function getItemdetails($url) { $dataResult = array(); $baseUrl = "http://www.dianping.com"; $totalUrl = $baseUrl . $url; cls_curl::set_gzip(true); $content = cls_curl::get($totalUrl); preg_match_all('#<h1 class=\\"shop-name\\">\\s(.*?)\\s</h1>#s', $content, $shopname); $name = preg_replace("/<(a.*?)>(.*?)<(\\/a.*?)>/si", "", $shopname[0][0]); $name = preg_replace("/<(\\/?h1.*?)>/si", "", $name); $dataResult['name'] = $name; preg_match_all('#<span class="info-name">营业时间:(.*?)</span>(.*?)<span class="item">\\s(.*?)\\s</span>#s', $content, $hour); $dataResult['hour'] = $hour[3][0]; preg_match_all('#<div class="expand-info address" itemprop="street-address">\\s(.*?)\\s</div>#s', $content, $address); $addr = preg_replace("/<(\\/?div.*?)>/si", "", $address[0][0]); $addr = preg_replace("/<(\\/?span.*?)>/si", "", $addr); $addr = preg_replace("/<(\\/?a.*?)>/si", "", $addr); $dataResult['address_detail'] = $addr; preg_match_all('#<p class="expand-info tel">\\s(.*?)\\s</p>#s', $content, $phone); $phoneNum = preg_replace("/<(\\/?p.*?)>/si", "", $phone[0][0]); $phoneNum = preg_replace("/<(\\/?span.*?)>/si", "", $phoneNum); $phoneNum = preg_replace("/<(a.*?)>(.*?)<(\\/a.*?)>/si", "", $phoneNum); $dataResult['tel'] = $phoneNum; $dataResult['province'] = '北京市'; $dataResult['city'] = '北京'; preg_match_all('#<span class="item">人均:(.*?)</span>#s', $content, $price); $dataResult['price'] = $price[1][0]; return $dataResult; }
/** * 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}"; set_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; }