Fixed #29049 -- Added slicing notation to F expressions.

Co-authored-by: Priyansh Saxena <askpriyansh@gmail.com>
Co-authored-by: Niclas Olofsson <n@niclasolofsson.se>
Co-authored-by: David Smith <smithdc@gmail.com>
Co-authored-by: Mariusz Felisiak <felisiak.mariusz@gmail.com>
Co-authored-by: Abhinav Yadav <abhinav.sny.2002@gmail.com>

8 files changed, 256 insertions, 4 deletions

diff --git a/django/contrib/postgres/fields/array.py b/django/contrib/postgres/fields/array.py
index c8e8e132e0..4171af82f9 100644
--- a/django/contrib/postgres/fields/array.py
+++ b/django/contrib/postgres/fields/array.py
@@ -234,6 +234,12 @@ class ArrayField(CheckFieldDefaultMixin, Field):
             }
         )
 
+    def slice_expression(self, expression, start, length):
+        # If length is not provided, don't specify an end to slice to the end
+        # of the array.
+        end = None if length is None else start + length - 1
+        return SliceTransform(start, end, expression)
+
 
 class ArrayRHSMixin:
     def __init__(self, lhs, rhs):
@@ -351,9 +357,11 @@ class SliceTransform(Transform):
 
     def as_sql(self, compiler, connection):
         lhs, params = compiler.compile(self.lhs)
-        if not lhs.endswith("]"):
-            lhs = "(%s)" % lhs
-        return "%s[%%s:%%s]" % lhs, (*params, self.start, self.end)
+        # self.start is set to 1 if slice start is not provided.
+        if self.end is None:
+            return f"({lhs})[%s:]", (*params, self.start)
+        else:
+            return f"({lhs})[%s:%s]", (*params, self.start, self.end)
 
 
 class SliceTransformFactory:
diff --git a/django/db/models/expressions.py b/django/db/models/expressions.py
index b67a2418d4..c20de5995a 100644
--- a/django/db/models/expressions.py
+++ b/django/db/models/expressions.py
@@ -851,6 +851,9 @@ class F(Combinable):
     def __repr__(self):
         return "{}({})".format(self.__class__.__name__, self.name)
 
+    def __getitem__(self, subscript):
+        return Sliced(self, subscript)
+
     def resolve_expression(
         self, query=None, allow_joins=True, reuse=None, summarize=False, for_save=False
     ):
@@ -925,6 +928,63 @@ class OuterRef(F):
         return self
 
 
+class Sliced(F):
+    """
+    An object that contains a slice of an F expression.
+
+    Object resolves the column on which the slicing is applied, and then
+    applies the slicing if possible.
+    """
+
+    def __init__(self, obj, subscript):
+        super().__init__(obj.name)
+        self.obj = obj
+        if isinstance(subscript, int):
+            if subscript < 0:
+                raise ValueError("Negative indexing is not supported.")
+            self.start = subscript + 1
+            self.length = 1
+        elif isinstance(subscript, slice):
+            if (subscript.start is not None and subscript.start < 0) or (
+                subscript.stop is not None and subscript.stop < 0
+            ):
+                raise ValueError("Negative indexing is not supported.")
+            if subscript.step is not None:
+                raise ValueError("Step argument is not supported.")
+            if subscript.stop and subscript.start and subscript.stop < subscript.start:
+                raise ValueError("Slice stop must be greater than slice start.")
+            self.start = 1 if subscript.start is None else subscript.start + 1
+            if subscript.stop is None:
+                self.length = None
+            else:
+                self.length = subscript.stop - (subscript.start or 0)
+        else:
+            raise TypeError("Argument to slice must be either int or slice instance.")
+
+    def __repr__(self):
+        start = self.start - 1
+        stop = None if self.length is None else start + self.length
+        subscript = slice(start, stop)
+        return f"{self.__class__.__qualname__}({self.obj!r}, {subscript!r})"
+
+    def resolve_expression(
+        self,
+        query=None,
+        allow_joins=True,
+        reuse=None,
+        summarize=False,
+        for_save=False,
+    ):
+        resolved = query.resolve_ref(self.name, allow_joins, reuse, summarize)
+        if isinstance(self.obj, (OuterRef, self.__class__)):
+            expr = self.obj.resolve_expression(
+                query, allow_joins, reuse, summarize, for_save
+            )
+        else:
+            expr = resolved
+        return resolved.output_field.slice_expression(expr, self.start, self.length)
+
+
 @deconstructible(path="django.db.models.Func")
 class Func(SQLiteNumericMixin, Expression):
     """An SQL function call."""
diff --git a/django/db/models/fields/__init__.py b/django/db/models/fields/__init__.py
index 41735d3b7f..5186f0c414 100644
--- a/django/db/models/fields/__init__.py
+++ b/django/db/models/fields/__init__.py
@@ -15,6 +15,7 @@ from django.core import checks, exceptions, validators
 from django.db import connection, connections, router
 from django.db.models.constants import LOOKUP_SEP
 from django.db.models.query_utils import DeferredAttribute, RegisterLookupMixin
+from django.db.utils import NotSupportedError
 from django.utils import timezone
 from django.utils.choices import (
     BlankChoiceIterator,
@@ -1143,6 +1144,10 @@ class Field(RegisterLookupMixin):
         """Return the value of this field in the given model instance."""
         return getattr(obj, self.attname)
 
