Genre Artist Breakdown

Implemented the generation of genre artist breakdown data for the genre
graphs. More test cases needed.

api/admin.py

@@ -1,8 +1,9 @@
 from django.contrib import admin
-from .models import Track, Artist, AudioFeatures, User
+from .models import Track, Artist, AudioFeatures, User, Genre
 
 # Register your models here.
 admin.site.register(Track)
 admin.site.register(Artist)
 admin.site.register(AudioFeatures)
 admin.site.register(User)
+admin.site.register(Genre)

api/models.py

@@ -30,6 +30,8 @@ class Artist(models.Model):
     id = models.CharField(primary_key=True, max_length=MAX_ID)
     name = models.CharField(max_length=50)
     genres = models.ManyToManyField(Genre, blank=True)
+    # genre = models.ForeignKey(Genre, on_delete=models.CASCADE, blank=True,
+    #                           null=True)
 
     def __str__(self):
         return self.name
@@ -45,7 +47,6 @@ class Track(models.Model):
         verbose_name_plural = "Tracks"
 
     id = models.CharField(primary_key=True, max_length=MAX_ID)
-    #  artist = models.ForeignKey(Artist, on_delete=models.CASCADE)
     artists = models.ManyToManyField(Artist, blank=True)
     year = models.PositiveSmallIntegerField()
     popularity = models.PositiveSmallIntegerField()
@@ -53,7 +54,7 @@ class Track(models.Model):
     name = models.CharField(max_length=200)
     users = models.ManyToManyField(User, blank=True)
     genre = models.ForeignKey(Genre, on_delete=models.CASCADE, blank=True,
-            null=True)
+                              null=True)
 
     def __str__(self):
         track_str = "{}, genre: {}, artists: [".format(self.name, self.genre)

api/tests.py

@@ -0,0 +1,50 @@
+from django.test import TestCase
+from api.models import Track, Genre, Artist
+from login.models import User
+from api import utils
+import math
+import pprint
+
+class GenreDataTestCase(TestCase):
+
+    def setUp(self):
+        test_user = User.objects.create(id="chrisshi", refresh_token="blah", access_token="blah",
+                         access_expires_in=10)
+        genre = Genre.objects.create(name="classical", num_songs=3)
+        artist_1 = Artist.objects.create(id='art1', name="Beethoven")
+        artist_2 = Artist.objects.create(id='art2', name="Mozart")
+        artist_3 = Artist.objects.create(id='art3', name='Chopin')
+
+        track_1 = Track.objects.create(id='track1', year=2013,
+                                       popularity=5, runtime=20,
+                                       name='concerto1',
+                                       genre=genre)
+        track_1.users.add(test_user)
+        track_1.artists.add(artist_1)
+        track_1.artists.add(artist_2)
+
+        track_2 = Track.objects.create(id='track2', year=2013,
+                                       popularity=5, runtime=20,
+                                       name='concerto2',
+                                       genre=genre)
+        track_2.users.add(test_user)
+        track_2.artists.add(artist_2)
+        track_2.artists.add(artist_3)
+        track_2.artists.add(artist_1)
+
+        track_3 = Track.objects.create(id='track3', year=2013,
+                                       popularity=5, runtime=20,
+                                       name='concerto3',
+                                       genre=genre)
+        track_3.users.add(test_user)
+        track_3.artists.add(artist_1)
+        track_3.artists.add(artist_3)
+
+    def test_get_artist_counts_in_genre(self):
+        test_user = User.objects.get(id='chrisshi')
+        artist_counts = utils.get_artists_in_genre(test_user, 'classical', 10)
+        # pprint.pprint(artist_counts)
+        self.assertTrue(math.isclose(artist_counts['Beethoven'], 1.3, rel_tol=0.05))
+        self.assertTrue(math.isclose(artist_counts['Mozart'], 0.85, rel_tol=0.05))
+        self.assertTrue(math.isclose(artist_counts['Chopin'], 0.85, rel_tol=0.05))
+        self.assertTrue(math.isclose(sum(artist_counts.values()), 3, rel_tol=0.01))

api/utils.py

@@ -11,6 +11,8 @@ from django.core import serializers
 from django.utils import timezone
 from .models import *
 from login.models import User
+from django.db.models import FloatField
+from django.db.models.functions import Cast
 
