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/api/controller/AITools.php

@@ -0,0 +1,299 @@
+<?php
+/**
+ * AI 工具接口 — 编辑辅助：AI 辅助写作风险分析
+ */
+
+namespace app\api\controller;
+
+use app\common\ArticleParserService;
+use app\common\mq\AiWritingRiskMqPublisher;
+use app\common\PlagiarismService;
+use app\common\service\AiWritingRiskTaskService;
+use think\Validate;
+
+class Aitools extends Base
+{
+    public function __construct(\think\Request $request = null)
+    {
+        parent::__construct($request);
+    }
+
+    /**
+     * @title 写作风险检测可视化页面（前后端分离入口）
+     * @description 仅渲染 HTML；检测与进度走 JSON 接口
+     */
+    public function index()
+    {
+        return $this->renderRiskPage();
+    }
+
+    /**
+     * @title 开始 AI 辅助写作风险检测（异步入队）
+     * @description 立即返回 task_id；后台 RabbitMQ 消费执行检测
+     * @param url 稿件链接地址（.docx）
+     * @param p_article_id 可选，业务稿件ID
+     */
+    public function realtimeWritingRiskDetect()
+    {
+        $data = $this->request->param();
+        $rule = new Validate([
+            'url' => 'require',
+        ]);
+        if (!$rule->check($data)) {
+            return $this->detectJsonError($rule->getError());
+        }
+        $url = trim((string) $data['url']);
+        $pArticleId = intval($data['p_article_id'] ?? 0);
+        try {
+            $taskService = new AiWritingRiskTaskService();
+            $created = $taskService->createTask($url, $pArticleId);
+            // 新任务，或复用但仍在排队（可能丢消息）时重新投递
+            if (empty($created['reused']) || intval($created['status']) === AiWritingRiskTaskService::STATUS_PENDING) {
+                (new AiWritingRiskMqPublisher())->publishTaskStart(intval($created['task_id']), 'enqueue');
+            }
+
+            return json([
+                'code' => 200,
+                'msg'  => !empty($created['reused']) ? '已有进行中的检测任务' : '已开始检测',
+                'data' => [
+                    'task_id'          => $created['task_id'],
+                    'task_no'          => $created['task_no'],
+                    'status'           => $created['status'],
+                    'status_text'      => intval($created['status']) === 1 ? 'running' : 'queued',
+                    'progress_percent' => 0,
+                    'message'          => $created['message'],
+                    'reused'           => !empty($created['reused']),
+                ],
+            ]);
+        } catch (\Exception $e) {
+            return $this->detectJsonError($e->getMessage());
+        }
+    }
+
+    /**
+     * @title 刷新 AI 辅助写作风险检测进度
+     * @description
+     *   - 浏览器 GET（无 format=json）→ 可视化页面：未完成显示进度，完成后显示结果
+     *   - POST 或 format=json → JSON（供前端轮询）
+     * @param task_id 任务ID（与 task_no 二选一）
+     * @param task_no 任务号（与 task_id 二选一）
+     */
+    public function realtimeWritingRiskProgress()
+    {
+        $data = $this->request->param();
+        $taskId = intval($data['task_id'] ?? 0);
+        $taskNo = trim((string) ($data['task_no'] ?? ''));
+        if ($taskId <= 0 && $taskNo === '') {
+            if ($this->wantsProgressJson()) {
+                return $this->detectJsonError('task_id 或 task_no 必填');
+            }
+            return $this->renderRiskPage();
+        }
+
+        try {
+            $progress = (new AiWritingRiskTaskService())->getProgress($taskNo, $taskId);
+        } catch (\Exception $e) {
+            if ($this->wantsProgressJson()) {
+                return $this->detectJsonError($e->getMessage());
+            }
+            return $this->renderRiskPage([
+                'page_error' => $e->getMessage(),
+                'task_id'    => $taskId,
+                'task_no'    => $taskNo,
+            ]);
+        }
+
+        if ($this->wantsProgressJson()) {
+            return json([
+                'code' => 200,
+                'msg'  => 'success',
+                'data' => $progress,
+            ]);
+        }
+
+        return $this->renderRiskPage([
+            'task_id'       => intval($progress['task_id'] ?? $taskId),
+            'task_no'       => (string) ($progress['task_no'] ?? $taskNo),
+            'initial_data'  => $progress,
+        ]);
+    }
+
+    /**
+     * 是否返回 JSON：POST、format=json、XHR/fetch 显式要 JSON
+     */
+    private function wantsProgressJson(): bool
+    {
+        if (strtolower((string) $this->request->param('format', '')) === 'json') {
+            return true;
+        }
+        if ($this->request->isAjax()) {
+            return true;
+        }
+        $accept = strtolower((string) $this->request->header('accept', ''));
+        if (strpos($accept, 'application/json') !== false) {
+            return true;
+        }
+        // 页面内 fetch 使用 POST
+        if ($this->request->isPost()) {
+            return true;
+        }
+        return false;
+    }
+
+    /**
+     * 渲染可视化页
+     * @param array $opts task_id/task_no/initial_data/page_error
+     */
+    private function renderRiskPage(array $opts = [])
+    {
+        $apiStart = (string) url('api/aitools/realtimeWritingRiskDetect');
