gh-57684: Add -P cmdline option and PYTHONSAFEPATH env var

Add the -P command line option and the PYTHONSAFEPATH environment variable to not prepend a potentially unsafe path to sys.path. * Add sys.flags.safe_path flag. * Add PyConfig.safe_path member. * Programs/_bootstrap_python.c uses config.safe_path=0. * Update subprocess._optim_args_from_interpreter_flags() to handle the -P command line option. * Modules/getpath.py sets safe_path to 1 if a "._pth" file is present.
vstinner · May 5, 2022 · Feb 23, 2022 · May 5, 2022 · May 5, 2022 · 33a26d9bea391ebaca0012f4c7cafcd8b4de1a71
commit 33a26d9bea391ebaca0012f4c7cafcd8b4de1a71
@@ -543,6 +543,24 @@ PyConfig
 
       See also the :c:member:`~PyConfig.orig_argv` member.
 
+   .. c:member:: int safe_path
+
+      If equals to zero, ``Py_RunMain()`` prepends a potentially unsafe path to
+      :data:`sys.path` at startup:
+
+      * If :c:member:`argv[0] <PyConfig.argv>` is equal to ``L"-m"``
+        (``python -m module``), prepend the current working directory.
+      * If running a script (``python script.py``), prepend the script path.
+        If it's a symbolic link, resolve symbolic links.
+      * Otherwise (``python -c code`` and ``python``), prepend an empty string.
+
+      Set to 1 by the :option:`-P` command line option and the
+      :envvar:`PYTHONSAFEPATH` environment variable.
+
+      Default: ``0`` in Python config, ``1`` in isolated config.
+
+      .. versionadded:: 3.11
+
    .. c:member:: wchar_t* base_exec_prefix
 
       :data:`sys.base_exec_prefix`.
@@ -809,13 +827,14 @@ PyConfig
 
       If greater than 0, enable isolated mode:
 
-      * :data:`sys.path` contains neither the script's directory (computed from
-        ``argv[0]`` or the current directory) nor the user's site-packages
-        directory.
+      * Set :c:member:`~PyConfig.safe_path` to 1:
+        don't prepend a potentially unsafe path to :data:`sys.path` at Python
+        startup.
+      * Set :c:member:`~PyConfig.use_environment` to 0.
+      * Set :c:member:`~PyConfig.user_site_directory` to 0: don't add the user
+        site directory to :data:`sys.path`.
       * Python REPL doesn't import :mod:`readline` nor enable default readline
         configuration on interactive prompts.
-      * Set :c:member:`~PyConfig.use_environment` and
-        :c:member:`~PyConfig.user_site_directory` to 0.
 
       Default: ``0`` in Python mode, ``1`` in isolated mode.
 
@@ -1029,12 +1048,13 @@ PyConfig
    .. c:member:: wchar_t* run_filename
 
       Filename passed on the command line: trailing command line argument
-      without :option:`-c` or :option:`-m`.
+      without :option:`-c` or :option:`-m`. It is used by the
+      :c:func:`Py_RunMain` function.
 
       For example, it is set to ``script.py`` by the ``python3 script.py arg``
-      command.
+      command line.
 
-      Used by :c:func:`Py_RunMain`.
+      See also the :c:member:`PyConfig.skip_source_first_line` option.
 
       Default: ``NULL``.
 
@@ -1419,9 +1439,16 @@ site-package directory to :data:`sys.path`.
 The following configuration files are used by the path configuration:
 
 * ``pyvenv.cfg``
-* ``python._pth`` (Windows only)
+* ``._pth`` file (ex: ``python._pth``)
 * ``pybuilddir.txt`` (Unix only)
 
+If a ``._pth`` file is present:
+
+* Set :c:member:`~PyConfig.isolated` to 1.
+* Set :c:member:`~PyConfig.use_environment` to 0.
+* Set :c:member:`~PyConfig.site_import` to 0.
+* Set :c:member:`~PyConfig.safe_path` to 1.
+
 The ``__PYVENV_LAUNCHER__`` environment variable is used to set
 :c:member:`PyConfig.base_executable`
 

diff --git a/Doc/library/sys.rst b/Doc/library/sys.rst
@@ -520,6 +520,7 @@ always available.
    :const:`hash_randomization`   :option:`-R`
    :const:`dev_mode`             :option:`-X dev <-X>` (:ref:`Python Development Mode <devmode>`)
    :const:`utf8_mode`            :option:`-X utf8 <-X>`
+   :const:`safe_path`            :option:`-P`
    ============================= ================================================================
 
    .. versionchanged:: 3.2
@@ -539,6 +540,9 @@ always available.
       Mode <devmode>` and the ``utf8_mode`` attribute for the new  :option:`-X`
       ``utf8`` flag.
 
