/**
* Transforms a twitter4j.Status object into a JSONObject
*
* @param pStatus a twitter4j.Status object
* @return JSONObject
*/
@SuppressWarnings("unchecked")
public static JSONObject getJson(Status pStatus) {
JSONObject jsonObj = null;
if (pStatus != null) {
jsonObj = new JSONObject();
jsonObj.put("createdAt", pStatus.getCreatedAt());
jsonObj.put("id", pStatus.getId());
jsonObj.put("text", pStatus.getText());
jsonObj.put("source", pStatus.getSource());
jsonObj.put("isTruncated", pStatus.isTruncated());
jsonObj.put("inReplyToStatusId", pStatus.getInReplyToStatusId());
jsonObj.put("inReplyToUserId", pStatus.getInReplyToUserId());
jsonObj.put("isFavorited", pStatus.isFavorited());
jsonObj.put("isRetweeted", pStatus.isRetweeted());
jsonObj.put("favoriteCount", pStatus.getFavoriteCount());
jsonObj.put("inReplyToScreenName", pStatus.getInReplyToScreenName());
jsonObj.put("geoLocation", pStatus.getGeoLocation());
jsonObj.put("place", pStatus.getPlace());
jsonObj.put("retweetCount", pStatus.getRetweetCount());
jsonObj.put("isPossiblySensitive", pStatus.isPossiblySensitive());
jsonObj.put("isoLanguageCode", pStatus.getIsoLanguageCode());
jsonObj.put("contributorsIDs", pStatus.getContributors());
jsonObj.put("retweetedStatus", pStatus.getRetweetedStatus());
jsonObj.put("userMentionEntities", pStatus.getUserMentionEntities());
jsonObj.put("urlEntities", pStatus.getURLEntities());
jsonObj.put("hashtagEntities", pStatus.getHashtagEntities());
jsonObj.put("mediaEntities", pStatus.getMediaEntities());
jsonObj.put("currentUserRetweetId", pStatus.getCurrentUserRetweetId());
jsonObj.put("user", pStatus.getUser());
}
return jsonObj;
}
public void persistirTweets(
List<Status> tweets, String query, String polaridadIndicada, String codFinalidadData) {
logger.debug("<==== Inicio Method persistirTweet ====>");
StatusTweet dto = null;
HashTag ht = null;
Url url = null;
UserMention um = null;
Media md = null;
List<UserMention> ums = null;
List<HashTag> hts = null;
List<Url> urls = null;
List<Media> mds = null;
UserMentionEntity[] ume = null;
URLEntity[] urle = null;
HashtagEntity[] hte = null;
MediaEntity[] mes = null;
for (Status tweet : tweets) {
ums = null;
hts = null;
urls = null;
mds = null;
ume = null;
urle = null;
hte = null;
mes = null;
dto = null;
ht = null;
url = null;
um = null;
md = null;
logger.debug(
"tweet completo: {}",
ToStringBuilder.reflectionToString(tweet, ToStringStyle.MULTI_LINE_STYLE));
dto = new StatusTweet();
dto.setId(tweet.getId());
dto.setQuery(query);
dto.setMensaje(tweet.getText());
if (tweet.getUser() != null) {
dto.setScreenNameUser(tweet.getUser().getScreenName());
}
dto.setContadorRetweets((int) tweet.getRetweetCount());
dto.setCodOpinionInicial(polaridadIndicada);
dto.setFechaHoraRegistro(new Date());
dto.setFlagEnProceso(FLAG_POR_PROCESAR);
dto.setCodIdioma(COD_LENGUAJE_CONSULTA_ESPANOL);
dto.setCodFinalidadData(codFinalidadData);
if (!ArrayUtils.isEmpty(tweet.getUserMentionEntities())) {
ums = new ArrayList<UserMention>();
ume = tweet.getUserMentionEntities();
int size = ume.length;
for (int i = 0; i < size; i++) {
um = new UserMention();
um.setScreenName(ume[i].getScreenName());
um.setIdTweet(tweet.getId());
ums.add(um);
}
}
if (!ArrayUtils.isEmpty(tweet.getURLEntities())) {
urls = new ArrayList<Url>();
urle = tweet.getURLEntities();
int size = urle.length;
for (int i = 0; i < size; i++) {
url = new Url();
url.setIdTweet(tweet.getId());
url.setUrlCorta(urle[i].getURL());
url.setUrlLarga(urle[i].getExpandedURL());
url.setUrlMostrada(urle[i].getDisplayURL());
url.setFlagEnProceso(FLAG_POR_PROCESAR);
urls.add(url);
}
}
if (!ArrayUtils.isEmpty(tweet.getMediaEntities())) {
mds = new ArrayList<Media>();
mes = tweet.getMediaEntities();
int size = mes.length;
for (int i = 0; i < size; i++) {
md = new Media();
md.setIdTweet(tweet.getId());
md.setUrl(mes[i].getURL());
md.setMediaUrl(mes[i].getMediaURL());
md.setMediaUrlHttps(mes[i].getMediaURLHttps());
md.setExpandedUrl(mes[i].getExpandedURL());
mds.add(md);
}
}
if (!ArrayUtils.isEmpty(tweet.getHashtagEntities())) {
hts = new ArrayList<HashTag>();
