<?php
/**
 * Visit counts.
 *
 * @package Dealroot
 */

namespace Dealroot;

defined( 'ABSPATH' ) || exit;

/**
 * Daily visit counters per channel, source, campaign and landing page.
 * Only aggregates are stored, nothing about individual visitors.
 */
class Visits {

	/**
	 * Raised when visits are deleted, so the kept sums of past days are counted again.
	 */
	const GENERATION = 'dealroot_visits_generation';

	/**
	 * Counts one visit for the touch that started it.
	 *
	 * @param array $touch   Sanitized touch.
	 * @param array $visitor Device, browser, country and city, see Visitor::columns().
	 * @return bool
	 */
	public static function record( array $touch, array $visitor = array() ) {
		global $wpdb;

		$row = array(
			'day'          => wp_date( 'Y-m-d' ),
			'channel'      => Attribution::channel( $touch ),
			'source'       => mb_substr( Attribution::source( $touch ), 0, 191 ),
			'medium'       => mb_substr( Attribution::medium( $touch ), 0, 191 ),
			'campaign'     => mb_substr( (string) ( $touch['utm_campaign'] ?? '' ), 0, 191 ),
			'landing_page' => mb_substr( (string) strtok( (string) ( $touch['lp'] ?? '' ), '?' ), 0, 191 ),
			'device'       => (string) ( $visitor['device'] ?? '' ),
			'browser'      => mb_substr( (string) ( $visitor['browser'] ?? '' ), 0, 40 ),
			'country'      => Geo::code( $visitor['country'] ?? '' ),
			'city'         => mb_substr( (string) ( $visitor['city'] ?? '' ), 0, 100 ),
		);

		// 0, 0 stands for an unknown place: no city lies there.
		$point = Geo::point( $visitor['lat'] ?? null, $visitor['lon'] ?? null );
		$lat   = (float) $point[0];
		$lon   = (float) $point[1];

		$row['hash'] = md5( implode( '|', $row ) . '|' . $lat . ',' . $lon );

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		$result = $wpdb->query(
			$wpdb->prepare(
				'INSERT INTO %i (day, hash, channel, source, medium, campaign, landing_page, device, browser, country, city, lat, lon, visits) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %f, %f, 1) ON DUPLICATE KEY UPDATE visits = visits + 1',
				Installer::visits_table(),
				$row['day'],
				$row['hash'],
				$row['channel'],
				$row['source'],
				$row['medium'],
				$row['campaign'],
				$row['landing_page'],
				$row['device'],
				$row['browser'],
				$row['country'],
				$row['city'],
				$lat,
				$lon
			)
		);

		return false !== $result;
	}

	/**
	 * Visits per group.
	 *
	 * @param string      $from   UTC start, empty for all time.
	 * @param string      $to     UTC end, empty for now.
	 * @param string      $column     Report column.
	 * @param bool|string $by_channel True to keep the bought clicks of a source apart, "campaign" to split visits without a campaign by channel.
	 * @return array|null Group => visits, or null when visits cannot be grouped by this column.
	 */
	public static function counts( $from, $to, $column, $by_channel = false ) {
		global $wpdb;

		$map = array(
			'channel'       => 'channel',
			'last_channel'  => 'channel',
			'source'        => 'source',
			'last_source'   => 'source',
			'campaign'      => 'campaign',
			'last_campaign' => 'campaign',
			'landing_page'  => 'landing_page',
			'device'        => 'device',
			'browser'       => 'browser',
			'country'       => 'country',
			'city'          => 'city',
		);

		if ( ! isset( $map[ $column ] ) ) {
			return null;
		}

		$group               = $map[ $column ];
		$channel             = $by_channel ? 'channel' : $group;
		list( $start, $end ) = Reports::sql_days( $from, $to );

		$rows = self::summed(
			'counts|' . $group . '|' . $channel,
			$start,
			$end,
			static function ( $first, $last ) use ( $wpdb, $group, $channel ) {
				// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
				return $wpdb->get_results(
					$wpdb->prepare(
						'SELECT %i AS grp, %i AS chn, SUM(visits) AS total FROM %i WHERE day >= %s AND day < %s GROUP BY %i, %i',
						$group,
						$channel,
						Installer::visits_table(),
						$first,
						$last,
						$group,
						$channel
					)
				);
			}
		);

