<?php
/**
 * Collation names a schema from another server can be created under here.
 *
 * @package Shiftora\Support\Db
 */

declare(strict_types=1);

namespace Shiftora\Support\Db;

// Prevent direct file access.
if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * A CREATE TABLE carries its source server's collation names, which this
 * server may not know (MySQL 8's utf8mb4_0900_ai_ci, MariaDB's
 * utf8mb4_uca1400_ai_ci). dbDelta swallows the "Unknown collation" error and
 * the table is never created.
 *
 * An unknown name becomes the closest collation of the same character set the
 * server has; only sort order can change.
 */
final class CollationMap {

	/**
	 * Closest stand-ins by character set, best first.
	 *
	 * @var array<string, string[]>
	 */
	private const FALLBACKS = [
		'utf8mb4' => [ 'utf8mb4_unicode_520_ci', 'utf8mb4_unicode_ci', 'utf8mb4_general_ci' ],
		'utf8mb3' => [ 'utf8mb3_unicode_520_ci', 'utf8mb3_unicode_ci', 'utf8mb3_general_ci', 'utf8_unicode_520_ci', 'utf8_unicode_ci', 'utf8_general_ci' ],
		'utf8'    => [ 'utf8_unicode_520_ci', 'utf8_unicode_ci', 'utf8_general_ci', 'utf8mb3_unicode_ci', 'utf8mb3_general_ci' ],
	];

	/** Rewrites the collations in a CREATE TABLE statement that this server does not support. */
	public static function adapt( \wpdb $wpdb, string $sql ): string {
		if ( ! preg_match_all( '/\bCOLLATE\s*=?\s*`?([A-Za-z0-9_]+)/i', $sql, $found ) ) {
			return $sql;
		}

		$known = self::known( $wpdb );
		// If the server will not list its collations, leave the statement as is.
		if ( [] === $known ) {
			return $sql;
		}

		$map = [];
		foreach ( array_unique( $found[1] ) as $name ) {
			if ( isset( $known[ strtolower( $name ) ] ) ) {
				continue;
			}
			$stand_in = self::stand_in( $name, $known );
			if ( null !== $stand_in ) {
				$map[ strtolower( $name ) ] = $stand_in;
			}
		}

		if ( [] === $map ) {
			return $sql;
		}

		return (string) preg_replace_callback(
			'/\b(COLLATE\s*=?\s*`?)([A-Za-z0-9_]+)/i',
			static fn( array $m ): string => $m[1] . ( $map[ strtolower( $m[2] ) ] ?? $m[2] ),
			$sql
		);
	}

	/** The closest supported collation for an unknown one, or null when none matches its charset. */
	private static function stand_in( string $name, array $known ): ?string {
		$charset    = strtolower( (string) strstr( $name, '_', true ) );
		$candidates = self::FALLBACKS[ $charset ] ?? [ $charset . '_general_ci', $charset . '_bin' ];

		foreach ( $candidates as $candidate ) {
			if ( isset( $known[ $candidate ] ) ) {
				return $candidate;
			}
		}

		return null;
	}

	/**
	 * Collations this server supports, read once per request.
	 *
	 * @return array<string, true>
	 */
	public static function known( \wpdb $wpdb ): array {
		static $known = null;

		if ( null === $known ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery -- server metadata, nothing to cache
			$names = $wpdb->get_col( 'SHOW COLLATION' );
			$known = is_array( $names ) && [] !== $names
				? array_fill_keys( array_map( 'strtolower', array_map( 'strval', $names ) ), true )
				: [];
		}

		return $known;
	}
}