Refs #373 -- Added tuple lookups.

6 files changed, 554 insertions, 68 deletions

diff --git a/AUTHORS b/AUTHORS
index 2915761b96..faf6420618 100644
--- a/AUTHORS
+++ b/AUTHORS
@@ -152,6 +152,7 @@ answer newbie questions, and generally made Django that much better:
     Ben Lomax <lomax.on.the.run@gmail.com>
     Ben Slavin <benjamin.slavin@gmail.com>
     Ben Sturmfels <ben@sturm.com.au>
+    Bendegúz Csirmaz <csirmazbendeguz@gmail.com>
     Berker Peksag <berker.peksag@gmail.com>
     Bernd Schlapsi
     Bernhard Essl <me@bernhardessl.com>
diff --git a/django/db/models/expressions.py b/django/db/models/expressions.py
index ffb9f3c816..4a242012ee 100644
--- a/django/db/models/expressions.py
+++ b/django/db/models/expressions.py
@@ -1295,6 +1295,52 @@ class Col(Expression):
         ) + self.target.get_db_converters(connection)
 
 
+class ColPairs(Expression):
+    def __init__(self, alias, targets, sources, output_field):
+        super().__init__(output_field=output_field)
+        self.alias, self.targets, self.sources = alias, targets, sources
+
+    def __len__(self):
+        return len(self.targets)
+
+    def __iter__(self):
+        return iter(self.get_cols())
+
+    def get_cols(self):
+        return [
+            Col(self.alias, target, source)
+            for target, source in zip(self.targets, self.sources)
+        ]
+
+    def get_source_expressions(self):
+        return self.get_cols()
+
+    def set_source_expressions(self, exprs):
+        assert all(isinstance(expr, Col) and expr.alias == self.alias for expr in exprs)
+        self.targets = [col.target for col in exprs]
+        self.sources = [col.field for col in exprs]
+
+    def as_sql(self, compiler, connection):
+        cols_sql = []
+        cols_params = []
+        cols = self.get_cols()
+
+        for col in cols:
+            sql, params = col.as_sql(compiler, connection)
+            cols_sql.append(sql)
+            cols_params.extend(params)
+
+        return ", ".join(cols_sql), cols_params
+
+    def relabeled_clone(self, relabels):
+        return self.__class__(
+            relabels.get(self.alias, self.alias), self.targets, self.sources, self.field
+        )
+
+    def resolve_expression(self, *args, **kwargs):
+        return self
+
+
 class Ref(Expression):
     """
     Reference to column alias of the query. For example, Ref('sum_cost') in
diff --git a/django/db/models/fields/related_lookups.py b/django/db/models/fields/related_lookups.py
index 07a06e1686..22fd17ab4f 100644
--- a/django/db/models/fields/related_lookups.py
+++ b/django/db/models/fields/related_lookups.py
@@ -1,3 +1,6 @@
+from django.db import NotSupportedError
+from django.db.models.expressions import ColPairs
+from django.db.models.fields.tuple_lookups import TupleIn, tuple_lookups
 from django.db.models.lookups import (
     Exact,
     GreaterThan,
@@ -9,34 +12,6 @@ from django.db.models.lookups import (
 )
 
 
-class MultiColSource:
-    contains_aggregate = False
-    contains_over_clause = False
-
-    def __init__(self, alias, targets, sources, field):
-        self.targets, self.sources, self.field, self.alias = (
-            targets,
-            sources,
-            field,
-            alias,
-        )
-        self.output_field = self.field
-
-    def __repr__(self):
-        return "{}({}, {})".format(self.__class__.__name__, self.alias, self.field)
-
-    def relabeled_clone(self, relabels):
-        return self.__class__(
-            relabels.get(self.alias, self.alias), self.targets, self.sources, self.field
-        )
-
-    def get_lookup(self, lookup):
-        return self.output_field.get_lookup(lookup)
-
-    def resolve_expression(self, *args, **kwargs):
-        return self
-
-
 def get_normalized_value(value, lhs):
     from django.db.models import Model
 
