List of usage examples for org.jsoup.nodes Element text
public String text()
From source file:net.meiolania.apps.habrahabr.fragments.posts.loader.PostShowLoader.java
@Override public PostsFullData loadInBackground() { PostsFullData data = new PostsFullData(); try {/*ww w . j a va 2s . c o m*/ Document document = Jsoup.connect(url).get(); Element title = document.select("span.post_title").first(); Element hubs = document.select("div.hubs").first(); Element content = document.select("div.content").first(); Element date = document.select("div.published").first(); Element author = document.select("div.author > a").first(); if (title != null) { data.setUrl(url); data.setTitle(title.text()); data.setHubs(hubs.text()); data.setContent(content.html()); data.setDate(date.text()); data.setAuthor(author.text()); } else data.setContent(context.getString(R.string.error_404)); } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.posts.loader.PostsLoader.java
@Override public ArrayList<PostsData> loadInBackground() { ArrayList<PostsData> data = new ArrayList<PostsData>(); try {/*from w w w .jav a2 s. co m*/ String readyUrl = url.replace("%page%", String.valueOf(page)); Log.i(TAG, "Loading a page: " + readyUrl); Document document = Jsoup.connect(readyUrl).get(); Elements posts = document.select("div.post"); for (Element post : posts) { PostsData postsData = new PostsData(); Element postTitle = post.select("a.post_title").first(); Element hubs = post.select("div.hubs").first(); Element date = post.select("div.published").first(); Element author = post.select("div.author > a").first(); Element comments = post.select("div.comments > span.all").first(); Element score = post.select("span.score").first(); postsData.setTitle(postTitle.text()); postsData.setUrl(postTitle.attr("abs:href")); postsData.setHubs(hubs.text()); postsData.setDate(date.text()); postsData.setAuthor(author != null ? author.text() : ""); postsData.setComments(comments != null ? comments.text() : "0"); postsData.setScore(score.text()); data.add(postsData); } } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.qa.loader.QaCommentsLoader.java
@Override public ArrayList<CommentsData> loadInBackground() { ArrayList<CommentsData> data = new ArrayList<CommentsData>(); try {/*from ww w .j a v a 2 s. c o m*/ Document document = Jsoup.connect(url).get(); Elements answers = document.select("div.answer"); for (Element answer : answers) { CommentsData commentsData = new CommentsData(); Element name = answer.select("a.username").first(); Element message = answer.select("div.message").first(); Element linkToComment = answer.select("a.link_to_comment").first(); Element score = answer.select("span.score").first(); commentsData.setUrl(linkToComment.attr("abs:href")); commentsData.setAuthor(name.text()); commentsData.setAuthorUrl(name.attr("abs:href")); commentsData.setComment(message.text()); commentsData.setLevel(0); commentsData.setScore(score.text()); data.add(commentsData); Elements comments = answer.select("div.comment_item"); for (Element comment : comments) { commentsData = new CommentsData(); name = comment.select("span.info > a").first(); message = comment.select("span.text").first(); commentsData.setUrl(linkToComment.attr("abs:href")); commentsData.setAuthorUrl(name.attr("abs:href")); commentsData.setAuthor(name.text()); commentsData.setComment(message.text()); commentsData.setLevel(1); commentsData.setScore(""); data.add(commentsData); } } } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.qa.loader.QaLoader.java
