参考文献type

2026-07-09 15:42:55 +08:00
parent 4fd967b1d4
commit 4cf26ea682
4 changed files with 278 additions and 16 deletions
--- a/application/api/controller/Production.php
+++ b/application/api/controller/Production.php
@@ -11,6 +11,7 @@ use think\Queue;
 use think\Validate;
 use think\log;
 use app\common\ArticleSymbolNormalizer;
+use app\common\ReferenceDispatchService;

 /**
 * @title 公共管理相关
@@ -1931,7 +1932,7 @@ class Production extends Base
        }

        $this->referToDoi($data['p_article_id']);
-        $this->doiTofrag($data['p_article_id']);
+        (new ReferenceDispatchService())->dispatchRefersByType($data['p_article_id']);
        return jsonSuccess([]);
    }

@@ -1998,19 +1999,7 @@ class Production extends Base

    public function doiTofrag($p_article_id)
    {
-        $p_info = $this->production_article_obj->where('p_article_id', $p_article_id)->find();
-        $refers = $this->production_article_refer_obj->where('p_article_id', $p_info['p_article_id'])->where('state', 0)->select();
-        foreach ($refers as $v) {
-            if ($v['refer_doi'] == '') {
-                $this->production_article_refer_obj->where('p_refer_id', $v['p_refer_id'])->update(['refer_frag' => $v['refer_content']]);
-            } else {
-
-                //修改队列兼容对接OPENAI接口 chengxiaoling 20251128 start
-                // Queue::push('app\api\job\ts@fire1', $v, 'ts');
-                Queue::push('app\api\job\ArticleReferDetailQueue@fire', $v, 'ArticleReferDetailQueue');
-                //修改队列兼容对接OPENAI接口 chengxiaoling 20251128 end
-            }
-        }
+        (new ReferenceDispatchService())->dispatchRefersByType($p_article_id);
        return jsonSuccess([]);
    }