+    def slice_expression(self, expression, start, length):
+        """Return a slice of this field."""
+        raise NotSupportedError("This field does not support slicing.")
+
 
 class BooleanField(Field):
     empty_strings_allowed = False
@@ -1303,6 +1308,11 @@ class CharField(Field):
             kwargs["db_collation"] = self.db_collation
         return name, path, args, kwargs
 
+    def slice_expression(self, expression, start, length):
+        from django.db.models.functions import Substr
+
+        return Substr(expression, start, length)
+
 
 class CommaSeparatedIntegerField(CharField):
     default_validators = [validators.validate_comma_separated_integer_list]
@@ -2497,6 +2507,11 @@ class TextField(Field):
             kwargs["db_collation"] = self.db_collation
         return name, path, args, kwargs
 
+    def slice_expression(self, expression, start, length):
+        from django.db.models.functions import Substr
+
+        return Substr(expression, start, length)
+
 
 class TimeField(DateTimeCheckMixin, Field):
     empty_strings_allowed = False
diff --git a/docs/ref/models/expressions.txt b/docs/ref/models/expressions.txt
index 9d85442d9c..67baef7dfc 100644
--- a/docs/ref/models/expressions.txt
+++ b/docs/ref/models/expressions.txt
@@ -183,6 +183,28 @@ the field value of each one, and saving each one back to the database::
 * getting the database, rather than Python, to do work
 * reducing the number of queries some operations require
 
+.. _slicing-using-f:
+
+Slicing ``F()`` expressions
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. versionadded:: 5.1
+
+For string-based fields, text-based fields, and
+:class:`~django.contrib.postgres.fields.ArrayField`, you can use Python's
+array-slicing syntax. The indices are 0-based and the ``step`` argument to
+``slice`` is not supported. For example:
+
+.. code-block:: pycon
+
+    >>> # Replacing a name with a substring of itself.
+    >>> writer = Writers.objects.get(name="Priyansh")
+    >>> writer.name = F("name")[1:5]
+    >>> writer.save()
+    >>> writer.refresh_from_db()
+    >>> writer.name
+    'riya'
+
 .. _avoiding-race-conditions-using-f:
 
 Avoiding race conditions using ``F()``
diff --git a/docs/releases/5.1.txt b/docs/releases/5.1.txt
index cc72346eef..b825e9be4f 100644
--- a/docs/releases/5.1.txt
+++ b/docs/releases/5.1.txt
@@ -184,6 +184,14 @@ Models
 * :meth:`.QuerySet.order_by` now supports ordering by annotation transforms
   such as ``JSONObject`` keys and ``ArrayAgg`` indices.
 
