• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1:mod:`rexec` --- Restricted execution framework
2===============================================
3
4.. module:: rexec
5   :synopsis: Basic restricted execution framework.
6   :deprecated:
7
8.. deprecated:: 2.6
9   The :mod:`rexec` module has been removed in Python 3.
10
11.. versionchanged:: 2.3
12   Disabled module.
13
14.. warning::
15
16   The documentation has been left in place to help in reading old code that uses
17   the module.
18
19This module contains the :class:`RExec` class, which supports :meth:`r_eval`,
20:meth:`r_execfile`, :meth:`r_exec`, and :meth:`r_import` methods, which are
21restricted versions of the standard Python functions :meth:`eval`,
22:meth:`execfile` and the :keyword:`exec` and :keyword:`import` statements. Code
23executed in this restricted environment will only have access to modules and
24functions that are deemed safe; you can subclass :class:`RExec` to add or remove
25capabilities as desired.
26
27.. warning::
28
29   While the :mod:`rexec` module is designed to perform as described below, it does
30   have a few known vulnerabilities which could be exploited by carefully written
31   code.  Thus it should not be relied upon in situations requiring "production
32   ready" security.  In such situations, execution via sub-processes or very
33   careful "cleansing" of both code and data to be processed may be necessary.
34   Alternatively, help in patching known :mod:`rexec` vulnerabilities would be
35   welcomed.
36
37.. note::
38
39   The :class:`RExec` class can prevent code from performing unsafe operations like
40   reading or writing disk files, or using TCP/IP sockets.  However, it does not
41   protect against code using extremely large amounts of memory or processor time.
42
43
44.. class:: RExec([hooks[, verbose]])
45
46   Returns an instance of the :class:`RExec` class.
47
48   *hooks* is an instance of the :class:`RHooks` class or a subclass of it. If it
49   is omitted or ``None``, the default :class:`RHooks` class is instantiated.
50   Whenever the :mod:`rexec` module searches for a module (even a built-in one) or
51   reads a module's code, it doesn't actually go out to the file system itself.
52   Rather, it calls methods of an :class:`RHooks` instance that was passed to or
53   created by its constructor.  (Actually, the :class:`RExec` object doesn't make
54   these calls --- they are made by a module loader object that's part of the
55   :class:`RExec` object.  This allows another level of flexibility, which can be
56   useful when changing the mechanics of :keyword:`import` within the restricted
57   environment.)
58
59   By providing an alternate :class:`RHooks` object, we can control the file system
60   accesses made to import a module, without changing the actual algorithm that
61   controls the order in which those accesses are made.  For instance, we could
62   substitute an :class:`RHooks` object that passes all filesystem requests to a
63   file server elsewhere, via some RPC mechanism such as ILU.  Grail's applet
64   loader uses this to support importing applets from a URL for a directory.
65
66   If *verbose* is true, additional debugging output may be sent to standard
67   output.
68
69It is important to be aware that code running in a restricted environment can
70still call the :func:`sys.exit` function.  To disallow restricted code from
71exiting the interpreter, always protect calls that cause restricted code to run
72with a :keyword:`try`/:keyword:`except` statement that catches the
73:exc:`SystemExit` exception.  Removing the :func:`sys.exit` function from the
74restricted environment is not sufficient --- the restricted code could still use
75``raise SystemExit``.  Removing :exc:`SystemExit` is not a reasonable option;
76some library code makes use of this and would break were it not available.
77
78
79.. seealso::
80
81   `Grail Home Page <http://grail.sourceforge.net/>`_
82      Grail is a Web browser written entirely in Python.  It uses the :mod:`rexec`
83      module as a foundation for supporting Python applets, and can be used as an
84      example usage of this module.
85
86
87.. _rexec-objects:
88
89RExec Objects
90-------------
91
92:class:`RExec` instances support the following methods:
93
94
95.. method:: RExec.r_eval(code)
96
97   *code* must either be a string containing a Python expression, or a compiled
98   code object, which will be evaluated in the restricted environment's
99   :mod:`__main__` module.  The value of the expression or code object will be
100   returned.
101
102
103.. method:: RExec.r_exec(code)
104
105   *code* must either be a string containing one or more lines of Python code, or a
106   compiled code object, which will be executed in the restricted environment's
107   :mod:`__main__` module.
108
109
110.. method:: RExec.r_execfile(filename)
111
112   Execute the Python code contained in the file *filename* in the restricted
113   environment's :mod:`__main__` module.
114
115Methods whose names begin with ``s_`` are similar to the functions beginning
116with ``r_``, but the code will be granted access to restricted versions of the
117standard I/O streams ``sys.stdin``, ``sys.stderr``, and ``sys.stdout``.
118
119
120.. method:: RExec.s_eval(code)
121
122   *code* must be a string containing a Python expression, which will be evaluated
123   in the restricted environment.
124
125
126.. method:: RExec.s_exec(code)
127
128   *code* must be a string containing one or more lines of Python code, which will
129   be executed in the restricted environment.
130
131
132.. method:: RExec.s_execfile(code)
133
134   Execute the Python code contained in the file *filename* in the restricted
135   environment.
136
137:class:`RExec` objects must also support various methods which will be
138implicitly called by code executing in the restricted environment. Overriding
139these methods in a subclass is used to change the policies enforced by a
140restricted environment.
141
142
143.. method:: RExec.r_import(modulename[, globals[, locals[, fromlist]]])
144
145   Import the module *modulename*, raising an :exc:`ImportError` exception if the
146   module is considered unsafe.
