uuid3 and uuid5 hard to use portably between Python 2 and 3

[rubasov]

BPO	34145
Nosy	@zware, @tirkarthi, @rubasov
PRs	[2.7] bpo-34145: Fix uuid3 and uuid5 to accept unicode names #10100

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2020-01-19.18:21:29.617>
created_at = <Date 2018-07-18.08:46:07.736>
labels = ['library']
title = 'uuid3 and uuid5 hard to use portably between Python 2 and 3'
updated_at = <Date 2020-01-19.18:21:29.615>
user = 'https://github.com/rubasov'

bugs.python.org fields:

activity = <Date 2020-01-19.18:21:29.615>
actor = 'zach.ware'
assignee = 'none'
closed = True
closed_date = <Date 2020-01-19.18:21:29.617>
closer = 'zach.ware'
components = ['Library (Lib)']
creation = <Date 2018-07-18.08:46:07.736>
creator = 'rubasov'
dependencies = []
files = []
hgrepos = []
issue_num = 34145
keywords = ['patch']
message_count = 2.0
messages = ['321870', '360265']
nosy_count = 3.0
nosy_names = ['zach.ware', 'xtreak', 'rubasov']
pr_nums = ['10100']
priority = 'normal'
resolution = 'out of date'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue34145'
versions = ['Python 2.7']

[rubasov]

The issue I'd like to report may not be an outright bug neither in cPython 2 nor in cPython 3, but more of a wishlist item to help Python programmers writing UUID-handling code that's valid in Python 2 and 3 at the same time.

Please consider these one-liners:

    $ python2 -c 'import uuid ; uuid.uuid5(namespace=uuid.UUID("850aeee8-e173-4da1-9d6b-dd06e4b06747"), name="foo")'
    $ python3 -c 'import uuid ; uuid.uuid5(namespace=uuid.UUID("850aeee8-e173-4da1-9d6b-dd06e4b06747"), name="foo")'

As long as the 'name' input to uuid.uuid3() or uuid.uuid5() is the literal string type of the relevant Python version there's no problem at all. However if you'd like to handle both unicode and non-unicode 'name' input in valid Python2/3 code then I find that's impossible to express without relying on Python version checking.

cPython2's uuid module is incompatible with unicode input:

    $ python2 -c 'import uuid ; uuid.uuid5(namespace=uuid.UUID("850aeee8-e173-4da1-9d6b-dd06e4b06747"), name=u"foo")'
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/usr/lib/python2.7/uuid.py", line 589, in uuid5
        hash = sha1(namespace.bytes + name).digest()
    UnicodeDecodeError: 'ascii' codec can't decode byte 0x85 in position 0: ordinal not in range(128)

cPython3's uuid module is incompatible with non-unicode input:

    $ python3 -c 'import uuid ; uuid.uuid5(namespace=uuid.UUID("850aeee8-e173-4da1-9d6b-dd06e4b06747"), name=b"foo")'
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/usr/lib/python3.5/uuid.py", line 608, in uuid5
        hash = sha1(namespace.bytes + bytes(name, "utf-8")).digest()
    TypeError: encoding without a string argument

The reason is obvious looking at the uuid modules' source code:

cPython 2.7:

cpython/Lib/uuid.py

Line 603 in ea9a099

hash = sha1(namespace.bytes + name).digest()

cPython 3.6:

cpython/Lib/uuid.py

Line 628 in e9e2fd7

hash = sha1(namespace.bytes + bytes(name, "utf-8")).digest()

Therefore portable code has to resort to version checking like this:

    import six
    import uuid

    if six.PY2:
        name = name.encode('utf-8')
    uuid.uuid5(namespace=namespace, name=name)

IMHO this inconvenience could be avoided if cPython2's uuid.uuid3() and uuid.uuid5() had been changed to also accept unicode 'name' arguments and encode() them implicitly.

What do you think?

[zware]

Python 2.7 has reached EOL, and so this change can no longer be made. Thanks for the idea report and idea anyway, Bence!