@@ -64,7 +39,7 @@ def get_normalized_value(value, lhs):
 
 class RelatedIn(In):
     def get_prep_lookup(self):
-        if not isinstance(self.lhs, MultiColSource):
+        if not isinstance(self.lhs, ColPairs):
             if self.rhs_is_direct_value():
                 # If we get here, we are dealing with single-column relations.
                 self.rhs = [get_normalized_value(val, self.lhs)[0] for val in self.rhs]
@@ -98,49 +73,33 @@ class RelatedIn(In):
         return super().get_prep_lookup()
 
     def as_sql(self, compiler, connection):
-        if isinstance(self.lhs, MultiColSource):
+        if isinstance(self.lhs, ColPairs):
             # For multicolumn lookups we need to build a multicolumn where clause.
             # This clause is either a SubqueryConstraint (for values that need
             # to be compiled to SQL) or an OR-combined list of
             # (col1 = val1 AND col2 = val2 AND ...) clauses.
-            from django.db.models.sql.where import (
-                AND,
-                OR,
-                SubqueryConstraint,
-                WhereNode,
-            )
+            from django.db.models.sql.where import SubqueryConstraint
 
-            root_constraint = WhereNode(connector=OR)
             if self.rhs_is_direct_value():
                 values = [get_normalized_value(value, self.lhs) for value in self.rhs]
-                for value in values:
-                    value_constraint = WhereNode()
-                    for source, target, val in zip(
-                        self.lhs.sources, self.lhs.targets, value
-                    ):
-                        lookup_class = target.get_lookup("exact")
-                        lookup = lookup_class(
-                            target.get_col(self.lhs.alias, source), val
-                        )
-                        value_constraint.add(lookup, AND)
-                    root_constraint.add(value_constraint, OR)
+                lookup = TupleIn(self.lhs, values)
+                return compiler.compile(lookup)
             else:
-                root_constraint.add(
+                return compiler.compile(
                     SubqueryConstraint(
                         self.lhs.alias,
                         [target.column for target in self.lhs.targets],
                         [source.name for source in self.lhs.sources],
                         self.rhs,
                     ),
-                    AND,
                 )
-            return root_constraint.as_sql(compiler, connection)
+
         return super().as_sql(compiler, connection)
 
 
 class RelatedLookupMixin:
     def get_prep_lookup(self):
-        if not isinstance(self.lhs, MultiColSource) and not hasattr(
+        if not isinstance(self.lhs, ColPairs) and not hasattr(
             self.rhs, "resolve_expression"
         ):
             # If we get here, we are dealing with single-column relations.
@@ -158,20 +117,16 @@ class RelatedLookupMixin:
         return super().get_prep_lookup()
 
     def as_sql(self, compiler, connection):
-        if isinstance(self.lhs, MultiColSource):
-            assert self.rhs_is_direct_value()
+        if isinstance(self.lhs, ColPairs):
+            if not self.rhs_is_direct_value():
+                raise NotSupportedError(
+                    f"'{self.lookup_name}' doesn't support multi-column subqueries."
+                )
             self.rhs = get_normalized_value(self.rhs, self.lhs)
-            from django.db.models.sql.where import AND, WhereNode
+            lookup_class = tuple_lookups[self.lookup_name]
+            lookup = lookup_class(self.lhs, self.rhs)
+            return compiler.compile(lookup)
 
-            root_constraint = WhereNode()
-            for target, source, val in zip(
-                self.lhs.targets, self.lhs.sources, self.rhs
-            ):
-                lookup_class = target.get_lookup(self.lookup_name)
-                root_constraint.add(
-                    lookup_class(target.get_col(self.lhs.alias, source), val), AND
-                )
-            return root_constraint.as_sql(compiler, connection)
         return super().as_sql(compiler, connection)
 
