django/tests/aggregation/tests.py

2178 lines
78 KiB
Python

import datetime
import math
import re
from decimal import Decimal
from django.core.exceptions import FieldError
from django.db import connection
from django.db.models import (
Avg,
Case,
Count,
DateField,
DateTimeField,
DecimalField,
DurationField,
Exists,
F,
FloatField,
IntegerField,
Max,
Min,
OuterRef,
Q,
StdDev,
Subquery,
Sum,
TimeField,
Value,
Variance,
When,
)
from django.db.models.expressions import Func, RawSQL
from django.db.models.functions import (
Cast,
Coalesce,
Greatest,
Least,
Lower,
Now,
Pi,
TruncDate,
TruncHour,
)
from django.test import TestCase
from django.test.testcases import skipUnlessDBFeature
from django.test.utils import Approximate, CaptureQueriesContext
from django.utils import timezone
from .models import Author, Book, Publisher, Store
class NowUTC(Now):
template = "CURRENT_TIMESTAMP"
output_field = DateTimeField()
def as_sql(self, compiler, connection, **extra_context):
if connection.features.test_now_utc_template:
extra_context["template"] = connection.features.test_now_utc_template
return super().as_sql(compiler, connection, **extra_context)
class AggregateTestCase(TestCase):
@classmethod
def setUpTestData(cls):
cls.a1 = Author.objects.create(name="Adrian Holovaty", age=34)
cls.a2 = Author.objects.create(name="Jacob Kaplan-Moss", age=35)
cls.a3 = Author.objects.create(name="Brad Dayley", age=45)
cls.a4 = Author.objects.create(name="James Bennett", age=29)
cls.a5 = Author.objects.create(name="Jeffrey Forcier", age=37)
cls.a6 = Author.objects.create(name="Paul Bissex", age=29)
cls.a7 = Author.objects.create(name="Wesley J. Chun", age=25)
cls.a8 = Author.objects.create(name="Peter Norvig", age=57)
cls.a9 = Author.objects.create(name="Stuart Russell", age=46)
cls.a1.friends.add(cls.a2, cls.a4)
cls.a2.friends.add(cls.a1, cls.a7)
cls.a4.friends.add(cls.a1)
cls.a5.friends.add(cls.a6, cls.a7)
cls.a6.friends.add(cls.a5, cls.a7)
cls.a7.friends.add(cls.a2, cls.a5, cls.a6)
cls.a8.friends.add(cls.a9)
cls.a9.friends.add(cls.a8)
cls.p1 = Publisher.objects.create(
name="Apress", num_awards=3, duration=datetime.timedelta(days=1)
)
cls.p2 = Publisher.objects.create(
name="Sams", num_awards=1, duration=datetime.timedelta(days=2)
)
cls.p3 = Publisher.objects.create(name="Prentice Hall", num_awards=7)
cls.p4 = Publisher.objects.create(name="Morgan Kaufmann", num_awards=9)
cls.p5 = Publisher.objects.create(name="Jonno's House of Books", num_awards=0)
cls.b1 = Book.objects.create(
isbn="159059725",
name="The Definitive Guide to Django: Web Development Done Right",
pages=447,
rating=4.5,
price=Decimal("30.00"),
contact=cls.a1,
publisher=cls.p1,
pubdate=datetime.date(2007, 12, 6),
)
cls.b2 = Book.objects.create(
isbn="067232959",
name="Sams Teach Yourself Django in 24 Hours",
pages=528,
rating=3.0,
price=Decimal("23.09"),
contact=cls.a3,
publisher=cls.p2,
pubdate=datetime.date(2008, 3, 3),
)
cls.b3 = Book.objects.create(
isbn="159059996",
name="Practical Django Projects",
pages=300,
rating=4.0,
price=Decimal("29.69"),
contact=cls.a4,
publisher=cls.p1,
pubdate=datetime.date(2008, 6, 23),
)
cls.b4 = Book.objects.create(
isbn="013235613",
name="Python Web Development with Django",
pages=350,
rating=4.0,
price=Decimal("29.69"),
contact=cls.a5,
publisher=cls.p3,
pubdate=datetime.date(2008, 11, 3),
)
cls.b5 = Book.objects.create(
isbn="013790395",
name="Artificial Intelligence: A Modern Approach",
pages=1132,
rating=4.0,
price=Decimal("82.80"),
contact=cls.a8,
publisher=cls.p3,
pubdate=datetime.date(1995, 1, 15),
)
cls.b6 = Book.objects.create(
isbn="155860191",
name=(
"Paradigms of Artificial Intelligence Programming: Case Studies in "
"Common Lisp"
),
pages=946,
rating=5.0,
price=Decimal("75.00"),
contact=cls.a8,
publisher=cls.p4,
pubdate=datetime.date(1991, 10, 15),
)
cls.b1.authors.add(cls.a1, cls.a2)
cls.b2.authors.add(cls.a3)
cls.b3.authors.add(cls.a4)
cls.b4.authors.add(cls.a5, cls.a6, cls.a7)
cls.b5.authors.add(cls.a8, cls.a9)
cls.b6.authors.add(cls.a8)
s1 = Store.objects.create(
name="Amazon.com",
original_opening=datetime.datetime(1994, 4, 23, 9, 17, 42),
friday_night_closing=datetime.time(23, 59, 59),
)
s2 = Store.objects.create(
name="Books.com",
original_opening=datetime.datetime(2001, 3, 15, 11, 23, 37),
friday_night_closing=datetime.time(23, 59, 59),
)
s3 = Store.objects.create(
name="Mamma and Pappa's Books",
original_opening=datetime.datetime(1945, 4, 25, 16, 24, 14),
friday_night_closing=datetime.time(21, 30),
)
s1.books.add(cls.b1, cls.b2, cls.b3, cls.b4, cls.b5, cls.b6)
s2.books.add(cls.b1, cls.b3, cls.b5, cls.b6)
s3.books.add(cls.b3, cls.b4, cls.b6)
def test_empty_aggregate(self):
self.assertEqual(Author.objects.aggregate(), {})
def test_aggregate_in_order_by(self):
msg = (
"Using an aggregate in order_by() without also including it in "
"annotate() is not allowed: Avg(F(book__rating)"
)
with self.assertRaisesMessage(FieldError, msg):
Author.objects.values("age").order_by(Avg("book__rating"))
def test_single_aggregate(self):
vals = Author.objects.aggregate(Avg("age"))
self.assertEqual(vals, {"age__avg": Approximate(37.4, places=1)})
def test_multiple_aggregates(self):
vals = Author.objects.aggregate(Sum("age"), Avg("age"))
self.assertEqual(
vals, {"age__sum": 337, "age__avg": Approximate(37.4, places=1)}
)
def test_filter_aggregate(self):
vals = Author.objects.filter(age__gt=29).aggregate(Sum("age"))
self.assertEqual(vals, {"age__sum": 254})
def test_related_aggregate(self):
vals = Author.objects.aggregate(Avg("friends__age"))
self.assertEqual(vals, {"friends__age__avg": Approximate(34.07, places=2)})
vals = Book.objects.filter(rating__lt=4.5).aggregate(Avg("authors__age"))
self.assertEqual(vals, {"authors__age__avg": Approximate(38.2857, places=2)})
vals = Author.objects.filter(name__contains="a").aggregate(Avg("book__rating"))
self.assertEqual(vals, {"book__rating__avg": 4.0})
vals = Book.objects.aggregate(Sum("publisher__num_awards"))
self.assertEqual(vals, {"publisher__num_awards__sum": 30})
vals = Publisher.objects.aggregate(Sum("book__price"))
self.assertEqual(vals, {"book__price__sum": Decimal("270.27")})
def test_aggregate_multi_join(self):
vals = Store.objects.aggregate(Max("books__authors__age"))
self.assertEqual(vals, {"books__authors__age__max": 57})
vals = Author.objects.aggregate(Min("book__publisher__num_awards"))
self.assertEqual(vals, {"book__publisher__num_awards__min": 1})
def test_aggregate_alias(self):
vals = Store.objects.filter(name="Amazon.com").aggregate(
amazon_mean=Avg("books__rating")
)
self.assertEqual(vals, {"amazon_mean": Approximate(4.08, places=2)})
def test_aggregate_transform(self):
vals = Store.objects.aggregate(min_month=Min("original_opening__month"))
self.assertEqual(vals, {"min_month": 3})
def test_aggregate_join_transform(self):
vals = Publisher.objects.aggregate(min_year=Min("book__pubdate__year"))
self.assertEqual(vals, {"min_year": 1991})
