private static List<Triple> generateTriples_ResourceVenue_upcoming(
URI uri, List<Triple> url_triples) throws URISyntaxException {
List<Triple> valid_triples = new ArrayList<Triple>();
valid_triples.add(new Triple(uri.toString(), "wam:domain", "upcoming.yahoo.com"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Resource"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Venue"));
valid_triples.add(new Triple(uri.toString(), "wam:URI", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:url", uri.toString()));
String title = "", locality = "";
if (url_triples.isEmpty()) {
/*
* GET MORE INFO WITH ANY23
*/
String n3s = MetadataEvalFunc.exec(uri.toString());
url_triples = new TripleNxParser().parse(n3s);
logger.info(" GET MORE INFO for Venue : " + uri.toString());
}
for (Triple triple : url_triples) {
if (triple.predicate.equals(Constants.NS_RDF + "type")
|| triple.predicate.equals(Constants.NS_VCARD + "latitude")
|| triple.predicate.equals(Constants.NS_VCARD + "longitude")
|| triple.predicate.equals(Constants.NS_VCARD + "fn")
|| triple.predicate.equals(Constants.NS_VCARD + "street-address")
|| triple.predicate.equals(Constants.NS_VCARD + "region")
|| triple.predicate.equals(Constants.NS_VCARD + "postal-code")) {
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
if ((triple.subject.equals(uri.toString())
&& triple.predicate.equals(Constants.NS_DC + "title"))) {
title = triple.object;
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
if (triple.predicate.equals(Constants.NS_VCARD + "locality")) {
locality = triple.object;
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
}
List<Triple> final_urltriples = new ArrayList<Triple>();
URI duplicate_uri = getDuplicate(uri, "wam:Venue", title, locality);
if (duplicate_uri != null) {
final_urltriples.add(new Triple(uri.toString(), "wam:hasResource", duplicate_uri.toString()));
} else {
if (!title.isEmpty()) {
final_urltriples.addAll(valid_triples);
}
}
return final_urltriples;
}
private static List<Triple> generateTriples_ResourceVideo(
URI uri, String title, String performer_name, String domain) {
logger.info(" TRIPLES VIDEOS: " + uri.toString() + ", videoTitle:" + title);
List<Triple> valid_triples = new ArrayList<Triple>();
valid_triples.add(new Triple(uri.toString(), "wam:domain", domain));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Resource"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Video"));
valid_triples.add(new Triple(uri.toString(), "wam:URI", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:url", uri.toString()));
valid_triples.add(new Triple(uri.toString(), Constants.NS_DC + "title", title));
if (!performer_name.isEmpty()) {
valid_triples.add(new Triple(uri.toString(), "wam:performerName", performer_name));
}
try {
String n3s = MetadataEvalFunc.exec(uri.toString());
List<Triple> url_triples = new TripleNxParser().parse(n3s);
if (!url_triples.isEmpty()) {
logger.info(" GET MORE INFO WITH ANY23 for VIDEO : " + uri.toString());
for (Triple triple : url_triples) {
if (triple.predicate.equals(Constants.NS_XHTML + "keywords")
|| triple.predicate.equals(Constants.NS_XHTML + "description")) {
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
}
}
} catch (Exception e) {
e.printStackTrace();
}
List<Triple> final_urltriples = new ArrayList<Triple>();
URI duplicate_uri = getDuplicate(uri, "wam:Video", title, "");
if (duplicate_uri != null) {
final_urltriples.add(new Triple(uri.toString(), "wam:hasResource", duplicate_uri.toString()));
} else {
if (!title.isEmpty()) {
final_urltriples.addAll(valid_triples);
}
}
return final_urltriples;
}
private static List<Triple> generateTriples_ResourcePerformer_eventful(
URI uri, List<Triple> url_triples) throws URISyntaxException {
List<Triple> valid_triples = new ArrayList<Triple>();
valid_triples.add(new Triple(uri.toString(), "wam:domain", "eventful.com"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Resource"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Performer"));
valid_triples.add(new Triple(uri.toString(), "wam:URI", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:url", uri.toString()));
String title = "";
if (url_triples.isEmpty()) {
/*
* GET MORE INFO WITH ANY23
*/
logger.info(" GET MORE INFO WITH ANY23: " + uri.toString());
String n3s = MetadataEvalFunc.exec(uri.toString());
url_triples = new TripleNxParser().parse(n3s);
}
for (Triple triple : url_triples) {
if (triple.predicate.equals(Constants.NS_OG + "type")
|| (triple.subject.equals(uri.toString())
&& triple.predicate.equals(Constants.NS_OG + "url"))) {
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
if (triple.predicate.equals(Constants.NS_OG + "title")) {
title = triple.object;
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
