analytics/app/Services/AnalyticsQueryService.php

<?php

namespace App\Services;

use App\Models\AnalyticsSession;
use App\Models\DailyAggregate;
use App\Models\PageDailyAggregate;
use App\Models\PageView;
use App\Models\Visitor;
use App\Models\Website;
use Illuminate\Support\Carbon;
use Illuminate\Support\Collection;
use Illuminate\Support\Facades\DB;

class AnalyticsQueryService
{
    private readonly Carbon $todayStart;

    private ?Collection $segmentVisitorIds = null;

    public function __construct(
        private readonly Website $website,
        private readonly Carbon $from,
        private readonly Carbon $to,
    ) {
        $this->todayStart = Carbon::today();
    }

    public static function for(Website $website, Carbon $from, Carbon $to): self
    {
        return new self($website, $from->startOfDay(), $to->endOfDay());
    }

    public function withSegment(?Collection $visitorIds): self
    {
        $this->segmentVisitorIds = $visitorIds;

        return $this;
    }

    public function isSegmented(): bool
    {
        return $this->segmentVisitorIds !== null;
    }

    private function applySegmentFilter($query, string $column = 'visitor_id')
    {
        if ($this->segmentVisitorIds !== null) {
            $query->whereIn($column, $this->segmentVisitorIds);
        }

        return $query;
    }

    /**
     * Split the requested date range into two parts:
     * - "aggregate range": completed days that can be served from aggregate tables
     * - "live range": today's partial data that must come from raw tables
     *
     * Returns [?Carbon $aggFrom, ?Carbon $aggTo, ?Carbon $liveFrom, ?Carbon $liveTo]
     */
    private function splitRange(): array
    {
        $aggFrom = null;
        $aggTo = null;
        $liveFrom = null;
        $liveTo = null;

        $yesterdayEnd = $this->todayStart->copy()->subSecond();

        if ($this->from->lt($this->todayStart)) {
            $aggFrom = $this->from->copy();
            $aggTo = $this->to->lt($this->todayStart) ? $this->to->copy() : $yesterdayEnd;
        }

        if ($this->to->gte($this->todayStart)) {
            $liveFrom = $this->from->gte($this->todayStart) ? $this->from->copy() : $this->todayStart->copy();
            $liveTo = $this->to->copy();
        }

        return [$aggFrom, $aggTo, $liveFrom, $liveTo];
    }

    private function hasAggregateRange(): bool
    {
        [$aggFrom] = $this->splitRange();

        return $aggFrom !== null;
    }

    private function hasLiveRange(): bool
    {
        [,, $liveFrom] = $this->splitRange();

        return $liveFrom !== null;
    }

    // ---- Summary Metrics ----

    public function totalPageViews(): int
    {
        if ($this->isSegmented()) {
            return $this->applySegmentFilter(
                PageView::where('website_id', $this->website->id)
                    ->whereBetween('viewed_at', [$this->from, $this->to])
            )->count();
        }

        $total = 0;
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();

        if ($aggFrom) {
            $total += (int) DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->sum('page_views');
        }

        if ($liveFrom) {
            $total += PageView::where('website_id', $this->website->id)
                ->whereBetween('viewed_at', [$liveFrom, $liveTo])
                ->count();
        }

        return $total;
    }

    public function uniqueVisitors(): int
    {
        if ($this->isSegmented()) {
            return $this->applySegmentFilter(
                PageView::where('website_id', $this->website->id)
                    ->whereBetween('viewed_at', [$this->from, $this->to])
            )->distinct('visitor_id')->count('visitor_id');
        }

        $total = 0;
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();

        if ($aggFrom) {
            $total += (int) DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->sum('visitors');
        }

        if ($liveFrom) {
            $total += PageView::where('website_id', $this->website->id)
                ->whereBetween('viewed_at', [$liveFrom, $liveTo])
                ->distinct('visitor_id')
                ->count('visitor_id');
        }

        return $total;
    }

    public function totalSessions(): int
    {
        if ($this->isSegmented()) {
            return $this->applySegmentFilter(
                AnalyticsSession::where('website_id', $this->website->id)
                    ->whereBetween('started_at', [$this->from, $this->to])
            )->count();
        }

