Files
python315/CVE-2025-12781-b64decode-alt-chars.patch
Matěj Cepl 63187f7b8b Fixing bsc#1257108 (CVE-2025-12781)
Add CVE-2025-12781-b64decode-alt-chars.patch fixing bsc#1257108
  (CVE-2025-12781) combining gh#python/cpython!141061,
  gh#python/cpython!141128, and gh#python/cpython!141153. All
  `*b64decode` functions should not accept non-altchars.
2026-02-03 16:42:01 +01:00

218 lines
9.9 KiB
Diff

From f922c02c529d25d61aa9c28a8192639c1fce8d4d Mon Sep 17 00:00:00 2001
From: Serhiy Storchaka <storchaka@gmail.com>
Date: Wed, 5 Nov 2025 20:12:31 +0200
Subject: [PATCH] gh-125346: Add more base64 tests
Add more tests for the altchars argument of b64decode() and for the map01
argument of b32decode().
---
Doc/library/base64.rst | 18 ++--
Doc/whatsnew/3.15.rst | 9 ++
Lib/base64.py | 40 +++++++-
Lib/test/test_base64.py | 45 ++++++++--
Misc/NEWS.d/next/Library/2025-11-06-12-03-29.gh-issue-125346.7Gfpgw.rst | 5 +
5 files changed, 100 insertions(+), 17 deletions(-)
Index: Python-3.15.0a3/Doc/library/base64.rst
===================================================================
--- Python-3.15.0a3.orig/Doc/library/base64.rst 2025-12-16 13:26:12.000000000 +0100
+++ Python-3.15.0a3/Doc/library/base64.rst 2026-02-03 16:15:09.480009542 +0100
@@ -77,15 +77,20 @@
A :exc:`binascii.Error` exception is raised
if *s* is incorrectly padded.
- If *validate* is ``False`` (the default), characters that are neither
+ If *validate* is false (the default), characters that are neither
in the normal base-64 alphabet nor the alternative alphabet are
- discarded prior to the padding check. If *validate* is ``True``,
- these non-alphabet characters in the input result in a
- :exc:`binascii.Error`.
+ discarded prior to the padding check, but the ``+`` and ``/`` characters
+ keep their meaning if they are not in *altchars* (they will be discarded
+ in future Python versions).
+ If *validate* is true, these non-alphabet characters in the input
+ result in a :exc:`binascii.Error`.
For more information about the strict base64 check, see :func:`binascii.a2b_base64`
- May assert or raise a :exc:`ValueError` if the length of *altchars* is not 2.
+ .. deprecated:: next
+ Accepting the ``+`` and ``/`` characters with an alternative alphabet
+ is now deprecated.
+
.. function:: standard_b64encode(s)
@@ -116,6 +121,9 @@
``/`` in the standard Base64 alphabet, and return the decoded
:class:`bytes`.
+ .. deprecated:: next
+ Accepting the ``+`` and ``/`` characters is now deprecated.
+
.. function:: b32encode(s)
Index: Python-3.15.0a3/Doc/whatsnew/3.15.rst
===================================================================
--- Python-3.15.0a3.orig/Doc/whatsnew/3.15.rst 2025-12-16 13:26:12.000000000 +0100
+++ Python-3.15.0a3/Doc/whatsnew/3.15.rst 2026-02-03 16:15:09.480398395 +0100
@@ -992,6 +992,15 @@
New deprecations
----------------
+* :mod:`base64`:
+
+ * Accepting the ``+`` and ``/`` characters with an alternative alphabet in
+ :func:`~base64.b64decode` and :func:`~base64.urlsafe_b64decode` is now
+ deprecated.
+ In future Python versions they will be errors in the strict mode and
+ discarded in the non-strict mode.
+ (Contributed by Serhiy Storchaka in :gh:`125346`.)
+
* CLI:
* Deprecate :option:`-b` and :option:`!-bb` command-line options
Index: Python-3.15.0a3/Lib/base64.py
===================================================================
--- Python-3.15.0a3.orig/Lib/base64.py 2026-02-03 16:13:28.034477672 +0100
+++ Python-3.15.0a3/Lib/base64.py 2026-02-03 16:15:09.480836931 +0100
@@ -69,20 +69,39 @@
The result is returned as a bytes object. A binascii.Error is raised if
s is incorrectly padded.
- If validate is False (the default), characters that are neither in the
+ If validate is false (the default), characters that are neither in the
normal base-64 alphabet nor the alternative alphabet are discarded prior
- to the padding check. If validate is True, these non-alphabet characters
+ to the padding check. If validate is true, these non-alphabet characters
in the input result in a binascii.Error.
For more information about the strict base64 check, see:
https://docs.python.org/3.11/library/binascii.html#binascii.a2b_base64
"""
s = _bytes_from_decode_data(s)
+ badchar = None
if altchars is not None:
altchars = _bytes_from_decode_data(altchars)
- assert len(altchars) == 2, repr(altchars)
+ if len(altchars) != 2:
+ raise ValueError(f'invalid altchars: {altchars!r}')
+ for b in b'+/':
+ if b not in altchars and b in s:
+ badchar = b
+ break
s = s.translate(bytes.maketrans(altchars, b'+/'))
- return binascii.a2b_base64(s, strict_mode=validate)
+ result = binascii.a2b_base64(s, strict_mode=validate)
+ if badchar is not None:
+ import warnings
+ if validate:
+ warnings.warn(f'invalid character {chr(badchar)!a} in Base64 data '
+ f'with altchars={altchars!r} and validate=True '
+ f'will be an error in future Python versions',
+ DeprecationWarning, stacklevel=2)
+ else:
+ warnings.warn(f'invalid character {chr(badchar)!a} in Base64 data '
+ f'with altchars={altchars!r} and validate=False '
+ f'will be discarded in future Python versions',
+ FutureWarning, stacklevel=2)
+ return result
def standard_b64encode(s):
@@ -127,8 +146,19 @@
The alphabet uses '-' instead of '+' and '_' instead of '/'.
