Validation messages
Validation messages reside in the separate schema, decoupled from the validation logic. The reasons for the decoupling are the following:
- Establish a clear separation of concerns.
- Prevent from garbaging complex validation rules with the messages.
- Have a single validation schema associated with any amount of localized messages schema.
Any message can be a plain string, or a resolver function that returns a string.
Resolvers are useful when a validation message derives not only from the field's validity, but also from its state, fields or a form. This way you can include the current value of a field in the error message, for example.
type MessageResolver = string | (params) => string
Parameter name | Type | Description |
value | any | The value of a field. |
fieldProps | Object | Props of a field. |
fields | Object | State of all fields. |
form | Object | Form component reference. |
extra | Object | Extra parameters passed in the async validation payload. |
Validation messages schema supports the following message types:
Type | Description |
missing | A message for an empty required field. |
invalid | A message for an invalid field. |
async |
Selector | Description |
name | Name-specific messages. |
type | Type-specific messages. |
general | General messages. Those are used as fallback values in case more specific messages are not present. |
type ValidationMessages = {
[selectorValue: string]: MessageResolver,
},
}
It is possible to associate validation rules with the specific validation messages. To do that, provide the
{ ruleName: message } map as the value of the rule key in the respective selector:export default {
type: {
password: {
missing: 'Please provide the password',
invalid: 'The passwords is invalid',
rule: {
minLength: 'Password must be at least 6 characters long',
},
},
},
}
Note: Named resolver must have the corresponding validation rule with the same name in order to resolve. Otherwise, the closest validation message will be returned by the resolver. Considering the example above, ifminLengthrule doesn't exist and thetype=["password"]field is invalid, the closestinvalidresolver will be used (which istype.password.invalidin this case).
Whenever the resolver cannot resolve it would attempt to grab the closest resolver recursively and resolve it instead. Fallback resolvers are iterated by their specificity, which can be presented as follows:
Here is the resolver paths (listed by priority, from highest to lowest):
- 1.
messages.name[fieldName][ruleName] - 2.
messages.name[fieldName].invalid/messages.name[fieldName].missing - 3.
messages.type[fieldType][ruleName] - 4.
messages.type[fieldType]invalid/messages.type[fieldType].missing - 5.
messages.general.invalid/messages.general.missing
Consider the following validation messages:
export default {
general: {
invalid: 'General invalid message',
},
type: {
email: {
invalid: 'E-mail is invalid',
},
},
name: {
userEmail: {
invalid: 'User e-mail is invalid',
rule: {
includesAt: 'E-mail must include "@" character',
},
},
},
}
And the following component layout:
<Form>
<Input
type="email"
name="userEmail"
value="foo" />
</Form>
With the given scenario the
includesAt validation rule would reject, marking the field as unexpected. This must be reflected in the UI using the validation messages schema.This is the priority sequence in which resolvers will attempt to resolve the validation message:
- 1.
name.userEmail.rule.includesAt - 2.
name.userEmail.invalid - 3.
type.email.invalid - 4.
general.invalid
The validation message resolves as soon as the resolver returns the value. The same sequence applies for the type-specific named resolvers.
Last modified 4yr ago