@Override public ArrayList<QaData> loadInBackground() { ArrayList<QaData> data = new ArrayList<QaData>(); try {/* w w w. jav a2 s . c o m*/ String readyUrl = url.replace("%page%", String.valueOf(page)); Log.i(TAG, "Loading a page: " + readyUrl); Document document = Jsoup.connect(readyUrl).get(); Elements qaList = document.select("div.post"); for (Element qa : qaList) { QaData qaData = new QaData(); Element title = qa.select("a.post_title").first(); Element hubs = qa.select("div.hubs").first(); Element answers = qa.select("div.informative").first(); Element date = qa.select("div.published").first(); Element author = qa.select("div.author > a").first(); Element score = qa.select("span.score").first(); qaData.setTitle(title.text()); qaData.setUrl(title.attr("abs:href")); qaData.setHubs(hubs.text()); qaData.setAnswers(answers.text()); qaData.setDate(date.text()); qaData.setAuthor(author.text()); qaData.setScore(score.text()); data.add(qaData); } } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.qa.loader.QaShowLoader.java
@Override public QaFullData loadInBackground() { QaFullData data = new QaFullData(); try {/*w w w . java 2 s. c om*/ Log.i(TAG, "Loading a page: " + url); Document document = Jsoup.connect(url).get(); Element title = document.select("span.post_title").first(); Element hubs = document.select("div.hubs").first(); Element content = document.select("div.content").first(); Element tags = document.select("ul.tags").first(); Element date = document.select("div.published").first(); Element author = document.select("div.author > a").first(); Element answers = document.select("span#comments_count").first(); data.setTitle(title.text()); data.setHubs(hubs.text()); data.setContent(content.html()); data.setTags(tags.text()); data.setDate(date.text()); data.setAuthor(author.text()); data.setAnswers(answers.text()); } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.users.loader.UsersLoader.java
@Override public ArrayList<UsersData> loadInBackground() { ArrayList<UsersData> data = new ArrayList<UsersData>(); try {// w w w.j a v a 2s . c om Log.i(TAG, "Loading a page: " + url); Document document = Jsoup.connect(url).get(); Elements users = document.select("div.user"); for (Element user : users) { UsersData usersData = new UsersData(); Element rating = user.select("div.rating").first(); Element karma = user.select("div.karma").first(); Element avatar = user.select("div.avatar > a > img").first(); Element name = user.select("div.userlogin > div.username > a").first(); Element lifetime = user.select("div.info > div.lifetime").first(); usersData.setName(name.text()); usersData.setUrl(name.attr("abs:href")); usersData.setRating(rating.text()); usersData.setKarma(karma.text()); usersData.setAvatar(avatar.attr("src")); usersData.setLifetime(lifetime.text()); data.add(usersData); } } catch (IOException e) { } return data; }
From source file:net.meiolania.apps.habrahabr.fragments.users.loader.UsersShowLoader.java
@Override public UsersFullData loadInBackground() { UsersFullData data = new UsersFullData(); try {/*from w w w . ja v a 2s . c om*/ Log.i(TAG, "Loading a page: " + url); Document document = Jsoup.connect(url).get(); Element avatar = document.select("a.avatar > img").first(); Element karma = document.select("div.karma > div.score > div.num").first(); Element rating = document.select("div.rating > div.num").first(); Element birthday = document.select("dd.bday").first(); Element fullname = document.select("div.fullname").first(); Element summary = document.select("dd.summary").first(); Element interests = document.select("dl.interests > dd").first(); data.setAvatar(avatar.attr("src")); data.setKarma(karma.text()); data.setRating(rating.text()); data.setBirthday(birthday != null ? birthday.text() : ""); data.setFullname(fullname != null ? fullname.text() : ""); data.setSummary(summary != null ? summary.text() : ""); data.setInterests(interests != null ? interests.text() : ""); } catch (IOException e) { } return data; }
From source file:net.trustie.model.OpenHubProject_Model.java
