JSOUP是偶然看到的一个处理HTML的JAVA 类库,其官方网址是:http://jsoup.org/
1、编写相关的试用程序(只需要在工程中引用jsoup-1.3.3.jar即可):
import java.io.File;
import java.io.IOException;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.select.Elements;
public class Test {
public static void main(String[] args) {
Test t = new Test();
t.parseFile();
}
public void parseString() {
String html = "<html><head><title>blog</title></head><body οnlοad='test()'><p>Parsed HTML into a doc.</p></body></html>";
Document doc = Jsoup.parse(html);
System.out.println(doc);
Elements es = doc.body().getAllElements();
System.out.println(es.attr("onload"));
System.out.println(es.select("p"));
}
public void parseUrl() {
try {
Document doc = Jsoup.connect("http://www.baidu.com/").get();
Elements hrefs = doc.select("a[href]");
System.out.println(hrefs);
System.out.println("------------------");
System.out.println(hrefs.select("[href^=http]"));
} catch (IOException e) {
e.printStackTrace();
}
}
public void parseFile() {
try {
File input = new File("input.html");
Document doc = Jsoup.parse(input, "UTF-8");
// 提取出所有的编号
Elements codes = doc.body().select("td[title^=IA] > a[href^=javascript:view]");
System.out.println(codes);
System.out.println("------------------");
System.out.println(codes.html());
} catch (IOException e) {
e.printStackTrace();
}
}
}
2、parseString的输出:
<html> <head> <title>blog</title> </head> <body οnlοad="test()"> <p>Parsed HTML into a doc.</p> </body> </html> test() <p>Parsed HTML into a doc.</p>
3、parseUrl的输出:
<a href="/gaoji/preferences.html">设置</a> <a href="http://passport.baidu.com/?login&tpl=mn">登录</a> <a href="http://news.baidu.com">新 闻</a> <a href="http://tieba.baidu.com">贴 吧</a> <a href="http://zhidao.baidu.com">知 道</a> <a href="http://mp3.baidu.com">MP3</a> <a href="http://image.baidu.com">图 片</a> <a href="http://video.baidu.com">视 频</a> <a href="http://map.baidu.com">地 图</a> <a href="#" name="ime_hw">手写</a> <a href="#" name="ime_py">拼音</a> <a href="#" name="ime_cl">关闭</a> <a href="http://hi.baidu.com">空间</a> <a href="http://baike.baidu.com">百科</a> <a href="http://www.hao123.com">hao123</a> <a href="/more/">更多>></a> <a id="st" οnclick="this.style.behavior='url(#default#homepage)';this.setHomePage('http://www.baidu.com')" href="http://utility.baidu.com/traf/click.php?id=215&url=http://www.baidu.com">把百度设为主页</a> <a href="http://e.baidu.com/?refer=888">加入百度推广</a> <a href="http://top.baidu.com">搜索风云榜</a> <a href="http://home.baidu.com">关于百度</a> <a href="http://ir.baidu.com">About Baidu</a> <a href="/duty/">使用百度前必读</a> <a href="http://www.miibeian.gov.cn" target="_blank">京ICP证030173号</a> ------------------ <a href="http://passport.baidu.com/?login&tpl=mn">登录</a> <a href="http://news.baidu.com">新 闻</a> <a href="http://tieba.baidu.com">贴 吧</a> <a href="http://zhidao.baidu.com">知 道</a> <a href="http://mp3.baidu.com">MP3</a> <a href="http://image.baidu.com">图 片</a> <a href="http://video.baidu.com">视 频</a> <a href="http://map.baidu.com">地 图</a> <a href="http://hi.baidu.com">空间</a> <a href="http://baike.baidu.com">百科</a> <a href="http://www.hao123.com">hao123</a> <a id="st" οnclick="this.style.behavior='url(#default#homepage)';this.setHomePage('http://www.baidu.com')" href="http://utility.baidu.com/traf/click.php?id=215&url=http://www.baidu.com">把百度设为主页</a> <a href="http://e.baidu.com/?refer=888">加入百度推广</a> <a href="http://top.baidu.com">搜索风云榜</a> <a href="http://home.baidu.com">关于百度</a> <a href="http://ir.baidu.com">About Baidu</a> <a href="http://www.miibeian.gov.cn" target="_blank">京ICP证030173号</a>
3、parseFile的输出:
<a href="javascript:view('67530','67530','0');">IA100908-002</a> <a href="javascript:view('67529','67529','0');">IA100908-001</a> <a href="javascript:view('67544','67544','0');">IA100908-016</a> <a href="javascript:view('67364','67364','0');">IA100903-008</a> <a href="javascript:view('67363','67363','0');">IA100903-007</a> <a href="javascript:view('66104','66104','0');">IA100710-013</a> <a href="javascript:view('57916','57916','0');">IA100515-013</a> <a href="javascript:view('56962','56962','0');">IA100430-022</a> <a href="javascript:view('66958','66958','0');">IA100830-001</a> <a href="javascript:view('66319','66319','0');">IA100713-003</a> <a href="javascript:view('66317','66317','0');">IA100713-001</a> <a href="javascript:view('66321','66321','0');">IA100713-005</a> <a href="javascript:view('66967','66967','0');">IA100830-010</a> <a href="javascript:view('66999','66999','0');">IA100831-001</a> <a href="javascript:view('67377','67377','0');">IA100904-004</a> <a href="javascript:view('67378','67378','0');">IA100904-005</a> <a href="javascript:view('3271','3271','0');">IA080115-031</a> ------------------ IA100908-002 IA100908-001 IA100908-016 IA100903-008 IA100903-007 IA100710-013 IA100515-013 IA100430-022 IA100830-001 IA100713-003 IA100713-001 IA100713-005 IA100830-010 IA100831-001 IA100904-004 IA100904-005 IA080115-031
补充下,input.html的基本结果如图: