1.首先下载HtmlUnit的jar包，导入。

package com.shu.htmlUnit;

import com.gargoylesoftware.htmlunit.BrowserVersion;

import com.gargoylesoftware.htmlunit.WebClient;

import com.gargoylesoftware.htmlunit.html.HtmlPage;

public class HtmlUnit01 {

public static void main(String[] args) throws Exception{

String url = “http://www.baidu.com”;

String str;

//创建一个webClient,模拟浏览器

//WebClient webClient = new WebClient();

//使用FireFox读取网页

WebClient webClient = new WebClient(BrowserVersion.FIREFOX_38);

//使用Chrome读取网页

//WebClient webClient = new WebClient(BrowserVersion.CHROME);

//打开的话，就是执行javaScript/Css

webClient.getOptions().setJavaScriptEnabled(false);

webClient.getOptions().setCssEnabled(false);

//获取页面

HtmlPage page = webClient.getPage(url);

//获取页面的title

str = page.getTitleText();

System.out.println(“Title:——“+str);

//获取页面的XML代码

str = page.asXml();

System.out.println(“Xml:——“+str);

//获取页面的文本

str = page.asText();

System.out.println(“Text:——“+str);

//关闭webClient

webClient.close();

}

}

package com.shu.htmlUnit;

import java.io.IOException;

import java.net.MalformedURLException;

import com.gargoylesoftware.htmlunit.WebClient;

import com.gargoylesoftware.htmlunit.html.HtmlInput;

import com.gargoylesoftware.htmlunit.html.HtmlPage;

/**

* 找到页面中特定的元素

* @author xiangni123

*

*/

public class HtmlUnit02 {

public static void main(String[] args) throws Exception{

String url = “http://www.baidu.com”;

String str;

//创建一个webClient,模拟浏览器

WebClient webClient = new WebClient();

//打开的话，就是执行javaScript/Css

webClient.getOptions().setJavaScriptEnabled(false);

webClient.getOptions().setCssEnabled(false);

//获取页面

HtmlPage page = webClient.getPage(url);

//通过id获得“百度一下”按钮.(在此页面中改按钮的id为”su”)

HtmlInput btn = (HtmlInput)page.getHtmlElementById(“su”);

System.out.println(btn.getDefaultValue());

//关闭webClient

webClient.close();

}

}

package com.shu.htmlUnit;

import java.util.List;

import com.gargoylesoftware.htmlunit.BrowserVersion;

import com.gargoylesoftware.htmlunit.WebClient;

import com.gargoylesoftware.htmlunit.html.HtmlDivision;

import com.gargoylesoftware.htmlunit.html.HtmlInput;

import com.gargoylesoftware.htmlunit.html.HtmlPage;

public class HtmlUnit03 {

public static void main(String[] args) throws Exception{

String url = “http://www.baidu.com”;

String str;

//创建一个webClient,模拟浏览器

WebClient webClient = new WebClient();

//打开的话，就是执行javaScript/Css

webClient.getOptions().setJavaScriptEnabled(false);

webClient.getOptions().setCssEnabled(false);

//获取页面

HtmlPage page = webClient.getPage(url);

//查找所有div

System.out.println(“查找所有div—————“);

List dList = (List) page.getByXPath(“//div”);

for(HtmlDivision d: dList){

System.out.println(d.toString());

}

//查找第一个div

System.out.println(“查找第一个div————–“);

HtmlDivision d = (HtmlDivision)dList.get(0);

System.out.println(d.toString());

//查找所有的input

System.out.println(“查找并获取特定的input”);

List> iList = page.getByXPath(“//input[@id=’su’]”);

HtmlInput input = (HtmlInput)iList.get(0);

System.out.println(input.toString());

//关闭webClient

webClient.close();

}

}

package com.shu.htmlUnit;

import java.util.List;

import com.gargoylesoftware.htmlunit.BrowserVersion;

import com.gargoylesoftware.htmlunit.WebClient;

import com.gargoylesoftware.htmlunit.html.HtmlDivision;

import com.gargoylesoftware.htmlunit.html.HtmlInput;

import com.gargoylesoftware.htmlunit.html.HtmlPage;

public class HtmlUnit04 {

public static void main(String[] args) throws Exception{

String url = “http://www.baidu.com”;

String str;

//创建一个webClient,模拟浏览器

WebClient webClient = new WebClient();

//打开的话，就是执行javaScript/Css

webClient.getOptions().setJavaScriptEnabled(false);

webClient.getOptions().setCssEnabled(false);

//获取页面

HtmlPage page = webClient.getPage(url);

//获取搜索输入框并提交搜索内容

System.out.println(“获取搜索输入框并提交搜索内容————“);

HtmlInput input = (HtmlInput)page.getHtmlElementById(“kw”);

System.out.println(input.toString());

input.setValueAttribute(“传世”);

System.out.println(input.toString());

//获取搜索按钮并点击

System.out.println(“获取搜索按钮并点击——————-“);

HtmlInput btn = (HtmlInput)page.getHtmlElementById(“su”);

HtmlPage page2 = btn.click();

//输出新页面文本

System.out.println(“输出新页面文本———————-“);

System.out.println(page2.asText());

//关闭webClient

webClient.close();

}

}