<?php
/**
 * Pro module — AI referral tracking: human visits that arrive from an AI assistant.
 *
 * @fs_premium_only
 *
 *
 * When a real person clicks a link inside a ChatGPT / Perplexity / Gemini /
 * Copilot / Claude answer and lands on the site, the browser sends a Referer
 * header pointing at that assistant. This module recognizes those hosts and
 * keeps a compact, privacy-preserving aggregate: totals per source, a per-day
 * series and the top landing pages. No IP, no full referrer URL, no PII.
 *
 * This is distinct from bot analytics (ADG_AIM_Analytics), which counts the
 * crawlers themselves. Referrals count the humans the crawlers send back.
 *
 * @package ADG_AI_Markdown
 */

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

class ADG_AIM_Referrals {

	/** Days kept in the by_day breakdown. */
	const DAYS_KEEP = 90;

	/** Soft cap on stored landing pages before trimming to the top ones. */
	const PAGES_CAP = 200;

	/** Target size after trimming landing pages. */
	const PAGES_TARGET = 100;

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

	/**
	 * @param ADG_AIM_Bot_Detector $detector Bot detector (to skip crawler hits).
	 */
	public function __construct( ADG_AIM_Bot_Detector $detector ) {
		$this->detector = $detector;
	}

	/**
	 * Register hooks. Runs after the negotiator (which exits for served bots).
	 */
	public function hooks() {
		add_action( 'template_redirect', array( $this, 'maybe_track' ), 20 );
	}

	/**
	 * Map of AI-assistant referrer hosts → source key + human label.
	 * The host is matched as an exact host or a dot-suffix.
	 *
	 * @return array host => array( key, label ).
	 */
	public static function source_map() {
		$map = array(
			'chatgpt.com'          => array( 'chatgpt', 'ChatGPT' ),
			'chat.openai.com'      => array( 'chatgpt', 'ChatGPT' ),
			'openai.com'           => array( 'chatgpt', 'ChatGPT' ),
			'perplexity.ai'        => array( 'perplexity', 'Perplexity' ),
			'gemini.google.com'    => array( 'gemini', 'Gemini' ),
			'bard.google.com'      => array( 'gemini', 'Gemini' ),
			'copilot.microsoft.com'=> array( 'copilot', 'Copilot' ),
			'claude.ai'            => array( 'claude', 'Claude' ),
			'you.com'              => array( 'other', 'Other AI' ),
			'poe.com'              => array( 'other', 'Other AI' ),
			'phind.com'            => array( 'other', 'Other AI' ),
		);

		/**
		 * Filter the AI-referrer host map.
		 *
		 * @param array $map host => array( source_key, label ).
		 */
		return (array) apply_filters( 'adg_aim_referral_sources', $map );
	}

	/**
	 * Human labels keyed by source key (for rendering/CSV).
	 *
	 * @return array key => label.
	 */
	public static function source_labels() {
		$labels = array();
		foreach ( self::source_map() as $entry ) {
			$labels[ $entry[0] ] = $entry[1];
		}
		return $labels;
	}

	/**
	 * Default aggregate structure.
	 *
	 * @return array
	 */
	public static function defaults() {
		return array(
			'totals'     => array(), // source_key => int.
			'by_day'     => array(), // 'Y-m-d' => int (all sources).
			'by_page'    => array(), // path => int.
			'first_seen' => '',
			'last_seen'  => '',
		);
	}

	/**
	 * Get the aggregate (backfilling missing keys).
	 *
	 * @return array
	 */
	public static function get() {
		$data = get_option( ADG_AIM_OPT_REFERRALS, array() );
		if ( ! is_array( $data ) ) {
			$data = array();
		}
		return wp_parse_args( $data, self::defaults() );
	}

	/**
	 * Reset all referral data.
	 */
	public static function reset() {
		delete_option( ADG_AIM_OPT_REFERRALS );
	}

