Xpath和JSoup

1.Jsoup

JSON有两种获取方式
第一种：

Document document = Jsoup.parse(file,"UTF-8");

第二种

Document document = Jsoup.connect(url);

2.JSoupUtils

public class GetHtmlUtils {
    /**
    * @Description: 通过JSoup获取到新闻列表html
    * @Param: [pathname] html路径
    * @return: java.lang.String  返回html的字符串
    * @Author: 董云川
    * @Date: 2021/7/28
    */
    public static String getHtmlByJSoup(String pathname) throws IOException {
        File file = new File(pathname);
        Document document = Jsoup.parse(file,"UTF-8");
        String html = document.toString();
//        System.out.println(html);
        return html;
//        Element elementse = document.getElementById("copyrightId");
//        System.out.println(elementse.text());
    }

    public static Document getHtmlByURL(String pathname) throws IOException {
        //设置连接时长，防止不断请求
        Document document = Jsoup.connect(pathname)
                .timeout(8000)//超时时间
                .get();;
        return document;
    }

}