参考文献作者堆叠

参考文献相关性检测
作者ai写作辅助检测工作

application/common/BackgroundCheckService.php

@@ -104,6 +104,64 @@ class BackgroundCheckService
         return ['count' => count($list), 'list' => $list, 'source' => 'openalex'];
     }
 
+    /**
+     * 按 DOI 从 OpenAlex 获取文献及作者身份（用于引用堆叠精准作者定位）
+     */
+    public function fetchOpenAlexWorkByDoi($doi)
+    {
+        $doi = $this->cleanDoi($doi);
+        if ($doi === '') {
+            return ['success' => false, 'error' => 'DOI为空'];
+        }
+
+        $res = $this->openAlexGet('/works/https://doi.org/' . rawurlencode($doi));
+        if (!$res['success']) {
+            return $res;
+        }
+
+        return ['success' => true, 'work' => $res['data']];
+    }
+
+    /**
+     * 解析 OpenAlex work 中的作者身份（OpenAlex ID + ORCID）
+     *
+     * @return array<int, array{openalex_id:string,orcid:string,display_name:string,author_position:string,is_corresponding:bool,identity_keys:string[]}>
+     */
+    public function parseWorkAuthorships(array $work)
+    {
+        $list = [];
+        foreach ($work['authorships'] ?? [] as $auth) {
+            if (!is_array($auth)) {
+                continue;
+            }
+            $author = is_array($auth['author'] ?? null) ? $auth['author'] : [];
+            $openalexId = $this->extractOpenAlexId($author['id'] ?? '');
+            $orcid = $this->cleanOrcid($author['orcid'] ?? '');
+            if ($openalexId === '' && $orcid === '') {
+                continue;
+            }
+
+            $identityKeys = [];
+            if ($openalexId !== '') {
+                $identityKeys[] = 'openalex:' . $openalexId;
+            }
+            if ($orcid !== '') {
+                $identityKeys[] = 'orcid:' . $orcid;
+            }
+
+            $list[] = [
+                'openalex_id'      => $openalexId,
+                'orcid'            => $orcid,
+                'display_name'     => trim((string)($author['display_name'] ?? '')),
+                'author_position'  => (string)($auth['author_position'] ?? ''),
+                'is_corresponding' => !empty($auth['is_corresponding']),
+                'identity_keys'    => $identityKeys,
+            ];
+        }
+
+        return $list;
+    }
+
     public function fetchRecentWorks($openAlexId, $limit = 5)
     {
         $res = $this->openAlexGet('/works', [
@@ -604,7 +662,7 @@ class BackgroundCheckService
         ];
     }
 
-    private function parseCrossRefAuthors($authorList)
+    public function parseCrossRefAuthors($authorList)
     {
         if (empty($authorList) || !is_array($authorList)) {
             return [];
@@ -612,16 +670,52 @@ class BackgroundCheckService
 
         $result = [];
         foreach ($authorList as $a) {
+            $orcid = '';
+            if (!empty($a['ORCID']) && is_array($a['ORCID'])) {
+                $orcid = $this->cleanOrcid((string)($a['ORCID'][0] ?? ''));
+            } elseif (!empty($a['ORCID']) && is_string($a['ORCID'])) {
+                $orcid = $this->cleanOrcid($a['ORCID']);
+            }
+
             $result[] = [
                 'given'  => $a['given'] ?? '',
                 'family' => $a['family'] ?? '',
                 'name'   => isset($a['name']) ? $a['name'] : trim(($a['given'] ?? '') . ' ' . ($a['family'] ?? '')),
-                'orcid'  => $a['ORCID'] ?? '',
+                'orcid'  => $orcid,
             ];
         }
         return $result;
     }
 
+    /**
+     * Crossref 作者列表 → 带 identity_keys 的结构（ORCID 兜底）
+     *
+     * @return array<int, array{openalex_id:string,orcid:string,display_name:string,author_position:string,is_corresponding:bool,identity_keys:string[]}>
+     */
+    public function authorshipsFromCrossrefAuthors(array $authorList)
+    {
+        $parsed = $this->parseCrossRefAuthors($authorList);
+        $list = [];
+        $total = count($parsed);
+        foreach ($parsed as $i => $a) {
+            $orcid = trim((string)($a['orcid'] ?? ''));
+            if ($orcid === '') {
+                continue;
+            }
+            $displayName = trim((string)($a['name'] ?? ''));
+            $position = ($i === 0) ? 'first' : (($i === $total - 1) ? 'last' : 'middle');
+            $list[] = [
+                'openalex_id'      => '',
+                'orcid'            => $orcid,
+                'display_name'     => $displayName,
+                'author_position'  => $position,
+                'is_corresponding' => false,
+                'identity_keys'    => ['orcid:' . $orcid],
+            ];
+        }
+        return $list;
+    }
+
     private function parseDateParts($dateObj)
     {
         if (!isset($dateObj['date-parts'][0])) {