vkontakte-api

vkontakte-api

TypeScript library to make requests performing to VK API simple

Installation

yarn add vkontakte-api

or

npm i vkontakte-api

Description

vkontakte-api follows repositories based concept where each repository is a class and represents some namespace in API. The main purpose of repository is to format request config for VKAPI instance, so he could perform request and return data.

Each request is added to queue and executed only after timeout, calculated according to rps property, is elapsed. So, there is no case when you were banned due to sending too many requests per second.

There is a multi-thread (multi-workers) support for those projects which are launched in this mode.

Usage

Creating instance

First, it is required to create VKAPI instance:

import {VKAPI} from 'vkontakte-api';

const api = new VKAPI();

It is allowed to pass rps property which means requests per second. VK API has its restriction, so make sure you have passed correct value.

Additionally, you can pass properties accessToken and lang which will be used as default parameters for each request. So, you have no need to pass them each time until overriding is needed.

Here is how it looks like:

const api = new VKAPI({
  rps: 20,
  accessToken: 'my default token',
  lang: 'uk',
});

Performing requests

VKAPI instance contains a list of repositories which generate request parameters to send to API. Each repository is named according to its name in API.

Simple example of sending request and logging data:

import {VKAPI} from 'vkontakte-api';

const api = new VKAPI();

api.users.get({userIds: ['vladkibenko']}).then(console.log);

Sending some notification:

api.notifications.sendMessage({
  userIds: ['vladkibenko'],
  message: 'Hello Vlad!',
});

Overriding default lang and accessToken:

import {LangEnum, VKAPI} from 'vkontakte-api';

const api = new VKAPI({accessToken: 'my personal user token'});

// Here we will get english-localized data from some application's face
api.users.get({
  userIds: ['vladkibenko'],
  accessToken: 'some application token',
  // Or you could just use 'uk' or 1
  lang: LangEnum.UK,
}).then(console.log);

Some of methods are not currently typed or realised. So, you are free to perform custom requests. Make sure, all of Params and Response fields are camel cased, because internally, vkontakte-api moves them from snake to camel case for easier usage:

import {VKAPI} from 'vkontakte-api';

const api = new VKAPI({accessToken: 'my token'});

// Description of parameters
interface Params {
  cityIds: string;
}

// Description of response
type Response = Array<{
  id: number;
  title: string;
}>;

// @see https://vk.com/dev/database.getCitiesById
api.addRequestToQueue<Params, Response>({
  method: 'database.getCitiesById',
  params: {
    cityIds: [1].join(','),
  },
}).then(console.log);

Errors

Sometimes, API throws errors. To detect if error was thrown by VK, you could use function isVKError which detects if error is VKError. It contains property data which contains all error data (it is typed, by the way).

Moreover, lib contains enum ErrorsEnum which is a set of all known errors.

Multi-threading support

In case your project is ran in multi cluster mode, you could use VKAPIMaster and VKAPISlave.

VKAPIMaster should be used in main thread. To create its instance, it is required to pass VKAPI instance which will perform all of the requests and list of workers, containing VKAPISlave which should communicate with VKAPIMaster.

With this scheme we are getting single VKAPI instance for project and not overflowing API restriction connected with requests per second.

You can find more complex example here.

Here is example:

import {fork, isMaster, Worker} from 'cluster';
import os from 'os';
import {VKAPI, VKAPIMaster, VKAPISlave, VKAPIInterface} from 'vkontakte-api';

// Runs http server. Accepts an object which looks like VKAPI instance. So,
// he does not know what api exactly is. It could be real VKAPI instance or
// VKAPISlave, we dont care, it is the same in this context
function http(api: VKAPIInterface) {
  // Here we can use all of the VKAPI methods due to they are defined
  // interface. We does not really care what api is
  api.users.get({userIds: ['vladkibenko']}).then(console.log);
}

// Just a stub. You can use the logic you need
const isDev = process.env.NODE_ENVIRONMENT === 'development';

// In development mode, let us run single thread. So, no VKAPIMaster and
// VKAPISlave are needed
if (isDev) {
  const api = new VKAPI();
  
  // Run http server
  return http(api);
}

// In production mode, we do create as many forks as processor support
if (isMaster) {
  const cpuCount = os.cpus().length;
  const workers: Worker[] = [];

  for (let i = 0; i < cpuCount; i++) {
    workers.push(fork());
  }

  // In master we do create VKAPI instance, because slaves should
  // communicate with single its instance, which is in VKAPIMaster
  const master = new VKAPIMaster({workers, instance: new VKAPI()});
  master.init();
} 
// In slave workers, we just create http server with VKAPISlave
else {
  // Slave has no its own settings due to it is just a class, which communicates
  // with master. So, all settings are set in VKAPIMaster 
  http(new VKAPISlave());
}