// @Transactional
public int importData(
final SyndFeed feed,
Set<KiWiUriResource> types,
Set<ContentItem> tags,
final User user,
final Collection<ContentItem> output) {
log.info(
"importing entries from #0 feed '#1' found at '#2'",
feed.getFeedType(),
feed.getTitle(),
feed.getUri());
if (types == null) {
types = new HashSet<KiWiUriResource>();
}
if (tags == null) {
tags = new HashSet<ContentItem>();
}
final Set<ContentItem> my_tags = tags;
final Set<KiWiUriResource> my_types = types;
// a hack for importing facebook activity streams: if the type is kiwi:FacebookPost,
// turn facebook activity stream mode on; in this mode, we will skip all entries where
// the remote author name and local user name are not identical
boolean facebookImport = false;
String t_facebookPost = Constants.NS_KIWI_CORE + "FacebookPost";
for (KiWiUriResource r : types) {
if (r.getUri().equals(t_facebookPost)) {
facebookImport = true;
break;
}
}
for (final SyndEntry entry : (List<SyndEntry>) feed.getEntries()) {
// facebook hack ... (see above)
if (facebookImport
&& !entry.getAuthor().equalsIgnoreCase(user.getFirstName() + " " + user.getLastName())) {
log.info("Facebook import: skipping friend post with title", entry.getTitle());
continue;
}
new RunAsOperation() {
@Override
public void execute() {
importEntry(feed, entry, my_types, my_tags, user, output);
}
}.addRole("admin").run();
}
// entityManager.flush();
log.info("#0 content items have been imported from RSS/Atom feed", feed.getEntries().size());
return feed.getEntries().size();
}
public void testReadAtom1() throws FeedException {
SyndFeedInput input = new SyndFeedInput();
SyndFeed result =
input.build(
new InputSource(getClass().getResourceAsStream("/be/hikage/xml/rome/atom_1.0.xml")));
assertEquals("RSS Veille Techno", result.getTitle());
assertEquals("http://svn.cyg.be/", result.getLink());
assertEquals(1, result.getEntries().size());
}
public static void main(String[] args) {
boolean ok = false;
if (args.length == 1) {
try {
URL feedUrl = new URL(args[0]);
FeedFetcherCache feedInfoCache = HashMapFeedInfoCache.getInstance();
FeedFetcher fetcher = new HttpURLFeedFetcher(feedInfoCache);
FetcherEventListenerImpl listener = new FetcherEventListenerImpl();
fetcher.addFetcherEventListener(listener);
System.err.println("Retrieving feed " + feedUrl);
// Retrieve the feed.
// We will get a Feed Polled Event and then a
// Feed Retrieved event (assuming the feed is valid)
SyndFeed feed = fetcher.retrieveFeed(feedUrl);
System.err.println(feedUrl + " retrieved");
System.err.println(
feedUrl
+ " has a title: "
+ feed.getTitle()
+ " and contains "
+ feed.getEntries().size()
+ " entries.");
// We will now retrieve the feed again. If the feed is unmodified
// and the server supports conditional gets, we will get a "Feed
// Unchanged" event after the Feed Polled event
System.err.println("Polling " + feedUrl + " again to test conditional get support.");
SyndFeed feed2 = fetcher.retrieveFeed(feedUrl);
System.err.println(
"If a \"Feed Unchanged\" event fired then the server supports conditional gets.");
ok = true;
} catch (Exception ex) {
System.out.println("ERROR: " + ex.getMessage());
ex.printStackTrace();
}
}
if (!ok) {
System.out.println();
System.out.println("FeedReader reads and prints any RSS/Atom feed type.");
System.out.println("The first parameter must be the URL of the feed to read.");
System.out.println();
}
}
protected WireFeed createRealFeed(String type, SyndFeed syndFeed) {
Channel channel = new Channel(type);
channel.setModules(ModuleUtils.cloneModules(syndFeed.getModules()));
channel.setEncoding(syndFeed.getEncoding());
channel.setTitle(syndFeed.getTitle());
channel.setLink(syndFeed.getLink());
channel.setDescription(syndFeed.getDescription());
SyndImage sImage = syndFeed.getImage();
if (sImage != null) {
channel.setImage(createRSSImage(sImage));
}
List sEntries = syndFeed.getEntries();
if (sEntries != null) {
channel.setItems(createRSSItems(sEntries));
}
return channel;
}
/**
* Generate an ebook from an RSS DOM Document.
