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

<?php

namespace App\Console\Commands;

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

/**
 * Phase 1 — Schritt 1: Duplikate identifizieren
 *
 * Sucht Kunden-Datensätze, die vermutlich dieselbe Person repräsentieren.
 * Drei Erkennungs-Stufen (absteigend nach Konfidenz):
 *
 *   HIGH   — gleiche E-Mail (nicht leer)
 *   MEDIUM — gleicher Name + Vorname + Geburtsdatum
 *   LOW    — gleicher Name + Vorname + PLZ
 *
 * Verwendung:
 *   php artisan contacts:find-duplicates
 *   php artisan contacts:find-duplicates --export=duplicates.csv
 *   php artisan contacts:find-duplicates --confidence=HIGH
 */
class ContactsFindDuplicates extends Command
{
    protected $signature = 'contacts:find-duplicates
                            {--export= : Pfad zur CSV-Ausgabedatei}
                            {--confidence= : Nur diese Konfidenz-Stufe ausgeben (HIGH|MEDIUM|LOW)}';

    protected $description = 'Identifiziert doppelte Customer-Datensätze anhand von E-Mail, Name/Geburtsdatum oder Name/PLZ';

    public function handle(): int
    {
        $this->info('Suche nach Duplikaten in der customer-Tabelle...');
        $this->newLine();

        $groups = collect();

        // ── HIGH: gleiche E-Mail ──────────────────────────────────────────
        if ($this->shouldCheck('HIGH')) {
            $emailDupes = DB::table('contacts')
                ->select('email', DB::raw('COUNT(*) as cnt'), DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC) as ids'))
                ->whereNotNull('email')
                ->where('email', '!=', '')
                ->whereNull('merged_into_id')
                ->groupBy('email')
                ->having('cnt', '>', 1)
                ->get();

            foreach ($emailDupes as $row) {
                $groups->push([
                    'confidence' => 'HIGH',
                    'reason'     => 'E-Mail: ' . $row->email,
                    'ids'        => $row->ids,
                    'count'      => $row->cnt,
                ]);
            }

            $this->line(sprintf('<fg=green>HIGH</> (gleiche E-Mail):   %d Gruppen', $emailDupes->count()));
        }

        // ── MEDIUM: Name + Vorname + Geburtsdatum ────────────────────────
        if ($this->shouldCheck('MEDIUM')) {
            $nameBdDupes = DB::table('contacts')
                ->select(
                    'name', 'firstname', 'birthdate',
                    DB::raw('COUNT(*) as cnt'),
                    DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC) as ids')
                )
                ->whereNotNull('name')
                ->whereNotNull('firstname')
                ->whereNotNull('birthdate')
                ->whereNull('merged_into_id')
                ->groupBy('name', 'firstname', 'birthdate')
                ->having('cnt', '>', 1)
                ->get();

            foreach ($nameBdDupes as $row) {
                $groups->push([
                    'confidence' => 'MEDIUM',
                    'reason'     => "Name: {$row->firstname} {$row->name}, GD: {$row->birthdate}",
                    'ids'        => $row->ids,
                    'count'      => $row->cnt,
                ]);
            }

            $this->line(sprintf('<fg=yellow>MEDIUM</> (Name+GD):         %d Gruppen', $nameBdDupes->count()));
        }

        // ── LOW: Name + Vorname + PLZ ─────────────────────────────────────
        if ($this->shouldCheck('LOW')) {
            $nameZipDupes = DB::table('contacts')
                ->select(
                    'name', 'firstname', 'zip',
                    DB::raw('COUNT(*) as cnt'),
                    DB::raw('GROUP_CONCAT(id ORDER BY updated_at DESC) as ids')
                )
                ->whereNotNull('name')
                ->whereNotNull('firstname')
                ->whereNotNull('zip')
                ->where('zip', '!=', '')
                ->whereNull('merged_into_id')
                ->groupBy('name', 'firstname', 'zip')
                ->having('cnt', '>', 1)
                ->get();

            foreach ($nameZipDupes as $row) {
                $groups->push([
                    'confidence' => 'LOW',
                    'reason'     => "Name: {$row->firstname} {$row->name}, PLZ: {$row->zip}",
                    'ids'        => $row->ids,
                    'count'      => $row->cnt,
                ]);
            }

            $this->line(sprintf('<fg=red>LOW</> (Name+PLZ):           %d Gruppen', $nameZipDupes->count()));
        }

        $this->newLine();
        $this->info(sprintf('Gesamt: %d Duplikat-Gruppen gefunden', $groups->count()));

        if ($groups->isEmpty()) {
            $this->info('Keine Duplikate — nichts zu tun.');
            return self::SUCCESS;
        }

        // Tabellen-Ausgabe
        $this->table(
            ['Konfidenz', 'Grund', 'IDs (neueste zuerst)', 'Anzahl'],
            $groups->map(fn ($g) => [$g['confidence'], $g['reason'], $g['ids'], $g['count']])->all()
        );

        // CSV-Export
        if ($export = $this->option('export')) {
            $this->exportCsv($groups->all(), $export);
            $this->info("CSV gespeichert: {$export}");
        } else {
            $this->newLine();
            $this->line('Tipp: --export=duplicates.csv für CSV-Export');
            $this->line('Tipp: php artisan contacts:merge-duplicates --dry-run  zum Prüfen');
        }

        return self::SUCCESS;
    }

    private function shouldCheck(string $level): bool
    {
        $filter = strtoupper((string) $this->option('confidence'));
        return $filter === '' || $filter === $level;
    }

    private function exportCsv(array $groups, string $path): void
    {
        $handle = fopen($path, 'w');
        fputcsv($handle, ['Konfidenz', 'Grund', 'IDs (neueste zuerst)', 'Anzahl']);
        foreach ($groups as $group) {
            fputcsv($handle, [$group['confidence'], $group['reason'], $group['ids'], $group['count']]);
        }
        fclose($handle);
    }
}