}
List<Triple> final_urltriples = new ArrayList<Triple>();
URI duplicate_uri = getDuplicate(uri, "wam:Performer", title, "");
if (duplicate_uri != null) {
final_urltriples.add(new Triple(uri.toString(), "wam:hasResource", duplicate_uri.toString()));
} else {
if (!title.isEmpty()) {
final_urltriples.addAll(valid_triples);
}
}
return final_urltriples;
}
private static List<Triple> generateTriples_ResourceEvent_upcoming(
URI uri, List<Triple> url_triples) {
List<Triple> valid_triples = new ArrayList<Triple>();
valid_triples.add(new Triple(uri.toString(), "wam:URI", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:url", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Resource"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Event"));
valid_triples.add(new Triple(uri.toString(), "wam:domain", "upcoming.yahoo.com"));
String title = "", locality = "", venue = "";
for (Triple triple : url_triples) {
if ((triple.subject.equals(uri.toString())
&& triple.predicate.equals(Constants.NS_RDF + "type"))
|| triple.predicate.equals(Constants.NS_ICAL + "summary")
|| triple.predicate.equals(Constants.NS_ICAL + "location")
|| triple.predicate.equals(Constants.NS_ICAL + "dtstart")
|| triple.predicate.equals(Constants.NS_ICAL + "categories")
|| triple.predicate.equals(Constants.NS_OG + "region")
|| triple.predicate.equals(Constants.NS_OG + "postal-code")) {
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
if ((triple.subject.equals(uri.toString())
&& triple.predicate.equals(Constants.NS_DC + "title"))) {
title = triple.object;
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
if (triple.predicate.equals(Constants.NS_OG + "locality")) {
locality = triple.object;
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
/*
* Get Venue
*/
if (triple.predicate.equals(Constants.NS_VCARD + "urlofvenue") && venue.isEmpty()) {
venue = triple.object;
try {
/*
* Check first for duplicate resources
*/
URI uri_venue = new URI(venue);
valid_triples.add(new Triple(uri.toString(), "wam:hasVenue", uri_venue.toString()));
valid_triples.addAll(
generateTriples_ResourceVenue_upcoming(uri_venue, new ArrayList<Triple>()));
} catch (URISyntaxException e) {
e.printStackTrace();
}
}
}
/*
* Get Video of event (when no performer given)
* First, GET Performer-URLS BY HTML-EXTRACTION
*/
MetadataEvalFunc extractor = new MetadataEvalFunc();
Map<String, String> performers = new HashMap<String, String>();
logger.info(" GET URLS BY HTML-EXTRACTION: " + uri.toString());
performers = extractor.scrapPage(uri.toString(), "upcoming.yahoo.com");
String performer_name = "";
String video_text_query = "";
for (String performer_url : performers.keySet()) {
logger.info("performer_url: " + performer_url);
try {
URI uri_performer = new URI(performer_url);
performer_name = performers.get(performer_url);
valid_triples.add(new Triple(uri.toString(), "wam:hasPerformer", uri_performer.toString()));
valid_triples.addAll(
generateTriples_ResourcePerformer_eventful(uri_performer, new ArrayList<Triple>()));
/*
* Get Video of Performer
*/
video_text_query = performer_name;
// logger.info(" GET VIDEOS PERFORMER WITH API: '" + performer_name+"'" +",
// "+video_text_query);
// YouTubeManager youtube_manager = new YouTubeManager();
// List<YouTubeVideo> videos = youtube_manager.retrieveVideos(video_text_query,
// video_max_results, true, youtube_timeout);
// for(YouTubeVideo video: videos){
// URI uri_video = new URI(video.getWebPlayerUrl());
// valid_triples.add(new Triple(uri_performer.toString(), "wam:hasVideo",
// uri_video.toString()));
// valid_triples.addAll(generateTriples_ResourceVideo(uri_video, video.getTitle(),
// performer_name, "upcoming.yahoo.com"));
//
// }
} catch (URISyntaxException e) {
e.printStackTrace();
}
}
List<Triple> final_urltriples = new ArrayList<Triple>();
URI duplicate_uri = getDuplicate(uri, "wam:Event", title, locality);
if (duplicate_uri != null) {
final_urltriples.add(new Triple(uri.toString(), "wam:hasResource", duplicate_uri.toString()));
} else {
if (!title.isEmpty()) {
final_urltriples.addAll(valid_triples);
}
}
return final_urltriples;
}
private static List<Triple> generateTriples_ResourceEvent_eventful(
URI uri, List<Triple> url_triples) {
List<Triple> valid_triples = new ArrayList<Triple>();
valid_triples.add(new Triple(uri.toString(), "wam:URI", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:url", uri.toString()));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Resource"));
valid_triples.add(new Triple(uri.toString(), "wam:type", "wam:Event"));
valid_triples.add(new Triple(uri.toString(), "wam:domain", "eventful.com"));
String title = "";
String addressLocality = "", addressRegion = "";
String venue = "";
for (Triple triple : url_triples) {
