/Doc/library/struct.rst

  1
  2:mod:`struct` --- Interpret strings as packed binary data
  3=========================================================
  4
  5.. module:: struct
  6   :synopsis: Interpret strings as packed binary data.
  7
  8.. index::
  9   pair: C; structures
 10   triple: packing; binary; data
 11
 12This module performs conversions between Python values and C structs represented
 13as Python strings.  It uses :dfn:`format strings` (explained below) as compact
 14descriptions of the lay-out of the C structs and the intended conversion to/from
 15Python values.  This can be used in handling binary data stored in files or from
 16network connections, among other sources.
 17
 18The module defines the following exception and functions:
 19
 20
 21.. exception:: error
 22
 23   Exception raised on various occasions; argument is a string describing what is
 24   wrong.
 25
 26
 27.. function:: pack(fmt, v1, v2, ...)
 28
 29   Return a string containing the values ``v1, v2, ...`` packed according to the
 30   given format.  The arguments must match the values required by the format
 31   exactly.
 32
 33
 34.. function:: pack_into(fmt, buffer, offset, v1, v2, ...)
 35
 36   Pack the values ``v1, v2, ...`` according to the given format, write the packed
 37   bytes into the writable *buffer* starting at *offset*. Note that the offset is
 38   a required argument.
 39
 40   .. versionadded:: 2.5
 41
 42
 43.. function:: unpack(fmt, string)
 44
 45   Unpack the string (presumably packed by ``pack(fmt, ...)``) according to the
 46   given format.  The result is a tuple even if it contains exactly one item.  The
 47   string must contain exactly the amount of data required by the format
 48   (``len(string)`` must equal ``calcsize(fmt)``).
 49
 50
 51.. function:: unpack_from(fmt, buffer[,offset=0])
 52
 53   Unpack the *buffer* according to tthe given format. The result is a tuple even
 54   if it contains exactly one item. The *buffer* must contain at least the amount
 55   of data required by the format (``len(buffer[offset:])`` must be at least
 56   ``calcsize(fmt)``).
 57
 58   .. versionadded:: 2.5
 59
 60
 61.. function:: calcsize(fmt)
 62
 63   Return the size of the struct (and hence of the string) corresponding to the
 64   given format.
 65
 66Format characters have the following meaning; the conversion between C and
 67Python values should be obvious given their types:
 68
 69+--------+-------------------------+--------------------+-------+
 70| Format | C Type                  | Python             | Notes |
 71+========+=========================+====================+=======+
 72| ``x``  | pad byte                | no value           |       |
 73+--------+-------------------------+--------------------+-------+
 74| ``c``  | :ctype:`char`           | string of length 1 |       |
 75+--------+-------------------------+--------------------+-------+
 76| ``b``  | :ctype:`signed char`    | integer            |       |
 77+--------+-------------------------+--------------------+-------+
 78| ``B``  | :ctype:`unsigned char`  | integer            |       |
 79+--------+-------------------------+--------------------+-------+
 80| ``?``  | :ctype:`_Bool`          | bool               | \(1)  |
 81+--------+-------------------------+--------------------+-------+
 82| ``h``  | :ctype:`short`          | integer            |       |
 83+--------+-------------------------+--------------------+-------+
 84| ``H``  | :ctype:`unsigned short` | integer            |       |
 85+--------+-------------------------+--------------------+-------+
 86| ``i``  | :ctype:`int`            | integer            |       |
 87+--------+-------------------------+--------------------+-------+
 88| ``I``  | :ctype:`unsigned int`   | integer or long    |       |
 89+--------+-------------------------+--------------------+-------+
 90| ``l``  | :ctype:`long`           | integer            |       |
 91+--------+-------------------------+--------------------+-------+
 92| ``L``  | :ctype:`unsigned long`  | long               |       |
 93+--------+-------------------------+--------------------+-------+
 94| ``q``  | :ctype:`long long`      | long               | \(2)  |
 95+--------+-------------------------+--------------------+-------+
 96| ``Q``  | :ctype:`unsigned long   | long               | \(2)  |
 97|        | long`                   |                    |       |
 98+--------+-------------------------+--------------------+-------+
 99| ``f``  | :ctype:`float`          | float              |       |
100+--------+-------------------------+--------------------+-------+
101| ``d``  | :ctype:`double`         | float              |       |
102+--------+-------------------------+--------------------+-------+
103| ``s``  | :ctype:`char[]`         | string             |       |
104+--------+-------------------------+--------------------+-------+
105| ``p``  | :ctype:`char[]`         | string             |       |
106+--------+-------------------------+--------------------+-------+
107| ``P``  | :ctype:`void \*`        | long               |       |
108+--------+-------------------------+--------------------+-------+
109
110Notes:
111
112(1)
113   The ``'?'`` conversion code corresponds to the :ctype:`_Bool` type defined by
114   C99. If this type is not available, it is simulated using a :ctype:`char`. In
115   standard mode, it is always represented by one byte.