def test_annotate_basic(self):
self.assertQuerySetEqual(
Book.objects.annotate().order_by("pk"),
[
"The Definitive Guide to Django: Web Development Done Right",
"Sams Teach Yourself Django in 24 Hours",
"Practical Django Projects",
"Python Web Development with Django",
"Artificial Intelligence: A Modern Approach",
"Paradigms of Artificial Intelligence Programming: Case Studies in "
"Common Lisp",
],
lambda b: b.name,
)
books = Book.objects.annotate(mean_age=Avg("authors__age"))
b = books.get(pk=self.b1.pk)
self.assertEqual(
b.name, "The Definitive Guide to Django: Web Development Done Right"
)
self.assertEqual(b.mean_age, 34.5)
def test_annotate_defer(self):
qs = (
Book.objects.annotate(page_sum=Sum("pages"))
.defer("name")
.filter(pk=self.b1.pk)
)
rows = [
(
self.b1.id,
"159059725",
447,
"The Definitive Guide to Django: Web Development Done Right",
)
]
self.assertQuerySetEqual(
qs.order_by("pk"), rows, lambda r: (r.id, r.isbn, r.page_sum, r.name)
)
def test_annotate_defer_select_related(self):
qs = (
Book.objects.select_related("contact")
.annotate(page_sum=Sum("pages"))
.defer("name")
.filter(pk=self.b1.pk)
)
rows = [
(
self.b1.id,
"159059725",
447,
"Adrian Holovaty",
"The Definitive Guide to Django: Web Development Done Right",
)
]
self.assertQuerySetEqual(
qs.order_by("pk"),
rows,
lambda r: (r.id, r.isbn, r.page_sum, r.contact.name, r.name),
)
def test_annotate_m2m(self):
books = (
Book.objects.filter(rating__lt=4.5)
.annotate(Avg("authors__age"))
.order_by("name")
)
self.assertQuerySetEqual(
books,
[
("Artificial Intelligence: A Modern Approach", 51.5),
("Practical Django Projects", 29.0),
("Python Web Development with Django", Approximate(30.3, places=1)),
("Sams Teach Yourself Django in 24 Hours", 45.0),
],
lambda b: (b.name, b.authors__age__avg),
)
books = Book.objects.annotate(num_authors=Count("authors")).order_by("name")
self.assertQuerySetEqual(
books,
[
("Artificial Intelligence: A Modern Approach", 2),
(
"Paradigms of Artificial Intelligence Programming: Case Studies in "
"Common Lisp",
1,
),
("Practical Django Projects", 1),
("Python Web Development with Django", 3),
("Sams Teach Yourself Django in 24 Hours", 1),
("The Definitive Guide to Django: Web Development Done Right", 2),
],
lambda b: (b.name, b.num_authors),
)
def test_backwards_m2m_annotate(self):
authors = (
Author.objects.filter(name__contains="a")
.annotate(Avg("book__rating"))
.order_by("name")
)
self.assertQuerySetEqual(
authors,
[
("Adrian Holovaty", 4.5),
("Brad Dayley", 3.0),
("Jacob Kaplan-Moss", 4.5),
("James Bennett", 4.0),
("Paul Bissex", 4.0),
("Stuart Russell", 4.0),
],
lambda a: (a.name, a.book__rating__avg),
)
authors = Author.objects.annotate(num_books=Count("book")).order_by("name")
self.assertQuerySetEqual(
authors,
[
("Adrian Holovaty", 1),
("Brad Dayley", 1),
("Jacob Kaplan-Moss", 1),
("James Bennett", 1),
("Jeffrey Forcier", 1),
("Paul Bissex", 1),
("Peter Norvig", 2),
("Stuart Russell", 1),
("Wesley J. Chun", 1),
],
lambda a: (a.name, a.num_books),
)
def test_reverse_fkey_annotate(self):
books = Book.objects.annotate(Sum("publisher__num_awards")).order_by("name")
self.assertQuerySetEqual(
books,
[
("Artificial Intelligence: A Modern Approach", 7),
(
"Paradigms of Artificial Intelligence Programming: Case Studies in "
"Common Lisp",
9,
),
("Practical Django Projects", 3),
("Python Web Development with Django", 7),
("Sams Teach Yourself Django in 24 Hours", 1),
("The Definitive Guide to Django: Web Development Done Right", 3),
],
lambda b: (b.name, b.publisher__num_awards__sum),
)
publishers = Publisher.objects.annotate(Sum("book__price")).order_by("name")
self.assertQuerySetEqual(
publishers,
[
("Apress", Decimal("59.69")),
("Jonno's House of Books", None),
("Morgan Kaufmann", Decimal("75.00")),
("Prentice Hall", Decimal("112.49")),
("Sams", Decimal("23.09")),
],
lambda p: (p.name, p.book__price__sum),
)
def test_annotate_values(self):
books = list(
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values()
)
self.assertEqual(
books,
[
{
"contact_id": self.a1.id,
"id": self.b1.id,
"isbn": "159059725",
"mean_age": 34.5,
"name": (
"The Definitive Guide to Django: Web Development Done Right"
),
"pages": 447,
"price": Approximate(Decimal("30")),
"pubdate": datetime.date(2007, 12, 6),
"publisher_id": self.p1.id,
"rating": 4.5,
}
],
)
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values("pk", "isbn", "mean_age")
)
self.assertEqual(
list(books),
[
{
"pk": self.b1.pk,
"isbn": "159059725",
"mean_age": 34.5,
}
],
)
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values("name")
)
self.assertEqual(
list(books),
[{"name": "The Definitive Guide to Django: Web Development Done Right"}],
)
books = (
Book.objects.filter(pk=self.b1.pk)
.values()
.annotate(mean_age=Avg("authors__age"))
)
self.assertEqual(
list(books),
[
{
"contact_id": self.a1.id,
"id": self.b1.id,
"isbn": "159059725",
"mean_age": 34.5,
"name": (
"The Definitive Guide to Django: Web Development Done Right"
),
"pages": 447,
"price": Approximate(Decimal("30")),
"pubdate": datetime.date(2007, 12, 6),
"publisher_id": self.p1.id,
"rating": 4.5,
}
],
)
books = (
Book.objects.values("rating")
.annotate(n_authors=Count("authors__id"), mean_age=Avg("authors__age"))
.order_by("rating")
)
self.assertEqual(
list(books),
[
{
"rating": 3.0,
"n_authors": 1,
"mean_age": 45.0,
},
{
"rating": 4.0,
"n_authors": 6,
"mean_age": Approximate(37.16, places=1),
},
{
"rating": 4.5,
"n_authors": 2,
"mean_age": 34.5,
},
{
"rating": 5.0,
"n_authors": 1,
"mean_age": 57.0,
},
],
)
authors = Author.objects.annotate(Avg("friends__age")).order_by("name")
self.assertQuerySetEqual(
authors,
[
("Adrian Holovaty", 32.0),
("Brad Dayley", None),
("Jacob Kaplan-Moss", 29.5),
("James Bennett", 34.0),
("Jeffrey Forcier", 27.0),
("Paul Bissex", 31.0),
("Peter Norvig", 46.0),
("Stuart Russell", 57.0),
("Wesley J. Chun", Approximate(33.66, places=1)),
],
lambda a: (a.name, a.friends__age__avg),
)
def test_count(self):
vals = Book.objects.aggregate(Count("rating"))
self.assertEqual(vals, {"rating__count": 6})
def test_count_star(self):
with self.assertNumQueries(1) as ctx:
Book.objects.aggregate(n=Count("*"))
sql = ctx.captured_queries[0]["sql"]
self.assertIn("SELECT COUNT(*) ", sql)
def test_count_distinct_expression(self):
aggs = Book.objects.aggregate(
distinct_ratings=Count(
Case(When(pages__gt=300, then="rating")), distinct=True
),
)
self.assertEqual(aggs["distinct_ratings"], 4)
def test_distinct_on_aggregate(self):
for aggregate, expected_result in (
(Avg, 4.125),
(Count, 4),
(Sum, 16.5),
):
with self.subTest(aggregate=aggregate.__name__):
books = Book.objects.aggregate(
ratings=aggregate("rating", distinct=True)
)
self.assertEqual(books["ratings"], expected_result)
def test_non_grouped_annotation_not_in_group_by(self):
"""
An annotation not included in values() before an aggregate should be
excluded from the group by clause.
