Code

Ticket #4682: uuid.py

File uuid.py, 19.7 KB (added by anonymous, 7 years ago)
Line 
1r"""UUID objects (universally unique identifiers) according to RFC 4122.
2
3This module provides immutable UUID objects (class UUID) and the functions
4uuid1(), uuid3(), uuid4(), uuid5() for generating version 1, 3, 4, and 5
5UUIDs as specified in RFC 4122.
6
7If all you want is a unique ID, you should probably call uuid1() or uuid4().
8Note that uuid1() may compromise privacy since it creates a UUID containing
9the computer's network address.  uuid4() creates a random UUID.
10
11Typical usage:
12
13    >>> import uuid
14
15    # make a UUID based on the host ID and current time
16    >>> uuid.uuid1()
17    UUID('a8098c1a-f86e-11da-bd1a-00112444be1e')
18
19    # make a UUID using an MD5 hash of a namespace UUID and a name
20    >>> uuid.uuid3(uuid.NAMESPACE_DNS, 'python.org')
21    UUID('6fa459ea-ee8a-3ca4-894e-db77e160355e')
22
23    # make a random UUID
24    >>> uuid.uuid4()
25    UUID('16fd2706-8baf-433b-82eb-8c7fada847da')
26
27    # make a UUID using a SHA-1 hash of a namespace UUID and a name
28    >>> uuid.uuid5(uuid.NAMESPACE_DNS, 'python.org')
29    UUID('886313e1-3b8a-5372-9b90-0c9aee199e5d')
30
31    # make a UUID from a string of hex digits (braces and hyphens ignored)
32    >>> x = uuid.UUID('{00010203-0405-0607-0809-0a0b0c0d0e0f}')
33
34    # convert a UUID to a string of hex digits in standard form
35    >>> str(x)
36    '00010203-0405-0607-0809-0a0b0c0d0e0f'
37
38    # get the raw 16 bytes of the UUID
39    >>> x.bytes
40    '\x00\x01\x02\x03\x04\x05\x06\x07\x08\t\n\x0b\x0c\r\x0e\x0f'
41
42    # make a UUID from a 16-byte string
43    >>> uuid.UUID(bytes=x.bytes)
44    UUID('00010203-0405-0607-0809-0a0b0c0d0e0f')
45"""
46
47__author__ = 'Ka-Ping Yee <ping@zesty.ca>'
48
49RESERVED_NCS, RFC_4122, RESERVED_MICROSOFT, RESERVED_FUTURE = [
50    'reserved for NCS compatibility', 'specified in RFC 4122',
51    'reserved for Microsoft compatibility', 'reserved for future definition']
52
53class UUID(object):
54    """Instances of the UUID class represent UUIDs as specified in RFC 4122.
55    UUID objects are immutable, hashable, and usable as dictionary keys.
56    Converting a UUID to a string with str() yields something in the form
57    '12345678-1234-1234-1234-123456789abc'.  The UUID constructor accepts
58    five possible forms: a similar string of hexadecimal digits, or a tuple
59    of six integer fields (with 32-bit, 16-bit, 16-bit, 8-bit, 8-bit, and
60    48-bit values respectively) as an argument named 'fields', or a string
61    of 16 bytes (with all the integer fields in big-endian order) as an
62    argument named 'bytes', or a string of 16 bytes (with the first three
63    fields in little-endian order) as an argument named 'bytes_le', or a
64    single 128-bit integer as an argument named 'int'.
