PHP CURL 模拟登录 SINA 微博

最近工作上有一个需求, 需要获取 http://weibo.com/at/weibo 的数据, 就是 @我自己 的数据, 没有接口, 只能通过抓页面. 下面贴下部分代码 来自: http://summerbluet.com/704

1. /**
2. * 用于模拟新浪微博登录! by CJ ( http://www.summerbluet.com )
3. */
5. /** 定义项目路径 */
6. define('PROJECT_ROOT_PATH' , dirname(__FILE__));
7. define('COOKIE_PATH' , PROJECT_ROOT_PATH );
9. // 通用时间戳
10. define('TIMESTAMP', time());
12. // 出现问题的时候可以开启, 调试用的, 会在当前文件夹下面创建 LOG 文件
13. define('DEBUG', false);
15. /** 用来做模拟登录的新浪帐号 */
16. $username = "";
17. $password = "";
19. /* Fire Up */
20. $weiboLogin = new weiboLogin( $username, $password );
21. exit($weiboLogin->showTestPage( 'http://weibo.com/at/comment' ));
23. class weiboLogin {
25. private $cookiefile;
26. private $username;
27. private $password;
29. function __construct( $username, $password )
30. {
31. ( $username =='' || $password=='' ) && exit( "请填写用户名密码" );
33. $this->cookiefile = COOKIE_PATH.'/cookie_sina_'.substr(base64_encode($username), 0, 10);
34. $this->username = $username;
35. $this->password = $password;
36. }
38. /**
39. * CURL请求
40. * @param String $url 请求地址
41. * @param Array $data 请求数据
42. */
43. function curlRequest($url, $data = false)
44. {
45. $ch = curl_init();
47. $option = array(
48. CURLOPT_URL => $url,
49. CURLOPT_HEADER => 0,
50. CURLOPT_HTTPHEADER => array('Accept-Language: zh-cn','Connection: Keep-Alive','Cache-Control: no-cache'),
51. CURLOPT_USERAGENT => "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.1 (KHTML, like Gecko) Chrome/21.0.1180.79 Safari/537.1",
52. CURLOPT_FOLLOWLOCATION => TRUE,
53. CURLOPT_MAXREDIRS => 4,
54. CURLOPT_RETURNTRANSFER => TRUE,
55. CURLOPT_COOKIEJAR => $this->cookiefile,
56. CURLOPT_COOKIEFILE => $this->cookiefile
57. );
59. if ( $data ) {
60. $option[CURLOPT_POST] = 1;
61. $option[CURLOPT_POSTFIELDS] = $data;
62. }
64. curl_setopt_array($ch, $option);
65. $response = curl_exec($ch);
67. if (curl_errno($ch) > 0) {
68. exit("CURL ERROR:$url " . curl_error($ch));
69. }
70. curl_close($ch);
71. return $response;
72. }
74. /** @desc CURL 模拟新浪登录 */
75. function doSinaLogin()
76. {
77. // Step 1 : Get tickit
78. $preLoginData = $this->curlRequest('http://login.sina.com.cn/sso/prelogin.php?entry=weibo&callback=sinaSSOController.preloginCallBack&su=' .
79. base64_encode($this->username) . '&client=ssologin.js(v1.3.16)');
80. preg_match('/sinaSSOController.preloginCallBack\((.*)\)/', $preLoginData, $preArr);
81. $jsonArr = json_decode($preArr[1], true);
83. $this->debug('debug_1_Tickit', $preArr[1]);
85. if (is_array($jsonArr)) {
86. // Step 2 : Do Certification
87. $postArr = array( 'entry' => 'weibo',
88. 'gateway' => 1,
89. 'from' => '',
90. 'vsnval' => '',
91. 'savestate' => 7,
92. 'useticket' => 1,
93. 'ssosimplelogin' => 1,
94. 'su' => base64_encode(urlencode($this->username)),
95. 'service' => 'miniblog',
96. 'servertime' => $jsonArr['servertime'],
97. 'nonce' => $jsonArr['nonce'],
98. 'pwencode' => 'wsse',
99. 'sp' => sha1(sha1(sha1($this->password)) . $jsonArr['servertime'] . $jsonArr['nonce']),
100. 'encoding' => 'UTF-8',
101. 'url' => 'http://weibo.com/ajaxlogin.php?framelogin=1&callback=parent.sinaSSOController.feedBackUrlCallBack',
102. 'returntype' => 'META');
104. $loginData = $this->curlRequest('http://login.sina.com.cn/sso/login.php?client=ssologin.js(v1.3.19)', $postArr);
106. $this->debug('debug_2_Certification_raw', $loginData);
108. // Step 3 : SSOLoginState
109. if ($loginData) {
111. $matchs = $loginResultArr =array();
112. preg_match('/replace\(\'(.*?)\'\)/', $loginData, $matchs);
114. $this->debug('debug_3_Certification_result', $matchs[1]);
116. $loginResult = $this->curlRequest( $matchs[1] );
117. preg_match('/feedBackUrlCallBack\((.*?)\)/', $loginResult, $loginResultArr);
119. $userInfo = json_decode($loginResultArr[1],true);
121. $this->debug('debug_4_UserInfo', $loginResultArr[1]);
122. } else {
123. exit('Login sina fail.');
124. }
125. } else {
126. exit('Server tickit fail');
127. }
128. }
130. /** 测试登录情况, 调用参考 */
131. function showTestPage( $url ) {
132. $file_holder = $this->curlRequest( $url );
134. // 如果未登录情况, 登录后再尝试
135. $isLogin = strpos( $file_holder, 'class="user_name"');
136. if ( !$isLogin ){
137. unset($file_holder);
138. $this->doSinaLogin();
139. $file_holder = $this->curlRequest( $url );
140. }
141. return $file_holder ;
142. }
144. /** 调试 */
145. function debug( $file_name, $data ) {
146. if ( DEBUG ) {
147. file_put_contents( $file_name.'.txt', $data );
148. }
149. }
151. }

复制代码