Page Contents

Synopsis

Generates models and types from an OpenAPI spec for a LoopBack application. It supports both server and client sides. Controllers are generated for the server side while service proxies and datasources with loopback-connector-openapi are created for the client side.

This command allows us to generate skeleton implementations of an OpenAPI spec and/or strongly-typed service proxies to access the endpoint conforming to the spec.

lb4 openapi [<url>] [options]

Options

  • --url: URL or file path of the OpenAPI spec.
  • --validate: Validate the OpenAPI spec. Default: false.
  • --promote-anonymous-schemas: Promote anonymous schemas as models classes. Default: false.
  • --client: Generate client-side service proxies for the OpenAPI spec Default: false.
  • --datasource: A valid datasource name
  • --positional: A flag to control if service methods use positional parameters or an object with named properties. Default: true.

Arguments

<url>: URL or file path of the OpenAPI spec. Type: String. Required: false.

Supported OpenAPI spec versions

Please note Swagger 2.0 specs are converted to OpenAPI 3.0 internally using swagger2openapi.

Interactive Prompts

The tool will prompt you for:

  • Please select the datasource You can select an existing datasource with openapi connector when --client is specified.
  • URL or file path of the OpenAPI spec If the url or file path is supplied from the command line, the prompt is skipped.
  • Select controllers to be generated You can select what controllers will be generated based on OpenAPI tags.
  • DataSource name: openapi Provide the datasource name for the OpenAPI service if --client is specified. If a datasource is selected, the prompt is skipped.

Generated artifacts

Models and types

The command generates the following model classes and TypeScript types:

1. Schemas

For each schema under components.schemas, a model class or type declaration is generated as src/models/<model-or-type-name>.model.ts.

Simple types, array types, composite types (allOf/anyOf/oneOf) are mapped to TypeScript type declarations. Object types are mapped to TypeScript classes.

For example,

src/models/message.model.ts

export type Message = string;

src/models/order-enum.model.ts

export type OrderEnum = 'ascending' | 'descending';

src/models/comments.model.ts

import {Comment} from './comment.model';
export type Comments = Comment[];

src/models/cart.model.ts

import {model, property} from '@loopback/repository';
import {CartShippingZone} from './cart-shipping-zone.model';
import {CartStoreInfo} from './cart-store-info.model';
import {CartWarehouse} from './cart-warehouse.model';

/**
 * The model class is generated from OpenAPI schema - Cart
 * Cart
 */
@model({name: 'Cart'})
export class Cart {
  constructor(data?: Partial<Cart>) {
    if (data != null && typeof data === 'object') {
      Object.assign(this, data);
    }
  }

  @property({name: 'additional_fields'})
  additional_fields?: {};

  @property({name: 'custom_fields'})
  custom_fields?: {};

  @property({name: 'db_prefix'})
  db_prefix?: string;

  @property({name: 'name'})
  name?: string;

  @property({name: 'shipping_zones'})
  shipping_zones?: CartShippingZone[];

  @property({name: 'stores_info'})
  stores_info?: CartStoreInfo[];

  @property({name: 'url'})
  url?: string;

  @property({name: 'version'})
  version?: string;

  @property({name: 'warehouses'})
  warehouses?: CartWarehouse[];
}

src/models/id-type.model.ts

export type IdType = string | number;

src/models/pet.model.ts

import {NewPet} from './new-pet.model.ts';
export type Pet = NewPet & {id: number};

2. Anonymous schemas

Anonymous schemas of object/array types are generated as inline TypeScript type literals or separate model classes/types depending on --promote-anonymous-schemas flag (default to false).

For example, the following OpenAPI spec snippet uses anonymous schemas for request and response body objects.

