<?php
/**
 * Report site value object.
 *
 * @package TurtleReport
 */

namespace TurtleReport\Profiles;

use TurtleReport\Policy\Options;
use TurtleReport\Policy\Schedule;
use TurtleReport\Policy\Segments;
use TurtleReport\Policy\TrafficMetrics;

defined( 'ABSPATH' ) || exit;

/**
 * A "Site" in the admin UI: one website's report.
 *
 * It holds the Google sources (Search Console, GA4, Business Profile), what
 * the report collects from them, and the roles that may view it. Individual
 * user grants live in user meta (see AccessPolicy) so WordPress can query
 * them efficiently.
 */
final class Profile {

	const BASIS_FIRST_USER = 'firstUser';
	const BASIS_SESSION    = 'session';

	const SCOPE_ORGANIC = 'organic';
	const SCOPE_ALL     = 'all';

	/** Report sections an admin can switch on or off per site. */
	const SECTION_SEARCH      = 'search';
	const SECTION_CONVERSIONS = 'conversions';
	const SECTION_AI          = 'ai';
	const SECTION_TRAFFIC     = 'traffic';
	const SECTION_BUSINESS    = 'business';

	const SEARCH_TYPES = array( 'web', 'image', 'video', 'news' );

	/** Google data sources. Each can read through its own connection. */
	const SOURCE_SEARCH    = 'search';
	const SOURCE_ANALYTICS = 'analytics';
	const SOURCE_BUSINESS  = 'business';

	/** GA4 allows at most 10 metrics per report request. */
	const MAX_TRAFFIC_METRICS = 10;

	/** Most conversion groups per site. */
	const MAX_GROUPS = 8;

	/**
	 * Site ID.
	 *
	 * @var string
	 */
	public $id = '';

	/**
	 * Display name.
	 *
	 * @var string
	 */
	public $name = '';

	/**
	 * Connection used to read this site's data (every source, unless one is
	 * overridden in $source_connections).
	 *
	 * @var string
	 */
	public $connection_id = '';

	/**
	 * Optional connection per source (SOURCE_* => connection ID). Empty means
	 * "use $connection_id".
	 *
	 * @var array<string,string>
	 */
	public $source_connections = array();

	/**
	 * Search Console property, e.g. "sc-domain:example.com" or "https://example.com/".
	 *
	 * @var string
	 */
	public $gsc_site = '';

	/**
	 * Search Console search type: web, image, video or news.
	 *
	 * @var string
	 */
	public $search_type = 'web';

	/**
	 * GA4 property resource name, e.g. "properties/123456".
	 *
	 * @var string
	 */
	public $ga_property = '';

	/**
	 * GA4 property display name.
	 *
	 * @var string
	 */
	public $ga_property_name = '';

	/**
	 * Business Profile location resource name, e.g. "locations/987".
	 *
	 * @var string
	 */
	public $gbp_location = '';

	/**
	 * Business Profile location display name.
	 *
	 * @var string
	 */
	public $gbp_location_name = '';

	/**
	 * Enabled report sections (SECTION_* constants).
	 *
	 * @var string[]
	 */
	public $sections = array();

	/**
	 * What this site calls its conversions, e.g. "Leads", "Bookings" or "Sales".
	 *
	 * @var string
	 */
	public $conversion_label = '';

	/**
	 * Conversion groups: each has a label and the GA4 event names it counts.
	 *
	 * @var array[] Items of ['label' => string, 'events' => string[]].
	 */
	public $conversion_groups = array();

	/**
	 * Which GA4 channel dimension decides "organic".
	 *
	 * @var string
	 */
	public $lead_basis = self::BASIS_FIRST_USER;

	/**
	 * GA4 traffic metrics to show (keys of TrafficMetrics::all()).
	 *
	 * @var string[]
	 */
	public $traffic_metrics = array();

	/**
	 * Whether traffic metrics cover organic search only or all channels.
	 *
	 * @var string
	 */
	public $traffic_scope = self::SCOPE_ORGANIC;

	/**
	 * Segments (named query or page filters).
	 *
	 * @var array[] See Segments::normalize().
	 */
	public $segments = array();

	/**
	 * Client logo attachment ID, shown on the PDF cover and in emails.
	 *
	 * @var int
	 */
	public $client_logo_id = 0;

	/**
	 * Scheduled email report settings.
	 *
	 * @var array See Email\Schedule::normalize().
	 */
	public $email = array();

