Socket
Socket
Sign inDemoInstall

class-validator-jsonschema

Package Overview
Dependencies
11
Maintainers
1
Versions
24
Alerts
File Explorer

Advanced tools

Install Socket

Detect and block malicious and high-risk dependencies

Install

class-validator-jsonschema

Convert class-validator-decorated classes into JSON schema


Version published
Maintainers
1
Weekly downloads
59,161
decreased by-9.89%

Weekly downloads

Readme

Source

class-validator-jsonschema

codecov npm version

Convert class-validator-decorated classes into OpenAPI-compatible JSON Schema. The aim is to provide a best-effort conversion: since some of the class-validator decorators lack a direct JSON Schema counterpart, the conversion is bound to be somewhat opinionated. To account for this multiple extension points are available.

Installation

npm install class-validator-jsonschema

Note the peer dependency versions in package.json. Try installing a previous major version of class-validator-jsonschema in case you're stuck with older peer dependencies.

Usage

import { IsOptional, IsString, MaxLength } from 'class-validator'
import { validationMetadatasToSchemas } from 'class-validator-jsonschema'

class BlogPost {
  @IsString() id: string

  @IsOptional()
  @MaxLength(20, { each: true })
  tags: string[]
}

const schemas = validationMetadatasToSchemas()
console.log(schemas)

which prints out:

{
  "BlogPost": {
    "properties": {
      "id": {
        "type": "string"
      },
      "tags": {
        "items": {
          "maxLength": 20,
          "type": "string"
        },
        "type": "array"
      }
    },
    "required": ["id"],
    "type": "object"
  }
}

validationMetadatasToSchemas takes an options object as an optional parameter. Check available configuration objects and defaults at options.ts.

Adding and overriding default converters

With options.additionalConverters you can add new validation metadata converters or override the existing ones. Let's say we want to, for example, add a handy description field to each @IsString()-decorated property:

import { ValidationTypes } from 'class-validator'

// ...

const schemas = validationMetadatasToSchemas({
  additionalConverters: {
    [ValidationTypes.IS_STRING]: {
      description: 'A string value',
      type: 'string',
    },
  },
})

which now outputs:

{
  "BlogPost": {
    "properties": {
      "id": {
        "description": "A string value",
        "type": "string"
      }
      // ...
    }
  }
}

An additional converter can also be supplied in form of a function that receives the validation metadata item and global options, outputting a JSON Schema property object (see below for usage):

type SchemaConverter = (
  meta: ValidationMetadata,
  options: IOptions
) => SchemaObject | void

Custom validation classes

class-validator allows you to define custom validation classes. You might for example validate that a string's length is between given two values:

import {
  Validate,
  ValidationArguments,
  ValidatorConstraint,
  ValidatorConstraintInterface,
} from 'class-validator'

// Implementing the validator:

@ValidatorConstraint()
export class CustomTextLength implements ValidatorConstraintInterface {
  validate(text: string, validationArguments: ValidationArguments) {
    const [min, max] = validationArguments.constraints
    return text.length >= min && text.length <= max
  }
}

// ...and putting it to use:

class Post {
  @Validate(CustomTextLength, [0, 11])
  title: string
}

Now to handle your custom validator's JSON Schema conversion include a CustomTextLength converter in options.additionalConverters:

const schemas = validationMetadatasToSchemas({
  additionalConverters: {
    CustomTextLength: (meta) => ({
      maxLength: meta.constraints[1],
      minLength: meta.constraints[0],
      type: 'string',
    }),
  },
})

Decorating with additional properties

Validation classes can also be supplemented with the JSONSchema decorator. JSONSchema can be applied both to classes and individual properties; any given keywords are then merged into the JSON Schema derived from class-validator decorators:

import { JSONSchema } from 'class-validator-jsonschema'

@JSONSchema({
  description: 'A User object',
  example: { id: '123' },
})
class BlogPost {
  @IsString()
  @JSONSchema({
    description: 'User primary key',
    format: 'custom-id',
  })
  id: string
}

Results in the following schema:

{
  "BlogPost": {
    "description": "A User object",
    "example": { "id": "123" },
    "properties": {
      "id": {
        "description": "User primary key",
        "format": "custom-id",
        "type": "string"
      }
    },
    "required": ["id"],
    "type": "object"
  }
}

JSONSchema decorators also flow down from parent classes into inherited validation decorators. Note though that if the inherited class uses JSONSchema to redecorate a property from the parent class, the parent class JSONSchema gets overwritten - i.e. there's no merging logic.

