mb_{detect/convert}_encoding. Again. Or maybe pdo?

[gzhegow1991]

Description

(For all tests online tool is used: https://onlinephp.io/)

The following code:

<?php

// > this is \PDOException message in Russian, that represents `Server is not responding` (it means - server configuration is not a solution)
$str = base64_decode('U1FMU1RBVEVbSFkwMDBdIFsyMDAyXSDP7uTq6/735e3o5SDt5SDz8fLg7e7i6+Xt7iwg8i7qLiDq7u3l9+376SDq7uzv/P7y5fAg7vLi5fDjIOfg7/Du8SDt4CDv7uTq6/735e3o5S4NCiAoU1FMOiBTRVQgRk9SRUlHTl9LRVlfQ0hFQ0tTPTA7KQ==');

// > and it looks like
// ###
// Warning: Your output contains characters that could not be displayed. Make sure you encode the output when working with special characters or binary data. [Click here for an example on how to do this](https://onlinephp.io/code/utf8-in-the-sandbox)
// SQLSTATE[HY000] [2002] ����������� �� �����������, �.�. �������� ��������� ������ ������ �� �����������.
 (SQL: SET FOREIGN_KEY_CHECKS=0;)
// ###


$mbListEncodings = mb_list_encodings();


$detect = mb_detect_encoding($str);

// PHP_VERSION_ID < 80300 -> 'UTF-8'
// PHP_VERSION_ID >= 80300 -> 'ASCII'
var_dump($detect);


$detect2 = mb_detect_encoding($str, $mbListEncodings, true);
// $detect2 = mb_detect_encoding($str, $mbListEncodings); // > same result, actually isnt, without $strict = true, it may return 'ASCII' if provided below results is not an option, with $strict it returns FALSE then

// PHP_VERSION_ID < 80100 -> 'ISO-8859-1'
// PHP_VERSION_ID >= 80100 -> 'Windows-1252'
var_dump($detect2);


// > accidentally IT WORKS HERE but PHP_VERSION_ID >= 80100
array_unshift($mbListEncodings, 'CP1251');
array_unshift($mbListEncodings, 'Windows-1251');
$detect3 = mb_detect_encoding($str, $mbListEncodings, true);

// PHP_VERSION_ID < 80100 -> 'ISO-8859-1' // > !!! seems as old bug
// PHP_VERSION_ID >= 80100 -> 'Windows-1251'
var_dump($detect3);


$cpDetectedWrong = 'Windows-1252';

$converted = mb_convert_encoding($str, 'UTF-8', $cpDetectedWrong);
$converted_b64 = base64_encode($converted);

var_dump($converted); // string(207) "SQLSTATE[HY000] [2002] Ïîäêëþ÷åíèå íå óñòàíîâëåíî, ò.ê. êîíå÷íûé êîìïüþòåð îòâåðã çàïðîñ íà ïîäêëþ÷åíèå.
 (SQL: SET FOREIGN_KEY_CHECKS=0;)"

var_dump($converted_b64); // string(276) "U1FMU1RBVEVbSFkwMDBdIFsyMDAyXSDDj8Ouw6TDqsOrw77Dt8Olw63DqMOlIMOtw6Ugw7PDscOyw6DDrcOuw6LDq8Olw63Driwgw7Iuw6ouIMOqw67DrcOlw7fDrcO7w6kgw6rDrsOsw6/DvMO+w7LDpcOwIMOuw7LDosOlw7DDoyDDp8Ogw6/DsMOuw7Egw63DoCDDr8Ouw6TDqsOrw77Dt8Olw63DqMOlLg0KIChTUUw6IFNFVCBGT1JFSUdOX0tFWV9DSEVDS1M9MDsp"

But I expected this output instead:

<?php

$detect = mb_detect_encoding($str, mb_list_encodings(), true);
var_dump($detect); // 'Windows-1251'

$cpDectectedCorrect = 'Windows-1251';

$converted = mb_convert_encoding($str, 'UTF-8', $cpDectectedCorrect);
$converted_b64 = base64_encode($converted);

var_dump($converted); // string(207) "SQLSTATE[HY000] [2002] Подключение не установлено, т.к. конечный компьютер отверг запрос на подключение.
 (SQL: SET FOREIGN_KEY_CHECKS=0;)"

var_dump($converted_b64); // string(276) "U1FMU1RBVEVbSFkwMDBdIFsyMDAyXSDQn9C+0LTQutC70Y7Rh9C10L3QuNC1INC90LUg0YPRgdGC0LDQvdC+0LLQu9C10L3Qviwg0YIu0LouINC60L7QvdC10YfQvdGL0Lkg0LrQvtC80L/RjNGO0YLQtdGAINC+0YLQstC10YDQsyDQt9Cw0L/RgNC+0YEg0L3QsCDQv9C+0LTQutC70Y7Rh9C10L3QuNC1Lg0KIChTUUw6IFNFVCBGT1JFSUdOX0tFWV9DSEVDS1M9MDsp"

I've tried using mb_check_encoding()... I've played for few hours with mb_detect_order(), mb_list_encodings()... I've even tried to split known encodings by groups by first letters or their slugs and apply mb_convert_encoding for better detection for each group.

No. Just dont work, and should be fixed like

<?php
set_exception_handler(function ($e) {
   $phpMessage = $e->getMessage();

   if ($e instanceof \PDOException) {
     $isUtf8 = preg_match('//u', $phpMessage) === 1;
     if (! $isUtf8) {
       $isWindows = (strtoupper(substr(PHP_OS, 0, 3)) === 'WIN');
       if ($isWindows) {
         $phpMessage = mb_convert_encoding($phpMessage, 'UTF-8', 'CP1251');
       }
     }
   }

   /// ...code
});

PHP Version

PHP 8.4

Operating System

Windows 10