Java Elements.select方法代码示例

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Map<ContentsActivity.parameter, Object> getContent(String baseUrl, String currentUrl, byte[] result, Map<ContentsActivity.parameter, Object> resultMap) throws UnsupportedEncodingException {
    List<AlbumInfo> urls = new ArrayList<>();
    Document document = Jsoup.parse(new String(result, "gbk"));
    Elements elements = document.select("#sliding li");
    for (Element element : elements) {
        AlbumInfo temp = new AlbumInfo();

        Elements title = element.select(".p-title");
        if (title.size() > 0)
            temp.setTitle(title.get(0).text());

        Elements album = element.select("a:has(img)");
        temp.setAlbumUrl(baseUrl + album.attr("href"));
        Elements pic = album.select("img");
        if (pic.size() > 0)
            temp.setPicUrl(pic.get(0).attr("src"));
        urls.add(temp);
    }
    resultMap.put(ContentsActivity.parameter.CURRENT_URL, currentUrl);
    resultMap.put(ContentsActivity.parameter.RESULT, urls);
    return resultMap;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Result process(Request request, Page page) {
    Result result = new Result();

    Elements elements = page.document().select("div.mainPanel table.table tr");

    List<Map> list = new ArrayList<Map>();
    for (Element element : elements) {
        Elements tdElements = elements.select("td");
        if (tdElements.size() > 5) {
            Map<String, String> info = new HashMap<String, String>();

            info.put("ip", tdElements.get(0).text());
            info.put("port", tdElements.get(1).text());
            info.put("country", tdElements.get(2).text());
            info.put("anonymous", tdElements.get(3).text());
            info.put("type", tdElements.get(4).text());

            list.add(info);
        }
    }

    result.put("list", list);
    return result;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
public HashMap<String, ArrayList<String>> getSynonyms() throws IOException, InterruptedException {
    if (synonyms.isEmpty()) {
        for (String word : wordList) {
            try {
                Document doc = Jsoup.connect("http://www.thesaurus.com/browse/" + word)
                        .userAgent("Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0")
                        .timeout(3000)
                        .get();
                Elements block = doc.getElementsByClass("relevancy-block");
                Elements list = block.select(".relevancy-list");
                Elements text = list.select(".text");
                ArrayList<String> synonymList = new ArrayList<>();
                for (int j = 0; j < text.size(); j++) {
                    synonymList.add(text.get(j).text());
                }
                synonyms.put(word, synonymList);
            } catch (IOException e) {
            }
        }
    }
    return synonyms;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private ArrayList<String> retrieveSimilarImageFromHTML(Elements body) {
    ArrayList<String> imageRes = new ArrayList<String>();
    Elements images = body.select("div.rg_meta");

    Gson gson = new Gson();
    JsonObject jsImage;
    String imageUrl;
    for (Element image : images){

        jsImage = gson.fromJson(image.text(), JsonObject.class);
        imageUrl = jsImage.get("ou").getAsString();
        try {
            new URL(imageUrl);
            imageRes.add(imageUrl);
        } catch (MalformedURLException e) {
            printMalformedError(e);
        }
    }
    return imageRes;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Map<ContentsActivity.parameter, Object> getContent(String baseUrl, String currentUrl, byte[] result, Map<ContentsActivity.parameter, Object> resultMap) throws UnsupportedEncodingException {
    List<AlbumInfo> urls = new ArrayList<>();
    Document document = Jsoup.parse(new String(result, "utf-8"));
    Elements elements = document.select("div.album");
    for (Element element : elements) {
        AlbumInfo temp = new AlbumInfo();

        Elements title = element.select("span.name");
        if (title.size() > 0)
            temp.setTitle(title.get(0).text());

        Elements album = element.select(".pic_box a");
        temp.setAlbumUrl(album.attr("href"));
        Elements pic = album.select("img");
        if (pic.size() > 0)
            temp.setPicUrl(pic.get(0).attr("src"));
        urls.add(temp);
    }
    resultMap.put(ContentsActivity.parameter.CURRENT_URL, currentUrl);
    resultMap.put(ContentsActivity.parameter.RESULT, urls);
    return resultMap;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Map<ContentsActivity.parameter, Object> getContent(String baseUrl, String currentUrl, byte[] result, Map<ContentsActivity.parameter, Object> resultMap) throws UnsupportedEncodingException {
    List<AlbumInfo> data = new ArrayList<>();
    Document document = Jsoup.parse(new String(result, "utf-8"));
    Elements elements = document.select("section");
    for (Element element : elements) {
        AlbumInfo temp = new AlbumInfo();

        Elements album = element.select("a:has(img)");
        if (album.size() > 0) {
            temp.setAlbumUrl(album.attr("href"));
            Elements pic = album.select("img");
            if (pic.size() > 0)
                temp.setPicUrl(pic.get(0).attr("data-src"));
        }

        data.add(temp);
    }

    resultMap.put(ContentsActivity.parameter.CURRENT_URL, currentUrl);
    resultMap.put(ContentsActivity.parameter.RESULT, data);
    return resultMap;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private ArrayList<String> retrieveDescriptionFromHTML(Elements body) {
    ArrayList<String> descRes = new ArrayList<String>();
    Elements descriptions = body.select("span.st");

    for (Element desc : descriptions)
        descRes.add(desc.text());

    return descRes;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public void run() {
    Document doc;

    try {
        doc = get(mUrl);
    } catch (IOException e) {
        e.printStackTrace();
        failedOnUI(e.getMessage());
        return;
    }

