<?php
/**
 * AI assistant traffic report.
 *
 * @package TurtleReport
 */

namespace TurtleReport\Reports;

use TurtleReport\Google\Analytics;
use TurtleReport\Policy\Segments;
use TurtleReport\Profiles\Profile;

defined( 'ABSPATH' ) || exit;

/**
 * Sessions and leads from AI assistants such as ChatGPT, Gemini and Claude.
 *
 * Uses GA4's "AI Assistant" default channel. Properties that don't show that
 * channel yet fall back to matching known AI sources by name.
 */
final class AiTrafficReport extends Report {

	const SOURCE_PATTERN = 'chatgpt|openai|perplexity|gemini|bard|copilot|claude|anthropic|deepseek|you\\.com|meta\\.ai|grok|mistral|poe\\.com';

	/**
	 * GA4 client for the current run.
	 *
	 * @var Analytics
	 */
	private $ga;

	/**
	 * Site for the current run.
	 *
	 * @var Profile
	 */
	private $site;

	/**
	 * Segment filter for the current run, or null.
	 *
	 * @var array|null
	 */
	private $segment_filter = null;

	/**
	 * Cache lifetime for the current run.
	 *
	 * @var int
	 */
	private $cache_ttl = 0;

	/**
	 * {@inheritDoc}
	 *
	 * @return string
	 */
	public function id() {
		return 'ai-traffic';
	}

	/**
	 * {@inheritDoc}
	 *
	 * @return string
	 */
	public function section() {
		return Profile::SECTION_AI;
	}

	/**
	 * {@inheritDoc}
	 *
	 * @return string[]
	 */
	public function segment_types() {
		return array( Segments::TYPE_PAGE );
	}

	/**
	 * {@inheritDoc}
	 *
	 * @return array
	 */
	public function args() {
		$date = array(
			'type'     => 'string',
			'pattern'  => '^\d{4}-\d{2}-\d{2}$',
			'required' => true,
		);
		return array(
			'prev_start' => $date,
			'prev_end'   => $date,
		);
	}

	/**
	 * {@inheritDoc}
	 *
	 * @param Profile $site   Site.
	 * @param array   $params Parameters.
	 * @return array
	 */
	public function run( Profile $site, array $params ) {
		$this->site           = $site;
		$this->ga             = $this->google->for_site( $site, Profile::SOURCE_ANALYTICS )->analytics();
		$this->cache_ttl      = $this->ttl( $params['end'] );
		$segment              = $this->segment( $site, $params );
		$this->segment_filter = $segment ? SegmentFilter::ga( $segment ) : null;
		$events               = $site->all_conversion_events();
		$s                    = $params['start'];
		$e                    = $params['end'];

		$mode = 'channel';
		$cur  = $this->totals( $mode, $s, $e );
		if ( 0 === $cur['sessions'] ) {
			$alt = $this->totals( 'source', $s, $e );
			if ( $alt['sessions'] > 0 ) {
				$mode = 'source';
				$cur  = $alt;
			}
		}

		return array(
			'mode'       => $mode,
			'cur'        => $cur,
			'prev'       => $this->totals( $mode, $params['prev_start'], $params['prev_end'] ),
			'ev'         => self::map( $this->event_counts( $mode, $events, $s, $e ) ),
			'pev'        => self::map( $this->event_counts( $mode, $events, $params['prev_start'], $params['prev_end'] ) ),
			'pages'      => $this->top_pages( $mode, $s, $e ),
			'page_leads' => self::map( $this->page_leads( $mode, $events, $s, $e ) ),
			'sources'    => $this->sources( $mode, $s, $e ),
			'all'        => $this->totals( null, $s, $e )['sessions'],
		);
	}

	/**
	 * Filter for AI traffic.
	 *
	 * @param string|null $mode channel, source, or null for all traffic.
	 * @return array|null
	 */
	private function ai_filter( $mode ) {
		return Filters::all( $this->base_filter( $mode ), $this->segment_filter );
	}

	/**
	 * AI channel or source filter, before segments.
	 *
	 * @param string|null $mode channel, source, or null for all traffic.
	 * @return array|null
	 */
	private function base_filter( $mode ) {
		if ( null === $mode ) {
			return null;
		}
		if ( 'source' === $mode ) {
			/**
			 * Filters the pattern used to spot AI assistant sources when GA4
			 * has no "AI Assistant" channel for a property.
			 *
			 * @param string $pattern RE2 regular expression.
			 */
			return Filters::regex( 'sessionSource', (string) apply_filters( 'turtlereport_ai_source_pattern', self::SOURCE_PATTERN ) );
		}
		return Filters::exact( 'sessionDefaultChannelGroup', 'AI Assistant' );
	}