public void afterProcess(Page page) { this.urlMd5 = DigestUtils.md5Hex(page.getPageUrl()); SimpleDateFormat bartDateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); this.collectTime = bartDateFormat.format(new Date()); Document doc = page.getHtml().getDocument(); // use_count//from w w w . j a v a 2s . co m Elements useCountLabel = doc .select("div#widgets div[itemprop=interactionCount] div.float_right div.use_count a"); if (useCountLabel.size() > 0) { String strUseCount = useCountLabel.get(0).text(); this.useCount = Integer.parseInt(StringUtils.remove(strUseCount, ',')); } // tags Elements tags = doc.select("div#project_tags p a"); String tag = null; List<String> listTags = new ArrayList<String>(); for (Element e : tags) { tag = e.text(); listTags.add(tag); } this.tags = StringHandler.combineTags(listTags); // quick reference // Elements quickRefs = doc.select("div.span6 div.well dl."); Elements quickRefs = doc.select("div.span6 div.well dl"); if (quickRefs.size() > 0) { Element quickRef = quickRefs.get(0); handleQuickRef(quickRef); } // nutshell // if (codeInfos.size() > 0) { // String nutShell0 = codeInfos.get(0); // handleNutShell0(nutShell0); // String nutShell1 = codeInfos.get(1); // handleNutShell1(nutShell1); // String nutShell2 = codeInfos.get(2); // handleNutShell2(nutShell2); // String nutShell3 = codeInfos.get(3); // handleNutShell3(nutShell3, page.getTime()); // } int tmplength = codeInfos.size(); if (tmplength > 0) { if (tmplength == 1) { String nutShell0 = codeInfos.get(0); handleNutShell0(nutShell0); } else if (tmplength == 2) { String nutShell0 = codeInfos.get(0); handleNutShell0(nutShell0); String nutShell1 = codeInfos.get(1); handleNutShell1(nutShell1); } else if (tmplength == 3) { String nutShell0 = codeInfos.get(0); handleNutShell0(nutShell0); String nutShell1 = codeInfos.get(1); handleNutShell1(nutShell1); String nutShell2 = codeInfos.get(2); handleNutShell2(nutShell2); } else { String nutShell0 = codeInfos.get(0); handleNutShell0(nutShell0); String nutShell1 = codeInfos.get(1); handleNutShell1(nutShell1); String nutShell2 = codeInfos.get(2); handleNutShell2(nutShell2); String nutShell3 = codeInfos.get(3); handleNutShell3(nutShell3, page.getTime()); } } List<String> eList = new ArrayList<String>(); List<String> valueList = new ArrayList<String>(); // language for (int i = 0; i < languages.size(); i++) { Element e = Jsoup.parse(languages.get(i)); Element eValue = Jsoup.parse(percentages.get(i)); // String strE = e.text(); eList.add(e.text()); // String strValue = eValue.text(); valueList.add(eValue.text()); } this.languagePercentages = StringHandler.assemblyOSSEANMap(eList, valueList); // activity String strtmp1 = ""; String strtmp2 = ""; String strtmp3 = ""; String strtmp4 = ""; // this.dCommitNumber = Integer.parseInt(dayActivityInfos.get(0)); // this.dContributorNumber = Integer.parseInt(dayActivityInfos.get(1)); // if (dayActivityInfos.size() > 0) { // strtmp1 = dayActivityInfos.get(0); // this.daysCommitNumber = getInt(strtmp1); // strtmp2 = dayActivityInfos.get(1); // this.daysContributorNumber = getInt(strtmp2); // strtmp3 = monthActivityInfos.get(0); // this.monthsCommitNumber = getInt(strtmp3); // strtmp4 = monthActivityInfos.get(1); // this.monthsContributorNumber = getInt(strtmp4); // // } int length = dayActivityInfos.size(); if (length > 0) { if (length == 1) { strtmp1 = dayActivityInfos.get(0); this.daysCommitNumber = getInt(strtmp1); } else { strtmp1 = dayActivityInfos.get(0); this.daysCommitNumber = getInt(strtmp1); strtmp2 = dayActivityInfos.get(1); this.daysContributorNumber = getInt(strtmp2); } } int length1 = monthActivityInfos.size(); if (length1 > 0) { if (length1 == 1) { strtmp3 = monthActivityInfos.get(0); this.monthsCommitNumber = getInt(strtmp3); } else { strtmp3 = monthActivityInfos.get(0); this.monthsCommitNumber = getInt(strtmp3); strtmp4 = monthActivityInfos.get(1); this.monthsContributorNumber = getInt(strtmp4); } } this.newContriNum = getInt(this.newContributor); // this.theCommitTrend = allTrend.get(0); // this.theContriTrend = allTrend.get(1); if (allTrend.size() > 0) { if (allTrend.size() == 1) { this.theCommitTrend = allTrend.get(0); } else { this.theCommitTrend = allTrend.get(0); this.theContriTrend = allTrend.get(1); } } // community // this.rateNum = getInt(this.rateInfo); this.ReContributor = StringHandler.combineTags(this.RecenctContributors); // button