--- a/application/common/ProductionArticleRefer.php
+++ b/application/common/ProductionArticleRefer.php
@@ -54,7 +54,7 @@ class ProductionArticleRefer
        return json_encode(['status' => 1,'msg' => 'Add to reference processing queue']);
    }
    /**
-     * 处理参考文献
+     * 处理参考文献（单个）
     *
     * @return void
     */
@@ -71,10 +71,17 @@ class ProductionArticleRefer
        }
        //查询未处理过的数据
        $aWhere = ['p_refer_id' => $iPReferId,'p_article_id' => $iPArticleId,'state' => 0];
-        $aRefer = Db::name('production_article_refer')->field('refer_doi,refer_content')->where($aWhere)->find();
+        $aRefer = Db::name('production_article_refer')->field('refer_doi,refer_content,refer_type')->where($aWhere)->find();
        if(empty($aRefer)){
            return json_encode(array('status' => 2,'msg' => 'No reference records found'.json_encode($aParam)));
        }
+
+        // 非期刊类型已在分流阶段处理，队列不再走期刊解析
+        $referType = strtolower(trim((string)($aRefer['refer_type'] ?? 'journal')));
+        if (in_array($referType, ['book', 'other'], true)) {
+            return json_encode(['status' => 1, 'msg' => 'Skipped non-journal reference']);
+        }
+
        if(empty($aRefer['refer_doi'])){
            return json_encode(['status' => 4,'msg' => 'Reference DOI is empty'.json_encode($aParam)]);
        }
--- a/application/common/ReferenceDispatchService.php
+++ b/application/common/ReferenceDispatchService.php
@@ -0,0 +1,251 @@
+<?php
+
+namespace app\common;
+
+use think\Db;
+use think\Env;
+use think\Queue;
+
+/**
+ * 参考文献分流处理：freshRefers 入口之后，按 journal / book / other 三路处理。
+ *
+ * - journal：有 DOI 入 ArticleReferDetailQueue；无 DOI 保留原文
+ * - book：填充结构化字段 author/title/joura/dateno/isbn（与 Preaccept、References 约定一致）
+ * - other：refer_frag = refer_content，不做结构化
+ */
+class ReferenceDispatchService
+{
+    /** @var ReferenceTypeClassifier */
+    private $classifier;
+
+    public function __construct()
+    {
+        $this->classifier = new ReferenceTypeClassifier(['use_llm' => true]);
+    }
+
+    /**
+     * 对某篇生产文章的全部参考文献按类型分流处理
+     */
+    public function dispatchRefersByType($pArticleId)
+    {
+        $pArticleId = intval($pArticleId);
+        if ($pArticleId <= 0) {
+            return;
+        }
+
+        $refers = Db::name('production_article_refer')
+            ->where('p_article_id', $pArticleId)
+            ->where('state', 0)
+            ->order('index asc, p_refer_id asc')
+            ->select();
+
+        if (empty($refers)) {
+            return;
+        }
+
+        $crossref = new CrossrefService([
+            'mailto' => trim((string)Env::get('crossref_mailto', '')),
+        ]);
+
+        foreach ($refers as $refer) {
+            $summary = null;
+            $crossrefType = '';
+            if (trim((string)$refer['refer_doi']) !== '') {
+                $doiNorm = $this->normalizeDoi($refer['refer_doi']);
+                if ($doiNorm !== '') {
+                    $summary = $crossref->fetchWorkSummary($doiNorm);
+                    if ($summary && !empty($summary['raw']['type'])) {
+                        $crossrefType = (string)$summary['raw']['type'];
+                    }
+                }
+            }
+
+            $typeInfo = $this->classifier->classify((string)$refer['refer_content'], $crossrefType);
+            $dispatchType = $this->classifier->normalizeDispatchType($typeInfo['type']);
+
+            Db::name('production_article_refer')->where('p_refer_id', $refer['p_refer_id'])->update([
+                'refer_type'  => $dispatchType,
+                'update_time' => time(),
+            ]);
+            $refer['refer_type'] = $dispatchType;
+
+            switch ($dispatchType) {
+                case ReferenceTypeClassifier::TYPE_BOOK:
+                    $this->processBookRefer($refer, $summary, $crossref);
+                    break;
+                case ReferenceTypeClassifier::TYPE_OTHER:
+                    $this->processOtherRefer($refer);
+                    break;
+                default:
+                    $this->processJournalRefer($refer);
+                    break;
+            }
+        }
+    }
+
+    private function processJournalRefer(array $refer)
+    {
+        $pReferId = intval($refer['p_refer_id']);
+        if (trim((string)$refer['refer_doi']) === '') {
+            Db::name('production_article_refer')->where('p_refer_id', $pReferId)->update([
+                'refer_frag'  => $refer['refer_content'],
+                'refer_type'  => ReferenceTypeClassifier::TYPE_JOURNAL,
+                'is_deal'     => 1,
+                'update_time' => time(),
+            ]);
+            return;
+        }
+
+        Queue::push('app\api\job\ArticleReferDetailQueue@fire', $refer, 'ArticleReferDetailQueue');
+    }
+
