vk-music-dl/index.js

const Cheerio = require('cheerio');
const { promises: fs, createWriteStream } = require('fs');
const debug = require('debug')('vk-music-dl');
const vm = require('vm');
const Promise = require('bluebird');
const _ = require('lodash');
const leven = require('leven');
const { URL, URLSearchParams } = require('url');
const fetch = require('node-fetch');
const m3u = require('m3u');
require('dotenv').config();


async function getSpotifyPlaylist(){
  const spotifyPlaylistPageContent = await fs.readFile('./test.html');
  const $ = Cheerio.load(spotifyPlaylistPageContent);

  let playlist;
  $('script').each(function(i, elem){
    const content = $(this).html();
    if(/Spotify.Entity/.test(content)){
      playlist = vm.runInNewContext(content);
    }
  });
  return playlist;
}

async function searchOnVkMusic(query){
  const url = new URL('https://api.vk.com/method/audio.search');
  url.search = new URLSearchParams({
    v: '5.71',
    access_token: process.env.ACCESS_TOKEN,
    q: query,
    count: 200,
  }).toString();

  return await fetch(url, {
    headers: {
      'User-Agent': 'KateMobileAndroid/48.2 lite-433 (Android 8.1.0; SDK 27; arm64-v8a; Google Pixel 2 XL; en)',
    }
  })
  .then(res => res.json())
  .then(responseBody => {
    if(responseBody.error){
      throw _.assign(responseBody.error, new Error(responseBody.error.error_msg));
    } else {
      return responseBody;
    }
  });
}

function matchScore(spotifyMetas, vkmusicMetas){
  const originalArtistNames = _.map(spotifyMetas.artists, 'name').join(', ').toLowerCase();
  const originalTitle = spotifyMetas.name.toLowerCase();
  const originalDuration = Math.round(spotifyMetas.duration_ms/1000);

  const matchArtistScore = 1 - (leven(originalArtistNames, vkmusicMetas.artist.toLowerCase()) / Math.max(originalArtistNames.length, vkmusicMetas.artist.length));
  const matchTitleScore =  1 - (leven(originalTitle, vkmusicMetas.title.toLowerCase()) / Math.max(originalTitle.length, vkmusicMetas.title.length));
  const matchDurationScore = 1 - (Math.abs(originalDuration - vkmusicMetas.duration) / originalDuration); // TODO this can return more than 1 or less than 0
  debug('matchArtistScore=%f matchTitleScore=%f matchDurationScore=%f', matchArtistScore, matchTitleScore, matchDurationScore);
  return matchArtistScore + matchTitleScore + matchDurationScore;
}

async function generateM3U8Playlist(filesWithMetas){
  const m3uWriter = m3u.extendedWriter();
  filesWithMetas.forEach(fileWithMetas => m3uWriter.file(fileWithMetas.path, fileWithMetas.duration, `${fileWithMetas.artist} - ${fileWithMetas.title}`));
  await fs.writeFile(`playlist.m3u8`, m3uWriter.toString());
}

async function main(){
  const playlist = await getSpotifyPlaylist();
  const vkPlaylist = await Promise.map(playlist.tracks.items, async ({track}) => {
    const artistNames = _.map(track.artists, 'name').join(', ');
    debug('%s - %s', track.name, artistNames);
    const { response } = await searchOnVkMusic(`${artistNames} ${track.name}`);
    debug('response=%O', response);
    const bestMatch = _.chain(response.items).map((item) => {
      item.score = matchScore(track, item);
      return item;
    }).sortBy('score').last().value();
    debug('bestMatch=%O', bestMatch);
    if(!bestMatch){
      console.log(`You are on your own for ${track.name} - ${artistNames}`);
      return;
    }
    bestMatch.path = `${bestMatch.artist} - ${bestMatch.title}.mp3`;
    await fs.access(bestMatch.path).catch(async () => {
      await fetch(bestMatch.url).then(res => {
        res.body.pipe(createWriteStream(bestMatch.path));
      });
      debug('Done downloading');
    });
    return bestMatch;
  }, {concurrency: 1});

  await generateM3U8Playlist(_.compact(vkPlaylist));
}

//main().catch(err => console.error(err));

async function test(query){
  const { response } = await searchOnVkMusic(query);
  console.log(response);
}

test(process.argv[2]);