<?php

/**
 * Frontend tracker.
 *
 * @package A360_Insights
 */

defined('ABSPATH') || exit;

/**
 * Tracks post views.
 */
class A360_Insights_Tracker
{

	/**
	 * Init hooks.
	 *
	 * @return void
	 */
	public function init()
	{
		add_action('template_redirect', array($this, 'maybe_track'), 99);
	}

	/**
	 * Track single public post view.
	 *
	 * @return void
	 */
	public function maybe_track()
	{
		static $tracked = false;

		if ($tracked) {
			return;
		}

		if (is_admin() || wp_doing_ajax() || is_preview() || is_feed() || is_robots()) {
			return;
		}

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

		if (! is_singular()) {
			return;
		}

		$post_id = get_queried_object_id();

		if (! $post_id || 'publish' !== get_post_status($post_id)) {
			return;
		}

		if (is_user_logged_in()) {
			$settings = A360_Insights_Settings::get();

			if (! empty($settings['exclude_logged_in_users'])) {
				return;
			}

			if (! empty($settings['exclude_administrators']) && current_user_can('manage_options')) {
				return;
			}
		}

		if (! A360_Insights_Settings::should_track_post($post_id)) {
			return;
		}

		A360_Insights_DB::insert_view($post_id, $this->source_data());
		$tracked = true;
	}


	/**
	 * Determine whether the current request should be ignored before tracking.
	 *
	 * @return bool
	 */
	private function should_ignore_request()
	{
		$settings = A360_Insights_Settings::get();

		if (! empty($settings['exclude_cron']) && wp_doing_cron()) {
			return true;
		}

		if (! empty($settings['exclude_cli']) && defined('WP_CLI') && WP_CLI) {
			return true;
		}

		if (! empty($settings['ignore_prefetch']) && $this->is_prefetch_request()) {
			return true;
		}

		if (! empty($settings['exclude_bots']) && $this->is_bot_request()) {
			return true;
		}

		return false;
	}

	/**
	 * Detect common browser prefetch, prerender and preview requests.
	 *
	 * @return bool
	 */
	private function is_prefetch_request()
	{
		$headers = array(
			'HTTP_X_MOZ',
			'HTTP_PURPOSE',
			'HTTP_SEC_PURPOSE',
			'HTTP_X_PURPOSE',
		);

		foreach ($headers as $header) {
			if (empty($_SERVER[$header])) {
				continue;
			}

			$value = strtolower(sanitize_text_field(wp_unslash($_SERVER[$header])));

			if (false !== strpos($value, 'prefetch') || false !== strpos($value, 'prerender') || false !== strpos($value, 'preview')) {
				return true;
			}
		}

		return false;
	}

	/**
	 * Detect common bots, crawlers, link previewers and uptime monitors.
	 *
	 * @return bool
	 */
	private function is_bot_request()
	{
		if (empty($_SERVER['HTTP_USER_AGENT'])) {
			return false;
		}

		$user_agent = strtolower(sanitize_text_field(wp_unslash($_SERVER['HTTP_USER_AGENT'])));

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

		$patterns = array(
			'bot',
			'crawler',
			'spider',
			'crawl',
			'slurp',
			'googlebot',
			'bingbot',
			'duckduckbot',
			'baiduspider',
			'yandex',
			'sogou',
			'exabot',
			'facebot',
			'facebookexternalhit',
			'linkedinbot',
			'twitterbot',
			'pinterest',
			'whatsapp',
			'telegrambot',
			'discordbot',
			'skypeuripreview',
			'vkshare',
			'ahrefs',
			'semrush',
			'mj12bot',
			'dotbot',
			'petalbot',
			'bytespider',
			'uptimerobot',
			'statuscake',
			'pingdom',
			'gtmetrix',
			'pagespeed',
			'lighthouse',
			'headlesschrome',
			'phantomjs',
			'python-requests',
			'curl/',
			'wget/',
			'httpclient',
		);

		foreach ($patterns as $pattern) {
			if (false !== strpos($user_agent, $pattern)) {
				return true;
			}
		}

		return false;
	}


	/**
	 * Source data.
	 *
	 * @return array
	 */
	private function source_data()
	{
		$settings = A360_Insights_Settings::get();

		// phpcs:ignore WordPress.Security.NonceVerification.Recommended -- Public read-only UTM tracking parameter.
		$utm_source = isset($_GET['utm_source']) && ! empty($settings['store_utm_parameters']) ? sanitize_text_field(wp_unslash($_GET['utm_source'])) : '';

		// phpcs:ignore WordPress.Security.NonceVerification.Recommended -- Public read-only UTM tracking parameter.
		$utm_medium = isset($_GET['utm_medium']) && ! empty($settings['store_utm_parameters']) ? sanitize_text_field(wp_unslash($_GET['utm_medium'])) : '';

		// phpcs:ignore WordPress.Security.NonceVerification.Recommended -- Public read-only UTM tracking parameter.
		$utm_campaign = isset($_GET['utm_campaign']) && ! empty($settings['store_utm_parameters']) ? sanitize_text_field(wp_unslash($_GET['utm_campaign'])) : '';

		$referrer_url  = isset($_SERVER['HTTP_REFERER']) && ! empty($settings['store_referrer_details'])
			? esc_url_raw(wp_unslash($_SERVER['HTTP_REFERER']))
			: '';

		$referrer_host = $referrer_url
			? wp_parse_url($referrer_url, PHP_URL_HOST)
			: '';