 
diff --git a/django/db/models/fields/tuple_lookups.py b/django/db/models/fields/tuple_lookups.py
new file mode 100644
index 0000000000..468826f224
--- /dev/null
+++ b/django/db/models/fields/tuple_lookups.py
@@ -0,0 +1,244 @@
+import itertools
+
+from django.core.exceptions import EmptyResultSet
+from django.db.models.expressions import ColPairs, Func, Value
+from django.db.models.lookups import (
+    Exact,
+    GreaterThan,
+    GreaterThanOrEqual,
+    In,
+    IsNull,
+    LessThan,
+    LessThanOrEqual,
+)
+from django.db.models.sql.where import AND, OR, WhereNode
+
+
+class Tuple(Func):
+    function = ""
+
+
+class TupleLookupMixin:
+    def get_prep_lookup(self):
+        self.check_tuple_lookup()
+        return super().get_prep_lookup()
+
+    def check_tuple_lookup(self):
+        assert isinstance(self.lhs, ColPairs)
+        self.check_rhs_is_tuple_or_list()
+        self.check_rhs_length_equals_lhs_length()
+
+    def check_rhs_is_tuple_or_list(self):
+        if not isinstance(self.rhs, (tuple, list)):
+            raise ValueError(
+                f"'{self.lookup_name}' lookup of '{self.lhs.field.name}' field "
+                "must be a tuple or a list"
+            )
+
+    def check_rhs_length_equals_lhs_length(self):
+        if len(self.lhs) != len(self.rhs):
+            raise ValueError(
+                f"'{self.lookup_name}' lookup of '{self.lhs.field.name}' field "
+                f"must have {len(self.lhs)} elements"
+            )
+
+    def check_rhs_is_collection_of_tuples_or_lists(self):
+        if not all(isinstance(vals, (tuple, list)) for vals in self.rhs):
+            raise ValueError(
+                f"'{self.lookup_name}' lookup of '{self.lhs.field.name}' field "
+                f"must be a collection of tuples or lists"
+            )
+
+    def check_rhs_elements_length_equals_lhs_length(self):
+        if not all(len(self.lhs) == len(vals) for vals in self.rhs):
+            raise ValueError(
+                f"'{self.lookup_name}' lookup of '{self.lhs.field.name}' field "
+                f"must have {len(self.lhs)} elements each"
+            )
+
+    def as_sql(self, compiler, connection):
+        # e.g.: (a, b, c) == (x, y, z) as SQL:
+        # WHERE (a, b, c) = (x, y, z)
+        vals = [
+            Value(val, output_field=col.output_field)
+            for col, val in zip(self.lhs, self.rhs)
+        ]
+        lookup_class = self.__class__.__bases__[-1]
+        lookup = lookup_class(Tuple(self.lhs), Tuple(*vals))
+        return lookup.as_sql(compiler, connection)
+
+
+class TupleExact(TupleLookupMixin, Exact):
+    def as_oracle(self, compiler, connection):
+        # e.g.: (a, b, c) == (x, y, z) as SQL:
+        # WHERE a = x AND b = y AND c = z
+        cols = self.lhs.get_cols()
+        lookups = [Exact(col, val) for col, val in zip(cols, self.rhs)]
+        root = WhereNode(lookups, connector=AND)
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleIsNull(IsNull):
+    def as_sql(self, compiler, connection):
+        # e.g.: (a, b, c) is None as SQL:
+        # WHERE a IS NULL AND b IS NULL AND c IS NULL
+        vals = self.rhs
+        if isinstance(vals, bool):
+            vals = [vals] * len(self.lhs)
+
+        cols = self.lhs.get_cols()
+        lookups = [IsNull(col, val) for col, val in zip(cols, vals)]
+        root = WhereNode(lookups, connector=AND)
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleGreaterThan(TupleLookupMixin, GreaterThan):
+    def as_oracle(self, compiler, connection):
+        # e.g.: (a, b, c) > (x, y, z) as SQL:
+        # WHERE a > x OR (a = x AND (b > y OR (b = y AND c > z)))
+        cols = self.lhs.get_cols()
+        lookups = itertools.cycle([GreaterThan, Exact])
+        connectors = itertools.cycle([OR, AND])
+        cols_list = [col for col in cols for _ in range(2)]
+        vals_list = [val for val in self.rhs for _ in range(2)]
+        cols_iter = iter(cols_list[:-1])
+        vals_iter = iter(vals_list[:-1])
+        col, val = next(cols_iter), next(vals_iter)
+        lookup, connector = next(lookups), next(connectors)
+        root = node = WhereNode([lookup(col, val)], connector=connector)
+
+        for col, val in zip(cols_iter, vals_iter):
+            lookup, connector = next(lookups), next(connectors)
+            child = WhereNode([lookup(col, val)], connector=connector)
+            node.children.append(child)
+            node = child
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleGreaterThanOrEqual(TupleLookupMixin, GreaterThanOrEqual):
+    def as_oracle(self, compiler, connection):
+        # e.g.: (a, b, c) >= (x, y, z) as SQL:
+        # WHERE a > x OR (a = x AND (b > y OR (b = y AND (c > z OR c = z))))
+        cols = self.lhs.get_cols()
+        lookups = itertools.cycle([GreaterThan, Exact])
+        connectors = itertools.cycle([OR, AND])
+        cols_list = [col for col in cols for _ in range(2)]
+        vals_list = [val for val in self.rhs for _ in range(2)]
+        cols_iter = iter(cols_list)
+        vals_iter = iter(vals_list)
+        col, val = next(cols_iter), next(vals_iter)
+        lookup, connector = next(lookups), next(connectors)
+        root = node = WhereNode([lookup(col, val)], connector=connector)
+
