@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);
}
}
}
