<?php
/**
 * Fabricates a deterministic Search Console searchAnalytics.query response.
 *
 * @package Tagsmith
 */

declare(strict_types=1);

namespace Tagsmith\Demo;

defined( 'ABSPATH' ) || exit;

require_once __DIR__ . '/Baseline.php';
require_once __DIR__ . '/FakeData.php';
require_once __DIR__ . '/FakeReports.php';

/**
 * Builds a SearchAnalyticsQueryResponse-shaped array from a request body.
 * Used by FakeGoogle.
 */
class FakeSearchConsole {

	/**
	 * Share of Organic Search sessions that Search Console reports as
	 * clicks (the rest are other search engines).
	 */
	private const CLICKS_PER_ORGANIC_SESSION = 0.85;

	/**
	 * Share of a scope's clicks covered by the listed queries (Google leaves
	 * out anonymized and long-tail queries).
	 */
	private const LISTED_QUERY_SHARE = 0.5;

	/**
	 * Builds a query response for a decoded request body, from the same
	 * site baseline as the Analytics reports (see Baseline): the site's
	 * clicks are a share of its Organic Search sessions, pages get their
	 * page share of those clicks, and queries share the site's (or the
	 * filtered page's) clicks. Deterministic for the same request.
	 *
	 * @param array<string, mixed> $body SearchAnalyticsQueryRequest body.
	 * @return array<string, mixed>
	 */
	public static function query( array $body ): array {
		$dims   = array_values( array_map( 'strval', (array) ( $body['dimensions'] ?? array() ) ) );
		$limit  = isset( $body['rowLimit'] ) ? max( 1, (int) $body['rowLimit'] ) : 25;
		$dates  = self::dates( (string) ( $body['startDate'] ?? '' ), (string) ( $body['endDate'] ?? '' ) );
		$page   = self::extract_page_filter( is_array( $body['dimensionFilterGroups'] ?? null ) ? $body['dimensionFilterGroups'] : array() );
		$search = FakeData::CHANNEL_SHARES['Organic Search'] * self::CLICKS_PER_ORGANIC_SESSION;

		if ( in_array( 'page', $dims, true ) ) {
			$keys = array();

			foreach ( Baseline::page_shares( array_keys( FakeReports::path_pool() ) ) as $path => $share ) {
				$keys[ (string) $path ] = array( home_url( (string) $path ), $share );
			}
		} else {
			$scope = null !== $page ? FakeReports::paths_share( array( (string) wp_parse_url( $page, PHP_URL_PATH ) ) ) : 1.0;
			$keys  = array();

			foreach ( Baseline::ranked( self::keywords( $page ), self::LISTED_QUERY_SHARE * $scope ) as $keyword => $share ) {
				$keys[ (string) $keyword ] = array( (string) $keyword, $share );
			}
		}

		// Floored once over the whole range (not per day), so small queries
		// still get their clicks; the sum of floors never exceeds the floor
		// of the sum, so queries still add up to at most their page.
		$sessions = Baseline::sum( $dates )['sessions'];
		$rows     = array();

		foreach ( $keys as $id => [ $key, $share ] ) {
			$clicks      = (int) floor( $sessions * $share * $search );
			$ctr         = 0.02 + 0.12 * Baseline::hash01( 'ctr|' . $id );
			$impressions = $clicks > 0 ? (int) round( $clicks / $ctr ) : 5 + (int) round( 60 * Baseline::hash01( 'imp|' . $id ) );

			$rows[] = array(
				'keys'        => array( $key ),
				'clicks'      => $clicks,
				'impressions' => $impressions,
				'ctr'         => $impressions > 0 ? round( $clicks / $impressions, 4 ) : 0.0,
				'position'    => round( 2 + 30 * Baseline::hash01( 'pos|' . $id ), 1 ),
			);
		}

		// The real API returns rows by clicks, descending.
		usort( $rows, static fn( array $a, array $b ): int => $b['clicks'] <=> $a['clicks'] );