"""
qs = (
Book.objects.annotate(xprice=F("price"))
.filter(rating=4.0)
.values("rating")
.annotate(count=Count("publisher_id", distinct=True))
.values("count", "rating")
.order_by("count")
)
self.assertEqual(list(qs), [{"rating": 4.0, "count": 2}])
def test_grouped_annotation_in_group_by(self):
"""
An annotation included in values() before an aggregate should be
included in the group by clause.
"""
qs = (
Book.objects.annotate(xprice=F("price"))
.filter(rating=4.0)
.values("rating", "xprice")
.annotate(count=Count("publisher_id", distinct=True))
.values("count", "rating")
.order_by("count")
)
self.assertEqual(
list(qs),
[
{"rating": 4.0, "count": 1},
{"rating": 4.0, "count": 2},
],
)
def test_fkey_aggregate(self):
explicit = list(Author.objects.annotate(Count("book__id")))
implicit = list(Author.objects.annotate(Count("book")))
self.assertCountEqual(explicit, implicit)
def test_annotate_ordering(self):
books = (
Book.objects.values("rating")
.annotate(oldest=Max("authors__age"))
.order_by("oldest", "rating")
)
self.assertEqual(
list(books),
[
{"rating": 4.5, "oldest": 35},
{"rating": 3.0, "oldest": 45},
{"rating": 4.0, "oldest": 57},
{"rating": 5.0, "oldest": 57},
],
)
books = (
Book.objects.values("rating")
.annotate(oldest=Max("authors__age"))
.order_by("-oldest", "-rating")
)
self.assertEqual(
list(books),
[
{"rating": 5.0, "oldest": 57},
{"rating": 4.0, "oldest": 57},
{"rating": 3.0, "oldest": 45},
{"rating": 4.5, "oldest": 35},
],
)
def test_aggregate_annotation(self):
vals = Book.objects.annotate(num_authors=Count("authors__id")).aggregate(
Avg("num_authors")
)
self.assertEqual(vals, {"num_authors__avg": Approximate(1.66, places=1)})
def test_avg_duration_field(self):
# Explicit `output_field`.
self.assertEqual(
Publisher.objects.aggregate(Avg("duration", output_field=DurationField())),
{"duration__avg": datetime.timedelta(days=1, hours=12)},
)
# Implicit `output_field`.
self.assertEqual(
Publisher.objects.aggregate(Avg("duration")),
{"duration__avg": datetime.timedelta(days=1, hours=12)},
)
def test_sum_duration_field(self):
self.assertEqual(
Publisher.objects.aggregate(Sum("duration", output_field=DurationField())),
{"duration__sum": datetime.timedelta(days=3)},
)
def test_sum_distinct_aggregate(self):
"""
Sum on a distinct() QuerySet should aggregate only the distinct items.
"""
authors = Author.objects.filter(book__in=[self.b5, self.b6])
self.assertEqual(authors.count(), 3)
distinct_authors = authors.distinct()
self.assertEqual(distinct_authors.count(), 2)
# Selected author ages are 57 and 46
age_sum = distinct_authors.aggregate(Sum("age"))
self.assertEqual(age_sum["age__sum"], 103)
def test_filtering(self):
p = Publisher.objects.create(name="Expensive Publisher", num_awards=0)
Book.objects.create(
name="ExpensiveBook1",
pages=1,
isbn="111",
rating=3.5,
price=Decimal("1000"),
publisher=p,
contact_id=self.a1.id,
pubdate=datetime.date(2008, 12, 1),
)
Book.objects.create(
name="ExpensiveBook2",
pages=1,
isbn="222",
rating=4.0,
price=Decimal("1000"),
publisher=p,
contact_id=self.a1.id,
pubdate=datetime.date(2008, 12, 2),
)
Book.objects.create(
name="ExpensiveBook3",
pages=1,
isbn="333",
rating=4.5,
price=Decimal("35"),
publisher=p,
contact_id=self.a1.id,
pubdate=datetime.date(2008, 12, 3),
)
publishers = (
Publisher.objects.annotate(num_books=Count("book__id"))
.filter(num_books__gt=1)
.order_by("pk")
)
self.assertQuerySetEqual(
publishers,
["Apress", "Prentice Hall", "Expensive Publisher"],
lambda p: p.name,
)
publishers = Publisher.objects.filter(book__price__lt=Decimal("40.0")).order_by(
"pk"
)
self.assertQuerySetEqual(
publishers,
[
"Apress",
"Apress",
"Sams",
"Prentice Hall",
"Expensive Publisher",
],
lambda p: p.name,
)
publishers = (
Publisher.objects.annotate(num_books=Count("book__id"))
.filter(num_books__gt=1, book__price__lt=Decimal("40.0"))
.order_by("pk")
)
self.assertQuerySetEqual(
publishers,
["Apress", "Prentice Hall", "Expensive Publisher"],
lambda p: p.name,
)
publishers = (
Publisher.objects.filter(book__price__lt=Decimal("40.0"))
.annotate(num_books=Count("book__id"))
.filter(num_books__gt=1)
.order_by("pk")
)
self.assertQuerySetEqual(publishers, ["Apress"], lambda p: p.name)
publishers = (
Publisher.objects.annotate(num_books=Count("book"))
.filter(num_books__range=[1, 3])
.order_by("pk")
)
self.assertQuerySetEqual(
publishers,
[
"Apress",
"Sams",
"Prentice Hall",
"Morgan Kaufmann",
"Expensive Publisher",
],
lambda p: p.name,
)
publishers = (
Publisher.objects.annotate(num_books=Count("book"))
.filter(num_books__range=[1, 2])
.order_by("pk")
)
self.assertQuerySetEqual(
publishers,
["Apress", "Sams", "Prentice Hall", "Morgan Kaufmann"],
lambda p: p.name,
)
publishers = (
Publisher.objects.annotate(num_books=Count("book"))
.filter(num_books__in=[1, 3])
.order_by("pk")
)
self.assertQuerySetEqual(
publishers,
["Sams", "Morgan Kaufmann", "Expensive Publisher"],
lambda p: p.name,
)
publishers = Publisher.objects.annotate(num_books=Count("book")).filter(
num_books__isnull=True
)
self.assertEqual(len(publishers), 0)
def test_annotation(self):
vals = Author.objects.filter(pk=self.a1.pk).aggregate(Count("friends__id"))
self.assertEqual(vals, {"friends__id__count": 2})
books = (
Book.objects.annotate(num_authors=Count("authors__name"))
.filter(num_authors__exact=2)
.order_by("pk")
)
self.assertQuerySetEqual(
books,
[
"The Definitive Guide to Django: Web Development Done Right",
"Artificial Intelligence: A Modern Approach",
],
lambda b: b.name,
)
authors = (
Author.objects.annotate(num_friends=Count("friends__id", distinct=True))
.filter(num_friends=0)
.order_by("pk")
)
self.assertQuerySetEqual(authors, ["Brad Dayley"], lambda a: a.name)
publishers = (
Publisher.objects.annotate(num_books=Count("book__id"))
.filter(num_books__gt=1)
.order_by("pk")
)
self.assertQuerySetEqual(
publishers, ["Apress", "Prentice Hall"], lambda p: p.name
)
publishers = (
Publisher.objects.filter(book__price__lt=Decimal("40.0"))
.annotate(num_books=Count("book__id"))
.filter(num_books__gt=1)
)
self.assertQuerySetEqual(publishers, ["Apress"], lambda p: p.name)
books = Book.objects.annotate(num_authors=Count("authors__id")).filter(
authors__name__contains="Norvig", num_authors__gt=1
)
self.assertQuerySetEqual(
books, ["Artificial Intelligence: A Modern Approach"], lambda b: b.name
)
def test_more_aggregation(self):
a = Author.objects.get(name__contains="Norvig")
b = Book.objects.get(name__contains="Done Right")
b.authors.add(a)
b.save()
vals = (
Book.objects.annotate(num_authors=Count("authors__id"))
.filter(authors__name__contains="Norvig", num_authors__gt=1)
.aggregate(Avg("rating"))
)
self.assertEqual(vals, {"rating__avg": 4.25})
def test_even_more_aggregate(self):
publishers = (
Publisher.objects.annotate(
earliest_book=Min("book__pubdate"),
)
.exclude(earliest_book=None)
.order_by("earliest_book")
.values(
"earliest_book",
"num_awards",
"id",
"name",
)
)
self.assertEqual(
list(publishers),
[
{
"earliest_book": datetime.date(1991, 10, 15),
"num_awards": 9,
"id": self.p4.id,
"name": "Morgan Kaufmann",
},
{
"earliest_book": datetime.date(1995, 1, 15),
"num_awards": 7,
"id": self.p3.id,
"name": "Prentice Hall",
},
{
"earliest_book": datetime.date(2007, 12, 6),
"num_awards": 3,
"id": self.p1.id,
"name": "Apress",
},
{
"earliest_book": datetime.date(2008, 3, 3),
"num_awards": 1,
"id": self.p2.id,
"name": "Sams",
},
],
)
vals = Store.objects.aggregate(
Max("friday_night_closing"), Min("original_opening")
)
self.assertEqual(
vals,
{
"friday_night_closing__max": datetime.time(23, 59, 59),
"original_opening__min": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
)
def test_annotate_values_list(self):
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values_list("pk", "isbn", "mean_age")
)
self.assertEqual(list(books), [(self.b1.id, "159059725", 34.5)])
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values_list("isbn")
)
self.assertEqual(list(books), [("159059725",)])
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values_list("mean_age")
)
self.assertEqual(list(books), [(34.5,)])
books = (
Book.objects.filter(pk=self.b1.pk)
.annotate(mean_age=Avg("authors__age"))
.values_list("mean_age", flat=True)
)
self.assertEqual(list(books), [34.5])
books = (
Book.objects.values_list("price")
.annotate(count=Count("price"))
.order_by("-count", "price")
)
self.assertEqual(
list(books),
[
(Decimal("29.69"), 2),
(Decimal("23.09"), 1),
(Decimal("30"), 1),
(Decimal("75"), 1),
(Decimal("82.8"), 1),
],
)
def test_dates_with_aggregation(self):
"""
.dates() returns a distinct set of dates when applied to a
QuerySet with aggregation.
