sbnews/server/app/common/service/ai/AiAssistantWebNewsRetriever.php

<?php

namespace app\common\service\ai;

use app\common\model\ai\AiConfig;

class AiAssistantWebNewsRetriever
{
    public static function retrieve(string $message, array $plan, array $historyMessages = []): array
    {
        if (!self::shouldRetrieve($message, $plan)) {
            return [];
        }

        $client = new DeepSeekClient();
        $result = $client->chatJson(self::systemPrompt(), self::userPrompt($message, $plan, $historyMessages), self::modelOptions());
        if (empty($result['success']) || !is_array($result['parsed'] ?? null)) {
            return [];
        }

        return self::normalizeItems($result['parsed'], $message);
    }

    private static function shouldRetrieve(string $message, array $plan): bool
    {
        if ((int) AiConfig::getVal('assistant_enable_web_news_search', '1') !== 1) {
            return false;
        }

        $intent = (string) ($plan['intent'] ?? '');
        $preferredDomain = (string) ($plan['preferred_domain'] ?? '');
        $domains = (array) ($plan['domains'] ?? []);

        if (in_array($intent, ['crypto_quote', 'match_schedule', 'match_result', 'lottery_analysis'], true)) {
            return false;
        }

        if ($preferredDomain === 'article' || in_array('article', $domains, true) || in_array('post', $domains, true)) {
            return true;
        }

        return (bool) preg_match('/资讯|新闻|消息|报道|热点|刚刚|最新|头条/u', $message);
    }

    private static function systemPrompt(): string
    {
        $today = date('Y-m-d');
        return '你是世博头条 AI 助手的联网资讯检索器。'
            . '请调用 web search 工具搜索与用户问题最相关、尽量最新的体育资讯或新闻。'
            . '今天日期是 ' . $today . '。'
            . '只输出 JSON：{"items":[{"title":"","source":"","url":"","summary":"","published_at":""}]}。'
            . '要求：最多返回 5 条；title 和 url 必填；summary 只写 1-2 句；没有可靠结果时返回 {"items":[]}。';
    }

    private static function userPrompt(string $message, array $plan, array $historyMessages): string
    {
        $history = [];
        foreach (array_slice($historyMessages, -4) as $row) {
            $content = trim((string) ($row['content'] ?? ''));
            if ($content === '') {
                continue;
            }
            $history[] = [
                'role' => (string) ($row['role'] ?? 'user'),
                'content' => mb_substr($content, 0, 200),
            ];
        }

        return "用户问题：\n"
            . $message
            . "\n\n查询规划：\n"
            . json_encode([
                'intent' => $plan['intent'] ?? '',
                'rewritten_query' => $plan['rewritten_query'] ?? '',
                'alternate_queries' => $plan['alternate_queries'] ?? [],
                'entities' => $plan['entities'] ?? [],
                'date_range' => $plan['date_range'] ?? [],
            ], JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT)
            . "\n\n最近对话：\n"
            . json_encode($history, JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT)
            . "\n\n请检索最新外网资讯并输出 JSON。";
    }

    private static function modelOptions(): array
    {
        return DeepSeekClient::defaultModelOptions([
            'max_tokens' => 900,
            'temperature' => 0.2,
            'retry_times' => 1,
            'retry_sleep_ms' => 600,
            'tools' => [
                ['type' => 'web_search'],
            ],
            'tool_choice' => 'required',
            'include' => ['web_search_call.action.sources'],
        ]);
    }

    private static function normalizeItems(array $parsed, string $message): array
    {
        $items = [];
        foreach ((array) ($parsed['items'] ?? []) as $row) {
            $title = trim((string) ($row['title'] ?? ''));
            $url = trim((string) ($row['url'] ?? ''));
            if ($title === '' || $url === '' || !preg_match('#^https?://#i', $url)) {
                continue;
            }

            $sourceName = trim((string) ($row['source'] ?? ''));
            $publishedAt = trim((string) ($row['published_at'] ?? ''));
            $summary = trim((string) ($row['summary'] ?? ''));
            $contextSummary = trim(implode(' ', array_filter([
                $sourceName !== '' ? '来源：' . $sourceName : '',
                $publishedAt !== '' ? '时间：' . $publishedAt : '',
                $summary,
            ])));

            $items[] = [
                'context' => [
                    'domain' => 'web_news',
                    'title' => $title,
                    'summary' => $contextSummary !== '' ? $contextSummary : $message,
                    'url' => $url,
                    'source' => $sourceName,
                    'published_at' => $publishedAt,
                ],
                'source' => [
                    'type' => 'web_news',
                    'title' => $title,
                    'summary' => self::clipText($contextSummary !== '' ? $contextSummary : $summary, 140),
                    'path' => '/pages/webview/webview?url=' . urlencode($url),
                    'source_id' => 0,
                ],
            ];
        }

        return array_slice($items, 0, 5);
    }

    private static function clipText(string $text, int $length): string
    {
        $text = trim(preg_replace('/\s+/u', ' ', $text) ?: '');
        if (mb_strlen($text) <= $length) {
            return $text;
        }
        return mb_substr($text, 0, $length) . '...';
    }
}