special vs built-in schema + custom validation?

[jindong-zhannng]

Hi folks,

I just found basically both of these 2 ways can do the same thing. e.g.

special

const NumericStringSchema = v.special<string>((i) => typeof i === 'string' && !isNaN(Number(i)))

built-in schema + custom

const NumericStringSchema = v.string([v.custom((i) => !isNaN(Number(i)))])

Which way is more preferred?

[fabian-hiller]

I recommend using string with custom. special is for more special or custom data types. What exactly are you trying to validate? Maybe it makes sense to add a new action validation so you can write the following:

const NumericStringSchema = v.string([v.numeric()]);

[jindong-zhannng]

My use case is parsing string-based form inputs to typed data.
For example if I have a input box that should only allow number inputs. When I use transform api to do the parsing, I'd like to validate it with v.string([v.numeric()]) firstly, then transform string inputs to numbers. So that I can gain meaningful error messages from the validation if input is invalid.

Maybe it makes sense to add a new action validation

Yes a numeric validation for strings may makes sense if other people like it also.
But is there an api I can use to define my own validation actions? The custom api? As I mentioned above, in my use case I may need more various validations to match different input types. I hope I can get more meaningful errors when validation fails. e.g. by overriding context or expected properties, inserting received value into error message. However the custom api seems unable to do these now.

[fabian-hiller]

Yes a numeric validation for strings may makes sense if other people like it also.

I like this idea and think we should add it. Feel free to create an issue or PR that merges it into the branch of #502.

I hope I can get more meaningful errors when validation fails. e.g. by overriding context or expected properties

This will require you to write your own validation action. But it is quite easy, you can just copy the source code of a similar validation action from our source code and change it to your needs.