Shortcuts on this page
r m x toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
1# dialects/postgresql/json.py
2# Copyright (C) 2005-2024 the SQLAlchemy authors and contributors
3# <see AUTHORS file>
4#
5# This module is part of SQLAlchemy and is released under
6# the MIT License: https://www.opensource.org/licenses/mit-license.php
7from __future__ import absolute_import
9from ... import types as sqltypes
10from ... import util
11from ...sql import operators
14__all__ = ("JSON", "JSONB")
16idx_precedence = operators._PRECEDENCE[operators.json_getitem_op]
18ASTEXT = operators.custom_op(
19 "->>",
20 precedence=idx_precedence,
21 natural_self_precedent=True,
22 eager_grouping=True,
23)
25JSONPATH_ASTEXT = operators.custom_op(
26 "#>>",
27 precedence=idx_precedence,
28 natural_self_precedent=True,
29 eager_grouping=True,
30)
33HAS_KEY = operators.custom_op(
34 "?",
35 precedence=idx_precedence,
36 natural_self_precedent=True,
37 eager_grouping=True,
38)
40HAS_ALL = operators.custom_op(
41 "?&",
42 precedence=idx_precedence,
43 natural_self_precedent=True,
44 eager_grouping=True,
45)
47HAS_ANY = operators.custom_op(
48 "?|",
49 precedence=idx_precedence,
50 natural_self_precedent=True,
51 eager_grouping=True,
52)
54CONTAINS = operators.custom_op(
55 "@>",
56 precedence=idx_precedence,
57 natural_self_precedent=True,
58 eager_grouping=True,
59)
61CONTAINED_BY = operators.custom_op(
62 "<@",
63 precedence=idx_precedence,
64 natural_self_precedent=True,
65 eager_grouping=True,
66)
69class JSONPathType(sqltypes.JSON.JSONPathType):
70 def bind_processor(self, dialect):
71 super_proc = self.string_bind_processor(dialect)
73 def process(value):
74 assert isinstance(value, util.collections_abc.Sequence)
75 tokens = [util.text_type(elem) for elem in value]
76 value = "{%s}" % (", ".join(tokens))
77 if super_proc:
78 value = super_proc(value)
79 return value
81 return process
83 def literal_processor(self, dialect):
84 super_proc = self.string_literal_processor(dialect)
86 def process(value):
87 assert isinstance(value, util.collections_abc.Sequence)
88 tokens = [util.text_type(elem) for elem in value]
89 value = "{%s}" % (", ".join(tokens))
90 if super_proc:
91 value = super_proc(value)
92 return value
94 return process
97class JSON(sqltypes.JSON):
98 """Represent the PostgreSQL JSON type.
100 :class:`_postgresql.JSON` is used automatically whenever the base
101 :class:`_types.JSON` datatype is used against a PostgreSQL backend,
102 however base :class:`_types.JSON` datatype does not provide Python
103 accessors for PostgreSQL-specific comparison methods such as
104 :meth:`_postgresql.JSON.Comparator.astext`; additionally, to use
105 PostgreSQL ``JSONB``, the :class:`_postgresql.JSONB` datatype should
106 be used explicitly.
108 .. seealso::
110 :class:`_types.JSON` - main documentation for the generic
111 cross-platform JSON datatype.
113 The operators provided by the PostgreSQL version of :class:`_types.JSON`
114 include:
116 * Index operations (the ``->`` operator)::
118 data_table.c.data['some key']
120 data_table.c.data[5]
123 * Index operations returning text (the ``->>`` operator)::
125 data_table.c.data['some key'].astext == 'some value'
127 Note that equivalent functionality is available via the
128 :attr:`.JSON.Comparator.as_string` accessor.
130 * Index operations with CAST
131 (equivalent to ``CAST(col ->> ['some key'] AS <type>)``)::
133 data_table.c.data['some key'].astext.cast(Integer) == 5
135 Note that equivalent functionality is available via the
136 :attr:`.JSON.Comparator.as_integer` and similar accessors.
138 * Path index operations (the ``#>`` operator)::
140 data_table.c.data[('key_1', 'key_2', 5, ..., 'key_n')]
142 * Path index operations returning text (the ``#>>`` operator)::
144 data_table.c.data[('key_1', 'key_2', 5, ..., 'key_n')].astext == 'some value'
146 .. versionchanged:: 1.1 The :meth:`_expression.ColumnElement.cast`
147 operator on
148 JSON objects now requires that the :attr:`.JSON.Comparator.astext`
149 modifier be called explicitly, if the cast works only from a textual
150 string.
152 Index operations return an expression object whose type defaults to
153 :class:`_types.JSON` by default,
154 so that further JSON-oriented instructions
155 may be called upon the result type.
157 Custom serializers and deserializers are specified at the dialect level,
158 that is using :func:`_sa.create_engine`. The reason for this is that when
159 using psycopg2, the DBAPI only allows serializers at the per-cursor
160 or per-connection level. E.g.::
162 engine = create_engine("postgresql://scott:tiger@localhost/test",
163 json_serializer=my_serialize_fn,
164 json_deserializer=my_deserialize_fn
165 )
167 When using the psycopg2 dialect, the json_deserializer is registered
168 against the database using ``psycopg2.extras.register_default_json``.
