<?php
/**
 * Sanitizers for values that legitimately contain percent signs.
 *
 * WordPress's sanitize_text_field() and sanitize_textarea_field() strip
 * percent-encoded octets — preg_replace( '/%[a-f0-9]{2}/i', '', $s ). That is
 * correct for user text, but it silently destroys any template variable whose
 * first two characters after the % are hex digits:
 *
 *   %category%  ->  tegory%   (%ca is a valid octet)
 *   %date%      ->  te%       (%da is a valid octet)
 *
 * while %title%, %author% and %sitename% pass through untouched. The result is
 * a template that works for most variables and quietly breaks for two of them,
 * which is far worse than failing outright.
 *
 * So template strings are sanitized with the tokens held out of the way.
 *
 * @package AiRankSuite
 */

declare(strict_types=1);

namespace AiRankSuite\Support;

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

final class Sanitize {

	/** Matches %token% — letters, digits and underscores only. */
	private const TOKEN = '/%([a-z0-9_]+)%/i';

	/**
	 * Sanitize a single-line template, preserving %variable% tokens.
	 */
	public static function template( string $value ): string {
		return self::guarded( $value, static fn( string $s ): string => sanitize_text_field( $s ) );
	}

	/**
	 * Sanitize a multi-line template, preserving %variable% tokens.
	 */
	public static function template_multiline( string $value ): string {
		return self::guarded( $value, static fn( string $s ): string => sanitize_textarea_field( $s ) );
	}

	/**
	 * Run $sanitizer with every %token% swapped for an inert placeholder, then
	 * put the tokens back.
	 *
	 * The placeholder uses a NUL-delimited marker: it cannot appear in real
	 * input (sanitization strips control characters, and any that survived
	 * would not match the restore pattern), and it contains no percent sign,
	 * so the sanitizer has nothing to strip.
	 *
	 * @param callable(string): string $sanitizer WordPress sanitizer to apply.
	 */
	private static function guarded( string $value, callable $sanitizer ): string {
		$tokens = array();

		$masked = (string) preg_replace_callback(
			self::TOKEN,
			static function ( array $m ) use ( &$tokens ): string {
				$index            = count( $tokens );
				$tokens[ $index ] = $m[0];
				return 'AIRSTOKEN' . $index . 'ENDTOKEN';
			},
			$value
		);

		$clean = (string) $sanitizer( $masked );

		return (string) preg_replace_callback(
			'/AIRSTOKEN(\d+)ENDTOKEN/',
			static function ( array $m ) use ( $tokens ): string {
				// A placeholder with no matching token can only come from input
				// that already looked like one — leave it as literal text.
				return $tokens[ (int) $m[1] ] ?? $m[0];
			},
			$clean
		);
	}
}