List of usage examples for twitter4j TwitterStream addConnectionLifeCycleListener
TwitterStream addConnectionLifeCycleListener(ConnectionLifeCycleListener listener);
From source file:ar.com.zauber.commons.social.twitter.impl.streaming.TweetFetcher.java
License:Apache License
/** * Creates the {@link TwitterStream}//from ww w . j ava 2s . c om * * @param user * @param password * @return */ private TwitterStream createStream(final String user, final String password) { final TwitterStream stream = new TwitterStreamFactory().getInstance(new BasicAuthorization(user, password)); stream.addListener(new StatusListener() { @Override public void onStatus(final Status status) { try { T t = transformer.transform(status); closure.execute(t); } catch (Throwable ex) { logger.error("Exception en onStatus", ex); } } @Override public void onException(final Exception e) { logger.error("Exception on TwitterStream", e); } @Override public void onTrackLimitationNotice(final int numberOfLimitedStatuses) { logger.warn("onTrackLimitationNotice: Number of limited " + "statuses: {}", numberOfLimitedStatuses); } @Override public void onDeletionNotice(final StatusDeletionNotice statusDeletionNotice) { logger.warn("statusDeletionNotice: {}", statusDeletionNotice); } @Override public void onScrubGeo(final long userId, final long upToStatusId) { logger.warn("scrubGeo: {} {}", userId, upToStatusId); } }); stream.addConnectionLifeCycleListener(new ConnectionLifeCycleListener() { @Override public void onDisconnect() { logger.warn("Disconnected from Twitter!"); } @Override public void onConnect() { } @Override public void onCleanUp() { } }); return stream; }
From source file:druid.examples.twitter.TwitterSpritzerFirehoseFactory.java
License:Open Source License
@Override public Firehose connect() throws IOException { final ConnectionLifeCycleListener connectionLifeCycleListener = new ConnectionLifeCycleListener() { @Override// w w w . ja va 2 s . c o m public void onConnect() { log.info("Connected_to_Twitter"); } @Override public void onDisconnect() { log.info("Disconnect_from_Twitter"); } /** * called before thread gets cleaned up */ @Override public void onCleanUp() { log.info("Cleanup_twitter_stream"); } }; // ConnectionLifeCycleListener final TwitterStream twitterStream; final StatusListener statusListener; final int QUEUE_SIZE = 2000; /** This queue is used to move twitter events from the twitter4j thread to the druid ingest thread. */ final BlockingQueue<Status> queue = new ArrayBlockingQueue<Status>(QUEUE_SIZE); final LinkedList<String> dimensions = new LinkedList<String>(); final long startMsec = System.currentTimeMillis(); dimensions.add("htags"); dimensions.add("lang"); dimensions.add("utc_offset"); // // set up Twitter Spritzer // twitterStream = new TwitterStreamFactory().getInstance(); twitterStream.addConnectionLifeCycleListener(connectionLifeCycleListener); statusListener = new StatusListener() { // This is what really gets called to deliver stuff from twitter4j @Override public void onStatus(Status status) { // time to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } try { boolean success = queue.offer(status, 15L, TimeUnit.SECONDS); if (!success) { log.warn("queue too slow!"); } } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } @Override public void onDeletionNotice(StatusDeletionNotice statusDeletionNotice) { //log.info("Got a status deletion notice id:" + statusDeletionNotice.getStatusId()); } @Override public void onTrackLimitationNotice(int numberOfLimitedStatuses) { // This notice will be sent each time a limited stream becomes unlimited. // If this number is high and or rapidly increasing, it is an indication that your predicate is too broad, and you should consider a predicate with higher selectivity. log.warn("Got track limitation notice:" + numberOfLimitedStatuses); } @Override public void onScrubGeo(long userId, long upToStatusId) { //log.info("Got scrub_geo event userId:" + userId + " upToStatusId:" + upToStatusId); } @Override public void onException(Exception ex) { ex.printStackTrace(); } @Override