List of usage examples for org.jsoup.nodes Element children
public Elements children()
From source file:org.structr.web.common.microformat.MicroformatParser.java
private void recurse(final Element element, final Map<String, Object> values, final int depth) { final Tag tag = element.tag(); final Set<String> classes = element.classNames(); final String link = element.attr("href"); final Object content = extractChildContent(element); if (!classes.isEmpty()) { removeEmpty(classes);/*from w w w. j ava2 s . co m*/ // toplevel classes define type if (tag.isBlock()) { if (depth == 0) { // store type attribute values.put("type", classes); for (final Element child : element.children()) { recurse(child, values, depth + 1); } } else { final Map<String, Object> childMap = new LinkedHashMap<>(); values.put(classes.iterator().next(), childMap); if (content != null) { childMap.put("name", content); } for (final Element child : element.children()) { recurse(child, childMap, depth + 1); } } } else if (tag.isInline()) { // extract href and store as URL if (classes.contains("url") && StringUtils.isNotBlank(link)) { values.put("url", link); classes.remove("url"); } if (content != null) { for (final String type : classes) { values.put(type, content); } } } } }
From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java
@Override public MediaMetadata getTvShowMetadata(MediaScrapeOptions options) throws Exception { MediaMetadata md = new MediaMetadata(providerInfo.getId()); String id = ""; String langu = options.getLanguage().name(); // id from result if (options.getResult() != null) { id = options.getResult().getId(); }/*from w w w .ja v a 2s . com*/ // do we have an id from the options? if (StringUtils.isEmpty(id)) { id = options.getId(providerInfo.getId()); } if (StringUtils.isEmpty(id)) { return md; } trackConnections(); // call API http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid=4242 String url = "http://api.anidb.net:9001/httpapi?request=anime&client=tinymediamanager&clientver=2&protover=1&aid=" + id; Document doc = null; try { CachedUrl cachedUrl = new CachedUrl(url); doc = Jsoup.parse(cachedUrl.getInputStream(), "UTF-8", "", Parser.xmlParser()); } catch (Exception e) { LOGGER.error("failed to get TV show metadata: " + e.getMessage()); // clear cache CachedUrl.removeCachedFileForUrl(url); } if (doc == null || doc.children().size() == 0) { return md; } md.setId(providerInfo.getId(), id); Element anime = doc.child(0); for (Element e : anime.children()) { if ("startdate".equalsIgnoreCase(e.tagName())) { md.storeMetadata(MediaMetadata.RELEASE_DATE, e.text()); try { Date date = org.tinymediamanager.scraper.util.StrgUtils.parseDate(e.text()); md.storeMetadata(MediaMetadata.YEAR, new SimpleDateFormat("yyyy").format(date)); } catch (Exception ex) { } } if ("titles".equalsIgnoreCase(e.tagName())) { parseTitle(md, langu, e); } if ("description".equalsIgnoreCase(e.tagName())) { md.storeMetadata(MediaMetadata.PLOT, e.text()); } if ("ratings".equalsIgnoreCase(e.tagName())) { getRating(md, e); } if ("picture".equalsIgnoreCase(e.tagName())) { md.storeMetadata(MediaMetadata.POSTER_URL, IMAGE_SERVER + e.text()); } if ("characters".equalsIgnoreCase(e.tagName())) { getActors(md, e); } } // add static "Anime" genre md.addGenre(MediaGenres.ANIME); return md; }
From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java
