Specify isA class-string types for falsey context

[herndlm]

The title is misleading already, sorry. Should be more like "Extract type determination logic from is_subclass_of and use it in is_a" :)

Closes phpstan/phpstan#6704

While I initially was working on another solution, the changes in #1039 were finished and I could just completely take them over here. Thank you @arnaud-lb!

is_a and is_subclass_of seem to be having only 2 differences

• is_a has $allow_string false by default, while is_subclass_of has it true by default. See https://www.php.net/manual/en/function.is-a.php and https://www.php.net/manual/en/function.is-subclass-of.php
• is_a also returns true if the object is of the class to check, while is_subclass_of doesn't

Because there is still phpstan/phpstan#6305 open, the second one is fine here. For the first one I just changed the default when taking over the changes.

[herndlm]

I'm gonna extract the common logic and re-use that

[ondrejmirtes]

Rebased it to see the CI results :)

[herndlm]

Should I push a change extracting the type determination? I basically moved the TypeTraverse::map into IsAFunctionTypeSpecifyingHelper::determineType(Type $classType, bool $allowString). What do you think? Unsure about the name still..
The idea is to later add another boolean argument or so that helps fixing phpstan/phpstan#6305

I'm just worried that either one of those extensions receives fixes/changes but the other is not udpated :/

[ondrejmirtes]

Yes, common logic extraction 👍

When fixing something in the common logic, regression tests for both is_a and is_subclass_of should still be added, it doesn't hurt.

[ondrejmirtes]

Please mark this as ready, we're good to go :)

[ondrejmirtes]

Thank you!