feat: spotify html scrape from url

README.md

@@ -10,12 +10,12 @@ Inspired by [DatMusic](https://github.com/alashow/datmusic-api), [MyFreeMp3](htt
 
 *For now*
 
-* Parse a Spotify Playlist page
-* Download a title from the playlist
+* Scrape Spotify playlist page
+* Download all the titles from MyFreeMp3 aka VK Music
+* Generate a .m3u8 playlist file
 
 *TODO*
 
-* Download the html spotify page
 * Dowload all songs in a folder
 * Add proper command line options
 * Document how to get a VK access token (need php and other things this
@@ -38,6 +38,5 @@ BEWARE WIP
 
 1. `git clone THIS`
 1. `npm install`
-1. Fill the `.env` with a `ACCESS_TOKEN=` for VK [HowTo](https://github.com/vodka2/vk-audio-token)
-1. `curl https://open.spotify.com/playlist/6LgeEhc97Azxq6sinJQt6w > test.html`
-1. `node . 0` => Download the first title of the playlist
+1. (Optional) Only for direct VK search, fill the `.env` with a `ACCESS_TOKEN=` for VK [HowTo](https://github.com/vodka2/vk-audio-token)
+1. `node . https://open.spotify.com/playlist/6LgeEhc97Azxq6sinJQt6w`

index.js

@@ -11,8 +11,10 @@ const m3u = require('m3u');
 require('dotenv').config();
 
 
-async function getSpotifyPlaylist(){
-  const spotifyPlaylistPageContent = await fs.readFile('./test.html');
+async function getSpotifyPlaylist(playListUrl){
+  const spotifyPlaylistPageContent = await fetch(playListUrl)
+    .then(res => res.text())
+
   const $ = Cheerio.load(spotifyPlaylistPageContent);
 
   let playlist;
@@ -44,19 +46,33 @@ async function searchOnVkMusic(query){
     if(responseBody.error){
       throw _.assign(responseBody.error, new Error(responseBody.error.error_msg));
     } else {
-      return responseBody;
+      return responseBody.response.items;
     }
   });
 }
 
+async function searchOnMyFreeMp3(query){
+  const url = new URL('https://myfreemp3cc.com/api/search.php?callback=callback');
+
+  return await fetch(url, {
+    method: 'POST',
+    body: new URLSearchParams({
+      q: query,
+      page: 0,
+    }),
+  })
+  .then(res => res.text())
+  .then(jsonp => vm.runInNewContext(jsonp, { callback: (payload) => payload.response }));
+}
+
 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
+  const matchArtistScore = 1 - (leven(originalArtistNames, _.get(vkmusicMetas, 'artist', '').toLowerCase()) / Math.max(originalArtistNames.length, _.get(vkmusicMetas, 'artist', '').length));
+  const matchTitleScore =  1 - (leven(originalTitle, _.get(vkmusicMetas, 'title', '').toLowerCase()) / Math.max(originalTitle.length, _.get(vkmusicMetas, 'title', '').length));
+  const matchDurationScore = 1 - (Math.abs(originalDuration - _.get(vkmusicMetas, 'duration', 0)) / 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;
 }
@@ -67,24 +83,24 @@ async function generateM3U8Playlist(filesWithMetas){
   await fs.writeFile(`playlist.m3u8`, m3uWriter.toString());
 }
 
-async function main(){
-  const playlist = await getSpotifyPlaylist();
+async function main(playlistUrl){
+  const playlist = await getSpotifyPlaylist(playlistUrl);
   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} ${track.name}`);
-    debug('response=%O', response);
-    const bestMatch = _.chain(response.items).map((item) => {
+    const items = await searchOnMyFreeMp3(`${artistNames} ${track.name}`);
+    debug('items=%O', items);
+    const bestMatch = _.chain(items).map((item) => {
       item.score = matchScore(track, item);
       return item;
-    }).sortBy('score').last().value();
-    debug('bestMatch=%O', bestMatch);
+    }).filter(item => item.score && item.url).sortBy('score').last().value();
     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 () => {
+    debug('bestMatch=%O', bestMatch);
+    await fs.access(bestMatch.path).catch(async () => { // TODO find a proper way to not re-download, lower/uppercase problems
       await fetch(bestMatch.url).then(res => {
         res.body.pipe(createWriteStream(bestMatch.path));
       });
@@ -96,11 +112,11 @@ async function main(){
   await generateM3U8Playlist(_.compact(vkPlaylist));
 }
 
-//main().catch(err => console.error(err));
+main(process.argv[2]).catch(err => console.error(err));
 
 async function test(query){
-  const { response } = await searchOnVkMusic(query);
+  const response = await searchOnMyFreeMp3(query);
   console.log(response);
 }
 
-test(process.argv[2]);
+//test(process.argv[2]);