private void getActors(MediaMetadata md, Element e) { for (Element character : e.children()) { MediaCastMember member = new MediaCastMember(CastType.ACTOR); for (Element characterInfo : character.children()) { if ("name".equalsIgnoreCase(characterInfo.tagName())) { member.setCharacter(characterInfo.text()); }/*from ww w.j a v a 2 s .c o m*/ if ("seiyuu".equalsIgnoreCase(characterInfo.tagName())) { member.setName(characterInfo.text()); String image = characterInfo.attr("picture"); if (StringUtils.isNotBlank(image)) { member.setImageUrl("http://img7.anidb.net/pics/anime/" + image); } } } md.addCastMember(member); } }
From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java
private void getRating(MediaMetadata md, Element e) { for (Element rating : e.children()) { if ("temporary".equalsIgnoreCase(rating.tagName())) { try { md.storeMetadata(MediaMetadata.RATING, Float.parseFloat(rating.text())); md.storeMetadata(MediaMetadata.VOTE_COUNT, Integer.parseInt(rating.attr("count"))); break; } catch (NumberFormatException ex) { }/*from w ww.j a v a2 s .co m*/ } } }
From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java
private void parseTitle(MediaMetadata md, String langu, Element e) { String titleEN = ""; String titleScraperLangu = ""; String titleFirst = ""; for (Element title : e.children()) { // store first title if neither the requested one nor the english one available if (StringUtils.isBlank(titleFirst)) { titleFirst = title.text();/*from w ww. j av a2 s. c o m*/ } // store the english one for fallback if ("en".equalsIgnoreCase(title.attr("xml:lang"))) { titleEN = title.text(); } // search for the requested one if (langu.equalsIgnoreCase(title.attr("xml:lang"))) { titleScraperLangu = title.text(); } } if (StringUtils.isNotBlank(titleScraperLangu)) { md.storeMetadata(MediaMetadata.TITLE, titleScraperLangu); } else if (StringUtils.isNotBlank(titleEN)) { md.storeMetadata(MediaMetadata.TITLE, titleEN); } else { md.storeMetadata(MediaMetadata.TITLE, titleFirst); } }
From source file:org.tinymediamanager.scraper.anidb.AniDBMetadataProvider.java
private List<Episode> parseEpisodes(Document doc) { List<Episode> episodes = new ArrayList<Episode>(); Element anime = doc.child(0); Element eps = null;//from w w w .j a v a2s . co m // find the "episodes" child for (Element e : anime.children()) { if ("episodes".equalsIgnoreCase(e.tagName())) { eps = e; break; } } if (eps == null) { return episodes; } for (Element e : eps.children()) { // filter out the desired episode if ("episode".equals(e.tagName())) { Episode episode = new Episode(); try { episode.id = Integer.parseInt(e.attr("id")); } catch (NumberFormatException ex) { } for (Element episodeInfo : e.children()) { if ("epno".equalsIgnoreCase(episodeInfo.tagName())) { try { episode.episode = Integer.parseInt(episodeInfo.text()); // looks like anidb is storing anything in a single season, so put 1 to season, if type = 1 if ("1".equals(episodeInfo.attr("type"))) { episode.season = 1; } else { // else - we see them as "specials" episode.season = 0; } } catch (NumberFormatException ex) { } continue; } if ("length".equalsIgnoreCase(episodeInfo.tagName())) { try { episode.runtime = Integer.parseInt(episodeInfo.text()); } catch (NumberFormatException ex) { } continue; } if ("airdate".equalsIgnoreCase(episodeInfo.tagName())) { episode.airdate = episodeInfo.text(); continue; } if ("rating".equalsIgnoreCase(episodeInfo.tagName())) { try { episode.rating = Float.parseFloat(episodeInfo.text()); } catch (NumberFormatException ex) { } continue; } if ("title".equalsIgnoreCase(episodeInfo.tagName())) { try { episode.titles.put(episodeInfo.attr("xml:lang").toLowerCase(), episodeInfo.text()); } catch (Exception ex) { } continue; } if ("summary".equalsIgnoreCase(episodeInfo.tagName())) { episode.summary = episodeInfo.text(); continue; } } episodes.add(episode); } } return episodes; }
From source file:org.tinymediamanager.scraper.imdb.ImdbMetadataProvider.java
