RSS 피드 자동 큐레이션

// RSS 큐레이션 시스템 - 테이블 생성 function map_rss_create_tables() { global $wpdb; $charset = $wpdb->get_charset_collate(); // 피드 소스 테이블 $sql1 = "CREATE TABLE IF NOT EXISTS {$wpdb->prefix}map_rss_feeds ( id BIGINT UNSIGNED AUTO_INCREMENT PRIMARY KEY, name VARCHAR(200) NOT NULL, url VARCHAR(500) NOT NULL, category VARCHAR(100) DEFAULT '', max_items INT DEFAULT 10, fetch_interval VARCHAR(20) DEFAULT 'hourly', is_active TINYINT(1) DEFAULT 1, last_fetched DATETIME DEFAULT NULL, last_error TEXT DEFAULT NULL, total_items_fetched INT DEFAULT 0, created_at DATETIME DEFAULT CURRENT_TIMESTAMP, INDEX idx_active (is_active), UNIQUE KEY uk_url (url(255)) ) {$charset};"; // 수집 아이템 테이블 $sql2 = "CREATE TABLE IF NOT EXISTS {$wpdb->prefix}map_rss_items ( id BIGINT UNSIGNED AUTO_INCREMENT PRIMARY KEY, feed_id BIGINT UNSIGNED NOT NULL, guid VARCHAR(64) NOT NULL, title VARCHAR(500) NOT NULL, link VARCHAR(1000) NOT NULL, description TEXT, ai_summary TEXT DEFAULT NULL, ai_category VARCHAR(100) DEFAULT NULL, ai_relevance_score DECIMAL(5,2) DEFAULT NULL, source_name VARCHAR(200), author VARCHAR(200) DEFAULT '', pub_date DATETIME, category VARCHAR(100) DEFAULT '', status ENUM('pending','approved','rejected','used','spam') DEFAULT 'pending', post_id BIGINT DEFAULT NULL, created_at DATETIME DEFAULT CURRENT_TIMESTAMP, UNIQUE KEY uk_guid (guid), INDEX idx_status (status), INDEX idx_feed_id (feed_id), INDEX idx_pub_date (pub_date) ) {$charset};"; require_once ABSPATH . 'wp-admin/includes/upgrade.php'; dbDelta($sql1); dbDelta($sql2); } register_activation_hook(__FILE__, 'map_rss_create_tables');

// RSS 피드 수집 및 큐레이션 핵심 클래스 class MAP_RSS_Curator { private $items_table; private $feeds_table; private $spam_filter; public function __construct() { global $wpdb; $this->items_table = $wpdb->prefix . 'map_rss_items'; $this->feeds_table = $wpdb->prefix . 'map_rss_feeds'; $this->spam_filter = new MAP_RSS_Spam_Filter(); // 매시간 RSS 피드 체크 add_action('map_fetch_rss_feeds', [$this, 'fetch_all_feeds']); if (!wp_next_scheduled('map_fetch_rss_feeds')) { wp_schedule_event(time(), 'hourly', 'map_fetch_rss_feeds'); } } // 피드 소스 추가 public function add_feed($name, $url, $category = '', $max_items = 10) { global $wpdb; // URL 유효성 검증 if (!filter_var($url, FILTER_VALIDATE_URL)) { return new WP_Error('invalid_url', '유효하지 않은 URL입니다.'); } // RSS 피드인지 확인 include_once ABSPATH . WPINC . '/feed.php'; $test = fetch_feed($url); if (is_wp_error($test)) { return new WP_Error('invalid_feed', 'RSS 피드를 파싱할 수 없습니다: ' . $test->get_error_message()); } return $wpdb->insert($this->feeds_table, [ 'name' => sanitize_text_field($name), 'url' => esc_url_raw($url), 'category' => sanitize_text_field($category), 'max_items'=> absint($max_items), ]); } // ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ // 모든 활성 RSS 피드 수집 // ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ public function fetch_all_feeds() { global $wpdb; $feeds = $wpdb->get_results( "SELECT * FROM {$this->feeds_table} WHERE is_active = 1" ); $total_new = 0; foreach ($feeds as $feed) { $count = $this->fetch_feed($feed); $total_new += $count; } error_log(sprintf( '[RSS Curator] %d개 피드에서 %d개 새 아이템 수집', count($feeds), $total_new )); return $total_new; } // 개별 피드 수집 private function fetch_feed($feed) { global $wpdb; include_once ABSPATH . WPINC . '/feed.php'; // SimplePie로 RSS 파싱 $rss = fetch_feed($feed->url); if (is_wp_error($rss)) { // 에러 기록 $wpdb->update($this->feeds_table, [ 'last_error' => $rss->get_error_message(), ], ['id' => $feed->id]); error_log("[RSS] 피드 오류 [{$feed->name}]: " . $rss->get_error_message()); return 0; } $items = $rss->get_items(0, $feed->max_items); $new_count = 0; foreach ($items as $item) { $guid = md5($item->get_permalink()); // 중복 체크 $exists = $wpdb->get_var($wpdb->prepare( "SELECT COUNT(*) FROM {$this->items_table} WHERE guid = %s", $guid )); if ($exists) continue; // 설명 텍스트 정리 (HTML 제거, 100단어 제한) $description = wp_trim_words( strip_tags($item->get_description()), 100 ); // 스팸 필터 검사 $status = $this->spam_filter->check($item->get_title(), $description) ? 'spam' : 'pending'; // 새 아이템 저장 $wpdb->insert($this->items_table, [ 'feed_id' => $feed->id, 'guid' => $guid, 'title' => sanitize_text_field($item->get_title()), 'link' => esc_url_raw($item->get_permalink()), 'description' => $description, 'author' => sanitize_text_field($item->get_author() ? $item->get_author()->get_name() : ''), 'source_name' => $feed->name, 'pub_date' => $item->get_date('Y-m-d H:i:s') ?: current_time('mysql'), 'category' => $feed->category, 'status' => $status, ]); $new_count++; } // 피드 메타 업데이트 $wpdb->update($this->feeds_table, [ 'last_fetched' => current_time('mysql'), 'last_error' => null, 'total_items_fetched' => $feed->total_items_fetched + $new_count, ], ['id' => $feed->id]); return $new_count; } }