65
66    UUIDs have these read-only attributes:
67
68        bytes       the UUID as a 16-byte string (containing the six
69                    integer fields in big-endian byte order)
70
71        bytes_le    the UUID as a 16-byte string (with time_low, time_mid,
72                    and time_hi_version in little-endian byte order)
73
74        fields      a tuple of the six integer fields of the UUID,
75                    which are also available as six individual attributes
76                    and two derived attributes:
77
78            time_low                the first 32 bits of the UUID
79            time_mid                the next 16 bits of the UUID
80            time_hi_version         the next 16 bits of the UUID
81            clock_seq_hi_variant    the next 8 bits of the UUID
82            clock_seq_low           the next 8 bits of the UUID
83            node                    the last 48 bits of the UUID
84
85            time                    the 60-bit timestamp
86            clock_seq               the 14-bit sequence number
87
88        hex         the UUID as a 32-character hexadecimal string
89
90        int         the UUID as a 128-bit integer
91
92        urn         the UUID as a URN as specified in RFC 4122
93
94        variant     the UUID variant (one of the constants RESERVED_NCS,
95                    RFC_4122, RESERVED_MICROSOFT, or RESERVED_FUTURE)
96
97        version     the UUID version number (1 through 5, meaningful only
98                    when the variant is RFC_4122)
99    """
100
101    def __init__(self, hex=None, bytes=None, bytes_le=None, fields=None,
102                       int=None, version=None):
103        r"""Create a UUID from either a string of 32 hexadecimal digits,
104        a string of 16 bytes as the 'bytes' argument, a string of 16 bytes
105        in little-endian order as the 'bytes_le' argument, a tuple of six
106        integers (32-bit time_low, 16-bit time_mid, 16-bit time_hi_version,
107        8-bit clock_seq_hi_variant, 8-bit clock_seq_low, 48-bit node) as
108        the 'fields' argument, or a single 128-bit integer as the 'int'
109        argument.  When a string of hex digits is given, curly braces,
110        hyphens, and a URN prefix are all optional.  For example, these
111        expressions all yield the same UUID:
112
113        UUID('{12345678-1234-5678-1234-567812345678}')
114        UUID('12345678123456781234567812345678')
115        UUID('urn:uuid:12345678-1234-5678-1234-567812345678')
116        UUID(bytes='\x12\x34\x56\x78'*4)
117        UUID(bytes_le='\x78\x56\x34\x12\x34\x12\x78\x56' +
118                      '\x12\x34\x56\x78\x12\x34\x56\x78')
119        UUID(fields=(0x12345678, 0x1234, 0x5678, 0x12, 0x34, 0x567812345678))
120        UUID(int=0x12345678123456781234567812345678)
121
122        Exactly one of 'hex', 'bytes', 'bytes_le', 'fields', or 'int' must
123        be given.  The 'version' argument is optional; if given, the resulting
124        UUID will have its variant and version set according to RFC 4122,
125        overriding the given 'hex', 'bytes', 'bytes_le', 'fields', or 'int'.
126        """
127
128        if [hex, bytes, bytes_le, fields, int].count(None) != 4:
129            raise TypeError('need one of hex, bytes, bytes_le, fields, or int')
130        if hex is not None:
131            hex = hex.replace('urn:', '').replace('uuid:', '')
132            hex = hex.strip('{}').replace('-', '')
133            if len(hex) != 32:
134                raise ValueError('badly formed hexadecimal UUID string')
135            int = long(hex, 16)
136        if bytes_le is not None:
137            if len(bytes_le) != 16:
138                raise ValueError('bytes_le is not a 16-char string')
139            bytes = (bytes_le[3] + bytes_le[2] + bytes_le[1] + bytes_le[0] +
140                     bytes_le[5] + bytes_le[4] + bytes_le[7] + bytes_le[6] +
141                     bytes_le[8:])
142        if bytes is not None:
143            if len(bytes) != 16:
144                raise ValueError('bytes is not a 16-char string')
145            int = long(('%02x'*16) % tuple(map(ord, bytes)), 16)
146        if fields is not None:
147            if len(fields) != 6:
148                raise ValueError('fields is not a 6-tuple')
149            (time_low, time_mid, time_hi_version,
150             clock_seq_hi_variant, clock_seq_low, node) = fields
151            if not 0 <= time_low < 1<<32L:
152                raise ValueError('field 1 out of range (need a 32-bit value)')
153            if not 0 <= time_mid < 1<<16L:
154                raise ValueError('field 2 out of range (need a 16-bit value)')
155            if not 0 <= time_hi_version < 1<<16L:
156                raise ValueError('field 3 out of range (need a 16-bit value)')
157            if not 0 <= clock_seq_hi_variant < 1<<8L:
158                raise ValueError('field 4 out of range (need an 8-bit value)')
159            if not 0 <= clock_seq_low < 1<<8L:
160                raise ValueError('field 5 out of range (need an 8-bit value)')
161            if not 0 <= node < 1<<48L:
162                raise ValueError('field 6 out of range (need a 48-bit value)')
163            clock_seq = (clock_seq_hi_variant << 8L) | clock_seq_low
164            int = ((time_low << 96L) | (time_mid << 80L) |
165                   (time_hi_version << 64L) | (clock_seq << 48L) | node)
166        if int is not None:
167            if not 0 <= int < 1<<128L:
168                raise ValueError('int is out of range (need a 128-bit value)')
169        if version is not None:
170            if not 1 <= version <= 5:
171                raise ValueError('illegal version number')
172            # Set the variant to RFC 4122.
173            int &= ~(0xc000 << 48L)
174            int |= 0x8000 << 48L
175            # Set the version number.
