bpo-31116: Add Z85 variant to base64 (GH-30598)

Z85  specification: https://rfc.zeromq.org/spec/32/

2 files changed, 114 insertions, 2 deletions

diff --git a/Lib/base64.py b/Lib/base64.py
index e3e983b..25164d1 100755
--- a/Lib/base64.py
+++ b/Lib/base64.py
@@ -18,7 +18,7 @@ __all__ = [
     'b64encode', 'b64decode', 'b32encode', 'b32decode',
     'b32hexencode', 'b32hexdecode', 'b16encode', 'b16decode',
     # Base85 and Ascii85 encodings
-    'b85encode', 'b85decode', 'a85encode', 'a85decode',
+    'b85encode', 'b85decode', 'a85encode', 'a85decode', 'z85encode', 'z85decode',
     # Standard Base64 encoding
     'standard_b64encode', 'standard_b64decode',
     # Some common Base64 alternatives.  As referenced by RFC 3458, see thread
@@ -497,6 +497,33 @@ def b85decode(b):
         result = result[:-padding]
     return result
 
+_z85alphabet = (b'0123456789abcdefghijklmnopqrstuvwxyz'
+                b'ABCDEFGHIJKLMNOPQRSTUVWXYZ.-:+=^!/*?&<>()[]{}@%$#')
+# Translating b85 valid but z85 invalid chars to b'\x00' is required
+# to prevent them from being decoded as b85 valid chars.
+_z85_b85_decode_diff = b';_`|~'
+_z85_decode_translation = bytes.maketrans(
+    _z85alphabet + _z85_b85_decode_diff,
+    _b85alphabet + b'\x00' * len(_z85_b85_decode_diff)
+)
+_z85_encode_translation = bytes.maketrans(_b85alphabet, _z85alphabet)
+
+def z85encode(s):
+    """Encode bytes-like object b in z85 format and return a bytes object."""
+    return b85encode(s).translate(_z85_encode_translation)
+
+def z85decode(s):
+    """Decode the z85-encoded bytes-like object or ASCII string b
+
+    The result is returned as a bytes object.
+    """
+    s = _bytes_from_decode_data(s)
+    s = s.translate(_z85_decode_translation)
+    try:
+        return b85decode(s)
+    except ValueError as e:
+        raise ValueError(e.args[0].replace('base85', 'z85')) from None
+
 # Legacy interface.  This code could be cleaned up since I don't believe
 # binascii has any line length limitations.  It just doesn't seem worth it
 # though.  The files should be opened in binary mode.
diff --git a/Lib/test/test_base64.py b/Lib/test/test_base64.py
index f6171d3..409c8c1 100644
--- a/Lib/test/test_base64.py
+++ b/Lib/test/test_base64.py
@@ -545,6 +545,40 @@ class BaseXYTestCase(unittest.TestCase):
         self.check_other_types(base64.b85encode, b"www.python.org",
                                b'cXxL#aCvlSZ*DGca%T')
 
+    def test_z85encode(self):
+        eq = self.assertEqual
+
+        tests = {
+            b'': b'',
+            b'www.python.org': b'CxXl-AcVLsz/dgCA+t',
+            bytes(range(255)): b"""009c61o!#m2NH?C3>iWS5d]J*6CRx17-skh9337x"""
+                b"""ar.{NbQB=+c[cR@eg&FcfFLssg=mfIi5%2YjuU>)kTv.7l}6Nnnj=AD"""
+                b"""oIFnTp/ga?r8($2sxO*itWpVyu$0IOwmYv=xLzi%y&a6dAb/]tBAI+J"""
+                b"""CZjQZE0{D[FpSr8GOteoH(41EJe-<UKDCY&L:dM3N3<zjOsMmzPRn9P"""
+                b"""Q[%@^ShV!$TGwUeU^7HuW6^uKXvGh.YUh4]Z})[9-kP:p:JqPF+*1CV"""
+                b"""^9Zp<!yAd4/Xb0k*$*&A&nJXQ<MkK!>&}x#)cTlf[Bu8v].4}L}1:^-"""
+                b"""@qDP""",
+            b"""abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"""
+                b"""0123456789!@#0^&*();:<>,. []{}""":
+                b"""vpA.SwObN*x>?B1zeKohADlbxB-}$ND3R+ylQTvjm[uizoh55PpF:[^"""
+                b"""q=D:$s6eQefFLssg=mfIi5@cEbqrBJdKV-ciY]OSe*aw7DWL""",
+            b'no padding..': b'zF{UpvpS[.zF7NO',
+            b'zero compression\x00\x00\x00\x00': b'Ds.bnay/tbAb]JhB7]Mg00000',
+            b'zero compression\x00\x00\x00': b'Ds.bnay/tbAb]JhB7]Mg0000',
+            b"""Boundary:\x00\x00\x00\x00""": b"""lt}0:wmoI7iSGcW00""",
+            b'Space compr:    ': b'q/DePwGUG3ze:IRarR^H',
+            b'\xff': b'@@',
+            b'\xff'*2: b'%nJ',
+            b'\xff'*3: b'%nS9',
+            b'\xff'*4: b'%nSc0',
+        }
+
+        for data, res in tests.items():
+            eq(base64.z85encode(data), res)
+
+        self.check_other_types(base64.z85encode, b"www.python.org",
+                               b'CxXl-AcVLsz/dgCA+t')
+
     def test_a85decode(self):
         eq = self.assertEqual
 