+        for col, val in zip(cols_iter, vals_iter):
+            lookup, connector = next(lookups), next(connectors)
+            child = WhereNode([lookup(col, val)], connector=connector)
+            node.children.append(child)
+            node = child
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleLessThan(TupleLookupMixin, LessThan):
+    def as_oracle(self, compiler, connection):
+        # e.g.: (a, b, c) < (x, y, z) as SQL:
+        # WHERE a < x OR (a = x AND (b < y OR (b = y AND c < z)))
+        cols = self.lhs.get_cols()
+        lookups = itertools.cycle([LessThan, Exact])
+        connectors = itertools.cycle([OR, AND])
+        cols_list = [col for col in cols for _ in range(2)]
+        vals_list = [val for val in self.rhs for _ in range(2)]
+        cols_iter = iter(cols_list[:-1])
+        vals_iter = iter(vals_list[:-1])
+        col, val = next(cols_iter), next(vals_iter)
+        lookup, connector = next(lookups), next(connectors)
+        root = node = WhereNode([lookup(col, val)], connector=connector)
+
+        for col, val in zip(cols_iter, vals_iter):
+            lookup, connector = next(lookups), next(connectors)
+            child = WhereNode([lookup(col, val)], connector=connector)
+            node.children.append(child)
+            node = child
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleLessThanOrEqual(TupleLookupMixin, LessThanOrEqual):
+    def as_oracle(self, compiler, connection):
+        # e.g.: (a, b, c) <= (x, y, z) as SQL:
+        # WHERE a < x OR (a = x AND (b < y OR (b = y AND (c < z OR c = z))))
+        cols = self.lhs.get_cols()
+        lookups = itertools.cycle([LessThan, Exact])
+        connectors = itertools.cycle([OR, AND])
+        cols_list = [col for col in cols for _ in range(2)]
+        vals_list = [val for val in self.rhs for _ in range(2)]
+        cols_iter = iter(cols_list)
+        vals_iter = iter(vals_list)
+        col, val = next(cols_iter), next(vals_iter)
+        lookup, connector = next(lookups), next(connectors)
+        root = node = WhereNode([lookup(col, val)], connector=connector)
+
+        for col, val in zip(cols_iter, vals_iter):
+            lookup, connector = next(lookups), next(connectors)
+            child = WhereNode([lookup(col, val)], connector=connector)
+            node.children.append(child)
+            node = child
+
+        return root.as_sql(compiler, connection)
+
+
+class TupleIn(TupleLookupMixin, In):
+    def check_tuple_lookup(self):
+        assert isinstance(self.lhs, ColPairs)
+        self.check_rhs_is_tuple_or_list()
+        self.check_rhs_is_collection_of_tuples_or_lists()
+        self.check_rhs_elements_length_equals_lhs_length()
+
+    def as_sql(self, compiler, connection):
+        if not self.rhs:
+            raise EmptyResultSet
+
+        # e.g.: (a, b, c) in [(x1, y1, z1), (x2, y2, z2)] as SQL:
+        # WHERE (a, b, c) IN ((x1, y1, z1), (x2, y2, z2))
+        rhs = []
+        for vals in self.rhs:
+            rhs.append(
+                Tuple(
+                    *[
+                        Value(val, output_field=col.output_field)
+                        for col, val in zip(self.lhs, vals)
+                    ]
+                )
+            )
+
+        lookup = In(Tuple(self.lhs), Tuple(*rhs))
+        return lookup.as_sql(compiler, connection)
+
+    def as_sqlite(self, compiler, connection):
+        if not self.rhs:
+            raise EmptyResultSet
+
+        # e.g.: (a, b, c) in [(x1, y1, z1), (x2, y2, z2)] as SQL:
+        # WHERE (a = x1 AND b = y1 AND c = z1) OR (a = x2 AND b = y2 AND c = z2)
+        root = WhereNode([], connector=OR)
+        cols = self.lhs.get_cols()
+
+        for vals in self.rhs:
+            lookups = [Exact(col, val) for col, val in zip(cols, vals)]
+            root.children.append(WhereNode(lookups, connector=AND))
+
+        return root.as_sql(compiler, connection)
+
+
+tuple_lookups = {
+    "exact": TupleExact,
+    "gt": TupleGreaterThan,
+    "gte": TupleGreaterThanOrEqual,
+    "lt": TupleLessThan,
+    "lte": TupleLessThanOrEqual,
+    "in": TupleIn,
+    "isnull": TupleIsNull,
+}
diff --git a/django/db/models/sql/query.py b/django/db/models/sql/query.py
index 9a57af2bf3..09916277bc 100644
--- a/django/db/models/sql/query.py
+++ b/django/db/models/sql/query.py
@@ -23,6 +23,7 @@ from django.db.models.constants import LOOKUP_SEP
 from django.db.models.expressions import (
     BaseExpression,
     Col,
+    ColPairs,
     Exists,
     F,
     OuterRef,
@@ -32,7 +33,6 @@ from django.db.models.expressions import (
     Value,
 )
 from django.db.models.fields import Field
-from django.db.models.fields.related_lookups import MultiColSource
 from django.db.models.lookups import Lookup
 from django.db.models.query_utils import (
     Q,
@@ -1549,9 +1549,7 @@ class Query(BaseExpression):
             if len(targets) == 1:
                 col = self._get_col(targets[0], join_info.final_field, alias)
             else:
-                col = MultiColSource(
-                    alias, targets, join_info.targets, join_info.final_field
-                )
+                col = ColPairs(alias, targets, join_info.targets, join_info.final_field)
         else:
             col = self._get_col(targets[0], join_info.final_field, alias)
 