openapi: 3.0.0
// ...
paths:
  // ...
  /{dataset}/{version}/records:
    post:
      // ...
      operationId: perform-search
      parameters:
        // ...
      responses:
        '200':
          description: successful operation
          content:
            application/json:
              schema:
                type: array
                items:
                  type: object
                  additionalProperties:
                    type: object
        '404':
          description: No matching record found for the given criteria.
      requestBody:
        content:
          application/x-www-form-urlencoded:
            schema:
              type: object
              properties:
                criteria:
                  description: >-
                    Uses Lucene Query Syntax in the format of
                    propertyName:value, propertyName:[num1 TO num2] and date
                    range format: propertyName:[yyyyMMdd TO yyyyMMdd]. In the
                    response please see the 'docs' element which has the list of
                    record objects. Each record structure would consist of all
                    the fields and their corresponding values.
                  type: string
                  default: '*:*'
                start:
                  description: Starting record number. Default value is 0.
                  type: integer
                  default: 0
                rows:
                  description: >-
                    Specify number of rows to be returned. If you run the search
                    with default values, in the response you will see 'numFound'
                    attribute which will tell the number of records available in
                    the dataset.
                  type: integer
                  default: 100
              required:
                - criteria

Without --promote-anonymous-schemas, no separate files are generated for anonymous schemas. The controller class uses inline TypeScript type literals as shown below.

src/controllers/search.controller.ts

@operation('post', '/{dataset}/{version}/records')
  async performSearch(
    @requestBody()
    body: {
      criteria: string;
      start?: number;
      rows?: number;
    },
    @param({name: 'version', in: 'path'}) version: string,
    @param({name: 'dataset', in: 'path'}) dataset: string,
  ): Promise<
    {
      [additionalProperty: string]: {};
    }[]
  > {
    throw new Error('Not implemented');
  }

On contrast, if lb4 openapi --promote-anonymous-schemas is used, two additional model files are generated:

src/models/perform-search-body.model.ts

/* eslint-disable @typescript-eslint/no-explicit-any */
import {model, property} from '@loopback/repository';

/**
 * The model class is generated from OpenAPI schema - performSearchBody
 * performSearchBody
 */
@model({name: 'performSearchBody'})
export class PerformSearchBody {
  constructor(data?: Partial<PerformSearchBody>) {
    if (data != null && typeof data === 'object') {
      Object.assign(this, data);
    }
  }

  /**
   * Uses Lucene Query Syntax in the format of propertyName:value, propertyName:[num1 TO num2] and date range format: propertyName:[yyyyMMdd TO yyyyMMdd]. In the response please see the 'docs' element which has the list of record objects. Each record structure would consist of all the fields and their corresponding values.
   */
  @property({name: 'criteria'})
  criteria: string = '*:*';

  /**
   * Starting record number. Default value is 0.
   */
  @property({name: 'start'})
  start?: number = 0;

  /**
   * Specify number of rows to be returned. If you run the search with default values, in the response you will see 'numFound' attribute which will tell the number of records available in the dataset.
   */
  @property({name: 'rows'})
  rows?: number = 100;
}

src/models/perform-search-response-body.model.ts

export type PerformSearchResponseBody = {
  [additionalProperty: string]: {};
}[];

3. Operations

The generator groups operations (paths.<path>.<verb>) by tags. If no tag is present, it defaults to OpenApi. For each tag, a controller class is generated as src/controllers/<tag-name>.controller.ts to hold all operations with the same tag.

Server-side controllers

Controller class names are derived from tag names. The x-controller-name property of an operation can be used to customize the controller name. Method names are derived from operationIds. They can be configured using x-operation-name.

For example,

import {operation, param} from '@loopback/rest';
import {DateTime} from '../models/date-time.model';

/**
 * The controller class is generated from OpenAPI spec with operations tagged
 * by account
 *
 */
export class AccountController {
  constructor() {}

  /**
   * Get list of carts.
   */
  @operation('get', '/account.cart.list.json')
  async accountCartList(
    @param({name: 'params', in: 'query'}) params: string,
    @param({name: 'exclude', in: 'query'}) exclude: string,
    @param({name: 'request_from_date', in: 'query'}) request_from_date: string,
    @param({name: 'request_to_date', in: 'query'}) request_to_date: string,
  ): Promise<{
    result?: {
      carts?: {
        cart_id?: string;
        id?: string;
        store_key?: string;
        total_calls?: string;
        url?: string;
      }[];
      carts_count?: number;
    };
    return_code?: number;
    return_message?: string;
  }> {
    throw new Error('Not implemented');
  }