public void onStallWarning(StallWarning warning) { System.out.println("Got stall warning:" + warning); } }; twitterStream.addListener(statusListener); twitterStream.sample(); // creates a generic StatusStream log.info("returned from sample()"); return new Firehose() { private final Runnable doNothingRunnable = new Runnable() { public void run() { } }; private long rowCount = 0L; private boolean waitIfmax = (maxEventCount < 0L); private final Map<String, Object> theMap = new HashMap<String, Object>(2); // DIY json parsing // private final ObjectMapper omapper = new ObjectMapper(); private boolean maxTimeReached() { if (maxRunMinutes <= 0) { return false; } else { return (System.currentTimeMillis() - startMsec) / 60000L >= maxRunMinutes; } } private boolean maxCountReached() { return maxEventCount >= 0 && rowCount >= maxEventCount; } @Override public boolean hasMore() { if (maxCountReached() || maxTimeReached()) { return waitIfmax; } else { return true; } } @Override public InputRow nextRow() { // Interrupted to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } // all done? if (maxCountReached() || maxTimeReached()) { if (waitIfmax) { // sleep a long time instead of terminating try { log.info("reached limit, sleeping a long time..."); sleep(2000000000L); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } else { // allow this event through, and the next hasMore() call will be false } } if (++rowCount % 1000 == 0) { log.info("nextRow() has returned %,d InputRows", rowCount); } Status status; try { status = queue.take(); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } HashtagEntity[] hts = status.getHashtagEntities(); if (hts != null && hts.length > 0) { List<String> hashTags = Lists.newArrayListWithExpectedSize(hts.length); for (HashtagEntity ht : hts) { hashTags.add(ht.getText()); } theMap.put("htags", Arrays.asList(hashTags.get(0))); } long retweetCount = status.getRetweetCount(); theMap.put("retweet_count", retweetCount); User user = status.getUser(); if (user != null) { theMap.put("follower_count", user.getFollowersCount()); theMap.put("friends_count", user.getFriendsCount()); theMap.put("lang", user.getLang()); theMap.put("utc_offset", user.getUtcOffset()); // resolution in seconds, -1 if not available? theMap.put("statuses_count", user.getStatusesCount()); } return new MapBasedInputRow(status.getCreatedAt().getTime(), dimensions, theMap); } @Override public Runnable commit() { // ephemera in, ephemera out. return doNothingRunnable; // reuse the same object each time } @Override public void close() throws IOException { log.info("CLOSE twitterstream"); twitterStream.shutdown(); // invokes twitterStream.cleanUp() } }; }
From source file:io.druid.examples.twitter.TwitterSpritzerFirehoseFactory.java
License:Apache License
@Override public Firehose connect(InputRowParser parser) throws IOException { final ConnectionLifeCycleListener connectionLifeCycleListener = new ConnectionLifeCycleListener() { @Override/* w w w . java2 s . c o m*/ public void onConnect() { log.info("Connected_to_Twitter"); } @Override public void onDisconnect() { log.info("Disconnect_from_Twitter"); } /** * called before thread gets cleaned up */ @Override public void onCleanUp() { log.info("Cleanup_twitter_stream"); } }; // ConnectionLifeCycleListener final TwitterStream twitterStream; final StatusListener statusListener; final int QUEUE_SIZE = 2000; /** This queue is used to move twitter events from the twitter4j thread to the druid ingest thread. */ final BlockingQueue<Status> queue = new ArrayBlockingQueue<Status>(QUEUE_SIZE); final long startMsec = System.currentTimeMillis(); // // set up Twitter Spritzer // twitterStream = new TwitterStreamFactory().getInstance(); twitterStream.addConnectionLifeCycleListener(connectionLifeCycleListener); statusListener = new StatusListener() { // This is what really gets called to deliver stuff from twitter4j @Override public void onStatus(Status status) { // time to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } try { boolean success = queue.offer(status, 