自动查重

application/common/UserFieldAiService.php

@@ -0,0 +1,404 @@
+<?php
+
+namespace app\common;
+
+use think\Db;
+use think\Env;
+use think\Exception;
+use think\Queue;
+
+/**
+ * 根据投稿记录 / 审稿人资料，用大模型总结用户主领域（中文）写入 field_ai。
+ * 队列链：UserFieldAiFill → 处理一条 → enqueueNextFieldAi → 下一条。
+ */
+class UserFieldAiService
+{
+    const QUEUE_NAME = 'UserFieldAi';
+
+    const STATUS_PENDING = 0;
+    const STATUS_DONE = 1;
+    const STATUS_INSUFFICIENT = 2;
+    const STATUS_FAILED = 3;
+
+    private $logFile;
+
+    public function __construct()
+    {
+        $this->logFile = ROOT_PATH . 'runtime' . DS . 'user_field_ai.log';
+    }
+
+    /**
+     * 启动链式处理（从 user_id=0 之后找第一个待处理用户）。
+     *
+     * @param bool $force  true 时重算已生成用户
+     * @return bool 是否已推入首条 job
+     */
+    public function startChain($force = false, $delay = 1, $queue = '')
+    {
+        return $this->enqueueNextFieldAi($delay, $queue, 0, $force);
+    }
+
+    /**
+     * 链式：找 user_id > $afterUserId 的下一位待处理用户并入队。
+     */
+    public function enqueueNextFieldAi($delay = 1, $queue = '', $afterUserId = 0, $force = false)
+    {
+        if ($queue === '') {
+            $queue = self::QUEUE_NAME;
+        }
+        $afterUserId = intval($afterUserId);
+        $userId = $this->findNextPendingUserId($afterUserId, $force);
+        if ($userId <= 0) {
+            $this->log('[FieldAi] chain finished after user_id=' . $afterUserId . ' force=' . ($force ? '1' : '0'));
+            return false;
+        }
+
+        $data = [
+            'user_id' => $userId,
+            'queue'   => $queue,
+            'force'   => $force ? 1 : 0,
+        ];
+        $jobClass = 'app\\api\\job\\UserFieldAiFill@fire';
+        if ($delay > 0) {
+            Queue::later($delay, $jobClass, $data, $queue);
+        } else {
+            Queue::push($jobClass, $data, $queue);
+        }
+        $this->log('[FieldAi] enqueued user_id=' . $userId . ' queue=' . $queue);
+        return true;
+    }
+
+    /**
+     * 处理单个用户（队列 Job 或同步调试）。
+     *
+     * @return array{ok:bool, skipped?:bool, insufficient?:bool, field_ai?:string, error?:string}
+     */
+    public function processUser($userId, $force = false)
+    {
+        $userId = intval($userId);
+        if ($userId <= 0) {
+            return ['ok' => false, 'error' => 'invalid user_id'];
+        }
+
+        $this->ensureReviewerInfoRow($userId);
+        $uri = Db::name('user_reviewer_info')->where('reviewer_id', $userId)->find();
+        if (!$uri) {
+            return ['ok' => false, 'error' => 'reviewer_info missing'];
+        }
+
+        if (!$force && intval($uri['field_ai_status']) === self::STATUS_DONE && trim((string)$uri['field_ai']) !== '') {
+            return ['ok' => true, 'skipped' => true, 'field_ai' => (string)$uri['field_ai']];
+        }
+
+        if (!$this->isEligible($userId, $uri)) {
+            $this->updateFieldAi($userId, '', self::STATUS_INSUFFICIENT, 'insufficient profile/articles');
+            return ['ok' => true, 'insufficient' => true];
+        }
+
+        try {
+            $context = $this->buildContext($userId, $uri);
+            $fieldAi = $this->summarizeWithLlm($context);
+            if ($fieldAi === '') {
+                throw new Exception('LLM returned empty field');
+            }
+            $this->updateFieldAi($userId, $fieldAi, self::STATUS_DONE, '');
+            return ['ok' => true, 'field_ai' => $fieldAi];
+        } catch (\Throwable $e) {
+            $this->updateFieldAi($userId, '', self::STATUS_FAILED, mb_substr($e->getMessage(), 0, 500));
+            $this->log('[FieldAi] user_id=' . $userId . ' fail: ' . $e->getMessage());
+            return ['ok' => false, 'error' => $e->getMessage()];
+        }
+    }
+
+    /**
+     * 是否满足「可总结」：有投稿 或 审稿人资料较全。
+     */
