<?php
/**
 * Full-page-cache bypass for AI crawlers.
 *
 * The problem this solves: page-cache plugins (WP Rocket, W3 Total Cache, WP Super
 * Cache, LiteSpeed Cache) serve cached HTML from an `advanced-cache.php` drop-in and
 * `exit()` *before* WordPress loads. Aiso_Tracker is hooked on `init`, which never
 * runs on a cache HIT — so a real AI-crawler request to an already-cached page is
 * silently lost. (Confirmed on caddooffices.com with WP Rocket 3.21: three GPTBot
 * requests to a warm-cached "/" produced zero rows; the same UA on an uncached URL
 * records fine.)
 *
 * The fix: tell each cache layer to treat known AI user-agents as uncacheable, so
 * those requests fall through to PHP and the *existing* tracker records them exactly
 * once. The UA list is sourced from Aiso_Bots::ua_patterns() (the same directory used
 * for detection) so the two can never drift. Human traffic is completely untouched:
 * non-AI requests are still served from cache as before.
 *
 * Nothing here writes a hit or changes verification — it only removes the cache
 * barrier so the normal path (Aiso_Tracker on `init` -> Aiso_Store::insert_hit) can
 * fire. That guarantees one row per request and identical verified/"—"/spoof states.
 *
 * @package Aiso_AI_Crawler_Analytics
 */

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

class Aiso_Cache {

	public function __construct() {
		// WP Rocket: append our UA regexes to its cache-reject list. Rocket consults
		// this from its advanced-cache drop-in, so matching bots bypass the page cache
		// on both read (HIT) and write. Harmless when WP Rocket is not installed.
		add_filter( 'rocket_cache_reject_ua', array( $this, 'rocket_reject_ua' ) );

		// W3TC / WP Super Cache / LiteSpeed: emit a per-request "do not cache" signal
		// when the current request is a known AI bot. Hooked at init:0 — after all
		// plugins have loaded (so the cache plugin's listeners exist) but well before
		// the response is buffered and stored.
		add_action( 'init', array( $this, 'signal_nocache' ), 0 );

		// Keep WP Rocket's baked config in sync whenever our options change, so a new
		// UA (or toggling the setting) takes effect without a manual cache clear.
		add_action( 'update_option_' . Aiso_Settings::OPTION, array( __CLASS__, 'refresh_cache_rules' ) );
	}

	/**
	 * Append AI-crawler UA patterns to WP Rocket's reject-UA list.
	 *
	 * @param array $uas Existing reject-UA fragments.
	 * @return array
	 */
	public function rocket_reject_ua( $uas ) {
		if ( ! $this->bypass_enabled() ) {
			return is_array( $uas ) ? $uas : array();
		}
		$uas = is_array( $uas ) ? $uas : array();
		return array_merge( $uas, Aiso_Bots::ua_patterns() );
	}

	/**
	 * For cache plugins without a UA-reject hook, signal "don't cache this request"
	 * when the current visitor is a known AI crawler.
	 */
	public function signal_nocache() {
		if ( ! $this->should_bypass() ) {
			return;
		}

		// W3 Total Cache, WP Super Cache, and many other layers honour this constant
		// to skip storing the current response.
		if ( ! defined( 'DONOTCACHEPAGE' ) ) {
			define( 'DONOTCACHEPAGE', true );
		}

		// LiteSpeed Cache: explicit no-cache control for the current request.
		do_action( 'litespeed_control_set_nocache', 'aiso ai bot' );
	}

	/**
	 * Regenerate WP Rocket's baked config so the reject-UA list is live in its
	 * advanced-cache drop-in. No-op when WP Rocket is absent. Called on settings
	 * save and on plugin activation.
	 */
	public static function refresh_cache_rules() {
		if ( function_exists( 'rocket_generate_config_file' ) ) {
			rocket_generate_config_file();
		}
	}

	/**
	 * Is the cache-bypass feature enabled? (Default on.)
	 *
	 * @return bool
	 */
	private function bypass_enabled() {
		return (bool) Aiso_Analytics::option( 'bypass_cache_for_ai_bots', true );
	}

	/**
	 * Should the current request bypass the page cache? Mirrors Aiso_Tracker's
	 * should_track() exclusions (admin, cron, ajax, REST, XML-RPC, CLI; GET/HEAD
	 * only) and additionally requires a known AI user-agent, so caching is never
	 * disabled for human traffic.
	 *
	 * @return bool
	 */
	private function should_bypass() {
		if ( ! $this->bypass_enabled() ) {
			return false;
		}
		if ( is_admin() ) {
			return false;
		}
		if ( function_exists( 'wp_doing_cron' ) && wp_doing_cron() ) {
			return false;
		}
		if ( function_exists( 'wp_doing_ajax' ) && wp_doing_ajax() ) {
			return false;
		}
		if ( defined( 'WP_CLI' ) && WP_CLI ) {
			return false;
		}
		if ( defined( 'XMLRPC_REQUEST' ) && XMLRPC_REQUEST ) {
			return false;
		}
		if ( defined( 'REST_REQUEST' ) && REST_REQUEST ) {
			return false;
		}

		$method = isset( $_SERVER['REQUEST_METHOD'] ) ? strtoupper( sanitize_text_field( wp_unslash( $_SERVER['REQUEST_METHOD'] ) ) ) : 'GET';
		if ( ! in_array( $method, array( 'GET', 'HEAD' ), true ) ) {
			return false;
		}

		$user_agent = isset( $_SERVER['HTTP_USER_AGENT'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) : '';
		if ( '' === $user_agent ) {
			return false;
		}

		return Aiso_Bots::is_ai_user_agent( $user_agent );
	}
}