+   .. versionchanged:: 3.11
+      Added the ``safe_path`` attribute for :option:`-P` option.
+
 
 .. data:: float_info
 
@@ -1138,15 +1142,19 @@ always available.
    the environment variable :envvar:`PYTHONPATH`, plus an installation-dependent
    default.
 
-   As initialized upon program startup, the first item of this list, ``path[0]``,
-   is the directory containing the script that was used to invoke the Python
-   interpreter.  If the script directory is not available (e.g.  if the interpreter
-   is invoked interactively or if the script is read from standard input),
-   ``path[0]`` is the empty string, which directs Python to search modules in the
-   current directory first.  Notice that the script directory is inserted *before*
-   the entries inserted as a result of :envvar:`PYTHONPATH`.
+   By default, as initialized upon program startup, a potentially unsafe path
+   is prepended to :data:`sys.path` (*before* the entries inserted as a result
+   of :envvar:`PYTHONPATH`):
+
+   * ``python -m module`` command line: prepend the current working
+     directory.
+   * ``python script.py`` command line: prepend the script path.
+     If it's a symbolic link, resolve symbolic links.
+   * ``python -c code`` and ``python`` (REPL) command lines: prepend an empty
+     string.
 
-   The initialization of :data:`sys.path` is documented at :ref:`sys-path-init`.
+   To not prepend this potentially unsafe path, use the :option:`-P` command
+   line option or the :envvar:`PYTHONSAFEPATH` environment variable?
 
    A program is free to modify this list for its own purposes.  Only strings
    and bytes should be added to :data:`sys.path`; all other data types are

diff --git a/Doc/using/cmdline.rst b/Doc/using/cmdline.rst
@@ -257,6 +257,8 @@ Miscellaneous options
    Ignore all :envvar:`PYTHON*` environment variables, e.g.
    :envvar:`PYTHONPATH` and :envvar:`PYTHONHOME`, that might be set.
 
+   See also the :option:`-P` and :option:`-I` (isolated) options.
+
 
 .. cmdoption:: -i
 
@@ -271,7 +273,9 @@ Miscellaneous options
 
 .. cmdoption:: -I
 
-   Run Python in isolated mode. This also implies -E and -s.
+   Run Python in isolated mode. This also implies :option:`-E`, :option:`-P`
+   and :option:`-s` options.
+
    In isolated mode :data:`sys.path` contains neither the script's directory nor
    the user's site-packages directory. All :envvar:`PYTHON*` environment
    variables are ignored, too. Further restrictions may be imposed to prevent
@@ -301,6 +305,23 @@ Miscellaneous options
       Modify ``.pyc`` filenames according to :pep:`488`.
 
 
+.. cmdoption:: -P
+
+   Don't prepend a potentially unsafe path to :data:`sys.path`:
+
+   * ``python -m module`` command line: Don't prepend the current working
+     directory.
+   * ``python script.py`` command line: Don't prepend the script path.
-   * ``python script.py`` command line: Don't prepend the script path.
+   * ``python script.py`` command line: Don't prepend the script's directory.
-   * ``python script.py`` command line: Don't prepend the script path.
+   * ``python script.py`` command line: Don't prepend the script's directory.
+     If it's a symbolic link, resolve symbolic links.
+   * ``python -c code`` and ``python`` (REPL) command lines: Don't prepend an
+     empty string.
+
+   See also the :envvar:`PYTHONSAFEPATH` environment variable, and :option:`-E`
+   and :option:`-I` (isolated) options.
+
+   .. versionadded:: 3.11
+
+
 .. cmdoption:: -q
 
    Don't display the copyright and version messages even in interactive mode.
@@ -583,6 +604,14 @@ conflict.
    within a Python program as the variable :data:`sys.path`.
 
 
+.. envvar:: PYTHONSAFEPATH
+
+   If this is set to a non-empty string, don't prepend a potentially unsafe
+   path to :data:`sys.path`: see the :option:`-P` option for details.
+
+   .. versionadded:: 3.11
+
+
 .. envvar:: PYTHONPLATLIBDIR
 
    If this is set to a non-empty string, it overrides the :data:`sys.platlibdir`

@@ -362,6 +362,11 @@ Other Language Changes
   pickles instance attributes implemented as :term:`slots <__slots__>`.
   (Contributed by Serhiy Storchaka in :issue:`26579`.)
 