+    public function isEligible($userId, $uri = null)
+    {
+        if ($this->hasSubmittedArticles($userId)) {
+            return true;
+        }
+        if ($uri === null) {
+            $uri = Db::name('user_reviewer_info')->where('reviewer_id', $userId)->find();
+        }
+        return $this->isReviewerProfileComplete($uri);
+    }
+
+    public function hasSubmittedArticles($userId)
+    {
+        $n = Db::name('article')
+            ->where('user_id', intval($userId))
+            ->where('title', '<>', '')
+            ->count();
+        return $n > 0;
+    }
+
+    /**
+     * 审稿人资料字段填充数达到阈值视为「较全」。
+     */
+    public function isReviewerProfileComplete($uri)
+    {
+        if (!$uri || !is_array($uri)) {
+            return false;
+        }
+        $minFilled = max(3, (int) Env::get('user_field_ai.min_profile_fields', 4));
+        $keys = ['field', 'company', 'country', 'technical', 'introduction', 'department', 'website'];
+        $filled = 0;
+        foreach ($keys as $k) {
+            if (!empty($uri[$k]) && trim((string)$uri[$k]) !== '') {
+                $filled++;
+            }
+        }
+        if (!empty($uri['major']) && trim((string)$uri['major']) !== '' && trim((string)$uri['major']) !== '0') {
+            $filled++;
+        }
+        $majorCount = Db::name('major_to_user')->where('user_id', intval($uri['reviewer_id']))->where('state', 0)->count();
+        if ($majorCount > 0) {
+            $filled++;
+        }
+        return $filled >= $minFilled;
+    }
+
+    private function findNextPendingUserId($afterUserId, $force)
+    {
+        $batch = max(20, (int) Env::get('user_field_ai.scan_batch', 80));
+        $cursor = intval($afterUserId);
+
+        while (true) {
+            $query = Db::name('user')->alias('u')
+                ->leftJoin('t_user_reviewer_info uri', 'uri.reviewer_id = u.user_id')
+                ->where('u.user_id', '>', $cursor);
+            if (!$force) {
+                $query->where(function ($q) {
+                    $q->where('uri.field_ai_status', self::STATUS_PENDING)
+                        ->whereOr('uri.field_ai_status', self::STATUS_FAILED)
+                        ->whereOr('uri.reviewer_info_id', 'null');
+                });
+            }
+            $ids = $query->order('u.user_id asc')->limit($batch)->column('u.user_id');
+
+            if (empty($ids)) {
+                return 0;
+            }
+
+            foreach ($ids as $uid) {
+                $uid = intval($uid);
+                $cursor = $uid;
+                $this->ensureReviewerInfoRow($uid);
+                $uri = Db::name('user_reviewer_info')->where('reviewer_id', $uid)->find();
+                if (!$force && intval($uri['field_ai_status']) === self::STATUS_DONE) {
+                    continue;
+                }
+                if (!$force && intval($uri['field_ai_status']) === self::STATUS_INSUFFICIENT) {
+                    continue;
+                }
+                if ($this->isEligible($uid, $uri)) {
+                    return $uid;
+                }
+                if (!$force) {
+                    $this->updateFieldAi($uid, '', self::STATUS_INSUFFICIENT, 'auto skip: insufficient data');
+                }
+            }
+        }
+    }
+
+    private function buildContext($userId, array $uri)
+    {
+        $user = Db::name('user')->where('user_id', $userId)->field('user_id,realname,email,account')->find();
+        $majorTitles = $this->resolveMajorTitles($userId, $uri);
+
+        $maxArticles = max(1, min(10, (int) Env::get('user_field_ai.max_articles', 5)));
+        $articles = Db::name('article')
+            ->where('user_id', $userId)
+            ->where('title', '<>', '')
+            ->order('article_id desc')
+            ->limit($maxArticles)
