This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author rubasov
Recipients rubasov
Date 2018-07-18.08:46:07
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1531903567.77.0.56676864532.issue34145@psf.upfronthosting.co.za>
In-reply-to
Content
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:
https://github.com/python/cpython/blob/ea9a0994cd0f4bd37799b045c34097eb21662b3d/Lib/uuid.py#L603

cPython 3.6:
https://github.com/python/cpython/blob/e9e2fd75ccbc6e9a5221cf3525e39e9d042d843f/Lib/uuid.py#L628

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?
History
Date User Action Args
2018-07-18 08:46:07rubasovsetrecipients: + rubasov
2018-07-18 08:46:07rubasovsetmessageid: <1531903567.77.0.56676864532.issue34145@psf.upfronthosting.co.za>
2018-07-18 08:46:07rubasovlinkissue34145 messages
2018-07-18 08:46:07rubasovcreate