+* Add :option:`-P` command line option and :envvar:`PYTHONSAFEPATH` environment
+  variable to not prepend a potentially unsafe path to :data:`sys.path` such as
+  the current directory, script directory or empty string.
+  (Contributed by Victor Stinner in :gh:`57684`.)
+
 
 Other CPython Implementation Changes
 ====================================
@@ -636,6 +641,9 @@ sys
   (equivalent to ``sys.exc_info()[1]``).
   (Contributed by Irit Katriel in :issue:`46328`.)
 
+* Add the :data:`sys.flags.safe_path <sys.flags>` flag.
+  (Contributed by Victor Stinner in :gh:`57684`.)
+
 
 sysconfig
 ---------
@@ -1480,6 +1488,8 @@ New Features
   representation of exceptions.
   (Contributed by Irit Katriel in :issue:`46343`.)
 
+* Added the :c:member:`PyConfig.safe_path` member.
+  (Contributed by Victor Stinner in :gh:`57684`.)
 
 Porting to Python 3.11
 ----------------------

@@ -176,6 +176,7 @@ typedef struct PyConfig {
 #endif
     wchar_t *check_hash_pycs_mode;
     int use_frozen_modules;
+    int safe_path;
 
     /* --- Path configuration inputs ------------ */
     int pathconfig_warnings;

@@ -313,12 +313,14 @@ def _args_from_interpreter_flags():
             args.append('-E')
         if sys.flags.no_user_site:
             args.append('-s')
+        if sys.flags.safe_path:
+            args.append('-P')
 
     # -W options
     warnopts = sys.warnoptions[:]
-    bytes_warning = sys.flags.bytes_warning
     xoptions = getattr(sys, '_xoptions', {})
-    dev_mode = ('dev' in xoptions)
+    bytes_warning = sys.flags.bytes_warning
+    dev_mode = sys.flags.dev_mode
 
     if bytes_warning > 1:
         warnopts.remove("error::BytesWarning")

diff --git a/Lib/test/test_cmd_line.py b/Lib/test/test_cmd_line.py
@@ -579,13 +579,13 @@ def test_unknown_options(self):
                      'Cannot run -I tests when PYTHON env vars are required.')
     def test_isolatedmode(self):
         self.verify_valid_flag('-I')
-        self.verify_valid_flag('-IEs')
+        self.verify_valid_flag('-IEPs')
         rc, out, err = assert_python_ok('-I', '-c',
             'from sys import flags as f; '
-            'print(f.no_user_site, f.ignore_environment, f.isolated)',
+            'print(f.no_user_site, f.ignore_environment, f.isolated, f.safe_path)',
             # dummyvar to prevent extraneous -E
             dummyvar="")
-        self.assertEqual(out.strip(), b'1 1 1')
+        self.assertEqual(out.strip(), b'1 1 1 True')
         with os_helper.temp_cwd() as tmpdir:
             fake = os.path.join(tmpdir, "uuid.py")
             main = os.path.join(tmpdir, "main.py")
@@ -880,14 +880,16 @@ def test_sys_flags_not_set(self):
         # Issue 31845: a startup refactoring broke reading flags from env vars
         expected_outcome = """
             (sys.flags.debug == sys.flags.optimize ==
-             sys.flags.dont_write_bytecode == sys.flags.verbose == 0)
+             sys.flags.dont_write_bytecode ==
+             sys.flags.verbose == sys.flags.safe_path == 0)
         """
         self.run_ignoring_vars(
             expected_outcome,
             PYTHONDEBUG="1",
             PYTHONOPTIMIZE="1",
             PYTHONDONTWRITEBYTECODE="1",
             PYTHONVERBOSE="1",
+            PYTHONSAFEPATH="1",
         )
 
 class SyntaxErrorTests(unittest.TestCase):

diff --git a/Lib/test/test_embed.py b/Lib/test/test_embed.py
@@ -479,6 +479,7 @@ class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
         '_init_main': 1,
         '_isolated_interpreter': 0,
         'use_frozen_modules': not Py_DEBUG,
+        'safe_path': 0,
         '_is_python_build': IGNORE_CONFIG,
     }
     if MS_WINDOWS:
@@ -496,6 +497,7 @@ class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
         isolated=1,
         use_environment=0,
         user_site_directory=0,
+        safe_path=1,
         dev_mode=0,
         install_signal_handlers=0,
         use_hash_seed=0,
@@ -855,6 +857,7 @@ def test_init_from_config(self):
             'faulthandler': 1,
             'platlibdir': 'my_platlibdir',
             'module_search_paths': self.IGNORE_CONFIG,
+            'safe_path': 1,
 