 #  }}} imports # 
 
@@ -34,16 +36,13 @@ def update_track_genres(user_obj):
     user_tracks = Track.objects.filter(users__exact=user_obj)
     for track in user_tracks:
         # just using this variable to save another call to db
-        track_artists = track.artists.all()
+        track_artists = list(track.artists.all())
-        # set genres to first artist's genres then find intersection with others
+        # TODO: Use the most popular genre of the first artist as the Track genre
-        shared_genres = track_artists.first().genres.all()
+        first_artist_genres = track_artists[0].genres.all().order_by('-num_songs')
-        for artist in track_artists:
-            shared_genres = shared_genres.intersection(artist.genres.all())
-        shared_genres = shared_genres.order_by('-num_songs')
 
         undefined_genre_obj = Genre.objects.get(name="undefined")
-        most_common_genre = shared_genres.first() if shared_genres.first() is \
+        most_common_genre = first_artist_genres.first() if first_artist_genres.first() is \
-                not undefined_genre_obj else shared_genres[1]
+                not undefined_genre_obj else first_artist_genres[1]
         track.genre = most_common_genre if most_common_genre is not None \
                 else undefined_genre_obj
         track.save()
@@ -143,8 +142,7 @@ def process_artist_genre(genre_name, artist_obj):
     :returns: None
 
     """
-    genre_obj, created = Genre.objects.get_or_create(name=genre_name,
+    genre_obj, created = Genre.objects.get_or_create(name=genre_name, defaults={'num_songs': 1})
-            defaults={'num_songs':1})
     if not created:
         genre_obj.num_songs = F('num_songs') + 1
         genre_obj.save()
@@ -192,7 +190,7 @@ def get_artists_in_genre(user, genre, max_songs):
     """Return count of artists in genre.
 
     :user: User object to return data for.
-    :genre: genre to count artists for.
+    :genre: genre to count artists for. (string)
     :max_songs: max total songs to include to prevent overflow due to having
     multiple artists on each track.
 
@@ -200,19 +198,22 @@ def get_artists_in_genre(user, genre, max_songs):
     have. 
     """
     genre_obj = Genre.objects.get(name=genre)
-    artist_counts = (Artist.objects.filter(track__users=user)
+    tracks_in_genre = Track.objects.filter(genre=genre_obj, users=user)
-            .filter(genres=genre_obj) 
+    track_count = tracks_in_genre.count()
-            .annotate(num_songs=Count('track', distinct=True))
+    user_artists = Artist.objects.filter(track__users=user)  # use this variable to save on db queries
-            .order_by('-num_songs')
+    total_artist_counts = tracks_in_genre.aggregate(counts=Count('artists'))['counts']
-            )
+
     processed_artist_counts = {}
-    songs_added = 0
+    # songs_added = 0
-    for artist in artist_counts:
+    for artist in user_artists:
         # hacky way to not have total count overflow due to there being multiple
         # artists on a track
-        if songs_added + artist.num_songs <= max_songs:
+        # if songs_added + artist.num_songs <= max_songs:
-            processed_artist_counts[artist.name] = artist.num_songs
+        #     processed_artist_counts[artist.name] = artist.num_songs
-            songs_added += artist.num_songs
+        #     songs_added += artist.num_songs
+        processed_artist_counts[artist.name] = round(artist.track_set
+                                                     .filter(genre=genre_obj, users=user)
+                                                     .count() * track_count / total_artist_counts, 2)
     #  processed_artist_counts = [{'name': artist.name, 'num_songs': artist.num_songs} for artist in artist_counts]
     #  processed_artist_counts = {artist.name: artist.num_songs for artist in artist_counts}
     #  pprint.pprint(processed_artist_counts)

api/views.py

@@ -81,7 +81,7 @@ def parse_library(request, user_secret):
                     track_artists, user_obj)
 
