2021.8.6笔记 jsoup

jsoup对象

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.io.File;
import java.io.IOException;
import java.net.URL;

public class xmlJsoupPra {
    public static void main(String[] args) throws IOException {

        /**
         * 1.parse(文件（需要传入path）, 编码方式)
         */
        String path = xmlJsoupPra.class.getClassLoader().getResource("student.xml").getPath();
        Document document = Jsoup.parse(new File(path), "utf-8");
        Elements eles = document.getElementsByTag("name");
        Element element = eles.get(0);
        String text = element.text();
        System.out.println(text);
        /**
         * 2.String
         */
        String str = "<?xml version=\"1.0\" encoding=\"utf-8\" ?>\n" +
                "<students>\n" +
                "    <studnt number=\"1001\">\n" +
                "        <name>zhangsan</name>\n" +
                "        <age>18</age>\n" +
                "        <grender>male</grender>\n" +
                "    </studnt>\n" +
                "    <studnt number=\"1002\">\n" +
                "        <name>lisi</name>\n" +
                "        <age>15</age>\n" +
                "        <grender>female</grender>\n" +
                "    </studnt>\n" +
                "</students>";
        Document parse1 = Jsoup.parse(str);
        System.out.println(parse1);

        /**
         * 3.url
         */
        URL url = new URL("https://www.baidu.com");
        Document parse2 = Jsoup.parse(url, 10000);
        System.out.println(parse2);
    }
}

Document对象

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.io.File;
import java.io.IOException;

public class DocumentObj {
    public static void main(String[] args) throws IOException {
        String path = DocumentObj.class.getClassLoader().getResource("student.xml").getPath();
        Document docu1 = Jsoup.parse(new File(path), "utf-8");

        Elements ele1 = docu1.getElementsByAttribute("number");
        Element element = ele1.get(0);
        System.out.println(element);

        Elements ele2 = docu1.getElementsByAttributeValue("number", "1001");
        System.out.println(ele2);

        Element ele3 = docu1.getElementById("i5");
        System.out.println(ele3);
    }
}

Element对象

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.io.File;
import java.io.IOException;

public class ElementObj {
    public static void main(String[] args) throws IOException {
        String path = ElementObj.class.getClassLoader().getResource("student.xml").getPath();
        Document docu1 = Jsoup.parse(new File(path), "utf-8");

        Element ele1 = docu1.getElementsByTag("student").get(0);

        Elements ele2 = ele1.getElementsByTag("name");
        System.out.println(ele2);

        String num = ele1.attr("number");
        System.out.println(num);

        Element age = ele1.getElementsByTag("age").get(0);
        String text = age.text();
        System.out.println(text);
        String htmlAge = age.html();
        System.out.println(htmlAge);
    }
}

根据选择器查询

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.select.Elements;

import java.io.File;
import java.io.IOException;

public class SelectObj {
    public static void main(String[] args) throws IOException {
        String path = SelectObj.class.getClassLoader().getResource("student.xml").getPath();
        Document docu = Jsoup.parse(new File(path), "utf-8");

        Elements nameEle = docu.select("name");
        System.out.println(nameEle);

        Elements IdEle= docu.select("#i5");
        System.out.println(IdEle);

        Elements newEle = docu.select("name[id=\"i5\"]");
        System.out.println(newEle);

        Elements newEle2 = docu.select("student[number=\"1001\"] age"); //>的话是一级子标签
        System.out.println(newEle2);
    }
}