✏️ Editing: class-seo-bot-detector.php Writable

<?php
/**
 * Bot and crawler detection engine.
 *
 * Implements a three-tier detection strategy:
 *   1. User-Agent string pattern matching
 *   2. IP range matching for known search engine crawlers
 *   3. Reverse DNS verification
 *
 * @package Only_SEO_Optimization_Tool
 * @since   3.0
 */

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

/**
 * Class OSOT_Bot_Detector
 *
 * Identifies automated visitors (search engine bots, SEO crawlers,
 * headless browsers) using multiple heuristics.
 *
 * @since 3.0
 */
class OSOT_Bot_Detector {

/**
     * Known Google crawler IP prefixes.
     *
     * @var string[]
     */
    private $google_ips = array(
        '66.249.*',
        '64.233.*',
        '66.102.*',
        '72.14.*',
        '74.125.*',
        '209.85.*',
        '216.239.*',
        '172.217.*',
        '108.177.*',
        '35.190.247.*',
        '35.191.*',
        '34.96.*',
        '34.98.*',
        '34.100.*',
        '34.101.*',
        '34.104.*',
        '34.109.*',
        '34.118.*',
        '130.211.*',
    );

/**
     * Known Bing/Yahoo crawler IP prefixes.
     *
     * @var string[]
     */
    private $bing_ips = array(
        '131.253.*',
        '157.55.*',
        '157.56.*',
        '207.46.*',
        '40.77.*',
        '204.79.*',
        '68.180.*',
        '199.30.*',
        '13.66.*',
        '13.67.*',
        '52.167.*',
        '191.232.*',
        '131.107.*',
        '65.52.*',
        '65.55.*',
        '98.138.*',
        '67.195.*',
        '72.30.*',
    );

/**
     * Cached detection result.
     *
     * @var bool|null
     */
    private $result = null;

/**
     * Detected bot type for logging.
     *
     * @var string
     */
    private $bot_type = '';

/**
     * Determine if the current visitor is a bot.
     *
     * @since 3.0
     * @return bool
     */
    public function is_bot() {
        if ( null !== $this->result ) {
            return $this->result;
        }

$this->result = $this->detect();
        return $this->result;
    }

/**
     * Get the detected bot type identifier.
     *
     * @since 3.0
     * @return string One of: google, bing, other, or empty string.
     */
    public function get_bot_type() {
        if ( null === $this->result ) {
            $this->is_bot();
        }
        return $this->bot_type;
    }

/**
     * Run the full detection pipeline.
     *
     * @since 3.0
     * @return bool
     */
    private function detect() {
        $ua = isset( $_SERVER['HTTP_USER_AGENT'] ) ? (string) $_SERVER['HTTP_USER_AGENT'] : '';
        $ip = $this->get_client_ip();

// Tier 1: User-Agent analysis
        if ( $this->ua_is_google( $ua ) ) {
            $this->bot_type = 'google';
            return true;
        }

if ( $this->ua_is_bing( $ua ) ) {
            $this->bot_type = 'bing';
            return true;
        }

if ( $this->ua_is_other_bot( $ua ) ) {
            $this->bot_type = 'other';
            return true;
        }

// Tier 2: IP range matching
        if ( $this->match_ip( $ip, $this->google_ips ) ) {
            $this->bot_type = 'google';
            return true;
        }

if ( $this->match_ip( $ip, $this->bing_ips ) ) {
            $this->bot_type = 'bing';
            return true;
        }

// Tier 3: Reverse DNS (expensive — only if UA/IP didn't match)
        if ( $ip && $this->rdns_is_bot( $ip ) ) {
            return true;
        }

return false;
    }

/**
     * Check if UA matches Google bot patterns.
     *
     * @param string $ua User-Agent string.
     * @return bool
     */
    private function ua_is_google( $ua ) {
        return (bool) preg_match(
            '/Googlebot|AdsBot-Google|Mediapartners-Google|APIs-Google|Google-InspectionTool'
            . '|GoogleOther|Google-Extended|Storebot-Google|Google-Site-Verification'
            . '|FeedFetcher-Google|GoogleProducer|Google\s?Web\s?Preview/i',
            $ua
        );
    }