+        $apiProgress = (string) url('api/aitools/realtimeWritingRiskProgress');
+        $initial = isset($opts['initial_data']) && is_array($opts['initial_data'])
+            ? $opts['initial_data']
+            : null;
+
+        $this->assign([
+            'api_start_js'     => json_encode($apiStart, JSON_UNESCAPED_SLASHES),
+            'api_progress_js'  => json_encode($apiProgress, JSON_UNESCAPED_SLASHES),
+            'init_task_id_js'  => json_encode(intval($opts['task_id'] ?? 0)),
+            'init_task_no_js'  => json_encode((string) ($opts['task_no'] ?? ''), JSON_UNESCAPED_UNICODE),
+            'init_data_js'     => json_encode($initial, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES),
+            'page_error_js'    => json_encode((string) ($opts['page_error'] ?? ''), JSON_UNESCAPED_UNICODE),
+        ]);
+        return $this->fetch('aiTools/index');
+    }
+
+    /**
+     * @title AI 辅助写作风险分析（编辑辅助）
+     * @description 解析稿件结构，并从模版句、重复短语、句长特征等维度输出风险报告
+     * @param url 稿件链接地址（.docx）
+     */
+    public function writingRiskAnalysis()
+    {
+        $data = $this->request->param();
+        $rule = new Validate([
+            'url' => 'require',
+        ]);
+        if (!$rule->check($data)) {
+            return jsonError($rule->getError());
+        }
+
+        try {
+            $filePath = $this->resolveManuscriptUrl($data['url']);
+        } catch (\Exception $e) {
+            return jsonError($e->getMessage());
+        }
+
+        $result = ArticleParserService::analyzeWritingRisk($filePath);
+        if (empty($result['status']) || intval($result['status']) !== 1) {
+            return jsonError(empty($result['msg']) ? '稿件分析失败' : $result['msg']);
+        }
+
+        return jsonSuccess([
+            'manuscript'     => $this->pickManuscriptFields($result['data']),
+            'risk_analysis'  => $result['data']['risk_analysis'] ?? [],
+            'metrics'        => $this->pickMetricFields($result['data']),
+        ]);
+    }
+
+    /**
+     * @title 根据稿件链接解析稿件结构
+     * @description 从稿件地址解析标题、摘要、关键词及 IMRaD 正文分节（不含风险综述）
+     * @param url 稿件链接地址（.docx）
+     */
+    public function parseManuscript()
+    {
+        $data = $this->request->param();
+        $rule = new Validate([
+            'url' => 'require',
+        ]);
+        if (!$rule->check($data)) {
+            return jsonError($rule->getError());
+        }
+
+        try {
+            $filePath = $this->resolveManuscriptUrl($data['url']);
+        } catch (\Exception $e) {
+            return jsonError($e->getMessage());
+        }
+
+        $result = ArticleParserService::parseManuscriptStructure($filePath);
+        if (empty($result['status']) || intval($result['status']) !== 1) {
+            return jsonError(empty($result['msg']) ? '稿件解析失败' : $result['msg']);
+        }
+
+        return jsonSuccess($result['data']);
+    }
+
+    private function detectJsonError($msg)
+    {
+        return json(['code' => 500, 'msg' => (string) $msg, 'data' => []]);
+    }
+
+    private function pickManuscriptFields(array $data): array
+    {
+        $keys = ['title', 'abstract', 'keywords', 'Introduction', 'Methods', 'Results', 'Discussion', 'Conclusion', 'References'];
+        $out = [];
+        foreach ($keys as $key) {
+            if (array_key_exists($key, $data)) {
+                $out[$key] = $data[$key];
+            }
+        }
+        return $out;
+    }
+
+    private function pickMetricFields(array $data): array
+    {
+        $keys = [
+            'sentence_stats',
+            'sentence_stats_by_section',
+            'repeated_phrases',
+            'repeated_phrases_by_section',
+            'template_sentence_stats',
+        ];
+        $out = [];
+        foreach ($keys as $key) {
+            if (array_key_exists($key, $data)) {
+                $out[$key] = $data[$key];
+            }
+        }
+        return $out;
+    }
+
+    /**
+     * 将稿件链接解析为本地可读文件路径
+     */
+    private function resolveManuscriptUrl($url)
+    {
+        $url = trim((string)$url);
+        if ($url === '') {
+            throw new \Exception('稿件地址不能为空');
+        }
+
+        if (preg_match('/^([a-zA-Z]:[\\\\\/]|\/)/', $url) && is_file($url)) {
+            return $url;
+        }
+
+        if (preg_match('#^https?://#i', $url)) {
+            try {
+                return (new PlagiarismService())->resolveFileUrlToLocal($url);
+            } catch (\Exception $e) {
+                $path = parse_url($url, PHP_URL_PATH);
+                if ($path) {
+                    $local = rtrim(ROOT_PATH, '/') . $path;
+                    if (is_file($local)) {
+                        return $local;
+                    }
+                }
+                throw $e;
+            }
+        }
+
+        $local = rtrim(ROOT_PATH, '/') . '/public/' . ltrim(ltrim($url, '/'), 'public');
+        if (is_file($local)) {
+            return $local;
+        }
+
+        throw new \Exception('无法解析稿件文件路径: ' . $url);
+    }
+}