Custom handlers

Alternatively JSONSchema can take a function of type (existingSchema: SchemaObject, options: IOptions) => SchemaObject. The return value of this function is then not automatically merged into existing schema (i.e. the one derived from class-validator decorators). Instead you can handle merging yourself in whichever way is preferred, the idea being that removal of existing keywords and other more complex overwrite scenarios can be implemented here.

@ValidateNested and arrays

class-validator supports validating nested objects via the @ValidateNested decorator. Likewise JSON Schema generation is supported out-of-the-box for nested properties such as

@ValidateNested()
user: UserClass

However, due to limitations in Typescript's reflection system we cannot infer the inner type of a generic class. In effect this means that properties like

@ValidateNested({ each: true })
users: UserClass[]

@ValidateNested()
user: Promise<UserClass>

would resolve to classes Array and Promise in JSON Schema. To work around this limitation we can use @Type from class-transformer to explicitly define the nested property's inner type:

import { Type } from 'class-transformer'
import { validationMetadatasToSchemas } from 'class-validator-jsonschema'
const { defaultMetadataStorage } = require('class-transformer/cjs/storage') // See https://github.com/typestack/class-transformer/issues/563 for alternatives

class User {
  @ValidateNested({ each: true })
  @Type(() => BlogPost) // 1) Explicitly define the nested property type
  blogPosts: BlogPost[]
}

const schemas = validationMetadatasToSchemas({
  classTransformerMetadataStorage: defaultMetadataStorage, // 2) Define class-transformer metadata in options
})

Note also how the classTransformerMetadataStorage option has to be defined for @Type decorator to take effect.

Using a custom validation metadataStorage

Under the hood we grab validation metadata from the default storage returned by class-validator's getMetadataStorage(). In case of a version clash or something you might want to manually pass in the storage:

const schemas = validationMetadatasToSchemas({
  classValidatorMetadataStorage: myCustomMetadataStorage,
})

Limitations

There's no handling for class-validators validation groups or conditional decorator (@ValidateIf) out-of-the-box. The above-mentioned extension methods can be used to fill the gaps if necessary.

The OpenAPI spec doesn't currently support the new JSON Schema draft-06 keywords const and contains. This means that constant value decorators such as @IsEqual() and @ArrayContains() translate to quite complicated schemas. Hopefully in a not too distant future these keywords are adopted into the spec and we'll be able to provide neater conversion.

Handling null values is also tricky since OpenAPI doesn't support JSON Schema's type: null, providing its own nullable keyword instead. The default @IsEmpty() converter for example opts for nullable but you can use type: null instead via options.additionalConverters:

// ...
additionalConverters: {
  [ValidationTypes.IS_EMPTY]: {
    anyOf: [
      {type: 'string', enum: ['']},
      {type: 'null'}
    ]
  }
}

TODO

  • handle skipMissingProperties and @isDefined()
  • decorators for overwriting prop schemas
  • optional property descriptions (e.g. A Base64-encoded string)
  • optional draft-06 keywords

Keywords

FAQs

Last updated on 14 Jan 2023

Did you know?

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts

SocketSocket SOC 2 Logo

Product

  • Package Alerts
  • Integrations
  • Docs
  • Pricing
  • FAQ
  • Roadmap

Stay in touch

Get open source security insights delivered straight into your inbox.


  • Terms
  • Privacy
  • Security

Made with ⚡️ by Socket Inc