@@ -626,6 +660,41 @@ class BaseXYTestCase(unittest.TestCase):
         self.check_other_types(base64.b85decode, b'cXxL#aCvlSZ*DGca%T',
                                b"www.python.org")
 
+    def test_z85decode(self):
+        eq = self.assertEqual
+
+        tests = {
+            b'': b'',
+            b'CxXl-AcVLsz/dgCA+t': b'www.python.org',
+            b"""009c61o!#m2NH?C3>iWS5d]J*6CRx17-skh9337x"""
+                b"""ar.{NbQB=+c[cR@eg&FcfFLssg=mfIi5%2YjuU>)kTv.7l}6Nnnj=AD"""
+                b"""oIFnTp/ga?r8($2sxO*itWpVyu$0IOwmYv=xLzi%y&a6dAb/]tBAI+J"""
+                b"""CZjQZE0{D[FpSr8GOteoH(41EJe-<UKDCY&L:dM3N3<zjOsMmzPRn9P"""
+                b"""Q[%@^ShV!$TGwUeU^7HuW6^uKXvGh.YUh4]Z})[9-kP:p:JqPF+*1CV"""
+                b"""^9Zp<!yAd4/Xb0k*$*&A&nJXQ<MkK!>&}x#)cTlf[Bu8v].4}L}1:^-"""
+                b"""@qDP""": bytes(range(255)),
+            b"""vpA.SwObN*x>?B1zeKohADlbxB-}$ND3R+ylQTvjm[uizoh55PpF:[^"""
+                b"""q=D:$s6eQefFLssg=mfIi5@cEbqrBJdKV-ciY]OSe*aw7DWL""":
+                b"""abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ"""
+                b"""0123456789!@#0^&*();:<>,. []{}""",
+            b'zF{UpvpS[.zF7NO': b'no padding..',
+            b'Ds.bnay/tbAb]JhB7]Mg00000': b'zero compression\x00\x00\x00\x00',
+            b'Ds.bnay/tbAb]JhB7]Mg0000': b'zero compression\x00\x00\x00',
+            b"""lt}0:wmoI7iSGcW00""": b"""Boundary:\x00\x00\x00\x00""",
+            b'q/DePwGUG3ze:IRarR^H': b'Space compr:    ',
+            b'@@': b'\xff',
+            b'%nJ': b'\xff'*2,
+            b'%nS9': b'\xff'*3,
+            b'%nSc0': b'\xff'*4,
+        }
+
+        for data, res in tests.items():
+            eq(base64.z85decode(data), res)
+            eq(base64.z85decode(data.decode("ascii")), res)
+
+        self.check_other_types(base64.z85decode, b'CxXl-AcVLsz/dgCA+t',
+                               b'www.python.org')
+
     def test_a85_padding(self):
         eq = self.assertEqual
 
@@ -707,6 +776,21 @@ class BaseXYTestCase(unittest.TestCase):
         self.assertRaises(ValueError, base64.b85decode, b'|NsC')
         self.assertRaises(ValueError, base64.b85decode, b'|NsC1')
 
+    def test_z85decode_errors(self):
+        illegal = list(range(33)) + \
+                  list(b'"\',;_`|\\~') + \
+                  list(range(128, 256))
+        for c in illegal:
+            with self.assertRaises(ValueError, msg=bytes([c])):
+                base64.z85decode(b'0000' + bytes([c]))
+
+        # b'\xff\xff\xff\xff' encodes to b'%nSc0', the following will overflow:
+        self.assertRaises(ValueError, base64.z85decode, b'%')
+        self.assertRaises(ValueError, base64.z85decode, b'%n')
+        self.assertRaises(ValueError, base64.z85decode, b'%nS')
+        self.assertRaises(ValueError, base64.z85decode, b'%nSc')
+        self.assertRaises(ValueError, base64.z85decode, b'%nSc1')
+
     def test_decode_nonascii_str(self):
         decode_funcs = (base64.b64decode,
                         base64.standard_b64decode,
@@ -714,7 +798,8 @@ class BaseXYTestCase(unittest.TestCase):
                         base64.b32decode,
                         base64.b16decode,
                         base64.b85decode,
-                        base64.a85decode)
+                        base64.a85decode,
+                        base64.z85decode)
         for f in decode_funcs:
             self.assertRaises(ValueError, f, 'with non-ascii \xcb')