application/api/controller/Base.php

@@ -272,6 +272,14 @@ class Base extends Controller
         $this->production_article_refer_obj->where('p_article_id', $refer_info['p_article_id'])->where('index', ">", $refer_info['index'])->where('state', 0)->setDec('index');
         $this->production_article_refer_obj->where('p_refer_id', $p_refer_id)->update(['state' => 1]);
 
+        try {
+            Db::name('production_article_refer_author')
+                ->where('p_refer_id', intval($p_refer_id))
+                ->delete();
+        } catch (\Exception $e) {
+            \think\Log::error('delOneRefer delete refer_author p_refer_id=' . $p_refer_id . ' ' . $e->getMessage());
+        }
+
         // 文献集合已变更，原校对结果的 reference_no 已全部错位，整篇标记为未校对
         try {
             (new \app\common\ReferenceCheckService())

application/api/controller/Preaccept.php

@@ -7,6 +7,8 @@ use think\Env;
 use think\Queue;
 use think\Validate;
 use app\common\CrossrefService;
+use app\common\ReferenceCheckService;
+use app\common\ReferenceReferAuthorService;
 use app\common\ReferenceRelevanceCheckService;
 
 class Preaccept extends Base
@@ -36,6 +38,78 @@ class Preaccept extends Base
         }
     }
 