	/**
	 * Delivery options: PDF attachment, secure link, review mode, incomplete reports, insights.
	 *
	 * @var array See Delivery\Options::normalize().
	 */
	public $delivery = array();

	/**
	 * Goals and priorities: main goal, important pages, priority GA4 events.
	 *
	 * @var array See Priorities::normalize().
	 */
	public $priorities = array();

	/**
	 * Roles allowed to view this site.
	 *
	 * @var string[]
	 */
	public $roles = array();

	/**
	 * Creation time as a Unix timestamp.
	 *
	 * @var int
	 */
	public $created = 0;

	/**
	 * A new site starts with the same defaults as stored data with fields missing:
	 * the default sections and metrics ticked, and a complete (off) email schedule.
	 */
	public function __construct() {
		$this->sections           = self::default_sections();
		$this->traffic_metrics    = TrafficMetrics::defaults();
		$this->email              = Schedule::normalize( array() );
		$this->delivery           = Options::defaults();
		$this->source_connections = self::empty_sources();
		$this->priorities         = Priorities::defaults();
		foreach ( ConversionPresets::default_groups() as $group ) {
			$this->conversion_groups[] = array(
				'label'  => (string) $group['label'],
				'events' => array_values( array_map( 'strval', (array) $group['events'] ) ),
			);
		}
	}

	/**
	 * Builds a site from stored data. Missing fields get defaults.
	 *
	 * @param array $data Stored data.
	 * @return self
	 */
	public static function from_array( array $data ) {
		$p                    = new self();
		$str                  = static function ( $key, $fallback = '' ) use ( $data ) {
			return isset( $data[ $key ] ) && is_scalar( $data[ $key ] ) ? (string) $data[ $key ] : $fallback;
		};
		$list                 = static function ( $key, $fallback = array() ) use ( $data ) {
			return isset( $data[ $key ] ) && is_array( $data[ $key ] ) ? array_values( array_map( 'strval', array_filter( $data[ $key ], 'is_scalar' ) ) ) : $fallback;
		};
		$p->id                = $str( 'id' );
		$p->name              = $str( 'name' );
		$p->connection_id     = $str( 'connection_id' );
		$p->gsc_site          = $str( 'gsc_site' );
		$p->search_type       = in_array( $str( 'search_type' ), self::SEARCH_TYPES, true ) ? $str( 'search_type' ) : 'web';
		$p->ga_property       = $str( 'ga_property' );
		$p->ga_property_name  = $str( 'ga_property_name' );
		$p->gbp_location      = $str( 'gbp_location' );
		$p->gbp_location_name = $str( 'gbp_location_name' );
		$p->sections          = array_values( array_intersect( $list( 'sections', self::default_sections() ), self::all_sections() ) );
		$p->conversion_label  = $str( 'conversion_label' );
		$p->lead_basis        = self::BASIS_SESSION === $str( 'lead_basis' ) ? self::BASIS_SESSION : self::BASIS_FIRST_USER;
		$p->traffic_metrics   = array_values( array_intersect( $list( 'traffic_metrics', TrafficMetrics::defaults() ), array_keys( TrafficMetrics::all() ) ) );
		$p->traffic_scope     = self::SCOPE_ALL === $str( 'traffic_scope' ) ? self::SCOPE_ALL : self::SCOPE_ORGANIC;
		$p->roles             = $list( 'roles' );
		$p->created           = isset( $data['created'] ) ? (int) $data['created'] : 0;
		$p->segments          = isset( $data['segments'] ) && is_array( $data['segments'] ) ? Segments::normalize( $data['segments'] ) : array();
		$p->client_logo_id    = isset( $data['client_logo_id'] ) ? max( 0, (int) $data['client_logo_id'] ) : 0;
		$p->email             = Schedule::normalize( isset( $data['email'] ) && is_array( $data['email'] ) ? $data['email'] : array() );
		// Missing delivery fields use the same defaults as a newly created site.
		$p->delivery           = isset( $data['delivery'] ) && is_array( $data['delivery'] ) ? Options::normalize( $data['delivery'] ) : Options::defaults();
		$p->source_connections = self::empty_sources();
		// Existing 1.0 sites have no priorities: retain the usual order until goals are chosen.
		$p->priorities = Priorities::normalize( isset( $data['priorities'] ) ? $data['priorities'] : array() );
		if ( isset( $data['source_connections'] ) && is_array( $data['source_connections'] ) ) {
			foreach ( array_keys( $p->source_connections ) as $source ) {
				$value                            = isset( $data['source_connections'][ $source ] ) && is_scalar( $data['source_connections'][ $source ] ) ? (string) $data['source_connections'][ $source ] : '';
				$p->source_connections[ $source ] = preg_match( '/^[a-z0-9]{2,20}$/', $value ) ? $value : '';
			}
		}