Refs #18056. Previously, .dates() would return distinct (date_kind,
aggregation) sets, in this case (year, num_authors), so 2008 would be
returned twice because there are books from 2008 with a different
number of authors.
"""
dates = Book.objects.annotate(num_authors=Count("authors")).dates(
"pubdate", "year"
)
self.assertSequenceEqual(
dates,
[
datetime.date(1991, 1, 1),
datetime.date(1995, 1, 1),
datetime.date(2007, 1, 1),
datetime.date(2008, 1, 1),
],
)
def test_values_aggregation(self):
# Refs #20782
max_rating = Book.objects.values("rating").aggregate(max_rating=Max("rating"))
self.assertEqual(max_rating["max_rating"], 5)
max_books_per_rating = (
Book.objects.values("rating")
.annotate(books_per_rating=Count("id"))
.aggregate(Max("books_per_rating"))
)
self.assertEqual(max_books_per_rating, {"books_per_rating__max": 3})
def test_ticket17424(self):
"""
Doing exclude() on a foreign model after annotate() doesn't crash.
"""
all_books = list(Book.objects.values_list("pk", flat=True).order_by("pk"))
annotated_books = Book.objects.order_by("pk").annotate(one=Count("id"))
# The value doesn't matter, we just need any negative
# constraint on a related model that's a noop.
excluded_books = annotated_books.exclude(publisher__name="__UNLIKELY_VALUE__")
# Try to generate query tree
str(excluded_books.query)
self.assertQuerySetEqual(excluded_books, all_books, lambda x: x.pk)
# Check internal state
self.assertIsNone(annotated_books.query.alias_map["aggregation_book"].join_type)
self.assertIsNone(excluded_books.query.alias_map["aggregation_book"].join_type)
def test_ticket12886(self):
"""
Aggregation over sliced queryset works correctly.
"""
qs = Book.objects.order_by("-rating")[0:3]
vals = qs.aggregate(average_top3_rating=Avg("rating"))["average_top3_rating"]
self.assertAlmostEqual(vals, 4.5, places=2)
def test_ticket11881(self):
"""
Subqueries do not needlessly contain ORDER BY, SELECT FOR UPDATE or
select_related() stuff.
"""
qs = (
Book.objects.select_for_update()
.order_by("pk")
.select_related("publisher")
.annotate(max_pk=Max("pk"))
)
with CaptureQueriesContext(connection) as captured_queries:
qs.aggregate(avg_pk=Avg("max_pk"))
self.assertEqual(len(captured_queries), 1)
qstr = captured_queries[0]["sql"].lower()
self.assertNotIn("for update", qstr)
forced_ordering = connection.ops.force_no_ordering()
if forced_ordering:
# If the backend needs to force an ordering we make sure it's
# the only "ORDER BY" clause present in the query.
self.assertEqual(
re.findall(r"order by (\w+)", qstr),
[", ".join(f[1][0] for f in forced_ordering).lower()],
)
else:
self.assertNotIn("order by", qstr)
self.assertEqual(qstr.count(" join "), 0)
def test_decimal_max_digits_has_no_effect(self):
Book.objects.all().delete()
a1 = Author.objects.first()
p1 = Publisher.objects.first()
thedate = timezone.now()
for i in range(10):
Book.objects.create(
isbn="abcde{}".format(i),
name="none",
pages=10,
rating=4.0,
price=9999.98,
contact=a1,
publisher=p1,
pubdate=thedate,
)
book = Book.objects.aggregate(price_sum=Sum("price"))
self.assertEqual(book["price_sum"], Decimal("99999.80"))
def test_nonaggregate_aggregation_throws(self):
with self.assertRaisesMessage(TypeError, "fail is not an aggregate expression"):
Book.objects.aggregate(fail=F("price"))
def test_nonfield_annotation(self):
book = Book.objects.annotate(val=Max(Value(2))).first()
self.assertEqual(book.val, 2)
book = Book.objects.annotate(
val=Max(Value(2), output_field=IntegerField())
).first()
self.assertEqual(book.val, 2)
book = Book.objects.annotate(val=Max(2, output_field=IntegerField())).first()
self.assertEqual(book.val, 2)
def test_annotation_expressions(self):
authors = Author.objects.annotate(
combined_ages=Sum(F("age") + F("friends__age"))
).order_by("name")
authors2 = Author.objects.annotate(
combined_ages=Sum("age") + Sum("friends__age")
).order_by("name")
for qs in (authors, authors2):
self.assertQuerySetEqual(
qs,
[
("Adrian Holovaty", 132),
("Brad Dayley", None),
("Jacob Kaplan-Moss", 129),
("James Bennett", 63),
("Jeffrey Forcier", 128),
("Paul Bissex", 120),
("Peter Norvig", 103),
("Stuart Russell", 103),
("Wesley J. Chun", 176),
],
lambda a: (a.name, a.combined_ages),
)
def test_aggregation_expressions(self):
a1 = Author.objects.aggregate(av_age=Sum("age") / Count("*"))
a2 = Author.objects.aggregate(av_age=Sum("age") / Count("age"))
a3 = Author.objects.aggregate(av_age=Avg("age"))
self.assertEqual(a1, {"av_age": 37})
self.assertEqual(a2, {"av_age": 37})
self.assertEqual(a3, {"av_age": Approximate(37.4, places=1)})
def test_avg_decimal_field(self):
v = Book.objects.filter(rating=4).aggregate(avg_price=(Avg("price")))[
"avg_price"
]
self.assertIsInstance(v, Decimal)
self.assertEqual(v, Approximate(Decimal("47.39"), places=2))
def test_order_of_precedence(self):
p1 = Book.objects.filter(rating=4).aggregate(avg_price=(Avg("price") + 2) * 3)
self.assertEqual(p1, {"avg_price": Approximate(Decimal("148.18"), places=2)})
p2 = Book.objects.filter(rating=4).aggregate(avg_price=Avg("price") + 2 * 3)
self.assertEqual(p2, {"avg_price": Approximate(Decimal("53.39"), places=2)})
def test_combine_different_types(self):
msg = (
"Cannot infer type of '+' expression involving these types: FloatField, "
"DecimalField. You must set output_field."