176            int &= ~(0xf000 << 64L)
177            int |= version << 76L
178        self.__dict__['int'] = int
179
180    def __cmp__(self, other):
181        if isinstance(other, UUID):
182            return cmp(self.int, other.int)
183        return NotImplemented
184
185    def __hash__(self):
186        return hash(self.int)
187
188    def __int__(self):
189        return self.int
190
191    def __repr__(self):
192        return 'UUID(%r)' % str(self)
193
194    def __setattr__(self, name, value):
195        raise TypeError('UUID objects are immutable')
196
197    def __str__(self):
198        hex = '%032x' % self.int
199        return '%s-%s-%s-%s-%s' % (
200            hex[:8], hex[8:12], hex[12:16], hex[16:20], hex[20:])
201
202    def get_bytes(self):
203        bytes = ''
204        for shift in range(0, 128, 8):
205            bytes = chr((self.int >> shift) & 0xff) + bytes
206        return bytes
207
208    bytes = property(get_bytes)
209
210    def get_bytes_le(self):
211        bytes = self.bytes
212        return (bytes[3] + bytes[2] + bytes[1] + bytes[0] +
213                bytes[5] + bytes[4] + bytes[7] + bytes[6] + bytes[8:])
214
215    bytes_le = property(get_bytes_le)
216
217    def get_fields(self):
218        return (self.time_low, self.time_mid, self.time_hi_version,
219                self.clock_seq_hi_variant, self.clock_seq_low, self.node)
220
221    fields = property(get_fields)
222
223    def get_time_low(self):
224        return self.int >> 96L
225
226    time_low = property(get_time_low)
227
228    def get_time_mid(self):
229        return (self.int >> 80L) & 0xffff
230
231    time_mid = property(get_time_mid)
232
233    def get_time_hi_version(self):
234        return (self.int >> 64L) & 0xffff
235
236    time_hi_version = property(get_time_hi_version)
237
238    def get_clock_seq_hi_variant(self):
239        return (self.int >> 56L) & 0xff
240
241    clock_seq_hi_variant = property(get_clock_seq_hi_variant)
242
243    def get_clock_seq_low(self):
244        return (self.int >> 48L) & 0xff
245
246    clock_seq_low = property(get_clock_seq_low)
247
248    def get_time(self):
249        return (((self.time_hi_version & 0x0fffL) << 48L) |
250                (self.time_mid << 32L) | self.time_low)
251
252    time = property(get_time)
253
254    def get_clock_seq(self):
255        return (((self.clock_seq_hi_variant & 0x3fL) << 8L) |
256                self.clock_seq_low)
257
258    clock_seq = property(get_clock_seq)
259
260    def get_node(self):
261        return self.int & 0xffffffffffff
262
263    node = property(get_node)
264
265    def get_hex(self):
266        return '%032x' % self.int
267
268    hex = property(get_hex)
269
270    def get_urn(self):
271        return 'urn:uuid:' + str(self)
272
273    urn = property(get_urn)
274
275    def get_variant(self):
276        if not self.int & (0x8000 << 48L):
277            return RESERVED_NCS
278        elif not self.int & (0x4000 << 48L):
279            return RFC_4122
280        elif not self.int & (0x2000 << 48L):
281            return RESERVED_MICROSOFT
282        else:
283            return RESERVED_FUTURE
284
285    variant = property(get_variant)
286
287    def get_version(self):
288        # The version bits are only meaningful for RFC 4122 UUIDs.
289        if self.variant == RFC_4122:
290            return int((self.int >> 76L) & 0xf)
291
292    version = property(get_version)
293
294def _find_mac(command, args, hw_identifiers, get_index):
295    import os
296    for dir in ['', '/sbin/', '/usr/sbin']:
297        executable = os.path.join(dir, command)
298        if not os.path.exists(executable):
299            continue
300
301        try:
302            # LC_ALL to get English output, 2>/dev/null to
303            # prevent output on stderr
304            cmd = 'LC_ALL=C %s %s 2>/dev/null' % (executable, args)
305            pipe = os.popen(cmd)
306        except IOError:
307            continue
308
309        for line in pipe:
310            words = line.lower().split()
311            for i in range(len(words)):
312                if words[i] in hw_identifiers:
313                    return int(words[get_index(i)].replace(':', ''), 16)
314    return None
315
316def _ifconfig_getnode():
317    """Get the hardware address on Unix by running ifconfig."""
318
319    # This works on Linux ('' or '-a'), Tru64 ('-av'), but not all Unixes.