informations // String headHref = "https://www.openhub.net"; // Document docTmp = Jsoup.parse(this.linkInfos.get(0)); // this.links = docTmp.getElementsByTag("a"); // // for(Element link : links){ // String url = headHref + link.attr("href"); // ansTmp += url +"**"; // } // // if (this.firstLinks.size() > 0) { // // this.newsLink = tmphref + this.firstLinks.get(0); // this.langLink = tmphref + this.firstLinks.get(1); // // this.commitsLink = tmphref + this.firstLinks.get(2); // this.settingLink = tmphref + this.otherLinks.get(0); // // this.sharingwidgetsLink = tmphref + this.otherLinks.get(1); // this.relatedprojectsLink = tmphref + this.otherLinks.get(2); // this.costestLink = tmphref + this.otherLinks.get(3); // this.contributorLink = tmphref + this.otherLinks.get(4); // } this.rateInfo = this.rateInfo.trim(); this.rateLevel = this.rateLevel.trim(); //?projectlinks?? this.projectLinks = this.projectLinks.trim(); if (this.projectLinks.contains("")) { this.projectLinks = this.projectLinks.replace("", " "); this.projectLinks = this.projectLinks.trim(); } this.activity = this.activity.trim(); SimpleDateFormat s = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss"); this.crawlerTime = s.format(page.getTime()); this.projectUrl = page.getPageUrl(); }
From source file:net.trustie.model.OpenHubProject_Model.java
private void handleQuickRef(Element quickRef) { Elements itemNames = quickRef.select("dt"); Elements itemValues = quickRef.select("dd"); Element e = null; Element eValue = null;//from w w w.j a v a 2 s . c o m for (int i = 0; i < itemNames.size(); i++) { e = itemNames.get(i); eValue = itemValues.get(i); String refName = e.text(); switch (refName) { case "Organization:": { this.organization = eValue.text(); break; } case "Project Links:": { Elements links = eValue.select("a"); String[] tmp = new String[links.size()]; Element ele = null; for (int j = 0; j < links.size(); j++) { ele = links.get(j); tmp[j] = ele.text() + Seperator.SOURCE_SEPERATOR + ele.attr("href"); } this.projectLinks = StringUtils.join(tmp, Seperator.OSSEAN_SEPERATOR); break; } case "Code Locations:": { Elements locs = eValue.select("a"); if (locs.size() == 0) { this.codeLocation = eValue.text(); } else { Element link = locs.get(0); this.codeLocation = link.text() + Seperator.SOURCE_SEPERATOR + link.attr("href"); } break; } case "Licenses:": { Elements links = eValue.select("a"); List<String> listLicenses = new ArrayList<String>(); // String[] tmp = new String[links.size()]; for (int j = 0; j < links.size(); j++) { listLicenses.add(links.get(j).text()); // tmp[j] = links.get(j).text(); } this.licenses = StringHandler.combineTags(listLicenses); break; } case "Similar Projects:": { // System.out.println(eValue); Elements projects = eValue.select("td[width=49%]"); // System.out.println(projects.size()); String[] tmp = new String[projects.size()]; Element ele = null; for (int j = 0; j < projects.size(); j++) { ele = projects.get(j); Element project = ele.select("a").get(0); tmp[j] = project.text() + Seperator.SOURCE_SEPERATOR + project.attr("href"); } this.similarProjects = StringUtils.join(tmp, Seperator.OSSEAN_SEPERATOR); break; } case "Managers:": { if ("Become the first manager for BugSystem".equals(eValue.text())) { break; } Elements users = eValue.select("a"); String[] tmp = new String[users.size()]; Element ele = null; for (int j = 0; j < users.size(); j++) { ele = users.get(j); tmp[j] = ele.text() + Seperator.SOURCE_SEPERATOR + ele.attr("href"); } this.managers = StringUtils.join(tmp, Seperator.OSSEAN_SEPERATOR); break; } default: { break; } } } }
From source file:net.trustie.model.OpenHubProject_Model.java
private void handleNutShell0(String nutshell) { Elements eles = getAElements(nutshell); Element ele = null; ele = eles.get(0);/*w ww. j av a 2s .co m*/ this.commitNum = getInt(ele.text()); ele = eles.get(1); this.contributorNum = getInt(ele.text()); ele = eles.get(2); this.codeLinesNum = getInt(ele.text()); }