@Override public MediaMetadata getMetadata(MediaScrapeOptions options) throws Exception { LOGGER.debug("getMetadata() " + options.toString()); // check if there is a md in the result if (options.getResult() != null && options.getResult().getMetadata() != null) { LOGGER.debug("IMDB: getMetadata from cache: " + options.getResult()); return options.getResult().getMetadata(); }/*from ww w. j av a 2s . c o m*/ MediaMetadata md = new MediaMetadata(providerInfo.getId()); String imdbId = ""; // imdbId from searchResult if (options.getResult() != null) { imdbId = options.getResult().getIMDBId(); } // imdbid from scraper option if (!MetadataUtil.isValidImdbId(imdbId)) { imdbId = options.getImdbId(); } if (!MetadataUtil.isValidImdbId(imdbId)) { return md; } LOGGER.debug("IMDB: getMetadata(imdbId): " + imdbId); md.setId(MediaMetadata.IMDBID, imdbId); ExecutorCompletionService<Document> compSvcImdb = new ExecutorCompletionService<Document>(executor); ExecutorCompletionService<MediaMetadata> compSvcTmdb = new ExecutorCompletionService<MediaMetadata>( executor); // worker for imdb request (/combined) (everytime from akas.imdb.com) // StringBuilder sb = new StringBuilder(imdbSite.getSite()); StringBuilder sb = new StringBuilder(ImdbSiteDefinition.IMDB_COM.getSite()); sb.append("title/"); sb.append(imdbId); sb.append("/combined"); Callable<Document> worker = new ImdbWorker(sb.toString(), options.getLanguage().name(), options.getCountry().getAlpha2()); Future<Document> futureCombined = compSvcImdb.submit(worker); // worker for imdb request (/plotsummary) (from chosen site) Future<Document> futurePlotsummary = null; sb = new StringBuilder(imdbSite.getSite()); sb.append("title/"); sb.append(imdbId); sb.append("/plotsummary"); worker = new ImdbWorker(sb.toString(), options.getLanguage().name(), options.getCountry().getAlpha2()); futurePlotsummary = compSvcImdb.submit(worker); // worker for tmdb request Future<MediaMetadata> futureTmdb = null; if (options.isScrapeImdbForeignLanguage() || options.isScrapeCollectionInfo()) { Callable<MediaMetadata> worker2 = new TmdbWorker(imdbId, options.getLanguage(), options.getCountry()); futureTmdb = compSvcTmdb.submit(worker2); } Document doc; doc = futureCombined.get(); /* * title and year have the following structure * * <div id="tn15title"><h1>Merida - Legende der Highlands <span>(<a href="/year/2012/">2012</a>) <span class="pro-link">...</span> <span * class="title-extra">Brave <i>(original title)</i></span> </span></h1> </div> */ // parse title and year Element title = doc.getElementById("tn15title"); if (title != null) { Element element = null; // title Elements elements = title.getElementsByTag("h1"); if (elements.size() > 0) { element = elements.first(); String movieTitle = cleanString(element.ownText()); md.storeMetadata(MediaMetadata.TITLE, movieTitle); } // year elements = title.getElementsByTag("span"); if (elements.size() > 0) { element = elements.first(); String content = element.text(); // search year Pattern yearPattern = Pattern.compile("\\(([0-9]{4})|/\\)"); Matcher matcher = yearPattern.matcher(content); while (matcher.find()) { if (matcher.group(1) != null) { String movieYear = matcher.group(1); md.storeMetadata(MediaMetadata.YEAR, movieYear); break; } } } // original title elements = title.getElementsByAttributeValue("class", "title-extra"); if (elements.size() > 0) { element = elements.first(); String content = element.text(); content = content.replaceAll("\\(original title\\)", "").trim(); md.storeMetadata(MediaMetadata.ORIGINAL_TITLE, content); } } // poster Element poster = doc.getElementById("primary-poster"); if (poster != null) { String posterUrl = poster.attr("src"); posterUrl = posterUrl.replaceAll("SX[0-9]{2,4}_", "SX400_"); posterUrl = posterUrl.replaceAll("SY[0-9]{2,4}_", "SY400_"); processMediaArt(md, MediaArtworkType.POSTER, "Poster", posterUrl); } /* * <div class="starbar-meta"> <b>7.4/10</b> <a href="ratings" class="tn15more">52,871 votes</a> » </div> */ // rating and rating count Element ratingElement = doc.getElementById("tn15rating"); if (ratingElement != null) { Elements elements = ratingElement.getElementsByClass("starbar-meta"); if (elements.size() > 0) { Element div = elements.get(0); // rating comes in <b> tag Elements b = div.getElementsByTag("b"); if (b.size() == 1) { String ratingAsString = b.text(); Pattern ratingPattern = Pattern.compile("([0-9]\\.[0-9])/10"); Matcher matcher = ratingPattern.matcher(ratingAsString); while (matcher.find()) { if (matcher.group(1) != null) { float rating = 0; try { rating = Float.valueOf(matcher.group(1)); } catch (Exception e) { } md.storeMetadata(MediaMetadata.RATING, rating); break; } } } // count Elements a = div.getElementsByAttributeValue("href", "ratings"); if (a.size() == 1) { String countAsString = a.text().replaceAll("[.,]|votes", "").trim(); int voteCount = 0; try { voteCount = Integer.parseInt(countAsString); } catch (Exception e) { } md.storeMetadata(MediaMetadata.VOTE_COUNT, voteCount); } } // top250 elements = ratingElement.getElementsByClass("starbar-special"); if (elements.size() > 0) { Elements a = elements.get(0).getElementsByTag("a"); if (a.size() > 0) { Element anchor = a.get(0); Pattern topPattern = Pattern.compile("Top 250: #([0-9]{1,3})"); Matcher matcher = topPattern.matcher(anchor.ownText()); while (matcher.find()) { if (matcher.group(1) != null) { int top250 = 0; try { top250 = Integer.parseInt(matcher.group(1)); } catch (Exception e) { } md.storeMetadata(MediaMetadata.TOP_250, top250); } } } } } // parse all items coming by <div class="info"> Elements elements = doc.getElementsByClass("info"); for (Element element : elements) { // only parse divs if (!"div".equals(element.tag().getName())) { continue; } // elements with h5 are the titles of the values Elements h5 = element.getElementsByTag("h5"); if (h5.size() > 0) { Element firstH5 = h5.first(); String h5Title = firstH5.text(); // release date /* * <div class="info"><h5>Release Date:</h5><div class="info-content">5 January 1996 (USA)<a class="tn15more inline" * href="/title/tt0114746/releaseinfo" * onclick="(new Image()).src='/rg/title-tease/releasedates/images/b.gif?link=/title/tt0114746/releaseinfo';"> See more</a> </div></div> */ if (h5Title.matches("(?i)" + ImdbSiteDefinition.IMDB_COM.getReleaseDate() + ".*")) { Elements div = element.getElementsByClass("info-content"); if (div.size() > 0) { Element releaseDateElement = div.first(); String releaseDate = cleanString(releaseDateElement.ownText().replaceAll("", "")); Pattern pattern = Pattern.compile("(.*)\\(.*\\)"); Matcher matcher = pattern.matcher(releaseDate); if (matcher.find()) { try { SimpleDateFormat sdf = new SimpleDateFormat("d MMM yyyy"); Date parsedDate = sdf.parse(matcher.group(1)); sdf = new SimpleDateFormat("dd-MM-yyyy"); md.storeMetadata(MediaMetadata.RELEASE_DATE, sdf.format(parsedDate)); } catch (Exception e) { } } } } /* * <div class="info"><h5>Tagline:</h5><div class="info-content"> (7) To Defend Us... <a class="tn15more inline" * href="/title/tt0472033/taglines" onClick= "(new Image()).src='/rg/title-tease/taglines/images/b.gif?link=/title/tt0472033/taglines';" >See * more</a> » </div></div> */ // tagline if (h5Title.matches("(?i)" + ImdbSiteDefinition.IMDB_COM.getTagline() + ".