
ProcessRobotsTask.emptyQueue(
urls, GroupingKey.DEFERRED_GROUPING_KEY, bufferCall.getOutputCollector(), flowProcess);
} catch (Throwable t) {
LOGGER.error(
"Caught an unexpected throwable - robots handling rejected our request for "
+ protocolAndDomain,
t);
_flowProcess.increment(FetchCounters.DOMAINS_REJECTED, 1);
_flowProcess.increment(FetchCounters.URLS_REJECTED, urls.size());
ProcessRobotsTask.emptyQueue(
urls, GroupingKey.DEFERRED_GROUPING_KEY, bufferCall.getOutputCollector(), flowProcess);
}
}
@Override
public void operate(FlowProcess flowProcess, BufferCall bufferCall) {
Iterator<TupleEntry> it = bufferCall.getArgumentsIterator();
HyperLogLog merged = null;
try {
while (it.hasNext()) {
TupleEntry tupleEntry = it.next();
byte[] serialized = (byte[]) tupleEntry.getObject(0);
HyperLogLog hll = HyperLogLog.Builder.build(serialized);
if (merged == null) {
merged = hll;
} else {
merged = (HyperLogLog) merged.merge(hll);
}
}
} catch (Exception e) {
throw new RuntimeException(e);
}
}
public void operate(FlowProcess flowProcess, BufferCall<TupleEntryCollector> bufferCall) {
if (bufferCall.getJoinerClosure() != null)
throw new IllegalStateException("joiner closure should be null");
if (insertHeader) bufferCall.getOutputCollector().add(new Tuple(value));
Iterator<TupleEntry> iterator = bufferCall.getArgumentsIterator();
while (iterator.hasNext()) {
TupleEntry arguments = iterator.next(); // must be called
if (expectedSize != -1 && arguments.size() != expectedSize)
throw new RuntimeException("arguments wrong size");
if (path != null) bufferCall.getContext().add(arguments);
if (value != null) bufferCall.getOutputCollector().add(new Tuple(value));
else bufferCall.getOutputCollector().add(arguments); // copy
}
if (insertFooter) bufferCall.getOutputCollector().add(new Tuple(value));
iterator.hasNext(); // regression
}
@Override
public void operate(FlowProcess process, BufferCall<NullContext> buffCall) {
QueuedValues values = new QueuedValues(buffCall.getArgumentsIterator());
_collector = buffCall.getOutputCollector();
FetcherPolicy fetcherPolicy = _fetcher.getFetcherPolicy();
// Each value is a PreFetchedDatum that contains a set of URLs to fetch in one request from
// a single server, plus other values needed to set state properly.
while (!Thread.interrupted() && !fetcherPolicy.isTerminateFetch() && !values.isEmpty()) {
FetchSetDatum datum = values.nextOrNull(_fetcherMode);
try {
if (datum == null) {
trace("Nothing ready to fetch, sleeping...");
process.keepAlive();
Thread.sleep(NOTHING_TO_FETCH_SLEEP_TIME);
} else {
List<ScoredUrlDatum> urls = datum.getUrls();
String ref = datum.getGroupingRef();
trace("Processing %d URLs for %s", urls.size(), ref);
Runnable doFetch = new FetchTask(this, _fetcher, urls, ref);
if (datum.isLastList()) {
makeActive(ref, 0L);
trace("Executing fetch of %d URLs from %s (last batch)", urls.size(), ref);
} else {
Long nextFetchTime = System.currentTimeMillis() + datum.getFetchDelay();
makeActive(ref, nextFetchTime);
trace(
"Executing fetch of %d URLs from %s (next fetch time %d)",
urls.size(), ref, nextFetchTime);
}
long startTime = System.currentTimeMillis();
try {
_executor.execute(doFetch);
} catch (RejectedExecutionException e) {
// should never happen.
LOGGER.error("Fetch pool rejected our fetch list for " + ref);
finished(ref);
skipUrls(
urls,
UrlStatus.SKIPPED_DEFERRED,
String.format("Execution rejection skipped %d URLs", urls.size()));
}
// Adjust for how long it took to get the request queued.
adjustActive(ref, System.currentTimeMillis() - startTime);
}
} catch (InterruptedException e) {
LOGGER.warn("FetchBuffer interrupted!");
Thread.currentThread().interrupt();
}
}
// Skip all URLs that we've got left.
if (!values.isEmpty()) {
trace("Found unprocessed URLs");
UrlStatus status =
Thread.interrupted() ? UrlStatus.SKIPPED_INTERRUPTED : UrlStatus.SKIPPED_TIME_LIMIT;
while (!values.isEmpty()) {
FetchSetDatum datum = values.drain();
List<ScoredUrlDatum> urls = datum.getUrls();
trace(
"Skipping %d urls from %s (e.g. %s) ",
urls.size(), datum.getGroupingRef(), urls.get(0).getUrl());
skipUrls(urls, status, null);
}
}
}