+            ->field('article_id,title,keywords,abstrart,journal_id,ctime')
+            ->select();
+
+        $journalNames = [];
+        if (!empty($articles)) {
+            $jids = array_unique(array_filter(array_column($articles, 'journal_id')));
+            if (!empty($jids)) {
+                $journalNames = Db::name('journal')->where('journal_id', 'in', $jids)->column('title', 'journal_id');
+            }
+        }
+
+        $articleBlocks = [];
+        foreach ($articles as $a) {
+            $jid = intval($a['journal_id']);
+            $articleBlocks[] = [
+                'title'    => (string) $a['title'],
+                'journal'  => isset($journalNames[$jid]) ? (string) $journalNames[$jid] : '',
+                'keywords' => (string) ($a['keywords'] ?? ''),
+                'abstract' => mb_substr(trim((string) ($a['abstrart'] ?? '')), 0, 800),
+            ];
+        }
+
+        return [
+            'user'     => [
+                'realname' => $user ? (string) $user['realname'] : '',
+                'email'    => $user ? (string) $user['email'] : '',
+            ],
+            'profile'  => [
+                'field'        => trim((string) ($uri['field'] ?? '')),
+                'technical'    => trim((string) ($uri['technical'] ?? '')),
+                'company'      => trim((string) ($uri['company'] ?? '')),
+                'department'   => trim((string) ($uri['department'] ?? '')),
+                'country'      => trim((string) ($uri['country'] ?? '')),
+                'introduction' => mb_substr(trim((string) ($uri['introduction'] ?? '')), 0, 1200),
+                'website'      => trim((string) ($uri['website'] ?? '')),
+                'majors'       => $majorTitles,
+            ],
+            'articles' => $articleBlocks,
+        ];
+    }
+
+    private function resolveMajorTitles($userId, array $uri)
+    {
+        $titles = [];
+        $ids = Db::name('major_to_user')->where('user_id', $userId)->where('state', 0)->column('major_id');
+        if (!empty($ids)) {
+            $titles = Db::name('reviewer_major')->where('major_id', 'in', $ids)->where('state', 0)->column('title');
+        }
+        if (empty($titles) && !empty($uri['major'])) {
+            $legacy = array_filter(array_map('intval', explode(',', (string) $uri['major'])));
+            if (!empty($legacy)) {
+                $titles = Db::name('reviewer_major')->where('major_id', 'in', $legacy)->column('title');
+            }
+        }
+        return array_values(array_unique(array_filter(array_map('trim', $titles))));
+    }
+
+    private function summarizeWithLlm(array $context)
+    {
+        $url = trim((string) Env::get('user_field_ai.chat_url', Env::get('expert_country_chat_url', Env::get('citation_chat_url', ''))));
+        $model = trim((string) Env::get('user_field_ai.chat_model', Env::get('expert_country_chat_model', Env::get('citation_chat_model', 'gpt-4.1'))));
+        $apiKey = trim((string) Env::get('user_field_ai.chat_api_key', Env::get('expert_country_chat_api_key', Env::get('citation_chat_api_key', ''))));
+        if ($url === '' || $model === '') {
+            throw new Exception('user_field_ai chat not configured (chat_url / chat_model)');
+        }
+
+        $payloadJson = json_encode($context, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES);
+        $messages = [
+            [
+                'role'    => 'system',
+                'content' => '你是学术领域分类助手。根据用户的投稿与个人资料，用简体中文给出该用户最主要的研究领域总结。'
+                    . '要求：精确、简洁，1～3 个中文领域词或短短语，用顿号分隔；不要解释、不要英文、不要 JSON 以外的多余文字。'
+                    . '只输出 JSON：{"field_ai":"..."}。',
+            ],
+            [
+                'role'    => 'user',
+                'content' => "请根据以下 JSON 资料总结该用户的主要研究领域：\n" . $payloadJson,
+            ],
+        ];
+
+        $body = [
