gx

@@ -164,16 +164,19 @@ class ProofreadingController extends BaseController
         $pattern = '/>([^<]+)</'; // 定义匹配中间内容不是标签的正则表达式
         $matches = array();
         preg_match_all($pattern, $strippedContent, $matches);
-        $data = [];
-        $textContentArray = $matches[1];
         $textContentArray = array_filter($matches[1], function($item) {
             return !empty(trim($item));
         });
-        var_dump($textContentArray);
-        die();
-        $uniqueArray = array_unique($data);
+        $textContentArray = array_values($textContentArray);
+        $uniqueArray = array_unique($textContentArray);
         $textContentArray = array_values($uniqueArray);
-        return $textContentArray;
+        $data = [];
+        foreach ($textContentArray as $v){
+            $content = trim($v);
+            $trimmedString = preg_replace('/\s+/', ' ', $content);
+            $data[] = $trimmedString;
+        }
+        return $data;
     }
 
     /**