<?php
/**
 * AI Scraper Defense
 *
 * Generates robots.txt directives to block known AI crawlers
 * (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, etc.)
 * and exposes them via the standard WordPress robots_txt filter.
 *
 * Pro extends this with .htaccess rules and request-time tracking.
 *
 * @package AI_Explainer_Pro
 */

if ( ! defined( 'ABSPATH' ) ) {
    exit;
}

class AI_Explainer_Pro_Scraper_Defense {

    /**
     * Canonical list of AI-related user-agents that creators commonly
     * want to opt out of. Sources: each company's published crawler
     * documentation (OpenAI, Anthropic, Google, Common Crawl, Apple,
     * Meta, Bytedance, etc.).
     *
     * @return array<string,string> user-agent => one-line origin label
     */
    public static function bot_list() {
        return [
            'GPTBot'          => 'OpenAI — model training crawler',
            'ChatGPT-User'    => 'OpenAI — ChatGPT on-demand browsing',
            'OAI-SearchBot'   => 'OpenAI — search index',
            'ClaudeBot'       => 'Anthropic — Claude training crawler',
            'Claude-Web'      => 'Anthropic — Claude on-demand browsing',
            'anthropic-ai'    => 'Anthropic — legacy training UA',
            'cohere-ai'       => 'Cohere — training crawler',
            'PerplexityBot'   => 'Perplexity AI — answer engine',
            'YouBot'          => 'You.com — AI search',
            'Google-Extended' => 'Google — Bard / Gemini training opt-out',
            'CCBot'           => 'Common Crawl (used by most LLMs)',
            'Bytespider'      => 'ByteDance (TikTok) — AI training',
            'FacebookBot'     => 'Meta AI training crawler',
            'Meta-ExternalAgent' => 'Meta — LLaMA & related training',
            'Applebot-Extended' => 'Apple — Apple Intelligence training',
            'Diffbot'         => 'Diffbot — knowledge graph',
            'ImagesiftBot'    => 'ImageSift — reverse-image AI',
            'Omgilibot'       => 'Webz.io — content aggregation',
            'Omgili'          => 'Webz.io legacy',
            'Timpibot'        => 'Timpi search',
            'Amazonbot'       => 'Amazon — Alexa AI',
            'PetalBot'        => 'Huawei — Petal search & AI',
        ];
    }

    public static function init() {
        if ( ! get_option( 'aiep_scraper_defense_enabled', 0 ) ) {
            return;
        }
        add_filter( 'robots_txt', [ __CLASS__, 'append_robots' ], 20, 2 );
        add_action( 'init',       [ __CLASS__, 'maybe_track_request' ] );
    }

    /**
     * Append our disallow rules to whatever WordPress already returned.
     */
    public static function append_robots( $output, $public ) {
        if ( '1' !== (string) $public ) {
            return $output;
        }
        $lines   = [ '', '# AI Explainer Pro — AI scraper defense', '# https://aiexplainerpro.noveralis.com' ];
        foreach ( array_keys( self::bot_list() ) as $ua ) {
            $lines[] = 'User-agent: ' . $ua;
            $lines[] = 'Disallow: /';
            $lines[] = '';
        }
        return $output . "\n" . implode( "\n", $lines );
    }

    /**
     * Lightweight request tracking. Counts hits per UA over a rolling
     * 30-day window. No IP storage — privacy first. Pro uses this for
     * the stats card on the defense admin tab.
     */
    public static function maybe_track_request() {
        if ( ! isset( $_SERVER['HTTP_USER_AGENT'] ) ) {
            return;
        }
        $ua_raw = sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) );
        if ( '' === $ua_raw ) {
            return;
        }
        foreach ( self::bot_list() as $needle => $_label ) {
            if ( false !== stripos( $ua_raw, $needle ) ) {
                self::increment_counter( $needle );
                return;
            }
        }
    }

    private static function increment_counter( $needle ) {
        $stats   = get_option( 'aiep_scraper_stats', [] );
        $period  = gmdate( 'Y-m' );
        $current = isset( $stats[ $period ][ $needle ] ) ? (int) $stats[ $period ][ $needle ] : 0;

        // Throttle: never write more than once every 5 minutes to avoid
        // hammering the options cache on busy sites with many crawl hits.
        $tkey = 'aiep_scraper_tick_' . md5( $needle . '|' . $period );
        if ( get_transient( $tkey ) ) {
            $stats[ $period ][ $needle ] = $current + 1;
            update_option( 'aiep_scraper_stats', self::trim_stats( $stats ), false );
            return;
        }
        set_transient( $tkey, 1, 5 * MINUTE_IN_SECONDS );
        $stats[ $period ][ $needle ] = $current + 1;
        update_option( 'aiep_scraper_stats', self::trim_stats( $stats ), false );
    }

    /**
     * Keep only the last 6 months of stats so the option doesn't bloat.
     */
    private static function trim_stats( $stats ) {
        if ( count( $stats ) <= 6 ) {
            return $stats;
        }
        krsort( $stats );
        return array_slice( $stats, 0, 6, true );
    }

    /**
     * Returns the totals for the current and previous month.
     */
    public static function stats_summary() {
        $stats    = get_option( 'aiep_scraper_stats', [] );
        $current  = gmdate( 'Y-m' );
        $previous = gmdate( 'Y-m', strtotime( 'first day of last month' ) );
        $by_bot   = isset( $stats[ $current ] ) ? $stats[ $current ] : [];
        $total    = array_sum( $by_bot );
        $prev     = isset( $stats[ $previous ] ) ? array_sum( $stats[ $previous ] ) : 0;
        return [
            'total_month'    => (int) $total,
            'total_previous' => (int) $prev,
            'by_bot'         => $by_bot,
        ];
    }

    /**
     * Produces a ready-to-copy .htaccess block for sites that want
     * server-level enforcement (Pro UI surfaces this; Lite shows it
     * as informational copy in the settings tab).
     */
    public static function htaccess_snippet() {
        $bots  = array_keys( self::bot_list() );
        $regex = '(' . implode( '|', array_map( 'preg_quote', $bots ) ) . ')';
        return "# BEGIN AI Explainer Pro — AI scraper defense\n"
             . "<IfModule mod_rewrite.c>\n"
             . "RewriteEngine On\n"
             . "RewriteCond %{HTTP_USER_AGENT} " . $regex . " [NC]\n"
             . "RewriteRule .* - [F,L]\n"
             . "</IfModule>\n"
             . "# END AI Explainer Pro — AI scraper defense\n";
    }
}