+    /**
+     * 新增参考文献后同步作者明细到 t_production_article_refer_author
+     */
+    private function syncReferAuthorsAfterInsert($pReferId, $pArticleId, array $refer)
+    {
+        $pReferId = intval($pReferId);
+        $pArticleId = intval($pArticleId);
+        if ($pReferId <= 0 || $pArticleId <= 0) {
+            return;
+        }
+        try {
+            (new ReferenceReferAuthorService())->syncOneRefer($pReferId, $pArticleId, $refer);
+        } catch (\Exception $e) {
+            \think\Log::error(
+                'syncReferAuthorsAfterInsert p_refer_id=' . $pReferId
+                . ' p_article_id=' . $pArticleId . ' ' . $e->getMessage()
+            );
+        }
+    }
+
+    /**
+     * 编辑参考文献后同步作者明细；refer_doi 变化时先删后重新抓取入库
+     */
+    private function syncReferAuthorsAfterUpdate($pReferId, $pArticleId, array $oldRefer, array $updata)
+    {
+        $pReferId = intval($pReferId);
+        $pArticleId = intval($pArticleId);
+        if ($pReferId <= 0 || $pArticleId <= 0 || empty($oldRefer)) {
+            return;
+        }
+
+        $newRefer = array_merge($oldRefer, $updata);
+        $refUtil = new ReferenceCheckService();
+        $oldDoi = $this->normalizeReferDoi($refUtil->extractDoiFromRefer($oldRefer));
+        $newDoi = $this->normalizeReferDoi($refUtil->extractDoiFromRefer($newRefer));
+        $doiChanged = $oldDoi !== $newDoi;
+
+        $authorChanged = false;
+        if (array_key_exists('author', $updata)) {
+            $authorChanged = trim((string)($oldRefer['author'] ?? '')) !== trim((string)$updata['author']);
+        }
+
+        if (!$doiChanged && !$authorChanged) {
+            return;
+        }
+
+        try {
+            $svc = new ReferenceReferAuthorService();
+            if ($doiChanged) {
+                Db::name('production_article_refer_author')->where('p_refer_id', $pReferId)->delete();
+                $svc->syncOneRefer($pReferId, $pArticleId, $newRefer);
+            } elseif ($authorChanged) {
+                $svc->syncFromReferAuthorField(
+                    $pReferId,
+                    $pArticleId,
+                    (string)($newRefer['author'] ?? '')
+                );
+            }
+        } catch (\Exception $e) {
+            \think\Log::error(
+                'syncReferAuthorsAfterUpdate p_refer_id=' . $pReferId
+                . ' p_article_id=' . $pArticleId . ' ' . $e->getMessage()
+            );
+        }
+    }
+
+    private function normalizeReferDoi($doi)
+    {
+        $doi = preg_replace('#^https?://(dx\.)?doi\.org/#i', '', trim((string)$doi));
+        return strtolower(trim($doi, " \t\n\r\0\x0B/"));
+    }
+
 
     /**获取文章参考文献列表
      * @return \think\response\Json
@@ -165,6 +239,11 @@ class Preaccept extends Base
         $adId= $this->production_article_refer_obj->insertGetId($insert);
         $this->production_article_refer_obj->where('p_article_id', $p_info['p_article_id'])->where("p_refer_id", "<>", $adId)->where("index", ">", $pre_refer['index'])->where('state', 0)->setInc('index');
         $this->resetArticleChecksOnReferChange(intval($p_info['p_article_id']), 'addRefer');
+        $this->syncReferAuthorsAfterInsert(
+            $adId,
+            intval($p_info['p_article_id']),
+            array_merge($insert, ['p_refer_id' => $adId, 'p_article_id' => $p_info['p_article_id']])
+        );
         return jsonSuccess([]);
 
 
@@ -222,6 +301,12 @@ class Preaccept extends Base
         $adId= $this->production_article_refer_obj->insertGetId($insert);
         $this->production_article_refer_obj->where('p_article_id', $p_info['p_article_id'])->where("p_refer_id", "<>", $adId)->where("index", ">", $pre_refer['index'])->where('state', 0)->setInc('index');
         $this->resetArticleChecksOnReferChange(intval($p_info['p_article_id']), 'addReferByParticleid');
+        $this->syncReferAuthorsAfterInsert(
+            $adId,
+            intval($p_info['p_article_id']),
+            array_merge($insert, ['p_refer_id' => $adId, 'p_article_id' => $p_info['p_article_id']])
+        );
+
         return jsonSuccess([]);
     }
 
@@ -498,6 +583,13 @@ class Preaccept extends Base
             \think\Log::error('editRefer enqueueRecheckByPReferId p_refer_id=' . $data['p_refer_id'] . ' ' . $e->getMessage());
         }
 
+        $this->syncReferAuthorsAfterUpdate(
+            intval($data['p_refer_id']),
+            intval($old_refer_info['p_article_id']),
+            $old_refer_info,
+            $updata
+        );
+
         return jsonSuccess([]);
     }
 

application/api/controller/References.php

@@ -13,6 +13,8 @@ use think\Env;
 use think\Queue;
 use app\common\ReferenceCheckService;
 use app\common\ReferenceRelevanceCheckService;
+use app\common\ReferenceStackingStatsService;
+use app\common\ReferenceReferAuthorService;
 use app\common\DbReconnectHelper;
 /**
  * @title 参考文献
@@ -1818,5 +1820,117 @@ class References extends Base
         return $frag;
     }
 
