java网页数据抓取源代码(抓取电话和身份证信息为例）

说明：源链接的代码是会报错，这个代码是修改过确定可以运行使用的。
对于加密的网站还没去研究，不知道能不能抓取，现在只是对一些没有加密的网站进行网页数据抓取。刚刚开始写的时候以为很多网站都能抓取，但是发现很多都加密了，本来以为一些地址可以通过网页数据检测工具测出他的数据变化，但是只能监测到一些通过js显示的数据，依然不能抓取到加密的网站。嗨，这个问题以后再说吧。

import java.net.* ;

import java.io.* ;

import java.util.regex.* ;

public class Capture{
public static void main(String args[])throws Exception{
System.out.println("*************************手机号查询************************") ;
System.out.println("我的位置是：" + new GrabMobile().grabMobileLocation("18323483580")) ;
System.out.println("手机卡类型是：" + new GrabMobile().grabMobileType("18323483580")) ;
System.out.println("我的邮编是：" + new GrabMobile().grabMobilePost("18323483580")) ;
System.out.println("*************************身份证查询************************") ;
System.out.println("我的性别是：" + new GrabIdentity().grabIdentitySex("362203199202243575")) ;
System.out.println("我的生日是：" + new GrabIdentity().grabIdentityBirth("362203199202243575")) ;
System.out.println("我的家乡是：" + new GrabIdentity().grabIdentityHome("362203199202243575")) ;
}

}

//1):抓取手机查询信息

class GrabMobile{

//1，手机归属地
public String grabMobileLocation(String m)throws Exception{
String strUrl = "http://www.ip138.com:8080/search.asp?action=mobile&mobile=" + m;
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = "卡号归属地" ;
String strEnd = "卡 类 型";
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;

int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+42,end-33) ;
result = drawChMob(result) ;
return result ;
}

//2，卡号类型
public String grabMobileType(String m)throws Exception{
String strUrl = "http://www.ip138.com:8080/search.asp?action=mobile&mobile=" + m;
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = "卡 类 型" ;
String strEnd = "<TD align=\"center\">区 号</TD>";
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;

int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+12,end) ;
result = drawChMob(result) ;
result = result.substring(1) ;
return result ;
}

//3， 邮编
public String grabMobilePost(String m)throws Exception{
String strUrl = "http://www.ip138.com:8080/search.asp?action=mobile&mobile=" + m;
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = "邮 编" ;
String strEnd = "更详细的..";
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;

int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+40,end-55) ;
return result ;
}

//正则
public String drawChMob(String str){
StringBuffer strBuf = new StringBuffer() ;
String regex="([\u4e00-\u9fa5]+)";
Matcher matcher = Pattern.compile(regex).matcher(str);
while(matcher.find()){
strBuf.append(matcher.group(0)).toString() ;
}
return strBuf.toString() ;
}

}

//2):抓取身份证查询信息

class GrabIdentity{

//1，性别
public String grabIdentitySex(String userid)throws Exception{
String strUrl = "http://qq.ip138.com/idsearch/index.asp?action=idcard&userid=" + userid + "&B1=%B2%E9+%D1%AF";
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = " 别" ;
String strEnd = "出生日期";
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;

int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+7,end) ;
result = drawCh(result) ;
return result ;
}

//2，出生日期
public String grabIdentityBirth(String userid)throws Exception{
String strUrl = "http://qq.ip138.com/idsearch/index.asp?action=idcard&userid=" + userid + "&B1=%B2%E9+%D1%AF";
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = "出生日期：</td><td class=\"tdc2\">" ;
String strEnd = "</td><tr><tr><td class=";
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;
int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+27,end) ;
return result ;
}

//3，身份证归属地
public String grabIdentityHome(String userid)throws Exception{
String strUrl = "http://qq.ip138.com/idsearch/index.asp?action=idcard&userid=" + userid + "&B1=%B2%E9+%D1%AF";
URL url = new URL(strUrl) ;
HttpURLConnection httpUrlCon = (HttpURLConnection)url.openConnection() ;
InputStreamReader inRead = new InputStreamReader(httpUrlCon.getInputStream(),"GBK") ;
BufferedReader bufRead = new BufferedReader(inRead) ;
StringBuffer strBuf = new StringBuffer() ;
String line = "" ;
while ((line = bufRead.readLine()) != null) {
strBuf.append(line);
}
String strStart = "证 地：</td><td class=\"tdc2\">" ;
String strEnd = "<br/></td></tr><tr><td class=\"tdc3\" valign=\"top\" align=\"right\">部分或" ;
String strAll = strBuf.toString() ;

int start = strAll.indexOf(strStart) ;
int end = strAll.indexOf(strEnd) ;

String result = strAll.substring(start+31,end) ;
return result ;
}

//正则
public String drawCh(String str){
StringBuffer strBuf = new StringBuffer() ;
String regex="([\u4e00-\u9fa5]+)";
Matcher matcher = Pattern.compile(regex).matcher(str);
if(matcher.find()){
str = strBuf.append(matcher.group(0)).toString() ;
}
return str ;
}

}

源码：http://www.2cto.com/kf/201211/166624.html