no message

server/app/common/service/article/ArticleAiCommentService.php

@@ -0,0 +1,285 @@
+<?php
+
+namespace app\common\service\article;
+
+use app\common\model\article\Article;
+use app\common\model\article\ArticleAiCommentTask;
+use app\common\model\article\ArticleComment;
+use app\common\model\user\User;
+use app\common\service\ai\AiService;
+use think\facade\Config;
+use think\facade\Db;
+
+class ArticleAiCommentService
+{
+    private const VIRTUAL_ACCOUNT_PREFIX = 'ai_comment_';
+    private const VIRTUAL_NICKNAME_PREFIX = 'AI评论员';
+    private const VIRTUAL_POOL_SIZE = 20;
+
+    public static function tick(int $seedLimit = 10, int $processLimit = 20): array
+    {
+        $virtualUsers = self::ensureVirtualUsers(self::VIRTUAL_POOL_SIZE);
+        $seeded = self::seedTasks($virtualUsers, $seedLimit);
+        $processed = self::processDueTasks($processLimit);
+
+        return [
+            'virtual_user_count' => count($virtualUsers),
+            'seeded' => $seeded,
+            'processed' => $processed,
+        ];
+    }
+
+    protected static function seedTasks(array $virtualUsers, int $limit = 10): array
+    {
+        if (empty($virtualUsers)) {
+            return ['created' => 0, 'skipped' => 0];
+        }
+
+        $articles = Article::field('id,title,desc,content,author,ext,click_virtual,click_actual,comment_count,create_time,published_at')
+            ->where('is_show', 1)
+            ->whereNull('delete_time')
+            ->where('create_time', '>=', strtotime('-14 days'))
+            ->orderRaw('comment_count asc, (click_actual + click_virtual) desc, id desc')
+            ->limit($limit)
+            ->select()
+            ->toArray();
+
+        $created = 0;
+        $skipped = 0;
+        foreach ($articles as $index => $article) {
+            $articleId = (int) ($article['id'] ?? 0);
+            if ($articleId <= 0) {
+                $skipped++;
+                continue;
+            }
+
+            $existingPending = ArticleAiCommentTask::where('article_id', $articleId)
+                ->whereIn('status', [0, 1])
+                ->count();
+            if ($existingPending > 0) {
+                $skipped++;
+                continue;
+            }
+
+            $desiredCount = self::resolveTaskCount($article);
+            if ($desiredCount <= 0) {
+                $skipped++;
+                continue;
+            }
+
+            for ($i = 0; $i < $desiredCount; $i++) {
+                $virtualUserId = $virtualUsers[($articleId + $index + $i) % count($virtualUsers)];
+                $comment = self::generateCommentContent($article, $virtualUserId);
+                if ($comment === '') {
+                    continue;
+                }
+
+                $runAt = time() + random_int(600, 43200);
+                ArticleAiCommentTask::create([
+                    'article_id' => $articleId,
+                    'virtual_user_id' => $virtualUserId,
+                    'comment_content' => $comment,
+                    'run_at' => $runAt,
+                    'status' => 0,
+                    'create_time' => time(),
+                    'update_time' => time(),
+                ]);
+                $created++;
+            }
+        }
+
+        return compact('created', 'skipped');
+    }
+
+    protected static function processDueTasks(int $limit = 20): array
+    {
+        $tasks = ArticleAiCommentTask::where('status', 0)
+            ->where('run_at', '<=', time())
+            ->order('run_at', 'asc')
+            ->limit($limit)
+            ->select()
+            ->toArray();
+
+        $done = 0;
+        $failed = 0;
+
+        foreach ($tasks as $task) {
+            $articleId = (int) ($task['article_id'] ?? 0);
+            $content = trim((string) ($task['comment_content'] ?? ''));
+            $virtualUserId = (int) ($task['virtual_user_id'] ?? 0);
+
+            if ($articleId <= 0 || $content === '' || $virtualUserId <= 0) {
+                self::markTaskFailed((int) $task['id'], '任务数据不完整');
+                $failed++;
+                continue;
+            }
+
+            $article = Article::where('id', $articleId)
+                ->where('is_show', 1)
+                ->whereNull('delete_time')
+                ->findOrEmpty();
+
+            if ($article->isEmpty()) {
+                self::markTaskFailed((int) $task['id'], '文章不存在或已下架');
+                $failed++;
+                continue;
+            }
+
+            Db::startTrans();
+            try {
+                ArticleComment::create([
+                    'article_id' => $articleId,