        $total = 0;
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();

        if ($aggFrom) {
            $total += (int) DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->sum('sessions');
        }

        if ($liveFrom) {
            $total += AnalyticsSession::where('website_id', $this->website->id)
                ->whereBetween('started_at', [$liveFrom, $liveTo])
                ->count();
        }

        return $total;
    }

    public function bounceRate(): float
    {
        $totalSessions = $this->totalSessions();
        if ($totalSessions === 0) {
            return 0;
        }

        $bounced = 0;
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();

        if ($aggFrom) {
            $bounced += (int) DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->sum('bounce_count');
        }

        if ($liveFrom) {
            $bounced += AnalyticsSession::where('website_id', $this->website->id)
                ->whereBetween('started_at', [$liveFrom, $liveTo])
                ->where('is_bounce', true)
                ->count();
        }

        return round(($bounced / $totalSessions) * 100, 1);
    }

    public function avgSessionDuration(): float
    {
        $totalSessions = $this->totalSessions();
        if ($totalSessions === 0) {
            return 0;
        }

        $totalDuration = 0;
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();

        if ($aggFrom) {
            $totalDuration += (int) DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->sum('total_duration');
        }

        if ($liveFrom) {
            $totalDuration += (int) AnalyticsSession::where('website_id', $this->website->id)
                ->whereBetween('started_at', [$liveFrom, $liveTo])
                ->sum('duration');
        }

        return round($totalDuration / $totalSessions);
    }

    public function avgPagesPerSession(): float
    {
        $totalSessions = $this->totalSessions();
        if ($totalSessions === 0) {
            return 0;
        }

        $totalPageViews = $this->totalPageViews();

        return round($totalPageViews / $totalSessions, 1);
    }

    public function summary(): array
    {
        return [
            'page_views' => $this->totalPageViews(),
            'visitors' => $this->uniqueVisitors(),
            'sessions' => $this->totalSessions(),
            'bounce_rate' => $this->bounceRate(),
            'avg_duration' => $this->avgSessionDuration(),
            'avg_pages' => $this->avgPagesPerSession(),
        ];
    }

    // ---- Time Series ----

    public function visitorsOverTime(): Collection
    {
        $days = $this->from->diffInDays($this->to);
        $isHourly = $days <= 2;
        $groupFormat = $isHourly ? '%Y-%m-%d %H:00' : '%Y-%m-%d';
        $carbonFormat = $isHourly ? 'Y-m-d H:00' : 'Y-m-d';

        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();
        $aggregated = collect();
        $liveData = collect();

        if ($aggFrom && ! $isHourly) {
            $aggregated = DailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->select('date as period', 'page_views', 'visitors', 'sessions')
                ->get()
                ->mapWithKeys(fn ($row) => [
                    Carbon::parse($row->period)->format('Y-m-d') => [
                        'page_views' => $row->page_views,
                        'visitors' => $row->visitors,
                        'sessions' => $row->sessions,
                    ],
                ]);
        }

        $liveQueryFrom = ($aggFrom && ! $isHourly) ? ($liveFrom ?? $this->todayStart) : $this->from;
        $liveQueryTo = $liveTo ?? $this->to;
        $needsLive = $isHourly || $liveFrom;

        if ($needsLive || ! $aggFrom) {
            $rawFrom = $isHourly ? $this->from : $liveQueryFrom;
            $rawTo = $isHourly ? $this->to : $liveQueryTo;

            $pvRaw = PageView::where('website_id', $this->website->id)
                ->whereBetween('viewed_at', [$rawFrom, $rawTo])
                ->select(
                    DB::raw($this->dateFormat($groupFormat, 'viewed_at').' as period'),
                    DB::raw('COUNT(*) as page_views'),
                    DB::raw('COUNT(DISTINCT visitor_id) as visitors'),
                )
                ->groupBy('period')
                ->get()
                ->keyBy('period');