147
148
149.. method:: RExec.r_open(filename[, mode[, bufsize]])
150
151   Method called when :func:`open` is called in the restricted environment.  The
152   arguments are identical to those of :func:`open`, and a file object (or a class
153   instance compatible with file objects) should be returned.  :class:`RExec`'s
154   default behaviour is allow opening any file for reading, but forbidding any
155   attempt to write a file.  See the example below for an implementation of a less
156   restrictive :meth:`r_open`.
157
158
159.. method:: RExec.r_reload(module)
160
161   Reload the module object *module*, re-parsing and re-initializing it.
162
163
164.. method:: RExec.r_unload(module)
165
166   Unload the module object *module* (remove it from the restricted environment's
167   ``sys.modules`` dictionary).
168
169And their equivalents with access to restricted standard I/O streams:
170
171
172.. method:: RExec.s_import(modulename[, globals[, locals[, fromlist]]])
173
174   Import the module *modulename*, raising an :exc:`ImportError` exception if the
175   module is considered unsafe.
176
177
178.. method:: RExec.s_reload(module)
179
180   Reload the module object *module*, re-parsing and re-initializing it.
181
182
183.. method:: RExec.s_unload(module)
184
185   Unload the module object *module*.
186
187   .. XXX what are the semantics of this?
188
189
190.. _rexec-extension:
191
192Defining restricted environments
193--------------------------------
194
195The :class:`RExec` class has the following class attributes, which are used by
196the :meth:`__init__` method.  Changing them on an existing instance won't have
197any effect; instead, create a subclass of :class:`RExec` and assign them new
198values in the class definition. Instances of the new class will then use those
199new values.  All these attributes are tuples of strings.
200
201
202.. attribute:: RExec.nok_builtin_names
203
204   Contains the names of built-in functions which will *not* be available to
205   programs running in the restricted environment.  The value for :class:`RExec` is
206   ``('open', 'reload', '__import__')``. (This gives the exceptions, because by far
207   the majority of built-in functions are harmless.  A subclass that wants to
208   override this variable should probably start with the value from the base class
209   and concatenate additional forbidden functions --- when new dangerous built-in
210   functions are added to Python, they will also be added to this module.)
211
212
213.. attribute:: RExec.ok_builtin_modules
214
215   Contains the names of built-in modules which can be safely imported. The value
216   for :class:`RExec` is ``('audioop', 'array', 'binascii', 'cmath', 'errno',
217   'imageop', 'marshal', 'math', 'md5', 'operator', 'parser', 'regex', 'select',
218   'sha', '_sre', 'strop', 'struct', 'time')``.  A similar remark about overriding
219   this variable applies --- use the value from the base class as a starting point.
220
221
222.. attribute:: RExec.ok_path
223
224   Contains the directories which will be searched when an :keyword:`import` is
225   performed in the restricted environment.   The value for :class:`RExec` is the
226   same as ``sys.path`` (at the time the module is loaded) for unrestricted code.
227
228
229.. attribute:: RExec.ok_posix_names
230
231   Contains the names of the functions in the :mod:`os` module which will be
232   available to programs running in the restricted environment.  The value for
233   :class:`RExec` is ``('error', 'fstat', 'listdir', 'lstat', 'readlink', 'stat',
234   'times', 'uname', 'getpid', 'getppid', 'getcwd', 'getuid', 'getgid', 'geteuid',
235   'getegid')``.
236
237   .. Should this be called ok_os_names?
238
239
240.. attribute:: RExec.ok_sys_names
241
242   Contains the names of the functions and variables in the :mod:`sys` module which
243   will be available to programs running in the restricted environment.  The value
244   for :class:`RExec` is ``('ps1', 'ps2', 'copyright', 'version', 'platform',
245   'exit', 'maxint')``.
246
247
248.. attribute:: RExec.ok_file_types
249
250   Contains the file types from which modules are allowed to be loaded. Each file
251   type is an integer constant defined in the :mod:`imp` module. The meaningful
252   values are :const:`PY_SOURCE`, :const:`PY_COMPILED`, and :const:`C_EXTENSION`.
253   The value for :class:`RExec` is ``(C_EXTENSION, PY_SOURCE)``.  Adding
254   :const:`PY_COMPILED` in subclasses is not recommended; an attacker could exit
255   the restricted execution mode by putting a forged byte-compiled file
256   (:file:`.pyc`) anywhere in your file system, for example by writing it to
257   :file:`/tmp` or uploading it to the :file:`/incoming` directory of your public
258   FTP server.
259
260
261An example
262----------
263
264Let us say that we want a slightly more relaxed policy than the standard
265:class:`RExec` class.  For example, if we're willing to allow files in
266:file:`/tmp` to be written, we can subclass the :class:`RExec` class::
267
268   class TmpWriterRExec(rexec.RExec):
269       def r_open(self, file, mode='r', buf=-1):
270           if mode in ('r', 'rb'):
271               pass
272           elif mode in ('w', 'wb', 'a', 'ab'):
273               # check filename: must begin with /tmp/
274               if file[:5]!='/tmp/':
275                   raise IOError("can't write outside /tmp")
276               elif (string.find(file, '/../') >= 0 or
277                    file[:3] == '../' or file[-3:] == '/..'):
278                   raise IOError("'..' in filename forbidden")
279           else: raise IOError("Illegal open() mode")
280           return open(file, mode, buf)
281
282Notice that the above code will occasionally forbid a perfectly valid filename;
283for example, code in the restricted environment won't be able to open a file
284called :file:`/tmp/foo/../bar`.  To fix this, the :meth:`r_open` method would
285have to simplify the filename to :file:`/tmp/bar`, which would require splitting
286apart the filename and performing various operations on it.  In cases where
287security is at stake, it may be preferable to write simple code which is
288sometimes overly restrictive, instead of more general code that is also more
289complex and may harbor a subtle security hole.
290