		$counts = array();
		foreach ( (array) $rows as $row ) {
			$key = Reports::split_key( $by_channel, (string) $row->grp, (string) $row->chn );

			$counts[ $key ] = ( $counts[ $key ] ?? 0 ) + (int) $row->total;
		}

		return $counts;
	}

	/**
	 * Rows summed from the visits of some days. The days before today no longer change, so their
	 * sums are kept until the next day; only today is counted again. The rows of the two parts
	 * come one after the other, for the caller to add up.
	 *
	 * @param string   $name  What is summed, part of the cache key.
	 * @param string   $start First day, Y-m-d.
	 * @param string   $end   Day after the last, Y-m-d.
	 * @param callable $query Runs the query for a first day and the day after the last.
	 * @return object[]
	 */
	public static function summed( $name, $start, $end, callable $query ) {
		$today = wp_date( 'Y-m-d' );
		$rows  = array();

		if ( $start < $today ) {
			$last = min( $end, $today );
			$key  = 'dealroot_visits_' . md5( $name . '|' . $start . '|' . $last . '|' . (int) get_option( self::GENERATION, 0 ) );
			$past = get_transient( $key );

			if ( ! is_array( $past ) ) {
				$past = array_map( 'get_object_vars', (array) $query( $start, $last ) );
				set_transient( $key, $past, DAY_IN_SECONDS );
			}

			$rows = array_map(
				static function ( $row ) {
					return (object) $row;
				},
				$past
			);
		}

		if ( $end > $today ) {
			$rows = array_merge( $rows, (array) $query( max( $start, $today ), $end ) );
		}

		return $rows;
	}

	/**
	 * Starts the sums of past days afresh, after visits were deleted.
	 */
	public static function forget_sums() {
		update_option( self::GENERATION, (int) get_option( self::GENERATION, 0 ) + 1 );
	}

	/**
	 * Visits counted today and the days before.
	 *
	 * @param int $days Days back, today included.
	 * @return int
	 */
	public static function recent( $days ) {
		global $wpdb;

		$since = wp_date( 'Y-m-d', time() - ( max( 1, (int) $days ) - 1 ) * DAY_IN_SECONDS );

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		return (int) $wpdb->get_var( $wpdb->prepare( 'SELECT SUM(visits) FROM %i WHERE day >= %s', Installer::visits_table(), $since ) );
	}

	/**
	 * Whether any visit has been counted yet.
	 *
	 * @return bool
	 */
	public static function has_data() {
		global $wpdb;

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		return (bool) $wpdb->get_var( $wpdb->prepare( 'SELECT 1 FROM %i LIMIT 1', Installer::visits_table() ) );
	}

	/**
	 * Removes counters older than the retention window.
	 *
	 * @param int $days Age in days.
	 * @return int Rows deleted.
	 */
	public static function purge_older_than( $days ) {
		global $wpdb;

		if ( $days < 1 ) {
			return 0;
		}

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
		$deleted = (int) $wpdb->query( $wpdb->prepare( 'DELETE FROM %i WHERE day < %s', Installer::visits_table(), wp_date( 'Y-m-d', time() - (int) $days * DAY_IN_SECONDS ) ) );

		if ( $deleted ) {
			self::forget_sums();
		}

		return $deleted;
	}

	/**
	 * Whether the user agent looks like a crawler or monitoring tool.
	 *
	 * @param string $user_agent User agent.
	 * @return bool
	 */
	public static function is_bot( $user_agent ) {
		if ( '' === $user_agent ) {
			return true;
		}

		return (bool) preg_match( '/bot|crawl|spider|slurp|headless|lighthouse|pagespeed|gtmetrix|pingdom|uptime|monitor|preview|facebookexternalhit|curl|wget|python|java\\/|go-http|axios|node-fetch/i', $user_agent );
	}
}