自动查重

application/common/ArticleParserService.php

@@ -1153,12 +1153,12 @@ class ArticleParserService
     }
 
     /**
-     * 提取 Word 文档中的参考文献列表（仅返回数组，不做入库）
-     * @return array 每条为一个参考文献的纯文本字符串
+     * 按段落提取 Word 全文行（供正文裁切、参考文献识别等复用）
+     * @return array<int,string>
      */
-    public static function getReferencesFromWord($filePath): array
+    public static function collectParagraphLines($filePath): array
     {
-        $othis = new self($filePath) ;
+        $othis = new self($filePath);
         if (empty($othis->sections)) {
             return [];
         }
@@ -1166,13 +1166,26 @@ class ArticleParserService
         $lines = [];
         foreach ($othis->sections as $section) {
             foreach ($section->getElements() as $element) {
-                $text = $othis->getTextFromElement($element);
-                $text = trim((string)$text);
-                if ($text === '') continue;
-                $lines[] = $text;
+                $text = trim((string) $othis->getTextFromElement($element));
+                if ($text === '') {
+                    continue;
+                }
+                if (!mb_check_encoding($text, 'UTF-8')) {
+                    $text = mb_convert_encoding($text, 'UTF-8', 'GBK');
+                }
+                $lines[] = preg_replace('/\s+/u', ' ', $text);
             }
         }
+        return $lines;
+    }
 
+    /**
+     * 提取 Word 文档中的参考文献列表（仅返回数组，不做入库）
+     * @return array 每条为一个参考文献的纯文本字符串
+     */
+    public static function getReferencesFromWord($filePath): array
+    {
+        $lines = self::collectParagraphLines($filePath);
         if (empty($lines)) {
             return [];
         }