            $sessRaw = AnalyticsSession::where('website_id', $this->website->id)
                ->whereBetween('started_at', [$rawFrom, $rawTo])
                ->select(
                    DB::raw($this->dateFormat($groupFormat, 'started_at').' as period'),
                    DB::raw('COUNT(*) as sessions'),
                )
                ->groupBy('period')
                ->get()
                ->keyBy('period');

            foreach ($pvRaw as $key => $row) {
                $liveData[$key] = [
                    'page_views' => $row->page_views,
                    'visitors' => $row->visitors,
                    'sessions' => $sessRaw->get($key)?->sessions ?? 0,
                ];
            }

            foreach ($sessRaw as $key => $row) {
                if (! isset($liveData[$key])) {
                    $liveData[$key] = [
                        'page_views' => 0,
                        'visitors' => 0,
                        'sessions' => $row->sessions,
                    ];
                }
            }
        }

        $merged = $aggregated->union($liveData);

        $periods = collect();
        $cursor = $this->from->copy();
        $increment = $isHourly ? 'addHour' : 'addDay';

        while ($cursor->lte($this->to)) {
            $key = $cursor->format($carbonFormat);
            $data = $merged->get($key);

            $periods->push([
                'period' => $key,
                'page_views' => $data['page_views'] ?? 0,
                'visitors' => $data['visitors'] ?? 0,
                'sessions' => $data['sessions'] ?? 0,
            ]);

            $cursor->$increment();
        }

        return $periods;
    }

    // ---- Audience: Geographic ----

    public function visitorsByCountry(int $limit = 50): Collection
    {
        return Visitor::where('website_id', $this->website->id)
            ->whereNotNull('country')
            ->whereHas('pageViews', fn ($q) => $q->whereBetween('viewed_at', [$this->from, $this->to]))
            ->select('country', DB::raw('COUNT(*) as visitors'))
            ->groupBy('country')
            ->orderByDesc('visitors')
            ->limit($limit)
            ->get();
    }

    public function visitorsByCity(int $limit = 50): Collection
    {
        return Visitor::where('website_id', $this->website->id)
            ->whereNotNull('city')
            ->whereHas('pageViews', fn ($q) => $q->whereBetween('viewed_at', [$this->from, $this->to]))
            ->select('country', 'region', 'city', DB::raw('COUNT(*) as visitors'))
            ->groupBy('country', 'region', 'city')
            ->orderByDesc('visitors')
            ->limit($limit)
            ->get();
    }

    // ---- Audience: Technology ----

    public function visitorsByBrowser(int $limit = 20): Collection
    {
        return $this->visitorGroupedBy('browser', $limit);
    }

    public function visitorsByOS(int $limit = 20): Collection
    {
        return $this->visitorGroupedBy('os', $limit);
    }

    public function visitorsByDeviceType(int $limit = 10): Collection
    {
        return $this->visitorGroupedBy('device_type', $limit);
    }

    public function visitorsByScreenResolution(int $limit = 20): Collection
    {
        return $this->visitorGroupedBy('screen_resolution', $limit);
    }

    // ---- Audience: Language ----

    public function visitorsByLanguage(int $limit = 20): Collection
    {
        return $this->visitorGroupedBy('language', $limit);
    }

    // ---- Audience: New vs Returning ----

    public function newVsReturning(): array
    {
        $activeVisitorIds = PageView::where('website_id', $this->website->id)
            ->whereBetween('viewed_at', [$this->from, $this->to])
            ->distinct()
            ->pluck('visitor_id');

        $total = $activeVisitorIds->count();
        if ($total === 0) {
            return ['new' => 0, 'returning' => 0, 'new_pct' => 0, 'returning_pct' => 0];
        }

        $returning = AnalyticsSession::where('website_id', $this->website->id)
            ->whereIn('visitor_id', $activeVisitorIds)
            ->select('visitor_id')
            ->groupBy('visitor_id')
            ->havingRaw('COUNT(*) > 1')
            ->get()
            ->count();

