@Override
@SuppressWarnings({"rawtypes", "unchecked"})
public void prepare(Map stormConf, TopologyContext context, OutputCollector collect) {
collector = collect;
metadataTransfer = MetadataTransfer.getInstance(stormConf);
sniffWhenNoMDKey = ConfUtils.getBoolean(stormConf, "feed.sniffContent", false);
filterHoursSincePub = ConfUtils.getInt(stormConf, "feed.filter.hours.since.published", -1);
urlFilters = URLFilters.fromConf(stormConf);
parseFilters = ParseFilters.fromConf(stormConf);
}
private List<Outlink> parseFeed(String url, byte[] content, Metadata parentMetadata)
throws MalformedURLException {
List<Outlink> links = new ArrayList<>();
SyndFeed feed = null;
try (ByteArrayInputStream is = new ByteArrayInputStream(content)) {
SyndFeedInput input = new SyndFeedInput();
feed = input.build(new InputSource(is));
} catch (Exception e) {
LOG.error("Exception parsing feed from DOM {}", url);
return links;
}
URL sURL = new URL(url);
List<SyndEntry> entries = feed.getEntries();
for (SyndEntry entry : entries) {
String targetURL = entry.getLink();
// build an absolute URL
try {
targetURL = URLUtil.resolveURL(sURL, targetURL).toExternalForm();
} catch (MalformedURLException e) {
LOG.debug("MalformedURLException on {}", targetURL);
continue;
}
targetURL = urlFilters.filter(sURL, parentMetadata, targetURL);
if (StringUtils.isBlank(targetURL)) continue;
Outlink newLink = new Outlink(targetURL);
Metadata targetMD = metadataTransfer.getMetaForOutlink(targetURL, url, parentMetadata);
newLink.setMetadata(targetMD);
String title = entry.getTitle();
if (StringUtils.isNotBlank(title)) {
targetMD.setValue("feed.title", title.trim());
}
Date publishedDate = entry.getPublishedDate();
if (publishedDate != null) {
// filter based on the published date
if (filterHoursSincePub != -1) {
Calendar rightNow = Calendar.getInstance();
rightNow.add(Calendar.HOUR, -filterHoursSincePub);
if (publishedDate.before(rightNow.getTime())) {
LOG.info(
"{} has a published date {} which is more than {} hours old",
targetURL,
publishedDate.toString(),
filterHoursSincePub);
continue;
}
}
targetMD.setValue("feed.publishedDate", publishedDate.toString());
}
SyndContent description = entry.getDescription();
if (description != null && StringUtils.isNotBlank(description.getValue())) {
targetMD.setValue("feed.description", description.getValue());
}
links.add(newLink);
}
return links;
}