blob: 1ffc6effc7142d9fce1b8ae6b0b42c8b461c8bcc [file] [log] [blame]
Georg Brandl116aa622007-08-15 14:28:22 +00001:mod:`marshal` --- Internal Python object serialization
2=======================================================
3
4.. module:: marshal
5 :synopsis: Convert Python objects to streams of bytes and back (with different
6 constraints).
7
Terry Jan Reedyfa089b92016-06-11 15:02:54 -04008--------------
Georg Brandl116aa622007-08-15 14:28:22 +00009
10This module contains functions that can read and write Python values in a binary
11format. The format is specific to Python, but independent of machine
12architecture issues (e.g., you can write a Python value to a file on a PC,
13transport the file to a Sun, and read it back there). Details of the format are
14undocumented on purpose; it may change between Python versions (although it
15rarely does). [#]_
16
17.. index::
18 module: pickle
19 module: shelve
Georg Brandl116aa622007-08-15 14:28:22 +000020
21This is not a general "persistence" module. For general persistence and
22transfer of Python objects through RPC calls, see the modules :mod:`pickle` and
23:mod:`shelve`. The :mod:`marshal` module exists mainly to support reading and
24writing the "pseudo-compiled" code for Python modules of :file:`.pyc` files.
25Therefore, the Python maintainers reserve the right to modify the marshal format
26in backward incompatible ways should the need arise. If you're serializing and
Georg Brandl9afde1c2007-11-01 20:32:30 +000027de-serializing Python objects, use the :mod:`pickle` module instead -- the
28performance is comparable, version independence is guaranteed, and pickle
29supports a substantially wider range of objects than marshal.
Georg Brandl116aa622007-08-15 14:28:22 +000030
31.. warning::
32
33 The :mod:`marshal` module is not intended to be secure against erroneous or
34 maliciously constructed data. Never unmarshal data received from an
35 untrusted or unauthenticated source.
36
Tommy Beadlee9b84032016-06-02 19:26:51 -040037.. index:: object; code, code object
38
Georg Brandl116aa622007-08-15 14:28:22 +000039Not all Python object types are supported; in general, only objects whose value
40is independent from a particular invocation of Python can be written and read by
Georg Brandl35ac3a82009-09-03 12:34:10 +000041this module. The following types are supported: booleans, integers, floating
42point numbers, complex numbers, strings, bytes, bytearrays, tuples, lists, sets,
43frozensets, dictionaries, and code objects, where it should be understood that
44tuples, lists, sets, frozensets and dictionaries are only supported as long as
Martin Panter3ee62702016-06-04 04:57:19 +000045the values contained therein are themselves supported. The
Georg Brandl35ac3a82009-09-03 12:34:10 +000046singletons :const:`None`, :const:`Ellipsis` and :exc:`StopIteration` can also be
47marshalled and unmarshalled.
Kristján Valur Jónssond7009c62013-03-19 18:02:10 -070048For format *version* lower than 3, recursive lists, sets and dictionaries cannot
49be written (see below).
Georg Brandl116aa622007-08-15 14:28:22 +000050
Georg Brandl116aa622007-08-15 14:28:22 +000051There are functions that read/write files as well as functions operating on
52strings.
53
54The module defines these functions:
55
56
57.. function:: dump(value, file[, version])
58
59 Write the value on the open file. The value must be a supported type. The
60 file must be an open file object such as ``sys.stdout`` or returned by
61 :func:`open` or :func:`os.popen`. It must be opened in binary mode (``'wb'``
62 or ``'w+b'``).
63
64 If the value has (or contains an object that has) an unsupported type, a
65 :exc:`ValueError` exception is raised --- but garbage data will also be written
66 to the file. The object will not be properly read back by :func:`load`.
67
Georg Brandl55ac8f02007-09-01 13:51:09 +000068 The *version* argument indicates the data format that ``dump`` should use
69 (see below).
Georg Brandl116aa622007-08-15 14:28:22 +000070
71
72.. function:: load(file)
73
74 Read one value from the open file and return it. If no valid value is read
75 (e.g. because the data has a different Python version's incompatible marshal
76 format), raise :exc:`EOFError`, :exc:`ValueError` or :exc:`TypeError`. The
77 file must be an open file object opened in binary mode (``'rb'`` or
78 ``'r+b'``).
79
Georg Brandle720c0a2009-04-27 16:20:50 +000080 .. note::
Georg Brandl116aa622007-08-15 14:28:22 +000081
82 If an object containing an unsupported type was marshalled with :func:`dump`,
83 :func:`load` will substitute ``None`` for the unmarshallable type.
84
85
86.. function:: dumps(value[, version])
87
88 Return the string that would be written to a file by ``dump(value, file)``. The
89 value must be a supported type. Raise a :exc:`ValueError` exception if value
90 has (or contains an object that has) an unsupported type.
91
Georg Brandl55ac8f02007-09-01 13:51:09 +000092 The *version* argument indicates the data format that ``dumps`` should use
93 (see below).
Georg Brandl116aa622007-08-15 14:28:22 +000094
95
96.. function:: loads(string)
97
98 Convert the string to a value. If no valid value is found, raise
99 :exc:`EOFError`, :exc:`ValueError` or :exc:`TypeError`. Extra characters in the
100 string are ignored.
101
102
103In addition, the following constants are defined:
104
105.. data:: version
106
Georg Brandle6bcc912008-05-12 18:05:20 +0000107 Indicates the format that the module uses. Version 0 is the historical
108 format, version 1 shares interned strings and version 2 uses a binary format
Kristján Valur Jónssond7009c62013-03-19 18:02:10 -0700109 for floating point numbers.
110 Version 3 adds support for object instancing and recursion.
Benjamin Petersonb4610002015-01-04 16:29:48 -0600111 The current version is 4.
Georg Brandl116aa622007-08-15 14:28:22 +0000112
Georg Brandl116aa622007-08-15 14:28:22 +0000113
114.. rubric:: Footnotes
115
116.. [#] The name of this module stems from a bit of terminology used by the designers of
117 Modula-3 (amongst others), who use the term "marshalling" for shipping of data
118 around in a self-contained form. Strictly speaking, "to marshal" means to
119 convert some data from internal to external form (in an RPC buffer for instance)
120 and "unmarshalling" for the reverse process.
121