Awesome

Introduction

DTO Classes is a TypeScript library for modelling data transfer objects in HTTP JSON APIs.

It gives you the following, a bundle I've found missing for TypeScript/Node:

• Class-based schemas that serialize and deserialize:
  • Parse/validate JSON to internal objects
  • Format internal objects to JSON
• Static types by default without an additional infer call
• Custom validation by adding methods to a schema class
• Simple way to access additional context (eg the request object) when parsing
• Async parsing & formatting to play nice with ORMs
• An API broadly similar to OpenAPI and JSON Schema

Example:

import { DTObject, ArrayField, BooleanField, StringField, DateTimeField } from "dto-classes";

class UserDto extends DTObject {
    name = StringField.bind()
    nickName = StringField.bind({ required: false })
    birthday = DateTimeField.bind()
    active = BooleanField.bind({ default: true })
    hobbies = ArrayField.bind({ items: StringField.bind() })
    favoriteColor = StringField.bind({ allowNull: true })
}

const userDto = await UserDto.parse({
    name: "Michael Scott",
    birthday: '1962-08-16',
    hobbies: ["Comedy", "Paper"],
    favoriteColor: "Red"
});

VS Code:

Table of Contents

• Introduction
• Table of Contents
• Installation
  • From npm
  • Config
• Basic Usage
  • Parsing & Validating Objects
  • Formatting Objects
• Fields
  • StringField: string
  • BooleanField: boolean
  • NumberField: number
  • DateTimeField: Date
  • ArrayField: Array<T>
  • CombineField: oneOf, anyOf
  • Nested Objects: DTObject
• Error Handling
• Custom Parsing/Validation
• Custom Formatting
• Recursive Objects
• Standalone Fields
• NestJS
  • Simple Validation Pipe
  • Validation Pipe with Request Context

Installation

TypeScript 4.5+ is required.

From npm

npm install dto-classes

Config

You'll get more accurate type hints with strict set to true in your tsconfig.json:

{
  "compilerOptions": {
    // ...
    "strict": true
    // ...
}

If that's not practical, you'll still get useful type hints by setting strictNullChecks to true:

{
  "compilerOptions": {
    // ...
    "strictNullChecks": true
    // ...
}

Basic Usage

The library handles both parsing, the process of transforming inputs to the most relevant types, and validating, the process of ensuring values meet the correct criteria.

This aligns with the robustness principle. When consuming an input for an age field, most applications will want the string "25" auto-converted to the number 25. However, you can override this default behavior with your own custom NumberField.

Parsing & Validating Objects

Let's start by defining some schema classes. Extend the DTObject class and define its fields:

import { DTObject, StringField, DateTimeField } from 'dto-classes';

class DirectorDto extends DTObject {
    name = StringField.bind()
}

class MovieDto extends DTObject {
    title = StringField.bind()
    releaseDate = DateTimeField.bind()
    director = DirectorDto.bind(),
    genre = StringField.bind({required: false})
}

There's some incoming data:

const data = {
    "title": "The Departed",
    "releaseDate": '2006-10-06',
    "director": {"name": "Martin Scorsese"}
}

We can coerce and validate the data by calling the static method parse(data) which will return a newly created DTO instance:

const movieDto = await MovieDto.parse(data);

If it succeeds, it will return a strongly typed instance of the class. If it fails, it will raise a validation error:

import { ParseError } from "dto-classes";

try {
    const movieDto = await MovieDto.parse(data);
} catch (error) {
    if (error instanceof ParseError) {
        // 
    }
}

For incoming PATCH requests, the convention is to make all fields optional, even if they'd otherwise be required.

You can pass partial: true for validation to succeed in these scenarios:

const data = {
    "releaseDate": '2006-10-06'
}

const movieDto = await MovieDto.parse(data, {partial: true});

Formatting Objects

You can also format internal data types to JSON data that can be returned in an HTTP response.

A common example is model instances originating from an ORM:

const movie = await repo.fetchMovie(45).join('director')
const jsonData = await MovieDto.format(movie);
return HttpResponse(jsonData);

Special types, like JavaScript's Date object, will be converted to JSON compatible output:

{
    "title": "The Departed",
    "releaseDate": "2006-10-06",
    "director": {"name": "Martin Scorsese"}
}

Any internal properties not specified will be ommitted from the formatted output.

Fields

Fields handle converting between primitive values and internal datatypes. They also deal with validating input values. They are attached to a DTObject using the bind(options) static method.

