presseportale/app/Services/PressRelease/CrawlerHitReport.php

<?php

namespace App\Services\PressRelease;

use App\Models\PressReleaseCrawlerHit;
use App\Support\CrawlerIdentifier;
use Illuminate\Support\Carbon;

/**
 * Admin-Auswertung des Crawler-Abruf-Protokolls (Roadmap 1c.0): welche
 * Such- und KI-Dienste in welchem Umfang Meldungen geholt haben. Zählt
 * Abrufe, nicht Zitate — das Protokoll weiß nicht, was ein Dienst mit der
 * Seite gemacht hat.
 */
class CrawlerHitReport
{
    /**
     * Je Dienst: Abrufe und betroffene Meldungen im Zeitraum, Abrufe seit
     * Beginn des Protokolls, letzter Abruf. Sortiert nach Abrufen im Zeitraum.
     *
     * @return list<array{service: string, label: string, kind: string, fetches: int, releases: int, fetches_total: int, last_at: Carbon}>
     */
    public function byService(int $days = 30): array
    {
        $since = now()->subDays($days);

        $recent = PressReleaseCrawlerHit::query()
            ->where('created_at', '>=', $since)
            ->selectRaw('service, count(*) as fetches, count(distinct press_release_id) as releases')
            ->groupBy('service')
            ->toBase()
            ->get()
            ->keyBy('service');

        return PressReleaseCrawlerHit::query()
            ->selectRaw('service, count(*) as fetches_total, max(created_at) as last_at')
            ->groupBy('service')
            ->toBase()
            ->get()
            ->map(fn (object $row): array => [
                'service' => (string) $row->service,
                'label' => CrawlerIdentifier::label((string) $row->service),
                'kind' => CrawlerIdentifier::kind((string) $row->service),
                'fetches' => (int) ($recent[$row->service]->fetches ?? 0),
                'releases' => (int) ($recent[$row->service]->releases ?? 0),
                'fetches_total' => (int) $row->fetches_total,
                'last_at' => Carbon::parse((string) $row->last_at),
            ])
            ->sortByDesc(fn (array $row): array => [$row['fetches'], $row['fetches_total']])
            ->values()
            ->all();
    }

    /**
     * Abrufe je Tag über die letzten N Tage, lückenlos, getrennt nach Suche
     * und KI (Art aus config/crawlers.php).
     *
     * @return list<array{date: Carbon, search: int, ai: int}>
     */
    public function daily(int $days = 14): array
    {
        $start = now()->startOfDay()->subDays($days - 1);

        $counts = PressReleaseCrawlerHit::query()
            ->where('created_at', '>=', $start)
            ->selectRaw('date(created_at) as day, service, count(*) as fetches')
            ->groupBy('day', 'service')
            ->toBase()
            ->get()
            ->groupBy('day');

        $series = [];

        for ($i = 0; $i < $days; $i++) {
            $day = $start->copy()->addDays($i);
            $row = ['date' => $day, 'search' => 0, 'ai' => 0];

            foreach ($counts->get($day->toDateString(), collect()) as $entry) {
                $row[CrawlerIdentifier::kind((string) $entry->service) === 'search' ? 'search' : 'ai'] += (int) $entry->fetches;
            }

            $series[] = $row;
        }

        return $series;
    }

    /**
     * Am häufigsten abgerufene Meldungen im Zeitraum.
     *
     * @return list<array{press_release_id: int, title: string, fetches: int, services: int}>
     */
    public function topReleases(int $days = 30, int $limit = 15): array
    {
        return PressReleaseCrawlerHit::query()
            ->join('press_releases', 'press_releases.id', '=', 'press_release_crawler_hits.press_release_id')
            ->where('press_release_crawler_hits.created_at', '>=', now()->subDays($days))
            ->selectRaw('press_release_crawler_hits.press_release_id, press_releases.title, count(*) as fetches, count(distinct service) as services')
            ->groupBy('press_release_crawler_hits.press_release_id', 'press_releases.title')
            ->orderByDesc('fetches')
            ->limit($limit)
            ->toBase()
            ->get()
            ->map(fn (object $row): array => [
                'press_release_id' => (int) $row->press_release_id,
                'title' => (string) $row->title,
                'fetches' => (int) $row->fetches,
                'services' => (int) $row->services,
            ])
            ->all();
    }
}