private static Record buildURLEntity(Schema schemaURLObject, URLEntity urlEntity) {
GenericRecordBuilder builderURLObject = new GenericRecordBuilder(schemaURLObject);
builderURLObject.set("url", urlEntity.getURL());
builderURLObject.set("text", urlEntity.getText());
builderURLObject.set("expanded_url", urlEntity.getExpandedURL());
builderURLObject.set("display_url", urlEntity.getDisplayURL());
builderURLObject.set("start", urlEntity.getStart());
builderURLObject.set("end", urlEntity.getEnd());
return builderURLObject.build();
}
/**
* Does the passed status match the filter pattern? This method checks the status text and the
* expanded url entities
*
* @param status Status to check
* @return True if the filter expression matches, false otherwise
*/
private boolean matchesFilter(Status status) {
boolean shouldFilter = false;
if (filterPattern != null) {
Matcher m = filterPattern.matcher(status.getText());
if (m.matches()) shouldFilter = true;
if (status.getURLEntities() != null) {
for (URLEntity ue : status.getURLEntities()) {
URL expUrl = ue.getExpandedURL();
if (expUrl != null && expUrl.toString() != null) {
m = filterPattern.matcher(expUrl.toString());
if (m.matches()) shouldFilter = true;
}
}
}
}
return shouldFilter;
}
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;
}