)
qs = Book.objects.annotate(sums=Sum("rating") + Sum("pages") + Sum("price"))
with self.assertRaisesMessage(FieldError, msg):
qs.first()
with self.assertRaisesMessage(FieldError, msg):
qs.first()
b1 = Book.objects.annotate(
sums=Sum(F("rating") + F("pages") + F("price"), output_field=IntegerField())
).get(pk=self.b4.pk)
self.assertEqual(b1.sums, 383)
b2 = Book.objects.annotate(
sums=Sum(F("rating") + F("pages") + F("price"), output_field=FloatField())
).get(pk=self.b4.pk)
self.assertEqual(b2.sums, 383.69)
b3 = Book.objects.annotate(
sums=Sum(F("rating") + F("pages") + F("price"), output_field=DecimalField())
).get(pk=self.b4.pk)
self.assertEqual(b3.sums, Approximate(Decimal("383.69"), places=2))
def test_complex_aggregations_require_kwarg(self):
with self.assertRaisesMessage(
TypeError, "Complex annotations require an alias"
):
Author.objects.annotate(Sum(F("age") + F("friends__age")))
with self.assertRaisesMessage(TypeError, "Complex aggregates require an alias"):
Author.objects.aggregate(Sum("age") / Count("age"))
with self.assertRaisesMessage(TypeError, "Complex aggregates require an alias"):
Author.objects.aggregate(Sum(1))
def test_aggregate_over_complex_annotation(self):
qs = Author.objects.annotate(combined_ages=Sum(F("age") + F("friends__age")))
age = qs.aggregate(max_combined_age=Max("combined_ages"))
self.assertEqual(age["max_combined_age"], 176)
age = qs.aggregate(max_combined_age_doubled=Max("combined_ages") * 2)
self.assertEqual(age["max_combined_age_doubled"], 176 * 2)
age = qs.aggregate(
max_combined_age_doubled=Max("combined_ages") + Max("combined_ages")
)
self.assertEqual(age["max_combined_age_doubled"], 176 * 2)
age = qs.aggregate(
max_combined_age_doubled=Max("combined_ages") + Max("combined_ages"),
sum_combined_age=Sum("combined_ages"),
)
self.assertEqual(age["max_combined_age_doubled"], 176 * 2)
self.assertEqual(age["sum_combined_age"], 954)
age = qs.aggregate(
max_combined_age_doubled=Max("combined_ages") + Max("combined_ages"),
sum_combined_age_doubled=Sum("combined_ages") + Sum("combined_ages"),
)
self.assertEqual(age["max_combined_age_doubled"], 176 * 2)
self.assertEqual(age["sum_combined_age_doubled"], 954 * 2)
def test_values_annotation_with_expression(self):
# ensure the F() is promoted to the group by clause
qs = Author.objects.values("name").annotate(another_age=Sum("age") + F("age"))
a = qs.get(name="Adrian Holovaty")
self.assertEqual(a["another_age"], 68)
qs = qs.annotate(friend_count=Count("friends"))
a = qs.get(name="Adrian Holovaty")
self.assertEqual(a["friend_count"], 2)
qs = (
qs.annotate(combined_age=Sum("age") + F("friends__age"))
.filter(name="Adrian Holovaty")
.order_by("-combined_age")
)
self.assertEqual(
list(qs),
[
{
"name": "Adrian Holovaty",
"another_age": 68,
"friend_count": 1,
"combined_age": 69,
},
{
"name": "Adrian Holovaty",
"another_age": 68,
"friend_count": 1,
"combined_age": 63,
},
],
)
vals = qs.values("name", "combined_age")
self.assertEqual(
list(vals),
[
{"name": "Adrian Holovaty", "combined_age": 69},
{"name": "Adrian Holovaty", "combined_age": 63},
],
)
def test_annotate_values_aggregate(self):
alias_age = (
Author.objects.annotate(age_alias=F("age"))
.values(
"age_alias",
)
.aggregate(sum_age=Sum("age_alias"))
)
age = Author.objects.values("age").aggregate(sum_age=Sum("age"))
self.assertEqual(alias_age["sum_age"], age["sum_age"])
def test_annotate_over_annotate(self):
author = (
Author.objects.annotate(age_alias=F("age"))
.annotate(sum_age=Sum("age_alias"))
.get(name="Adrian Holovaty")
)
other_author = Author.objects.annotate(sum_age=Sum("age")).get(
name="Adrian Holovaty"
)
self.assertEqual(author.sum_age, other_author.sum_age)
def test_aggregate_over_aggregate(self):
msg = "Cannot resolve keyword 'age_agg' into field."
with self.assertRaisesMessage(FieldError, msg):
Author.objects.aggregate(
age_agg=Sum(F("age")),
avg_age=Avg(F("age_agg")),
)
def test_annotated_aggregate_over_annotated_aggregate(self):
with self.assertRaisesMessage(
FieldError, "Cannot compute Sum('id__max'): 'id__max' is an aggregate"
):
Book.objects.annotate(Max("id")).annotate(Sum("id__max"))
class MyMax(Max):
def as_sql(self, compiler, connection):
self.set_source_expressions(self.get_source_expressions()[0:1])
return super().as_sql(compiler, connection)
with self.assertRaisesMessage(
FieldError, "Cannot compute Max('id__max'): 'id__max' is an aggregate"
):
Book.objects.annotate(Max("id")).annotate(my_max=MyMax("id__max", "price"))
def test_multi_arg_aggregate(self):
class MyMax(Max):
output_field = DecimalField()
def as_sql(self, compiler, connection):
copy = self.copy()
copy.set_source_expressions(copy.get_source_expressions()[0:1])
return super(MyMax, copy).as_sql(compiler, connection)
with self.assertRaisesMessage(TypeError, "Complex aggregates require an alias"):
Book.objects.aggregate(MyMax("pages", "price"))
with self.assertRaisesMessage(
TypeError, "Complex annotations require an alias"
):
Book.objects.annotate(MyMax("pages", "price"))
Book.objects.aggregate(max_field=MyMax("pages", "price"))
def test_add_implementation(self):
class MySum(Sum):
pass
# test completely changing how the output is rendered
def lower_case_function_override(self, compiler, connection):
sql, params = compiler.compile(self.source_expressions[0])
substitutions = {
"function": self.function.lower(),
"expressions": sql,
"distinct": "",
}
substitutions.update(self.extra)
return self.template % substitutions, params
setattr(MySum, "as_" + connection.vendor, lower_case_function_override)
qs = Book.objects.annotate(
sums=MySum(
F("rating") + F("pages") + F("price"), output_field=IntegerField()
)
)
self.assertEqual(str(qs.query).count("sum("), 1)
b1 = qs.get(pk=self.b4.pk)
self.assertEqual(b1.sums, 383)
# test changing the dict and delegating
def lower_case_function_super(self, compiler, connection):
self.extra["function"] = self.function.lower()
return super(MySum, self).as_sql(compiler, connection)
setattr(MySum, "as_" + connection.vendor, lower_case_function_super)
qs = Book.objects.annotate(
sums=MySum(
F("rating") + F("pages") + F("price"), output_field=IntegerField()
)
)
self.assertEqual(str(qs.query).count("sum("), 1)
b1 = qs.get(pk=self.b4.pk)
self.assertEqual(b1.sums, 383)
# test overriding all parts of the template
def be_evil(self, compiler, connection):
substitutions = {"function": "MAX", "expressions": "2", "distinct": ""}
substitutions.update(self.extra)
return self.template % substitutions, ()
setattr(MySum, "as_" + connection.vendor, be_evil)
qs = Book.objects.annotate(
sums=MySum(
F("rating") + F("pages") + F("price"), output_field=IntegerField()
)
)
self.assertEqual(str(qs.query).count("MAX("), 1)
b1 = qs.get(pk=self.b4.pk)
self.assertEqual(b1.sums, 2)
def test_complex_values_aggregation(self):
max_rating = Book.objects.values("rating").aggregate(
double_max_rating=Max("rating") + Max("rating")
)
self.assertEqual(max_rating["double_max_rating"], 5 * 2)
max_books_per_rating = (
Book.objects.values("rating")
.annotate(books_per_rating=Count("id") + 5)
.aggregate(Max("books_per_rating"))
)
self.assertEqual(max_books_per_rating, {"books_per_rating__max": 3 + 5})
def test_expression_on_aggregation(self):
qs = (
Publisher.objects.annotate(
price_or_median=Greatest(
Avg("book__rating", output_field=DecimalField()), Avg("book__price")
)
)
.filter(price_or_median__gte=F("num_awards"))
.order_by("num_awards")
)
self.assertQuerySetEqual(qs, [1, 3, 7, 9], lambda v: v.num_awards)
qs2 = (
Publisher.objects.annotate(
rating_or_num_awards=Greatest(
Avg("book__rating"), F("num_awards"), output_field=FloatField()
)
)
.filter(rating_or_num_awards__gt=F("num_awards"))
.order_by("num_awards")
)
self.assertQuerySetEqual(qs2, [1, 3], lambda v: v.num_awards)
def test_arguments_must_be_expressions(self):
msg = "QuerySet.aggregate() received non-expression(s): %s."