hte = tweet.getHashtagEntities();
int size = hte.length;
for (int i = 0; i < size; i++) {
ht = new HashTag();
ht.setTextoTag(hte[i].getText());
ht.setIdTweet(tweet.getId());
hts.add(ht);
}
}
dto.setUrls(urls);
dto.setUsuariosMencionados(ums);
dto.setHashTags(hts);
dto.setMedias(mds);
persistirTweet(dto);
}
}
private static Record buildEntities(Schema schemaEntities, Status status) {
GenericRecordBuilder builderEntities = new GenericRecordBuilder(schemaEntities);
if (status.getHashtagEntities().length > 0) {
Schema schemaHashtagObject = schemaEntities.getField("hashtags").schema().getElementType();
List<GenericRecord> listHashtagObjects = new ArrayList<>();
for (HashtagEntity hashtagEntity : status.getHashtagEntities()) {
GenericRecordBuilder builderHashtagObject = new GenericRecordBuilder(schemaHashtagObject);
builderHashtagObject.set("text", hashtagEntity.getText());
builderHashtagObject.set("start", hashtagEntity.getStart());
builderHashtagObject.set("end", hashtagEntity.getEnd());
listHashtagObjects.add(builderHashtagObject.build());
}
builderEntities.set("hashtags", listHashtagObjects);
} else builderEntities.set("hashtags", Collections.emptyList());
if (status.getSymbolEntities().length > 0) {
Schema schemaSymbolObject = schemaEntities.getField("symbols").schema().getElementType();
List<GenericRecord> listSymbolObject = new ArrayList<>();
for (SymbolEntity symbolEntity : status.getSymbolEntities()) {
GenericRecordBuilder builderSymbolObject = new GenericRecordBuilder(schemaSymbolObject);
builderSymbolObject.set("text", symbolEntity.getText());
builderSymbolObject.set("start", symbolEntity.getStart());
builderSymbolObject.set("end", symbolEntity.getEnd());
listSymbolObject.add(builderSymbolObject.build());
}
builderEntities.set("symbols", listSymbolObject);
} else builderEntities.set("symbols", Collections.emptyList());
if (status.getMediaEntities().length > 0) {
Schema schemaMediaObject = schemaEntities.getField("media").schema().getElementType();
List<GenericRecord> listMediaObject = new ArrayList<>();
for (MediaEntity mediaEntity : status.getMediaEntities()) {
GenericRecordBuilder builderMediaObject = new GenericRecordBuilder(schemaMediaObject);
builderMediaObject.set("url", mediaEntity.getURL());
builderMediaObject.set("display_url", mediaEntity.getDisplayURL());
builderMediaObject.set("expanded_url", mediaEntity.getExpandedURL());
builderMediaObject.set("id", mediaEntity.getId());
builderMediaObject.set("media_url", mediaEntity.getMediaURL());
builderMediaObject.set("media_url_https", mediaEntity.getMediaURLHttps());
builderMediaObject.set("type", mediaEntity.getType());
builderMediaObject.set("text", mediaEntity.getText());
builderMediaObject.set("start", mediaEntity.getStart());
builderMediaObject.set("end", mediaEntity.getEnd());
Schema schemaSize = schemaMediaObject.getField("sizes").schema().getValueType();
GenericRecordBuilder builderSize = new GenericRecordBuilder(schemaSize);
Map<String, GenericRecord> mapSizes = new HashMap<>(4);
for (int key : mediaEntity.getSizes().keySet()) {
Size size = mediaEntity.getSizes().get(key);
builderSize.set("h", size.getHeight());
builderSize.set("w", size.getWidth());
builderSize.set("resize", size.getResize());
mapSizes.put(Integer.toString(key), builderSize.build());
}
builderMediaObject.set("sizes", mapSizes);
listMediaObject.add(builderMediaObject.build());
}
builderEntities.set("media", listMediaObject);
} else builderEntities.set("media", Collections.emptyList());
if (status.getURLEntities().length > 0) {
Schema schemaURLObject = schemaEntities.getField("urls").schema().getElementType();
List<GenericRecord> listURLObject1 = new ArrayList<>();
for (URLEntity urlEntity : status.getURLEntities())
listURLObject1.add(buildURLEntity(schemaURLObject, urlEntity));
builderEntities.set("urls", listURLObject1);
} else builderEntities.set("urls", Collections.emptyList());
if (status.getUserMentionEntities().length > 0) {
Schema schemaUserMentionObject =
schemaEntities.getField("user_mentions").schema().getElementType();
List<GenericRecord> listUserMentionObject = new ArrayList<>();