		// No dimensions: one site-wide row, like the real API.
		if ( array() === $dims ) {
			$clicks      = (int) array_sum( array_column( $rows, 'clicks' ) );
			$impressions = (int) array_sum( array_column( $rows, 'impressions' ) );
			$weighted    = array_sum( array_map( static fn( array $row ): float => $row['position'] * $row['impressions'], $rows ) );

			return array(
				'rows' => array(
					array(
						'keys'        => array(),
						'clicks'      => $clicks,
						'impressions' => $impressions,
						'ctr'         => $impressions > 0 ? round( $clicks / $impressions, 4 ) : 0.0,
						'position'    => $impressions > 0 ? round( $weighted / $impressions, 1 ) : 0.0,
					),
				),
			);
		}

		return array( 'rows' => array_slice( $rows, 0, $limit ) );
	}

	/**
	 * Every date (Ymd) from start to end (Y-m-d), inclusive.
	 *
	 * @param string $start Start date.
	 * @param string $end   End date.
	 * @return string[]
	 */
	private static function dates( string $start, string $end ): array {
		$cursor = strtotime( $start );
		$stop   = strtotime( $end );
		$dates  = array();

		while ( false !== $cursor && false !== $stop && $cursor <= $stop ) {
			$dates[] = gmdate( 'Ymd', $cursor );
			$cursor += DAY_IN_SECONDS;
		}

		return $dates;
	}

	/**
	 * The query keywords: from the filtered page's title, or the generic
	 * pool.
	 *
	 * @param string|null $page_filter The filtered page URL, if any.
	 * @return string[]
	 */
	private static function keywords( ?string $page_filter ): array {
		return null !== $page_filter ? array_values( array_unique( self::keywords_from_page( $page_filter ) ) ) : FakeData::GSC_KEYWORDS;
	}

	/**
	 * The page URL of a dimensionFilterGroups 'page' filter, if any: the
	 * expression of an equals filter, or the site URL of the path in an
	 * includingRegex filter of the form ^https?://[^/]+/path/?$.
	 *
	 * @param array<int, mixed> $groups dimensionFilterGroups.
	 */
	private static function extract_page_filter( array $groups ): ?string {
		foreach ( $groups as $group ) {
			foreach ( (array) ( is_array( $group ) ? ( $group['filters'] ?? array() ) : array() ) as $filter ) {
				if ( ! is_array( $filter ) || 'page' !== ( $filter['dimension'] ?? '' ) ) {
					continue;
				}

				$expression = (string) ( $filter['expression'] ?? '' );

				if ( 'includingRegex' === ( $filter['operator'] ?? '' ) && 1 === preg_match( '#^\^https\?://\[\^/\]\+(.*)/\?\$$#', $expression, $m ) ) {
					return home_url( (string) preg_replace( '/\\\\(.)/', '$1', $m[1] ) . '/' );
				}

				return $expression;
			}
		}

		return null;
	}

	/**
	 * Search keywords derived from a page's post title (two-word phrases),
	 * falling back to the generic pool when the page has no matching post.
	 *
	 * @param string $page_url The filtered page URL.
	 * @return string[]
	 */
	private static function keywords_from_page( string $page_url ): array {
		$post_id = url_to_postid( $page_url );
		$title   = $post_id > 0 ? (string) get_the_title( $post_id ) : '';

		if ( '' === $title ) {
			return FakeData::GSC_KEYWORDS;
		}

		$words = array_values( array_filter( (array) preg_split( '/\s+/', strtolower( $title ) ) ) );

		if ( count( $words ) < 2 ) {
			return array_merge( array( strtolower( $title ) ), FakeData::GSC_KEYWORDS );
		}

		$phrases    = array( strtolower( $title ) );
		$word_count = count( $words );

		$stop = array( 'a', 'an', 'and', 'as', 'for', 'how', 'in', 'is', 'made', 'of', 'on', 'our', 'that', 'the', 'to', 'what', 'with' );

		for ( $i = 0; $i < $word_count - 1; $i++ ) {
			if ( in_array( $words[ $i ], $stop, true ) || in_array( $words[ $i + 1 ], $stop, true ) ) {
				continue;
			}

			$phrases[] = $words[ $i ] . ' ' . $words[ $i + 1 ];
		}

		return array_merge( $phrases, FakeData::GSC_KEYWORDS );
	}
}