oneOf, notOneOf swallowing multiple errors

[akmjenkins]

The following gives me a ValidatorError with three errors:

        yup
          .string()
          .length('10')
          .min('10')
          .required()
          .validateSync('', { abortEarly: false });

The following gives me a ValidationError with only one error:

        yup
          .string()
          .oneOf(['1234'])
          .length('10')
          .min('10')
          .required()
          .validateSync('', { abortEarly: false });

Seems counterintuitive.

[jquense]

seems like a bug

[krnlde]

Cool

[akmjenkins]

@jquense It seems like this "bug" was done intentionally, by reading the code and here.

It's not immediately clear to me why a subset of validations need/should be run first? But in any case, if a typeError, blacklist, or whitelist error are encountered, then no other validations are run, regardless of abortEarly. The behaviour should probably be documented if it can't be changed.

[brianle1301]

@akmjenkins Whitelisted and blacklisted values are checked first, this is the same as Joi. When these values are met, validation is stopped without further testing.
I think this was intentional @jquense . If it's not, then it should be.
Joi's source code reference: https://github.com/hapijs/joi/blob/master/lib/validator.js#L275-L284

Edit Sorry my bad. Joi.valid() doesn't stop further errors from being collected. :(

[szamanr]

why is this closed? here's my use case:

password: Yup.string()
  .required("Required")
  .notOneOf([email], "You cannot use your email address as password."),

when the field is left blank, the validation says "You cannot use your email address as password." instead of "Required"

[IBonkI]

why is this closed? here's my use case:

password: Yup.string()
  .required("Required")
  .notOneOf([email], "You cannot use your email address as password."),

when the field is left blank, the validation says "You cannot use your email address as password." instead of "Required"

i also got this issue

[akmjenkins]

@jquense this is still a bug, I suspected it got closed due to inactivity. I just opened a PR as the (dis)allowed value tests being run first are not documented behavior in the README or the test suite. Seems like a pretty straightforward change!

[nisarg2023]

why is this closed? here's my use case:

password: Yup.string()
  .required("Required")
  .notOneOf([email], ({value})=>value==="""?(''Required"):("You cannot use your email address as password.")),

try this

[mi-mazouz]

Same issue

....
confirmedNewPassword: yup
      .string()
      .matches(/(?=.{8,})/, {
        name: "MIN_LENGTH_8",
      })
      .matches(/(?=.*[0-9])/, {
        name: "AT_LEAST_ONE_DIGIT",
      })
      .matches(/(?=.*[a-z])/, {
        name: "AT_LEAST_ONE_LOWER_CASE",
      })
      .matches(/(?=.*[A-Z])/, {
        name: "AT_LEAST_ONE_UPPER_CASE",
      })
      .matches(/(?=.*[!@#$%^&*])/, {
        name: "AT_LEAST_ONE_SPECIAL_CHAR",
      })
      .oneOf([yup.ref("newPassword")], "Passwords must be the same")
      .required()
...

Setting abortEarly to false has no effect