mein-sterntours/app/Console/Commands/ContactsMergeDuplicates.php

<?php

namespace App\Console\Commands;

use Illuminate\Console\Command;
use Illuminate\Support\Facades\DB;

/**
 * Phase 1 — Schritt 2: Duplikate zusammenführen
 *
 * Strategie: Der neueste Datensatz (höchstes updated_at, dann höchste id)
 * wird Master. Alle anderen Datensätze derselben Gruppe erhalten
 * merged_into_id = master_id und werden nicht mehr zurückgegeben.
 *
 * Alle FK-Referenzen in lead, booking, customer_mails, lead_mails
 * werden auf den Master umgestellt.
 *
 * Verwendung:
 *   php artisan contacts:merge-duplicates --dry-run          # Vorschau, keine Änderung
 *   php artisan contacts:merge-duplicates --confidence=HIGH  # Nur sichere Duplikate
 *   php artisan contacts:merge-duplicates                    # Ausführen
 */
class ContactsMergeDuplicates extends Command
{
    protected $signature = 'contacts:merge-duplicates
                            {--dry-run : Zeigt was passieren würde, ohne Daten zu ändern}
                            {--confidence= : Nur diese Konfidenz-Stufe verarbeiten (HIGH|MEDIUM|LOW)}
                            {--force : Überspringt Sicherheitsabfrage}';

    protected $description = 'Führt doppelte Customer-Datensätze zusammen (neuester wird Master)';

    private bool $dryRun = false;
    private int $mergedCount = 0;
    private int $updatedLeads = 0;
    private int $updatedBookings = 0;

    public function handle(): int
    {
        $this->dryRun = (bool) $this->option('dry-run');

        if ($this->dryRun) {
            $this->warn('DRY-RUN Modus — keine Daten werden verändert');
        } else {
            $this->warn('ACHTUNG: Diese Operation verändert Produktionsdaten.');
            if (!$this->option('force') && !$this->confirm('Fortfahren?')) {
                $this->info('Abgebrochen.');
                return self::SUCCESS;
            }
        }

        $this->newLine();

        DB::transaction(function () {
            $this->processLevel(
                'HIGH',
                fn () => $this->findByEmail()
            );

            $this->processLevel(
                'MEDIUM',
                fn () => $this->findByNameBirthdate()
            );

            $this->processLevel(
                'LOW',
                fn () => $this->findByNameZip()
            );
        });

        $this->newLine();
        $this->info(sprintf(
            '%s %d Duplikate zusammengeführt | %d leads aktualisiert | %d bookings aktualisiert',
            $this->dryRun ? '[DRY-RUN]' : '',
            $this->mergedCount,
            $this->updatedLeads,
            $this->updatedBookings
        ));

        if ($this->dryRun) {
            $this->newLine();
            $this->line('Zum Ausführen: php artisan contacts:merge-duplicates');
        }

        return self::SUCCESS;
    }

    // ─────────────────────────────────────────────────────────────────────────
    // Duplikat-Gruppen ermitteln
    // ─────────────────────────────────────────────────────────────────────────

    private function findByEmail(): array
    {
        return DB::table('contacts')
            ->select('email', DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC, id DESC) as ids'))
            ->whereNotNull('email')
            ->where('email', '!=', '')
            ->whereNull('merged_into_id')
            ->groupBy('email')
            ->having(DB::raw('COUNT(*)'), '>', 1)
            ->pluck('ids')
            ->map(fn ($ids) => explode(',', $ids))
            ->all();
    }

    private function findByNameBirthdate(): array
    {
        return DB::table('contacts')
            ->select(DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC, id DESC) as ids'))
            ->whereNotNull('name')
            ->whereNotNull('firstname')
            ->whereNotNull('birthdate')
            ->whereNull('merged_into_id')
            ->groupBy('name', 'firstname', 'birthdate')
            ->having(DB::raw('COUNT(*)'), '>', 1)
            ->pluck('ids')
            ->map(fn ($ids) => explode(',', $ids))
            ->all();
    }