diff --git a/tests/foreign_object/test_tuple_lookups.py b/tests/foreign_object/test_tuple_lookups.py
new file mode 100644
index 0000000000..cf080d084b
--- /dev/null
+++ b/tests/foreign_object/test_tuple_lookups.py
@@ -0,0 +1,242 @@
+import unittest
+
+from django.db import NotSupportedError, connection
+from django.test import TestCase
+
+from .models import Contact, Customer
+
+
+class TupleLookupsTests(TestCase):
+    @classmethod
+    def setUpTestData(cls):
+        super().setUpTestData()
+        cls.customer_1 = Customer.objects.create(customer_id=1, company="a")
+        cls.customer_2 = Customer.objects.create(customer_id=1, company="b")
+        cls.customer_3 = Customer.objects.create(customer_id=2, company="c")
+        cls.customer_4 = Customer.objects.create(customer_id=3, company="d")
+        cls.customer_5 = Customer.objects.create(customer_id=1, company="e")
+        cls.contact_1 = Contact.objects.create(customer=cls.customer_1)
+        cls.contact_2 = Contact.objects.create(customer=cls.customer_1)
+        cls.contact_3 = Contact.objects.create(customer=cls.customer_2)
+        cls.contact_4 = Contact.objects.create(customer=cls.customer_3)
+        cls.contact_5 = Contact.objects.create(customer=cls.customer_1)
+        cls.contact_6 = Contact.objects.create(customer=cls.customer_5)
+
+    def test_exact(self):
+        test_cases = (
+            (self.customer_1, (self.contact_1, self.contact_2, self.contact_5)),
+            (self.customer_2, (self.contact_3,)),
+            (self.customer_3, (self.contact_4,)),
+            (self.customer_4, ()),
+            (self.customer_5, (self.contact_6,)),
+        )
+
+        for customer, contacts in test_cases:
+            with self.subTest(customer=customer, contacts=contacts):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer=customer).order_by("id"), contacts
+                )
+
+    def test_exact_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'exact' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer=subquery).order_by("id"), ()
+            )
+
+    def test_in(self):
+        cust_1, cust_2, cust_3, cust_4, cust_5 = (
+            self.customer_1,
+            self.customer_2,
+            self.customer_3,
+            self.customer_4,
+            self.customer_5,
+        )
+        c1, c2, c3, c4, c5, c6 = (
+            self.contact_1,
+            self.contact_2,
+            self.contact_3,
+            self.contact_4,
+            self.contact_5,
+            self.contact_6,
+        )
+        test_cases = (
+            ((), ()),
+            ((cust_1,), (c1, c2, c5)),
+            ((cust_1, cust_2), (c1, c2, c3, c5)),
+            ((cust_1, cust_2, cust_3), (c1, c2, c3, c4, c5)),
+            ((cust_1, cust_2, cust_3, cust_4), (c1, c2, c3, c4, c5)),
+            ((cust_1, cust_2, cust_3, cust_4, cust_5), (c1, c2, c3, c4, c5, c6)),
+        )
+
+        for contacts, customers in test_cases:
+            with self.subTest(contacts=contacts, customers=customers):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer__in=contacts).order_by("id"),
+                    customers,
+                )
+
+    @unittest.skipIf(
+        connection.vendor == "mysql",
+        "MySQL doesn't support LIMIT & IN/ALL/ANY/SOME subquery",
+    )
+    def test_in_subquery(self):
+        subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+        self.assertSequenceEqual(
+            Contact.objects.filter(customer__in=subquery).order_by("id"),
+            (self.contact_1, self.contact_2, self.contact_5),
+        )
+
+    def test_lt(self):
+        c1, c2, c3, c4, c5, c6 = (
+            self.contact_1,
+            self.contact_2,
+            self.contact_3,
+            self.contact_4,
+            self.contact_5,
+            self.contact_6,
+        )
+        test_cases = (
+            (self.customer_1, ()),
+            (self.customer_2, (c1, c2, c5)),
+            (self.customer_5, (c1, c2, c3, c5)),
+            (self.customer_3, (c1, c2, c3, c5, c6)),
+            (self.customer_4, (c1, c2, c3, c4, c5, c6)),
+        )
+
+        for customer, contacts in test_cases:
+            with self.subTest(customer=customer, contacts=contacts):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer__lt=customer).order_by("id"),
+                    contacts,
+                )
+
+    def test_lt_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'lt' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__lt=subquery).order_by("id"), ()
+            )
+
+    def test_lte(self):
+        c1, c2, c3, c4, c5, c6 = (
+            self.contact_1,
+            self.contact_2,
+            self.contact_3,
+            self.contact_4,
+            self.contact_5,
+            self.contact_6,
+        )
+        test_cases = (
+            (self.customer_1, (c1, c2, c5)),
+            (self.customer_2, (c1, c2, c3, c5)),
+            (self.customer_5, (c1, c2, c3, c5, c6)),
+            (self.customer_3, (c1, c2, c3, c4, c5, c6)),
+            (self.customer_4, (c1, c2, c3, c4, c5, c6)),
+        )
+
+        for customer, contacts in test_cases:
+            with self.subTest(customer=customer, contacts=contacts):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer__lte=customer).order_by("id"),
+                    contacts,
+                )
+
+    def test_lte_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'lte' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__lte=subquery).order_by("id"), ()
+            )
+
+    def test_gt(self):
+        test_cases = (
+            (self.customer_1, (self.contact_3, self.contact_4, self.contact_6)),
+            (self.customer_2, (self.contact_4, self.contact_6)),
+            (self.customer_5, (self.contact_4,)),
+            (self.customer_3, ()),
+            (self.customer_4, ()),
+        )
+
+        for customer, contacts in test_cases:
+            with self.subTest(customer=customer, contacts=contacts):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer__gt=customer).order_by("id"),
+                    contacts,
+                )
+
+    def test_gt_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'gt' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__gt=subquery).order_by("id"), ()
+            )
+
+    def test_gte(self):
+        c1, c2, c3, c4, c5, c6 = (
+            self.contact_1,
+            self.contact_2,
+            self.contact_3,
+            self.contact_4,
+            self.contact_5,
+            self.contact_6,
+        )
+        test_cases = (
+            (self.customer_1, (c1, c2, c3, c4, c5, c6)),
+            (self.customer_2, (c3, c4, c6)),
+            (self.customer_5, (c4, c6)),
+            (self.customer_3, (c4,)),
+            (self.customer_4, ()),
+        )
+
+        for customer, contacts in test_cases:
+            with self.subTest(customer=customer, contacts=contacts):
+                self.assertSequenceEqual(
+                    Contact.objects.filter(customer__gte=customer).order_by("pk"),
+                    contacts,
+                )
+
+    def test_gte_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'gte' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=self.customer_1.id)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__gte=subquery).order_by("id"), ()
+            )
+
+    def test_isnull(self):
+        with self.subTest("customer__isnull=True"):
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__isnull=True).order_by("id"),
+                (),
+            )
+        with self.subTest("customer__isnull=False"):
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__isnull=False).order_by("id"),
+                (
+                    self.contact_1,
+                    self.contact_2,
+                    self.contact_3,
+                    self.contact_4,
+                    self.contact_5,
+                    self.contact_6,
+                ),
+            )
+
+    def test_isnull_subquery(self):
+        with self.assertRaisesMessage(
+            NotSupportedError, "'isnull' doesn't support multi-column subqueries."
+        ):
+            subquery = Customer.objects.filter(id=0)[:1]
+            self.assertSequenceEqual(
+                Contact.objects.filter(customer__isnull=subquery).order_by("id"), ()
+            )