All field types accept some core options:

interface BaseFieldOptions {
    required?: boolean;  
    allowNull?: boolean;
    readOnly?: boolean;  
    writeOnly?: boolean; 
    default?: any;
    partial?: boolean;
    formatSource?: string;
    ignoreInput?: boolean;
    context?: {[key: string]: any};
}

StringField: string

• Parses input to strings. Coerces numbers, other types invalid.
• Formats all value types to strings.

interface StringOptions extends BaseFieldOptions {
    allowBlank?: boolean;
    trimWhitespace?: boolean;
    maxLength?: number;
    minLength?: number;
    pattern?: RegExp,
    format?: 'email' | 'url'
}

BooleanField: boolean

• Parses input to booleans. Coerces certain bool-y strings. Other types invalid.
• Formats values to booleans.

Truthy inputs:

['t', 'T', 'y', 'Y', 'yes', 'Yes', 'YES', 'true', 'True', 'TRUE', 'on', 'On', 'ON', '1', 1, true]

Falsey inputs:

['f', 'F', 'n', 'N', 'no', 'No', 'NO', 'false', 'False', 'FALSE', 'off', 'Off', 'OFF', '0', 0, 0.0, false]

NumberField: number

• Parses input to numbers. Coerces numeric strings. Other types invalid.
• Formats values to numbers.

interface NumberOptions extends BaseFieldOptions {
    maxValue?: number;
    minValue?: number;
}

DateTimeField: Date

• Parses input to Date instances. Coercing date-ish strings using Date.parse().
• Formats values to strings with Date.toISOString().

interface DateTimeFieldOptions extends BaseFieldOptions {
    maxDate?: Date;
    minDate?: Date;
}

ArrayField: Array<T>

• Parses and formats a list of fields or nested objects.

interface ArrayOptions extends BaseFieldOptions {
    items: BaseField | DTObject;
    maxLength?: number;
    minLength?: number;
}

Examples:

class ActionDto extends DTObject {
    name = String.bind()
    timestamp = DateTimeField.bind()
}

class UserDto extends DTObject {
    actions = ArrayField.bind({ items: ActionDto.bind() })
    emailAddresses = ArrayField.bind({ items: StringField.bind() })
}

CombineField: oneOf, anyOf

Fields or objects can be composed together, using JSON Schema's oneOf or anyOf to parse & validate with OR or XOR logic. That is, the input must match at least one (anyOf) or exactly one (oneOf) of the specified sub-fields.

interface CombineField extends BaseFieldOptions {
    anyOf?: Array<BaseField | DTObject>;
    oneOf?: Array<BaseField | DTObject>;
}

Example: oneOf

import { CombineField } from "dto-classes";

// friendSchema must match a person or dog object, but not both

class Person extends DTObject {
    name = StringField.bind()
    hasTwoLegs = BooleanField.bind()
}

class Dog extends DTObject {
    name = StringField.bind()
    hasFourLegs = BooleanField.bind()
}

const friendSchema = new CombineField({
    oneOf: [
        Person.bind(),
        Dog.bind()
    ]
});

Example: anyOf

// Quantity must be between 1-5 or 50-100

class InventoryItem extends DTObject {
    quantity = CombineField.bind({
        anyOf: [
            NumberField.bind({ minValue: 1, maxValue: 5 }),
            NumberField.bind({ minValue: 50, maxValue: 100 })
        ]
    })
}

Nested Objects: DTObject

DTObject classes can be nested under parent DTObject classes and configured with the same core BaseFieldOptions:

import { DTObject, StringField, DateTimeField } from 'dto-classes';


class Plot extends DTObject {
    content: StringField.bind()
}

class MovieDto extends DTObject {
    title = StringField.bind()
    plot = Plot.bind({required: false, allowNull: true})
}

Error Handling

If parsing fails for any reason -- the input data could not be parsed or a validation constraint failed -- a ParseError is thrown.

The error can be inspected:

class ParseError extends Error {
  issues: ValidationIssue[];
}

interface ValidationIssue {
    path: string[]; // path to the field that raised the error
    message: string; // English description of the problem
}

Example:

import { ParseError } from "dto-classes";


class DirectorDto extends DTObject {
    name = StringField.bind()
}

class MovieDto extends DTObject {
    title = StringField.bind()
    director = DirectorDto.bind(),
}

try {
    const movieDto = await MovieDto.parse({
        title: 'Clifford', 
        director: {}
    });
} catch (error) {
    if (error instanceof ParseError) {
        console.log(error.issues);
        /* [
            {
                "path": ["director", "name"],
                "message": "This field is required"
            }
        ] */
    }
}

Custom Parsing/Validation

For custom validation and rules that must examine the whole object, methods can be added to the DTObject class.

