I am using a SQLite backend with a simple show - season - episode schema:
class Show(BaseModel):
name = CharField()
class Season(BaseModel):
show = ForeignKeyField(Show, related_name='seasons')
season_number = IntegerField()
class Episode(BaseModel):
season = ForeignKeyField(Season, related_name='episodes')
episode_number = IntegerField()
and I would need the following query :
seasons = (Season.select(Season, Episode)
.join(Episode)
.where(Season.show == SHOW_ID)
.order_by(Season.season_number.desc(), Episode.episode_number.desc())
.aggregate_rows())
SHOW_ID being the id of the show for which I want the list of seasons. But when I iterate over the query with the following code :
for season in seasons:
for episode in season.episodes:
print(episode.episode_number)
... I get something which is not ordered at all, and which does not even follow the order I would get without using order_by(), i.e. the insertion order.
I activated the debug logs to see the outgoing query, and the query does contain the ORDER BY clause, and manually applying it returns the proper descending order.
I am new to peewee, and I have seen so many examples making use of a join() combines with an order_by(), but I can still not find out what I am doing wrong.
This was due to a bug in the processing of nested collections in the aggregate query result wrapper.
The github issue is: https://github.com/coleifer/peewee/issues/519
The fix has been merged here: https://github.com/coleifer/peewee/commit/ec0e87f1a480695d98bf1f0d7f2e63aed8dfc440
So, to get the fix you'll need to either clone master or wait til the next release which should be in the next week or two (2.4.7).