slake-school-service/app/Services/Brief/WeeklyBriefService.php

<?php

namespace App\Services\Brief;

use App\Models\News;
use App\Models\Paper;
use App\Models\Teacher;
use App\Models\WeeklyBrief;
use Carbon\Carbon;
use Illuminate\Support\Collection;
use Illuminate\Support\Str;

class WeeklyBriefService
{
    /** @var list<string> */
    protected const LAB_KEYWORDS = ['实验室', '研究院', '人工智能学院', 'PI', '入职', '加入', '团队', '学院', '研究中心'];

    /** @var list<string> */
    protected const COOP_KEYWORDS = ['校企', '合作', '联合实验室', '联合', '入驻', '签约', '成立'];

    /** @var list<string> */
    protected const TECH_KEYWORDS = ['突破', '发布', '开源', '模型', '大模型', 'Token', '智能体', '算力', 'DeepSeek', 'Kimi', 'Qwen'];

    /** @var list<string> */
    protected const RANKING_KEYWORDS = ['排名', 'CSRankings', '软科', 'AIRankings', '学科实力', '上榜', '登顶'];

    /**
     * @return array{0: Carbon, 1: Carbon}
     */
    public function resolvePreviousWeek(?Carbon $reference = null): array
    {
        $ref = ($reference ?? now())->copy()->timezone('Asia/Shanghai')->startOfDay();
        $lastSunday = $ref->copy()->previous(Carbon::SUNDAY);
        if ($ref->isSunday()) {
            $lastSunday = $ref->copy();
        }
        $weekStart = $lastSunday->copy()->startOfWeek(Carbon::MONDAY);
        $weekEnd = $lastSunday->copy()->endOfDay();

        return [$weekStart, $weekEnd];
    }

    public function generate(Carbon $weekStart, Carbon $weekEnd, ?int $adminUserId = null, bool $replace = true): WeeklyBrief
    {
        $weekStart = $weekStart->copy()->timezone('Asia/Shanghai')->startOfDay();
        $weekEnd = $weekEnd->copy()->timezone('Asia/Shanghai')->endOfDay();

        $papers = $this->fetchPapers($weekStart, $weekEnd);
        $newsItems = $this->fetchNews($weekStart, $weekEnd);
        $teachers = $this->fetchTeachers($weekStart, $weekEnd);

        $groupedNews = $this->groupNews($newsItems);
        $labNews = $groupedNews['lab'];
        $coopNews = $groupedNews['coop'];
        $techNews = $groupedNews['tech'];
        $rankingNews = $groupedNews['ranking'];
        $otherNews = $groupedNews['other'];

        $generatedAt = now()->timezone('Asia/Shanghai');
        $titleDate = $this->formatChineseDate($generatedAt);
        $periodMonth = $weekStart->format('Y年n月');

        $title = "中国高校AI科技成果简报 | {$titleDate}";

        $lines = [];
        $lines[] = "# {$title}";
        $lines[] = '';
        $lines[] = '> 本简报汇总'.$periodMonth.'国内重点高校在人工智能领域的最新研究论文、技术突破、校企合作项目及实验室动态。';
        $lines[] = '';
        $lines[] = '---';
        $lines[] = '';

        $refIndex = 1;
        $references = [];

        $lines[] = '## 一、重要AI论文与研究成果';
        $lines[] = '';
        if ($papers->isEmpty() && $otherNews->isEmpty()) {
            $lines[] = '_本周暂无相关论文入库。_';
            $lines[] = '';
        } else {
            $index = 1;
            foreach ($papers as $paper) {
                $lines = array_merge($lines, $this->renderPaperEntry($index, $paper, $refIndex, $references));
                $index++;
            }
            foreach ($otherNews as $item) {
                $lines = array_merge($lines, $this->renderNewsEntry($index, $item, $refIndex, $references, '核心成果'));
                $index++;
            }
        }

        $lines[] = '---';
        $lines[] = '';
        $lines[] = '## 二、高校AI实验室与研究院动态';
        $lines[] = '';
        $lines = array_merge($lines, $this->renderNewsSection($labNews, $teachers, $refIndex, $references, 1));

        $lines[] = '---';
        $lines[] = '';
        $lines[] = '## 三、校企合作项目与产业落地';
        $lines[] = '';
        $lines = array_merge($lines, $this->renderNewsSection($coopNews, collect(), $refIndex, $references, 1));

        $lines[] = '---';
        $lines[] = '';
        $lines[] = '## 四、国内AI领域突破性技术进展';
        $lines[] = '';
        $lines = array_merge($lines, $this->renderNewsSection($techNews, collect(), $refIndex, $references, 1));