320    for args in ('', '-a', '-av'):
321        mac = _find_mac('ifconfig', args, ['hwaddr', 'ether'], lambda i: i+1)
322        if mac:
323            return mac
324
325    import socket
326    ip_addr = socket.gethostbyname(socket.gethostname())
327
328    # Try getting the MAC addr from arp based on our IP address (Solaris).
329    mac = _find_mac('arp', '-an', [ip_addr], lambda i: -1)
330    if mac:
331        return mac
332
333    # This might work on HP-UX.
334    mac = _find_mac('lanscan', '-ai', ['lan0'], lambda i: 0)
335    if mac:
336        return mac
337
338    return None
339
340def _ipconfig_getnode():
341    """Get the hardware address on Windows by running ipconfig.exe."""
342    import os, re
343    dirs = ['', r'c:\windows\system32', r'c:\winnt\system32']
344    try:
345        import ctypes
346        buffer = ctypes.create_string_buffer(300)
347        ctypes.windll.kernel32.GetSystemDirectoryA(buffer, 300)
348        dirs.insert(0, buffer.value.decode('mbcs'))
349    except:
350        pass
351    for dir in dirs:
352        try:
353            pipe = os.popen(os.path.join(dir, 'ipconfig') + ' /all')
354        except IOError:
355            continue
356        for line in pipe:
357            value = line.split(':')[-1].strip().lower()
358            if re.match('([0-9a-f][0-9a-f]-){5}[0-9a-f][0-9a-f]', value):
359                return int(value.replace('-', ''), 16)
360
361def _netbios_getnode():
362    """Get the hardware address on Windows using NetBIOS calls.
363    See http://support.microsoft.com/kb/118623 for details."""
364    import win32wnet, netbios
365    ncb = netbios.NCB()
366    ncb.Command = netbios.NCBENUM
367    ncb.Buffer = adapters = netbios.LANA_ENUM()
368    adapters._pack()
369    if win32wnet.Netbios(ncb) != 0:
370        return
371    adapters._unpack()
372    for i in range(adapters.length):
373        ncb.Reset()
374        ncb.Command = netbios.NCBRESET
375        ncb.Lana_num = ord(adapters.lana[i])
376        if win32wnet.Netbios(ncb) != 0:
377            continue
378        ncb.Reset()
379        ncb.Command = netbios.NCBASTAT
380        ncb.Lana_num = ord(adapters.lana[i])
381        ncb.Callname = '*'.ljust(16)
382        ncb.Buffer = status = netbios.ADAPTER_STATUS()
383        if win32wnet.Netbios(ncb) != 0:
384            continue
385        status._unpack()
386        bytes = map(ord, status.adapter_address)
387        return ((bytes[0]<<40L) + (bytes[1]<<32L) + (bytes[2]<<24L) +
388                (bytes[3]<<16L) + (bytes[4]<<8L) + bytes[5])
389
390# Thanks to Thomas Heller for ctypes and for his help with its use here.
391
392# If ctypes is available, use it to find system routines for UUID generation.
393_uuid_generate_random = _uuid_generate_time = _UuidCreate = None
394try:
395    import ctypes, ctypes.util
396    _buffer = ctypes.create_string_buffer(16)
397
398    # The uuid_generate_* routines are provided by libuuid on at least
399    # Linux and FreeBSD, and provided by libc on Mac OS X.
400    for libname in ['uuid', 'c']:
401        try:
402            lib = ctypes.CDLL(ctypes.util.find_library(libname))
403        except:
404            continue
405        if hasattr(lib, 'uuid_generate_random'):
406            _uuid_generate_random = lib.uuid_generate_random
407        if hasattr(lib, 'uuid_generate_time'):
408            _uuid_generate_time = lib.uuid_generate_time
409
410    # On Windows prior to 2000, UuidCreate gives a UUID containing the
411    # hardware address.  On Windows 2000 and later, UuidCreate makes a
412    # random UUID and UuidCreateSequential gives a UUID containing the
413    # hardware address.  These routines are provided by the RPC runtime.
414    # NOTE:  at least on Tim's WinXP Pro SP2 desktop box, while the last
415    # 6 bytes returned by UuidCreateSequential are fixed, they don't appear
416    # to bear any relationship to the MAC address of any network device
417    # on the box.
418    try:
419        lib = ctypes.windll.rpcrt4
420    except:
421        lib = None
422    _UuidCreate = getattr(lib, 'UuidCreateSequential',
423                          getattr(lib, 'UuidCreate', None))
424except:
425    pass
426
427def _unixdll_getnode():
428    """Get the hardware address on Unix using ctypes."""