with self.assertRaisesMessage(TypeError, msg % FloatField()):
Book.objects.aggregate(FloatField())
with self.assertRaisesMessage(TypeError, msg % True):
Book.objects.aggregate(is_book=True)
with self.assertRaisesMessage(
TypeError, msg % ", ".join([str(FloatField()), "True"])
):
Book.objects.aggregate(FloatField(), Avg("price"), is_book=True)
def test_aggregation_subquery_annotation(self):
"""Subquery annotations are excluded from the GROUP BY if they are
not explicitly grouped against."""
latest_book_pubdate_qs = (
Book.objects.filter(publisher=OuterRef("pk"))
.order_by("-pubdate")
.values("pubdate")[:1]
)
publisher_qs = Publisher.objects.annotate(
latest_book_pubdate=Subquery(latest_book_pubdate_qs),
).annotate(count=Count("book"))
with self.assertNumQueries(1) as ctx:
list(publisher_qs)
self.assertEqual(ctx[0]["sql"].count("SELECT"), 2)
# The GROUP BY should not be by alias either.
self.assertEqual(ctx[0]["sql"].lower().count("latest_book_pubdate"), 1)
def test_aggregation_subquery_annotation_exists(self):
latest_book_pubdate_qs = (
Book.objects.filter(publisher=OuterRef("pk"))
.order_by("-pubdate")
.values("pubdate")[:1]
)
publisher_qs = Publisher.objects.annotate(
latest_book_pubdate=Subquery(latest_book_pubdate_qs),
count=Count("book"),
)
self.assertTrue(publisher_qs.exists())
def test_aggregation_filter_exists(self):
publishers_having_more_than_one_book_qs = (
Book.objects.values("publisher")
.annotate(cnt=Count("isbn"))
.filter(cnt__gt=1)
)
query = publishers_having_more_than_one_book_qs.query.exists()
_, _, group_by = query.get_compiler(connection=connection).pre_sql_setup()
self.assertEqual(len(group_by), 1)
def test_aggregation_exists_annotation(self):
published_books = Book.objects.filter(publisher=OuterRef("pk"))
publisher_qs = Publisher.objects.annotate(
published_book=Exists(published_books),
count=Count("book"),
).values_list("name", flat=True)
self.assertCountEqual(
list(publisher_qs),
[
"Apress",
"Morgan Kaufmann",
"Jonno's House of Books",
"Prentice Hall",
"Sams",
],
)
def test_aggregation_subquery_annotation_values(self):
"""
Subquery annotations and external aliases are excluded from the GROUP
BY if they are not selected.
"""
books_qs = (
Book.objects.annotate(
first_author_the_same_age=Subquery(
Author.objects.filter(
age=OuterRef("contact__friends__age"),
)
.order_by("age")
.values("id")[:1],
)
)
.filter(
publisher=self.p1,
first_author_the_same_age__isnull=False,
)
.annotate(
min_age=Min("contact__friends__age"),
)
.values("name", "min_age")
.order_by("name")
)
self.assertEqual(
list(books_qs),
[
{"name": "Practical Django Projects", "min_age": 34},
{
"name": (
"The Definitive Guide to Django: Web Development Done Right"
),
"min_age": 29,
},
],
)
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_aggregation_subquery_annotation_values_collision(self):
books_rating_qs = Book.objects.filter(
pk=OuterRef("book"),
).values("rating")
publisher_qs = (
Publisher.objects.filter(
book__contact__age__gt=20,
)
.annotate(
rating=Subquery(books_rating_qs),
)
.values("rating")
.annotate(total_count=Count("*"))
.order_by("rating")
)
self.assertEqual(
list(publisher_qs),
[
{"rating": 3.0, "total_count": 1},
{"rating": 4.0, "total_count": 3},
{"rating": 4.5, "total_count": 1},
{"rating": 5.0, "total_count": 1},
],
)
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_aggregation_subquery_annotation_multivalued(self):
"""
Subquery annotations must be included in the GROUP BY if they use
potentially multivalued relations (contain the LOOKUP_SEP).
"""
subquery_qs = Author.objects.filter(
pk=OuterRef("pk"),
book__name=OuterRef("book__name"),
).values("pk")
author_qs = Author.objects.annotate(
subquery_id=Subquery(subquery_qs),
).annotate(count=Count("book"))
self.assertEqual(author_qs.count(), Author.objects.count())
def test_aggregation_order_by_not_selected_annotation_values(self):
result_asc = [
self.b4.pk,
self.b3.pk,
self.b1.pk,
self.b2.pk,
self.b5.pk,
self.b6.pk,
]
result_desc = result_asc[::-1]
tests = [
("min_related_age", result_asc),
("-min_related_age", result_desc),
(F("min_related_age"), result_asc),
(F("min_related_age").asc(), result_asc),
(F("min_related_age").desc(), result_desc),
]
for ordering, expected_result in tests:
with self.subTest(ordering=ordering):
books_qs = (
Book.objects.annotate(
min_age=Min("authors__age"),
)
.annotate(
min_related_age=Coalesce("min_age", "contact__age"),
)
.order_by(ordering)
.values_list("pk", flat=True)
)
self.assertEqual(list(books_qs), expected_result)
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_group_by_subquery_annotation(self):
"""
Subquery annotations are included in the GROUP BY if they are
grouped against.
"""
long_books_count_qs = (
Book.objects.filter(
publisher=OuterRef("pk"),
pages__gt=400,
)
.values("publisher")
.annotate(count=Count("pk"))
.values("count")
)
groups = [
Subquery(long_books_count_qs),
long_books_count_qs,
long_books_count_qs.query,
]
for group in groups:
with self.subTest(group=group.__class__.__name__):
long_books_count_breakdown = Publisher.objects.values_list(
group,
).annotate(total=Count("*"))
self.assertEqual(dict(long_books_count_breakdown), {None: 1, 1: 4})
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_group_by_exists_annotation(self):
"""
Exists annotations are included in the GROUP BY if they are
grouped against.
"""
long_books_qs = Book.objects.filter(
publisher=OuterRef("pk"),
pages__gt=800,
)
has_long_books_breakdown = Publisher.objects.values_list(
Exists(long_books_qs),
).annotate(total=Count("*"))
self.assertEqual(dict(has_long_books_breakdown), {True: 2, False: 3})
def test_group_by_nested_expression_with_params(self):
books_qs = (
Book.objects.annotate(greatest_pages=Greatest("pages", Value(600)))
.values(
"greatest_pages",
)
.annotate(
min_pages=Min("pages"),
least=Least("min_pages", "greatest_pages"),
)
.values_list("least", flat=True)
)
self.assertCountEqual(books_qs, [300, 946, 1132])
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_aggregation_subquery_annotation_related_field(self):
publisher = Publisher.objects.create(name=self.a9.name, num_awards=2)
book = Book.objects.create(
isbn="159059999",
name="Test book.",
pages=819,
rating=2.5,
price=Decimal("14.44"),
contact=self.a9,
publisher=publisher,
pubdate=datetime.date(2019, 12, 6),
)
book.authors.add(self.a5, self.a6, self.a7)
books_qs = (
Book.objects.annotate(
contact_publisher=Subquery(
Publisher.objects.filter(
pk=OuterRef("publisher"),
name=OuterRef("contact__name"),
).values("name")[:1],
)
)
.filter(
contact_publisher__isnull=False,
)
.annotate(count=Count("authors"))
)
with self.assertNumQueries(1) as ctx:
self.assertSequenceEqual(books_qs, [book])
if connection.features.allows_group_by_select_index:
self.assertEqual(ctx[0]["sql"].count("SELECT"), 3)
@skipUnlessDBFeature("supports_subqueries_in_group_by")
def test_aggregation_nested_subquery_outerref(self):
publisher_with_same_name = Publisher.objects.filter(
id__in=Subquery(
Publisher.objects.filter(
name=OuterRef(OuterRef("publisher__name")),
).values("id"),
),
).values(publisher_count=Count("id"))[:1]
books_breakdown = Book.objects.annotate(
publisher_count=Subquery(publisher_with_same_name),
authors_count=Count("authors"),
).values_list("publisher_count", flat=True)
self.assertSequenceEqual(books_breakdown, [1] * 6)
def test_aggregation_exists_multivalued_outeref(self):
self.assertCountEqual(
Publisher.objects.annotate(
books_exists=Exists(
Book.objects.filter(publisher=OuterRef("book__publisher"))
),
books_count=Count("book"),
),
Publisher.objects.all(),
)
def test_filter_in_subquery_or_aggregation(self):
"""
Filtering against an aggregate requires the usage of the HAVING clause.