*
* @param url The URL from where the Document was fetched (used only to set the author metadata)
* @param doc The DOM Document of the feed.
* @return An ebook.
* @throws IllegalArgumentException
* @throws FeedException
* @throws IOException
*/
private static Book createBookFromFeed(URL url, Document doc, List<Keyword> keywords)
throws IllegalArgumentException, FeedException, IOException {
Book book = new Book();
// start parsing our feed and have the above onItem methods called
SyndFeedInput input = new SyndFeedInput();
SyndFeed feed = input.build(doc);
System.out.println(feed);
// Set the title
book.getMetadata().addTitle(feed.getTitle());
// Add an Author
String author = feed.getAuthor();
if (author == null || "".equals(author.trim())) {
author = url.getHost();
}
book.getMetadata().addAuthor(new Author(author));
if (feed.getPublishedDate() != null) {
book.getMetadata().addDate(new nl.siegmann.epublib.domain.Date(feed.getPublishedDate()));
}
if (feed.getDescription() != null) {
book.getMetadata().addDescription(feed.getDescription());
}
if (feed.getCopyright() != null) {
book.getMetadata().getRights().add(feed.getCopyright());
}
// Set cover image - This has never worked.
// if (feed.getImage() != null) {
// System.out.println("There is an image for the feed");
// Promise<HttpResponse> futureImgResponse =
// WS.url(feed.getImage().getUrl()).getAsync();
// HttpResponse imgResponse = await(futureImgResponse);
// System.out.println("Content-type: " + imgResponse.getContentType());
// if (imgResponse.getContentType().startsWith("image/")) {
// String extension =
// imgResponse.getContentType().substring("image/".length());
// InputStream imageStream = imgResponse.getStream();
// book.getMetadata().setCoverImage(new Resource(imageStream, "cover." +
// extension));
// System.out.println("Using default cover");
// imageStream =
// VirtualFile.fromRelativePath("assets/cover.png").inputstream();
// if (imageStream != null) {
// System.out.println("Using default cover");
// book.getMetadata().setCoverImage(new Resource(imageStream,
// "cover.png"));
// } else {
// System.out.println("Could not load default cover");
// }
// }
// }
int entryNumber = 0;
List<SyndEntry> entries = feed.getEntries();
for (SyndEntry entry : entries) {
if (matchesKeyword(entry, keywords)) {
StringBuilder title = new StringBuilder(100);
if (entry.getTitle() != null) {
title.append(entry.getTitle());
}
if (entry.getAuthor() != null) {
title.append(" - ").append(entry.getAuthor());
}
StringBuilder content = new StringBuilder();
// Add title inside text
content.append("<h2>").append(title).append("</h2>");
if (entry.getDescription() != null) {
SyndContent syndContent = (SyndContent) entry.getDescription();
if (!syndContent.getType().contains("html")) {
content.append("<pre>\n");
}
content.append(syndContent.getValue());
if (!syndContent.getType().contains("html")) {
content.append("\n</pre>");
}
content.append("<hr/>");
}
if (entry.getContents().size() > 0) {
SyndContent syndContent = (SyndContent) entry.getContents().get(0);
if (!syndContent.getType().contains("html")) {
content.append("<pre>\n");
}
content.append(syndContent.getValue());
if (!syndContent.getType().contains("html")) {
content.append("\n</pre>");
}
}
String strContent = clean(content.toString());
// Add Chapter
try {
entryNumber++;
book.addSection(
title.toString(),
new Resource(new StringReader(strContent), "entry" + entryNumber + ".xhtml"));
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
}
return book;
}
/**
* Fills in the feed and entry-level metadata from DSpace objects.