15L, TimeUnit.SECONDS); if (!success) { log.warn("queue too slow!"); } } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } @Override public void onDeletionNotice(StatusDeletionNotice statusDeletionNotice) { //log.info("Got a status deletion notice id:" + statusDeletionNotice.getStatusId()); } @Override public void onTrackLimitationNotice(int numberOfLimitedStatuses) { // This notice will be sent each time a limited stream becomes unlimited. // If this number is high and or rapidly increasing, it is an indication that your predicate is too broad, and you should consider a predicate with higher selectivity. log.warn("Got track limitation notice:" + numberOfLimitedStatuses); } @Override public void onScrubGeo(long userId, long upToStatusId) { //log.info("Got scrub_geo event userId:" + userId + " upToStatusId:" + upToStatusId); } @Override public void onException(Exception ex) { ex.printStackTrace(); } @Override public void onStallWarning(StallWarning warning) { System.out.println("Got stall warning:" + warning); } }; twitterStream.addListener(statusListener); twitterStream.sample(); // creates a generic StatusStream log.info("returned from sample()"); return new Firehose() { private final Runnable doNothingRunnable = new Runnable() { public void run() { } }; private long rowCount = 0L; private boolean waitIfmax = (getMaxEventCount() < 0L); private final Map<String, Object> theMap = new TreeMap<>(); // DIY json parsing // private final ObjectMapper omapper = new ObjectMapper(); private boolean maxTimeReached() { if (getMaxRunMinutes() <= 0) { return false; } else { return (System.currentTimeMillis() - startMsec) / 60000L >= getMaxRunMinutes(); } } private boolean maxCountReached() { return getMaxEventCount() >= 0 && rowCount >= getMaxEventCount(); } @Override public boolean hasMore() { if (maxCountReached() || maxTimeReached()) { return waitIfmax; } else { return true; } } @Override public InputRow nextRow() { // Interrupted to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } // all done? if (maxCountReached() || maxTimeReached()) { if (waitIfmax) { // sleep a long time instead of terminating try { log.info("reached limit, sleeping a long time..."); sleep(2000000000L); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } else { // allow this event through, and the next hasMore() call will be false } } if (++rowCount % 1000 == 0) { log.info("nextRow() has returned %,d InputRows", rowCount); } Status status; try { status = queue.take(); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } theMap.clear(); HashtagEntity[] hts = status.getHashtagEntities(); String text = status.getText(); theMap.put("text", (null == text) ? "" : text); theMap.put("htags", (hts.length > 0) ? Lists.transform(Arrays.asList(hts), new Function<HashtagEntity, String>() { @Nullable @Override public String apply(HashtagEntity input) { return input.getText(); } }) : ImmutableList.<String>of()); long[] lcontrobutors = status.getContributors(); List<String> contributors = new ArrayList<>(); for (long contrib : lcontrobutors) { contributors.add(String.format("%d", contrib)); } theMap.put("contributors", contributors); GeoLocation geoLocation = status.getGeoLocation(); if (null != geoLocation) { double lat = status.getGeoLocation().getLatitude(); double lon = status.getGeoLocation().getLongitude(); theMap.put("lat", lat); theMap.put("lon", lon); } else { theMap.put("lat", null); theMap.put("lon", null); } if (status.getSource() != null) { Matcher m = sourcePattern.matcher(status.getSource()); theMap.put("source", m.find() ? m.group(1) : status.getSource()); } theMap.put("retweet", status.isRetweet()); if (status.isRetweet()) { Status original = status.getRetweetedStatus(); theMap.put("retweet_count", original.getRetweetCount()); User originator = original.getUser(); theMap.put("originator_screen_name", originator != null ? originator.getScreenName() : ""); theMap.put("originator_follower_count", originator != null ? originator.getFollowersCount() : ""); theMap.put("originator_friends_count", originator != null ? originator.getFriendsCount() : ""); theMap.put("originator_verified", originator != null ? originator.isVerified() : ""); } User user = status.getUser(); final boolean hasUser = (null != user); theMap.put("follower_count", hasUser ? user.getFollowersCount() : 0); theMap.put("friends_count", hasUser ? user.getFriendsCount() : 0); theMap.put("lang", hasUser ? user.getLang() : ""); theMap.put("utc_offset", hasUser ? user.getUtcOffset() : -1); // resolution in seconds, -1 if not available? theMap.put("statuses_count", hasUser ? user.getStatusesCount() : 0); theMap.put("user_id", hasUser ? String.format("%d", user.getId()) : ""); theMap.put("screen_name", hasUser ? user.getScreenName() : ""); theMap.put("location", hasUser ? user.getLocation() : ""); theMap.put("verified", hasUser ? user.isVerified() : ""); theMap.put("ts", status.getCreatedAt().getTime()); List<String> dimensions = Lists.newArrayList(theMap.keySet()); return new MapBasedInputRow(status.getCreatedAt().getTime(), dimensions, theMap); } @Override public Runnable commit() { // ephemera in, ephemera out. return doNothingRunnable; // reuse the same object each time } @Override public void close() throws IOException { log.info("CLOSE twitterstream"); twitterStream.shutdown(); // invokes twitterStream.cleanUp() } }; }
From source file:org.apache.druid.examples.twitter.TwitterSpritzerFirehoseFactory.java
License:Apache License
@Override public Firehose connect(InputRowParser parser, File temporaryDirectory) { final ConnectionLifeCycleListener connectionLifeCycleListener = new ConnectionLifeCycleListener() { @Override/* w w w. j ava 2s .c om*/ public void onConnect() { log.info("Connected_to_Twitter"); } @Override public void onDisconnect() { log.info("Disconnect_from_Twitter"); } /** * called before thread gets cleaned up */ @Override public void onCleanUp() { log.info("Cleanup_twitter_stream"); } }; // ConnectionLifeCycleListener final TwitterStream twitterStream; final StatusListener statusListener; final int QUEUE_SIZE = 2000; /** This queue is used to move twitter events from the twitter4j thread to the druid ingest thread. */ final BlockingQueue<Status> queue = new ArrayBlockingQueue<Status>(QUEUE_SIZE); final long startMsec = System.currentTimeMillis(); // // set up Twitter Spritzer // twitterStream = new TwitterStreamFactory().getInstance(); twitterStream.addConnectionLifeCycleListener(connectionLifeCycleListener); statusListener = new StatusListener() { // This is what really gets called to deliver stuff from twitter4j @Override public void onStatus(Status status) { // time to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } try { boolean success = queue.offer(status, 15L, TimeUnit.SECONDS); if (!success) { log.warn("queue too slow!"); } } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } @Override public void onDeletionNotice(StatusDeletionNotice statusDeletionNotice) { //log.info("Got a status deletion notice id:" + statusDeletionNotice.getStatusId()); } @Override public void onTrackLimitationNotice(int numberOfLimitedStatuses) { // This notice will be sent each time a limited stream becomes unlimited. // If this number is high and or rapidly increasing, it is an indication that your predicate is too broad, and you should consider a predicate with higher selectivity. log.warn("Got track limitation notice:" + numberOfLimitedStatuses); } @Override public void onScrubGeo(long userId, long upToStatusId) { //log.info("Got scrub_geo event userId:" + userId + " upToStatusId:" + upToStatusId); } @Override public void onException(Exception ex) { log.error(ex, "Got exception"); } @Override public void onStallWarning(StallWarning warning) { log.warn("Got stall warning: %s", warning); } }; twitterStream.addListener(statusListener); twitterStream.sample(); // creates a generic StatusStream log.info("returned from sample()"); return new Firehose() { private final Runnable doNothingRunnable = new Runnable() { @Override public void run() { } }; private long rowCount = 0L; private boolean waitIfmax = (getMaxEventCount() < 0L); private final