170 .. seealso::
172 :class:`_types.JSON` - Core level JSON type
174 :class:`_postgresql.JSONB`
176 .. versionchanged:: 1.1 :class:`_postgresql.JSON` is now a PostgreSQL-
177 specific specialization of the new :class:`_types.JSON` type.
179 """ # noqa
181 astext_type = sqltypes.Text()
183 def __init__(self, none_as_null=False, astext_type=None):
184 """Construct a :class:`_types.JSON` type.
186 :param none_as_null: if True, persist the value ``None`` as a
187 SQL NULL value, not the JSON encoding of ``null``. Note that
188 when this flag is False, the :func:`.null` construct can still
189 be used to persist a NULL value::
191 from sqlalchemy import null
192 conn.execute(table.insert(), data=null())
194 .. versionchanged:: 0.9.8 - Added ``none_as_null``, and :func:`.null`
195 is now supported in order to persist a NULL value.
197 .. seealso::
199 :attr:`_types.JSON.NULL`
201 :param astext_type: the type to use for the
202 :attr:`.JSON.Comparator.astext`
203 accessor on indexed attributes. Defaults to :class:`_types.Text`.
205 .. versionadded:: 1.1
207 """
208 super(JSON, self).__init__(none_as_null=none_as_null)
209 if astext_type is not None:
210 self.astext_type = astext_type
212 class Comparator(sqltypes.JSON.Comparator):
213 """Define comparison operations for :class:`_types.JSON`."""
215 @property
216 def astext(self):
217 """On an indexed expression, use the "astext" (e.g. "->>")
218 conversion when rendered in SQL.
220 E.g.::
222 select(data_table.c.data['some key'].astext)
224 .. seealso::
226 :meth:`_expression.ColumnElement.cast`
228 """
229 if isinstance(self.expr.right.type, sqltypes.JSON.JSONPathType):
230 return self.expr.left.operate(
231 JSONPATH_ASTEXT,
232 self.expr.right,
233 result_type=self.type.astext_type,
234 )
235 else:
236 return self.expr.left.operate(
237 ASTEXT, self.expr.right, result_type=self.type.astext_type
238 )
240 comparator_factory = Comparator
243class JSONB(JSON):
244 """Represent the PostgreSQL JSONB type.
246 The :class:`_postgresql.JSONB` type stores arbitrary JSONB format data,
247 e.g.::
249 data_table = Table('data_table', metadata,
250 Column('id', Integer, primary_key=True),
251 Column('data', JSONB)
252 )
254 with engine.connect() as conn:
255 conn.execute(
256 data_table.insert(),
257 data = {"key1": "value1", "key2": "value2"}
258 )
260 The :class:`_postgresql.JSONB` type includes all operations provided by
261 :class:`_types.JSON`, including the same behaviors for indexing
262 operations.
263 It also adds additional operators specific to JSONB, including
264 :meth:`.JSONB.Comparator.has_key`, :meth:`.JSONB.Comparator.has_all`,
265 :meth:`.JSONB.Comparator.has_any`, :meth:`.JSONB.Comparator.contains`,
266 and :meth:`.JSONB.Comparator.contained_by`.
268 Like the :class:`_types.JSON` type, the :class:`_postgresql.JSONB`
269 type does not detect
270 in-place changes when used with the ORM, unless the
271 :mod:`sqlalchemy.ext.mutable` extension is used.
273 Custom serializers and deserializers
274 are shared with the :class:`_types.JSON` class,
275 using the ``json_serializer``
276 and ``json_deserializer`` keyword arguments. These must be specified
277 at the dialect level using :func:`_sa.create_engine`. When using
278 psycopg2, the serializers are associated with the jsonb type using
279 ``psycopg2.extras.register_default_jsonb`` on a per-connection basis,
280 in the same way that ``psycopg2.extras.register_default_json`` is used
281 to register these handlers with the json type.
283 .. versionadded:: 0.9.7
285 .. seealso::
287 :class:`_types.JSON`
289 """
291 __visit_name__ = "JSONB"
293 class Comparator(JSON.Comparator):
294 """Define comparison operations for :class:`_types.JSON`."""
296 def has_key(self, other):
297 """Boolean expression. Test for presence of a key. Note that the
298 key may be a SQLA expression.
299 """
300 return self.operate(HAS_KEY, other, result_type=sqltypes.Boolean)
302 def has_all(self, other):
303 """Boolean expression. Test for presence of all keys in jsonb"""
304 return self.operate(HAS_ALL, other, result_type=sqltypes.Boolean)
306 def has_any(self, other):
307 """Boolean expression. Test for presence of any key in jsonb"""
308 return self.operate(HAS_ANY, other, result_type=sqltypes.Boolean)
310 def contains(self, other, **kwargs):
311 """Boolean expression. Test if keys (or array) are a superset
312 of/contained the keys of the argument jsonb expression.
314 kwargs may be ignored by this operator but are required for API
315 conformance.
316 """
317 return self.operate(CONTAINS, other, result_type=sqltypes.Boolean)
319 def contained_by(self, other):
320 """Boolean expression. Test if keys are a proper subset of the
321 keys of the argument jsonb expression.
322 """
323 return self.operate(
324 CONTAINED_BY, other, result_type=sqltypes.Boolean
325 )
327 comparator_factory = Comparator