if (triple.predicate.equals(Constants.NS_OG + "type")
|| triple.predicate.equals(Constants.NS_OG + "title")
|| triple.predicate.equals(Constants.NS_OG + "url")
|| triple.predicate.equals(Constants.NS_XHTML + "description")
|| triple.predicate.equals(Constants.NS_XHTML + "keywords")
|| triple.predicate.equals(Constants.NS_GEO + "Point")
|| triple.predicate.equals(Constants.NS_GEO + "lat")
|| triple.predicate.equals(Constants.NS_GEO + "long")) {
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
}
/*
* Get Location
*/
if (triple.predicate.equals(Constants.NS_DC + "title") && title.isEmpty()) {
/*
* Example
* subject http://purl.org/dc/terms/title Tommee Profitt in Grand Haven, MI - Jul 1, 2012 7:00 pm | Eventful
*/
title = triple.object.toString();
try {
if (title.contains(",") && title.indexOf(",") > title.indexOf("in ")) {
addressLocality =
title.substring(title.lastIndexOf("in ") + 3, title.indexOf(",")).trim();
}
if (title.contains("-") && title.indexOf("-") > title.indexOf(",")) {
addressRegion = title.substring(title.indexOf(",") + 1, title.lastIndexOf("-")).trim();
}
} catch (Exception e) {
e.printStackTrace();
}
valid_triples.add(new Triple(uri.toString(), triple.predicate, triple.object));
valid_triples.add(
new Triple(
uri.toString(), Constants.NS_SCHEMA_ORG + "addressLocality", addressLocality));
valid_triples.add(
new Triple(uri.toString(), Constants.NS_SCHEMA_ORG + "addressRegion", addressRegion));
}
/*
* Get Venue
*/
if (triple.predicate.equals(Constants.NS_XHTML + "bookmark")
&& triple.object.matches(".*\\/venues\\/.*")
&& venue.isEmpty()) {
venue = triple.object;
valid_triples.add(triple);
valid_triples.add(new Triple(uri.toString(), "wam:hasVenue", (triple.object)));
try {
URI uri_venue = new URI(triple.object);
valid_triples.addAll(
generateTriples_ResourceVenue_eventful(uri_venue, new ArrayList<Triple>()));
} catch (URISyntaxException e) {
e.printStackTrace();
}
}
}
/*
* Get Performer
* GET URLS BY HTML-EXTRACTION
*/
MetadataEvalFunc extractor = new MetadataEvalFunc();
Map<String, String> performers = new HashMap<String, String>();
logger.info(" GET URLS BY HTML-EXTRACTION: " + uri.toString());
performers = extractor.scrapPage(uri.toString(), "eventful.com");
String performer_name = "";
String video_text_query = "";
for (String performer_url : performers.keySet()) {
logger.info("performer_url: " + performer_url);
try {
URI uri_performer = new URI(performer_url);
performer_name = performers.get(performer_url);
valid_triples.add(new Triple(uri.toString(), "wam:hasPerformer", uri_performer.toString()));
valid_triples.addAll(
generateTriples_ResourcePerformer_eventful(uri_performer, new ArrayList<Triple>()));
/*
* Get Video of Performer
*/
video_text_query = performer_name;
// logger.info(" GET VIDEOS PERFORMER WITH API: '" + performer_name+"'" +",
// "+video_text_query);
// YouTubeManager youtube_manager = new YouTubeManager();
// List<YouTubeVideo> videos = youtube_manager.retrieveVideos(video_text_query,
// video_max_results, true, youtube_timeout);
// for(YouTubeVideo video: videos){
// URI uri_video = new URI(video.getWebPlayerUrl());
// valid_triples.add(new Triple(uri_performer.toString(), "wam:hasVideo",
// uri_video.toString()));
// valid_triples.addAll(generateTriples_ResourceVideo(uri_video, video.getTitle(),
// performer_name, "eventful.com"));
//
// }
} catch (URISyntaxException e) {
e.printStackTrace();
}
}
/*
* Get Video of event (when no performer)
*/
// try {
// if(performers.isEmpty() && !title.isEmpty()){
// video_text_query = title;
// logger.info(" GET VIDEOS EVENT WITH API: " + uri.toString() +", "+video_text_query);
// YouTubeManager youtube_manager = new YouTubeManager();//youtube_clientID
// List<YouTubeVideo> videos_event = youtube_manager.retrieveVideos(video_text_query,
// video_max_results, true, youtube_timeout);
// for(YouTubeVideo video: videos_event){
// logger.info("video: " + video.getWebPlayerUrl() + ", " + video.getTitle());
// URI uri_video = new URI(video.getWebPlayerUrl());
// valid_triples.add(new Triple(uri.toString(), "wam:hasVideo", uri_video.toString()));
// valid_triples.addAll(generateTriples_ResourceVideo(uri_video, video.getTitle(), "",
// "eventful.com"));
//
// }
// }
// } catch (URISyntaxException e) {
// e.printStackTrace();
// } catch (Exception e) {
// e.printStackTrace();
// }
List<Triple> final_urltriples = new ArrayList<Triple>();
URI duplicate_uri = getDuplicate(uri, "wam:Event", title, addressLocality);
if (duplicate_uri != null) {
final_urltriples.add(new Triple(uri.toString(), "wam:hasResource", duplicate_uri.toString()));
} else {
if (!title.isEmpty()) {
logger.info("\t title: " + title);
final_urltriples.addAll(valid_triples);
}
}
return final_urltriples;
}
