<?php
/**
 * Path normalization for redirect matching.
 *
 * @package Unbreak
 */

namespace Unbreak\Redirects;

use Unbreak\Support\Url;

defined( 'ABSPATH' ) || exit;

/**
 * Lowercase, decoded, no trailing slash; query handling depends on the "redirect_query" setting.
 */
final class Path {

	/**
	 * Normalize a source path, a same-site URL or a request URI. Returns null for other hosts or junk.
	 *
	 * @param string $input      Path ("/Old-Page/?a=1") or absolute same-site URL.
	 * @param bool   $keep_query Keep the query string.
	 */
	public static function normalize( string $input, bool $keep_query = false ): ?string {
		$input = trim( $input );
		if ( '' === $input ) {
			return null;
		}
		if ( preg_match( '#^https?://#i', $input ) ) {
			if ( ! Url::is_internal( $input ) ) {
				return null;
			}
			$parts = wp_parse_url( $input );
			$input = ( $parts['path'] ?? '/' ) . ( isset( $parts['query'] ) ? '?' . $parts['query'] : '' );
		}
		if ( '/' !== $input[0] ) {
			$input = '/' . $input;
		}
		$hash = strpos( $input, '#' );
		if ( false !== $hash ) {
			$input = substr( $input, 0, $hash );
		}
		$query = '';
		$qpos  = strpos( $input, '?' );
		if ( false !== $qpos ) {
			$query = substr( $input, $qpos + 1 );
			$input = substr( $input, 0, $qpos );
		}
		$path = strtolower( rawurldecode( $input ) );
		$path = (string) preg_replace( '#/+#', '/', $path );
		if ( strlen( $path ) > 1 ) {
			$path = rtrim( $path, '/' );
		}
		if ( '' === $path || preg_match( '/[\x00-\x1F\x7F]/', $path ) ) {
			return null;
		}
		return $keep_query && '' !== $query ? $path . '?' . $query : $path;
	}

	/**
	 * Path part of a normalized key (without query).
	 *
	 * @param string $key Normalized key.
	 */
	public static function strip_query( string $key ): string {
		$qpos = strpos( $key, '?' );
		return false === $qpos ? $key : substr( $key, 0, $qpos );
	}
}