		$referrer_host = is_string($referrer_host)
			? strtolower(sanitize_text_field($referrer_host))
			: '';

		if ($referrer_host) {
			$source = $this->classify_referrer($referrer_host, $referrer_url);
		} elseif ($utm_source) {
			$source = $this->classify_source_value($utm_source);
		} else {
			$source = 'direct';
		}
		if ('internal' === $source) {
			$source        = 'direct';
			$referrer_host = '';
			$referrer_url  = '';
		}

		return array(
			'source'        => $source,
			'referrer_host' => $referrer_host,
			'referrer_url'  => $referrer_url,
			'utm_source'    => $utm_source,
			'utm_medium'    => $utm_medium,
			'utm_campaign'  => $utm_campaign,
		);
	}
	private function classify_source_value($value)
	{
		$value = strtolower(sanitize_text_field($value));

		if (false !== strpos($value, 'codex')) {
			return 'codex';
		}

		if (false !== strpos($value, 'chatgpt') || false !== strpos($value, 'openai')) {
			return 'chatgpt';
		}

		if (false !== strpos($value, 'gemini') || false !== strpos($value, 'bard')) {
			return 'gemini';
		}

		if (false !== strpos($value, 'copilot')) {
			return 'copilot';
		}

		if (false !== strpos($value, 'perplexity')) {
			return 'perplexity';
		}

		if (false !== strpos($value, 'claude') || false !== strpos($value, 'anthropic')) {
			return 'claude';
		}
		if (false !== strpos($value, 'meta.ai') || false !== strpos($value, 'meta_ai')) {
			return 'meta_ai';
		}

		if (false !== strpos($value, 'mistral') || false !== strpos($value, 'lechat')) {
			return 'mistral';
		}

		if (false !== strpos($value, 'you.com') || 'you' === $value) {
			return 'you';
		}

		if (false !== strpos($value, 'poe')) {
			return 'poe';
		}

		if (false !== strpos($value, 'phind')) {
			return 'phind';
		}

		if (false !== strpos($value, 'deepseek')) {
			return 'deepseek';
		}

		if (false !== strpos($value, 'grok') || false !== strpos($value, 'x.ai')) {
			return 'grok';
		}
		if (false !== strpos($value, 'google')) {
			return 'google';
		}

		if (false !== strpos($value, 'bing')) {
			return 'bing';
		}

		return sanitize_key($value);
	}
	/**
	 * Classify referrer host.
	 *
	 * @param string $host Host.
	 * @param string $url  Full referrer URL.
	 * @return string
	 */
	private function classify_referrer($host, $url = '')
	{
		$site_host = wp_parse_url(home_url(), PHP_URL_HOST);

		$site_host = is_string($site_host)
			? strtolower(preg_replace('/^www\./', '', $site_host))
			: '';

		$host = strtolower(preg_replace('/^www\./', '', $host));
		$path = $url ? wp_parse_url($url, PHP_URL_PATH) : '';
		$path = is_string($path) ? strtolower($path) : '';

		if (
			$site_host &&
			(
				$host === $site_host ||
				substr($host, -1 * (strlen($site_host) + 1)) === '.' . $site_host
			)
		) {
			return 'internal';
		}
		if (
			preg_match('/^codex\.openai\.com$/', $host) ||
			(
				preg_match('/(^|\.)chatgpt\.com$|(^|\.)openai\.com$/', $host) &&
				false !== strpos($path, 'codex')
			)
		) {
			return 'codex';
		}

		if (preg_match('/(^|\.)chatgpt\.com$|(^|\.)openai\.com$/', $host)) {
			return 'chatgpt';
		}

		if (preg_match('/^gemini\.google\.com$|^bard\.google\.com$/', $host)) {
			return 'gemini';
		}

		if (preg_match('/^copilot\.microsoft\.com$/', $host)) {
			return 'copilot';
		}

		if (preg_match('/(^|\.)perplexity\.ai$/', $host)) {
			return 'perplexity';
		}

		if (preg_match('/(^|\.)claude\.ai$|(^|\.)anthropic\.com$/', $host)) {
			return 'claude';
		}
		if (preg_match('/(^|\.)meta\.ai$/', $host)) {
			return 'meta_ai';
		}

		if (preg_match('/(^|\.)mistral\.ai$|(^|\.)chat\.mistral\.ai$/', $host)) {
			return 'mistral';
		}

		if (preg_match('/(^|\.)you\.com$/', $host)) {
			return 'you';
		}

		if (preg_match('/(^|\.)poe\.com$/', $host)) {
			return 'poe';
		}

		if (preg_match('/(^|\.)phind\.com$/', $host)) {
			return 'phind';
		}

		if (preg_match('/(^|\.)deepseek\.com$/', $host)) {
			return 'deepseek';
		}

		if (preg_match('/(^|\.)grok\.com$|(^|\.)x\.ai$/', $host)) {
			return 'grok';
		}
		if (preg_match('/(^|\.)google\./', $host)) {
			return 'google';
		}

		if (preg_match('/(^|\.)facebook\.com$|(^|\.)fb\.com$|(^|\.)l\.facebook\.com$|(^|\.)m\.facebook\.com$/', $host)) {
			return 'facebook';
		}

		if (preg_match('/(^|\.)youtube\.com$|(^|\.)youtu\.be$/', $host)) {
			return 'youtube';
		}

		if (preg_match('/instagram\.com|x\.com|twitter\.com|linkedin\.com|tiktok\.com|pinterest\./', $host)) {
			return 'social';
		}

		return 'referral';
	}
}