*") && !options.isScrapeImdbForeignLanguage()) { Elements div = element.getElementsByClass("info-content"); if (div.size() > 0) { Element taglineElement = div.first(); String tagline = cleanString(taglineElement.ownText().replaceAll("", "")); md.storeMetadata(MediaMetadata.TAGLINE, tagline); } } /* * <div class="info-content"><a href="/Sections/Genres/Animation/">Animation</a> | <a href="/Sections/Genres/Action/">Action</a> | <a * href="/Sections/Genres/Adventure/">Adventure</a> | <a href="/Sections/Genres/Fantasy/">Fantasy</a> | <a * href="/Sections/Genres/Mystery/">Mystery</a> | <a href="/Sections/Genres/Sci-Fi/">Sci-Fi</a> | <a * href="/Sections/Genres/Thriller/">Thriller</a> <a class="tn15more inline" href="/title/tt0472033/keywords" onClick= * "(new Image()).src='/rg/title-tease/keywords/images/b.gif?link=/title/tt0472033/keywords';" > See more</a> » </div> */ // genres are only scraped from akas.imdb.com if (h5Title.matches("(?i)" + imdbSite.getGenre() + "(.*)")) { Elements div = element.getElementsByClass("info-content"); if (div.size() > 0) { Elements a = div.first().getElementsByTag("a"); for (Element anchor : a) { if (anchor.attr("href").matches("/Sections/Genres/.*")) { md.addGenre(getTmmGenre(anchor.ownText())); } } } } // } /* * <div class="info"><h5>Runtime:</h5><div class="info-content">162 min | 171 min (special edition) | 178 min (extended cut)</div></div> */ // runtime // if (h5Title.matches("(?i)" + imdbSite.getRuntime() + ".*")) { if (h5Title.matches("(?i)" + ImdbSiteDefinition.IMDB_COM.getRuntime() + ".*")) { Elements div = element.getElementsByClass("info-content"); if (div.size() > 0) { Element taglineElement = div.first(); String first = taglineElement.ownText().split("\\|")[0]; String runtimeAsString = cleanString(first.replaceAll("min", "")); int runtime = 0; try { runtime = Integer.parseInt(runtimeAsString); } catch (Exception e) { // try to filter out the first number we find Pattern runtimePattern = Pattern.compile("([0-9]{2,3})"); Matcher matcher = runtimePattern.matcher(runtimeAsString); if (matcher.find()) { runtime = Integer.parseInt(matcher.group(0)); } } md.storeMetadata(MediaMetadata.RUNTIME, runtime); } } /* * <div class="info"><h5>Country:</h5><div class="info-content"><a href="/country/fr">France</a> | <a href="/country/es">Spain</a> | <a * href="/country/it">Italy</a> | <a href="/country/hu">Hungary</a></div></div> */ // country if (h5Title.matches("(?i)Country.*")) { Elements a = element.getElementsByTag("a"); String countries = ""; for (Element anchor : a) { Pattern pattern = Pattern.compile("/country/(.*)"); Matcher matcher = pattern.matcher(anchor.attr("href")); if (matcher.matches()) { String country = matcher.group(1); if (StringUtils.isNotEmpty(countries)) { countries += ", "; } countries += country.toUpperCase(); } } md.storeMetadata(MediaMetadata.COUNTRY, countries); } /* * <div class="info"><h5>Language:</h5><div class="info-content"><a href="/language/en">English</a> | <a href="/language/de">German</a> | <a * href="/language/fr">French</a> | <a href="/language/it">Italian</a></div> */ // Spoken languages if (h5Title.matches("(?i)Language.*")) { Elements a = element.getElementsByTag("a"); String spokenLanguages = ""; for (Element anchor : a) { Pattern pattern = Pattern.compile("/language/(.