for (UserMentionEntity userMentionEntity : status.getUserMentionEntities()) {
GenericRecordBuilder builderUserMentionObject =
new GenericRecordBuilder(schemaUserMentionObject);
builderUserMentionObject.set("name", userMentionEntity.getName());
builderUserMentionObject.set("screen_name", userMentionEntity.getScreenName());
builderUserMentionObject.set("text", userMentionEntity.getText());
builderUserMentionObject.set("id", userMentionEntity.getId());
builderUserMentionObject.set("start", userMentionEntity.getStart());
builderUserMentionObject.set("end", userMentionEntity.getEnd());
listUserMentionObject.add(builderUserMentionObject.build());
}
builderEntities.set("user_mentions", listUserMentionObject);
} else builderEntities.set("user_mentions", Collections.emptyList());
if (status.getExtendedMediaEntities().length > 0) {
Schema schemaExtendedMediaObject =
schemaEntities.getField("extended_entities").schema().getElementType();
List<GenericRecord> listExtendedMediaObject = new ArrayList<>();
for (ExtendedMediaEntity extendedMediaEntity : status.getExtendedMediaEntities()) {
GenericRecordBuilder builderExtendedMediaObject =
new GenericRecordBuilder(schemaExtendedMediaObject);
builderExtendedMediaObject.set("url", extendedMediaEntity.getURL());
builderExtendedMediaObject.set("display_url", extendedMediaEntity.getDisplayURL());
builderExtendedMediaObject.set("expanded_url", extendedMediaEntity.getExpandedURL());
builderExtendedMediaObject.set("id", extendedMediaEntity.getId());
builderExtendedMediaObject.set("media_url", extendedMediaEntity.getMediaURL());
builderExtendedMediaObject.set("media_url_https", extendedMediaEntity.getMediaURLHttps());
builderExtendedMediaObject.set("type", extendedMediaEntity.getType());
builderExtendedMediaObject.set("text", extendedMediaEntity.getText());
builderExtendedMediaObject.set("start", extendedMediaEntity.getStart());
builderExtendedMediaObject.set("end", extendedMediaEntity.getEnd());
Schema schemaSize = schemaExtendedMediaObject.getField("sizes").schema().getValueType();
GenericRecordBuilder builderSize = new GenericRecordBuilder(schemaSize);
Map<String, GenericRecord> mapSizes = new HashMap<>(4);
for (int key : extendedMediaEntity.getSizes().keySet()) {
Size size = extendedMediaEntity.getSizes().get(key);
builderSize.set("h", size.getHeight());
builderSize.set("w", size.getWidth());
builderSize.set("resize", size.getResize());
mapSizes.put(Integer.toString(key), builderSize.build());
}
builderExtendedMediaObject.set("sizes", mapSizes);
Schema schemaVideoInfo = schemaExtendedMediaObject.getField("video_info").schema();
GenericRecordBuilder builderVideoInfo = new GenericRecordBuilder(schemaVideoInfo);
builderVideoInfo.set("h", extendedMediaEntity.getVideoAspectRatioHeight());
builderVideoInfo.set("w", extendedMediaEntity.getVideoAspectRatioWidth());
builderVideoInfo.set("duration_millis", extendedMediaEntity.getVideoDurationMillis());
Schema schemaVideoVariants = schemaVideoInfo.getField("variants").schema().getElementType();
List<GenericRecord> listVideoVariants = new ArrayList<>();
for (Variant extendedVideoVariant : extendedMediaEntity.getVideoVariants()) {
GenericRecordBuilder builderVideoVariant = new GenericRecordBuilder(schemaVideoVariants);
builderVideoVariant.set("bitrate", extendedVideoVariant.getBitrate());
builderVideoVariant.set("content_type", extendedVideoVariant.getContentType());
builderVideoVariant.set("url", extendedVideoVariant.getUrl());
listVideoVariants.add(builderVideoVariant.build());
}
builderVideoInfo.set("variants", listVideoVariants);
builderExtendedMediaObject.set("video_info", builderVideoInfo.build());
listExtendedMediaObject.add(builderExtendedMediaObject.build());
}
builderEntities.set("extended_entities", listExtendedMediaObject);
} else builderEntities.set("extended_entities", Collections.emptyList());
return builderEntities.build();
}
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;
}
private String getStatusText(Status status) {
return parseStatusText(status.getText(), status.getURLEntities(), status.getMediaEntities());
}