<?php
/**
 * AI-aware robots.txt management.
 *
 * Adds granular control over AI training bots vs retrieval/search bots.
 * Best practice 2026: Allow retrieval bots, optionally block training bots.
 *
 * @since 3.0.0
 */

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

class AIREWP_Robots_Txt {

    /**
     * AI training bots (scrape for model training).
     * Default: BLOCKED (opt-in to allow).
     */
    public const TRAINING_BOTS = [
        'GPTBot'             => 'OpenAI (training)',
        'ClaudeBot'          => 'Anthropic (training)',
        'Google-Extended'    => 'Google (training)',
        'Meta-ExternalAgent' => 'Meta (training)',
        'Bytespider'         => 'ByteDance (training)',
        'CCBot'              => 'Common Crawl (training)',
        'Applebot-Extended'  => 'Apple Intelligence (training)',
        'Amazonbot'          => 'Amazon (training)',
        'Diffbot'            => 'Diffbot (training)',
        'Omgilibot'          => 'Omgili (training)',
        'FacebookBot'        => 'Meta (training)',
        'ImagesiftBot'       => 'Imagesift (training)',
        'Cohere-ai'          => 'Cohere (training)',
    ];

    /**
     * AI retrieval/search bots (real-time answers, search results).
     * Default: ALLOWED (opt-out to block).
     */
    public const RETRIEVAL_BOTS = [
        'OAI-SearchBot'    => 'OpenAI Search',
        'ChatGPT-User'     => 'ChatGPT (browsing)',
        'Claude-SearchBot' => 'Anthropic Search',
        'Claude-User'      => 'Claude (browsing)',
        'PerplexityBot'    => 'Perplexity AI',
        'Perplexity-User'  => 'Perplexity (browsing)',
        'GoogleOther'      => 'Google AI',
    ];

    public function __construct() {
        $settings = AI_Ready_WP::get_settings();
        if ( ! empty( $settings['enable_robots_ai'] ) ) {
            add_filter( 'robots_txt', [ $this, 'modify_robots_txt' ], 100, 2 );
        }
    }

    /**
     * Add AI bot directives to WordPress virtual robots.txt.
     */
    public function modify_robots_txt( string $output, bool $public ): string {
        if ( ! $public ) {
            return $output;
        }

        $settings = AI_Ready_WP::get_settings();

        // Determine which bots to block
        $block_training  = ! empty( $settings['robots_block_training'] );
        $block_retrieval = ! empty( $settings['robots_block_retrieval'] );

        $lines = [];
        $lines[] = '';
        $lines[] = '# AI Bot Directives (managed by AI-Ready WP)';

        // Training bots
        if ( $block_training ) {
            $allowed_training = self::get_allowed_list( $settings, 'robots_allow_training' );
            foreach ( self::TRAINING_BOTS as $bot => $label ) {
                if ( in_array( $bot, $allowed_training, true ) ) {
                    continue; // Explicitly allowed
                }
                $lines[] = 'User-agent: ' . $bot;
                $lines[] = 'Disallow: /';
                $lines[] = '';
            }
        }

        // Retrieval bots (only block if explicitly set)
        if ( $block_retrieval ) {
            $allowed_retrieval = self::get_allowed_list( $settings, 'robots_allow_retrieval' );
            foreach ( self::RETRIEVAL_BOTS as $bot => $label ) {
                if ( in_array( $bot, $allowed_retrieval, true ) ) {
                    continue;
                }
                $lines[] = 'User-agent: ' . $bot;
                $lines[] = 'Disallow: /';
                $lines[] = '';
            }
        }

        // AI discovery pointers (always add)
        $lines[] = '# AI Discovery Files';
        $lines[] = 'Sitemap: ' . home_url( '/sitemap.xml' );

        // Without a working filesystem we cannot tell which files exist, so
        // robots.txt simply does not advertise them rather than fataling.
        $fs = AI_Ready_WP::filesystem();
        if ( null !== $fs && $fs->exists( AIREWP_ROOT_PATH . '.well-known/agents.json' ) ) {
            $lines[] = '# agents.json: ' . home_url( '/.well-known/agents.json' );
        }
        if ( null !== $fs && $fs->exists( AIREWP_ROOT_PATH . 'llms.txt' ) ) {
            $lines[] = '# llms.txt: ' . home_url( '/llms.txt' );
        }

        return $output . implode( "\n", $lines ) . "\n";
    }

    /**
     * Get the list of explicitly allowed bots from a comma-separated setting.
     */
    private static function get_allowed_list( array $settings, string $key ): array {
        $value = $settings[ $key ] ?? '';
        if ( empty( $value ) ) {
            return [];
        }
        return array_map( 'trim', explode( ',', $value ) );
    }

    /**
     * Get all known AI bot user-agent strings for analytics matching.
     */
    public static function get_all_bot_agents(): array {
        return array_merge(
            array_keys( self::TRAINING_BOTS ),
            array_keys( self::RETRIEVAL_BOTS )
        );
    }
}