    private function findByNameZip(): array
    {
        return DB::table('contacts')
            ->select(DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC, id DESC) as ids'))
            ->whereNotNull('name')
            ->whereNotNull('firstname')
            ->whereNotNull('zip')
            ->where('zip', '!=', '')
            ->whereNull('merged_into_id')
            ->groupBy('name', 'firstname', 'zip')
            ->having(DB::raw('COUNT(*)'), '>', 1)
            ->pluck('ids')
            ->map(fn ($ids) => explode(',', $ids))
            ->all();
    }

    // ─────────────────────────────────────────────────────────────────────────
    // Verarbeitung
    // ─────────────────────────────────────────────────────────────────────────

    private function processLevel(string $level, callable $finder): void
    {
        if ($filter = $this->option('confidence')) {
            if (strtoupper($filter) !== $level) {
                return;
            }
        }

        $groups = $finder();

        if (empty($groups)) {
            $this->line("[{$level}] Keine Duplikate.");
            return;
        }

        $this->line(sprintf('[%s] %d Gruppe(n) gefunden', $level, count($groups)));

        foreach ($groups as $ids) {
            $masterId  = (int) $ids[0];           // erster = neuester
            $duplicateIds = array_map('intval', array_slice($ids, 1));

            $this->line(sprintf(
                '  Master: #%d  ←  Duplikate: %s',
                $masterId,
                implode(', ', array_map(fn ($id) => '#' . $id, $duplicateIds))
            ));

            foreach ($duplicateIds as $dupeId) {
                $this->mergeInto($masterId, $dupeId);
            }
        }
    }

    private function mergeInto(int $masterId, int $dupeId): void
    {
        // 1. Leads umhängen
        $leadCount = DB::table('inquiries')->where('customer_id', $dupeId)->count();
        if ($leadCount > 0) {
            $this->line("    lead.customer_id: {$leadCount} Zeile(n) → #{$masterId}");
            if (!$this->dryRun) {
                DB::table('inquiries')
                    ->where('customer_id', $dupeId)
                    ->update(['customer_id' => $masterId]);
            }
            $this->updatedLeads += $leadCount;
        }

        // 2. Bookings umhängen
        $bookingCount = DB::table('booking')->where('customer_id', $dupeId)->count();
        if ($bookingCount > 0) {
            $this->line("    booking.customer_id: {$bookingCount} Zeile(n) → #{$masterId}");
            if (!$this->dryRun) {
                DB::table('booking')
                    ->where('customer_id', $dupeId)
                    ->update(['customer_id' => $masterId]);
            }
            $this->updatedBookings += $bookingCount;
        }

        // 3. customer_mails umhängen
        $mailCount = DB::table('customer_mails')->where('customer_id', $dupeId)->count();
        if ($mailCount > 0) {
            $this->line("    customer_mails.customer_id: {$mailCount} Zeile(n) → #{$masterId}");
            if (!$this->dryRun) {
                DB::table('customer_mails')
                    ->where('customer_id', $dupeId)
                    ->update(['customer_id' => $masterId]);
            }
        }

        // 4. lead_mails umhängen
        $leadMailCount = DB::table('lead_mails')->where('customer_id', $dupeId)->count();
        if ($leadMailCount > 0) {
            $this->line("    lead_mails.customer_id: {$leadMailCount} Zeile(n) → #{$masterId}");
            if (!$this->dryRun) {
                DB::table('lead_mails')
                    ->where('customer_id', $dupeId)
                    ->update(['customer_id' => $masterId]);
            }
        }

        // 5. Duplikat als zusammengeführt markieren
        if (!$this->dryRun) {
            DB::table('contacts')
                ->where('id', $dupeId)
                ->update([
                    'merged_into_id' => $masterId,
                    'merged_at'      => now(),
                ]);
        }

        $this->mergedCount++;
    }
}