	/**
	 * Frontend entry point: record a referral if this is a human view from an AI host.
	 */
	public function maybe_track() {
		if ( ! (bool) ADG_AIM_Plugin::get_setting( 'referrals_enabled', true ) ) {
			return;
		}

		// Only real, public GET page views by anonymous visitors.
		if ( is_admin() || is_feed() || is_trackback() || is_robots() || is_404() ) {
			return;
		}
		if ( ( defined( 'REST_REQUEST' ) && REST_REQUEST )
			|| ( defined( 'DOING_AJAX' ) && DOING_AJAX )
			|| ( defined( 'DOING_CRON' ) && DOING_CRON )
			|| ( defined( 'XMLRPC_REQUEST' ) && XMLRPC_REQUEST ) ) {
			return;
		}
		$method = isset( $_SERVER['REQUEST_METHOD'] ) ? strtoupper( sanitize_text_field( wp_unslash( $_SERVER['REQUEST_METHOD'] ) ) ) : 'GET';
		if ( 'GET' !== $method ) {
			return;
		}
		if ( is_user_logged_in() ) {
			return; // Do not count the site's own team navigating.
		}

		// Skip crawlers — this metric is about humans (UA match only, no IP cost).
		if ( false !== $this->detector->match_ua() ) {
			return;
		}

		$referer = isset( $_SERVER['HTTP_REFERER'] ) ? esc_url_raw( wp_unslash( $_SERVER['HTTP_REFERER'] ) ) : '';
		if ( '' === $referer ) {
			return;
		}

		$source = self::match_source( $referer );
		if ( null === $source ) {
			return;
		}

		$this->record( $source, $this->current_path() );
	}

	/**
	 * Match a referrer URL against the AI-source map.
	 *
	 * @param string $referer Referrer URL.
	 * @return string|null Source key or null.
	 */
	public static function match_source( $referer ) {
		$host = wp_parse_url( $referer, PHP_URL_HOST );
		if ( ! is_string( $host ) || '' === $host ) {
			return null;
		}
		$host = strtolower( $host );
		if ( 0 === strpos( $host, 'www.' ) ) {
			$host = substr( $host, 4 );
		}

		// Never count same-site navigation.
		$self = wp_parse_url( home_url( '/' ), PHP_URL_HOST );
		if ( is_string( $self ) && strtolower( ltrim( $self, 'www.' ) ) === $host ) {
			return null;
		}

		foreach ( self::source_map() as $needle => $entry ) {
			$needle = strtolower( $needle );
			if ( $host === $needle || substr( $host, - ( strlen( $needle ) + 1 ) ) === '.' . $needle ) {
				return $entry[0];
			}
		}
		return null;
	}

	/**
	 * The current request path (no query string, capped length).
	 *
	 * @return string
	 */
	private function current_path() {
		$uri = isset( $_SERVER['REQUEST_URI'] )
			? esc_url_raw( wp_unslash( $_SERVER['REQUEST_URI'] ) )
			: '/';
		$path = (string) wp_parse_url( $uri, PHP_URL_PATH );
		if ( '' === $path ) {
			$path = '/';
		}
		if ( strlen( $path ) > 300 ) {
			$path = substr( $path, 0, 300 );
		}
		return $path;
	}

	/**
	 * Record one referral into the aggregate.
	 *
	 * @param string $source Source key.
	 * @param string $path   Landing path.
	 */
	private function record( $source, $path ) {
		$data = self::get();
		$now  = current_time( 'mysql' );

		$data['totals'][ $source ] = ( isset( $data['totals'][ $source ] ) ? (int) $data['totals'][ $source ] : 0 ) + 1;

