<?php
if ( ! defined( 'ABSPATH' ) ) exit; // Exit if accessed directly

/**
 * Scanner for 404 Scout. Extracts links from posts and saves them to the DB.
 *
 * @since      1.0
 * @package    AILinkMaster
 * @subpackage AILinkMaster/modules/404-scout
 */

class AILM_Scout_Scanner {

    /**
     * Parse a single post and save its links to the monitor table.
     * @param int $post_id The ID of the post to scan.
     */
    public static function process_post( $post_id ) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ailm_link_monitor';

        // 1. Clear old records for this post to avoid duplicates on re-scan
        // phpcs:ignore WordPress.DB.DirectDatabaseQuery
        $wpdb->delete( $table_name, array( 'post_id' => $post_id ), array( '%d' ) );

        $post = get_post( $post_id );
        if ( ! $post ) {
            return;
        }

        $links_to_insert = array();
        $home_url = untrailingslashit( home_url() );

        // 2. Find "Manual" links in the raw content
        $content = $post->post_content;
        // FIX: Escaped quotes in regex
        if ( preg_match_all( '/<a\s+[^>]*href=[\'"]([^\'" ]+)[\'"][^>]*>(.*?)<\/a>/is', $content, $matches ) ) {
            foreach ( $matches[1] as $index => $url ) {
                $url = trim( html_entity_decode( $url ) );
                
                // Skip anchor links, emails, tel, and empty hrefs
                if ( empty( $url ) || strpos( $url, 'http' ) !== 0 ) {
                    continue; 
                }
                
                // Determine link type (internal or external)
                $is_internal = ( strpos( $url, $home_url ) === 0 );
                $link_type = $is_internal ? 'internal' : 'external';
                $anchor = wp_strip_all_tags( $matches[2][$index] );

                // Avoid duplicates within the same post
                $hash = md5( $url . $anchor );
                $links_to_insert[$hash] = array(
                    'post_id'     => $post_id,
                    'url'         => esc_url_raw( $url ),
                    'anchor'      => sanitize_text_field( $anchor ),
                    'link_origin' => 'manual',
                    'link_type'   => $link_type,
                    'status_code' => 0, // 0 = pending check
                    'last_checked'=> '0000-00-00 00:00:00'
                );
            }
        }

        // 3. Find dynamic plugin links (AILM) in the frozen map
        $frozen_map = get_post_meta( $post_id, '_ailm_frozen_map', true );
        if ( is_array( $frozen_map ) && ! empty( $frozen_map ) ) {
            foreach ( $frozen_map as $item ) {
                if ( empty( $item['url'] ) ) continue;
                
                $hash = md5( $item['url'] . $item['word'] . 'ailm' );
                $links_to_insert[$hash] = array(
                    'post_id'     => $post_id,
                    'url'         => esc_url_raw( $item['url'] ),
                    'anchor'      => sanitize_text_field( $item['word'] ),
                    'link_origin' => 'ailm',
                    'link_type'   => 'internal', // AILinkMaster links are always internal
                    'status_code' => 0,
                    'last_checked'=> '0000-00-00 00:00:00'
                );
            }
        }

        // 4. Insert all found links into the database
        if ( ! empty( $links_to_insert ) ) {
            foreach ( $links_to_insert as $data ) {
                // phpcs:ignore WordPress.DB.DirectDatabaseQuery
                $wpdb->insert( 
                    $table_name, 
                    $data, 
                    array( '%d', '%s', '%s', '%s', '%s', '%d', '%s' ) 
                );
            }
        }
    }
}