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

@@ -2,11 +2,17 @@
 
 
 namespace App\Console\Commands\Update;
 namespace App\Console\Commands\Update;
 
 
+use App\Helper\Arr;
+use App\Models\Blog\Blog;
 use App\Models\Collect\CollectSource;
 use App\Models\Collect\CollectSource;
 use App\Models\Collect\CollectTask;
 use App\Models\Collect\CollectTask;
 use App\Models\Com\UpdateLog;
 use App\Models\Com\UpdateLog;
 use App\Models\Com\UpdateOldInfo;
 use App\Models\Com\UpdateOldInfo;
+use App\Models\CustomModule\CustomModuleContent;
+use App\Models\News\News;
+use App\Models\Product\Product;
 use App\Models\RouteMap\RouteMap;
 use App\Models\RouteMap\RouteMap;
+use App\Models\Template\BCustomTemplate;
 use App\Services\CosService;
 use App\Services\CosService;
 use App\Services\ProjectServer;
 use App\Services\ProjectServer;
 use Illuminate\Console\Command;
 use Illuminate\Console\Command;
@@ -65,7 +71,7 @@ class HtmlCollect extends Command
         //设置数据库
         //设置数据库
         $project = ProjectServer::useProject($project_id);
         $project = ProjectServer::useProject($project_id);
         if ($project) {
         if ($project) {
-            $collect_info = CollectTask::select(['id', 'domain', 'route'])->where('id', $collect_id)->where('status', CollectTask::STATUS_UN)->where('language', '')->first();
+            $collect_info = CollectTask::select(['id', 'domain', 'route', 'source', 'source_id'])->where('id', $collect_id)->where('status', CollectTask::STATUS_UN)->where('language', '')->first();
 
 
             if (!$collect_info) {
             if (!$collect_info) {
                 sleep(2);
                 sleep(2);
@@ -101,6 +107,9 @@ class HtmlCollect extends Command
                     $new_html = str_replace($v64, '', $new_html);
                     $new_html = str_replace($v64, '', $new_html);
                 }
                 }
 
 
+                //提取页面tdk
+                $this->get_site_meta($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']);
                 $source_list = $this->html_preg($new_html, $project_id, $collect_info->domain, $old_info['web_url_domain'], $old_info['home_url']);
 
 
                 if ($source_list) {
                 if ($source_list) {
@@ -188,6 +197,73 @@ class HtmlCollect extends Command
         return $task_id;
         return $task_id;
     }
     }
 
 
+    //获取META信息
+    private function get_site_meta($html, $collect_info)
+    {
+        $meta = [];
+        if (!empty($html)) {
+            #Title
+            preg_match_all('/<title>([\w\W]*?)<\/title>/', $html, $matches);
+            if (!empty($matches[1])) {
+                $meta['title'] = substr($matches[1][0], 0, 70);
+            }
+
+            #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);
+            }
+
+            #Description
+            preg_match_all('/<meta name=[\'|\"]description[\'|\"]\s+[^>]*?content=[\'|\"]([\w\W]*?)[\'|\"]/', $html, $matches);
+            if (!empty($matches[1])) {
+                $meta['description'] = substr($matches[1][0], 0, 255);
+            }
+        }
+
+        if (!empty($meta)) {
+            $id = $collect_info->source_id;
+
+            switch ($collect_info->source) {
+                case RouteMap::SOURCE_PRODUCT:
+                    $seo_mate = [
+                        'title' => $meta['title'] ?? '',
+                        'keyword' => $meta['keyword'] ?? '',
+                        'description' => $meta['description'] ?? ''
+                    ];
+                    Product::where('id', $id)->update(['seo_mate' => Arr::a2s($seo_mate)]);
+                    break;
+                case RouteMap::SOURCE_NEWS:
+                    News::where('id', $id)->update([
+                        'seo_title' => $meta['title'] ?? '',
+                        'seo_keywords' => $meta['keyword'] ?? '',
+                        'seo_description' => $meta['description'] ?? ''
+                    ]);
+                    break;
+                case RouteMap::SOURCE_BLOG:
+                    Blog::where('id', $id)->update([
+                        'seo_title' => $meta['title'] ?? '',
+                        'seo_keywords' => $meta['keyword'] ?? '',
+                        'seo_description' => $meta['description'] ?? ''
+                    ]);
+                    break;
+                case RouteMap::SOURCE_PAGE:
+                    BCustomTemplate::where('id', $id)->update([
+                        'title' => $meta['title'] ?? '',
+                        'keywords' => $meta['keyword'] ?? '',
+                        'description' => $meta['description'] ?? ''
+                    ]);
+                    break;
+                default:
+                    CustomModuleContent::where('id', $id)->update([
+                        'seo_title' => $meta['title'] ?? '',
+                        'seo_keywords' => $meta['keyword'] ?? '',
+                        'seo_description' => $meta['description'] ?? ''
+                    ]);
+            }
+        }
+    }
+
     //正则匹配html资源
     //正则匹配html资源
     protected function html_preg($html, $project_id, $domain, $web_url_domain, $home_url)
     protected function html_preg($html, $project_id, $domain, $web_url_domain, $home_url)
     {
     {

@@ -157,7 +157,8 @@ class TranslateLogic extends BaseLogic
             $contentData[] = $trimmedString;
             $contentData[] = $trimmedString;
         }
         }
         $contentData = array_values($contentData);
         $contentData = array_values($contentData);
-        $pattern = '/<meta\s+[^>]*content=[\'"]([^\'"]+)[\'"]/i'; // 匹配 meta 标签的 content 属性的正则表达式
+        $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();
         $matches1 = array();
         preg_match_all($pattern, $strippedContent, $matches1);
         preg_match_all($pattern, $strippedContent, $matches1);
         $metaData = array();
         $metaData = array();