function collection($userArr) { global $db, $curl; $page = 1; while (true) { $url = "http://www.jianshu.com/users/{$userArr['name']}/subscriptions?page={$page}"; $curl->setUrl($url); $html = $curl->run(); $p = '/<h4><a href="\\/collection\\/(.*?)"/'; preg_match_all($p, $html, $match); if (is_array($match[1]) && !empty($match[1])) { foreach ($match[1] as $value) { $temp = $db->get('collection', '*', array('name' => $value)); if (!is_array($temp) || empty($temp)) { $id = getCollectionInfo($value); $temp['id'] = $id; } //关系表 $temp = $db->get('collection_user_relation', '*', array('AND' => array('cid' => $temp['id'], 'uid' => $userArr['id']))); if (!is_array($temp) || empty($temp)) { $db->insert('collection_user_relation', array('cid' => $temp['id'], 'uid' => $userArr['id'])); } $db->clear(); } } else { break; } $page++; echo "collection:{$page}\r\n"; } }
function getCollection() { global $curl; $page = 1; while (true) { //热门 //$url = "http://www.jianshu.com/collections?category_id=53&page={$page}&_=1476265581231"; //推荐 //$url = "http://www.jianshu.com/collections?order_by=score&page={$page}&_=1476259795034"; //城市 $url = "http://www.jianshu.com/collections?category_id=69&page={$page}&_=1476346480280"; $curl->setUrl($url); $html = $curl->run(); $p = '/"\\/collection\\/(.*?)\\"/'; preg_match_all($p, $html, $math); if (isset($math[1]) && !empty($math[1])) { $temp = array_unique($math[1]); // print_r($temp); foreach ($temp as $value) { getCollectionInfo($value); } } else { echo "Over!\r\n"; break; } $page++; echo $page, "\r\n"; } }