<?php

if (!defined('ABSPATH')) {
    exit;
}

class AIEO_Content_Text {

    const CURSOR_OPTION = 'aieo_content_text_cursor';

    
    const SCHEMA_VERSION = '2026-08-content-text-v1';

    
    private static $schema_checked = false;

    
    public static function table() {
        global $wpdb;
        return $wpdb->prefix . 'aieo_content_text';
    }

    
    public static function post_types() {
        return apply_filters('aieo_content_text_post_types', array('product', 'post', 'page'));
    }

    
    public static function ensure_table() {
        global $wpdb;
        $table = self::table();
        $charset_collate = aieo_get_charset_collate();
        
        $wpdb->query("CREATE TABLE IF NOT EXISTS {$table} (
            `object_type` varchar(32) NOT NULL,
            `object_id` bigint unsigned NOT NULL,
            `text_hash` char(32) NOT NULL DEFAULT '',
            `text_content` mediumtext NULL,
            `word_count` int unsigned NOT NULL DEFAULT 0,
            `source_modified` datetime DEFAULT NULL,
            `updated_at` datetime DEFAULT NULL,
            PRIMARY KEY (`object_type`, `object_id`),
            KEY `updated_at` (`updated_at`)
        ) {$charset_collate}");
    }

    
    public static function maybe_install() {
        if (self::$schema_checked) {
            return;
        }
        if (get_option('aieo_content_text_schema_version') !== self::SCHEMA_VERSION && AIEO_Schema_Lock::acquire('aieo_content_text_schema_version')) {
            self::ensure_table();
            update_option('aieo_content_text_schema_version', self::SCHEMA_VERSION, false);
        }
        self::$schema_checked = true;
    }

    
    public static function write_row($object_type, $object_id, $text, $source_modified = null) {
        global $wpdb;
        self::maybe_install();
        $text = (string) $text;
        
        
        
        
        
        $words = preg_split('/\s+/u', $text, -1, PREG_SPLIT_NO_EMPTY);
        if (!is_array($words)) { $words = preg_split('/\s+/', $text, -1, PREG_SPLIT_NO_EMPTY); }
        
        return false !== $wpdb->replace(self::table(), array(
            'object_type'     => (string) $object_type,
            'object_id'       => (int) $object_id,
            'text_hash'       => md5($text),
            'text_content'    => $text,
            'word_count'      => $text === '' ? 0 : count((array) $words),
            'source_modified' => $source_modified,
            'updated_at'      => current_time('mysql', true),
        ));
    }

    
    public static function refresh($batch = 300) {
        global $wpdb;
        self::maybe_install();
        $table = self::table();

        
        
        
        
        
        if (false !== get_option(self::CURSOR_OPTION, false)) {
            
            if (null === $wpdb->get_var("SELECT 1 FROM {$table} LIMIT 1")) {
                self::reseed();
            }
        }

        $cursor = get_option(self::CURSOR_OPTION, array('modified' => '1970-01-01 00:00:00', 'id' => 0));
        $modified = isset($cursor['modified']) ? (string) $cursor['modified'] : '1970-01-01 00:00:00';
        $after_id = isset($cursor['id']) ? (int) $cursor['id'] : 0;

        $types = self::post_types();
        $placeholders = implode(',', array_fill(0, count($types), '%s'));

        
        $rows = $wpdb->get_results($wpdb->prepare(
            "SELECT ID, post_type, post_content, post_modified
               FROM {$wpdb->posts}
              WHERE post_status = 'publish'
                AND post_type IN ({$placeholders})
                AND (post_modified > %s OR (post_modified = %s AND ID > %d))
              ORDER BY post_modified ASC, ID ASC
              LIMIT %d",
            array_merge($types, array($modified, $modified, $after_id, $batch))
        ));

        $scanned = count($rows);
        $written = 0;

        if ($scanned > 0) {
            
            $keys = array();
            foreach ($rows as $r) {
                $keys[] = $wpdb->prepare('(%s,%d)', $r->post_type, (int) $r->ID);
            }
            
            $existing_rows = $wpdb->get_results(
                "SELECT object_type, object_id, text_hash FROM {$table}
                  WHERE (object_type, object_id) IN (" . implode(',', $keys) . ')'
            );
            $existing = array();
            foreach ($existing_rows as $e) {
                $existing[$e->object_type . ':' . $e->object_id] = $e->text_hash;
            }

            $last_ok = null;
            $error   = '';
            foreach ($rows as $r) {
                $text = self::strip((string) $r->post_content);
                $hash = md5($text);
                $key = $r->post_type . ':' . $r->ID;
                if (isset($existing[$key]) && $existing[$key] === $hash) {
                    $last_ok = $r;
                    continue; 
                }
                if (!self::write_row($r->post_type, (int) $r->ID, $text, $r->post_modified)) {
                    
                    
                    
                    $error = (string) $wpdb->last_error;
                    break;
                }
                $written++;
                $last_ok = $r;
            }

            if (null !== $last_ok) {
                update_option(self::CURSOR_OPTION, array(
                    'modified' => $last_ok->post_modified,
                    'id'       => (int) $last_ok->ID,
                ), false);
            }

            if ('' !== $error) {
                
                
                
                return array('scanned' => $scanned, 'written' => $written, 'done' => true, 'error' => $error);
            }
        }

        $done = $scanned < $batch;
        if ($done) {
            self::gc();
        }

        return array('scanned' => $scanned, 'written' => $written, 'done' => $done);
    }

    
    public static function gc() {
        global $wpdb;
        $table = self::table();
        
        return (int) $wpdb->query(
            "DELETE ct FROM {$table} ct
              LEFT JOIN {$wpdb->posts} p
                     ON p.ID = ct.object_id AND p.post_type = ct.object_type
              WHERE p.ID IS NULL OR p.post_status <> 'publish'"
        );
    }

    
    public static function reseed() {
        delete_option(self::CURSOR_OPTION);
    }

    
    public static function strip($html) {
        if ($html === '') {
            return '';
        }
        
        
        $text = strip_shortcodes($html);
        $text = preg_replace('/\[\/?[a-zA-Z0-9_\-]+[^\]\n]{0,200}\]/u', ' ', $text);
        
        
        $text = preg_replace('/<(\/?)(p|div|br|li|tr|td|th|h[1-6]|section|article|blockquote)\b[^>]*>/iu', ' ', $text);
        $text = wp_strip_all_tags($text, false);
        $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
        
        $text = preg_replace('/[\x{00A0}\x{2000}-\x{200B}]/u', ' ', $text);
        $text = preg_replace('/\s+/u', ' ', trim($text));
        return $text;
    }

    
    public static function get_row($object_type, $object_id) {
        global $wpdb;
        self::maybe_install();
        $table = self::table();
        
        return $wpdb->get_row($wpdb->prepare(
            "SELECT text_hash, text_content, word_count, updated_at
               FROM {$table} WHERE object_type = %s AND object_id = %d",
            $object_type, $object_id
        ));
    }
}