*)"); Matcher matcher = pattern.matcher(anchor.attr("href")); if (matcher.matches()) { String langu = matcher.group(1); if (StringUtils.isNotEmpty(spokenLanguages)) { spokenLanguages += ", "; } spokenLanguages += langu; } } md.storeMetadata(MediaMetadata.SPOKEN_LANGUAGES, spokenLanguages); } /* * <div class="info"><h5>Certification:</h5><div class="info-content"><a href="/search/title?certificates=us:pg">USA:PG</a> <i>(certificate * #47489)</i> | <a href="/search/title?certificates=ca:pg">Canada:PG</a> <i>(Ontario)</i> | <a * href="/search/title?certificates=au:pg">Australia:PG</a> | <a href="/search/title?certificates=in:u">India:U</a> | <a * href="/search/title?certificates=ie:pg">Ireland:PG</a> ...</div></div> */ // certification // if (h5Title.matches("(?i)" + imdbSite.getCertification() + ".*")) { if (h5Title.matches("(?i)" + ImdbSiteDefinition.IMDB_COM.getCertification() + ".*")) { Elements a = element.getElementsByTag("a"); for (Element anchor : a) { // certification for the right country if (anchor.attr("href").matches( "(?i)/search/title\\?certificates=" + options.getCountry().getAlpha2() + ".*")) { Pattern certificationPattern = Pattern.compile(".*:(.*)"); Matcher matcher = certificationPattern.matcher(anchor.ownText()); Certification certification = null; while (matcher.find()) { if (matcher.group(1) != null) { certification = Certification.getCertification(options.getCountry(), matcher.group(1)); } } if (certification != null) { md.addCertification(certification); break; } } } } } /* * <div id="director-info" class="info"> <h5>Director:</h5> <div class="info-content"><a href="/name/nm0000416/" onclick= * "(new Image()).src='/rg/directorlist/position-1/images/b.gif?link=name/nm0000416/';" >Terry Gilliam</a><br/> </div> </div> */ // director if ("director-info".equals(element.id())) { Elements a = element.getElementsByTag("a"); for (Element anchor : a) { if (anchor.attr("href").matches("/name/nm.*")) { MediaCastMember cm = new MediaCastMember(CastType.DIRECTOR); cm.setName(anchor.ownText()); md.addCastMember(cm); } } } } /* * <table class="cast"> <tr class="odd"><td class="hs"><a href="http://pro.imdb.com/widget/resume_redirect/" onClick= * "(new Image()).src='/rg/resume/prosystem/images/b.gif?link=http://pro.imdb.com/widget/resume_redirect/';" ><img src= * "http://i.media-imdb.com/images/SF9113d6f5b7cb1533c35313ccd181a6b1/tn15/no_photo.png" width="25" height="31" border="0"></td><td class="nm"><a * href="/name/nm0577828/" onclick= "(new Image()).src='/rg/castlist/position-1/images/b.gif?link=/name/nm0577828/';" >Joseph Melito</a></td><td * class="ddd"> ... </td><td class="char"><a href="/character/ch0003139/">Young Cole</a></td></tr> <tr class="even"><td class="hs"><a * href="/name/nm0000246/" onClick= "(new Image()).src='/rg/title-tease/tinyhead/images/b.gif?link=/name/nm0000246/';" ><img src= * "http://ia.media-imdb.com/images/M/MV5BMjA0MjMzMTE5OF5BMl5BanBnXkFtZTcwMzQ2ODE3Mw@@._V1._SY30_SX23_.jpg" width="23" height="32" * border="0"></a><br></td><td class="nm"><a href="/name/nm0000246/" onclick= * "(new Image()).src='/rg/castlist/position-2/images/b.gif?link=/name/nm0000246/';" >Bruce Willis</a></td><td class="ddd"> ... </td><td * class="char"><a href="/character/ch0003139/">James Cole</a></td></tr> <tr class="odd"><td class="hs"><a href="/name/nm0781218/" onClick= * "(new Image()).src='/rg/title-tease/tinyhead/images/b.gif?link=/name/nm0781218/';" ><img src= * "http://ia.media-imdb.com/images/M/MV5BODI1MTA2MjkxM15BMl5BanBnXkFtZTcwMTcwMDg2Nw@@._V1._SY30_SX23_.jpg" width="23" height="32" * border="0"></a><br></td><td class="nm"><a href="/name/nm0781218/" onclick= * "(new Image()).src='/rg/castlist/position-3/images/b.gif?link=/name/nm0781218/';" >Jon Seda</a></td><td class="ddd"> ... </td><td * class="char"><a href="/character/ch0003143/">Jose</a></td></tr>...