Backup #1602

Dateiinhalt

<?php

declare(strict_types=1);

namespace Infrastructure\Persistence;

// @responsibility: CRUD-Operationen für Stopwords (Entity-Extraction Filter)

use Domain\Repository\StopwordRepositoryInterface;
use PDO;

final class StopwordRepository implements StopwordRepositoryInterface
{
    private PDO $db;

    /** @var array<string>|null Cached canonical forms */
    private ?array $canonicalCache = null;

    public function __construct(PDO $pdo)
    {
        $this->db = $pdo;
    }

    /**
     * {@inheritDoc}
     */
    public function findAll(bool $activeOnly = true, ?string $category = null): array
    {
        $sql = 'SELECT * FROM stopwords WHERE 1=1';
        $params = [];

        if ($activeOnly) {
            $sql .= ' AND is_active = 1';
        }

        if ($category !== null) {
            $sql .= ' AND category = :category';
            $params['category'] = $category;
        }

        $sql .= ' ORDER BY word ASC';

        $stmt = $this->db->prepare($sql);
        $stmt->execute($params);

        return $stmt->fetchAll();
    }

    /**
     * {@inheritDoc}
     */
    public function find(int $id): ?array
    {
        $stmt = $this->db->prepare('SELECT * FROM stopwords WHERE id = :id');
        $stmt->execute(['id' => $id]);
        $result = $stmt->fetch();

        return $result === false ? null : $result;
    }

    /**
     * {@inheritDoc}
     */
    public function getCanonicalForms(bool $activeOnly = true): array
    {
        if ($this->canonicalCache !== null && $activeOnly) {
            return $this->canonicalCache;
        }

        $sql = 'SELECT canonical_form FROM stopwords';
        if ($activeOnly) {
            $sql .= ' WHERE is_active = 1';
        }

        $stmt = $this->db->query($sql);
        $forms = $stmt->fetchAll(PDO::FETCH_COLUMN);

        if ($activeOnly) {
            $this->canonicalCache = $forms;
        }

        return $forms;
    }

    /**
     * {@inheritDoc}
     */
    public function isStopword(string $word): bool
    {
        $canonical = $this->normalize($word);
        $forms = $this->getCanonicalForms();

        return in_array($canonical, $forms, true);
    }

    /**
     * {@inheritDoc}
     */
    public function create(array $data): int
    {
        $canonical = $this->normalize($data['word'] ?? '');

        $stmt = $this->db->prepare(
            'INSERT INTO stopwords (word, canonical_form, category, reason, is_active)
             VALUES (:word, :canonical, :category, :reason, :active)'
        );

        $stmt->execute([
            'word' => $data['word'],
            'canonical' => $canonical,
            'category' => $data['category'] ?? 'generic',
            'reason' => $data['reason'] ?? null,
            'active' => $data['is_active'] ?? 1,
        ]);

        $this->canonicalCache = null;

        return (int) $this->db->lastInsertId();
    }

    /**
     * {@inheritDoc}
     */
    public function update(int $id, array $data): bool
    {
        $sets = [];
        $params = ['id' => $id];

        if (isset($data['word'])) {
            $sets[] = 'word = :word';
            $sets[] = 'canonical_form = :canonical';
            $params['word'] = $data['word'];
            $params['canonical'] = $this->normalize($data['word']);
        }

        if (isset($data['category'])) {
            $sets[] = 'category = :category';
            $params['category'] = $data['category'];
        }

        if (array_key_exists('reason', $data)) {
            $sets[] = 'reason = :reason';
            $params['reason'] = $data['reason'];
        }

        if (isset($data['is_active'])) {
            $sets[] = 'is_active = :active';
            $params['active'] = $data['is_active'];
        }

        if ($sets === []) {
            return false;
        }

        $sql = 'UPDATE stopwords SET ' . implode(', ', $sets) . ' WHERE id = :id';
        $stmt = $this->db->prepare($sql);
        $result = $stmt->execute($params);

        $this->canonicalCache = null;

        return $result && $stmt->rowCount() > 0;
    }

    /**
     * {@inheritDoc}
     */
    public function delete(int $id): bool
    {
        $stmt = $this->db->prepare('DELETE FROM stopwords WHERE id = :id');
        $result = $stmt->execute(['id' => $id]);

        $this->canonicalCache = null;

        return $result && $stmt->rowCount() > 0;
    }

    /**
     * {@inheritDoc}
     */
    public function toggleActive(int $id): bool
    {
        $stmt = $this->db->prepare(
            'UPDATE stopwords SET is_active = NOT is_active WHERE id = :id'
        );
        $result = $stmt->execute(['id' => $id]);

        $this->canonicalCache = null;

        return $result && $stmt->rowCount() > 0;
    }

    /**
     * {@inheritDoc}
     */
    public function getStats(): array
    {
        $stmt = $this->db->query(
            'SELECT category, COUNT(*) as count, SUM(is_active) as active
             FROM stopwords GROUP BY category'
        );

        $stats = ['total' => 0, 'active' => 0, 'by_category' => []];

        foreach ($stmt->fetchAll() as $row) {
            $stats['by_category'][$row['category']] = [
                'count' => (int) $row['count'],
                'active' => (int) $row['active'],
            ];
            $stats['total'] += (int) $row['count'];
            $stats['active'] += (int) $row['active'];
        }

        return $stats;
    }

    /**
     * Normalize word to canonical form.
     */
    private function normalize(string $word): string
    {
        $result = mb_strtolower(trim($word));

        // German umlauts
        $replacements = [
            'ä' => 'ae', 'ö' => 'oe', 'ü' => 'ue', 'ß' => 'ss',
        ];
        $result = str_replace(array_keys($replacements), array_values($replacements), $result);

        // Remove special chars, keep only alphanumeric and underscore
        $result = preg_replace('/[^a-z0-9_]/', '', $result);

        return $result ?? '';
    }
}

Vollständig herunterladen

Aktionen

← Zurück zur Übersicht