调整

application/api/controller/Aireview.php

@@ -129,7 +129,7 @@ class Aireview extends Base
                 $aFile = empty($aFile['data']) ? [] : $aFile['data'];
                 $aArticleMain = empty($aFile['mains']) ? [] : $aFile['mains'];
             }
-            $sContent = empty($aArticleMain) ? '' : implode("", array_unique($aArticleMain));
+            $sContent = empty($aArticleMain) ? '' : implode("\n", array_unique($aArticleMain));
             if(empty($sContent)){
                 return json_encode(array('status' => 4,'msg' => 'No article content found:'.$sQuestionFields));
             }
@@ -140,6 +140,7 @@ class Aireview extends Base
             $sContent = $oHelperFunction->filterAllTags($sContent);
             //将文章内容拆分参考文献
             $aDealContent = $this->dealContent($sContent);
+
             $sBefore= empty($aDealContent['before']) ? '' : $aDealContent['before'];
             $sReference = empty($aDealContent['after']) ?  '' :  $aDealContent['after'];
             if(in_array($sQuestionFields, ['attribute'])){//科学性和创新性
@@ -191,26 +192,67 @@ class Aireview extends Base
      * @title 将文章内容拆分参考文献
      * @param sContent 文章内容
      */
-    private function dealContent($sContent = ''){
+    private function dealContent($sContent = '',$regex = null){
         if(empty($sContent)){
-            return [];
+            return ['before' => '', 'after' => ''];
         }
 
-        //拆分字符串
+        // 1. 限制匹配范围（末尾30%）
-        $lastPos = strrpos($sContent, 'Reference');
+        $contentLength = strlen($sContent);
-        if ($lastPos === false) {
+        $searchStart = $contentLength > 5000 ? (int)($contentLength * 0.7) : 0;
-            $lastPos = strrpos($sContent, 'REFERENCE');
+        $searchContent = substr($sContent, $searchStart);
-            if($lastPos === false){
+
-                // 未找到 "Reference"，返回原字符串和空
+        // 2. 正则模式优化
-                return ['before' => $sContent, 'after' => ''];
+        if ($regex === null) {
-            }
+            $keywords = [
+                'references?', 'bibliograph(?:y|ies)', 
+                'works? cited', 'citation(?:s)?'
+            ];
+            $pattern = sprintf(
+                '/(?:^|\s)\s*#*\s*(%s)\s*[:\-–]?\s*(?:$|\s)/i',
+                implode('|', $keywords)
+            );
+            $regex = $pattern;
         }
 
-        // 拆分：关键词之前的内容
+        // 3. 匹配并处理结果
-        $before = substr($sContent, 0, $lastPos);
+        if (preg_match_all($regex, $searchContent, $matches, PREG_OFFSET_CAPTURE)) {
-        // 关键词及之后的内容（包含关键词本身）
+            $lastMatch = end($matches[0]);
-        $after = substr($sContent, $lastPos);
+            $refPosition = $searchStart + $lastMatch[1];
-        return ['before' => $before,'after' => $after];
+
+            // 4. 合并字符串处理
+            $before = substr($sContent, 0, $refPosition);
+            $after = substr($sContent, $refPosition);
+            
+            // 一次性处理空白和换行
+            $process = function($str) {
+                return str_replace("\n", '', trim($str));
+            };
+            return [
+                'before' => $process($before),
+                'after' => $process($after)
+            ];
+        }
+
+        // 未匹配时处理
+        return [
+            'before' => str_replace("\n", '', trim($sContent)),
+            'after' => ''
+        ];
+        // $lastPos = strrpos($sContent, 'Reference');
+        // if ($lastPos === false) {
+        //     $lastPos = strrpos($sContent, 'REFERENCE');
+        //     if($lastPos === false){
+        //         // 未找到 "Reference"，返回原字符串和空
+        //         return ['before' => $sContent, 'after' => ''];
+        //     }
+        // }
+        
+        // // 拆分：关键词之前的内容
+        // $before = substr($sContent, 0, $lastPos);
+        // // 关键词及之后的内容（包含关键词本身）
+        // $after = substr($sContent, $lastPos);
+        // return ['before' => $before,'after' => $after];
     }
 
     /**