• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1
2:mod:`mmap` --- Memory-mapped file support
3==========================================
4
5.. module:: mmap
6   :synopsis: Interface to memory-mapped files for Unix and Windows.
7
8
9Memory-mapped file objects behave like both strings and like file objects.
10Unlike normal string objects, however, these are mutable.  You can use mmap
11objects in most places where strings are expected; for example, you can use
12the :mod:`re` module to search through a memory-mapped file.  Since they're
13mutable, you can change a single character by doing ``obj[index] = 'a'``, or
14change a substring by assigning to a slice: ``obj[i1:i2] = '...'``.  You can
15also read and write data starting at the current file position, and
16:meth:`seek` through the file to different positions.
17
18A memory-mapped file is created by the :class:`~mmap.mmap` constructor, which is
19different on Unix and on Windows.  In either case you must provide a file
20descriptor for a file opened for update. If you wish to map an existing Python
21file object, use its :meth:`fileno` method to obtain the correct value for the
22*fileno* parameter.  Otherwise, you can open the file using the
23:func:`os.open` function, which returns a file descriptor directly (the file
24still needs to be closed when done).
25
26.. note::
27   If you want to create a memory-mapping for a writable, buffered file, you
28   should :func:`~io.IOBase.flush` the file first.  This is necessary to ensure
29   that local modifications to the buffers are actually available to the
30   mapping.
31
32For both the Unix and Windows versions of the constructor, *access* may be
33specified as an optional keyword parameter. *access* accepts one of three
34values: :const:`ACCESS_READ`, :const:`ACCESS_WRITE`, or :const:`ACCESS_COPY`
35to specify read-only, write-through or copy-on-write memory respectively.
36*access* can be used on both Unix and Windows.  If *access* is not specified,
37Windows mmap returns a write-through mapping.  The initial memory values for
38all three access types are taken from the specified file.  Assignment to an
39:const:`ACCESS_READ` memory map raises a :exc:`TypeError` exception.
40Assignment to an :const:`ACCESS_WRITE` memory map affects both memory and the
41underlying file.  Assignment to an :const:`ACCESS_COPY` memory map affects
42memory but does not update the underlying file.
43
44.. versionchanged:: 2.5
45   To map anonymous memory, -1 should be passed as the fileno along with the
46   length.
47
48.. versionchanged:: 2.6
49   mmap.mmap has formerly been a factory function creating mmap objects. Now
50   mmap.mmap is the class itself.
51
52.. class:: mmap(fileno, length[, tagname[, access[, offset]]])
53
54   **(Windows version)** Maps *length* bytes from the file specified by the
55   file handle *fileno*, and creates a mmap object.  If *length* is larger
56   than the current size of the file, the file is extended to contain *length*
57   bytes.  If *length* is ``0``, the maximum length of the map is the current
58   size of the file, except that if the file is empty Windows raises an
59   exception (you cannot create an empty mapping on Windows).
60
61   *tagname*, if specified and not ``None``, is a string giving a tag name for
62   the mapping.  Windows allows you to have many different mappings against
63   the same file.  If you specify the name of an existing tag, that tag is
64   opened, otherwise a new tag of this name is created.  If this parameter is
65   omitted or ``None``, the mapping is created without a name.  Avoiding the
66   use of the tag parameter will assist in keeping your code portable between
67   Unix and Windows.
68
69   *offset* may be specified as a non-negative integer offset. mmap references
70   will be relative to the offset from the beginning of the file. *offset*
71   defaults to 0.  *offset* must be a multiple of the ALLOCATIONGRANULARITY.
72
73
74.. class:: mmap(fileno, length[, flags[, prot[, access[, offset]]]])
75   :noindex:
76
77   **(Unix version)** Maps *length* bytes from the file specified by the file
78   descriptor *fileno*, and returns a mmap object.  If *length* is ``0``, the
79   maximum length of the map will be the current size of the file when
80   :class:`~mmap.mmap` is called.
81
82   *flags* specifies the nature of the mapping. :const:`MAP_PRIVATE` creates a
83   private copy-on-write mapping, so changes to the contents of the mmap
84   object will be private to this process, and :const:`MAP_SHARED` creates a
85   mapping that's shared with all other processes mapping the same areas of
86   the file.  The default value is :const:`MAP_SHARED`.
87
88   *prot*, if specified, gives the desired memory protection; the two most
89   useful values are :const:`PROT_READ` and :const:`PROT_WRITE`, to specify
90   that the pages may be read or written.  *prot* defaults to
91   :const:`PROT_READ \| PROT_WRITE`.
92
93   *access* may be specified in lieu of *flags* and *prot* as an optional
94   keyword parameter.  It is an error to specify both *flags*, *prot* and
95   *access*.  See the description of *access* above for information on how to
96   use this parameter.
