{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing the libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import json\n",
    "import spotipy\n",
    "import spotipy.oauth2 as oauth2\n",
    "from spotipy.oauth2 import SpotifyOAuth,SpotifyClientCredentials\n",
    "import yaml\n",
    "import re\n",
    "from tqdm import tqdm\n",
    "import multiprocessing as mp\n",
    "import time\n",
    "import random\n",
    "import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "stream= open(\"spotify/spotify.yaml\")\n",
    "spotify_details = yaml.safe_load(stream)\n",
    "auth_manager = SpotifyClientCredentials(client_id=spotify_details['Client_id'],\n",
    "                                        client_secret=spotify_details['client_secret'])\n",
    "sp = spotipy.client.Spotify(auth_manager=auth_manager)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing the dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Reading the 1 million playlists and keeping only the unique track URIs for the content-based recommendation system.\n",
    "- The first for loop Read the 1,000 JSON files one at a time.\n",
    "- The second for loop is for getting only the unique track."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def loop_slices(path, num_slices=20):\n",
    "  cnt = 0\n",
    "  cnt1 = 0\n",
    "  mpd_playlists = []\n",
    "  unique_tracks= pd.DataFrame()\n",
    "  filenames = os.listdir(path)\n",
    "  for fname in tqdm(sorted(filenames, key=len)):\n",
    "    if fname.startswith(\"mpd.slice.\") and fname.endswith(\".json\"):\n",
    "      cnt += 1\n",
    "      fullpath = os.sep.join((path, fname))\n",
    "      f = open(fullpath)\n",
    "      js = f.read()\n",
    "      f.close()\n",
    "      current_slice = json.loads(js)\n",
    "      # Create a list of all playlists\n",
    "      for playlist in current_slice['playlists']:\n",
    "        cnt1 +=1\n",
    "        mpd_playlists.append(playlist)\n",
    "        if cnt1 == 1000:\n",
    "          cnt1=0\n",
    "          temp=pd.DataFrame(mpd_playlists)\n",
    "          temp=temp.explode('tracks')\n",
    "          temp=pd.DataFrame(temp['tracks'].apply(pd.Series))\n",
    "          unique_tracks=pd.concat([unique_tracks,temp],axis=0)\n",
    "          unique_tracks.drop_duplicates(subset=['track_uri'],inplace=True)\n",
    "          mpd_playlists = []\n",
    "      if cnt == num_slices:\n",
    "        break\n",
    "  return unique_tracks\n",
    "# Path where the json files are extracted\n",
    "path = 'spotify_million_playlist_dataset/data/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = loop_slices(path, num_slices=1000)\n",
    "df.to_csv('data/1m.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[\"track_uri\"] = df[\"track_uri\"].apply(lambda x: re.findall(r'\\w+$', x)[0])\n",
    "df[\"artist_uri\"] = df[\"artist_uri\"].apply(lambda x: re.findall(r'\\w+$', x)[0])\n",
    "df[\"album_uri\"] = df[\"album_uri\"].apply(lambda x: re.findall(r'\\w+$', x)[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pd.read_csv('Data/1mV3.csv') #Dropped all columns except ['track_uri', 'artist_uri', 'album_uri']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['track_uri', 'artist_uri', 'album_uri'], dtype='object')"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "t_uri=df[\"track_uri\"].unique()\n",
    "a_uri=df[\"artist_uri\"].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature extraction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the Spotify API for Feature Extraction and Saving Results to a CSV File and Errors to a Log File\n",
    "\n",
    "I was using SP.track first, but I realised that it would take a lot of time and I would have to counter a lot of Api rate limits, so I used SP.tracks and SP.artists instead. They accept lists with a 50-URI maximum and handle them in a single request, so it took a lot less time."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = open('data/audio_features.csv','a')\n",
    "e=0\n",
    "for i in tqdm(range(0,len(t_uri),100)):\n",
    "    try:\n",
    "     track_feature = sp.audio_features(t_uri[i:i+100])\n",
    "     track_df = pd.DataFrame(track_feature)\n",
    "     csv_data = track_df.to_csv(header=False,index=False)\n",
    "     f.write(csv_data)\n",
    "    except Exception as error:\n",
    "        e+=1\n",
    "        r = open(\"audio_features_log.txt\", \"a\")\n",
    "        r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\": \"+str(error)+'\\n')\n",
    "        r.close()\n",
    "        time.sleep(3)\n",
    "        continue\n",
    "r = open(\"audio_features_log.txt\", \"a\")\n",
    "r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\" _________________________ \"+\"Total Number Of Errors : \"+str(e)+\" _________________________ \"+'\\n')\n",
    "r.close()\n",
    "f.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = open('data/track_features.csv','a')\n",
    "e=0\n",
    "for i in tqdm(range(0,len(t_uri),50)):\n",
    "    try:\n",
    "        track_features = sp.tracks(t_uri[i:i+50])\n",
    "        for x in range(50):\n",
    "            track_pop=pd.DataFrame([t_uri[i+x]])\n",
    "            track_pop['release_date']=track_features['tracks'][x]['album']['release_date']\n",
    "            track_pop['pop'] = track_features['tracks'][x][\"popularity\"]\n",
    "            csv_data = track_pop.to_csv(header=False,index=False)\n",
    "            f.write(csv_data)\n",
    "    except Exception as error:\n",
    "        e+=1\n",
    "        r = open(\"track_features.txt\", \"a\")\n",
    "        r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\": \"+str(error)+'\\n')\n",
    "        r.close()\n",
    "        time.sleep(3)\n",
    "        continue\n",
    "r = open(\"track_features.txt\", \"a\")\n",
    "r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\" _________________________ \"+\"Total Number Of Errors : \"+str(e)+\" _________________________ \"+'\\n')\n",
    "r.close()\n",
    "f.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "f = open('data/artist_features.csv','a')\n",
    "e=0\n",
    "for i in tqdm(range(0,len(a_uri),50)):\n",
    "    try:\n",
    "        artist_features = sp.artists(a_uri[i:i+50])\n",
    "        for x in range(50):\n",
    "            artist_df=pd.DataFrame([a_uri[i+x]])\n",
    "            artist_pop = artist_features['artists'][x][\"popularity\"]\n",
    "            artist_genres = artist_features['artists'][x][\"genres\"]\n",
    "            artist_df[\"artist_pop\"] = artist_pop\n",
    "            if artist_genres: \n",
    "                artist_df[\"genres\"] = \" \".join([re.sub(' ','_',i) for i in artist_genres])\n",
    "            else:\n",
    "              artist_df[\"genres\"] = \"unknown\"\n",
    "            csv_data = artist_df.to_csv(header=False,index=False)\n",
    "            f.write(csv_data)\n",
    "    except Exception as error:\n",
    "        e+=1\n",
    "        r = open(\"artist_features.txt\", \"a\")\n",
    "        r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\": \"+str(error)+'\\n')\n",
    "        r.close()\n",
    "        time.sleep(3)\n",
    "        continue\n",
    "r = open(\"artist_features.txt\", \"a\")\n",
    "r.write(datetime.datetime.now().strftime(\"%d.%b %Y %H:%M:%S\")+\" _________________________ \"+\"Total Number Of Errors : \"+str(e)+\" _________________________ \"+'\\n')\n",
    "r.close()\n",
    "f.close()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9.13 ('base')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "e246d2215c418239c9316a1ebf2d8abb44dc50b2e5b0e29defd87143398aa387"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}