If such a filter is unionized to a non-aggregate one the latter will
also need to be moved to the HAVING clause and have its grouping
columns used in the GROUP BY.
When this is done with a subquery the specialized logic in charge of
using outer reference columns to group should be used instead of the
subquery itself as the latter might return multiple rows.
"""
authors = Author.objects.annotate(
Count("book"),
).filter(Q(book__count__gt=0) | Q(pk__in=Book.objects.values("authors")))
self.assertCountEqual(authors, Author.objects.all())
def test_aggregation_random_ordering(self):
"""Random() is not included in the GROUP BY when used for ordering."""
authors = Author.objects.annotate(contact_count=Count("book")).order_by("?")
self.assertQuerySetEqual(
authors,
[
("Adrian Holovaty", 1),
("Jacob Kaplan-Moss", 1),
("Brad Dayley", 1),
("James Bennett", 1),
("Jeffrey Forcier", 1),
("Paul Bissex", 1),
("Wesley J. Chun", 1),
("Stuart Russell", 1),
("Peter Norvig", 2),
],
lambda a: (a.name, a.contact_count),
ordered=False,
)
def test_empty_result_optimization(self):
with self.assertNumQueries(0):
self.assertEqual(
Publisher.objects.none().aggregate(
sum_awards=Sum("num_awards"),
books_count=Count("book"),
),
{
"sum_awards": None,
"books_count": 0,
},
)
# Expression without empty_result_set_value forces queries to be
# executed even if they would return an empty result set.
raw_books_count = Func("book", function="COUNT")
raw_books_count.contains_aggregate = True
with self.assertNumQueries(1):
self.assertEqual(
Publisher.objects.none().aggregate(
sum_awards=Sum("num_awards"),
books_count=raw_books_count,
),
{
"sum_awards": None,
"books_count": 0,
},
)
def test_coalesced_empty_result_set(self):
with self.assertNumQueries(0):
self.assertEqual(
Publisher.objects.none().aggregate(
sum_awards=Coalesce(Sum("num_awards"), 0),
)["sum_awards"],
0,
)
# Multiple expressions.
with self.assertNumQueries(0):
self.assertEqual(
Publisher.objects.none().aggregate(
sum_awards=Coalesce(Sum("num_awards"), None, 0),
)["sum_awards"],
0,
)
# Nested coalesce.
with self.assertNumQueries(0):
self.assertEqual(
Publisher.objects.none().aggregate(
sum_awards=Coalesce(Coalesce(Sum("num_awards"), None), 0),
)["sum_awards"],
0,
)
# Expression coalesce.
with self.assertNumQueries(1):
self.assertIsInstance(
Store.objects.none().aggregate(
latest_opening=Coalesce(
Max("original_opening"),
RawSQL("CURRENT_TIMESTAMP", []),
),
)["latest_opening"],
datetime.datetime,
)
def test_aggregation_default_unsupported_by_count(self):
msg = "Count does not allow default."
with self.assertRaisesMessage(TypeError, msg):
Count("age", default=0)
def test_aggregation_default_unset(self):
for Aggregate in [Avg, Max, Min, StdDev, Sum, Variance]:
with self.subTest(Aggregate):
result = Author.objects.filter(age__gt=100).aggregate(
value=Aggregate("age"),
)
self.assertIsNone(result["value"])
def test_aggregation_default_zero(self):
for Aggregate in [Avg, Max, Min, StdDev, Sum, Variance]:
with self.subTest(Aggregate):
result = Author.objects.filter(age__gt=100).aggregate(
value=Aggregate("age", default=0),
)
self.assertEqual(result["value"], 0)
def test_aggregation_default_integer(self):
for Aggregate in [Avg, Max, Min, StdDev, Sum, Variance]:
with self.subTest(Aggregate):
result = Author.objects.filter(age__gt=100).aggregate(
value=Aggregate("age", default=21),
)
self.assertEqual(result["value"], 21)
def test_aggregation_default_expression(self):
for Aggregate in [Avg, Max, Min, StdDev, Sum, Variance]:
with self.subTest(Aggregate):
result = Author.objects.filter(age__gt=100).aggregate(
value=Aggregate("age", default=Value(5) * Value(7)),
)
self.assertEqual(result["value"], 35)
def test_aggregation_default_group_by(self):
qs = (
Publisher.objects.values("name")
.annotate(
books=Count("book"),
pages=Sum("book__pages", default=0),
)
.filter(books=0)
)
self.assertSequenceEqual(
qs,
[{"name": "Jonno's House of Books", "books": 0, "pages": 0}],
)
def test_aggregation_default_compound_expression(self):
# Scale rating to a percentage; default to 50% if no books published.
formula = Avg("book__rating", default=2.5) * 20.0
queryset = Publisher.objects.annotate(rating=formula).order_by("name")
self.assertSequenceEqual(
queryset.values("name", "rating"),
[
{"name": "Apress", "rating": 85.0},
{"name": "Jonno's House of Books", "rating": 50.0},
{"name": "Morgan Kaufmann", "rating": 100.0},
{"name": "Prentice Hall", "rating": 80.0},
{"name": "Sams", "rating": 60.0},
],
)
def test_aggregation_default_using_time_from_python(self):
expr = Min(
"store__friday_night_closing",
filter=~Q(store__name="Amazon.com"),
default=datetime.time(17),
)
if connection.vendor == "mysql":
# Workaround for #30224 for MySQL & MariaDB.
expr.default = Cast(expr.default, TimeField())
queryset = Book.objects.annotate(oldest_store_opening=expr).order_by("isbn")
self.assertSequenceEqual(
queryset.values("isbn", "oldest_store_opening"),
[
{"isbn": "013235613", "oldest_store_opening": datetime.time(21, 30)},
{
"isbn": "013790395",
"oldest_store_opening": datetime.time(23, 59, 59),
},
{"isbn": "067232959", "oldest_store_opening": datetime.time(17)},
{"isbn": "155860191", "oldest_store_opening": datetime.time(21, 30)},
{
"isbn": "159059725",
"oldest_store_opening": datetime.time(23, 59, 59),
},
{"isbn": "159059996", "oldest_store_opening": datetime.time(21, 30)},
],
)
def test_aggregation_default_using_time_from_database(self):
now = timezone.now().astimezone(datetime.timezone.utc)
expr = Min(
"store__friday_night_closing",
filter=~Q(store__name="Amazon.com"),
default=TruncHour(NowUTC(), output_field=TimeField()),
)
queryset = Book.objects.annotate(oldest_store_opening=expr).order_by("isbn")
self.assertSequenceEqual(
queryset.values("isbn", "oldest_store_opening"),
[
{"isbn": "013235613", "oldest_store_opening": datetime.time(21, 30)},
{
"isbn": "013790395",
"oldest_store_opening": datetime.time(23, 59, 59),
},
{"isbn": "067232959", "oldest_store_opening": datetime.time(now.hour)},
{"isbn": "155860191", "oldest_store_opening": datetime.time(21, 30)},
{
"isbn": "159059725",
"oldest_store_opening": datetime.time(23, 59, 59),
},
{"isbn": "159059996", "oldest_store_opening": datetime.time(21, 30)},
],
)
def test_aggregation_default_using_date_from_python(self):
expr = Min("book__pubdate", default=datetime.date(1970, 1, 1))
if connection.vendor == "mysql":
