Issue 26711: Fix comparison of plistlib.Data

Proposed patch fixes several bugs in plistlib.Data.__eq__().

* isinstance(other, str) was used instead of isinstance(other, bytes). Data always wraps bytes and should be comparable with bytes. str was correct type in Python 2.

* id(self) == id(other) is always false, because if other is self, the first condition (isinstance(other, self.__class__)) should be true. NotImplemented should be returned as fallback. This allows comparing with Data subclasses and correct work of __ne__().

* The __eq__() method should be used instead of the equality operator. This is needed for correct work in case if value is bytes subclass with overloaded __eq__().

I don't understand the explicit use of __eq__ in the bytes case. Why is that needed?

>>> import plistlib
>>> class MyData(bytes):
...     def __eq__(self, other):
...         if isinstance(other, plistlib.Data):
...             return super().__eq__(other.value)
...         return False
... 
>>> plistlib.Data(b'x') == MyData(b'x')
True

If use the equality operator the result is False.

I don't know if this is good example. In any case this is corner case and we can manage with "==".

Ronald, my second patch uses "==" instead of __eq__. Is it good to you?

Serhiy,

I slightly prefer the second patch, but either one would be fine.  

The reason I asked about explicitly calling __eq__ is that this is an uncommon pattern (at least for me). 

Ronald

New changeset 41afb83cffac by Serhiy Storchaka in branch '3.5':
Issue #26711: Fixed the comparison of plistlib.Data with other types.
https://hg.python.org/cpython/rev/41afb83cffac

New changeset dbdd5bc4df99 by Serhiy Storchaka in branch 'default':
Issue #26711: Fixed the comparison of plistlib.Data with other types.
https://hg.python.org/cpython/rev/dbdd5bc4df99

For simplicity I left "==" and fixed only obvious bug.

assignee: serhiy.storchaka
resolution: fixed
stage: patch review -> resolved