+            'model'       => $model,
+            'temperature' => 0.2,
+            'messages'    => $messages,
+        ];
+
+        $ch = curl_init();
+        curl_setopt_array($ch, [
+            CURLOPT_URL            => $url,
+            CURLOPT_POST           => true,
+            CURLOPT_POSTFIELDS     => json_encode($body, JSON_UNESCAPED_UNICODE),
+            CURLOPT_RETURNTRANSFER => true,
+            CURLOPT_CONNECTTIMEOUT => 15,
+            CURLOPT_TIMEOUT        => max(30, (int) Env::get('user_field_ai.timeout', 90)),
+            CURLOPT_HTTPHEADER     => array_filter([
+                'Content-Type: application/json',
+                $apiKey !== '' ? 'Authorization: Bearer ' . $apiKey : null,
+            ]),
+        ]);
+        $raw = curl_exec($ch);
+        $code = (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
+        $err = curl_error($ch);
+        curl_close($ch);
+
+        if ($raw === false) {
+            throw new Exception('LLM curl error: ' . $err);
+        }
+        if ($code < 200 || $code >= 300) {
+            throw new Exception('LLM HTTP ' . $code . ': ' . mb_substr((string) $raw, 0, 400));
+        }
+
+        $data = json_decode($raw, true);
+        $content = '';
+        if (is_array($data) && isset($data['choices'][0]['message']['content'])) {
+            $content = trim((string) $data['choices'][0]['message']['content']);
+        } elseif (is_string($raw)) {
+            $content = trim($raw);
+        }
+
+        $fieldAi = $this->parseFieldAiFromContent($content);
+        if ($fieldAi === '' && $content !== '') {
+            $fieldAi = $this->cleanFieldAiText($content);
+        }
+        return $fieldAi;
+    }
+
+    private function parseFieldAiFromContent($content)
+    {
+        $content = trim((string) $content);
+        if ($content === '') {
+            return '';
+        }
+        $content = preg_replace('/^```[a-zA-Z]*\s*|```$/m', '', $content);
+        if (preg_match('/\{.*\}/s', $content, $m)) {
+            $obj = json_decode($m[0], true);
+            if (is_array($obj) && !empty($obj['field_ai'])) {
+                return $this->cleanFieldAiText((string) $obj['field_ai']);
+            }
+        }
+        $obj = json_decode($content, true);
+        if (is_array($obj) && !empty($obj['field_ai'])) {
+            return $this->cleanFieldAiText((string) $obj['field_ai']);
+        }
+        return '';
+    }
+
+    private function cleanFieldAiText($text)
+    {
+        $text = trim((string) $text);
+        $text = trim($text, "\"' \t\n\r");
+        $text = preg_replace('/\s+/u', '', $text);
+        if (mb_strlen($text) > 200) {
+            $text = mb_substr($text, 0, 200);
+        }
+        return $text;
+    }
+
+    public function ensureReviewerInfoRow($userId)
+    {
+        $exists = Db::name('user_reviewer_info')->where('reviewer_id', $userId)->find();
+        if ($exists) {
+            return;
+        }
+        Db::name('user_reviewer_info')->insert([
+            'reviewer_id' => $userId,
+            'ctime'       => time(),
+            'state'       => 0,
+        ]);
+    }
+
+    private function updateFieldAi($userId, $fieldAi, $status, $note)
+    {
+        $data = [
+            'field_ai'        => mb_substr(trim((string) $fieldAi), 0, 512),
+            'field_ai_status' => intval($status),
+            'field_ai_utime'  => time(),
+        ];
+        Db::name('user_reviewer_info')->where('reviewer_id', $userId)->update($data);
+        if ($note !== '') {
+            $this->log('[FieldAi] user_id=' . $userId . ' status=' . $status . ' note=' . $note);
+        }
+    }
+
+    public function log($msg)
+    {
+        $line = date('Y-m-d H:i:s') . ' ' . $msg . PHP_EOL;
+        @file_put_contents($this->logFile, $line, FILE_APPEND);
+    }
+}