/**
* 获取知乎关联用户 和 用户信息
*
* Licensed under The MIT License
* For full copyright and license information, please see the MIT-LICENSE.txt
* Redistributions of files must retain the above copyright notice.
*
* @author seatle<*****@*****.**>
* @copyright seatle<*****@*****.**>
* @link http://www.epooll.com/
* @license http://www.opensource.org/licenses/mit-license.php MIT License
*/
function set_cookie()
{
//$cookie_jar = dirname(__FILE__)."/zhihu.cookie";
//cls_curl::set_cookie_file($cookie_jar);
$cookie = '_za=36643642-e546-4d60-a771-8af8dcfbd001; q_c1=a57a2b9f10964f909b8d8969febf3ab2|1437705596000|1437705596000; _xsrf=f0304fba4e44e1d008ec308d59bab029; cap_id="YWY1YmRmODlmZGVmNDc3MWJlZGFkZDg3M2E0M2Q5YjM=|1437705596|963518c454bb6f10d96775021c098c84e1e46f5a"; z_c0="QUFCQVgtRWZBQUFYQUFBQVlRSlZUVjR6NEZVUTgtRkdjTVc5UDMwZXRJZFdWZ2JaOWctNVhnPT0=|1438164574|aed6ef3707f246a7b64da4f1e8c089395d77ff2b"; __utma=51854390.1105113342.1437990174.1438160686.1438164116.10; __utmc=51854390; __utmz=51854390.1438134939.8.5.utmcsr=zhihu.com|utmccn=(referral)|utmcmd=referral|utmcct=/people/yangzetao; __utmv=51854390.100-1|2=registration_date=20131030=1^3=entry_date=20131030=1';
cls_curl::set_cookie($cookie);
}
/**
* 获取知乎关联用户 和 用户信息
*
* Licensed under The MIT License
* For full copyright and license information, please see the MIT-LICENSE.txt
* Redistributions of files must retain the above copyright notice.
*
* @author seatle<*****@*****.**>
* @copyright seatle<*****@*****.**>
* @link http://www.epooll.com/
* @license http://www.opensource.org/licenses/mit-license.php MIT License
*/
function set_cookie()
{
//$cookie_jar = dirname(__FILE__)."/zhihu.cookie";
//cls_curl::set_cookie_file($cookie_jar);
$cookie = '';
// 自己去知乎上面取
cls_curl::set_cookie($cookie);
}
/**
* 获取用户
*
* @param string $username
* @param string $user_type followees 、followers
* @return void
* @author seatle <*****@*****.**>
* @created time :2015-07-28 09:46
*/
function get_user_index($username, $user_type = 'followees', $worker)
{
$url = "http://www.zhihu.com/people/{$username}/{$user_type}";
// by suhy
$cookie = trim(file_get_contents("cookie.txt"));
cls_curl::set_cookie($cookie);
cls_curl::set_gzip(true);
$content = cls_curl::get($url);
if (empty($content)) {
return array();
}
$users = array();
// 用户不足20个的时候，从ajax取不到用户，所以首页这里还是要取一下
preg_match_all('#<h2 class="zm-list-content-title"><a data-tip=".*?" href="http://www.zhihu.com/people/(.*?)" class="zg-link" title=".*?">(.*?)</a></h2>#', $content, $out);
$count = count($out[1]);
for ($i = 0; $i < $count; $i++) {
$d_username = empty($out[1][$i]) ? '' : $out[1][$i];
$d_nickname = empty($out[2][$i]) ? '' : $out[2][$i];
if (!empty($d_username) && !empty($d_nickname)) {
$users[$d_username] = array('username' => $d_username, 'nickname' => $d_nickname);
}
}
$keyword = $user_type == 'followees' ? '关注了' : '关注者';
$worker->log("采集用户 --- " . $username . " --- {$keyword} --- 主页 --- 成功\n");
preg_match('#<span class="zg-gray-normal">' . $keyword . '</span><br />\\s<strong>(.*?)</strong><label> 人</label>#', $content, $out);
$user_count = empty($out[1]) ? 0 : intval($out[1]);
preg_match('#<input type="hidden" name="_xsrf" value="(.*?)"/>#', $content, $out);
$_xsrf = empty($out[1]) ? '' : trim($out[1]);
preg_match('#<div class="zh-general-list clearfix" data-init="(.*?)">#', $content, $out);
$url_params = empty($out[1]) ? '' : json_decode(html_entity_decode($out[1]), true);
if (!empty($_xsrf) && !empty($url_params) && is_array($url_params)) {
$url = "http://www.zhihu.com/node/" . $url_params['nodename'];
$params = $url_params['params'];
$j = 1;
for ($i = 0; $i < $user_count; $i = $i + 20) {
$params['offset'] = $i;
$post_data = array('method' => 'next', 'params' => json_encode($params), '_xsrf' => $_xsrf);
$content = cls_curl::post($url, $post_data);
if (empty($content)) {
$worker->log("采集用户 --- " . $username . " --- {$keyword} --- 第{$j}页 --- 失败\n");
continue;
}
$rows = json_decode($content, true);
if (empty($rows['msg']) || !is_array($rows['msg'])) {
$worker->log("采集用户 --- " . $username . " --- {$keyword} --- 第{$j}页 --- 失败\n");
continue;
}
$worker->log("采集用户 --- " . $username . " --- {$keyword} --- 第{$j}页 --- 成功\n");
foreach ($rows['msg'] as $row) {
preg_match_all('#<h2 class="zm-list-content-title"><a data-tip=".*?" href="http://www.zhihu.com/people/(.*?)" class="zg-link" title=".*?">(.*?)</a></h2>#', $row, $out);
$d_username = empty($out[1][0]) ? '' : $out[1][0];
$d_nickname = empty($out[2][0]) ? '' : $out[2][0];
if (!empty($d_username) && !empty($d_nickname)) {
$users[$d_username] = array('username' => $d_username, 'nickname' => $d_nickname);
}
}
$j++;
}
}
return $users;
}
