[6.x] Fix Str@contains for PHP 8.0.1

[rodrigopedra]

Fix issue described in #35891.

The issue only happens when running PHP >= 8.0.1. if the $haystack parameter when calling mb_strpos is not a string, PHP will throw and error.

After fixing it one test started failing. I then fixed Illuminate\Database\Query\Builder@where to explicitly cast the $haystack parameter to a string when calling Str@contains.

It was failing on the last assertion of the Illuminate\Tests\Database\DatabaseQueryBuilderTest@testMySqlWrappingJsonWithBoolean test.

Issue was, besides the Builder@where docblock accept \Closure|string|array the assertion was passing a Illuminate\Database\Query\Expression object as the column name on ->where() call.

I didn't fix the docblock for the Builder@where on this PR, as I think it should be fixed in a separate PR.

One alternative approach is to always cast the $haystack to a string inside the Str@contains implementation.

I prefer the implementation I sent in this PR (return false if not a string) instead, as it conforms with the method current dockblock and with the "Principle of least astonishment".

But I can update the PR if you prefer the alternative implementation.

[GrahamCampbell]

Note for Taylor: This PR fixes crashes on PHP 8.0.1 because the mb_strpos function now actually wants a genuine string as input, however Illuminate\Database\Query\Builder@where invokes this with an object with a __toString. I think this is likely a regression that should be fixed upstream, but until then (or in the case they don't want to fix it, because they actually want this new behaviour), we need to work around it.

[GrahamCampbell]

@rodrigopedra Can you raise this change at bugs.php.net as a bug. It seems odd this is the only function with that change, and laravel is not calling the function with strict types enabled, so casting should take place at the call site already.

[rodrigopedra]

Thanks for the review @GrahamCampbell . Will open the bug for PHP tomorrow.

[rodrigopedra]

Hey @GrahamCampbell ,

Before sending the bug to PHP I installed and tested it on PHP 8.0.1, and it actually works:

https://3v4l.org/bh9aV

Even when $haystack is null.

Maybe who opened #35891 have strict types enabled, or the issue is somewhere else.

I am closing this, as by reverting the unneeded cast, the only change would be one test assertion of a Stringable object, which most of the other tests don't use (it is only used to test the Str@is method).

Thanks for your and @crynobone 's time, and please apologize for sending this based only on that issue description.

[rodrigopedra]

If both of you think it is worth, I can send a new PR just adding assertions to test Stringable instances for all the tests in that test class.