        $new = $total - $returning;

        return [
            'new' => $new,
            'returning' => $returning,
            'new_pct' => round(($new / $total) * 100, 1),
            'returning_pct' => round(($returning / $total) * 100, 1),
        ];
    }

    // ---- Acquisition: Traffic Sources ----

    public function trafficSources(int $limit = 50): Collection
    {
        return AnalyticsSession::where('website_id', $this->website->id)
            ->whereBetween('started_at', [$this->from, $this->to])
            ->select(
                DB::raw("COALESCE(NULLIF(referrer_domain, ''), 'Direct') as source"),
                DB::raw('COUNT(*) as sessions'),
                DB::raw('SUM(CASE WHEN is_bounce = 1 THEN 1 ELSE 0 END) as bounces'),
                DB::raw('AVG(duration) as avg_duration'),
                DB::raw('AVG(pages_count) as avg_pages'),
            )
            ->groupBy('source')
            ->orderByDesc('sessions')
            ->limit($limit)
            ->get();
    }

    // ---- Acquisition: Campaigns (UTM) ----

    public function campaigns(int $limit = 50): Collection
    {
        return AnalyticsSession::where('website_id', $this->website->id)
            ->whereBetween('started_at', [$this->from, $this->to])
            ->whereNotNull('utm_source')
            ->select(
                'utm_source', 'utm_medium', 'utm_campaign',
                DB::raw('COUNT(*) as sessions'),
                DB::raw('SUM(CASE WHEN is_bounce = 1 THEN 1 ELSE 0 END) as bounces'),
                DB::raw('AVG(duration) as avg_duration'),
            )
            ->groupBy('utm_source', 'utm_medium', 'utm_campaign')
            ->orderByDesc('sessions')
            ->limit($limit)
            ->get();
    }

    // ---- Acquisition: Referrers ----

    public function referrers(int $limit = 50): Collection
    {
        return AnalyticsSession::where('website_id', $this->website->id)
            ->whereBetween('started_at', [$this->from, $this->to])
            ->whereNotNull('referrer_domain')
            ->where('referrer_domain', '!=', '')
            ->select(
                'referrer_domain',
                DB::raw('COUNT(*) as sessions'),
                DB::raw('SUM(CASE WHEN is_bounce = 1 THEN 1 ELSE 0 END) as bounces'),
                DB::raw('AVG(duration) as avg_duration'),
            )
            ->groupBy('referrer_domain')
            ->orderByDesc('sessions')
            ->limit($limit)
            ->get();
    }

    // ---- Behavior: Pages ----

    public function topPages(int $limit = 50): Collection
    {
        [$aggFrom, $aggTo, $liveFrom, $liveTo] = $this->splitRange();
        $aggData = collect();
        $liveData = collect();

        if ($aggFrom) {
            $aggData = PageDailyAggregate::where('website_id', $this->website->id)
                ->whereBetween('date', [$aggFrom->toDateString(), $aggTo->toDateString()])
                ->select(
                    'path',
                    DB::raw('SUM(views) as views'),
                    DB::raw('SUM(unique_views) as unique_views'),
                    DB::raw('CASE WHEN SUM(views) > 0 THEN SUM(total_time_on_page) * 1.0 / SUM(views) ELSE 0 END as avg_time'),
                )
                ->groupBy('path')
                ->get()
                ->keyBy('path');
        }

        if ($liveFrom) {
            $liveData = PageView::where('website_id', $this->website->id)
                ->whereBetween('viewed_at', [$liveFrom, $liveTo])
                ->select(
                    'path',
                    DB::raw('COUNT(*) as views'),
                    DB::raw('COUNT(DISTINCT visitor_id) as unique_views'),
                    DB::raw('AVG(time_on_page) as avg_time'),
                )
                ->groupBy('path')
                ->get()
                ->keyBy('path');
        }

        if ($aggData->isEmpty()) {
            return $liveData->values()->sortByDesc('views')->take($limit)->values();
        }

        if ($liveData->isEmpty()) {
            return $aggData->values()->sortByDesc('views')->take($limit)->values();
        }