-    
+    /**
+     * 参考文献引用堆叠统计（同作者>15%、同刊>20%、自引>10%，实时计算）
+     *
+     * POST/GET: p_article_id（必填）
+     */
+    public function referenceStackingStats()
+    {
+        $aParam = $this->request->post();
+        if (empty($aParam)) {
+            $aParam = $this->request->param();
+        }
+
+        $iPArticleId = empty($aParam['p_article_id']) ? 0 : intval($aParam['p_article_id']);
+        if ($iPArticleId <= 0) {
+            return json_encode(array('status' => 2, 'msg' => 'Please select an article'));
+        }
+
+        try {
+            $svc = new ReferenceStackingStatsService();
+            $result = $svc->getThresholdStackingByPArticleId($iPArticleId);
+            return jsonSuccess($result);
+        } catch (\Exception $e) {
+            return jsonError($e->getMessage());
+        }
+    }
+
+    /**
+     * 测试：同步参考文献作者明细到 t_production_article_refer_author
+     * POST/GET: p_article_id（必填）, p_refer_id（可选，仅同步单条）, include_authors（可选，默认 0；1 返回作者明细）, sleep_ms（可选，默认 120）
+     */
+    public function referenceReferAuthorSyncTest()
+    {
+        $aParam = $this->request->post();
+        if (empty($aParam)) {
+            $aParam = $this->request->param();
+        }
+
+        $iPArticleId = empty($aParam['p_article_id']) ? 0 : intval($aParam['p_article_id']);
+        if ($iPArticleId <= 0) {
+            return json_encode(array('status' => 2, 'msg' => 'Please select an article'));
+        }
+
+        $iPReferId = empty($aParam['p_refer_id']) ? 0 : intval($aParam['p_refer_id']);
+        $includeAuthors = !empty($aParam['include_authors']) && intval($aParam['include_authors']) === 1;
+        $sleepMs = isset($aParam['sleep_ms']) ? max(0, intval($aParam['sleep_ms'])) : 120;
+
+        try {
+            $svc = new ReferenceReferAuthorService();
+
+            if ($iPReferId > 0) {
+                $refer = Db::name('production_article_refer')
+                    ->where('p_refer_id', $iPReferId)
+                    ->where('p_article_id', $iPArticleId)
+                    ->where('state', 0)
+                    ->find();
+                if (empty($refer)) {
+                    return jsonError('Reference not found');
+                }
+
+                $refUtil = new ReferenceCheckService();
+                $doi = $refUtil->extractDoiFromRefer($refer);
+                $count = 0;
+                if ($doi !== '') {
+                    $crossref = new CrossrefService([
+                        'mailto' => trim((string)Env::get('crossref_mailto', '')),
+                    ]);
+                    $summary = $crossref->fetchWorkSummary($doi);
+                    if ($summary === null || empty($summary['doi'])) {
+                        $summary = ['doi' => $doi, 'raw' => []];
+                    }
+                    $count = $svc->syncFromWorkSummary($iPReferId, $iPArticleId, $doi, $summary);
+                }
+                if ($count <= 0) {
+                    $count = $svc->syncFromReferAuthorField($iPReferId, $iPArticleId, (string)($refer['author'] ?? ''));
+                }
+
+                $result = [
+                    'p_article_id' => $iPArticleId,
+                    'p_refer_id'   => $iPReferId,
+                    'reference_no' => intval($refer['index']) + 1,
+                    'doi'          => $doi,
+                    'authors'      => $count,
+                    'synced'       => $count > 0 ? 1 : 0,
+                    'failed'       => $count > 0 ? 0 : 1,
+                ];
+            } else {
+                $result = $svc->syncByPArticleId($iPArticleId, ['sleep_ms' => $sleepMs]);
+            }
+
+            if ($includeAuthors) {
+                $query = Db::name('production_article_refer_author')
+                    ->alias('a')
+                    ->join('production_article_refer r', 'r.p_refer_id = a.p_refer_id', 'LEFT')
+                    ->field('a.p_refer_id,r.index,a.author_seq,a.display_name,a.citation_name,a.orcid,a.openalex_id,a.identity_source')
+                    ->where('a.p_article_id', $iPArticleId);
+                if ($iPReferId > 0) {
+                    $query->where('a.p_refer_id', $iPReferId);
+                }
+                $rows = $query->order('r.index asc, a.author_seq asc')->select();
+                foreach ($rows as &$row) {
+                    $row['reference_no'] = intval($row['index']) + 1;
+                    unset($row['index']);
+                }
+                unset($row);
+                $result['author_rows'] = $rows;
+            }
+
+            return jsonSuccess($result);
+        } catch (\Exception $e) {
+            return jsonError($e->getMessage());
+        }
+    }
+
 }