| 1 | # $Id$
|
| 2 | #
|
| 3 | # Copyright (C) 2005 Gregory P. Smith (greg@krypto.org)
|
| 4 | # Licensed to PSF under a Contributor Agreement.
|
| 5 | #
|
| 6 |
|
| 7 | __doc__ = """hashlib module - A common interface to many hash functions.
|
| 8 |
|
| 9 | new(name, string='') - returns a new hash object implementing the
|
| 10 | given hash function; initializing the hash
|
| 11 | using the given string data.
|
| 12 |
|
| 13 | Named constructor functions are also available, these are much faster
|
| 14 | than using new():
|
| 15 |
|
| 16 | md5(), sha1(), sha224(), sha256(), sha384(), and sha512()
|
| 17 |
|
| 18 | More algorithms may be available on your platform but the above are guaranteed
|
| 19 | to exist. See the algorithms_guaranteed and algorithms_available attributes
|
| 20 | to find out what algorithm names can be passed to new().
|
| 21 |
|
| 22 | NOTE: If you want the adler32 or crc32 hash functions they are available in
|
| 23 | the zlib module.
|
| 24 |
|
| 25 | Choose your hash function wisely. Some have known collision weaknesses.
|
| 26 | sha384 and sha512 will be slow on 32 bit platforms.
|
| 27 |
|
| 28 | Hash objects have these methods:
|
| 29 | - update(arg): Update the hash object with the string arg. Repeated calls
|
| 30 | are equivalent to a single call with the concatenation of all
|
| 31 | the arguments.
|
| 32 | - digest(): Return the digest of the strings passed to the update() method
|
| 33 | so far. This may contain non-ASCII characters, including
|
| 34 | NUL bytes.
|
| 35 | - hexdigest(): Like digest() except the digest is returned as a string of
|
| 36 | double length, containing only hexadecimal digits.
|
| 37 | - copy(): Return a copy (clone) of the hash object. This can be used to
|
| 38 | efficiently compute the digests of strings that share a common
|
| 39 | initial substring.
|
| 40 |
|
| 41 | For example, to obtain the digest of the string 'Nobody inspects the
|
| 42 | spammish repetition':
|
| 43 |
|
| 44 | >>> import hashlib
|
| 45 | >>> m = hashlib.md5()
|
| 46 | >>> m.update("Nobody inspects")
|
| 47 | >>> m.update(" the spammish repetition")
|
| 48 | >>> m.digest()
|
| 49 | '\\xbbd\\x9c\\x83\\xdd\\x1e\\xa5\\xc9\\xd9\\xde\\xc9\\xa1\\x8d\\xf0\\xff\\xe9'
|
| 50 |
|
| 51 | More condensed:
|
| 52 |
|
| 53 | >>> hashlib.sha224("Nobody inspects the spammish repetition").hexdigest()
|
| 54 | 'a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2'
|
| 55 |
|
| 56 | """
|
| 57 |
|
| 58 | # This tuple and __get_builtin_constructor() must be modified if a new
|
| 59 | # always available algorithm is added.
|
| 60 | __always_supported = ('md5', 'sha1', 'sha224', 'sha256', 'sha384', 'sha512')
|
| 61 |
|
| 62 | algorithms_guaranteed = set(__always_supported)
|
| 63 | algorithms_available = set(__always_supported)
|
| 64 |
|
| 65 | algorithms = __always_supported
|
| 66 |
|
| 67 | __all__ = __always_supported + ('new', 'algorithms_guaranteed',
|
| 68 | 'algorithms_available', 'algorithms',
|
| 69 | 'pbkdf2_hmac')
|
| 70 |
|
| 71 |
|
| 72 | def __get_builtin_constructor(name):
|
| 73 | try:
|
| 74 | if name in ('SHA1', 'sha1'):
|
| 75 | import _sha
|
| 76 | return _sha.new
|
| 77 | elif name in ('MD5', 'md5'):
|
| 78 | import _md5
|
| 79 | return _md5.new
|
| 80 | elif name in ('SHA256', 'sha256', 'SHA224', 'sha224'):
|
| 81 | import _sha256
|
| 82 | bs = name[3:]
|
| 83 | if bs == '256':
|
| 84 | return _sha256.sha256
|
| 85 | elif bs == '224':
|
| 86 | return _sha256.sha224
|
| 87 | elif name in ('SHA512', 'sha512', 'SHA384', 'sha384'):
|
| 88 | import _sha512
|
| 89 | bs = name[3:]
|
| 90 | if bs == '512':
|
| 91 | return _sha512.sha512
|
| 92 | elif bs == '384':
|
| 93 | return _sha512.sha384
|
| 94 | except ImportError:
|
| 95 | pass # no extension module, this hash is unsupported.
|
| 96 |
|
| 97 | raise ValueError('unsupported hash type ' + name)
|
| 98 |
|
| 99 |
|
| 100 | def __get_openssl_constructor(name):
|
| 101 | try:
|
| 102 | f = getattr(_hashlib, 'openssl_' + name)
|
| 103 | # Allow the C module to raise ValueError. The function will be
|
| 104 | # defined but the hash not actually available thanks to OpenSSL.
|
| 105 | f()
|
| 106 | # Use the C function directly (very fast)
|
| 107 | return f
|
| 108 | except (AttributeError, ValueError):
|
| 109 | return __get_builtin_constructor(name)
|
| 110 |
|
| 111 |
|
| 112 | def __py_new(name, string=''):
|
| 113 | """new(name, string='') - Return a new hashing object using the named algorithm;
|
| 114 | optionally initialized with a string.