		$groups               = isset( $data['conversion_groups'] ) && is_array( $data['conversion_groups'] ) ? $data['conversion_groups'] : ConversionPresets::default_groups();
		$p->conversion_groups = array();
		foreach ( $groups as $group ) {
			if ( is_array( $group ) && isset( $group['label'] ) && is_scalar( $group['label'] ) && '' !== (string) $group['label'] ) {
				$p->conversion_groups[] = array(
					'label'  => (string) $group['label'],
					'events' => isset( $group['events'] ) && is_array( $group['events'] ) ? array_values( array_map( 'strval', array_filter( $group['events'], 'is_scalar' ) ) ) : array(),
				);
			}
		}
		return $p;
	}

	/**
	 * Data to store.
	 *
	 * @return array
	 */
	public function to_array() {
		return get_object_vars( $this );
	}

	/**
	 * Source keys, in display order.
	 *
	 * @return string[]
	 */
	public static function sources() {
		return array( self::SOURCE_SEARCH, self::SOURCE_ANALYTICS, self::SOURCE_BUSINESS );
	}

	/**
	 * Per-source overrides with nothing set.
	 *
	 * @return array<string,string>
	 */
	public static function empty_sources() {
		return array_fill_keys( self::sources(), '' );
	}

	/**
	 * The Google source a report section reads from.
	 *
	 * @param string $section SECTION_* constant.
	 * @return string SOURCE_* constant.
	 */
	public static function source_for_section( $section ) {
		switch ( $section ) {
			case self::SECTION_SEARCH:
				return self::SOURCE_SEARCH;
			case self::SECTION_BUSINESS:
				return self::SOURCE_BUSINESS;
			default:
				return self::SOURCE_ANALYTICS;
		}
	}

	/**
	 * Connection that reads a source: its own override, else the site's main connection.
	 *
	 * @param string $source SOURCE_* constant.
	 * @return string Connection ID, or '' when none is chosen.
	 */
	public function connection_for( $source ) {
		$own = isset( $this->source_connections[ $source ] ) ? (string) $this->source_connections[ $source ] : '';
		return '' !== $own ? $own : $this->connection_id;
	}

	/**
	 * Whether a source reads through a different connection than the main one.
	 *
	 * @param string $source SOURCE_* constant.
	 * @return bool
	 */
	public function has_own_connection( $source ) {
		return '' !== ( isset( $this->source_connections[ $source ] ) ? (string) $this->source_connections[ $source ] : '' ) && $this->source_connections[ $source ] !== $this->connection_id;
	}

	/**
	 * Every connection this site reads through.
	 *
	 * @return string[]
	 */
	public function connection_ids() {
		$ids = array( $this->connection_id );
		foreach ( self::sources() as $source ) {
			$ids[] = $this->connection_for( $source );
		}
		return array_values( array_unique( array_filter( $ids ) ) );
	}

	/**
	 * Sources that read through a connection.
	 *
	 * @param string $connection_id Connection ID.
	 * @return string[] SOURCE_* constants.
	 */
	public function sources_using( $connection_id ) {
		$out = array();
		foreach ( self::sources() as $source ) {
			if ( '' !== $connection_id && $this->connection_for( $source ) === $connection_id ) {
				$out[] = $source;
			}
		}
		return $out;
	}

	/**
	 * Every section key.
	 *
	 * @return string[]
	 */
	public static function all_sections() {
		return array( self::SECTION_SEARCH, self::SECTION_CONVERSIONS, self::SECTION_AI, self::SECTION_TRAFFIC, self::SECTION_BUSINESS );
	}

	/**
	 * Sections switched on for new sites.
	 *
	 * @return string[]
	 */
	public static function default_sections() {
		return self::all_sections();
	}

	/**
	 * Whether a section is switched on and its source is configured.
	 *
	 * The REST API refuses data for sections that are off, so turning a
	 * section off really stops it being collected and shown.
	 *
	 * @param string $section SECTION_* constant.
	 * @return bool
	 */
	public function wants( $section ) {
		if ( ! in_array( $section, $this->sections, true ) ) {
			return false;
		}
		switch ( $section ) {
			case self::SECTION_SEARCH:
				return '' !== $this->gsc_site;
			case self::SECTION_BUSINESS:
				return '' !== $this->gbp_location;
			default:
				return '' !== $this->ga_property;
		}
	}