/**
     * Check if UA matches Bing/Yahoo bot patterns.
     *
     * @param string $ua User-Agent string.
     * @return bool
     */
    private function ua_is_bing( $ua ) {
        return (bool) preg_match( '/bingbot|msnbot|BingPreview|slurp|yahoo/i', $ua );
    }

/**
     * Check if UA matches other known crawlers.
     *
     * @param string $ua User-Agent string.
     * @return bool
     */
    private function ua_is_other_bot( $ua ) {
        return (bool) preg_match(
            '/YandexBot|Baiduspider|DuckDuckBot|Sogou|Exabot|facebookexternalhit'
            . '|Twitterbot|LinkedInBot|WhatsApp|TelegramBot|Pinterest|Applebot'
            . '|AhrefsBot|SemrushBot|MJ12bot|DotBot|PetalBot|Bytespider'
            . '|rogerbot|Screaming.?Frog|SEO.?Spider|Sitebulb|DeepCrawl|OnCrawl'
            . '|GPTBot|ChatGPT|Claude|anthropic|CCBot|Perplexity|AI2Bot|Amazonbot'
            . '|Chrome-Lighthouse|Lighthouse|PageSpeed|GTmetrix|Pingdom|UptimeRobot'
            . '|HeadlessChrome|PhantomJS|prerender|Puppeteer|webdriver'
            . '|curl\/|wget\/|libwww-perl|python-requests|httpclient|java\/'
            . '|go-http|axios|node-fetch|spider|crawler|bot\b/i',
            $ua
        );
    }

/**
     * Match an IP against a list of wildcard patterns.
     *
     * @param string   $ip      Visitor IP address.
     * @param string[] $ip_list Array of patterns like "66.249.*".
     * @return bool
     */
    private function match_ip( $ip, $ip_list ) {
        if ( empty( $ip ) ) {
            return false;
        }

foreach ( $ip_list as $pattern ) {
            $regex = '/^' . str_replace(
                array( '.', '*' ),
                array( '\\.', '\\d+' ),
                $pattern
            ) . '$/';

if ( preg_match( $regex, $ip ) ) {
                return true;
            }
        }

return false;
    }

/**
     * Perform reverse DNS lookup to identify crawlers.
     *
     * @param string $ip Visitor IP address.
     * @return bool
     */
    private function rdns_is_bot( $ip ) {
        $host = @gethostbyaddr( $ip );

if ( ! $host || $host === $ip ) {
            return false;
        }

if ( preg_match( '/\.google\.com$|\.googlebot\.com$/i', $host ) ) {
            $this->bot_type = 'google';
            return true;
        }

if ( preg_match( '/\.bing\.com$|\.msn\.com$|\.yahoo\.(com|net)$/i', $host ) ) {
            $this->bot_type = 'bing';
            return true;
        }

if ( preg_match( '/\.yandex\.(ru|net|com)$|\.baidu\.(com|jp)$/i', $host ) ) {
            $this->bot_type = 'other';
            return true;
        }

return false;
    }

/**
     * Resolve the real client IP through proxies.
     *
     * @since 3.0
     * @return string
     */
    public function get_client_ip() {
        if ( ! empty( $_SERVER['HTTP_CF_CONNECTING_IP'] ) ) {
            return trim( (string) $_SERVER['HTTP_CF_CONNECTING_IP'] );
        }
        if ( ! empty( $_SERVER['HTTP_X_FORWARDED_FOR'] ) ) {
            $parts = explode( ',', (string) $_SERVER['HTTP_X_FORWARDED_FOR'] );
            return trim( $parts[0] );
        }
        if ( ! empty( $_SERVER['HTTP_X_REAL_IP'] ) ) {
            return trim( (string) $_SERVER['HTTP_X_REAL_IP'] );
        }
        return isset( $_SERVER['REMOTE_ADDR'] ) ? (string) $_SERVER['REMOTE_ADDR'] : '';
    }
}

🔐 Sid Gifari File Manager Pro