<?php

namespace AgoLab\Legal;

defined( 'ABSPATH' ) || exit;

class CookieScanner {

    /** Heuristic classification of cookies by name pattern. */
    private const PATTERNS = [
        'analytics'  => [ '_ga', '_gid', '_gat', '__utm', '_clck', '_clsk', 'AMP_', 'mp_', 'mixpanel', '_pk_', 'piwik', '_hjSession', '_hjAbsoluteSessionInProgress', 'hubspotutk' ],
        'marketing'  => [ '_fbp', '_fbc', 'fbsr_', 'IDE', 'NID', 'DSID', 'FLC', 'AID', 'TAID', 'gads', 'GPS', 'YSC', 'VISITOR_INFO', '__qca', '_uetsid', '_uetvid' ],
        'functional' => [ 'lang', 'locale', 'theme', 'sidebar', 'tz', 'cookieconsent', 'darkmode', 'i18n' ],
        'necessary'  => [ 'PHPSESSID', 'wordpress_logged_in', 'wp-settings-', 'wordpress_test_cookie', 'wp_lang', 'comment_author_', 'XSRF-TOKEN', 'csrf_token', '__cf', 'cf_clearance' ],
    ];

    /** Scan home URL, return array of cookies seen. */
    public static function scan( string $url = '' ): array {
        $url = $url ?: home_url( '/' );
        $args = [
            'timeout'     => 15,
            'redirection' => 3,
            'sslverify'   => false,
            'cookies'     => [],
            'headers'     => [ 'User-Agent' => 'aGo Legal Cookie Scanner/1.0' ],
        ];
        $response = wp_remote_get( $url, $args );
        if ( is_wp_error( $response ) ) {
            return [ 'ok' => false, 'error' => $response->get_error_message() ];
        }

        $cookies   = wp_remote_retrieve_cookies( $response );
        $body      = (string) wp_remote_retrieve_body( $response );
        $scanned   = [];

        // Headers Set-Cookie
        foreach ( $cookies as $c ) {
            $name = $c->name ?? '';
            if ( ! $name ) continue;
            $scanned[ $name ] = [
                'name'        => $name,
                'category'    => self::classify( $name ),
                'source'      => 'header',
                'description' => self::describe( $name ),
            ];
        }

        // Heuristic: scan body for "document.cookie =" or "Cookies.set('"
        if ( preg_match_all( "/document\\.cookie\\s*=\\s*['\"]([a-zA-Z0-9_\\-]+)=/", $body, $m ) ) {
            foreach ( $m[1] as $name ) {
                if ( ! isset( $scanned[ $name ] ) ) {
                    $scanned[ $name ] = [
                        'name'        => $name,
                        'category'    => self::classify( $name ),
                        'source'      => 'inline-js',
                        'description' => self::describe( $name ),
                    ];
                }
            }
        }
        // Detect third-party scripts
        $third_party = [];
        if ( preg_match_all( '/<script[^>]+src=["\']([^"\']+)["\']/i', $body, $m ) ) {
            foreach ( $m[1] as $src ) {
                $host = wp_parse_url( $src, PHP_URL_HOST );
                if ( $host && $host !== wp_parse_url( home_url(), PHP_URL_HOST ) ) {
                    $third_party[ $host ] = ( $third_party[ $host ] ?? 0 ) + 1;
                }
            }
        }

        return [
            'ok'           => true,
            'url'          => $url,
            'scanned_at'   => current_time( 'mysql' ),
            'cookies'      => array_values( $scanned ),
            'third_party'  => $third_party,
        ];
    }

    public static function classify( string $name ): string {
        foreach ( self::PATTERNS as $cat => $patterns ) {
            foreach ( $patterns as $p ) {
                if ( stripos( $name, $p ) !== false ) return $cat;
            }
        }
        return 'unclassified';
    }

    public static function describe( string $name ): string {
        $known = [
            '_ga'                 => 'Google Analytics - identificador de usuario',
            '_gid'                => 'Google Analytics - identificador de sesión',
            '_gat'                => 'Google Analytics - control de rate',
            '_fbp'                => 'Meta Pixel - identificador de navegador',
            '_fbc'                => 'Meta Pixel - identificador de click',
            'PHPSESSID'           => 'Sesión PHP - mantiene el estado del usuario en el servidor',
            'wordpress_logged_in' => 'WordPress - identifica usuario autenticado',
            '_hjSession'          => 'Hotjar - sesión de heatmap',
            '__cf_bm'             => 'Cloudflare Bot Management - prevención de bots',
            'cf_clearance'        => 'Cloudflare - protección anti-bot',
        ];
        foreach ( $known as $k => $desc ) {
            if ( stripos( $name, $k ) === 0 ) return $desc;
        }
        return '';
    }

    /** Save scan result as the active cookie table. */
    public static function persist_as_cookie_table( array $scan ): void {
        if ( empty( $scan['cookies'] ) ) return;
        $entries = [];
        foreach ( $scan['cookies'] as $c ) {
            $entries[] = [
                'name'        => $c['name'],
                'category'    => $c['category'],
                'description' => $c['description'],
                'duration'    => '',
            ];
        }
        $settings = get_option( 'agolegal_settings', [] );
        $settings['cookie_table'] = wp_json_encode( $entries );
        update_option( 'agolegal_settings', $settings );
    }
}