Inconsistency between Assert::alpha and Assert::alnum

[zerkms]

The documentation states

alpha($value, $message = '') | Check that a string contains letters only
alnum($value, $message = '') | Check that a string contains letters and digits only

But then implementations are:

    public static function alpha($value, $message = '')
    {
        static::string($value);

        $locale = \setlocale(LC_CTYPE, 0);
        \setlocale(LC_CTYPE, 'C');
        $valid = !\ctype_alpha($value);
        \setlocale(LC_CTYPE, $locale);

        if ($valid) {
            static::reportInvalidArgument(\sprintf(
                $message ?: 'Expected a value to contain only letters. Got: %s',
                static::valueToString($value)
            ));
        }
    }

    public static function alnum($value, $message = '')
    {
        $locale = \setlocale(LC_CTYPE, 0);
        \setlocale(LC_CTYPE, 'C');
        $valid = !\ctype_alnum($value);
        \setlocale(LC_CTYPE, $locale);

        if ($valid) {
            static::reportInvalidArgument(\sprintf(
                $message ?: 'Expected a value to contain letters and digits only. Got: %s',
                static::valueToString($value)
            ));
        }
    }

Note that alpha accepts mixed and checks if it's a string explicitly, but alnum expects it's a string already and does not make a runtime check.

I think both documentation and implementation should be changed to be consistent.

[zerkms]

/cc @Ocramius

[Ocramius]

IMO, changing these would be a BC break: it's worth doing it in a 2.0, but otherwise leave the inconsistency there and we keep the issue Marco Pivetta http://twitter.com/Ocramius http://ocramius.github.com/

…

On Sat, May 30, 2020 at 2:41 AM Ivan Kurnosov ***@***.***> wrote: /cc @Ocramius <https://github.com/Ocramius> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#197 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABFVECGTUG7DO2RG3XGH6DRUBI2FANCNFSM4NOORCRA> .

[zerkms]

Indeed, would be helpful if milestone was added here then.

[BackEndTea]

ctype is funny in that it just returns false if the entered value is not a string. Even an object that could be cast to a string is not.

So the end result should be the same (except for integers between -128 and 255)

[zerkms]

Yep, but isn't the general idea behind assertion libraries (in general) is: we pass you any arbitrary rubbish and if it's not what we expect - throw.

As of now - if you want to check it's an alnum in a variable - you must assert it's a string first.