429    _uuid_generate_time(_buffer)
430    return UUID(bytes=_buffer.raw).node
431
432def _windll_getnode():
433    """Get the hardware address on Windows using ctypes."""
434    if _UuidCreate(_buffer) == 0:
435        return UUID(bytes=_buffer.raw).node
436
437def _random_getnode():
438    """Get a random node ID, with eighth bit set as suggested by RFC 4122."""
439    import random
440    return random.randrange(0, 1<<48L) | 0x010000000000L
441
442_node = None
443
444def getnode():
445    """Get the hardware address as a 48-bit positive integer.
446
447    The first time this runs, it may launch a separate program, which could
448    be quite slow.  If all attempts to obtain the hardware address fail, we
449    choose a random 48-bit number with its eighth bit set to 1 as recommended
450    in RFC 4122.
451    """
452
453    global _node
454    if _node is not None:
455        return _node
456
457    import sys
458    if sys.platform == 'win32':
459        getters = [_windll_getnode, _netbios_getnode, _ipconfig_getnode]
460    else:
461        getters = [_unixdll_getnode, _ifconfig_getnode]
462
463    for getter in getters + [_random_getnode]:
464        try:
465            _node = getter()
466        except:
467            continue
468        if _node is not None:
469            return _node
470
471_last_timestamp = None
472
473def uuid1(node=None, clock_seq=None):
474    """Generate a UUID from a host ID, sequence number, and the current time.
475    If 'node' is not given, getnode() is used to obtain the hardware
476    address.  If 'clock_seq' is given, it is used as the sequence number;
477    otherwise a random 14-bit sequence number is chosen."""
478
479    # When the system provides a version-1 UUID generator, use it (but don't
480    # use UuidCreate here because its UUIDs don't conform to RFC 4122).
481    if _uuid_generate_time and node is clock_seq is None:
482        _uuid_generate_time(_buffer)
483        return UUID(bytes=_buffer.raw)
484
485    global _last_timestamp
486    import time
487    nanoseconds = int(time.time() * 1e9)
488    # 0x01b21dd213814000 is the number of 100-ns intervals between the
489    # UUID epoch 1582-10-15 00:00:00 and the Unix epoch 1970-01-01 00:00:00.
490    timestamp = int(nanoseconds/100) + 0x01b21dd213814000L
491    if timestamp <= _last_timestamp:
492        timestamp = _last_timestamp + 1
493    _last_timestamp = timestamp
494    if clock_seq is None:
495        import random
496        clock_seq = random.randrange(1<<14L) # instead of stable storage
497    time_low = timestamp & 0xffffffffL
498    time_mid = (timestamp >> 32L) & 0xffffL
499    time_hi_version = (timestamp >> 48L) & 0x0fffL
500    clock_seq_low = clock_seq & 0xffL
501    clock_seq_hi_variant = (clock_seq >> 8L) & 0x3fL
502    if node is None:
503        node = getnode()
504    return UUID(fields=(time_low, time_mid, time_hi_version,
505                        clock_seq_hi_variant, clock_seq_low, node), version=1)
506
507def uuid3(namespace, name):
508    """Generate a UUID from the MD5 hash of a namespace UUID and a name."""
509    import md5
510    hash = md5.md5(namespace.bytes + name).digest()
511    return UUID(bytes=hash[:16], version=3)
512
513def uuid4():
514    """Generate a random UUID."""
515
516    # When the system provides a version-4 UUID generator, use it.
517    if _uuid_generate_random:
518        _uuid_generate_random(_buffer)
519        return UUID(bytes=_buffer.raw)
520
521    # Otherwise, get randomness from urandom or the 'random' module.
522    try:
523        import os
524        return UUID(bytes=os.urandom(16), version=4)
525    except:
526        import random
527        bytes = [chr(random.randrange(256)) for i in range(16)]
528        return UUID(bytes=bytes, version=4)
529
530def uuid5(namespace, name):
531    """Generate a UUID from the SHA-1 hash of a namespace UUID and a name."""
532    import sha
533    hash = sha.sha(namespace.bytes + name).digest()
534    return UUID(bytes=hash[:16], version=5)
535
536# The following standard UUIDs are for use with uuid3() or uuid5().
537
538NAMESPACE_DNS = UUID('6ba7b810-9dad-11d1-80b4-00c04fd430c8')
539NAMESPACE_URL = UUID('6ba7b811-9dad-11d1-80b4-00c04fd430c8')
540NAMESPACE_OID = UUID('6ba7b812-9dad-11d1-80b4-00c04fd430c8')
541NAMESPACE_X500 = UUID('6ba7b814-9dad-11d1-80b4-00c04fd430c8')