97
98   *offset* may be specified as a non-negative integer offset. mmap references
99   will be relative to the offset from the beginning of the file. *offset*
100   defaults to 0.  *offset* must be a multiple of the PAGESIZE or
101   ALLOCATIONGRANULARITY.
102
103   To ensure validity of the created memory mapping the file specified
104   by the descriptor *fileno* is internally automatically synchronized
105   with physical backing store on Mac OS X and OpenVMS.
106
107   This example shows a simple way of using :class:`~mmap.mmap`::
108
109      import mmap
110
111      # write a simple example file
112      with open("hello.txt", "wb") as f:
113          f.write("Hello Python!\n")
114
115      with open("hello.txt", "r+b") as f:
116          # memory-map the file, size 0 means whole file
117          mm = mmap.mmap(f.fileno(), 0)
118          # read content via standard file methods
119          print mm.readline()  # prints "Hello Python!"
120          # read content via slice notation
121          print mm[:5]  # prints "Hello"
122          # update content using slice notation;
123          # note that new content must have same size
124          mm[6:] = " world!\n"
125          # ... and read again using standard file methods
126          mm.seek(0)
127          print mm.readline()  # prints "Hello  world!"
128          # close the map
129          mm.close()
130
131
132   The next example demonstrates how to create an anonymous map and exchange
133   data between the parent and child processes::
134
135      import mmap
136      import os
137
138      mm = mmap.mmap(-1, 13)
139      mm.write("Hello world!")
140
141      pid = os.fork()
142
143      if pid == 0:  # In a child process
144          mm.seek(0)
145          print mm.readline()
146
147          mm.close()
148
149
150   Memory-mapped file objects support the following methods:
151
152
153   .. method:: close()
154
155      Closes the mmap. Subsequent calls to other methods of the object will
156      result in a ValueError exception being raised. This will not close
157      the open file.
158
159
160   .. method:: find(string[, start[, end]])
161
162      Returns the lowest index in the object where the substring *string* is
163      found, such that *string* is contained in the range [*start*, *end*].
164      Optional arguments *start* and *end* are interpreted as in slice notation.
165      Returns ``-1`` on failure.
166
167
168   .. method:: flush([offset, size])
169
170      Flushes changes made to the in-memory copy of a file back to disk. Without
171      use of this call there is no guarantee that changes are written back before
172      the object is destroyed.  If *offset* and *size* are specified, only
173      changes to the given range of bytes will be flushed to disk; otherwise, the
174      whole extent of the mapping is flushed.
175
176      **(Windows version)** A nonzero value returned indicates success; zero
177      indicates failure.
178
179      **(Unix version)** A zero value is returned to indicate success. An
180      exception is raised when the call failed.
181
182
183   .. method:: move(dest, src, count)
184
185      Copy the *count* bytes starting at offset *src* to the destination index
186      *dest*.  If the mmap was created with :const:`ACCESS_READ`, then calls to
187      move will raise a :exc:`TypeError` exception.
188
189
190   .. method:: read(num)
191
192      Return a string containing up to *num* bytes starting from the current
193      file position; the file position is updated to point after the bytes that
194      were returned.
195
196
197   .. method:: read_byte()
198
199      Returns a string of length 1 containing the character at the current file
200      position, and advances the file position by 1.
201
202
203   .. method:: readline()
204
205      Returns a single line, starting at the current file position and up to the
206      next newline.
207
208
209   .. method:: resize(newsize)
210
211      Resizes the map and the underlying file, if any. If the mmap was created
212      with :const:`ACCESS_READ` or :const:`ACCESS_COPY`, resizing the map will
213      raise a :exc:`TypeError` exception.
214
215
216   .. method:: rfind(string[, start[, end]])
217
218      Returns the highest index in the object where the substring *string* is
219      found, such that *string* is contained in the range [*start*, *end*].
220      Optional arguments *start* and *end* are interpreted as in slice notation.
221      Returns ``-1`` on failure.
222
223
224   .. method:: seek(pos[, whence])
225
226      Set the file's current position.  *whence* argument is optional and
227      defaults to ``os.SEEK_SET`` or ``0`` (absolute file positioning); other
228      values are ``os.SEEK_CUR`` or ``1`` (seek relative to the current
229      position) and ``os.SEEK_END`` or ``2`` (seek relative to the file's end).
230
231
232   .. method:: size()
233
234      Return the length of the file, which can be larger than the size of the
235      memory-mapped area.
236
237
238   .. method:: tell()
239
240      Returns the current position of the file pointer.
241
242
243   .. method:: write(string)
244
245      Write the bytes in *string* into memory at the current position of the
246      file pointer; the file position is updated to point after the bytes that
247      were written. If the mmap was created with :const:`ACCESS_READ`, then
248      writing to it will raise a :exc:`TypeError` exception.
249
250
251   .. method:: write_byte(byte)
252
253      Write the single-character string *byte* into memory at the current
254      position of the file pointer; the file position is advanced by ``1``. If
255      the mmap was created with :const:`ACCESS_READ`, then writing to it will
256      raise a :exc:`TypeError` exception.
257