        $allPaths = $aggData->keys()->merge($liveData->keys())->unique();

        return $allPaths->map(function (string $path) use ($aggData, $liveData) {
            $agg = $aggData->get($path);
            $live = $liveData->get($path);

            $totalViews = ($agg->views ?? 0) + ($live->views ?? 0);
            $totalUnique = ($agg->unique_views ?? 0) + ($live->unique_views ?? 0);
            $avgTime = $totalViews > 0
                ? (($agg ? $agg->avg_time * $agg->views : 0) + ($live ? $live->avg_time * $live->views : 0)) / $totalViews
                : 0;

            return (object) [
                'path' => $path,
                'views' => $totalViews,
                'unique_views' => $totalUnique,
                'avg_time' => $avgTime,
            ];
        })->sortByDesc('views')->take($limit)->values();
    }

    // ---- Behavior: Landing Pages ----

    public function landingPages(int $limit = 50): Collection
    {
        return AnalyticsSession::where('website_id', $this->website->id)
            ->whereBetween('started_at', [$this->from, $this->to])
            ->whereNotNull('entry_page')
            ->select(
                'entry_page as path',
                DB::raw('COUNT(*) as sessions'),
                DB::raw('SUM(CASE WHEN is_bounce = 1 THEN 1 ELSE 0 END) as bounces'),
                DB::raw('AVG(duration) as avg_duration'),
            )
            ->groupBy('entry_page')
            ->orderByDesc('sessions')
            ->limit($limit)
            ->get();
    }

    // ---- Behavior: Exit Pages ----

    public function exitPages(int $limit = 50): Collection
    {
        return AnalyticsSession::where('website_id', $this->website->id)
            ->whereBetween('started_at', [$this->from, $this->to])
            ->whereNotNull('exit_page')
            ->select(
                'exit_page as path',
                DB::raw('COUNT(*) as exits'),
            )
            ->groupBy('exit_page')
            ->orderByDesc('exits')
            ->limit($limit)
            ->get();
    }

    // ---- Events: Overview ----

    public function eventsOverview(int $limit = 50): Collection
    {
        $driver = DB::connection()->getDriverName();
        $castType = $driver === 'sqlite' ? 'REAL' : 'DECIMAL(10,2)';

        return \App\Models\Event::where('website_id', $this->website->id)
            ->whereBetween('occurred_at', [$this->from, $this->to])
            ->select(
                'name',
                'category',
                DB::raw('COUNT(*) as total'),
                DB::raw('COUNT(DISTINCT visitor_id) as unique_visitors'),
                DB::raw("SUM(CAST(value AS {$castType})) as total_value"),
            )
            ->groupBy('name', 'category')
            ->orderByDesc('total')
            ->limit($limit)
            ->get();
    }

    // ---- Events: Detail ----

    public function eventDetail(string $eventName): Collection
    {
        $driver = DB::connection()->getDriverName();
        $castType = $driver === 'sqlite' ? 'REAL' : 'DECIMAL(10,2)';

        return \App\Models\Event::where('website_id', $this->website->id)
            ->whereBetween('occurred_at', [$this->from, $this->to])
            ->where('name', $eventName)
            ->select(
                'action', 'label',
                DB::raw('COUNT(*) as total'),
                DB::raw("SUM(CAST(value AS {$castType})) as total_value"),
            )
            ->groupBy('action', 'label')
            ->orderByDesc('total')
            ->get();
    }

    // ---- Events: Trending ----

    public function eventTrending(string $eventName): Collection
    {
        $days = $this->from->diffInDays($this->to);
        $groupFormat = $days <= 2 ? '%Y-%m-%d %H:00' : '%Y-%m-%d';
        $carbonFormat = $days <= 2 ? 'Y-m-d H:00' : 'Y-m-d';

        $data = \App\Models\Event::where('website_id', $this->website->id)
            ->whereBetween('occurred_at', [$this->from, $this->to])
            ->where('name', $eventName)
            ->select(
                DB::raw($this->dateFormat($groupFormat, 'occurred_at').' as period'),
                DB::raw('COUNT(*) as total'),
            )
            ->groupBy('period')
            ->orderBy('period')
            ->get()
            ->keyBy('period');

