更新提交

server/app/Services/ThirdPlatform/CheMenuService.php

@@ -50,10 +50,16 @@ class CheMenuService
     {
         try{
             $pageUrl = $url['url'];
+            SimpleLogs::writeLog('开始抓取'.$pageUrl.'分类页面数据', 'chemenu info', 'info');
             while(true){
                 $responseHtml = $this->apiService->getPage($pageUrl);
                 //file_put_contents('./test2.txt', $responseHtml);
                 if($responseHtml){
+                    if(preg_match('/<span[\s]*?class=[\\\'|\"]caption-subject font-blue-sharp bold[\\\'|\"]>Sorry ! You visit too frequently, please complete validation[\s\S]*?<\/span><\/h2>/i', $responseHtml)){
+                        SimpleLogs::writeLog('遇到反爬系统', 'chemenu info', 'info');
+                        return true;
+                    }
+                    SimpleLogs::writeLog('开始分析分类页面数据', 'chemenu info', 'info');
                     $titleDivRegx = '/<div[\s]*?class=[\\\'|\"]cate_desc[\\\'|\"]>[\S\s]*<h2[\s]*?class=[\\\'|\"]desc_title[\\\'|\"]>(?<title>[\s\S]+?)<\/h2>[\s\S]*?<p[\s]*?class=[\\\'|\"]desc_content[\\\'|\"]>(?<content>[\s\S]*?)<\/p>/i';
                     preg_match($titleDivRegx, $responseHtml,$titleHtml);
 
@@ -76,9 +82,12 @@ class CheMenuService
                             }
                             preg_match('/<li[\s]*?id=[\\\'|\"]pagination_next[\\\'|\"]>[\s]*?<a[\S\s]*?href=[\\\'|\"](?<nextHref>[\s\S]+?)[\\\'|\"]/i', $responseHtml, $nextHref);
                             $pageUrl = str_replace($this->apiService->getDomain(), '', $nextHref['nextHref']);
+
+                            SimpleLogs::writeLog('开始下一页抓取,链接：'.$pageUrl, 'chemenu info', 'info');
                         }
                     }
-                    usleep(rand(50000,150000));
+                    $randomFloat = 3 + mt_rand(0, 3) + (mt_rand(0, 999) / 1000);
+                    usleep($randomFloat * 1000000); // 微秒级延迟
                 }
             }
         }catch(\Throwable $exception){