+                    'user_id' => $virtualUserId,
+                    'parent_id' => 0,
+                    'reply_user_id' => 0,
+                    'content' => $content,
+                    'is_show' => 1,
+                    'create_time' => time(),
+                    'update_time' => time(),
+                ]);
+
+                Article::where('id', $articleId)->inc('comment_count')->update();
+                ArticleAiCommentTask::where('id', $task['id'])->update([
+                    'status' => 1,
+                    'update_time' => time(),
+                ]);
+
+                Db::commit();
+                $done++;
+            } catch (\Throwable $e) {
+                Db::rollback();
+                self::markTaskFailed((int) $task['id'], $e->getMessage());
+                $failed++;
+            }
+        }
+
+        return compact('done', 'failed');
+    }
+
+    protected static function ensureVirtualUsers(int $targetCount = self::VIRTUAL_POOL_SIZE): array
+    {
+        $existing = User::where('account', 'like', self::VIRTUAL_ACCOUNT_PREFIX . '%')
+            ->order('id', 'asc')
+            ->column('id');
+
+        $existing = array_values(array_map('intval', $existing));
+        $missing = max(0, $targetCount - count($existing));
+        if ($missing > 0) {
+            $avatar = (string) Config::get('project.default_image.user_avatar', '');
+            $passwordSalt = (string) Config::get('project.unique_identification', 'sport-era');
+
+            for ($i = 1; $i <= $missing; $i++) {
+                $accountIndex = count($existing) + $i;
+                $sn = User::createUserSn('9');
+                $password = create_password(bin2hex(random_bytes(6)), $passwordSalt);
+
+                $user = User::create([
+                    'sn' => $sn,
+                    'avatar' => $avatar,
+                    'real_name' => '',
+                    'nickname' => self::VIRTUAL_NICKNAME_PREFIX . str_pad((string) $accountIndex, 2, '0', STR_PAD_LEFT),
+                    'account' => self::VIRTUAL_ACCOUNT_PREFIX . str_pad((string) $accountIndex, 4, '0', STR_PAD_LEFT),
+                    'password' => $password,
+                    'mobile' => '',
+                    'sex' => 0,
+                    'channel' => 0,
+                    'is_disable' => 0,
+                    'is_new_user' => 0,
+                    'user_money' => 0,
+                    'total_recharge_amount' => 0,
+                ]);
+
+                $existing[] = (int) $user->id;
+            }
+        }
+
+        return $existing;
+    }
+
+    protected static function resolveTaskCount(array $article): int
+    {
+        $commentCount = (int) ($article['comment_count'] ?? 0);
+        $clickCount = (int) ($article['click_actual'] ?? 0) + (int) ($article['click_virtual'] ?? 0);
+
+        if ($commentCount >= 20) {
+            return 0;
+        }
+
+        if ($clickCount >= 500) {
+            return 2;
+        }
+
+        return 1;
+    }
+
+    protected static function generateCommentContent(array $article, int $virtualUserId): string
+    {
+        $title = trim((string) ($article['title'] ?? ''));
+        $abstract = trim((string) ($article['desc'] ?? ''));
+        $content = self::normalizeContent((string) ($article['content'] ?? ''));
+        if ($title === '' && $abstract === '' && $content === '') {
+            return '';
+        }
+
+        $personaPool = [
+            '资深球迷',
+            '战术观察员',
+            '数据控',
+            '赛事评论员',
+            '球迷视角',
+        ];
+        $persona = $personaPool[$virtualUserId % count($personaPool)];
+
+        $result = AiService::generateArticleComment($title, $abstract, $content, $persona);
+        if (empty($result['success'])) {
+            return self::fallbackComment($title, $abstract, $content);
+        }
+
+        $comment = trim((string) ($result['content'] ?? ''));
+        $comment = preg_replace('/^评论[:：]\s*/u', '', $comment);
+        $comment = preg_replace('/^[“"‘’\'\']|[”"‘’\'\']$/u', '', $comment);
+        $comment = preg_replace('/\s+/u', ' ', $comment);
+        $comment = trim($comment, " \t\n\r\0\x0B，。,.!！?？;；:：");
+
+        if ($comment === '' || mb_strlen($comment) < 8) {
+            return self::fallbackComment($title, $abstract, $content);
+        }
+
+        return mb_substr($comment, 0, 120);
+    }
+
+    protected static function fallbackComment(string $title, string $abstract, string $content): string
+    {
+        $focus = $title ?: ($abstract ?: mb_substr($content, 0, 18));
+        if ($focus === '') {
+            return '';
+        }
+