To run the logic after coercion, use the @AfterParse decorator.

import { AfterParse, BeforeParse, ParseError } from "dto-classes";

class MovieDto extends DTObject {
    title = StringField.bind()
    director = DirectorDto.bind()

    @AfterParse()
    rejectBadTitles() {
        if (this.title == 'Yet Another Superhero Movie') {
            throw new ParseError('No thanks');
        }
    }
}

The method can modify the object as well:

import { AfterParse, BeforeParse, ParseError } from "dto-classes";

class MovieDto extends DTObject {
    title = StringField.bind()
    director = DirectorDto.bind()

    @AfterParse()
    makeTitleExciting() {
        this.title = this.title + '!!';
    }
}

Custom Formatting

Override the static format method to apply custom formatting.

import { AfterParse, BeforeParse, ParseError } from "dto-classes";

class MovieDto extends DTObject {
    title = StringField.bind()
    director = DirectorDto.bind()

    static format(value: any) {
        const formatted = super.format(value);
        formatted['genre'] = formatted['director']['name'].includes("Mike Judge") ? 'comedy' : 'drama';
        return formatted;
    }
}

A nicer way to expose computed values is to use the @Format decorator. The single argument (e.g. "obj") will always be the full object initially passed to the static format() method:

class Person extends DTObject {
    firstName = StringField.bind()
    lastName = StringField.bind()

    @Format()
    fullName(obj: any) {
        return `${obj.firstName} ${obj.lastName}`;
    }
}

const formattedData = await Person.format({
    firstName: 'George',
    lastName: 'Washington',
});

expect(formattedData).toEqual({ 
    firstName: 'George', 
    lastName: 'Washington', 
    fullName: 'George Washington' 
});

You can customize the formatted field name by passing {fieldName: string} to the decorator:

class Person extends DTObject {
    firstName = StringField.bind()
    lastName = StringField.bind()

    @Format({fieldName: 'fullName'})
    makeFullName(obj: any) {
        return `${obj.firstName} ${obj.lastName}`;
    }
}

/*
{ 
    firstName: 'George', 
    lastName: 'Washington', 
    fullName: 'George Washington' 
}
*/

Recursive Objects

To prevent recursion errors (eg "Maximum call stack size exceeded"), wrap nested self-refrencing objects in a Recursive call:

import { ArrayField, Rescursive } from "dto-classes";

class MovieDto extends DTObject {
    title = StringField.bind()
    director = DirectorDto.bind()
    sequels: ArrayField({items: Recursive(MovieDto)})
}

Standalone Fields

It's possible to use fields outside of DTObject schemas:

import { DateTimeField } from "dto-classes";

const pastOrPresentday = DateTimeField.parse('2022-12-25', {maxDate: new Date()});

You can also create re-usable field schemas by calling the instance method parseValue():

const pastOrPresentSchema = new DateTimeField({maxDate: new Date()});

pastOrPresentSchema.parseValue('2021-04-16');
pastOrPresentSchema.parseValue('2015-10-23');

NestJS

DTObject classes can integrate easily with NestJS global pipes.

Two ready-to-use examples are included.

Simple Validation Pipe

Copy the pipe in nestjs-examples/dto-validation-pipe.ts to your project.

async function bootstrap() {
  const app = await NestFactory.create(AppModule);
  app.useGlobalPipes(new DTOValidationPipe());
  await app.listen(3000);
}
bootstrap();

Validation Pipe with Request Context

To implement more complex validation it's often useful to be able to access the current HTTP request object. For example, knowing the current user could affect whether validation succeeds. Or maybe you want to implement a hidden field that always returns the current user.

Copy the pipe in nestjs-examples/dto-context-validation-pipe.ts to your project.

Each request will construct its own pipe with the current request object so the pipe must be configured as a provider:

@Module({
  imports: [
    UsersModule
  ],
  controllers: [],
  providers: [
    {
        provide: APP_PIPE,
        useClass: DTOContextValidationPipe,
    }
  ],
})
export class AppModule { }

You can now easily set up a hidden input field that always returns the authenticated user of the request:

import { BaseField, BaseFieldOptions, ParseReturnType } from 'dto-classes';
import { User } from 'path-to-entities/user.entity';


export class CurrentUserField<T extends BaseFieldOptions> extends BaseField {
    _options: T;

    constructor(options?: T) {
        super(options);
        this._options.ignoreInput = true;
    }

    public async parseValue(value: any): ParseReturnType<User, T> {
        return this.getDefaultParseValue();
    }

    async getDefaultParseValue(): Promise<any> {
        const user = await this._context.request.fetchUser();
        return user;
    }
}