public function actionCheck($id = null)
{
if ($_SERVER['REQUEST_METHOD'] == 'POST') {
$id = $_POST['List']['collect_list_id'];
}
$ids = is_array($id) ? $id : array($id);
$ids = array_filter($ids);
if (!count($ids)) {
$this->redirect[] = array('text' => '', 'href' => $_SERVER['HTTP_REFERER']);
$this->message('请选择至少一项进行操作！', self::MSG_SUCCESS, true);
}
$value = intval($_REQUEST['value']) ? 1 : 0;
//1审核0取消审核
foreach ($ids as $id) {
$list = CollectList::get_list_by_id($id, false);
$task = CollectTask::get_task_by_id($list['collect_task_id'], false);
$template = CollectTemplate::get_template_by_id($task['collect_template_id'], false);
if ($value) {
/*$sql = "INSERT INTO {{game_media_data}} SET
`media_data_subject`='{$list[collect_list_title]}',
`media_data_href`='$list[collect_list_url]',
`media_data_thumb`='$list[collect_list_thumb]',
`media_data_dateline`=".time().",
`game_id`='{$list[game_id]}',
`content_class_id`='{$task[content_class_id]}',
`media_data_source`='{$template[collect_source_id]}',
`collect_data_time`=".strtotime($list['collect_list_day'])
;
$_cmd->execute();*/
$_cmd = Yii::app()->db->createCommand()->insert('{{game_media_data}}', array('media_data_subject' => $list[collect_list_title], 'media_data_href' => $list[collect_list_url], 'media_data_thumb' => $list[collect_list_thumb], 'media_data_dateline' => time(), 'game_id' => $list[game_id], 'content_class_id' => $task[content_class_id], 'media_data_source' => $template[collect_source_id], 'collect_data_time' => strtotime($list['collect_list_day'])));
$media_data_id = Yii::app()->db->getLastInsertID();
} else {
//取消审核
$media_data_id = 0;
if ($list["media_data_id"]) {
$sql = "DELETE FROM {{game_media_data}} WHERE media_data_id={$list[media_data_id]}";
$_cmd = Yii::app()->db->createCommand($sql);
$_cmd->execute();
}
}
$_sql = "UPDATE {{collect_list}} SET `collect_list_check`={$value},`media_data_id`={$media_data_id} WHERE collect_list_id={$id}";
$_cmd = Yii::app()->db->createCommand($_sql);
$_cmd->execute();
}
if (!isset($_GET['ajax'])) {
$this->redirect[] = array('text' => '', 'href' => $_SERVER['HTTP_REFERER']);
$this->message('操作成功', self::MSG_SUCCESS, true);
}
}
public function collectAction($id)
{
$task = CollectTaskModel::inst()->getTaskById($id);
if (empty($task)) {
$this->redirect[] = array('text' => '', 'href' => '/collect/task/index');
$this->message('采集规则不存在或已被删除', self::MSG_ERROR, true);
}
if ($_GET['type'] == 'list') {
$cacheKey = "jvod.collect.list.urls";
if ($this->redis->llen($cacheKey) == 0) {
$collect_list_urls = CollectTaskModel::getListUrls($task['collect_task_urls']);
foreach ($collect_list_urls as $_k => $_v) {
$this->redis->lpush($cacheKey, $_v);
}
}
for ($i = 0; $i < 20 && $this->redis->llen($cacheKey) > 0; $i++) {
$collect_list_url = $this->redis->rpop($cacheKey);
$collect_content_urls = CollectTaskModel::getContentUrls($collect_list_url, $task['collect_list_rules']['begin'], $task['collect_list_rules']['end']);
$repeat_number = 0;
$sql = "SELECT COUNT(collect_content_id) FROM {{collect_content}} WHERE collect_content_url=:collect_content_url";
foreach ($collect_content_urls as $_k => $_v) {
if ($this->db->queryScalar($sql, array(':collect_content_url' => $_v)) == 0) {
$this->db->insert('{{collect_content}}', array('collect_content_id' => 0, 'collect_task_id' => $id, 'collect_content_url' => $_v, 'is_published' => 0, 'is_collected' => 0, 'lasttime' => $_SERVER['REQUEST_TIME'], 'dateline' => $_SERVER['REQUEST_TIME']));
} else {
$repeat_number++;
}
if ($repeat_number > 20) {
// 超过20个重复地址，退出内容地址采集
break;
}
}
}
} else {
$cacheKey = "jvod.collect.content.urls";
if ($this->queue->llen($cacheKey) == 0) {