             #  add audio features {{{ # 
-            
+
             # if a new track is not created, the associated audio feature does
             # not need to be created again
             if track_created:
@@ -174,13 +174,12 @@ def get_genre_data(request, user_secret):
     """
     user = User.objects.get(secret=user_secret)
     genre_counts = (Track.objects.filter(users__exact=user)
-            .values('genre')
+                    .values('genre')
-            .order_by('genre')
+                    .order_by('genre')
-            .annotate(num_songs=Count('genre'))
+                    .annotate(num_songs=Count('genre'))
-            )
+                    )
     for genre_dict in genre_counts:
-        genre_dict['artists'] = get_artists_in_genre(user, genre_dict['genre'],
+        genre_dict['artists'] = get_artists_in_genre(user, genre_dict['genre'], genre_dict['num_songs'])
-                genre_dict['num_songs'])
     print("*** Genre Breakdown ***")
     pprint.pprint(list(genre_counts))
     return JsonResponse(data=list(genre_counts), safe=False) 

graphs/static/graphs/scripts/genre_graph.js

@@ -4,7 +4,7 @@ function create_genre_graph(data) {
     data.forEach(function(d) {
         d.num_songs = +d.num_songs;
         console.log(d.genre, d.num_songs);
-        var artist_names = Object.keys(d.artists);
+        let artist_names = Object.keys(d.artists);
         artist_names.forEach(function(e) {
             d.artists[e] = +d.artists[e];
             console.log(e, d.artists[e]);
@@ -31,22 +31,22 @@ function create_genre_graph(data) {
 
     // setup bar colors {{{ //
     
-    var max_artists = d3.max(data, function(d) {
+    let max_artists = d3.max(data, function(d) {
         return Object.keys(d.artists).length;
     });
-    var z = d3.scaleOrdinal().range(randomColor({
+    let z = d3.scaleOrdinal().range(randomColor({
         count: max_artists,
         luminosity: 'light',
     }));
     
     // }}} setup bar colors //
 
-    for (var genre_dict of data) {
+    for (let genre_dict of data) {
 		
         // process artist breakdown  {{{ //
         
-        var keys = Object.keys(genre_dict.artists);
+        let keys = Object.keys(genre_dict.artists);
-        var stack = d3.stack()
+        let stack = d3.stack()
             //.order(d3.stackOrderAscending)
             .order(d3.stackOrderDescending)
             .keys(keys)([genre_dict.artists])
@@ -112,7 +112,7 @@ function create_genre_graph(data) {
 // https://gist.github.com/guypursey/f47d8cd11a8ff24854305505dbbd8c07#file-index-html
 function wrap(text, width) {
     text.each(function() {
-        var text = d3.select(this),
+        let text = d3.select(this),
 			words = text.text().split(/\s+/).reverse(),
             word,
             line = [],

graphs/templates/graphs/genre_graph.html

@@ -26,16 +26,16 @@
 
   <svg width="1920" height="740"></svg>
   <script>
-    var svg = d3.select("svg"),
+    let svg = d3.select("svg"),
         margin = {top: 20, right: 20, bottom: 30, left: 40},
         width = +svg.attr("width") - margin.left - margin.right,
         height = +svg.attr("height") - margin.top - margin.bottom,
         g = svg.append("g").attr("transform", "translate(" + margin.left + "," + margin.top + ")");
-    var x = d3.scaleBand()
+    let x = d3.scaleBand()
         .rangeRound([0, width])
         .paddingInner(0.05)
         .align(0.1);
-    var y = d3.scaleLinear()
+    let y = d3.scaleLinear()
         .rangeRound([height, 0]);
 
     d3.json("{% url "api:get_genre_data" user_secret %}").then(create_genre_graph);