        $lines[] = '---';
        $lines[] = '';
        $lines[] = '## 五、高校排名与学科实力';
        $lines[] = '';
        $lines = array_merge($lines, $this->renderNewsSection($rankingNews, collect(), $refIndex, $references, 1));

        $lines[] = '---';
        $lines[] = '';
        $lines[] = '## 参考来源汇总';
        $lines[] = '';
        foreach ($references as $i => $ref) {
            $num = $i + 1;
            $lines[] = "{$num}. {$ref['label']} {$ref['url']}";
        }
        if ($references === []) {
            $lines[] = '_本周暂无外部来源链接。_';
        }

        $lines[] = '';
        $lines[] = '---';
        $lines[] = '';
        $lines[] = '*简报生成时间：'.$generatedAt->format('Y年n月j日 H:i').' (Asia/Shanghai)*';

        $markdown = implode("\n", $lines);

        $stats = [
            'papers_count' => $papers->count(),
            'news_count' => $newsItems->count(),
            'teachers_count' => $teachers->count(),
            'references_count' => count($references),
            'sections' => [
                'papers' => $papers->count(),
                'lab' => $labNews->count() + $teachers->count(),
                'coop' => $coopNews->count(),
                'tech' => $techNews->count(),
                'ranking' => $rankingNews->count(),
                'other' => $otherNews->count(),
            ],
        ];

        if ($replace) {
            WeeklyBrief::query()
                ->whereDate('week_start', $weekStart->toDateString())
                ->whereDate('week_end', $weekEnd->toDateString())
                ->delete();
        }

        return WeeklyBrief::query()->create([
            'week_start' => $weekStart->toDateString(),
            'week_end' => $weekEnd->toDateString(),
            'title' => $title,
            'markdown' => $markdown,
            'stats_json' => $stats,
            'admin_user_id' => $adminUserId,
            'generated_at' => $generatedAt,
        ]);
    }

    /**
     * @return Collection<int, Paper>
     */
    protected function fetchPapers(Carbon $weekStart, Carbon $weekEnd): Collection
    {
        return Paper::query()
            ->whereNotNull('crawl_job_id')
            ->whereBetween('created_at', [$weekStart, $weekEnd])
            ->orderByDesc('published_at')
            ->orderByDesc('id')
            ->limit(30)
            ->get();
    }

    /**
     * @return Collection<int, News>
     */
    protected function fetchNews(Carbon $weekStart, Carbon $weekEnd): Collection
    {
        return News::query()
            ->with('categoryItem')
            ->whereNotNull('crawl_job_id')
            ->whereBetween('created_at', [$weekStart, $weekEnd])
            ->orderByDesc('published_at')
            ->orderByDesc('id')
            ->limit(50)
            ->get();
    }

    /**
     * @return Collection<int, Teacher>
     */
    protected function fetchTeachers(Carbon $weekStart, Carbon $weekEnd): Collection
    {
        return Teacher::query()
            ->with('university')
            ->whereBetween('created_at', [$weekStart, $weekEnd])
            ->where(function ($q) {
                $q->whereNotNull('bio')
                    ->orWhereNotNull('department')
                    ->orWhereNotNull('title');
            })
            ->orderByDesc('id')
            ->limit(20)
            ->get();
    }

    /**
     * @param  Collection<int, News>  $newsItems
     * @return array{lab: Collection, coop: Collection, tech: Collection, ranking: Collection, other: Collection}
     */
    protected function groupNews(Collection $newsItems): array
    {
        $lab = collect();
        $coop = collect();
        $tech = collect();
        $ranking = collect();
        $other = collect();

        foreach ($newsItems as $item) {
            $text = $item->title.' '.($item->summary ?? '');
            if ($this->matchesKeywords($text, self::RANKING_KEYWORDS)) {
                $ranking->push($item);
            } elseif ($this->matchesKeywords($text, self::COOP_KEYWORDS)) {
                $coop->push($item);
            } elseif ($this->matchesKeywords($text, self::LAB_KEYWORDS)) {
                $lab->push($item);
            } elseif ($this->matchesKeywords($text, self::TECH_KEYWORDS)) {
                $tech->push($item);
            } else {
                $other->push($item);
            }
        }

        return compact('lab', 'coop', 'tech', 'ranking', 'other');
    }

    /**
     * @param  list<string>  $keywords
     */
    protected function matchesKeywords(string $text, array $keywords): bool
    {
        foreach ($keywords as $keyword) {
            if ($keyword !== '' && Str::contains($text, $keyword)) {
                return true;
            }
        }

        return false;
    }

    /**
     * @param  list<array{label: string, url: string}>  $references
     * @return list<string>
     */
    protected function renderPaperEntry(int $index, Paper $paper, int &$refIndex, array &$references): array
    {
        $school = trim((string) ($paper->school_name ?: '国内高校'));
        $headline = "{$school} | {$paper->title}";
        $lines = [];
        $lines[] = "### {$index}. {$headline}";

        if ($paper->published_at) {
            $lines[] = '- **发表时间**：'.$paper->published_at->format('Y年n月j日');
        }

