테크니컬 SEO 자동 감사

// 깨진 링크 종합 스캐너
class Broken_Link_Scanner {
    private $results = array();
    private $checked_urls = array();
    private $batch_size = 20;

    public function full_scan() {
        global $wpdb;

        $posts = $wpdb->get_results(
            "SELECT ID, post_content, post_title
             FROM {$wpdb->posts}
             WHERE post_status = 'publish'
             AND post_type IN ('post', 'page', 'product')"
        );

        foreach ($posts as $post) {
            $this->scan_post($post);
        }

        // 결과 저장
        update_option('broken_link_scan_results', $this->results);
        update_option('broken_link_scan_date', current_time('mysql'));

        return $this->results;
    }

    private function scan_post($post) {
        // 모든 링크 추출
        preg_match_all('/href=["\']([^"\']+)["\']/i', $post->post_content, $matches);

        foreach ($matches[1] as $url) {
            // 상대 URL을 절대 URL로 변환
            if (strpos($url, '/') === 0) {
                $url = home_url($url);
            }

            // 이미 체크한 URL 스킵
            if (isset($this->checked_urls[$url])) {
                if ($this->checked_urls[$url] !== 200) {
                    $this->add_result($post, $url, $this->checked_urls[$url]);
                }
                continue;
            }

            // HTTP 상태 코드 확인
            $status = $this->check_url($url);
            $this->checked_urls[$url] = $status;

            if ($status !== 200 && $status !== 301 && $status !== 302) {
                $this->add_result($post, $url, $status);
            }
        }
    }

    private function check_url($url) {
        $response = wp_remote_head($url, array(
            'timeout' => 10,
            'redirection' => 0,
            'sslverify' => false,
        ));

        if (is_wp_error($response)) return 0;
        return wp_remote_retrieve_response_code($response);
    }

    private function add_result($post, $url, $status) {
        $this->results[] = array(
            'post_id' => $post->ID,
            'post_title' => $post->post_title,
            'post_url' => get_permalink($post->ID),
            'broken_url' => $url,
            'status_code' => $status,
            'type' => $this->classify_error($status),
        );
    }

    private function classify_error($status) {
        if ($status === 0) return 'connection_failed';
        if ($status === 403) return 'forbidden';
        if ($status === 404) return 'not_found';
        if ($status >= 500) return 'server_error';
        return 'other_error';
    }
}

// 깨진 링크 자동 수정 시스템
function auto_fix_broken_links($scan_results) {
    $fixed = 0;
    $suggestions = array();

    foreach ($scan_results as $result) {
        if ($result['status_code'] !== 404) continue;

        $broken_url = $result['broken_url'];
        $post_id = $result['post_id'];

        // 1. 내부 링크인 경우: 리다이렉트 맵에서 대체 URL 찾기
        if (strpos($broken_url, home_url()) === 0) {
            $redirect_url = find_redirect_for_url($broken_url);
            if ($redirect_url) {
                replace_url_in_post($post_id, $broken_url, $redirect_url);
                $fixed++;
                continue;
            }

            // 유사한 slug의 페이지 찾기
            $path = parse_url($broken_url, PHP_URL_PATH);
            $slug = basename(rtrim($path, '/'));
            $similar = find_similar_post($slug);
            if ($similar) {
                $suggestions[] = array(
                    'post_id' => $post_id,
                    'broken_url' => $broken_url,
                    'suggested_url' => get_permalink($similar->ID),
                    'suggested_title' => $similar->post_title,
                    'confidence' => 'high',
                );
            }
        }

        // 2. 외부 링크인 경우: 제거 또는 Wayback Machine 링크로 대체
        else {
            $wayback_url = "https://web.archive.org/web/" . $broken_url;
            $suggestions[] = array(
                'post_id' => $post_id,
                'broken_url' => $broken_url,
                'suggested_url' => $wayback_url,
                'suggested_title' => 'Wayback Machine 아카이브',
                'confidence' => 'low',
            );
        }
    }

    return array('fixed' => $fixed, 'suggestions' => $suggestions);
}

function replace_url_in_post($post_id, $old_url, $new_url) {
    $content = get_post_field('post_content', $post_id);
    $content = str_replace($old_url, $new_url, $content);
    wp_update_post(array('ID' => $post_id, 'post_content' => $content));
}