# Workaround for #30224 for MySQL & MariaDB.
expr.default = Cast(expr.default, DateField())
queryset = Publisher.objects.annotate(earliest_pubdate=expr).order_by("name")
self.assertSequenceEqual(
queryset.values("name", "earliest_pubdate"),
[
{"name": "Apress", "earliest_pubdate": datetime.date(2007, 12, 6)},
{
"name": "Jonno's House of Books",
"earliest_pubdate": datetime.date(1970, 1, 1),
},
{
"name": "Morgan Kaufmann",
"earliest_pubdate": datetime.date(1991, 10, 15),
},
{
"name": "Prentice Hall",
"earliest_pubdate": datetime.date(1995, 1, 15),
},
{"name": "Sams", "earliest_pubdate": datetime.date(2008, 3, 3)},
],
)
def test_aggregation_default_using_date_from_database(self):
now = timezone.now().astimezone(datetime.timezone.utc)
expr = Min("book__pubdate", default=TruncDate(NowUTC()))
queryset = Publisher.objects.annotate(earliest_pubdate=expr).order_by("name")
self.assertSequenceEqual(
queryset.values("name", "earliest_pubdate"),
[
{"name": "Apress", "earliest_pubdate": datetime.date(2007, 12, 6)},
{"name": "Jonno's House of Books", "earliest_pubdate": now.date()},
{
"name": "Morgan Kaufmann",
"earliest_pubdate": datetime.date(1991, 10, 15),
},
{
"name": "Prentice Hall",
"earliest_pubdate": datetime.date(1995, 1, 15),
},
{"name": "Sams", "earliest_pubdate": datetime.date(2008, 3, 3)},
],
)
def test_aggregation_default_using_datetime_from_python(self):
expr = Min(
"store__original_opening",
filter=~Q(store__name="Amazon.com"),
default=datetime.datetime(1970, 1, 1),
)
if connection.vendor == "mysql":
# Workaround for #30224 for MySQL & MariaDB.
expr.default = Cast(expr.default, DateTimeField())
queryset = Book.objects.annotate(oldest_store_opening=expr).order_by("isbn")
self.assertSequenceEqual(
queryset.values("isbn", "oldest_store_opening"),
[
{
"isbn": "013235613",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
{
"isbn": "013790395",
"oldest_store_opening": datetime.datetime(2001, 3, 15, 11, 23, 37),
},
{
"isbn": "067232959",
"oldest_store_opening": datetime.datetime(1970, 1, 1),
},
{
"isbn": "155860191",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
{
"isbn": "159059725",
"oldest_store_opening": datetime.datetime(2001, 3, 15, 11, 23, 37),
},
{
"isbn": "159059996",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
],
)
def test_aggregation_default_using_datetime_from_database(self):
now = timezone.now().astimezone(datetime.timezone.utc)
expr = Min(
"store__original_opening",
filter=~Q(store__name="Amazon.com"),
default=TruncHour(NowUTC(), output_field=DateTimeField()),
)
queryset = Book.objects.annotate(oldest_store_opening=expr).order_by("isbn")
self.assertSequenceEqual(
queryset.values("isbn", "oldest_store_opening"),
[
{
"isbn": "013235613",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
{
"isbn": "013790395",
"oldest_store_opening": datetime.datetime(2001, 3, 15, 11, 23, 37),
},
{
"isbn": "067232959",
"oldest_store_opening": now.replace(
minute=0, second=0, microsecond=0, tzinfo=None
),
},
{
"isbn": "155860191",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
{
"isbn": "159059725",
"oldest_store_opening": datetime.datetime(2001, 3, 15, 11, 23, 37),
},
{
"isbn": "159059996",
"oldest_store_opening": datetime.datetime(1945, 4, 25, 16, 24, 14),
},
],
)
def test_aggregation_default_using_duration_from_python(self):
result = Publisher.objects.filter(num_awards__gt=3).aggregate(
value=Sum("duration", default=datetime.timedelta(0)),
)
self.assertEqual(result["value"], datetime.timedelta(0))
def test_aggregation_default_using_duration_from_database(self):
result = Publisher.objects.filter(num_awards__gt=3).aggregate(
value=Sum("duration", default=Now() - Now()),
)
self.assertEqual(result["value"], datetime.timedelta(0))
def test_aggregation_default_using_decimal_from_python(self):
result = Book.objects.filter(rating__lt=3.0).aggregate(
value=Sum("price", default=Decimal("0.00")),
)
self.assertEqual(result["value"], Decimal("0.00"))
def test_aggregation_default_using_decimal_from_database(self):
result = Book.objects.filter(rating__lt=3.0).aggregate(
value=Sum("price", default=Pi()),
)
self.assertAlmostEqual(result["value"], Decimal.from_float(math.pi), places=6)
def test_aggregation_default_passed_another_aggregate(self):
result = Book.objects.aggregate(
value=Sum("price", filter=Q(rating__lt=3.0), default=Avg("pages") / 10.0),
)
self.assertAlmostEqual(result["value"], Decimal("61.72"), places=2)
def test_aggregation_default_after_annotation(self):
result = Publisher.objects.annotate(
double_num_awards=F("num_awards") * 2,
).aggregate(value=Sum("double_num_awards", default=0))
self.assertEqual(result["value"], 40)
def test_aggregation_default_not_in_aggregate(self):
result = Publisher.objects.annotate(
avg_rating=Avg("book__rating", default=2.5),
).aggregate(Sum("num_awards"))
self.assertEqual(result["num_awards__sum"], 20)
def test_exists_none_with_aggregate(self):
qs = Book.objects.annotate(
count=Count("id"),
exists=Exists(Author.objects.none()),
)
self.assertEqual(len(qs), 6)
def test_alias_sql_injection(self):
crafted_alias = """injected_name" from "aggregation_author"; --"""
msg = (
"Column aliases cannot contain whitespace characters, quotation marks, "
"semicolons, or SQL comments."
)
with self.assertRaisesMessage(ValueError, msg):
Author.objects.aggregate(**{crafted_alias: Avg("age")})
def test_exists_extra_where_with_aggregate(self):
qs = Book.objects.annotate(
count=Count("id"),
exists=Exists(Author.objects.extra(where=["1=0"])),
)
self.assertEqual(len(qs), 6)
def test_aggregation_over_annotation_shared_alias(self):
self.assertEqual(
Publisher.objects.annotate(agg=Count("book__authors")).aggregate(
agg=Count("agg"),
),
{"agg": 5},
)
class AggregateAnnotationPruningTests(TestCase):
@classmethod
def setUpTestData(cls):
cls.a1 = Author.objects.create(age=1)
cls.a2 = Author.objects.create(age=2)
cls.p1 = Publisher.objects.create(num_awards=1)
cls.p2 = Publisher.objects.create(num_awards=0)
cls.b1 = Book.objects.create(
name="b1",
publisher=cls.p1,
pages=100,
rating=4.5,
price=10,
contact=cls.a1,
pubdate=datetime.date.today(),
)
cls.b1.authors.add(cls.a1)
cls.b2 = Book.objects.create(
name="b2",
publisher=cls.p2,
pages=1000,
rating=3.2,
price=50,
contact=cls.a2,
pubdate=datetime.date.today(),
)
cls.b2.authors.add(cls.a1, cls.a2)
def test_unused_aliased_aggregate_pruned(self):
with CaptureQueriesContext(connection) as ctx:
cnt = Book.objects.alias(
authors_count=Count("authors"),
).count()
self.assertEqual(cnt, 2)
sql = ctx.captured_queries[0]["sql"].lower()
self.assertEqual(sql.count("select"), 2, "Subquery wrapping required")
self.assertNotIn("authors_count", sql)
def test_non_aggregate_annotation_pruned(self):
with CaptureQueriesContext(connection) as ctx:
Book.objects.annotate(
name_lower=Lower("name"),
).count()
sql = ctx.captured_queries[0]["sql"].lower()
self.assertEqual(sql.count("select"), 1, "No subquery wrapping required")
self.assertNotIn("name_lower", sql)
def test_unreferenced_aggregate_annotation_pruned(self):
with CaptureQueriesContext(connection) as ctx:
cnt = Book.objects.annotate(
authors_count=Count("authors"),
).count()
self.assertEqual(cnt, 2)
sql = ctx.captured_queries[0]["sql"].lower()
self.assertEqual(sql.count("select"), 2, "Subquery wrapping required")
self.assertNotIn("authors_count", sql)
def test_referenced_aggregate_annotation_kept(self):
with CaptureQueriesContext(connection) as ctx:
Book.objects.annotate(
authors_count=Count("authors"),
).aggregate(Avg("authors_count"))
sql = ctx.captured_queries[0]["sql"].lower()
self.assertEqual(sql.count("select"), 2, "Subquery wrapping required")
self.assertEqual(sql.count("authors_count"), 2)