html代码里面所有的链接地址和链接名称

浏览数：23 / 时间：2015年06月09日

package parser;

import org.htmlparser.Parser;

import org.htmlparser.Node;

import org.htmlparser.NodeFilter;

import org.htmlparser.Parser;

importorg.htmlparser.filters.TagNameFilter;

import org.htmlparser.tags.LinkTag;

import org.htmlparser.tags.TableTag;

import org.htmlparser.util.NodeList;

import org.htmlparser.util.ParserException;

import org.htmlparser.visitors.HtmlPage;

/**

*htmlparser取得一段html代码里面所有的链接地址和链接名称

*@author chenguoyong

public class Testhtmlparser {

/**

* @param args

publicstatic void main(String[] args) {

Stringhtmlcode ="<HTML><HEAD><TITLE>AAA</TITLE></HEAD><BODY>"

+"<a href=‘http://topic.csdn.net/u/20080522/14/0ff402ef-c382-499a-8213-ba6b2f550425.html‘>连接1</a>"

+"<a href=‘http://topic.csdn.net‘>连接2</a></BODY></HTML>";

//创建Parser对象根据传给字符串和指定的编码

Parserparser = Parser.createParser(htmlcode, "GBK");

//创建HtmlPage对象HtmlPage(Parser parser)

HtmlPagepage = new HtmlPage(parser);

try{

//HtmlPage extends visitor,Apply the given visitor to the current

//page.

parser.visitAllNodesWith(page);

}catch (ParserException e1) {

e1= null;

}

//所有的节点

NodeListnodelist = page.getBody();

//建立一个节点filter用于过滤节点

NodeFilterfilter = new TagNameFilter("A");

//得到所有过滤后，想要的节点

nodelist= nodelist.extractAllNodesThatMatch(filter, true);

for(int i = 0; i < nodelist.size(); i++) {

LinkTaglink = (LinkTag) nodelist.elementAt(i);

//链接地址

System.out.println(link.getAttribute("href")+ "\n");

//链接名称

System.out.println(link.getStringText());

}

http://c.tieba.baidu.com/p/3476776824
http://c.tieba.baidu.com/p/3476808306
http://c.tieba.baidu.com/p/3476798710
http://c.tieba.baidu.com/p/3474281354
http://c.tieba.baidu.com/p/3474300101
http://c.tieba.baidu.com/p/3474294075
http://c.tieba.baidu.com/p/3474123295
http://c.tieba.baidu.com/p/3474314242
http://c.tieba.baidu.com/p/3474310411
http://c.tieba.baidu.com/p/3474304550
http://c.tieba.baidu.com/p/3475433945
http://c.tieba.baidu.com/p/3475430015
http://c.tieba.baidu.com/p/3475433348
http://c.tieba.baidu.com/p/3475431434
http://c.tieba.baidu.com/p/3474176863
http://c.tieba.baidu.com/p/3474159835
http://c.tieba.baidu.com/p/3474163941
http://c.tieba.baidu.com/p/3474156121
http://c.tieba.baidu.com/p/3474147660
http://c.tieba.baidu.com/p/3474151899
http://c.tieba.baidu.com/p/3474142287
http://c.tieba.baidu.com/p/3474136965
http://c.tieba.baidu.com/p/3474133165
http://c.tieba.baidu.com/p/3474128675
http://c.tieba.baidu.com/p/3474103896
http://c.tieba.baidu.com/p/3474099488
http://c.tieba.baidu.com/p/3474094120
http://c.tieba.baidu.com/p/3475431976
http://c.tieba.baidu.com/p/3474267991
http://c.tieba.baidu.com/p/3474259583
http://c.tieba.baidu.com/p/3474254990
http://c.tieba.baidu.com/p/3474228986
http://c.tieba.baidu.com/p/3474221626
http://c.tieba.baidu.com/p/3474215742
http://c.tieba.baidu.com/p/3474212122
http://c.tieba.baidu.com/p/3474188883
http://c.tieba.baidu.com/p/3474207722
http://c.tieba.baidu.com/p/3474184143
http://c.tieba.baidu.com/p/3474180522
http://c.tieba.baidu.com/p/3474171022
http://c.tieba.baidu.com/p/3474086627
http://c.tieba.baidu.com/p/3462847203
http://c.tieba.baidu.com/p/3462839334
http://c.tieba.baidu.com/p/3462834294
http://c.tieba.baidu.com/p/3462786130
http://c.tieba.baidu.com/p/3462782768
http://c.tieba.baidu.com/p/3461791753
http://c.tieba.baidu.com/p/3461784215
http://c.tieba.baidu.com/p/3461778008
http://c.tieba.baidu.com/p/3461772860
http://c.tieba.baidu.com/p/3461767442
http://c.tieba.baidu.com/p/3461736231
http://c.tieba.baidu.com/p/3461704953
http://c.tieba.baidu.com/p/3461692676
http://c.tieba.baidu.com/p/3461665341
http://c.tieba.baidu.com/p/3461656389
http://c.tieba.baidu.com/p/3461660595
http://c.tieba.baidu.com/p/3461566608
http://c.tieba.baidu.com/p/3461652243
http://c.tieba.baidu.com/p/3461561596
http://c.tieba.baidu.com/p/3461557067