*
* @param request request
* @param context context
* @param dso DSpaceObject
* @param items array of objects
* @param labels label map
*/
public void populate(
HttpServletRequest request,
Context context,
DSpaceObject dso,
List<? extends DSpaceObject> items,
Map<String, String> labels) {
String logoURL = null;
String objectURL = null;
String defaultTitle = null;
boolean podcastFeed = false;
this.request = request;
// dso is null for the whole site, or a search without scope
if (dso == null) {
defaultTitle = ConfigurationManager.getProperty("dspace.name");
feed.setDescription(localize(labels, MSG_FEED_DESCRIPTION));
objectURL = resolveURL(request, null);
logoURL = ConfigurationManager.getProperty("webui.feed.logo.url");
} else {
Bitstream logo = null;
if (dso.getType() == Constants.COLLECTION) {
Collection col = (Collection) dso;
defaultTitle = col.getName();
feed.setDescription(collectionService.getMetadata(col, "short_description"));
logo = col.getLogo();
String cols = ConfigurationManager.getProperty("webui.feed.podcast.collections");
if (cols != null && cols.length() > 1 && cols.contains(col.getHandle())) {
podcastFeed = true;
}
} else if (dso.getType() == Constants.COMMUNITY) {
Community comm = (Community) dso;
defaultTitle = comm.getName();
feed.setDescription(communityService.getMetadata(comm, "short_description"));
logo = comm.getLogo();
String comms = ConfigurationManager.getProperty("webui.feed.podcast.communities");
if (comms != null && comms.length() > 1 && comms.contains(comm.getHandle())) {
podcastFeed = true;
}
}
objectURL = resolveURL(request, dso);
if (logo != null) {
logoURL = urlOfBitstream(request, logo);
}
}
feed.setTitle(
labels.containsKey(MSG_FEED_TITLE) ? localize(labels, MSG_FEED_TITLE) : defaultTitle);
feed.setLink(objectURL);
feed.setPublishedDate(new Date());
feed.setUri(objectURL);
// add logo if we found one:
if (logoURL != null) {
// we use the path to the logo for this, the logo itself cannot
// be contained in the rdf. Not all RSS-viewers show this logo.
SyndImage image = new SyndImageImpl();
image.setLink(objectURL);
if (StringUtils.isNotBlank(feed.getTitle())) {
image.setTitle(feed.getTitle());
} else {
image.setTitle(localize(labels, MSG_LOGO_TITLE));
}
image.setUrl(logoURL);
feed.setImage(image);
}
// add entries for items
if (items != null) {
List<SyndEntry> entries = new ArrayList<SyndEntry>();
for (DSpaceObject itemDSO : items) {
if (itemDSO.getType() != Constants.ITEM) {
continue;
}
Item item = (Item) itemDSO;
boolean hasDate = false;
SyndEntry entry = new SyndEntryImpl();
entries.add(entry);
String entryURL = resolveURL(request, item);
entry.setLink(entryURL);
entry.setUri(entryURL);
String title = getOneDC(item, titleField);
entry.setTitle(title == null ? localize(labels, MSG_UNTITLED) : title);
// "published" date -- should be dc.date.issued
String pubDate = getOneDC(item, dateField);
if (pubDate != null) {
entry.setPublishedDate((new DCDate(pubDate)).toDate());
hasDate = true;
}
// date of last change to Item
entry.setUpdatedDate(item.getLastModified());
StringBuffer db = new StringBuffer();
for (String df : descriptionFields) {
// Special Case: "(date)" in field name means render as date
boolean isDate = df.indexOf("(date)") > 0;
if (isDate) {
df = df.replaceAll("\\(date\\)", "");
}
List<MetadataValue> dcv = itemService.getMetadataByMetadataString(item, df);
if (dcv.size() > 0) {
String fieldLabel = labels.get(MSG_METADATA + df);
if (fieldLabel != null && fieldLabel.length() > 0) {
db.append(fieldLabel).append(": ");
}
boolean first = true;
for (MetadataValue v : dcv) {
if (first) {
first = false;
} else {
db.append("; ");
}
db.append(isDate ? new DCDate(v.getValue()).toString() : v.getValue());
}
db.append("\n");
}
}
if (db.length() > 0) {
SyndContent desc = new SyndContentImpl();
desc.setType("text/plain");