</table> */ // cast elements = doc.getElementsByClass("cast"); if (elements.size() > 0) { Elements tr = elements.get(0).getElementsByTag("tr"); for (Element row : tr) { Elements td = row.getElementsByTag("td"); MediaCastMember cm = new MediaCastMember(); for (Element column : td) { // actor thumb if (column.hasClass("hs")) { Elements img = column.getElementsByTag("img"); if (img.size() > 0) { String thumbUrl = img.get(0).attr("src"); if (thumbUrl.contains("no_photo.png")) { cm.setImageUrl(""); } else { thumbUrl = thumbUrl.replaceAll("SX[0-9]{2,4}_", "SX400_"); thumbUrl = thumbUrl.replaceAll("SY[0-9]{2,4}_", ""); cm.setImageUrl(thumbUrl); } } } // actor name if (column.hasClass("nm")) { cm.setName(cleanString(column.text())); } // character if (column.hasClass("char")) { cm.setCharacter(cleanString(column.text())); } } if (StringUtils.isNotEmpty(cm.getName()) && StringUtils.isNotEmpty(cm.getCharacter())) { cm.setType(CastType.ACTOR); md.addCastMember(cm); } } } Element content = doc.getElementById("tn15content"); if (content != null) { elements = content.getElementsByTag("table"); for (Element table : elements) { // writers if (table.text().contains(ImdbSiteDefinition.IMDB_COM.getWriter())) { Elements anchors = table.getElementsByTag("a"); for (Element anchor : anchors) { if (anchor.attr("href").matches("/name/nm.*")) { MediaCastMember cm = new MediaCastMember(CastType.WRITER); cm.setName(anchor.ownText()); md.addCastMember(cm); } } } // producers if (table.text().contains(ImdbSiteDefinition.IMDB_COM.getProducers())) { Elements rows = table.getElementsByTag("tr"); for (Element row : rows) { if (row.text().contains(ImdbSiteDefinition.IMDB_COM.getProducers())) { continue; } Elements columns = row.children(); if (columns.size() == 0) { continue; } MediaCastMember cm = new MediaCastMember(CastType.PRODUCER); String name = cleanString(columns.get(0).text()); if (StringUtils.isBlank(name)) { continue; } cm.setName(name); if (columns.size() >= 3) { cm.setPart(cleanString(columns.get(2).text())); } md.addCastMember(cm); } } } } // Production companies elements = doc.getElementsByClass("blackcatheader"); for (Element blackcatheader : elements) { if (blackcatheader.ownText().equals(ImdbSiteDefinition.IMDB_COM.getProductionCompanies())) { Elements a = blackcatheader.nextElementSibling().getElementsByTag("a"); StringBuilder productionCompanies = new StringBuilder(); for (Element anchor : a) { if (StringUtils.isNotEmpty(productionCompanies)) { productionCompanies.append(", "); } productionCompanies.append(anchor.ownText()); } md.storeMetadata(MediaMetadata.PRODUCTION_COMPANY, productionCompanies.toString()); break; } } /* * plot from /plotsummary */ // build the url doc = null; doc = futurePlotsummary.get(); // imdb.com has another site structure if (imdbSite == ImdbSiteDefinition.IMDB_COM) { Elements zebraList = doc.getElementsByClass("zebraList"); if (zebraList != null && !zebraList.isEmpty()) { Elements odd = zebraList.get(0).getElementsByClass("odd"); if (odd.isEmpty()) { odd = zebraList.get(0).getElementsByClass("even"); // sometimes imdb has even } if (odd.size() > 0) { Elements p = odd.get(0).getElementsByTag("p"); if (p.size() > 0) { String plot = cleanString(p.get(0).ownText()); md.storeMetadata(MediaMetadata.PLOT, plot); } } } } else { Element wiki = doc.getElementById("swiki.2.1"); if (wiki != null) { String plot = cleanString(wiki.ownText()); md.storeMetadata(MediaMetadata.PLOT, plot); } } // title also from chosen site if we are not scraping akas.imdb.com if (imdbSite != ImdbSiteDefinition.IMDB_COM) { title = doc.getElementById("tn15title"); if (title != null) { Element element = null; // title elements = title.getElementsByClass("main"); if (elements.size() > 0) { element = elements.first(); String movieTitle = cleanString(element.ownText()); md.storeMetadata(MediaMetadata.TITLE, movieTitle); } } } // } // get data from tmdb? if (options.isScrapeImdbForeignLanguage() || options.isScrapeCollectionInfo()) { MediaMetadata tmdbMd = futureTmdb.get(); if (options.isScrapeImdbForeignLanguage() && tmdbMd != null && StringUtils.isNotBlank(tmdbMd.getStringValue(MediaMetadata.PLOT))) { // tmdbid md.setId(MediaMetadata.TMDBID, tmdbMd.getId(MediaMetadata.TMDBID)); // title md.storeMetadata(MediaMetadata.TITLE, tmdbMd.getStringValue(MediaMetadata.TITLE)); // original title md.storeMetadata(MediaMetadata.ORIGINAL_TITLE, tmdbMd.getStringValue(MediaMetadata.ORIGINAL_TITLE)); // tagline md.storeMetadata(MediaMetadata.TAGLINE, tmdbMd.getStringValue(MediaMetadata.TAGLINE)); // plot md.storeMetadata(MediaMetadata.PLOT, tmdbMd.getStringValue(MediaMetadata.PLOT)); // collection info md.storeMetadata(MediaMetadata.COLLECTION_NAME, tmdbMd.getStringValue(MediaMetadata.COLLECTION_NAME)); md.storeMetadata(MediaMetadata.TMDBID_SET, tmdbMd.getIntegerValue(MediaMetadata.TMDBID_SET)); } if (options.isScrapeCollectionInfo() && tmdbMd != null) { md.storeMetadata(MediaMetadata.TMDBID_SET, tmdbMd.getIntegerValue(MediaMetadata.TMDBID_SET)); md.storeMetadata(MediaMetadata.COLLECTION_NAME, tmdbMd.getStringValue(MediaMetadata.COLLECTION_NAME)); } } // if we have still no original title, take the title if (StringUtils.isBlank(md.getStringValue(MediaMetadata.ORIGINAL_TITLE))) { md.storeMetadata(MediaMetadata.ORIGINAL_TITLE, md.getStringValue(MediaMetadata.TITLE)); } return md; }
From source file:org.trec.liveqa.GetYAnswersPropertiesFromQid.java
private static Element findElement(Element e, ElementPredicate f) { if (f.check(e)) { return e; }/*ww w. j a v a 2 s. c o m*/ for (Element c : e.children()) { Element elem = findElement(c, f); if (elem != null) { return elem; } } return null; }
From source file:org.wandora.application.tools.extractors.bookmark.BookmarkExtractor.java
private void parseCategory(Element c, TopicMap t) throws TopicMapException, ParseException { Topic wc = getWandoraClass(t);/*ww w . j a va2 s. com*/ Topic root = getOrCreateTopic(t, ROOT_SI, "Bookmark"); root.setSubjectLocator(new Locator(ROOT_SI)); root.setDisplayName(LANG, "Bookmark"); makeSubclassOf(t, root, wc); for (Element child : c.children()) { if (child.tagName().equals("dt")) { for (Element grandChild : child.children()) { if (grandChild.tagName().equals("a")) parseItem(grandChild, root, t); else if (grandChild.tagName().equals("dl")) parseCategory(child, root, t); } } } parseCategory(c, root, t); }
From source file:org.wandora.application.tools.extractors.bookmark.BookmarkExtractor.java
private void parseCategory(Element c, Topic parent, TopicMap t) throws TopicMapException, ParseException { Topic cTopic = parent;/*w ww . jav a 2s. co m*/ Elements children = c.children(); for (Element child : children) { if (child.tagName().equals("h3")) { String cLocator = parent.getSubjectLocator().toString(); cLocator += "/" + urlEncode(child.html()); String cName = child.ownText(); cTopic = getOrCreateTopic(t, cLocator); cTopic.setSubjectLocator(new Locator(cLocator)); cTopic.setBaseName(cName + " (Bookmark)"); cTopic.setDisplayName(LANG, cName); makeSubclassOf(t, cTopic, parent); } } for (Element child : children) { if (!child.tagName().equals("dl")) continue; for (Element grandChild : child.children()) { if (!grandChild.tagName().equals("dt")) continue; for (Element ggChild : grandChild.children()) { if (ggChild.tagName().equals("a")) parseItem(ggChild, cTopic, t); else if (ggChild.tagName().equals("dl")) parseCategory(grandChild, cTopic, t); } } } }