Merge branch 'develop' of http://47.244.231.31:8099/zhl/globalso-v6 into bate

@@ -53,6 +53,7 @@ class HtmlCollect extends Command
 
     protected function start_collect()
     {
+        $tdk_project_ids = [714];
         $task_id = $this->get_task();
         if ($task_id === false) {
             //所有项目采集完成
@@ -108,7 +109,9 @@ class HtmlCollect extends Command
                 }
 
                 //提取页面tdk
-                $this->get_site_meta($new_html, $collect_info);
+                if(in_array($project_id,$tdk_project_ids)){
+                    $this->get_site_meta($new_html, $collect_info);
+                }
 
                 $source_list = $this->html_preg($new_html, $project_id, $collect_info->domain, $old_info['web_url_domain'], $old_info['home_url']);
 
@@ -205,13 +208,13 @@ class HtmlCollect extends Command
             #Title
             preg_match_all('/<title>([\w\W]*?)<\/title>/', $html, $matches);
             if (!empty($matches[1])) {
-                $meta['title'] = substr($matches[1][0], 0, 70);
+                $meta['title'] = substr($matches[1][0], 0, 255);
             }
 
             #Keywords
             preg_match_all('/<meta\s+[^>]*?name=[\'|\"]keywords[\'|\"]\s+[^>]*?content=[\'|\"]([\w\W]*?)[\'|\"]/', $html, $matches);
             if (!empty($matches[1])) {
-                $meta['keyword'] = substr($matches[1][0], 0, 200);
+                $meta['keyword'] = substr($matches[1][0], 0, 255);
             }
 
             #Description

@@ -239,7 +239,7 @@ class WebTraffic extends Command
             ->where('pdo.domain', '>', 0)
             ->where('poc.qa_status', OnlineCheck::STATUS_ONLINE_TRUE)
             ->whereIn('gl_project.type', [Project::TYPE_TWO, Project::TYPE_FOUR])
-            ->whereIn('pdo.project_id', [6, 25])   //todo 测试两个项目 后面删掉
+            ->where('gl_project.is_upgrade', 0)   //非升级项目
             ->where(function ($query) use ($type) {
            if($type == 1){
                //1-3个月项目

@@ -158,14 +158,11 @@ class TranslateLogic extends BaseLogic
         }
         $contentData = array_values($contentData);
         $pattern = '/<meta\s+[^>]*name=[\'"](keywords|description)[\'"][^>]*content=[\'"]([^\'"]+)[\'"]/i'; // 匹配 name 为 "keywords" 或 "description" 的 meta 标签的正则表达式
-
-        $matches1 = array();
-        preg_match_all($pattern, $strippedContent, $matches1);
+        $matches = array();
+        preg_match_all($pattern, $strippedContent, $matches);
         $metaData = array();
-        foreach ($matches1[1] as $content) {
-            if (!empty($content)) {
-                $metaData[] = $content;
-            }
+        foreach ($matches[2] as $index => $content) {
+            $metaData[] = $content;
         }
         $data = array_merge($metaData, $contentData);
         return $data;