schema is not generated with validation when comment includes Japanese

[manak1]

Hello, Thank you for creating a wonderful package!
I have a problem generating schema with validation using Japanese error message.
Is there any workaround for this problem? 👀

datasource db {
  provider  = "postgres"
}

generator zod {
  provider = "zod-prisma-types"
}

model User {
  id String @id @default(uuid())
  /// @zod.string({ required_error: 'required' })
  email String 
  /// @zod.string({ required_error: '必須項目です' })
  password String
}

Expected output

export const UserSchema = z.object({
  id: z.string().uuid(),
  email: z.string({ required_error: 'required' }),
  password: z.string({ required_error: '必須項目です' }),
})

Actual output

export const UserSchema = z.object({
  id: z.string().uuid(),
  email: z.string({ required_error: 'required' }),
  /**
   * ({ required_error: '必須項目です' })
   */
  password: z.string(),
})

[chrishoermann]

@manak1 thanks for pointing this out. since I'm using regexes quite heavily to validate if the user input is valid this problem happens becaues I do not check for Japanese characters since I do not speak nor can I read your language - so sadly Ihave no idea how to check for these characters 😢

maybe you could open a PR or provide some info on how to check for Japanese in a regex?

[manak1]

@chrishoermann
Thank you for your response!
I'm not really familiar with regex but I will check it out 👀

[dbtl88]

Not sure how helpful this still might be, but this should work to match all Japanese text:

[\p{sc=Hira}\p{sc=Kana}\p{sc=Han}]

Which is the same as (using the combined hiragana / katakana unicode script identifier):

[\p{sc=Hrkt}\p{sc=Han}]

Explanations can be found here:
how to do regex on Unicode script identifiers (section 1.2.6)
the list of Unicode script identifiers

Note I've also included the 'Han' script identifier, i.e. all of the possible Kanji.

EDIT: I've validated through a regex validator, and the "Hrkt" identifier doesn't work on ECMAScript regex, so please use the first version, with the separate hiragana / katakana identifiers.

[chrishoermann]

I implemented the behaviour as suggested by @dbtl88 and according to my tests it works. Would be nice if someone could confirm this with a more extended range of japanese characters than the ones I used in my tests.