/** @inheritDoc */
public Subscription fetchSubscription(String feedURL, Date lastModified) throws FetcherException {
if (feedURL == null) {
throw new IllegalArgumentException("feed url cannot be null");
}
// setup Rome feed fetcher
FeedFetcher feedFetcher = getRomeFetcher();
// fetch the feed
log.debug("Fetching feed: " + feedURL);
SyndFeed feed;
try {
feed = feedFetcher.retrieveFeed(new URL(feedURL));
} catch (Exception ex) {
throw new FetcherException("Error fetching subscription - " + feedURL, ex);
}
log.debug("Feed pulled, extracting data into Subscription");
// build planet subscription from fetched feed
Subscription newSub = new Subscription();
newSub.setFeedURL(feedURL);
newSub.setSiteURL(feed.getLink());
newSub.setTitle(feed.getTitle());
newSub.setAuthor(feed.getAuthor());
newSub.setLastUpdated(feed.getPublishedDate());
// normalize any data that couldn't be properly extracted
if (newSub.getSiteURL() == null) {
// set the site url to the feed url then
newSub.setSiteURL(newSub.getFeedURL());
}
if (newSub.getAuthor() == null) {
// set the author to the title
newSub.setAuthor(newSub.getTitle());
}
if (newSub.getLastUpdated() == null) {
// no update time specified in feed, so try consulting feed info cache
FeedFetcherCache feedCache = getRomeFetcherCache();
try {
SyndFeedInfo feedInfo = feedCache.getFeedInfo(new URL(newSub.getFeedURL()));
if (feedInfo.getLastModified() != null) {
long lastUpdatedLong = ((Long) feedInfo.getLastModified()).longValue();
if (lastUpdatedLong != 0) {
newSub.setLastUpdated(new Date(lastUpdatedLong));
}
}
} catch (MalformedURLException ex) {
// should never happen since we check this above
}
}
// check if feed is unchanged and bail now if so
if (lastModified != null
&& newSub.getLastUpdated() != null
&& !newSub.getLastUpdated().after(lastModified)) {
return null;
}
if (log.isDebugEnabled()) {
log.debug("Subscription is: " + newSub.toString());
}
// some kludge to deal with feeds w/ no entry dates
// we assign arbitrary dates chronologically by entry starting either
// from the current time or the last update time of the subscription
Calendar cal = Calendar.getInstance();
if (newSub.getLastUpdated() != null) {
cal.setTime(newSub.getLastUpdated());
} else {
cal.setTime(new Date());
cal.add(Calendar.DATE, -1);
}
// add entries
List<SyndEntry> feedEntries = feed.getEntries();
for (SyndEntry feedEntry : feedEntries) {
SubscriptionEntry newEntry = buildEntry(feedEntry);
// some kludge to handle feeds with no entry dates
if (newEntry.getPubTime() == null) {
log.debug("No published date, assigning fake date for " + feedURL);
newEntry.setPubTime(new Timestamp(cal.getTimeInMillis()));
cal.add(Calendar.DATE, -1);
}
if (newEntry != null) {
newSub.addEntry(newEntry);
}
}
log.debug(feedEntries.size() + " entries included");
return newSub;
}
