java正则表达式爬图片_利用正则表达式爬取网页图片 java

importjava.io.File;importjava.io.FileOutputStream;importjava.io.IOException;importjava.io.InputStream;importjava.net.URL;importjava.net.URLConnection;importjava.util.ArrayList;importjava.util.List;importjava.util.regex.Matcher;importjava.util.regex.Pattern;public classCatchImg {static String url="你想爬取得网址";//匹配的Url正则

static String url_reg="]*?>";//匹配Src的正则

static String src_reg="[a-zA-z]+://[^\\s]*";public static void main(String[] args) throwsException

{//获取HTML内容

String html=getHtml(url);//获取Url

List listUrl=getUrl(html);//获取Src

List listSrc=getSrc(listUrl);//下载到本地

Download(listSrc);

}public static String getHtml(String url) throwsException

{

URL uri=newURL(url);

URLConnection connection=uri.openConnection();

InputStream in=connection.getInputStream();

StringBuffer sb=newStringBuffer();byte[] buf=new byte[1024];int len=0;while((len=in.read(buf))!=-1)

{//这个地方要注意，如果用BufferedReader按line读取的话，每行后面要加上append("/n")

sb.append(new String(buf,0,len));

}

in.close();returnsb.toString();

}public static ListgetUrl(String html)

{

Matcher matcher=Pattern.compile(url_reg).matcher(html);

List listUrl=new ArrayList();while(matcher.find())

{

listUrl.add(matcher.group());

}returnlistUrl;

}public static List getSrc(ListlistUrl )

{

List listSrc=new ArrayList();for(String temp:listUrl)

{

Matcher matcher=Pattern.compile(src_reg).matcher(temp);while(matcher.find())

{//如果吧下面代码改成listSrc.add(matcher.group()),因为下面代码是把最后一个字符去掉的

listSrc.add(matcher.group().substring(0,matcher.group().length() - 1));

}

}returnlistSrc;

}private static void Download(ListlistImgSrc) {int count=0;try{for(String url : listImgSrc)

{

String imageName= count+".JPG";

URL uri= newURL(url);

InputStream in=uri.openStream();

FileOutputStream fo= new FileOutputStream(new File(imageName));//文件输出流

byte[] buf = new byte[1024];int length = 0;

System.out.println("开始下载:" +url);while ((length = in.read(buf)) != -1) {

fo.write(buf,0, length);

}

in.close();

fo.close();

System.out.println(imageName+ "下载完成");

count++;

}

}catch(Exception e) {

System.out.println("下载失败");

}

}

}