$sql = "SELECT collect_content_id, collect_task_id, collect_content_url, is_collected, is_published, lasttime, dateline FROM {{collect_content}} WHERE collect_task_id=:collect_task_id AND is_collected=:is_collected";
$ret = $this->db->queryAll($sql, array(':collect_task_id' => $id, ':is_collected' => 0));
foreach ($ret as $_k => $_v) {
$this->queue->lpush($cacheKey, json_encode(array($_v['collect_content_id'], $_v['collect_content_url'])));
}
}
for ($i = 0; $i < 20 && $this->queue->llen($cacheKey) > 0; $i++) {
list($collect_content_id, $collect_content_url) = json_decode($this->queue->rpop($cacheKey));
$collect_content_charset = 'utf-8';
$collect_content_body = CollectTask::get_url_contents($collect_content_url, $collect_content_charset);
if (strtolower($collect_content_charset) != 'utf-8') {
$collect_content_body = mb_convert_encoding($collect_content_body, 'UTF-8', $collect_content_charset);
}
$task['collect_content_data'] = array();
$data = array('collect_content_id' => $collect_content_id);
foreach ($task['collect_content_rules'] as $_k => $_v) {
if ($_v['begin'] && $_v['end']) {
preg_match("/{$_v['begin']}(.+?){$_v['end']}/s", $collect_content_body, $ret);
$data[$_v['collect_fields_identify']] = $ret[1] ? $ret[1] : '';
} else {
$data[$_v['collect_fields_identify']] = '';
}
}
$flag = $this->db->insert("{{collect_model_addons{$task['collect_model_identify']}}}", $data);
if ($flag) {
$this->db->update("{{collect_content}}", array('is_collected' => 1), 'collect_content_url=:collect_content_url', array(':collect_content_url' => $collect_content_url));
}
}
}
}
<?php
/**
* 关注/取消关注
*/
require_once 'lib/WriteLog.lib.php';
try {
require_once 'tasks/Collect/CollectTask.class.php';
$nType = isset($_GET['type']) ? $_GET['type'] : 0;
$collect = new CollectTask();
if ($nType == 0) {
$result = $collect->getMyDesigner();
}
if ($nType == 1) {
$result = $collect->getDesigner();
}
echo $result;
} catch (Exception $e) {
Log::write('designer exception:' . $e->getMessage(), 'log');
echo get_rsp_result(false, 'designer exception');
}
/**
* 采集
* @param int $id collect_task_id
* @url string 需要读取的文件地址
* @task array 采集任务记录数组
*/
public function ContentReg($id, $url, $task)
{
$re = CollectTask::read_html($url);
if ($re['success']) {
$content = $re['content'];
} else {
echo '<script>alert("' . $re['error'] . '");window.close();</script>';
exit;
}
$url_arr = parse_url($url);
if ($url_arr['path']) {
$a = array_filter(explode("/", $url_arr['path']));
array_pop($a);
$pa = implode("/", $a) . "/";
} else {
$pa = "";
}
$content = CollectTask::re2ab($content, $url_arr['scheme'] . "://" . $url_arr['host'] . "/" . $pa);
$template = CollectTemplate::get_template_by_id($task['collect_template_id'], false);
if ($task['collect_task_totalpagereg']) {
$totalpagereg = $task['collect_task_totalpagereg'];
} elseif ($template['collect_template_totalpagereg']) {
$totalpagereg = $template['collect_template_totalpagereg'];
} else {
$totalpagereg = "";
}
$pagerule = $task['collect_task_pagerule'] ? $task['collect_task_pagerule'] : $template['collect_template_pagerule'];
$totalpage = 0;
if ($totalpagereg) {
$totalpagereg = preg_replace("/\\[.*\\]/U", "(\\d)", $totalpagereg);
preg_match("/" . $totalpagereg . "/", $content, $p);
if (intval($p[1])) {
$totalpage = intval($p[1]);
}
}
$pageurl_arr = array();
if ($totalpage) {
for ($i = 2; $i <= $totalpage; $i++) {
$pageurl_arr[] = preg_replace("/{$pagerule}\\d/", $pagerule . $i, $url);
}
} else {
//不是js分页时，则查找分页地址
$pagestart = $task['collect_task_pagestart'] ? $task['collect_task_pagestart'] : $template['collect_template_pagestart'];
$pageend = $task['collect_task_pageend'] ? $task['collect_task_pageend'] : $template['collect_template_pageend'];
if ($pagestart && $pageend) {
