void emitLastRecord(Reporter reporter) throws IOException {
if (_flags != 0) {
if (_domainStats == null) {
reporter.incrCounter(Counters.EMITTING_URL_RECORD_WITH_NULL_DOMAINSTATS, 1);
} else {
reporter.incrCounter(Counters.EMITTING_URL_RECORD_WITH_DOMINSTATS, 1);
}
if (_crawlStatus != null) {
reporter.incrCounter(Counters.EMITTED_RECORD_HAD_CRAWLSTATUS, 1);
} else {
reporter.incrCounter(Counters.EMITTED_RECORD_HAD_NULL_CRAWLSTATUS, 1);
}
}
if (_contextURLBytes.getLength() >= 4097) {
reporter.incrCounter(Counters.SKIPPING_INVALID_LENGTH_URL, 1);
} else {
GoogleURL urlObject = new GoogleURL(_contextURLBytes.toString());
if (!skipRecord(urlObject, reporter)) {
if (urlObject.has_query()) {
reporter.incrCounter(Counters.LET_THROUGH_QUERY_URL, 1);
}
URLFPV2 fp = URLUtils.getURLFPV2FromURLObject(urlObject);
if (fp != null) {
if (_emittedURLSFilter.isPresent(fp)) {
reporter.incrCounter(Counters.SKIPPING_ALREADY_EMITTED_URL, 1);
} else {
_emittedURLSFilter.add(fp);
_emittedURLSInFilter++;
SegmentGeneratorItem itemValue = new SegmentGeneratorItem();
itemValue.setDomainFP(fp.getDomainHash());
itemValue.setRootDomainFP(fp.getRootDomainHash());
itemValue.setUrlFP(fp.getUrlHash());
itemValue.setUrl(urlObject.getCanonicalURL());
itemValue.setPageRank(0);
itemValue.setModifiedStatus((byte) 0);
items.add(itemValue);
if (items.size() >= SPILL_THRESHOLD) spillItems(reporter);
}
} else {
reporter.incrCounter(Counters.NULL_FP_FOR_URL, 1);
}
}
}
// reset stuff
_flags = 0;
_crawlStatus = null;
_contextURLBytes.clear();
_blogURLSkipFlag.set(true);
}
@Test
public void testSourceInputOutputWriters() throws IOException {
_sourceInputsBuffer = new DataOutputBuffer(16348 * 4);
_sourceInputsTrackingFilter = new URLFPBloomFilter(100000, NUM_HASH_FUNCTIONS, NUM_BITS);
String sourceDomainURL = "http://sourcedomain.com/foo";
URLFPV2 sourceFP = URLUtils.getURLFPV2FromCanonicalURL(sourceDomainURL);
String urls[] = {"http://somedomain.com/foo", "http://someother.com/bar"};
for (String url : urls) {
URLFPV2 fp = URLUtils.getURLFPV2FromCanonicalURL(url);
// double insert and validate actual single insertion
trackPotentialLinkSource(fp, url, sourceFP);
trackPotentialLinkSource(fp, url, sourceFP);
}
// validate data ...
TextBytes firstVersion = new TextBytes();
firstVersion.set(_sourceInputsBuffer.getData(), 0, _sourceInputsBuffer.getLength());
StringTokenizer tokenizer = new StringTokenizer(firstVersion.toString(), "\n");
int itemIndex = 0;
while (tokenizer.hasMoreElements()) {
String nextLine = tokenizer.nextToken();
String splits[] = nextLine.split("\t");
// validate fp
URLFPV2 fp = URLUtils.getURLFPV2FromCanonicalURL(urls[itemIndex]);
Assert.assertEquals(fp.getDomainHash(), Long.parseLong(splits[0]));
// validate actual url ...
Assert.assertEquals(splits[1], urls[itemIndex]);
itemIndex++;
}
// reset output buffer ...
_sourceInputsBuffer = new DataOutputBuffer(16348 * 4);
// and source bloom filter ...
_sourceInputsTrackingFilter = new URLFPBloomFilter(10000000, NUM_HASH_FUNCTIONS, NUM_BITS);
importLinkSourceData(sourceFP, firstVersion);
// second text should match first ..
TextBytes secondVersion = new TextBytes();
secondVersion.set(_sourceInputsBuffer.getData(), 0, _sourceInputsBuffer.getLength());
Assert.assertEquals(firstVersion, secondVersion);
}
/**
* update the model from the raw (generated tuples)
*
* @param tuple
* @throws Exception
*/
void updateModelFromInputTuple(Pair<TextBytes, TextBytes> tuple) throws Exception {
URLFPV2 fp = new URLFPV2();
// get key ...
fp.setRootDomainHash(
CrawlDBKey.getLongComponentFromKey(
tuple.e0, CrawlDBKey.ComponentId.ROOT_DOMAIN_HASH_COMPONENT_ID));
fp.setDomainHash(
CrawlDBKey.getLongComponentFromKey(
tuple.e0, CrawlDBKey.ComponentId.DOMAIN_HASH_COMPONENT_ID));
fp.setUrlHash(
CrawlDBKey.getLongComponentFromKey(
tuple.e0, CrawlDBKey.ComponentId.URL_HASH_COMPONENT_ID));
long recordType =
CrawlDBKey.getLongComponentFromKey(tuple.e0, CrawlDBKey.ComponentId.TYPE_COMPONENT_ID);
if (recordType == CrawlDBKey.Type.KEY_TYPE_CRAWL_STATUS.ordinal()
|| recordType == CrawlDBKey.Type.KEY_TYPE_HTML_LINK.ordinal()) {
// update model given key ...
URLStateModel urlModel = fpToModelMap.get(fp);
if (urlModel == null) {
urlModel = new URLStateModel();
urlModel.fp = fp;
fpToModelMap.put(fp, urlModel);
}
if (recordType == CrawlDBKey.Type.KEY_TYPE_CRAWL_STATUS.ordinal()) {
JsonObject redirectJSON = urlModel.updateModelGivenCrawlStatus(tuple.e1);
if (redirectJSON != null) {
URLFPV2 redirectFP =
URLUtils.getURLFPV2FromURL(redirectJSON.get("source_url").getAsString());
TextBytes key =
CrawlDBKey.generateKey(
redirectFP,
CrawlDBKey.Type.KEY_TYPE_CRAWL_STATUS,
redirectJSON.get("attempt_time").getAsLong());
Pair<TextBytes, TextBytes> redirectTuple =
new Pair<TextBytes, TextBytes>(key, new TextBytes(redirectJSON.toString()));
updateModelFromInputTuple(redirectTuple);
}
} else if (recordType == CrawlDBKey.Type.KEY_TYPE_HTML_LINK.ordinal()) {
urlModel.updateModelGivenLinkRecord(tuple.e1);
}
}
}
public void updateModelGivenLinkRecord(TextBytes linkJSON) {
JsonParser parser = new JsonParser();
JsonObject jsonObj = parser.parse(linkJSON.toString()).getAsJsonObject();
if (source_url == null) {
source_url = jsonObj.get("href").getAsString();
}
String sourceURL = jsonObj.get("source_url").getAsString();
URLFPV2 urlfp = URLUtils.getURLFPV2FromURL(sourceURL);
if (urlfp != null) {
if (urlfp.getRootDomainHash() != fp.getRootDomainHash()) {
if (!incoming.containsKey(urlfp.getRootDomainHash())) {
incoming.put(urlfp.getRootDomainHash(), sourceURL);
}
}
}
}
