math_cms/app/Support/TextEncoding.php

<?php

namespace App\Support;

class TextEncoding
{
    /**
     * Normalize unknown-encoding text to valid UTF-8 for safe JSON/JS rendering.
     */
    public static function toUtf8(string $content): string
    {
        // Remove null bytes that can break downstream processing.
        $content = str_replace("\0", '', $content);

        if ($content === '') {
            return $content;
        }

        if (function_exists('mb_check_encoding') && mb_check_encoding($content, 'UTF-8')) {
            return $content;
        }

        $candidates = [
            'UTF-8',
            'GB18030',
            'GBK',
            'GB2312',
            'BIG5',
            'ISO-8859-1',
            'Windows-1252',
        ];

        if (function_exists('mb_detect_encoding') && function_exists('mb_convert_encoding')) {
            $detected = mb_detect_encoding($content, $candidates, true);
            if (is_string($detected) && $detected !== '' && strtoupper($detected) !== 'UTF-8') {
                $converted = @mb_convert_encoding($content, 'UTF-8', $detected);
                if (is_string($converted) && $converted !== '' && mb_check_encoding($converted, 'UTF-8')) {
                    return $converted;
                }
            }
        }

        // Fallback: best-effort iconv conversion (drops invalid bytes).
        foreach (['GB18030', 'GBK', 'GB2312', 'BIG5', 'ISO-8859-1', 'Windows-1252'] as $from) {
            $converted = @iconv($from, 'UTF-8//IGNORE', $content);
            if (is_string($converted) && $converted !== '' && function_exists('mb_check_encoding') && mb_check_encoding($converted, 'UTF-8')) {
                return $converted;
            }
        }

        // Last resort: strip invalid sequences.
        $converted = @iconv('UTF-8', 'UTF-8//IGNORE', $content);
        return is_string($converted) ? $converted : '';
    }
}