  /**
   * Update configs in the API2Cart database.
   */
  @operation('put', '/account.config.update.json')
  async accountConfigUpdate(
    @param({name: 'db_tables_prefix', in: 'query'}) db_tables_prefix: string,
    @param({name: 'client_id', in: 'query'}) client_id: string,
    @param({name: 'bridge_url', in: 'query'}) bridge_url: string,
    @param({name: 'store_root', in: 'query'}) store_root: string,
    @param({name: 'shared_secret', in: 'query'}) shared_secret: string,
  ): Promise<{
    result?: {
      updated_items?: number;
    };
    return_code?: number;
    return_message?: string;
  }> {
    throw new Error('Not implemented');
  }

  /**
   * List webhooks that was not delivered to the callback.
   */
  @operation('get', '/account.failed_webhooks.json')
  async accountFailedWebhooks(
    @param({name: 'count', in: 'query'}) count: number,
    @param({name: 'start', in: 'query'}) start: number,
    @param({name: 'ids', in: 'query'}) ids: string,
  ): Promise<{
    result?: {
      all_failed_webhook?: string;
      webhook?: {
        entity_id?: string;
        time?: DateTime;
        webhook_id?: number;
      }[];
    };
    return_code?: number;
    return_message?: string;
  }> {
    throw new Error('Not implemented');
  }
}

Client-side datasource and service proxies

If --client is specified, a datasource is generated to configure the connection to the endpoint that exposes an OpenAPI spec.

src/datasources/test2.datasource.ts

import {inject, lifeCycleObserver, LifeCycleObserver} from '@loopback/core';
import {juggler} from '@loopback/repository';

const config = {
  name: 'test2',
  connector: 'openapi',
  spec: 'customer.yaml',
  validate: false,
  positional: true,
};

// Observe application's life cycle to disconnect the datasource when
// application is stopped. This allows the application to be shut down
// gracefully. The `stop()` method is inherited from `juggler.DataSource`.
// Learn more at https://loopback.io/doc/en/lb4/Life-cycle.html
@lifeCycleObserver('datasource')
export class Test2DataSource extends juggler.DataSource
  implements LifeCycleObserver {
  static dataSourceName = 'test2';
  static readonly defaultConfig = config;

  constructor(
    @inject('datasources.config.test2', {optional: true})
    dsConfig: object = config,
  ) {
    super(dsConfig);
  }
}

For each tag, a service interface and proxy are generated to provide strongly-typed access to the endpoints.

src/services/customer.service.ts

import {getService} from '@loopback/service-proxy';
import {inject, Provider} from '@loopback/core';
import {Test2DataSource} from '../datasources';

import {Customer} from '../models/customer.model';

/**
 * The service interface is generated from OpenAPI spec with operations tagged
 * by Customer
 * Customer resource
 */
export interface CustomerService {
  /**
   * Returns all customers (\/* customers *\/)
   *
   * @param _if if condition
   * @param limit maximum number of results to return
   * @param accessToken Access token (\/* access_token *\/)
   * @returns customer response
   */
  getCustomers(
    _if: string[],
    limit: number,
    accessToken: string,
  ): Promise<Customer[]>;

  /**
   * Creates a new customer
   *
   * @param _requestBody Customer to add
   * @param accessToken Access token (\/* access_token *\/)
   * @returns customer response
   */
  createCustomer(
    _requestBody: Customer,
    accessToken: string,
  ): Promise<Customer>;

  // If `--no-positional` is supplied, we generate an object as the params:
  /**
   * ```ts
   * createCustomer(params: {
   *   _requestBody: Customer;
   *   accessToken: string;
   * }): Promise<Customer>;
   * ```
   */

  /**
   * Returns a customer based on a single ID
   *
   * @param customerId ID of customer to fetch
   * @returns customer response
   */
  findCustomerById(customerId: number): Promise<Customer>;
}

export class CustomerServiceProvider implements Provider<CustomerService> {
  constructor(
    // test2 must match the name property in the datasource json file
    @inject('datasources.test2')
    protected dataSource: Test2DataSource = new Test2DataSource(),
  ) {}

  async value(): Promise<CustomerService> {
    const service = await getService<{apis: Record<string, CustomerService>}>(
      this.dataSource,
    );
    return service.apis['Customer'];
  }
}

OpenAPI Examples

Try out the following specs or your own with lb4 openapi:

For more real world OpenAPI specs, see https://api.apis.guru/v2/list.json.