desc.setValue(db.toString());
entry.setDescription(desc);
}
// This gets the authors into an ATOM feed
List<MetadataValue> authors = itemService.getMetadataByMetadataString(item, authorField);
if (authors.size() > 0) {
List<SyndPerson> creators = new ArrayList<SyndPerson>();
for (MetadataValue author : authors) {
SyndPerson sp = new SyndPersonImpl();
sp.setName(author.getValue());
creators.add(sp);
}
entry.setAuthors(creators);
}
// only add DC module if any DC fields are configured
if (dcCreatorField != null || dcDateField != null || dcDescriptionField != null) {
DCModule dc = new DCModuleImpl();
if (dcCreatorField != null) {
List<MetadataValue> dcAuthors =
itemService.getMetadataByMetadataString(item, dcCreatorField);
if (dcAuthors.size() > 0) {
List<String> creators = new ArrayList<String>();
for (MetadataValue author : dcAuthors) {
creators.add(author.getValue());
}
dc.setCreators(creators);
}
}
if (dcDateField != null && !hasDate) {
List<MetadataValue> v = itemService.getMetadataByMetadataString(item, dcDateField);
if (v.size() > 0) {
dc.setDate((new DCDate(v.get(0).getValue())).toDate());
}
}
if (dcDescriptionField != null) {
List<MetadataValue> v =
itemService.getMetadataByMetadataString(item, dcDescriptionField);
if (v.size() > 0) {
StringBuffer descs = new StringBuffer();
for (MetadataValue d : v) {
if (descs.length() > 0) {
descs.append("\n\n");
}
descs.append(d.getValue());
}
dc.setDescription(descs.toString());
}
}
entry.getModules().add(dc);
}
// iTunes Podcast Support - START
if (podcastFeed) {
// Add enclosure(s)
List<SyndEnclosure> enclosures = new ArrayList();
try {
List<Bundle> bunds = itemService.getBundles(item, "ORIGINAL");
if (bunds.get(0) != null) {
List<Bitstream> bits = bunds.get(0).getBitstreams();
for (Bitstream bit : bits) {
String mime = bit.getFormat(context).getMIMEType();
if (ArrayUtils.contains(podcastableMIMETypes, mime)) {
SyndEnclosure enc = new SyndEnclosureImpl();
enc.setType(bit.getFormat(context).getMIMEType());
enc.setLength(bit.getSize());
enc.setUrl(urlOfBitstream(request, bit));
enclosures.add(enc);
} else {
continue;
}
}
}
// Also try to add an external value from dc.identifier.other
// We are assuming that if this is set, then it is a media file
List<MetadataValue> externalMedia =
itemService.getMetadataByMetadataString(item, externalSourceField);
if (externalMedia.size() > 0) {
for (MetadataValue anExternalMedia : externalMedia) {
SyndEnclosure enc = new SyndEnclosureImpl();
enc.setType(
"audio/x-mpeg"); // We can't determine MIME of external file, so just picking
// one.
enc.setLength(1);
enc.setUrl(anExternalMedia.getValue());
enclosures.add(enc);
}
}
} catch (Exception e) {
System.out.println(e.getMessage());
}
entry.setEnclosures(enclosures);
// Get iTunes specific fields: author, subtitle, summary, duration, keywords
EntryInformation itunes = new EntryInformationImpl();
String author = getOneDC(item, authorField);
if (author != null && author.length() > 0) {
itunes.setAuthor(author); // <itunes:author>
}
itunes.setSubtitle(
title == null ? localize(labels, MSG_UNTITLED) : title); // <itunes:subtitle>
if (db.length() > 0) {
itunes.setSummary(db.toString()); // <itunes:summary>
}
String extent =
getOneDC(
item,
"dc.format.extent"); // assumed that user will enter this field with length of
// song in seconds
if (extent != null && extent.length() > 0) {
extent = extent.split(" ")[0];
Integer duration = Integer.parseInt(extent);
itunes.setDuration(new Duration(duration)); // <itunes:duration>
}
String subject = getOneDC(item, "dc.subject");
if (subject != null && subject.length() > 0) {
String[] subjects = new String[1];
subjects[0] = subject;
itunes.setKeywords(subjects); // <itunes:keywords>
}
entry.getModules().add(itunes);
}
}
feed.setEntries(entries);
}
}
/** @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;
}
