#1466065: add validate option to base64.b64decode

Patch by Neil Tallim.  This provides a mechanism for module
users to achieve RFC 3548 compliance in the cases where ignoring
non-base64-alphabet input characters is *not* mandated by the RFC that
references RFC 3548.

Doc/library/base64.rst

@@ -37,7 +37,7 @@ The modern interface provides:
    The encoded byte string is returned.
 
 
-.. function:: b64decode(s, altchars=None)
+.. function:: b64decode(s, altchars=None, validate=False)
 
    Decode a Base64 encoded byte string.
 
@@ -45,9 +45,13 @@ The modern interface provides:
    at least length 2 (additional characters are ignored) which specifies the
    alternative alphabet used instead of the ``+`` and ``/`` characters.
 
-   The decoded byte string is returned.  A :exc:`TypeError` is raised if *s* were
-   incorrectly padded or if there are non-alphabet characters present in the
-   string.
+   The decoded string is returned.  A `binascii.Error` is raised if *s* is
+   incorrectly padded.
+
+   If *validate* is ``False`` (the default), non-base64-alphabet characters are
+   discarded prior to the padding check.  If *validate* is ``True``,
+   non-base64-alphabet characters in the input result in a
+   :exc:`binascii.Error`.
 
 
 .. function:: standard_b64encode(s)

Lib/base64.py

@@ -65,16 +65,19 @@ def b64encode(s, altchars=None):
     return encoded
 
 
-def b64decode(s, altchars=None):
+def b64decode(s, altchars=None, validate=False):
     """Decode a Base64 encoded byte string.
 
     s is the byte string to decode.  Optional altchars must be a
     string of length 2 which specifies the alternative alphabet used
     instead of the '+' and '/' characters.
 
-    The decoded byte string is returned.  binascii.Error is raised if
-    s were incorrectly padded or if there are non-alphabet characters
-    present in the string.
+    The decoded string is returned.  A binascii.Error is raised if s is
+    incorrectly padded.
+
+    If validate is False (the default), non-base64-alphabet characters are
+    discarded prior to the padding check.  If validate is True,
+    non-base64-alphabet characters in the input result in a binascii.Error.
     """
     if not isinstance(s, bytes_types):
         raise TypeError("expected bytes, not %s" % s.__class__.__name__)
@@ -84,6 +87,8 @@ def b64decode(s, altchars=None):
                             % altchars.__class__.__name__)
         assert len(altchars) == 2, repr(altchars)
         s = _translate(s, {chr(altchars[0]): b'+', chr(altchars[1]): b'/'})
+    if validate and not re.match(b'^[A-Za-z0-9+/]*={0,2}$', s):
+        raise binascii.Error('Non-base64 digit found')
     return binascii.a2b_base64(s)
 
 

Lib/test/test_base64.py

@@ -138,9 +138,25 @@ class BaseXYTestCase(unittest.TestCase):
         eq(base64.urlsafe_b64decode(b'01a-b_cd'), b'\xd3V\xbeo\xf7\x1d')
         self.assertRaises(TypeError, base64.urlsafe_b64decode, "")
 
-    def test_b64decode_error(self):
+    def test_b64decode_padding_error(self):
         self.assertRaises(binascii.Error, base64.b64decode, b'abc')
 
+    def test_b64decode_invalid_chars(self):
+        # issue 1466065: Test some invalid characters.
+        tests = ((b'%3d==', b'\xdd'),
+                 (b'$3d==', b'\xdd'),
+                 (b'[==', b''),
+                 (b'YW]3=', b'am'),
+                 (b'3{d==', b'\xdd'),
+                 (b'3d}==', b'\xdd'),
+                 (b'@@', b''),
+                 (b'!', b''),
+                 (b'YWJj\nYWI=', b'abcab'))
+        for bstr, res in tests:
+            self.assertEquals(base64.b64decode(bstr), res)
+            with self.assertRaises(binascii.Error):
+                base64.b64decode(bstr, validate=True)
+
     def test_b32encode(self):
         eq = self.assertEqual
         eq(base64.b32encode(b''), b'')

Misc/NEWS

@@ -63,6 +63,9 @@ Core and Builtins
 Library
 -------
 
+- Issue #1466065: Add 'validate' option to base64.b64decode to raise
+  an error if there are non-base64 alphabet characters in the input.
+
 - Issue #10386: Add __all__ to token module; this simplifies importing
   in tokenize module and prevents leaking of private names through
   import *.