参考文献格式--book

application/common/ReferenceDispatchService.php

@@ -24,11 +24,15 @@ class ReferenceDispatchService
     /** @var BookMetadataService */
     private $bookMetadata;
 
+    /** @var BookCitationParser */
+    private $bookParser;
+
     public function __construct()
     {
         $this->classifier = new ReferenceTypeClassifier(['use_llm' => true]);
         $this->metadata = new ReferenceMetadataService();
         $this->bookMetadata = new BookMetadataService();
+        $this->bookParser = new BookCitationParser();
     }
 
     /**
@@ -181,7 +185,9 @@ class ReferenceDispatchService
         $summary = is_array($meta) ? $meta['crossref_summary'] : null;
         $raw = is_array($summary) ? ($summary['raw'] ?? []) : [];
         $hasMeta = is_array($meta)
-            && (trim((string)$meta['title']) !== '' || trim((string)$meta['author']) !== '');
+            && (trim((string)$meta['title']) !== '' || trim((string)$meta['author']) !== '')
+            // 图书著录里挂的 DOI 常常是被错误关联的期刊文章，题名对不上就不能拿来覆盖
+            && $this->metaMatchesContent($meta, $content);
 
         if ($hasMeta) {
             $authorCitation = trim((string)$meta['author']);
@@ -217,6 +223,44 @@ class ReferenceDispatchService
         Db::name('production_article_refer')->where('p_refer_id', $pReferId)->update($update);
     }
 
+    /**
+     * DOI 抓回来的题名是否确实是这条著录说的那本书
+     */
+    private function metaMatchesContent($meta, $content)
+    {
+        $title = $this->normalizeForMatch((string)$meta['title']);
+        if ($title === '') {
+            return false;
+        }
+        $blob = $this->normalizeForMatch($content);
+        if ($blob === '' || strpos($blob, $title) !== false) {
+            return true;
+        }
+
+        $words = array_filter(explode(' ', $title), function ($w) {
+            return strlen($w) > 3;
+        });
+        if (empty($words)) {
+            return false;
+        }
+        $hit = 0;
+        foreach ($words as $word) {
+            if (strpos($blob, $word) !== false) {
+                $hit++;
+            }
+        }
+
+        return ($hit / count($words)) >= 0.6;
+    }
+
+    private function normalizeForMatch($text)
+    {
+        $text = strtolower(trim((string)$text));
+        $text = preg_replace('/[^a-z0-9\x{4e00}-\x{9fff}\s]+/u', ' ', $text);
+
+        return trim(preg_replace('/\s+/u', ' ', $text));
+    }
+
     /**
      * 按书名/作者/年份/版次反查 ISBN，顺带补齐缺失的出版社与出版年
      *
@@ -299,50 +343,22 @@ class ReferenceDispatchService
     }
 
     /**
-     * 无 Crossref 时从原文尽量抽取 book 结构化字段
+     * 无可用 DOI 元数据时，从原文著录抽取 book 结构化字段
      */
     private function parseBookFromContent($content)
     {
-        $content = trim((string)$content);
-        $out = [
-            'author' => '',
-            'title'  => '',
-            'joura'  => '',
-            'dateno' => '',
-            'isbn'   => '',
+        $parsed = $this->bookParser->parse($content);
+
+        $author = trim((string)$parsed['author']);
+
+        return [
+            'author' => $author !== '' ? rtrim($author, '.') . '.' : '',
+            'title'  => trim((string)$parsed['title']),
+            'joura'  => trim((string)$parsed['publisher']),
+            'dateno' => trim((string)$parsed['year']),
+            'isbn'   => trim((string)$parsed['isbn']),
             'is_ja'  => 1,
         ];
-
-        if ($content === '') {
-            return $out;
-        }
-
-        if (preg_match('/\bISBN[:\s]*([\d\-Xx\s]+)/i', $content, $m)) {
-            $out['isbn'] = preg_replace('/\s+/', '-', trim($m[1]));
-        }
-
-        if (preg_match('/\b(19|20)\d{2}\b/', $content, $m)) {
-            $out['dateno'] = $m[0];
-        }
-
-        // Place: Publisher; Year  → joura 取 Publisher
-        if (preg_match('/:\s*([^;]+);\s*(19|20)\d{2}/', $content, $m)) {
-            $out['joura'] = trim($m[1]);
-        } elseif (preg_match('/\b([A-Z][A-Za-z .&]+(?:Press|Publishing|Publisher|Books?))\b/i', $content, $m)) {
-            $out['joura'] = trim($m[1]);
-        }
-
-        // 作者. 标题. ... 简单拆分
-        $parts = preg_split('/\.\s+/', $content, 3);
-        if (is_array($parts) && count($parts) >= 2) {
-            $out['author'] = trim($parts[0]);
-            if (substr($out['author'], -1) !== '.') {
-                $out['author'] .= '.';
-            }
-            $out['title'] = trim(rtrim($parts[1], '.'));
-        }
-
-        return $out;
     }
 
     private function normalizeDoi($doi)