+        return mb_substr($focus, 0, 18) . '这条信息量挺足，后续关键数据变化值得继续关注。';
+    }
+
+    protected static function markTaskFailed(int $taskId, string $message): void
+    {
+        if ($taskId <= 0) {
+            return;
+        }
+
+        ArticleAiCommentTask::where('id', $taskId)->update([
+            'status' => 2,
+            'error_msg' => mb_substr($message, 0, 255),
+            'update_time' => time(),
+        ]);
+    }
+
+    protected static function normalizeContent(string $content): string
+    {
+        if ($content === '') {
+            return '';
+        }
+
+        $content = strip_tags($content);
+        $content = html_entity_decode($content, ENT_QUOTES | ENT_HTML5, 'UTF-8');
+        $content = preg_replace('/\s+/u', ' ', $content);
+        return trim($content);
+    }
+}

server/app/common/service/article/ArticleImageProxyService.php

@@ -0,0 +1,241 @@
+<?php
+
+namespace app\common\service\article;
+
+use think\Response;
+
+class ArticleImageProxyService
+{
+    private const SOURCE_HOSTS = [
+        'hkjc_racingnews' => [
+            'res.hkjc.com',
+            'racingnews.hkjc.com',
+        ],
+        'taiwan_lottery' => [
+            'cdn.taiwanlottery.com.tw',
+            'www.taiwanlottery.com',
+            'taiwanlottery.com',
+        ],
+    ];
+
+    private const SOURCE_REFERERS = [
+        'hkjc_racingnews' => 'https://racingnews.hkjc.com/',
+        'taiwan_lottery' => 'https://www.taiwanlottery.com/',
+    ];
+
+    public static function rewriteArticlePayload(array $article): array
+    {
+        $ext = self::decodeExt($article['ext'] ?? []);
+        if (!self::shouldRewrite($ext)) {
+            return $article;
+        }
+
+        if (!empty($article['image'])) {
+            $article['image'] = self::buildProxyUrl((string) $article['image'], $ext);
+        }
+
+        if (!empty($article['content'])) {
+            $article['content'] = preg_replace_callback(
+                '/(<img\b[^>]*\bsrc=["\'])([^"\']+)(["\'][^>]*>)/i',
+                static function (array $matches) use ($ext) {
+                    $proxyUrl = self::buildProxyUrl($matches[2] ?? '', $ext);
+                    if ($proxyUrl === '' || $proxyUrl === ($matches[2] ?? '')) {
+                        return $matches[0];
+                    }
+                    return $matches[1]
+                        . htmlspecialchars($proxyUrl, ENT_QUOTES | ENT_SUBSTITUTE, 'UTF-8')
+                        . $matches[3];
+                },
+                (string) $article['content']
+            );
+        }
+
+        return $article;
+    }
+
+    public static function proxyRemoteImage(string $url): Response
+    {
+        $normalizedUrl = self::normalizeUrl($url);
+        if ($normalizedUrl === '' || !self::isAllowedRemoteUrl($normalizedUrl)) {
+            return response('image not found', 404)->header([
+                'Content-Type' => 'text/plain; charset=utf-8',
+                'Cache-Control' => 'public, max-age=300',
+            ]);
+        }
+
+        $source = self::detectSourceByUrl($normalizedUrl);
+        $headers = [
+            'Accept: image/avif,image/webp,image/apng,image/svg+xml,image/*,*/*;q=0.8',
+            'User-Agent: Mozilla/5.0 (compatible; SportEra/1.0)',
+        ];
+        $referer = self::SOURCE_REFERERS[$source] ?? '';
+        if ($referer !== '') {
+            $headers[] = 'Referer: ' . $referer;
+        }
+
+        $ch = curl_init($normalizedUrl);
+        curl_setopt_array($ch, [
+            CURLOPT_RETURNTRANSFER => true,
+            CURLOPT_FOLLOWLOCATION => true,
+            CURLOPT_MAXREDIRS => 3,
+            CURLOPT_CONNECTTIMEOUT => 10,
+            CURLOPT_TIMEOUT => 20,
+            CURLOPT_ENCODING => '',
+            CURLOPT_HTTPHEADER => $headers,
+        ]);
+
+        $body = curl_exec($ch);
+        $httpCode = (int) curl_getinfo($ch, CURLINFO_RESPONSE_CODE);
+        $contentType = (string) curl_getinfo($ch, CURLINFO_CONTENT_TYPE);
+        $effectiveUrl = (string) curl_getinfo($ch, CURLINFO_EFFECTIVE_URL);
+        $curlError = curl_error($ch);
+        curl_close($ch);
+
+        if ($body === false || $curlError !== '' || $httpCode < 200 || $httpCode >= 300) {
+            return response('image not found', 404)->header([
+                'Content-Type' => 'text/plain; charset=utf-8',
+                'Cache-Control' => 'public, max-age=300',
+            ]);
+        }
+
+        $effectiveUrl = $effectiveUrl !== '' ? $effectiveUrl : $normalizedUrl;
+        if (!self::isAllowedRemoteUrl($effectiveUrl)) {
+            return response('image not found', 404)->header([
