Beispiel #1
0
  public Holder doParse(String html, String url) {
    Holder holder = new Holder();
    holder.url = url;

    Document doc = Jsoup.parse(html, url);
    Elements typeElement =
        doc.select("body > div.main_w.clearfix > div.main.clearfix > ul > li:nth-child(5) > a");
    holder.dishType = typeElement.text();

    Elements titleElement =
        doc.select(
            "body > div.main_w.clearfix > div.main.clearfix > div.cp_header.clearfix > div.cp_main_info_w > div.info1 > h1 > a");

    holder.title = titleElement.text();

    Elements methodElement =
        doc.select(
            "body > div.main_w.clearfix > div.main.clearfix > div.cp_header.clearfix > div.cp_main_info_w > div.info2 > ul > li:nth-child(1) > a");
    holder.method = methodElement.text();

    Elements materialElement =
        doc.select(
            "body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix > div.cp_body_left > div.materials > div > div.yl.zl.clearfix > ul > li > div > h4 > a");

    holder.mainMaterial = materialElement.text();

    Elements stepE =
        doc.select(
            "body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix > div.cp_body_left > div.measure > div.editnew.edit > div.content.clearfix");
    //
    // body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix
    // > div.cp_body_left > div.measure > div.editnew.edit >
    // div.content.clearfix
    // body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix
    // > div.cp_body_left > div.measure > div.edit > p:nth-child(1) > em
    //

    if (stepE.size() == 0) {
      stepE =
          doc.select(
              "body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix > div.cp_body_left > div.measure > div.edit > p");
    }

    for (int i = 0; i < stepE.size(); i++) {
      Element e = stepE.get(i);

      if (e.children().hasClass("step")) {
        String step = e.text();
        if (!"".equals(step)) {
          holder.steps.add(step);
        }
      }
    }
    // body > div.main_w.clearfix > div.main.clearfix > div.cp_body.clearfix
    // > div.cp_body_left > div.measure > div.editnew.edit >
    // div:nth-child(1)

    return holder;
  }