116
117   .. versionadded:: 2.6
118
119(2)
120   The ``'q'`` and ``'Q'`` conversion codes are available in native mode only if
121   the platform C compiler supports C :ctype:`long long`, or, on Windows,
122   :ctype:`__int64`.  They are always available in standard modes.
123
124   .. versionadded:: 2.2
125
126A format character may be preceded by an integral repeat count.  For example,
127the format string ``'4h'`` means exactly the same as ``'hhhh'``.
128
129Whitespace characters between formats are ignored; a count and its format must
130not contain whitespace though.
131
132For the ``'s'`` format character, the count is interpreted as the size of the
133string, not a repeat count like for the other format characters; for example,
134``'10s'`` means a single 10-byte string, while ``'10c'`` means 10 characters.
135For packing, the string is truncated or padded with null bytes as appropriate to
136make it fit. For unpacking, the resulting string always has exactly the
137specified number of bytes.  As a special case, ``'0s'`` means a single, empty
138string (while ``'0c'`` means 0 characters).
139
140The ``'p'`` format character encodes a "Pascal string", meaning a short
141variable-length string stored in a fixed number of bytes. The count is the total
142number of bytes stored.  The first byte stored is the length of the string, or
143255, whichever is smaller.  The bytes of the string follow.  If the string
144passed in to :func:`pack` is too long (longer than the count minus 1), only the
145leading count-1 bytes of the string are stored.  If the string is shorter than
146count-1, it is padded with null bytes so that exactly count bytes in all are
147used.  Note that for :func:`unpack`, the ``'p'`` format character consumes count
148bytes, but that the string returned can never contain more than 255 characters.
149
150For the ``'I'``, ``'L'``, ``'q'`` and ``'Q'`` format characters, the return
151value is a Python long integer.
152
153For the ``'P'`` format character, the return value is a Python integer or long
154integer, depending on the size needed to hold a pointer when it has been cast to
155an integer type.  A *NULL* pointer will always be returned as the Python integer
156``0``. When packing pointer-sized values, Python integer or long integer objects
157may be used.  For example, the Alpha and Merced processors use 64-bit pointer
158values, meaning a Python long integer will be used to hold the pointer; other
159platforms use 32-bit pointers and will use a Python integer.
160
161For the ``'?'`` format character, the return value is either :const:`True` or
162:const:`False`. When packing, the truth value of the argument object is used.
163Either 0 or 1 in the native or standard bool representation will be packed, and
164any non-zero value will be True when unpacking.
165
166By default, C numbers are represented in the machine's native format and byte
167order, and properly aligned by skipping pad bytes if necessary (according to the
168rules used by the C compiler).
169
170Alternatively, the first character of the format string can be used to indicate
171the byte order, size and alignment of the packed data, according to the
172following table:
173
174+-----------+------------------------+--------------------+
175| Character | Byte order             | Size and alignment |
176+===========+========================+====================+
177| ``@``     | native                 | native             |
178+-----------+------------------------+--------------------+
179| ``=``     | native                 | standard           |
180+-----------+------------------------+--------------------+
181| ``<``     | little-endian          | standard           |
182+-----------+------------------------+--------------------+
183| ``>``     | big-endian             | standard           |
184+-----------+------------------------+--------------------+
185| ``!``     | network (= big-endian) | standard           |
186+-----------+------------------------+--------------------+
187
188If the first character is not one of these, ``'@'`` is assumed.
189
190Native byte order is big-endian or little-endian, depending on the host system.
191For example, Motorola and Sun processors are big-endian; Intel and DEC
192processors are little-endian.