|
| 115 | """
|
| 116 | return __get_builtin_constructor(name)(string)
|
| 117 |
|
| 118 |
|
| 119 | def __hash_new(name, string=''):
|
| 120 | """new(name, string='') - Return a new hashing object using the named algorithm;
|
| 121 | optionally initialized with a string.
|
| 122 | """
|
| 123 | try:
|
| 124 | return _hashlib.new(name, string)
|
| 125 | except ValueError:
|
| 126 | # If the _hashlib module (OpenSSL) doesn't support the named
|
| 127 | # hash, try using our builtin implementations.
|
| 128 | # This allows for SHA224/256 and SHA384/512 support even though
|
| 129 | # the OpenSSL library prior to 0.9.8 doesn't provide them.
|
| 130 | return __get_builtin_constructor(name)(string)
|
| 131 |
|
| 132 |
|
| 133 | try:
|
| 134 | import _hashlib
|
| 135 | new = __hash_new
|
| 136 | __get_hash = __get_openssl_constructor
|
| 137 | algorithms_available = algorithms_available.union(
|
| 138 | _hashlib.openssl_md_meth_names)
|
| 139 | except ImportError:
|
| 140 | new = __py_new
|
| 141 | __get_hash = __get_builtin_constructor
|
| 142 |
|
| 143 | for __func_name in __always_supported:
|
| 144 | # try them all, some may not work due to the OpenSSL
|
| 145 | # version not supporting that algorithm.
|
| 146 | try:
|
| 147 | globals()[__func_name] = __get_hash(__func_name)
|
| 148 | except ValueError:
|
| 149 | import logging
|
| 150 | logging.exception('code for hash %s was not found.', __func_name)
|
| 151 |
|
| 152 |
|
| 153 | try:
|
| 154 | # OpenSSL's PKCS5_PBKDF2_HMAC requires OpenSSL 1.0+ with HMAC and SHA
|
| 155 | from _hashlib import pbkdf2_hmac
|
| 156 | except ImportError:
|
| 157 | import binascii
|
| 158 | import struct
|
| 159 |
|
| 160 | _trans_5C = b"".join(chr(x ^ 0x5C) for x in range(256))
|
| 161 | _trans_36 = b"".join(chr(x ^ 0x36) for x in range(256))
|
| 162 |
|
| 163 | def pbkdf2_hmac(hash_name, password, salt, iterations, dklen=None):
|
| 164 | """Password based key derivation function 2 (PKCS #5 v2.0)
|
| 165 |
|
| 166 | This Python implementations based on the hmac module about as fast
|
| 167 | as OpenSSL's PKCS5_PBKDF2_HMAC for short passwords and much faster
|
| 168 | for long passwords.
|
| 169 | """
|
| 170 | if not isinstance(hash_name, str):
|
| 171 | raise TypeError(hash_name)
|
| 172 |
|
| 173 | if not isinstance(password, (bytes, bytearray)):
|
| 174 | password = bytes(buffer(password))
|
| 175 | if not isinstance(salt, (bytes, bytearray)):
|
| 176 | salt = bytes(buffer(salt))
|
| 177 |
|
| 178 | # Fast inline HMAC implementation
|
| 179 | inner = new(hash_name)
|
| 180 | outer = new(hash_name)
|
| 181 | blocksize = getattr(inner, 'block_size', 64)
|
| 182 | if len(password) > blocksize:
|
| 183 | password = new(hash_name, password).digest()
|
| 184 | password = password + b'\x00' * (blocksize - len(password))
|
| 185 | inner.update(password.translate(_trans_36))
|
| 186 | outer.update(password.translate(_trans_5C))
|
| 187 |
|
| 188 | def prf(msg, inner=inner, outer=outer):
|
| 189 | # PBKDF2_HMAC uses the password as key. We can re-use the same
|
| 190 | # digest objects and just update copies to skip initialization.
|
| 191 | icpy = inner.copy()
|
| 192 | ocpy = outer.copy()
|
| 193 | icpy.update(msg)
|
| 194 | ocpy.update(icpy.digest())
|
| 195 | return ocpy.digest()
|
| 196 |
|
| 197 | if iterations < 1:
|
| 198 | raise ValueError(iterations)
|
| 199 | if dklen is None:
|
| 200 | dklen = outer.digest_size
|
| 201 | if dklen < 1:
|
| 202 | raise ValueError(dklen)
|
| 203 |
|
| 204 | hex_format_string = "%%0%ix" % (new(hash_name).digest_size * 2)
|
| 205 |
|
| 206 | dkey = b''
|
| 207 | loop = 1
|
| 208 | while len(dkey) < dklen:
|
| 209 | prev = prf(salt + struct.pack(b'>I', loop))
|
| 210 | rkey = int(binascii.hexlify(prev), 16)
|
| 211 | for i in xrange(iterations - 1):
|
| 212 | prev = prf(prev)
|
| 213 | rkey ^= int(binascii.hexlify(prev), 16)
|
| 214 | loop += 1
|
| 215 | dkey += binascii.unhexlify(hex_format_string % rkey)
|
| 216 |
|
| 217 | return dkey[:dklen]
|
| 218 |
|
| 219 | # Cleanup locals()
|
| 220 | del __always_supported, __func_name, __get_hash
|
| 221 | del __py_new, __hash_new, __get_openssl_constructor
|