+    private function processOtherRefer(array $refer)
+    {
+        Db::name('production_article_refer')->where('p_refer_id', intval($refer['p_refer_id']))->update([
+            'refer_frag'  => $refer['refer_content'],
+            'refer_type'  => ReferenceTypeClassifier::TYPE_OTHER,
+            'cs'          => 0,
+            'is_deal'     => 1,
+            'update_time' => time(),
+        ]);
+    }
+
+    /**
+     * book：结构化字段，DOI 仅用于补数据
+     */
+    private function processBookRefer(array $refer, $summary, CrossrefService $crossref)
+    {
+        $pReferId = intval($refer['p_refer_id']);
+        $content = (string)$refer['refer_content'];
+        $update = [
+            'refer_type'  => ReferenceTypeClassifier::TYPE_BOOK,
+            'is_deal'     => 1,
+            'update_time' => time(),
+        ];
+
+        if (is_array($summary) && !empty($summary['raw'])) {
+            $raw = $summary['raw'];
+            $authorCitation = $crossref->getAuthorsCitation($raw, 3);
+            $update['author'] = $authorCitation !== '' ? rtrim($authorCitation, '.') . '.' : '';
+            $update['title'] = trim((string)($summary['title'] ?? ''));
+            $update['joura'] = $this->extractBookPublisher($raw, $summary);
+            $update['dateno'] = $this->extractBookDateno($raw);
+            $isbn = $this->extractIsbnFromRaw($raw);
+            if ($isbn === '' && !empty($refer['refer_doi'])) {
+                $doi = $this->normalizeDoi($refer['refer_doi']);
+                $isbn = $doi !== '' ? 'https://doi.org/' . $doi : '';
+            }
+            $update['isbn'] = $isbn;
+            $update['is_ja'] = 1;
+        } else {
+            $parsed = $this->parseBookFromContent($content);
+            $update = array_merge($update, $parsed);
+        }
+
+        $hasCore = trim((string)($update['author'] ?? '')) !== ''
+            && trim((string)($update['title'] ?? '')) !== '';
+        $update['cs'] = $hasCore ? 1 : 0;
+        if (!$hasCore) {
+            $update['refer_frag'] = $content;
+        }
+
+        Db::name('production_article_refer')->where('p_refer_id', $pReferId)->update($update);
+    }
+
+    private function extractBookPublisher(array $raw, array $summary)
+    {
+        $publisher = trim((string)($raw['publisher'] ?? ''));
+        if ($publisher !== '') {
+            return $publisher;
+        }
+        $pub = $summary['publisher'] ?? [];
+        if (!empty($pub['publisher'])) {
+            return trim((string)$pub['publisher']);
+        }
+        if (!empty($pub['title'])) {
+            return trim((string)$pub['title']);
+        }
+        return '';
+    }
+
+    private function extractBookDateno(array $raw)
+    {
+        if (!empty($raw['published']['date-parts'][0][0])) {
+            return (string)$raw['published']['date-parts'][0][0];
+        }
+        if (!empty($raw['issued']['date-parts'][0][0])) {
+            return (string)$raw['issued']['date-parts'][0][0];
+        }
+        if (!empty($raw['created']['date-parts'][0][0])) {
+            return (string)$raw['created']['date-parts'][0][0];
+        }
+        return '';
+    }
+
+    private function extractIsbnFromRaw(array $raw)
+    {
+        if (empty($raw['ISBN']) || !is_array($raw['ISBN'])) {
+            return '';
+        }
+        foreach ($raw['ISBN'] as $isbn) {
+            $isbn = trim((string)$isbn);
+            if ($isbn !== '') {
+                return $isbn;
+            }
+        }
+        return '';
+    }
+
+    /**
+     * 无 Crossref 时从原文尽量抽取 book 结构化字段
+     */
+    private function parseBookFromContent($content)
+    {
+        $content = trim((string)$content);
+        $out = [
+            'author' => '',
+            'title'  => '',
+            'joura'  => '',
+            'dateno' => '',
+            'isbn'   => '',
+            'is_ja'  => 1,
+        ];
+
+        if ($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)
+    {
+        $doi = preg_replace('#^https?://(dx\.)?doi\.org/#i', '', trim((string)$doi));
+        return trim($doi, " \t\n\r\0\x0B/");
+    }
+}
--- a/application/common/ReferenceTypeClassifier.php
+++ b/application/common/ReferenceTypeClassifier.php
@@ -81,6 +81,21 @@ class ReferenceTypeClassifier
    /**
     * CrossRef type → 内部枚举映射；未命中返回空串
     */
+    /**
+     * 将细分类归并为排版用的三类：journal / book / other
+     */
+    public function normalizeDispatchType($type)
+    {
+        $type = strtolower(trim((string)$type));
+        if ($type === self::TYPE_BOOK) {
+            return self::TYPE_BOOK;
+        }
+        if ($type === self::TYPE_JOURNAL) {
+            return self::TYPE_JOURNAL;
+        }
+        return self::TYPE_OTHER;
+    }
+
    public function mapCrossrefType($crossrefType)
    {
        $t = strtolower(trim((string)$crossrefType));