Error `datetime doesn't define __round__ method` when datetime used as dict key with some of of the `ignore_...` flags

[cunnane]

Receive error datetime doesn't define __round__ method when datetime is used as dict key and ignore_numeric_type_changes or ignore_string_case or ignore_string_type_changes is specified.

To Reproduce

import datetime
from deepdiff import DeepDiff
d1 = {datetime.datetime(2020, 5, 17, 22, 15): 10.0}
d2 = {datetime.datetime(2020, 5, 17, 22, 15): 10}
DeepDiff(d1, d2, ignore_numeric_type_changes=True)

Expected behavior
Output an empty dict

• OS: Windows
• Version Latest
• Python Version 3.11
• DeepDiff Version 8.5.0 to 7.0.1 (have not checked others)

[Mukller]

Root cause

number_to_string() in deepdiff/helper.py includes the datetimes tuple in the broader numbers tuple so that datetime objects can participate in numeric comparisons. However, the else branch (reached when the value is neither Decimal, Fraction, nor complex) calls:

number = round(number=number, ndigits=significant_digits)

datetime objects do not implement __round__, so this raises:

TypeError: type datetime.datetime doesn't define __round__ method

This is triggered whenever a datetime is used as a dict key and any of ignore_numeric_type_changes, ignore_string_case, or ignore_string_type_changes is set — because these flags cause _get_value_for_key to invoke number_to_string on the key.

Fix (PR linked above)

Add elif isinstance(number, datetimes): return number before the else branch. Datetime values are already unique and hashable; they need no rounding and can pass through unchanged. This short-circuits before the round() call.