193
194Native size and alignment are determined using the C compiler's
195``sizeof`` expression.  This is always combined with native byte order.
196
197Standard size and alignment are as follows: no alignment is required for any
198type (so you have to use pad bytes); :ctype:`short` is 2 bytes; :ctype:`int` and
199:ctype:`long` are 4 bytes; :ctype:`long long` (:ctype:`__int64` on Windows) is 8
200bytes; :ctype:`float` and :ctype:`double` are 32-bit and 64-bit IEEE floating
201point numbers, respectively. :ctype:`_Bool` is 1 byte.
202
203Note the difference between ``'@'`` and ``'='``: both use native byte order, but
204the size and alignment of the latter is standardized.
205
206The form ``'!'`` is available for those poor souls who claim they can't remember
207whether network byte order is big-endian or little-endian.
208
209There is no way to indicate non-native byte order (force byte-swapping); use the
210appropriate choice of ``'<'`` or ``'>'``.
211
212The ``'P'`` format character is only available for the native byte ordering
213(selected as the default or with the ``'@'`` byte order character). The byte
214order character ``'='`` chooses to use little- or big-endian ordering based on
215the host system. The struct module does not interpret this as native ordering,
216so the ``'P'`` format is not available.
217
218Examples (all using native byte order, size and alignment, on a big-endian
219machine)::
220
221   >>> from struct import *
222   >>> pack('hhl', 1, 2, 3)
223   '\x00\x01\x00\x02\x00\x00\x00\x03'
224   >>> unpack('hhl', '\x00\x01\x00\x02\x00\x00\x00\x03')
225   (1, 2, 3)
226   >>> calcsize('hhl')
227   8
228
229Hint: to align the end of a structure to the alignment requirement of a
230particular type, end the format with the code for that type with a repeat count
231of zero.  For example, the format ``'llh0l'`` specifies two pad bytes at the
232end, assuming longs are aligned on 4-byte boundaries.  This only works when
233native size and alignment are in effect; standard size and alignment does not
234enforce any alignment.
235
236Unpacked fields can be named by assigning them to variables or by wrapping
237the result in a named tuple::
238
239    >>> record = 'raymond   \x32\x12\x08\x01\x08'
240    >>> name, serialnum, school, gradelevel = unpack('<10sHHb', record)
241
242    >>> from collections import namedtuple
243    >>> Student = namedtuple('Student', 'name serialnum school gradelevel')
244    >>> Student._make(unpack('<10sHHb', s))
245    Student(name='raymond   ', serialnum=4658, school=264, gradelevel=8)
246
247.. seealso::
248
249   Module :mod:`array`
250      Packed binary storage of homogeneous data.
251
252   Module :mod:`xdrlib`
253      Packing and unpacking of XDR data.
254
255
256.. _struct-objects:
257
258Struct Objects
259--------------
260
261The :mod:`struct` module also defines the following type:
262
263
264.. class:: Struct(format)
265
266   Return a new Struct object which writes and reads binary data according to the
267   format string *format*.  Creating a Struct object once and calling its methods
268   is more efficient than calling the :mod:`struct` functions with the same format
269   since the format string only needs to be compiled once.
270
271   .. versionadded:: 2.5
272
273   Compiled Struct objects support the following methods and attributes:
274
275
276   .. method:: pack(v1, v2, ...)
277
278      Identical to the :func:`pack` function, using the compiled format.
279      (``len(result)`` will equal :attr:`self.size`.)
280
281
282   .. method:: pack_into(buffer, offset, v1, v2, ...)
283
284      Identical to the :func:`pack_into` function, using the compiled format.
285
286
287   .. method:: unpack(string)
288
289      Identical to the :func:`unpack` function, using the compiled format.
290      (``len(string)`` must equal :attr:`self.size`).
291
292
293   .. method:: unpack_from(buffer[, offset=0])
294
295      Identical to the :func:`unpack_from` function, using the compiled format.
296      (``len(buffer[offset:])`` must be at least :attr:`self.size`).
297
298
299   .. attribute:: format
300
301      The format string used to construct this Struct object.
302
303   .. attribute:: size
304
305      The calculated size of the struct (and hence of the string) corresponding
306      to :attr:`format`.
307