<?php
/**
 * Pro module — robots.txt management for AI crawlers.
 *
 * @fs_premium_only
 *
 * Appends per-bot Allow/Disallow rules to WordPress's virtual robots.txt via
 * the core `robots_txt` filter, following the training/citation access policy.
 * A bot whose category the site opted out of (or that is disabled) is served a
 * `Disallow: /`; the rest get an explicit `Allow: /`.
 *
 * NOTE: WordPress only serves (and filters) a *virtual* robots.txt. A physical
 * robots.txt at the web root is returned directly and bypasses these rules.
 *
 * @package ADG_AI_Markdown
 */

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

class ADG_AIM_Robots {

	/**
	 * @var ADG_AIM_Bot_Detector
	 */
	private $detector;

	/**
	 * @param ADG_AIM_Bot_Detector $detector Bot detector (source of the bot list).
	 */
	public function __construct( ADG_AIM_Bot_Detector $detector ) {
		$this->detector = $detector;
	}

	/**
	 * Register hooks.
	 */
	public function hooks() {
		add_filter( 'robots_txt', array( $this, 'filter_robots_txt' ), 20, 2 );
	}

	/**
	 * Append the AI-crawler rules to robots.txt output.
	 *
	 * @param string $output The robots.txt content so far.
	 * @param bool   $public Whether the site is set to be indexed (blog_public).
	 * @return string
	 */
	public function filter_robots_txt( $output, $public ) {
		// Pro-gated, and off by default.
		if ( ! ADG_AIM_Plugin::is_pro() || ! (bool) ADG_AIM_Plugin::get_setting( 'robots_manage', false ) ) {
			return $output;
		}

		// Non-public sites: core already emits a global Disallow; do not fight it.
		if ( ! $public ) {
			return $output;
		}

		$block = $this->build_block();
		if ( '' === $block ) {
			return $output;
		}

		$output = rtrim( (string) $output, "\n" );
		return $output . "\n\n" . $block . "\n";
	}

	/**
	 * Build the AI-crawler rules block (without leading/trailing blank lines).
	 *
	 * @return string
	 */
	public function build_block() {
		$lines = array( '# AI crawler policy — managed by ADG Markdown for AI Crawlers' );
		$seen  = array();
		$rules = 0;

		foreach ( $this->detector->get_bots() as $bot ) {
			$token = ADG_AIM_Bot_Detector::robots_token( $bot );
			if ( '' === $token ) {
				continue;
			}

			$key = strtolower( $token );
			if ( isset( $seen[ $key ] ) ) {
				continue;
			}
			$seen[ $key ] = true;

			$category = ADG_AIM_Bot_Detector::bot_category( $bot );
			$allowed  = ! empty( $bot['enabled'] ) && ADG_AIM_Plugin::category_allowed( $category );

			$lines[] = '';
			$lines[] = 'User-agent: ' . $token;
			$lines[] = $allowed ? 'Allow: /' : 'Disallow: /';
			++$rules;
		}

		if ( 0 === $rules ) {
			return '';
		}

		if ( (bool) ADG_AIM_Plugin::get_setting( 'llms_enabled', true ) ) {
			$lines[] = '';
			$lines[] = '# Structured index for AI: ' . home_url( '/llms.txt' );
		}

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

	/**
	 * Whether a physical robots.txt exists at the web root (bypasses these rules).
	 *
	 * @return bool
	 */
	public static function physical_robots_exists() {
		$path = untrailingslashit( ABSPATH ) . '/robots.txt';
		return @file_exists( $path ); // phpcs:ignore WordPress.PHP.NoSilencedErrors, WordPress.WP.AlternativeFunctions
	}
}