	/**
	 * Sessions and engaged sessions.
	 *
	 * @param string|null $mode  Filter mode.
	 * @param string      $start Start date.
	 * @param string      $end   End date.
	 * @return array
	 */
	private function totals( $mode, $start, $end ) {
		$body   = array(
			'dateRanges' => array(
				array(
					'startDate' => $start,
					'endDate'   => $end,
				),
			),
			'metrics'    => array( array( 'name' => 'sessions' ), array( 'name' => 'engagedSessions' ) ),
		);
		$filter = $this->ai_filter( $mode );
		if ( $filter ) {
			$body['dimensionFilter'] = $filter;
		}
		$rows = $this->ga->run_report( $this->site->ga_property, $body, $this->cache_ttl );
		return array(
			'sessions' => isset( $rows[0] ) ? (int) $rows[0]['m'][0] : 0,
			'engaged'  => isset( $rows[0] ) ? (int) $rows[0]['m'][1] : 0,
		);
	}

	/**
	 * Lead event counts from AI traffic.
	 *
	 * @param string   $mode   Filter mode.
	 * @param string[] $events Lead events.
	 * @param string   $start  Start date.
	 * @param string   $end    End date.
	 * @return array
	 */
	private function event_counts( $mode, array $events, $start, $end ) {
		if ( ! $events ) {
			return array();
		}
		$rows = $this->ga->run_report(
			$this->site->ga_property,
			array(
				'dateRanges'      => array(
					array(
						'startDate' => $start,
						'endDate'   => $end,
					),
				),
				'dimensions'      => array( array( 'name' => 'eventName' ) ),
				'metrics'         => array( array( 'name' => 'eventCount' ) ),
				'dimensionFilter' => Filters::all( $this->ai_filter( $mode ), Filters::events( $events ) ),
			),
			$this->cache_ttl
		);
		$out  = array();
		foreach ( $rows as $row ) {
			$out[ $row['d'][0] ] = (int) $row['m'][0];
		}
		return $out;
	}

	/**
	 * Top landing pages for AI traffic.
	 *
	 * @param string $mode  Filter mode.
	 * @param string $start Start date.
	 * @param string $end   End date.
	 * @return array[]
	 */
	private function top_pages( $mode, $start, $end ) {
		$rows = $this->ga->run_report(
			$this->site->ga_property,
			array(
				'dateRanges'      => array(
					array(
						'startDate' => $start,
						'endDate'   => $end,
					),
				),
				'dimensions'      => array( array( 'name' => 'landingPage' ) ),
				'metrics'         => array( array( 'name' => 'sessions' ) ),
				'dimensionFilter' => $this->ai_filter( $mode ),
				'orderBys'        => array(
					array(
						'metric' => array( 'metricName' => 'sessions' ),
						'desc'   => true,
					),
				),
				'limit'           => 10,
			),
			$this->cache_ttl
		);
		$out  = array();
		foreach ( $rows as $row ) {
			if ( '' !== $row['d'][0] && '(not set)' !== $row['d'][0] ) {
				$out[] = array(
					'p'        => $row['d'][0],
					'sessions' => (int) $row['m'][0],
				);
			}
		}
		return $out;
	}

	/**
	 * Lead events per landing page from AI traffic.
	 *
	 * @param string   $mode   Filter mode.
	 * @param string[] $events Lead events.
	 * @param string   $start  Start date.
	 * @param string   $end    End date.
	 * @return array
	 */
	private function page_leads( $mode, array $events, $start, $end ) {
		if ( ! $events ) {
			return array();
		}
		$rows = $this->ga->run_report(
			$this->site->ga_property,
			array(
				'dateRanges'      => array(
					array(
						'startDate' => $start,
						'endDate'   => $end,
					),
				),
				'dimensions'      => array( array( 'name' => 'landingPage' ), array( 'name' => 'eventName' ) ),
				'metrics'         => array( array( 'name' => 'eventCount' ) ),
				'dimensionFilter' => Filters::all( $this->ai_filter( $mode ), Filters::events( $events ) ),
				'limit'           => 1000,
			),
			$this->cache_ttl
		);
		$out  = array();
		foreach ( $rows as $row ) {
			$path                   = self::normalize_path( $row['d'][0] );
			$event                  = $row['d'][1];
			$out[ $path ][ $event ] = ( isset( $out[ $path ][ $event ] ) ? $out[ $path ][ $event ] : 0 ) + (int) $row['m'][0];
		}
		return $out;
	}

	/**
	 * Top AI sources.
	 *
	 * @param string $mode  Filter mode.
	 * @param string $start Start date.
	 * @param string $end   End date.
	 * @return array[]
	 */
	private function sources( $mode, $start, $end ) {
		$rows = $this->ga->run_report(
			$this->site->ga_property,
			array(
				'dateRanges'      => array(
					array(
						'startDate' => $start,
						'endDate'   => $end,
					),
				),
				'dimensions'      => array( array( 'name' => 'sessionSource' ) ),
				'metrics'         => array( array( 'name' => 'sessions' ) ),
				'dimensionFilter' => $this->ai_filter( $mode ),
				'orderBys'        => array(
					array(
						'metric' => array( 'metricName' => 'sessions' ),
						'desc'   => true,
					),
				),
				'limit'           => 8,
			),
			$this->cache_ttl
		);
		return array_map(
			static function ( $row ) {
				return array(
					'name'     => $row['d'][0],
					'sessions' => (int) $row['m'][0],
				);
			},
			$rows
		);
	}
}