+* :class:`F() <django.db.models.F>` and :class:`OuterRef()
+  <django.db.models.OuterRef>` expressions that output
+  :class:`~django.db.models.CharField`, :class:`~django.db.models.EmailField`,
+  :class:`~django.db.models.SlugField`, :class:`~django.db.models.URLField`,
+  :class:`~django.db.models.TextField`, or
+  :class:`~django.contrib.postgres.fields.ArrayField` can now be :ref:`sliced
+  <slicing-using-f>`.
+
 Requests and Responses
 ~~~~~~~~~~~~~~~~~~~~~~
 
diff --git a/tests/expressions/models.py b/tests/expressions/models.py
index 0a8a0a6584..bd4db9050e 100644
--- a/tests/expressions/models.py
+++ b/tests/expressions/models.py
@@ -106,3 +106,7 @@ class UUIDPK(models.Model):
 class UUID(models.Model):
     uuid = models.UUIDField(null=True)
     uuid_fk = models.ForeignKey(UUIDPK, models.CASCADE, null=True)
+
+
+class Text(models.Model):
+    name = models.TextField()
diff --git a/tests/expressions/tests.py b/tests/expressions/tests.py
index 073e2e9258..cbb441601c 100644
--- a/tests/expressions/tests.py
+++ b/tests/expressions/tests.py
@@ -84,6 +84,7 @@ from .models import (
     RemoteEmployee,
     Result,
     SimulationRun,
+    Text,
     Time,
 )
 
@@ -205,6 +206,100 @@ class BasicExpressionsTests(TestCase):
             ],
         )
 
+    def _test_slicing_of_f_expressions(self, model):
+        tests = [
+            (F("name")[:], "Example Inc.", "Example Inc."),
+            (F("name")[:7], "Example Inc.", "Example"),
+            (F("name")[:6][:5], "Example", "Examp"),  # Nested slicing.
+            (F("name")[0], "Examp", "E"),
+            (F("name")[5], "E", ""),
+            (F("name")[7:], "Foobar Ltd.", "Ltd."),
+            (F("name")[0:10], "Ltd.", "Ltd."),
+            (F("name")[2:7], "Test GmbH", "st Gm"),
+            (F("name")[1:][:3], "st Gm", "t G"),
+            (F("name")[2:2], "t G", ""),
+        ]
+        for expression, name, expected in tests:
+            with self.subTest(expression=expression, name=name, expected=expected):
+                obj = model.objects.get(name=name)
+                obj.name = expression
+                obj.save()
+                obj.refresh_from_db()
+                self.assertEqual(obj.name, expected)
+
+    def test_slicing_of_f_expressions_charfield(self):
+        self._test_slicing_of_f_expressions(Company)
+
+    def test_slicing_of_f_expressions_textfield(self):
+        Text.objects.bulk_create(
+            [Text(name=company.name) for company in Company.objects.all()]
+        )
+        self._test_slicing_of_f_expressions(Text)
+
+    def test_slicing_of_f_expressions_with_annotate(self):
+        qs = Company.objects.annotate(
+            first_three=F("name")[:3],
+            after_three=F("name")[3:],
+            random_four=F("name")[2:5],
+            first_letter_slice=F("name")[:1],
+            first_letter_index=F("name")[0],
+        )
+        tests = [
+            ("first_three", ["Exa", "Foo", "Tes"]),
+            ("after_three", ["mple Inc.", "bar Ltd.", "t GmbH"]),
+            ("random_four", ["amp", "oba", "st "]),
+            ("first_letter_slice", ["E", "F", "T"]),
+            ("first_letter_index", ["E", "F", "T"]),
+        ]
+        for annotation, expected in tests:
+            with self.subTest(annotation):
+                self.assertCountEqual(qs.values_list(annotation, flat=True), expected)
+
+    def test_slicing_of_f_expression_with_annotated_expression(self):
+        qs = Company.objects.annotate(
+            new_name=Case(
+                When(based_in_eu=True, then=Concat(Value("EU:"), F("name"))),
+                default=F("name"),
+            ),
+            first_two=F("new_name")[:3],
+        )
+        self.assertCountEqual(
+            qs.values_list("first_two", flat=True),
+            ["Exa", "EU:", "Tes"],
+        )
+
+    def test_slicing_of_f_expressions_with_negative_index(self):
+        msg = "Negative indexing is not supported."
+        indexes = [slice(0, -4), slice(-4, 0), slice(-4), -5]
+        for i in indexes:
+            with self.subTest(i=i), self.assertRaisesMessage(ValueError, msg):
+                F("name")[i]
+
+    def test_slicing_of_f_expressions_with_slice_stop_less_than_slice_start(self):
+        msg = "Slice stop must be greater than slice start."
