php curl采集数据携带cookie,THINKPHP 通用采集类，通过CURL提交请求，保持cookie，可变参数...

//采集公共类

class CurlApi extends HomeController {

/**

* reqeust_by_curl

* 通过Curl请求

*

* @param string $url 请求地址，必选

* @param array/string POST或PUT的数据参数

* @param string $mothod 请求类型 POST GET PUT DELETE

* @param string $cookie_dir cookie 保存目录，默认保存到系统临时目录

* @param string $cookie_file 已有 cookie 文件时，使用该 cookie 文件里的 cookie 访问

* @param string $http_version HTTP 版本，支持 1.0, 1.1, 2.0, 2TLS, 2_PRIOR_KNOWLEDGE ，默认 1.1

* @param string $referer 来源地址

* @param array $headers 自定义头部信息

* @param boolean $nobody 是否返回消息主体

* @param string $encoding 编码类型 identity, deflate, gzip

* @param int $request_timeout 请求超时时间

* @param int $response_timeout 响应超时时间

* @param string $user_agent 用户浏览器类型

* @return mixed array

* int httpcode 响应代码

* array cookies 数组保存的获取到的 cookie

* string cookie_file 本次保存的 cookie 文件

* string effective_url 跳转地址

* string header 响应头

* string body 响应主体

*/

public function request_by_curl()

{

// 获得传入的所有参数的数组

$args_arr = func_get_args();

// 数组分配变量，添加arg_头缀

extract($args_arr[0], EXTR_PREFIX_ALL, 'arg');

// 请求地址，必填，否则返回false;

$url = $arg_url;

if(empty($url)) {

return false;

}

isset($arg_data) ? $data = $arg_data : $data = null;

isset($arg_method) ? $method = $arg_method : $method = 'GET';

isset($arg_cookie_dir) ? $cookie_dir = $arg_cookie_dir : $cookie_dir = sys_get_temp_dir();

isset($arg_cookie_file) ? $cookie_file = $arg_cookie_file : $cookie_file = null;

isset($arg_http_version) ? $http_version = $arg_http_version : $http_version = '1.1';

isset($arg_referer) ? $referer = $arg_referer : $referer = null;

isset($arg_headers) ? $headers = $arg_headers : $headers = null;

isset($arg_nobody) ? $nobody = $arg_nobody : $nobody = false;

isset($arg_encoding) ? $encoding = $arg_encoding : $encoding = null;

isset($arg_request_timeout) ? $request_timeout = $arg_request_timeout : $request_timeout = 30;

isset($arg_response_timeout) ? $response_timeout = $arg_response_timeout : $response_timeout = 30;

isset($arg_user_agent) ? $user_agent = $arg_user_agent : $user_agent = 'Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.1; Trident/4.0;)';

$ch = curl_init();

// 请求地址

curl_setopt($ch, CURLOPT_URL, $url);

// cookie处理

if(empty($cookie_file)) {

$cookie_file = tempnam($cookie_dir, 'shuke_');

}

$size = strlen(file_get_contents($cookie_file));

if($size <= 0) {

curl_setopt($ch, CURLOPT_COOKIEJAR, $cookie_file);

} else {

curl_setopt($ch, CURLOPT_COOKIEFILE, $cookie_file);

}

// SSL HTTPS

if(stripos($url, 'https://') !== false) {

curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);

curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);

curl_setopt($ch, CURLOPT_SSLVERSION, 1);

}

// 设置同时返回响应头信息

curl_setopt($ch, CURLOPT_HEADER, TRUE);

// 设置不返回消息主体

if($nobody) {

curl_setopt($ch, CURLOPT_NOBODY, TRUE);

}

// 设置返回值给变量，不直接显示结果

curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE);

// 请求超时时间

curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $request_timeout);

// 响应超时时间

curl_setopt($ch, CURLOPT_TIMEOUT, $response_timeout);

// User-Agent

curl_setopt($ch, CURLOPT_USERAGENT, $user_agent);

// Accept-Encoding

if(!empty($encoding)) {

curl_setopt($ch, CURLOPT_HTTPHEADER, 'Accept-Encoding: '.$encoding);

curl_setopt($ch, CURLOPT_ENCODING, $encoding);

}

// 自定义header

if(!empty($headers)) {

curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);

}

// 来源Referer

if(!empty($referer)) {

curl_setopt($ch, CURLOPT_REFERER, $referer);

}

// 抓取跳转信息

curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE);

// HTTP版本选择，默认HTTP 1.1

switch($http_version) {

case '1.0':

curl_setopt($ch, CURLOPT_HTTP_VERSION , CURL_HTTP_VERSION_1_0);

break;

case '2.0':

curl_setopt($ch, CURLOPT_HTTP_VERSION , CURL_HTTP_VERSION_2_0);

break;

case '2TLS':

curl_setopt($ch, CURLOPT_HTTP_VERSION , CURL_HTTP_VERSION_2TLS);

break;

case '2_PRIOR_KNOWLEDGE':

curl_setopt($ch, CURLOPT_HTTP_VERSION , CURL_HTTP_VERSION_2_PRIOR_KNOWLEDGE);

break;

default:

curl_setopt($ch, CURLOPT_HTTP_VERSION , CURL_HTTP_VERSION_1_1);

}

//拼接参数

$query_string = '';

if(is_array($data)) {

foreach ($data as $key => $val) {

$query_string .= urlencode($key).'='.urlencode($val).'&';

}

$query_string = substr($query_string, 0, -1);

} else {

$query_string = $data;

}

//请求方法，默认GET

switch($method) {

case 'POST':

curl_setopt($ch, CURLOPT_POST, true);

curl_setopt($ch, CURLOPT_POSTFIELDS, $query_string);

break;

case 'PUT':

curl_setopt($ch, CURLOPT_CUSTOMREQUEST, 'PUT');

curl_setopt($ch, CURLOPT_POSTFIELDS, $query_string);

break;

case 'DELETE':

curl_setopt($ch, CURLOPT_CUSTOMREQUEST, 'DELETE');

break;

default:

}

$response = curl_exec($ch);

$httpcode = curl_getinfo($ch, CURLINFO_HTTP_CODE);

$effective_url = curl_getinfo($ch, CURLINFO_EFFECTIVE_URL);

curl_close($ch);

// 折分响应头和主体

list($header, $body) = explode("\r\n\r\n", $response, 2);

// 解析响应头 cookie

preg_match_all("/set\-cookie:([^\r\n]*)/i", $header, $matches);

$cookie = $matches[1];

$response_arr = array('httpcode' => $httpcode, 'cookies'=> $cookie, 'cookie_file'=> $cookie_file, 'effective_url' => $effective_url, 'header' => $header, 'body' => $body);

return $response_arr;

}

}

简单调用GET请求：

$curlGet = new CurlApi;

//GET请求，响应结果存入$result变量

$result = $curlGet->request_by_curl(array('url' => 'http://www.shukeiot.cn/api'));

//取得cookie_file存放目录，下次请求可以带着cookie

if($result['httpcode'] == 200) {

$cookie_file = $result['cookie_file'];

}

//模拟登陆, $arr_params参数根据需要选择即可，最简只需要传入url

$arr_params = array(

// 请求地址，支持 SSL

'url'=>$this->login_index,

// 请求方法 GET, POST, PUT, DELETE

'method' => 'POST',

// 传递的数据，可传 null, array, string

'data' => array(

'username' => $username,

'password' => $password

),

// 已有 cookie 文件时，使用该 cookie 文件里的 cookie 访问

'cookie_file' => $cookie_file,

);

$login_result = $curlGet->request_by_curl($arr_params);