On the tests for the rules, when we expect an error we check that the return value is an array and that it contains one element. Some times we check the error message, and some other times we don't. But this is not really following any rules.
For each possible error message, we should check the string once. Some other tests variations are ok just checking that there is an error, but the current way to check for that is ugly and repeated many times. It could be a common helper.
On the tests for the rules, when we expect an error we check that the return value is an array and that it contains one element. Some times we check the error message, and some other times we don't. But this is not really following any rules.
For each possible error message, we should check the string once. Some other tests variations are ok just checking that there is an error, but the current way to check for that is ugly and repeated many times. It could be a common helper.