             'check_hash_pycs_mode': 'always',
             'pathconfig_warnings': 0,
@@ -889,6 +892,7 @@ def test_init_compat_env(self):
             'warnoptions': ['EnvVar'],
             'platlibdir': 'env_platlibdir',
             'module_search_paths': self.IGNORE_CONFIG,
+            'safe_path': 1,
         }
         self.check_all_configs("test_init_compat_env", config, preconfig,
                                api=API_COMPAT)
@@ -919,6 +923,7 @@ def test_init_python_env(self):
             'warnoptions': ['EnvVar'],
             'platlibdir': 'env_platlibdir',
             'module_search_paths': self.IGNORE_CONFIG,
+            'safe_path': 1,
         }
         self.check_all_configs("test_init_python_env", config, preconfig,
                                api=API_PYTHON)
@@ -959,12 +964,13 @@ def test_preinit_parse_argv(self):
         }
         config = {
             'argv': ['script.py'],
-            'orig_argv': ['python3', '-X', 'dev', 'script.py'],
+            'orig_argv': ['python3', '-X', 'dev', '-P', 'script.py'],
             'run_filename': os.path.abspath('script.py'),
             'dev_mode': 1,
             'faulthandler': 1,
             'warnoptions': ['default'],
             'xoptions': ['dev'],
+            'safe_path': 1,
         }
         self.check_all_configs("test_preinit_parse_argv", config, preconfig,
                                api=API_PYTHON)
@@ -975,7 +981,7 @@ def test_preinit_dont_parse_argv(self):
             'isolated': 0,
         }
         argv = ["python3",
-               "-E", "-I",
+               "-E", "-I", "-P",
                "-X", "dev",
                "-X", "utf8",
                "script.py"]
@@ -990,6 +996,7 @@ def test_preinit_dont_parse_argv(self):
     def test_init_isolated_flag(self):
         config = {
             'isolated': 1,
+            'safe_path': 1,
             'use_environment': 0,
             'user_site_directory': 0,
         }
@@ -999,6 +1006,7 @@ def test_preinit_isolated1(self):
         # _PyPreConfig.isolated=1, _PyCoreConfig.isolated not set
         config = {
             'isolated': 1,
+            'safe_path': 1,
             'use_environment': 0,
             'user_site_directory': 0,
         }
@@ -1008,6 +1016,7 @@ def test_preinit_isolated2(self):
         # _PyPreConfig.isolated=0, _PyCoreConfig.isolated=1
         config = {
             'isolated': 1,
+            'safe_path': 1,
             'use_environment': 0,
             'user_site_directory': 0,
         }

diff --git a/Lib/test/test_support.py b/Lib/test/test_support.py
@@ -522,6 +522,7 @@ def test_args_from_interpreter_flags(self):
             ['-E'],
             ['-v'],
             ['-b'],
+            ['-P'],
             ['-q'],
             ['-I'],
             # same option multiple times
@@ -541,7 +542,8 @@ def test_args_from_interpreter_flags(self):
             with self.subTest(opts=opts):
                 self.check_options(opts, 'args_from_interpreter_flags')
 
-        self.check_options(['-I', '-E', '-s'], 'args_from_interpreter_flags',
+        self.check_options(['-I', '-E', '-s', '-P'],
+                           'args_from_interpreter_flags',
                            ['-I'])
 
     def test_optim_args_from_interpreter_flags(self):

diff --git a/Lib/test/test_sys.py b/Lib/test/test_sys.py
@@ -669,10 +669,10 @@ def test_sys_flags(self):
                  "dont_write_bytecode", "no_user_site", "no_site",
                  "ignore_environment", "verbose", "bytes_warning", "quiet",
                  "hash_randomization", "isolated", "dev_mode", "utf8_mode",
-                 "warn_default_encoding")
+                 "warn_default_encoding", "safe_path")
         for attr in attrs:
             self.assertTrue(hasattr(sys.flags, attr), attr)
-            attr_type = bool if attr == "dev_mode" else int
+            attr_type = bool if attr in ("dev_mode", "safe_path") else int
             self.assertEqual(type(getattr(sys.flags, attr)), attr_type, attr)
         self.assertTrue(repr(sys.flags))
         self.assertEqual(len(sys.flags), len(attrs))

diff --git a/Misc/NEWS.d/next/Security/2022-05-04-14-32-24.gh-issue-57684.HrlDrM.rst b/Misc/NEWS.d/next/Security/2022-05-04-14-32-24.gh-issue-57684.HrlDrM.rst
@@ -0,0 +1,3 @@
+Add the :option:`-P` command line option and the :envvar:`PYTHONSAFEPATH`
+environment variable to not prepend a potentially unsafe path to
+:data:`sys.path`. Patch by Victor Stinner.