+        with self.assertRaisesMessage(ValueError, msg):
+            F("name")[4:2]
+
+    def test_slicing_of_f_expressions_with_invalid_type(self):
+        msg = "Argument to slice must be either int or slice instance."
+        with self.assertRaisesMessage(TypeError, msg):
+            F("name")["error"]
+
+    def test_slicing_of_f_expressions_with_step(self):
+        msg = "Step argument is not supported."
+        with self.assertRaisesMessage(ValueError, msg):
+            F("name")[::4]
+
+    def test_slicing_of_f_unsupported_field(self):
+        msg = "This field does not support slicing."
+        with self.assertRaisesMessage(NotSupportedError, msg):
+            Company.objects.update(num_chairs=F("num_chairs")[:4])
+
+    def test_slicing_of_outerref(self):
+        inner = Company.objects.filter(name__startswith=OuterRef("ceo__firstname")[0])
+        outer = Company.objects.filter(Exists(inner)).values_list("name", flat=True)
+        self.assertSequenceEqual(outer, ["Foobar Ltd."])
+
     def test_arithmetic(self):
         # We can perform arithmetic operations in expressions
         # Make sure we have 2 spare chairs
@@ -2359,6 +2454,12 @@ class ReprTests(SimpleTestCase):
             repr(Func("published", function="TO_CHAR")),
             "Func(F(published), function=TO_CHAR)",
         )
+        self.assertEqual(
+            repr(F("published")[0:2]), "Sliced(F(published), slice(0, 2, None))"
+        )
+        self.assertEqual(
+            repr(OuterRef("name")[1:5]), "Sliced(OuterRef(name), slice(1, 5, None))"
+        )
         self.assertEqual(repr(OrderBy(Value(1))), "OrderBy(Value(1), descending=False)")
         self.assertEqual(repr(RawSQL("table.col", [])), "RawSQL(table.col, [])")
         self.assertEqual(
diff --git a/tests/postgres_tests/test_array.py b/tests/postgres_tests/test_array.py
index 8aaa7be077..386a0afa3a 100644
--- a/tests/postgres_tests/test_array.py
+++ b/tests/postgres_tests/test_array.py
@@ -10,7 +10,7 @@ from django.core import checks, exceptions, serializers, validators
 from django.core.exceptions import FieldError
 from django.core.management import call_command
 from django.db import IntegrityError, connection, models
-from django.db.models.expressions import Exists, OuterRef, RawSQL, Value
+from django.db.models.expressions import Exists, F, OuterRef, RawSQL, Value
 from django.db.models.functions import Cast, JSONObject, Upper
 from django.test import TransactionTestCase, override_settings, skipUnlessDBFeature
 from django.test.utils import isolate_apps
@@ -594,6 +594,40 @@ class TestQuerying(PostgreSQLTestCase):
             [None, [1], [2], [2, 3], [20, 30]],
         )
 
+    def test_slicing_of_f_expressions(self):
+        tests = [
+            (F("field")[:2], [1, 2]),
+            (F("field")[2:], [3, 4]),
+            (F("field")[1:3], [2, 3]),
+            (F("field")[3], [4]),
+            (F("field")[:3][1:], [2, 3]),  # Nested slicing.
+            (F("field")[:3][1], [2]),  # Slice then index.
+        ]
+        for expression, expected in tests:
+            with self.subTest(expression=expression, expected=expected):
+                instance = IntegerArrayModel.objects.create(field=[1, 2, 3, 4])
+                instance.field = expression
+                instance.save()
+                instance.refresh_from_db()
+                self.assertEqual(instance.field, expected)
+
+    def test_slicing_of_f_expressions_with_annotate(self):
+        IntegerArrayModel.objects.create(field=[1, 2, 3])
+        annotated = IntegerArrayModel.objects.annotate(
+            first_two=F("field")[:2],
+            after_two=F("field")[2:],
+            random_two=F("field")[1:3],
+        ).get()
+        self.assertEqual(annotated.first_two, [1, 2])
+        self.assertEqual(annotated.after_two, [3])
+        self.assertEqual(annotated.random_two, [2, 3])
+
+    def test_slicing_of_f_expressions_with_len(self):
+        queryset = NullableIntegerArrayModel.objects.annotate(
+            subarray=F("field")[:1]
+        ).filter(field__len=F("subarray__len"))
+        self.assertSequenceEqual(queryset, self.objs[:2])
+
     def test_usage_in_subquery(self):
         self.assertSequenceEqual(
             NullableIntegerArrayModel.objects.filter(