$page_area_reg = str_replace("/", "\\/", $pagestart) . "([\\s\\S]*)" . str_replace("/", "\\/", $pageend);
preg_match("/{$page_area_reg}/Ui", $content, $mm);
if ($mm[1]) {
preg_match_all('/\\<a.*href=["|\'](.*)?["|\'].*\\>(.*)\\<\\/a\\>/iU', $content, $arr);
if (count($arr[1])) {
$links = array_unique($arr[1]);
sort($links);
$source_url = preg_replace("/{$pagerule}\\d/", "", $url);
if ($pagerule) {
foreach ($links as $k => $v) {
if ($v != $url) {
$baseurl = preg_replace("/{$pagerule}\\d/", "", $v);
if ($baseurl == $source_url && $v != $url) {
$pageurl_arr[] = $v;
}
}
}
}
}
}
}
}
if (is_array($pageurl_arr) && count($pageurl_arr)) {
//取得所有分页内容
foreach ($pageurl_arr as $v) {
$c = CollectTask::read_html($v);
if ($c['content']) {
$c['content'] = CollectTask::re2ab($c['content'], $url_arr['scheme'] . "://" . $url_arr['host'] . "/");
$content .= $c['content'];
}
}
}
$content = str_replace("none", "block", $content);
//有些分页是默认不显示内容的
$filter = array();
if ($task['collect_task_filter']) {
$filter = array_merge($filter, json_decode($task['collect_task_filter']));
}
if ($template['collect_template_filter']) {
$filter = array_merge($filter, json_decode($template['collect_template_filter']));
}
$_charset = CollectTask::get_charset();
$content = mb_convert_encoding($content, "UTF-8", "gb2312,gbk,utf-8");
if (count($filter)) {
foreach ($filter as $f) {
$f = preg_replace("/\\[.*\\]/U", "([\\s\\S]*)", $f);
$f = str_replace("/", "\\/", $f);
$content = preg_replace("/" . $f . "/Ui", " ", $content);
}
}
if ($task['collect_task_rulearr']) {
eval('$task_rule_arr = ' . $task['collect_task_rulearr'] . ';');
$task_rule_arr = array_filter($task_rule_arr);
} else {
$task_rule_arr = array();
}
if ($template['collect_template_fieldsreg']) {
eval('$template_rule_arr = ' . $template['collect_template_fieldsreg'] . ';');
}
if (count($task_rule_arr)) {
foreach ($task_rule_arr as $k => $v) {
$template_rule_arr[$k] = $v;
}
}
$template_rule_arr = array_filter($template_rule_arr);
$fields = array();
if (count($template_rule_arr) && is_array($template_rule_arr)) {
foreach ($template_rule_arr as $k => $v) {
$reg = preg_replace("/\\[.*\\]/U", "([\\s\\S]*)", $v);
$reg = str_replace("/", "\\/", $reg);
preg_match_all("/" . $reg . "/U", $content, $ma);
if (count($ma[1])) {
foreach ($ma[1] as $k2 => $v2) {
if (CollectTask::is_image($v2)) {
if ($task['collect_task_saveimg']) {
$ma[1][$k2] = CollectTask::save_image($v2);
} else {
$ma[1][$k2] = $v2;
}
} else {
if ($task['collect_task_saveimg']) {
preg_match_all('/<img.*src=["|\']+(.*)?["|\']+>/Ui', $v2, $arr);
//正则出所有图片保存替换
if (is_array($arr[1]) && count($arr[1])) {
foreach ($arr[1] as $img) {
$ma[1][$k2] = str_replace($img, CollectTask::save_image($img), $v2);
}
}
}
}
}
preg_match("/\\[(.*)\\]/U", $v, $match);
$fields[] = array('identify' => $k, 'name' => $match[1], 'value' => $ma[1][0]);
}
}
}
return $fields;
}
<?php
/**
* 关注/取消关注
*/
require_once 'lib/WriteLog.lib.php';
try {
require_once 'tasks/Collect/CollectTask.class.php';
$collect = new CollectTask();
$nCollect = isset($_GET['collect']) ? $_GET['collect'] : 0;
#0取消关注/1添加关注/2获取关注状态
if ($nCollect == 2) {
$result = $collect->getCollect($nCollect);
} else {
$result = $collect->setCollect($nCollect);
}
echo $result;
} catch (Exception $e) {
Log::write('collect exception:' . $e->getMessage(), 'log');
echo get_rsp_result(false, 'collect exception');
}
if (is_array($f['value'])) {
foreach ($f['value'] as $v) {
if (CollectTask::is_image($v)) {
//echo CollectTask::signName($v).".".CollectTask::getFileType($v);
?>
<img src='<?php
echo $v;
?>
' /><br />
<?php
} else {
echo $v . '<br />';
}
}
} else {
if (CollectTask::is_image($f['value'])) {
?>
<img src='<?php
echo $v;
?>
' /><br />
<?php
} else {
echo $v . '<br />';
}
}
?>
</td>
</tr>
<?php
}
