ldif.LDIFRecordList handles all attribute names as utf-8 strings
and all attribute values as bytes. If we take the attribute value
and try to search for it in the entry (= dictionary), if it contains
the attribute name as a key (which is a string), their hashes match.
However, even if hashes match, Python needs to make sure those two
are the same in case of a hash collision, so it tries to compare them.
This causes BytesWarning exception when running in strict mode
because `bytes` and `str` instances cannot be compared. KeyError
would be thrown in a non-strict mode.
Also, when later passing the attr to replace_value(), we need for it
to be `str` otherwise the modifications handler fails because it
tries to sort the attributes it's modifying but that's a bit less
poetic issue than the first one.
https://pagure.io/freeipa/issue/7129
Reviewed-By: Michal Reznik <mreznik@redhat.com>
Stanislav Laznicka
0f13e663ca