		$day                  = current_time( 'Y-m-d' );
		$data['by_day'][ $day ] = ( isset( $data['by_day'][ $day ] ) ? (int) $data['by_day'][ $day ] : 0 ) + 1;
		$data['by_day']         = self::trim_days( $data['by_day'] );

		if ( '' !== $path ) {
			$data['by_page'][ $path ] = ( isset( $data['by_page'][ $path ] ) ? (int) $data['by_page'][ $path ] : 0 ) + 1;
			$data['by_page']          = self::trim_pages( $data['by_page'] );
		}

		if ( '' === $data['first_seen'] ) {
			$data['first_seen'] = $now;
		}
		$data['last_seen'] = $now;

		update_option( ADG_AIM_OPT_REFERRALS, $data, false );
	}

	/* ------------------------------------------------------------------ */

	/**
	 * Grand total across all sources.
	 *
	 * @param array $data Aggregate.
	 * @return int
	 */
	public static function total( array $data ) {
		$t = isset( $data['totals'] ) && is_array( $data['totals'] ) ? $data['totals'] : array();
		return (int) array_sum( array_map( 'intval', $t ) );
	}

	/**
	 * Totals per source, sorted desc, with labels.
	 *
	 * @param array $data Aggregate.
	 * @return array list of array( key, label, count ).
	 */
	public static function by_source( array $data ) {
		$labels = self::source_labels();
		$totals = isset( $data['totals'] ) && is_array( $data['totals'] ) ? $data['totals'] : array();
		arsort( $totals );
		$out = array();
		foreach ( $totals as $key => $count ) {
			$out[] = array(
				'key'   => $key,
				'label' => isset( $labels[ $key ] ) ? $labels[ $key ] : ucfirst( $key ),
				'count' => (int) $count,
			);
		}
		return $out;
	}

	/**
	 * "Last N days" series 'Y-m-d' => count (with zeros), TZ-stable.
	 *
	 * @param array $data Aggregate.
	 * @param int   $days Days back.
	 * @return array
	 */
	public static function daily_series( array $data, $days = 30 ) {
		$days   = max( 1, (int) $days );
		$by_day = isset( $data['by_day'] ) && is_array( $data['by_day'] ) ? $data['by_day'] : array();

		$tz    = new DateTimeZone( 'UTC' );
		$today = DateTimeImmutable::createFromFormat( 'Y-m-d', current_time( 'Y-m-d' ), $tz );
		if ( false === $today ) {
			$today = new DateTimeImmutable( 'now', $tz );
		}

		$series = array();
		for ( $i = $days - 1; $i >= 0; $i-- ) {
			$key            = $today->modify( "-{$i} days" )->format( 'Y-m-d' );
			$series[ $key ] = isset( $by_day[ $key ] ) ? (int) $by_day[ $key ] : 0;
		}
		return $series;
	}

	/**
	 * Top landing pages by referral count.
	 *
	 * @param array $data Aggregate.
	 * @param int   $n    Count.
	 * @return array path => count.
	 */
	public static function top_pages( array $data, $n = 10 ) {
		$pages = isset( $data['by_page'] ) && is_array( $data['by_page'] ) ? $data['by_page'] : array();
		arsort( $pages );
		return array_slice( $pages, 0, max( 1, (int) $n ), true );
	}

	/**
	 * Sum of the last N days across all sources.
	 *
	 * @param array $data Aggregate.
	 * @param int   $days Days.
	 * @return int
	 */
	public static function sum_last_days( array $data, $days = 7 ) {
		return (int) array_sum( self::daily_series( $data, $days ) );
	}

	/* ------------------------------------------------------------------ */

	/**
	 * Trim the by-day breakdown to the last DAYS_KEEP.
	 *
	 * @param array $days 'Y-m-d' => int.
	 * @return array
	 */
	private static function trim_days( array $days ) {
		if ( count( $days ) <= self::DAYS_KEEP ) {
			return $days;
		}
		ksort( $days );
		return array_slice( $days, - self::DAYS_KEEP, null, true );
	}

	/**
	 * Shrink landing pages on overflow.
	 *
	 * @param array $pages path => int.
	 * @return array
	 */
	private static function trim_pages( array $pages ) {
		if ( count( $pages ) <= self::PAGES_CAP ) {
			return $pages;
		}
		arsort( $pages );
		return array_slice( $pages, 0, self::PAGES_TARGET, true );
	}
}