+                'Content-Type' => 'text/plain; charset=utf-8',
+                'Cache-Control' => 'public, max-age=300',
+            ]);
+        }
+
+        $contentType = self::normalizeContentType($contentType, $effectiveUrl);
+        if (!str_starts_with($contentType, 'image/')) {
+            return response('image not found', 404)->header([
+                'Content-Type' => 'text/plain; charset=utf-8',
+                'Cache-Control' => 'public, max-age=300',
+            ]);
+        }
+
+        return response($body, 200)->header([
+            'Content-Type' => $contentType,
+            'Cache-Control' => 'public, max-age=86400',
+            'Access-Control-Allow-Origin' => '*',
+        ]);
+    }
+
+    private static function shouldRewrite(array $ext): bool
+    {
+        $source = (string) ($ext['source'] ?? '');
+        return isset(self::SOURCE_HOSTS[$source]);
+    }
+
+    private static function buildProxyUrl(string $url, array $ext = []): string
+    {
+        $normalizedUrl = self::normalizeUrl($url);
+        if ($normalizedUrl === '' || self::isLocalProxyUrl($normalizedUrl)) {
+            return $url;
+        }
+
+        if (!self::isAllowedRemoteUrl($normalizedUrl, $ext)) {
+            return $url;
+        }
+
+        $path = '/api/article/imageProxy?url=' . rawurlencode($normalizedUrl);
+        $domain = rtrim((string) request()->domain(), '/');
+        return $domain !== '' ? $domain . $path : $path;
+    }
+
+    private static function normalizeUrl(string $url): string
+    {
+        $url = trim(html_entity_decode($url, ENT_QUOTES | ENT_HTML5, 'UTF-8'));
+        if ($url === '') {
+            return '';
+        }
+        if (str_starts_with($url, '//')) {
+            return 'https:' . $url;
+        }
+        return $url;
+    }
+
+    private static function isAllowedRemoteUrl(string $url, array $ext = []): bool
+    {
+        $parts = parse_url($url);
+        $scheme = strtolower((string) ($parts['scheme'] ?? ''));
+        $host = strtolower((string) ($parts['host'] ?? ''));
+        if ($host === '' || !in_array($scheme, ['http', 'https'], true)) {
+            return false;
+        }
+
+        $allowedHosts = self::getAllowedHosts($ext);
+        foreach ($allowedHosts as $allowedHost) {
+            $allowedHost = strtolower($allowedHost);
+            if ($host === $allowedHost || str_ends_with($host, '.' . $allowedHost)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    private static function getAllowedHosts(array $ext = []): array
+    {
+        $source = (string) ($ext['source'] ?? '');
+        if ($source !== '' && isset(self::SOURCE_HOSTS[$source])) {
+            return self::SOURCE_HOSTS[$source];
+        }
+
+        $allHosts = [];
+        foreach (self::SOURCE_HOSTS as $hosts) {
+            $allHosts = array_merge($allHosts, $hosts);
+        }
+        return array_values(array_unique($allHosts));
+    }
+
+    private static function detectSourceByUrl(string $url): string
+    {
+        $parts = parse_url($url);
+        $host = strtolower((string) ($parts['host'] ?? ''));
+        foreach (self::SOURCE_HOSTS as $source => $hosts) {
+            foreach ($hosts as $allowedHost) {
+                $allowedHost = strtolower($allowedHost);
+                if ($host === $allowedHost || str_ends_with($host, '.' . $allowedHost)) {
+                    return $source;
+                }
+            }
+        }
+        return '';
+    }
+
+    private static function isLocalProxyUrl(string $url): bool
+    {
+        return str_contains($url, '/api/article/imageProxy?');
+    }
+
+    private static function normalizeContentType(string $contentType, string $url): string
+    {
+        $contentType = strtolower(trim(explode(';', $contentType)[0] ?? ''));
+        if (str_starts_with($contentType, 'image/')) {
+            return $contentType;
+        }
+
+        $path = strtolower((string) parse_url($url, PHP_URL_PATH));
+        return match (pathinfo($path, PATHINFO_EXTENSION)) {
+            'jpg', 'jpeg' => 'image/jpeg',
+            'png' => 'image/png',
+            'gif' => 'image/gif',
+            'webp' => 'image/webp',
+            'svg' => 'image/svg+xml',
+            default => 'application/octet-stream',
+        };
+    }
+
+    private static function decodeExt($ext): array
+    {
+        if (is_array($ext)) {
+            return $ext;
+        }
+        if (is_string($ext) && $ext !== '') {
+            $decoded = json_decode($ext, true);
+            return is_array($decoded) ? $decoded : [];
+        }
+        return [];
+    }
+}