使用Jsoup抓取携程旅游网全国城市信息

package step4;
import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
public class Task {

    //通过filePath文件路径获取Docment对象
    public Document getDoc(String filePath) throws IOException{
        /**********   Begin   **********/


        File file=new File("backups/hotels.ctrip.com_domestic-city-hotel.txt");
        Document doc=Jsoup.parse(file,"UTF-8","http://hotels.ctrip.com/");
        return doc;
        /**********   End   **********/
    }

    /**
     * 获取所有城市返回城市信息集合
     * @param doc
     * @return
     */
    public List<HotelCity> getAllCitys(Document doc){
        /**********   Begin   **********/



        List<HotelCity> cities = new ArrayList<HotelCity>();

        Elements aa= doc.getElementsByClass("pinyin_filter_detail layoutfix");
        Element pp = aa.first();
        Elements hh= pp.getElementsByTag("dd");
        Elements hts=pp.getElementsByTag("dt");

       for (int i = 0; i < hh.size(); i++) {
            Element bb = hts.get(i);
            Element head_hotelsLink = hh.get(i);
            Elements links = head_hotelsLink.children();

            for (Element link : links) {
                String pinyin_cityId = link.attr("href").replace("/hotel/", "");
                String pinyin = pinyin_cityId.replace(StringUtil.getNumbers(link.attr("href")), "");//截取拼音
                HotelCity city = new HotelCity();
                city.setCityId(StringUtil.getNumbers(link.attr("href"))); //截取cityId
                city.setCityName(link.text());
                city.setHeadPinyin(bb.text());
                city.setPinyin(pinyin);
                cities.add(city);
            }
        }
        return cities;
        /**********   End   **********/
    }
}