我如何在Django注释中返回列表？

Question

现在，我有以下非常缓慢但有效的代码：

crossover_list = {}
    for song_id in song_ids:
        crossover_set = list(dance_occurrences.filter(
            song_id=song_id).values_list('dance_name_id', flat=True).distinct())
        crossover_list[song_id] = crossover_set

返回字典，其中将歌曲ID用作字典键，并使用整数值列表作为值。前三个键如下：

crossover_list = {
        1:[38,37],
        2:[38],
        ....
}

这里有人知道将其包装到单个查询中的简洁方法吗？数据存在于具有三列的单个表中，其中每个song_id可以与多个dance_id相关联。

song_id | playlist_id | dance_id
      1             1         38
      1             2         37
      2             1         38

理想情况下，我试图弄清楚如何返回：

<QuerySet[{'song_id':1, [{'dance_id':38, 'dance_id':37}]}, {'song_id':2, [{'dance_id':38}]}]>

感谢任何想法或帮助。

编辑：根据要求，以下是相关模型：

# This model helps us do analysis on music/dance crossover density, song popularity within-genre,
# playlist viability within-genre (based on song occurrence counts per song within each playlist), etc.
class SongOccurrences(models.Model):
    song = models.ForeignKey(
        'Songs',
        on_delete=models.CASCADE,
    )
    playlist = models.ForeignKey(
        'Playlists',
        on_delete=models.CASCADE,
    )

    dance_name = models.ForeignKey(
        'DanceMasterTable',
        on_delete=models.CASCADE,
    )

    class Meta:
        constraints = [
            models.UniqueConstraint(fields=['song', 'playlist'], name='unique occurrence')
        ]

# This model contains relevant data from Spotify about each playlist
class Playlists(models.Model):
    spotify_playlist_uri = models.CharField(max_length=200, unique=True)
    spotify_playlist_owner = models.CharField(max_length=200)
    spotify_playlist_name = models.CharField(max_length=200)
    current_song_count = models.IntegerField()
    previous_song_count = models.IntegerField()
    dance_name = models.ForeignKey(
        'DanceMasterTable',
        on_delete=models.CASCADE,
    )

# This model contains all relavent data from Spotify about each song
class Songs(models.Model):
    title = models.CharField(max_length=200)
    first_artist = models.CharField(max_length=200)
    all_artists = models.CharField(max_length=200)
    album = models.CharField(max_length=200)
    release_date = models.DateField('Release Date', blank=True)
    genres = models.CharField(max_length=1000, blank=True)
    popularity = models.FloatField(blank=True)  # This value changes often
    explicit = models.BooleanField(blank=True)
    uri = models.CharField(max_length=200, unique=True)
    tempo = models.FloatField(blank=True)
    time_signature = models.IntegerField()
    energy = models.FloatField(blank=True)
    danceability = models.FloatField(blank=True)
    duration_ms = models.IntegerField()
    tonic = models.IntegerField(blank=True)
    mode = models.IntegerField(blank=True)
    acousticness = models.FloatField(blank=True)
    instrumentalness = models.FloatField(blank=True)
    liveness = models.FloatField(blank=True)
    loudness = models.FloatField(blank=True)
    speechiness = models.FloatField(blank=True)
    valence = models.FloatField(blank=True)

    class Meta:
        constraints = [
            models.UniqueConstraint(fields=['title', 'first_artist', 'all_artists'], name='unique song')
        ]

# This model contains the (static) master list of partner dances to be analyzed.
class DanceMasterTable(models.Model):
    dance_name = models.CharField(max_length=200, unique=True)

Answer 1

您正在循环内运行查询，因此它很慢。

您可以事先用所有dance_occurrences过滤song_ids，最后循环遍历这些值，将舞曲ID附加到其各自的歌曲ID。

示例：

song_dance_occurrences = dance_occurrences.filter(
    song_id__in=song_ids
).values_list('song_id', 'dance_id').distinct()

crossover_dict = {}
for song_id, dance_id in song_dance_occurrences:
    crossover_dict[song_id] = crossover_dict.get(song_id, [])
    crossover_dict[song_id].append(dance_id)