<?php
/**
 * PHP request gate.
 *
 * @package AIBlackout
 */

namespace AIBlackout\Enforcement;

use AIBlackout\Crawlers\UserAgentMatcher;

defined( 'ABSPATH' ) || exit;

/**
 * Returns 403 to blocked crawlers from within PHP.
 *
 * @since 1.0.0
 */
class PhpGate {

	/**
	 * User-agent matcher.
	 *
	 * @var UserAgentMatcher
	 */
	private $matcher;

	/**
	 * Constructor.
	 *
	 * @param UserAgentMatcher $matcher User-agent matcher.
	 */
	public function __construct( UserAgentMatcher $matcher ) {
		$this->matcher = $matcher;
	}

	/**
	 * Registers the hook.
	 *
	 * @return void
	 */
	public function register(): void {
		add_action( 'init', array( $this, 'maybe_block' ), 1 );
	}

	/**
	 * Blocks the current request when it comes from a blocked crawler.
	 *
	 * @return void
	 */
	public function maybe_block(): void {
		if ( ! $this->should_run() ) {
			return;
		}

		if ( $this->is_declaration_request() ) {
			return;
		}

		$token = $this->matcher->match_agent( $this->get_user_agent() );

		if ( null === $token ) {
			return;
		}

		$this->send_block_response( $token );
	}

	/**
	 * Whether the request is for one of the declaration endpoints.
	 *
	 * @return bool True when the request must not be blocked.
	 */
	private function is_declaration_request(): bool {
		if ( empty( $_SERVER['REQUEST_URI'] ) ) {
			return false;
		}

		$path = (string) wp_parse_url(
			sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ) ),
			PHP_URL_PATH
		);

		if ( '' === $path ) {
			return false;
		}

		$robots     = (string) wp_parse_url( home_url( '/robots.txt' ), PHP_URL_PATH );
		$well_known = (string) wp_parse_url( home_url( '/.well-known/' ), PHP_URL_PATH );

		if ( '' !== $robots && untrailingslashit( $path ) === untrailingslashit( $robots ) ) {
			return true;
		}

		return '' !== $well_known && 0 === strpos( $path, $well_known );
	}

	/**
	 * Whether this layer should evaluate the current request.
	 *
	 * @return bool True when the request is a candidate for blocking.
	 */
	private function should_run(): bool {
		if ( aiblackout_is_disabled() ) {
			return false;
		}

		// Never interfere with the admin, cron, CLI or a logged-in user.
		if ( is_admin() || wp_doing_cron() || ( function_exists( 'wp_doing_ajax' ) && wp_doing_ajax() ) ) {
			return false;
		}

		if ( defined( 'WP_CLI' ) && WP_CLI ) {
			return false;
		}

		if ( is_user_logged_in() ) {
			return false;
		}

		return true;
	}

	/**
	 * Returns the request's User-Agent header.
	 *
	 * @return string User agent, or an empty string.
	 */
	private function get_user_agent(): string {
		if ( empty( $_SERVER['HTTP_USER_AGENT'] ) ) {
			return '';
		}

		return sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) );
	}

	/**
	 * Sends the 403 response and stops.
	 *
	 * @param string $token The crawler token that matched.
	 * @return void
	 */
	private function send_block_response( string $token ): void {
		if ( ! headers_sent() ) {
			status_header( 403 );
			header( 'X-Robots-Tag: noai, noimageai', true );
			header( 'Content-Type: text/plain; charset=utf-8' );
			nocache_headers();
		}

		/**
		 * Fires when the PHP layer refuses a request from an AI crawler.
		 *
		 * @since 1.0.0
		 *
		 * @param string $token The crawler token that matched.
		 */
		do_action( 'aiblackout_crawler_blocked', $token );

		echo esc_html__( 'This site does not permit AI crawling. Your content is not training data.', 'ai-blackout' );

		exit;
	}
}