Map<String, Object> theMap = new TreeMap<>(); // DIY json parsing // private final ObjectMapper omapper = new ObjectMapper(); private boolean maxTimeReached() { if (getMaxRunMinutes() <= 0) { return false; } else { return (System.currentTimeMillis() - startMsec) / 60000L >= getMaxRunMinutes(); } } private boolean maxCountReached() { return getMaxEventCount() >= 0 && rowCount >= getMaxEventCount(); } @Override public boolean hasMore() { if (maxCountReached() || maxTimeReached()) { return waitIfmax; } else { return true; } } @Nullable @Override public InputRow nextRow() { // Interrupted to stop? if (Thread.currentThread().isInterrupted()) { throw new RuntimeException("Interrupted, time to stop"); } // all done? if (maxCountReached() || maxTimeReached()) { if (waitIfmax) { // sleep a long time instead of terminating try { log.info("reached limit, sleeping a long time..."); Thread.sleep(2000000000L); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } } else { // allow this event through, and the next hasMore() call will be false } } if (++rowCount % 1000 == 0) { log.info("nextRow() has returned %,d InputRows", rowCount); } Status status; try { status = queue.take(); } catch (InterruptedException e) { throw new RuntimeException("InterruptedException", e); } theMap.clear(); HashtagEntity[] hts = status.getHashtagEntities(); String text = status.getText(); theMap.put("text", (null == text) ? "" : text); theMap.put("htags", (hts.length > 0) ? Lists.transform(Arrays.asList(hts), new Function<HashtagEntity, String>() { @Nullable @Override public String apply(HashtagEntity input) { return input.getText(); } }) : ImmutableList.<String>of()); long[] lcontrobutors = status.getContributors(); List<String> contributors = new ArrayList<>(); for (long contrib : lcontrobutors) { contributors.add(StringUtils.format("%d", contrib)); } theMap.put("contributors", contributors); GeoLocation geoLocation = status.getGeoLocation(); if (null != geoLocation) { double lat = status.getGeoLocation().getLatitude(); double lon = status.getGeoLocation().getLongitude(); theMap.put("lat", lat); theMap.put("lon", lon); } else { theMap.put("lat", null); theMap.put("lon", null); } if (status.getSource() != null) { Matcher m = sourcePattern.matcher(status.getSource()); theMap.put("source", m.find() ? m.group(1) : status.getSource()); } theMap.put("retweet", status.isRetweet()); if (status.isRetweet()) { Status original = status.getRetweetedStatus(); theMap.put("retweet_count", original.getRetweetCount()); User originator = original.getUser(); theMap.put("originator_screen_name", originator != null ? originator.getScreenName() : ""); theMap.put("originator_follower_count", originator != null ? originator.getFollowersCount() : ""); theMap.put("originator_friends_count", originator != null ? originator.getFriendsCount() : ""); theMap.put("originator_verified", originator != null ? originator.isVerified() : ""); } User user = status.getUser(); final boolean hasUser = (null != user); theMap.put("follower_count", hasUser ? user.getFollowersCount() : 0); theMap.put("friends_count", hasUser ? user.getFriendsCount() : 0); theMap.put("lang", hasUser ? user.getLang() : ""); theMap.put("utc_offset", hasUser ? user.getUtcOffset() : -1); // resolution in seconds, -1 if not available? theMap.put("statuses_count", hasUser ? user.getStatusesCount() : 0); theMap.put("user_id", hasUser ? StringUtils.format("%d", user.getId()) : ""); theMap.put("screen_name", hasUser ? user.getScreenName() : ""); theMap.put("location", hasUser ? user.getLocation() : ""); theMap.put("verified", hasUser ? user.isVerified() : ""); theMap.put("ts", status.getCreatedAt().getTime()); List<String> dimensions = Lists.newArrayList(theMap.keySet()); return new MapBasedInputRow(status.getCreatedAt().getTime(), dimensions, theMap); } @Override public Runnable commit() { // ephemera in, ephemera out. return doNothingRunnable; // reuse the same object each time } @Override public void close() { log.info("CLOSE twitterstream"); twitterStream.shutdown(); // invokes twitterStream.cleanUp() } }; }