	/**
	 * GA4 events that appear in more than one conversion group.
	 *
	 * The overall total counts each event once, but each group's own number
	 * still includes it, so the groups can add up to more than the total.
	 *
	 * @return array<string,string[]> Event name => labels of the groups it is in.
	 */
	public function overlapping_events() {
		$seen = array();
		foreach ( $this->conversion_groups as $group ) {
			foreach ( array_unique( $group['events'] ) as $event ) {
				$seen[ $event ][] = $group['label'];
			}
		}
		return array_filter(
			$seen,
			static function ( $labels ) {
				return count( $labels ) > 1;
			}
		);
	}

	/**
	 * Every GA4 event name counted as a conversion, across all groups.
	 *
	 * @return string[]
	 */
	public function all_conversion_events() {
		$out = array();
		foreach ( $this->conversion_groups as $group ) {
			$out = array_merge( $out, $group['events'] );
		}
		return array_values( array_unique( $out ) );
	}

	/**
	 * Goals and priorities, cleaned.
	 *
	 * @return array{goal:string,custom:string,pages:string[],events:string[]}
	 */
	public function priorities() {
		return Priorities::normalize( $this->priorities );
	}

	/**
	 * Priority GA4 events that are still among this site's conversion events.
	 *
	 * An event removed from the conversion groups stops counting as a priority.
	 *
	 * @return string[]
	 */
	public function priority_events() {
		return array_values( array_intersect( $this->priorities()['events'], $this->all_conversion_events() ) );
	}

	/**
	 * Conversion label, falling back to "Leads".
	 *
	 * When the label is the default "Leads" but some of the GA4 events are
	 * clicks (a phone or email link, a button), the label says so, because a
	 * tap on a phone number is not a lead yet. A label chosen for the site
	 * (Enquiries, Sign-ups ...) is always kept as it is.
	 *
	 * @return string
	 */
	public function conversion_label() {
		$leads = __( 'Leads', 'turtlereport' );
		$label = '' !== $this->conversion_label ? $this->conversion_label : $leads;
		if ( $label !== $leads && 'Leads' !== $label ) {
			return $label;
		}
		$events = $this->all_conversion_events();
		$clicks = array_values( array_filter( $events, array( __CLASS__, 'is_click_event' ) ) );
		if ( ! $clicks ) {
			return $label;
		}
		$phone = count( $clicks ) === count( array_filter( $clicks, array( __CLASS__, 'is_phone_event' ) ) );
		if ( count( $clicks ) === count( $events ) ) {
			return $phone ? __( 'Phone clicks', 'turtlereport' ) : __( 'Contact clicks', 'turtlereport' );
		}
		return $phone ? __( 'Leads + phone clicks', 'turtlereport' ) : __( 'Leads + clicks', 'turtlereport' );
	}

	/**
	 * Whether a GA4 event records a click or tap rather than a submitted lead.
	 *
	 * @param string $event Event name.
	 * @return bool
	 */
	public static function is_click_event( $event ) {
		return (bool) preg_match( '/click|(^|[\s_\-])tap([\s_\-]|$)|outbound|^(tel|mailto|whatsapp)([\s_:\-]|$)/i', (string) $event );
	}

	/**
	 * Whether a click event is a phone (or WhatsApp) tap.
	 *
	 * @param string $event Event name.
	 * @return bool
	 */
	public static function is_phone_event( $event ) {
		return (bool) preg_match( '/phone|call|(^|[\s_\-])tel([\s_:\-]|$)|whatsapp/i', (string) $event );
	}

	/**
	 * Search Console site as a readable host name.
	 *
	 * @return string
	 */
	public function site_label() {
		$label = preg_replace( '#^(sc-domain:|https?://)#', '', $this->gsc_site );
		return rtrim( (string) $label, '/' );
	}

	/**
	 * Name to show in lists.
	 *
	 * @return string
	 */
	public function display_name() {
		if ( '' !== $this->name ) {
			return $this->name;
		}
		if ( '' !== $this->gsc_site ) {
			return $this->site_label();
		}
		return '' !== $this->ga_property_name ? $this->ga_property_name : __( 'Untitled site', 'turtlereport' );
	}
}