    Elements userProfileElements = doc.select("div.user-page .profile");
    if (userProfileElements.isEmpty()) {
        failedOnUI("获取用户资料失败");
        return;
    }

    Elements headerElements = userProfileElements.select("div.ui-header");
    if (headerElements.isEmpty()) {
        failedOnUI("获取用户资料失败");
        return;
    }

    UserProfile profile = new UserProfile();

    profile.setUrl(mUrl);
    profile.setAvatar(headerElements.select("img.avatar").attr("src"));
    profile.setUsername(headerElements.select("div.username").text());
    profile.setNumber(headerElements.select("div.user-number .number").text());
    profile.setSince(headerElements.select("div.user-number .since").text());

    if (profile.isValid()) {
        successOnUI(profile);
    } else {
        failedOnUI("获取用户资料出错");
    }
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
public static Map<Integer, Comment> getCommentsFromElements(Elements elements) {
    TreeMap<Integer, Comment> comments = new TreeMap<>();

    Elements replyItems = elements.select("div.reply-item");

    for (Element replyItem : replyItems) {
        Comment comment = getCommentFromElement(replyItem);
        comments.put(comment.getMeta().getFloor(), comment);
    }

    return comments.descendingMap();
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private void setData(List<FictionModel> list, int eq, String type) {
    Elements select = document.select("div.blockcontent").eq(eq);
    Elements itemElements = select.select("li");
    list.add(getMoreTitle(type, select));
    FictionModel fictionModel;
    for (Element element : itemElements) {
        fictionModel = new FictionModel();
        Elements a = element.select("a");
        fictionModel.title = a.text();
        fictionModel.detailUrl = a.attr("abs:href");
        fictionModel.type = TYPE_ITEM;
        list.add(fictionModel);
    }
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private void readStepContent(Elements elements) {
  Elements contentElements = elements.select(DESIGN_TAG_CONTENT);
  if (!contentElements.isEmpty()) {
    Element contentElement = contentElements.get(0);
    Elements children = contentElement.children();

    if (children.size() > 1) {
      String msg = "Only one child allowed for tag <" + DESIGN_TAG_CONTENT + ">";
      throw new IllegalArgumentException(msg);
    }

    Component component = designContext.readDesign(children.first());
    setContent(component);
  }
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Map<ContentsActivity.parameter, Object> getContent(String baseUrl, String currentUrl, byte[] result, Map<ContentsActivity.parameter, Object> resultMap) throws UnsupportedEncodingException {
    List<AlbumInfo> urls = new ArrayList<>();
    Document document = Jsoup.parse(new String(result, "gb2312"));
    Elements elements = document.select(".piclist li");
    for (Element element : elements) {
        AlbumInfo temp = new AlbumInfo();

        Elements title = element.select("a:has(img) span");
        if (title.size() > 0)
            temp.setTitle(title.get(0).text());

        Elements time = element.select(".b1");
        if (time.size() > 0)
            temp.setTime(time.get(0).text());

        Elements album = element.select("a:has(img)");
        temp.setAlbumUrl(baseUrl + album.attr("href"));
        Elements pic = album.select("img");
        if (pic.size() > 0)
            temp.setPicUrl(pic.get(0).attr("src"));
        urls.add(temp);
    }
    resultMap.put(ContentsActivity.parameter.CURRENT_URL, currentUrl);
    resultMap.put(ContentsActivity.parameter.RESULT, urls);
    return resultMap;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private ArrayList<String> retrieveTitleFromHTML(Elements body) {
    ArrayList<String> titleRes = new ArrayList<String>();
    Elements titles = body.select("h3.r");

    for (Element title : titles)
        titleRes.add(title.text());

    return titleRes;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
@Override
public Map<ContentsActivity.parameter, Object> getContent(String baseUrl, String currentUrl, byte[] result, Map<ContentsActivity.parameter, Object> resultMap) throws UnsupportedEncodingException {
    List<AlbumInfo> data = new ArrayList<>();
    Document document = Jsoup.parse(new String(result, "utf-8"));
    Elements elements = document.select("section");
    for (Element element : elements) {
        AlbumInfo temp = new AlbumInfo();

        Elements album = element.select(".card-bg > a:has(img)");
        if (album.size() > 0) {
            temp.setAlbumUrl(album.attr("href"));
            Elements pic = album.select("img");
            if (pic.size() > 0) {
                Log.e("Acg12", "getContent: " + pic.get(0).attr("data-src"));
                temp.setPicUrl(pic.get(0).attr("data-src"));
            }
        }

        Elements title = element.select("h3.title");
        if (title.size() > 0)
            temp.setTitle(title.get(0).text());

        Elements time = element.select("time");
        if (time.size() > 0)
            temp.setTime(time.attr("title") + " " + time.text());

        data.add(temp);
    }

    resultMap.put(ContentsActivity.parameter.CURRENT_URL, currentUrl);
    resultMap.put(ContentsActivity.parameter.RESULT, data);
    return resultMap;
}
 

import org.jsoup.select.Elements; //导入方法依赖的package包/类
private String retrieveBestGuessFromHTML(Elements body) {
    String bestGuessRes;
    Elements bestGuesses = body.select("a._gUb");
    bestGuessRes = bestGuesses.first().text();
    return bestGuessRes;
}