"""
s = _bytes_from_decode_data(s)
+ badchar = None
+ for b in b'+/':
+ if b in s:
+ badchar = b
+ break
s = s.translate(_urlsafe_decode_translation)
- return b64decode(s)
+ result = binascii.a2b_base64(s, strict_mode=False)
+ if badchar is not None:
+ import warnings
+ warnings.warn(f'invalid character {chr(badchar)!a} in URL-safe Base64 data '
+ f'will be discarded in future Python versions',
+ FutureWarning, stacklevel=2)
+ return result
Index: Python-3.15.0a3/Lib/test/test_base64.py
===================================================================
--- Python-3.15.0a3.orig/Lib/test/test_base64.py 2026-02-03 16:13:29.357990779 +0100
+++ Python-3.15.0a3/Lib/test/test_base64.py 2026-02-03 16:15:09.481134341 +0100
@@ -265,6 +265,25 @@
eq(base64.b64decode(data, altchars=altchars_str), res)
eq(base64.b64decode(data_str, altchars=altchars_str), res)
+ def test_b64decode_altchars(self):
+ # Test with arbitrary alternative characters
+ eq = self.assertEqual
+ res = b'\xd3V\xbeo\xf7\x1d'
+ for altchars in b'*$', b'+/', b'/+', b'+_', b'-+', b'-/', b'/_':
+ data = b'01a%cb%ccd' % tuple(altchars)
+ data_str = data.decode('ascii')
+ altchars_str = altchars.decode('ascii')
+
+ eq(base64.b64decode(data, altchars=altchars), res)
+ eq(base64.b64decode(data_str, altchars=altchars), res)
+ eq(base64.b64decode(data, altchars=altchars_str), res)
+ eq(base64.b64decode(data_str, altchars=altchars_str), res)
+
+ self.assertRaises(ValueError, base64.b64decode, b'', altchars=b'+')
+ self.assertRaises(ValueError, base64.b64decode, b'', altchars=b'+/-')
+ self.assertRaises(ValueError, base64.b64decode, '', altchars='+')
+ self.assertRaises(ValueError, base64.b64decode, '', altchars='+/-')
+
def test_b64decode_padding_error(self):
self.assertRaises(binascii.Error, base64.b64decode, b'abc')
self.assertRaises(binascii.Error, base64.b64decode, 'abc')
@@ -296,13 +315,25 @@
with self.assertRaises(binascii.Error):
base64.b64decode(bstr.decode('ascii'), validate=True)
- # Normal alphabet characters not discarded when alternative given
- res = b'\xfb\xef\xff'
- self.assertEqual(base64.b64decode(b'++//', validate=True), res)
- self.assertEqual(base64.b64decode(b'++//', '-_', validate=True), res)
- self.assertEqual(base64.b64decode(b'--__', '-_', validate=True), res)
- self.assertEqual(base64.urlsafe_b64decode(b'++//'), res)
- self.assertEqual(base64.urlsafe_b64decode(b'--__'), res)
+ # Normal alphabet characters will be discarded when alternative given
+ with self.assertWarns(FutureWarning):
+ self.assertEqual(base64.b64decode(b'++++', altchars=b'-_'),
+ b'\xfb\xef\xbe')
+ with self.assertWarns(FutureWarning):
+ self.assertEqual(base64.b64decode(b'////', altchars=b'-_'),
+ b'\xff\xff\xff')
+ with self.assertWarns(DeprecationWarning):
+ self.assertEqual(base64.b64decode(b'++++', altchars=b'-_', validate=True),
+ b'\xfb\xef\xbe')
+ with self.assertWarns(DeprecationWarning):
+ self.assertEqual(base64.b64decode(b'////', altchars=b'-_', validate=True),
+ b'\xff\xff\xff')
+ with self.assertWarns(FutureWarning):
+ self.assertEqual(base64.urlsafe_b64decode(b'++++'), b'\xfb\xef\xbe')
+ with self.assertWarns(FutureWarning):
+ self.assertEqual(base64.urlsafe_b64decode(b'////'), b'\xff\xff\xff')
+ with self.assertRaises(binascii.Error):
+ base64.b64decode(b'+/!', altchars=b'-_')
def _altchars_strategy():
"""Generate 'altchars' for base64 encoding."""
Index: Python-3.15.0a3/Misc/NEWS.d/next/Library/2025-11-06-12-03-29.gh-issue-125346.7Gfpgw.rst
===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ Python-3.15.0a3/Misc/NEWS.d/next/Library/2025-11-06-12-03-29.gh-issue-125346.7Gfpgw.rst 2026-02-03 16:15:09.481407075 +0100
@@ -0,0 +1,5 @@
+Accepting ``+`` and ``/`` characters with an alternative alphabet in
+:func:`base64.b64decode` and :func:`base64.urlsafe_b64decode` is now
+deprecated.
+In future Python versions they will be errors in the strict mode and
+discarded in the non-strict mode.