<?php // 默认超时 set_time_limit(0); // 定义应用目录 define('APP', dirname(dirname(__FILE__)) . DIRECTORY_SEPARATOR); // 载入框架引导文件 require APP . 'system/_shell.php'; require APP . 'funcs/spider.fn.php'; require APP . 'models/TaskModel.php'; system('echo -e "\\033[32m开始获取队列... \\033[0m"'); $taskModel = new TaskModel(); $task_list = $taskModel->getList(); $task_count = count($task_list); system('echo -e "获取到\\033[32m[' . $task_count . ']\\033[0m个任务队列, 开始抓取链接列表..."'); foreach ($task_list as $v) { $task_list_id = $v['id']; $task_status = $v['status']; $chatset = $v['charset']; $rules = array(); if ($task_status == 'yes') { system("echo -e '开始抓取\\033[34m[" . $v['node_name'] . "]\\033[0m...'"); $rules = json_decode($v['link_rules'], true); // 获取内容的链接 $link_list = array(); $link_list = get_link_list($rules['list_rule']); $link_list_count = count($link_list); system("echo -e '获取到\\033[32m[" . $link_list_count . "]\\033[0m个列表列表链接,准备获取文章链接...'"); foreach ($link_list as $vo) { system("echo -e '获取列表内容链接: \\033[32m" . $vo . "\\033[0m'"); $content = '';
/** * 获取单个任务列表 * * @author boxcore * @date 2014-09-17 * @return array [description] */ protected function _getTaskList() { $mark = 'qsbk'; $task_list = array(); if (!empty($mark)) { $taskModel = new TaskModel(); $task_id = $taskModel->getIdByMark($mark); if ($task_id > 0) { $task_list = $taskModel->getList(array('task_id' => $task_id)); if (!empty($task_list)) { return $task_list[0]; } } } return false; }