        $periods = collect();
        $cursor = $this->from->copy();
        $increment = $days <= 2 ? 'addHour' : 'addDay';

        while ($cursor->lte($this->to)) {
            $key = $cursor->format($carbonFormat);
            $periods->push([
                'period' => $key,
                'total' => $data->get($key)?->total ?? 0,
            ]);
            $cursor->$increment();
        }

        return $periods;
    }

    // ---- Helpers ----

    private function visitorGroupedBy(string $column, int $limit): Collection
    {
        return Visitor::where('website_id', $this->website->id)
            ->whereNotNull($column)
            ->where($column, '!=', '')
            ->whereHas('pageViews', fn ($q) => $q->whereBetween('viewed_at', [$this->from, $this->to]))
            ->select($column, DB::raw('COUNT(*) as visitors'))
            ->groupBy($column)
            ->orderByDesc('visitors')
            ->limit($limit)
            ->get();
    }

    /**
     * Get the appropriate date formatting function for the current database driver.
     */
    private function dateFormat(string $format, string $column): string
    {
        $driver = DB::connection()->getDriverName();

        if ($driver === 'sqlite') {
            return "strftime('{$format}', {$column})";
        }

        // MySQL/MariaDB
        // Convert SQLite strftime format to MySQL DATE_FORMAT
        $mysqlFormat = str_replace(
            ['%Y', '%m', '%d', '%H', '%M', '%S'],
            ['%Y', '%m', '%d', '%H', '%i', '%s'],
            $format
        );

        return "DATE_FORMAT({$column}, '{$mysqlFormat}')";
    }

    // ---- Conversions ----

    public function conversionsReport(): Collection
    {
        $goals = \App\Models\Goal::where('website_id', $this->website->id)->get();
        $totalSessions = $this->totalSessions();

        return $goals->map(function ($goal) use ($totalSessions) {
            $completions = \App\Models\GoalConversion::where('goal_id', $goal->id)
                ->whereBetween('converted_at', [$this->from, $this->to])
                ->count();

            $totalValue = \App\Models\GoalConversion::where('goal_id', $goal->id)
                ->whereBetween('converted_at', [$this->from, $this->to])
                ->sum('value');

            return (object) [
                'id' => $goal->id,
                'name' => $goal->name,
                'type' => $goal->type,
                'is_active' => $goal->is_active,
                'completions' => $completions,
                'conversion_rate' => $totalSessions > 0 ? round(($completions / $totalSessions) * 100, 2) : 0,
                'total_value' => $totalValue,
            ];
        });
    }

    public function goalCompletionsOverTime(): Collection
    {
        $days = $this->from->diffInDays($this->to);
        $groupFormat = $days <= 2 ? '%Y-%m-%d %H:00' : '%Y-%m-%d';
        $carbonFormat = $days <= 2 ? 'Y-m-d H:00' : 'Y-m-d';

        $data = \App\Models\GoalConversion::where('website_id', $this->website->id)
            ->whereBetween('converted_at', [$this->from, $this->to])
            ->select(
                DB::raw($this->dateFormat($groupFormat, 'converted_at').' as period'),
                DB::raw('COUNT(*) as completions'),
            )
            ->groupBy('period')
            ->orderBy('period')
            ->get()
            ->keyBy('period');

        $periods = collect();
        $cursor = $this->from->copy();
        $increment = $days <= 2 ? 'addHour' : 'addDay';

        while ($cursor->lte($this->to)) {
            $key = $cursor->format($carbonFormat);
            $periods->push([
                'period' => $key,
                'completions' => $data->get($key)?->completions ?? 0,
            ]);
            $cursor->$increment();
        }

        return $periods;
    }

    // ---- Real-Time ----

    public static function activeVisitors(Website $website, int $minutes = 5): int
    {
        return AnalyticsSession::where('website_id', $website->id)
            ->where('ended_at', '>=', Carbon::now()->subMinutes($minutes))
            ->distinct('visitor_id')
            ->count('visitor_id');
    }
}