        $summary = $this->cleanText($paper->summary);
        if ($summary !== '') {
            $lines[] = '- **核心成果**：'.$this->truncate($summary, 280);
        }

        if ($paper->authors) {
            $lines[] = '- **研究团队**：'.$this->truncate($this->cleanText($paper->authors), 120);
        }

        $source = $this->appendReference($paper->source_site ?: '论文来源', $paper->url, $refIndex, $references);
        $lines[] = '- **来源**：'.$source;
        $lines[] = '';

        return $lines;
    }

    /**
     * @param  list<array{label: string, url: string}>  $references
     * @return list<string>
     */
    protected function renderNewsEntry(int $index, News $news, int &$refIndex, array &$references, string $summaryLabel = '核心内容'): array
    {
        $headline = $news->title;
        $lines = [];
        $lines[] = "### {$index}. {$headline}";

        if ($news->published_at) {
            $lines[] = '- **时间**：'.$news->published_at->timezone('Asia/Shanghai')->format('Y年n月j日');
        }

        $summary = $this->cleanText($news->summary);
        if ($summary !== '') {
            $lines[] = "- **{$summaryLabel}**：".$this->truncate($summary, 280);
        }

        if ($news->source) {
            $lines[] = '- **来源媒体**：'.$this->cleanText($news->source);
        }

        $label = $news->source_site ?: ($news->source ?: '资讯来源');
        $source = $this->appendReference($label, $news->source_url ?: $news->cover_url, $refIndex, $references);
        $lines[] = '- **来源**：'.$source;
        $lines[] = '';

        return $lines;
    }

    /**
     * @param  Collection<int, News>  $newsItems
     * @param  Collection<int, Teacher>  $teachers
     * @param  list<array{label: string, url: string}>  $references
     * @return list<string>
     */
    protected function renderNewsSection(Collection $newsItems, Collection $teachers, int &$refIndex, array &$references, int $startIndex): array
    {
        $lines = [];
        $index = $startIndex;

        if ($newsItems->isEmpty() && $teachers->isEmpty()) {
            $lines[] = '_本周暂无相关动态。_';
            $lines[] = '';

            return $lines;
        }

        foreach ($newsItems as $item) {
            $lines = array_merge($lines, $this->renderNewsEntry($index, $item, $refIndex, $references));
            $index++;
        }

        foreach ($teachers as $teacher) {
            $uni = $teacher->university?->name ?: $teacher->university_text ?: '国内高校';
            $lines[] = "### {$index}. {$uni} | {$teacher->name}";
            if ($teacher->title) {
                $lines[] = '- **职称**：'.$this->cleanText($teacher->title);
            }
            if ($teacher->department) {
                $lines[] = '- **院系**：'.$this->cleanText($teacher->department);
            }
            if ($teacher->bio) {
                $lines[] = '- **简介**：'.$this->truncate($this->cleanText($teacher->bio), 220);
            }
            $lines[] = '';
            $index++;
        }

        return $lines;
    }

    /**
     * @param  list<array{label: string, url: string}>  $references
     */
    protected function appendReference(string $label, ?string $url, int &$refIndex, array &$references): string
    {
        $label = $this->cleanText($label) ?: '来源';
        $url = trim((string) $url);
        if ($url === '' || ! preg_match('#^https?://#i', $url)) {
            return $label;
        }

        foreach ($references as $ref) {
            if ($ref['url'] === $url) {
                return "[{$label}]({$url})";
            }
        }

        $references[] = ['label' => $label.' - '.$this->briefTitleFromUrl($url), 'url' => $url];
        $refIndex++;

        return "[{$label}]({$url})";
    }

    protected function briefTitleFromUrl(string $url): string
    {
        $host = parse_url($url, PHP_URL_HOST) ?: $url;

        return Str::limit($host, 48, '');
    }

    protected function formatChineseDate(Carbon $date): string
    {
        return $date->format('Y年n月j日');
    }

    protected function cleanText(?string $text): string
    {
        if ($text === null || trim($text) === '') {
            return '';
        }

        return trim(preg_replace('/\s+/u', ' ', strip_tags($text)) ?? '');
    }

    protected function truncate(string $text, int $limit): string
    {
        if (mb_strlen($text) <= $limit) {
            return $text;
        }

        return mb_substr($text, 0, $limit - 1).'…';
    }
}