public void onStatus(twitter4j.Status status) {
Logger.info(status.getUser().getName() + " : " + status.getText());
Tweet tweet = new Tweet(status);
tweet.conformsToTerms = checkMatch(status);
tweet.save();
if (tweet.conformsToTerms && esClient != null) {
String json = DataObjectFactory.getRawJSON(status);
json =
json.replaceAll(
"(\"geo\":\\{\"type\":\"Point\",\"coordinates\":)\\[([-0-9.,]*)\\]", "$1\"$2\"");
// Logger.debug("geo mangled json");
// Logger.debug(json);
IndexResponse response =
esClient.prepareIndex("twitter", "tweet").setSource(json).execute().actionGet();
}
}
protected boolean checkMatch(twitter4j.Status status) {
boolean result = false;
if (matchPattern.matcher(status.getText()).find()) result = true;
if (result) {
Logger.debug("Terms found in text");
Logger.debug(" \"" + status.getText() + "\"");
return result;
}
for (URLEntity ue : status.getURLEntities()) {
if (matchPattern.matcher(ue.getDisplayURL()).find()) result = true;
if (matchPattern.matcher(ue.getExpandedURL()).find()) result = true;
}
if (result) {
Logger.debug("Terms found in URL entities");
for (URLEntity ue : status.getURLEntities()) {
Logger.debug(" " + ue.getDisplayURL());
Logger.debug(" " + ue.getExpandedURL());
}
return result;
}
for (URLEntity ue : status.getMediaEntities()) {
if (matchPattern.matcher(ue.getDisplayURL()).find()) result = true;
if (matchPattern.matcher(ue.getExpandedURL()).find()) result = true;
}
if (result) {
Logger.debug("Terms found in Media entities");
for (URLEntity ue : status.getMediaEntities()) {
Logger.debug(" " + ue.getDisplayURL());
Logger.debug(" " + ue.getExpandedURL());
}
return result;
}
for (HashtagEntity he : status.getHashtagEntities()) {
if (matchPattern.matcher(he.getText()).find()) result = true;
}
if (result) {
Logger.debug("Terms found in Hashtag entities");
for (HashtagEntity he : status.getHashtagEntities()) {
Logger.debug(" " + he.getText());
}
return result;
}
for (UserMentionEntity me : status.getUserMentionEntities()) {
if (matchPattern.matcher(me.getScreenName()).find()) result = true;
}
if (result) {
Logger.debug("Terms found in User mention entities");
for (UserMentionEntity me : status.getUserMentionEntities()) {
Logger.debug(" " + me.getScreenName());
}
return result;
}
Logger.debug("Terms NOT FOUND");
Logger.debug(" Terms not found in URL entities");
for (URLEntity ue : status.getURLEntities()) {
Logger.debug(" " + ue.getDisplayURL());
Logger.debug(" " + ue.getExpandedURL());
}
Logger.debug(" Terms not found in Media entities");
for (URLEntity ue : status.getMediaEntities()) {
Logger.debug(" " + ue.getDisplayURL());
Logger.debug(" " + ue.getExpandedURL());
}
Logger.debug(" Terms not found in Hashtag entities");
for (HashtagEntity he : status.getHashtagEntities()) {
Logger.debug(" " + he.getText());
}
Logger.debug(" Terms not found in User mention entities");
for (UserMentionEntity me : status.getUserMentionEntities()) {
Logger.debug(" " + me.getScreenName());
}
return result;
}