apache HttpClient 4.3.4自动登录并抓取中国联通网页用户基本信息和账单数据

001

package

com.httpclient.asm.demo;

002

003

import

org.apache.http.HttpEntity;

004

import

org.apache.http.HttpResponse;

005

import

org.apache.http.client.CookieStore;

006

import

org.apache.http.client.HttpClient;

007

import

org.apache.http.client.methods.CloseableHttpResponse;

008

import

org.apache.http.client.methods.HttpGet;

009

import

org.apache.http.client.methods.HttpPost;

010

import

org.apache.http.cookie.Cookie;

011

import

org.apache.http.impl.client.*;

012

import

org.apache.http.util.EntityUtils;

013

014

import

javax.swing.*;

015

import

java.io.BufferedReader;

016

import

java.io.File;

017

import

java.io.FileOutputStream;

018

import

java.io.InputStream;

019

import

java.io.InputStreamReader;

020

021

public

class

LoginChinaUnicomWithCaptcha{

022

023

/**

024

*登录并抓取中国联通数据

025

*带验证码登录

026

*@authorEdson.di

027

*@date2015年3月4日

028

*@version1.0

029

*@throwsIOException

030

*/

031

public

static

void

main(Stringargs[])

throws

Exception{

032

033

Stringname=

"联通手机号"

;

034

Stringpwd=

"手机服务密码"

;

035

036

//生成验证码的链接

037

StringcreateCaptchaUrl=

"http://uac.10010.com/portal/Service/CreateImage"

;

038

HttpClienthttpClient=

new

DefaultHttpClient();

039

040

//这里可自定义所需要的cookie

041

CookieStorecookieStore=

new

BasicCookieStore();

042

043

CloseableHttpClienthttpclient=HttpClients.custom()

044

.setDefaultCookieStore(cookieStore)

045

.build();

046

047

//getcaptcha,获取验证码

048

HttpGetcaptchaHttpGet=

new

HttpGet(createCaptchaUrl);

049

HttpResponsecapthcaResponse=httpClient.execute(captchaHttpGet);

050

051

if

(capthcaResponse.getStatusLine().getStatusCode()==

200

){

052

//将验证码写入本地

053

saveToLocal(capthcaResponse.getEntity(),

"chinaunicom.capthca."

+System.currentTimeMillis()+

".png"

);

054

}

055

056

057

//手工输入验证码并验证

058

HttpResponseverifyResponse=

null

;

059

Stringcapthca=

null

;

060

Stringuvc=

null

;

061

062

do

{

063

//输入验证码,读入键盘输入

064

//1)

065

//InputStreaminputStream=System.in;

066

//BufferedReaderbufferedReader=newBufferedReader(newInputStreamReader(inputStream));

067

//System.out.println("请输入验证码:");

068

//capthca=bufferedReader.readLine();

069

070

capthca=JOptionPane.showInputDialog(

"请输入图片验证码:"

);

071

072

//2)

073

//Scannerscanner=newScanner(System.in);

074

//capthca=scanner.next();

075

//
http://uac.10010.com/portal/Service/CtaIdyChk?callback=jsonp1404716227598&verifyCode=4m3e&verifyType=1

076

StringverifyCaptchaUrl=

"
+capthca+
"&verifyType=1"
;

077

HttpGetverifyCapthcaGet=

new

HttpGet(verifyCaptchaUrl);

078

verifyResponse=httpClient.execute(verifyCapthcaGet);

079

AbstractHttpClientabstractHttpClient=(AbstractHttpClient)httpClient;

080

for

(Cookiecookie:abstractHttpClient.getCookieStore().getCookies()){

081

System.out.println(cookie.getName()+

":"

+cookie.getValue());

082

if

(cookie.getName().equals(

"uacverifykey"

)){

083

uvc=cookie.getValue();

084

}

085

}

086

}

while

(!EntityUtils.toString(verifyResponse.getEntity()).contains(

"true"

));

087

088

//登录

089

Stringloginurl=

"
+name+
"&password="

+pwd+
"&pwdType=01&productType=01&verifyCode="

+capthca+
"&redirectType=03&uvc="

+uvc;

090

HttpGetloginGet=

new

HttpGet(loginurl);

091

CloseableHttpResponseloginResponse=httpclient.execute(loginGet);

092

System.out.print(

"result:"

+EntityUtils.toString(loginResponse.getEntity()));

093

094

//抓取基本信息数据

095

//jsonp1404663560635({resultCode:"7072",redirectURL:"http://www.10010.com",errDesc:"null",msg:'系统忙，请稍后再试。',needvode:"1"});

096

HttpPostbasicHttpGet=

new

HttpPost(

"http://iservice.10010.com/ehallService/static/acctBalance/execute/YH102010005/QUERY_AcctBalance.processData/Result"

);

097

saveToLocal(httpclient.execute(basicHttpGet).getEntity(),

"chinaunicom.basic.html"

);

098

099

}

100

/**

101

*写文件到本地

102

*

103

*@paramhttpEntity

104

*@paramfilename

105

*/

106

public

static

void

saveToLocal(HttpEntityhttpEntity,Stringfilename){

107

108

try

{

109

110

Filedir=

new

File(

"/JEE/sz-588/workspace/maven-httpclient-demo"

);

111

if

(!dir.isDirectory()){

112

dir.mkdir();

113

}

114

115

Filefile=

new

File(dir.getAbsolutePath()+

"/"

+filename);

116

FileOutputStreamfileOutputStream=

new

FileOutputStream(file);

117

InputStreaminputStream=httpEntity.getContent();

118

119

if

(!file.exists()){

120

file.createNewFile();

121

}

122

byte

[]bytes=

new

byte

[

1024

];

123

int

length=

0

;

124

while

((length=inputStream.read(bytes))>

0

){

125

fileOutputStream.write(bytes,

0

,length);

126

}

127

inputStream.close();

128

fileOutputStream.close();

129

}

catch

(Exceptione){

130

e.printStackTrace();

131

}

132

133

}

134

}