mb_convert_encoding "\" (backslash) and "~" (tilde) BC breaks to Shift_JIS-2004

[youkidearitai]

Description

The following code:

<?php

$backslash_and_tilde = "~\\";

$sjis2004_backslash_and_tilde = mb_convert_encoding($backslash_and_tilde, "SJIS-2004", "UTF-8");
var_dump(
        bin2hex($sjis2004_backslash_and_tilde),
        $re_convert = mb_convert_encoding($sjis2004_backslash_and_tilde, "UTF-8", "SJIS-2004"),
        bin2hex($re_convert)
);

Resulted in this output:

string(8) "8160815f"
string(6) "〜＼"
string(12) "e3809cefbcbc"

But I expected this output instead:

string(4) "7e5c"
string(5) "‾¥"
string(10) "e280bec2a5"

Similar to #8281 . This behavior is from 5.4 to 8.0 (3v4l: https://3v4l.org/42pAL ). Please keep to Backward Compatible for PHP 8.1.

PHP Version

PHP 8.1.x

Operating System

No response

[cmb69]

Similar to #8281 .

And I think we should fix it like that.

@alexdowad, what do you think?

[alexdowad]

@cmb69 Agreed.

[alexdowad]

@youkidearitai Are you aware of any other Shift-JIS-2004 mappings which have changed?

In this case, your concern is just about Unicode -> JIS mappings, not JIS -> Unicode, is that correct?

[youkidearitai]

@alexdowad Thanks for reply.

Are you aware of any other Shift-JIS-2004 mappings which have changed?

Sorry, I'm not know any changed Shift_JIS-2004 in PHP, But Shift_JIS-2004 is upper compatible of Shift_JIS that normally no any changes.

In this case, your concern is just about Unicode -> JIS mappings, not JIS -> Unicode, is that correct?

Yes this case I concern Unicode -> JIS mappings it is.

[alexdowad]

@youkidearitai Thanks. Hopefully I might prepare a PR later today...

Thanks again.