pandas.Series.to_json() incorrectly localizes tz-naive datetimes to UTC

[simplydac]

Seemingly related but not identical to #12997

How to reproduce the bug

Checked with pandas 0.24.2, 1.1.5 and 1.2.0.

import pandas as pd

index = pd.date_range(
    start='2020-12-28 00:00:00',
    end='2020-12-28 02:00:00',
    freq='1H',
)
a = pd.Series(
    data=range(3),
    index=index,
)
a.to_json(date_format='iso')

Output

'{"2020-12-28T00:00:00.000Z":0,"2020-12-28T01:00:00.000Z":1,"2020-12-28T02:00:00.000Z":2}'

(Notice the "Z" indicating UTC)

Expected output

'{"2020-12-28T00:00:00.000":0,"2020-12-28T01:00:00.000":1,"2020-12-28T02:00:00.000":2}'

(without "Z")

[simplydac]

Solving #12997 should close this issue as well.

[jreback]

Seemingly related but not identical to #12997

how so?

[simplydac]

how so?

Related:

• to_json converts to UTC when encoding ISO formatted datetimes #12997 may be at the root of the present issue because pandas.Series.to_json() calls pandas.io.json.to_json(), which is the subject of to_json converts to UTC when encoding ISO formatted datetimes #12997. See https://github.com/pandas-dev/pandas/blob/v1.2.0/pandas/core/generic.py#L2196-L2477, lines 2453 and 2463.
• to_json converts to UTC when encoding ISO formatted datetimes #12997 does mention pandas.DataFrame.to_json()

Not identical:

• to_json converts to UTC when encoding ISO formatted datetimes #12997 mentions timezone conversion on a tz-aware timestamp (which points to a same instant, so somehow correct though clearly unexpected), while the present issue mentions localization of tz-naive timestamps (which is plainly incorrect):

import pandas as pd

pd.Timestamp('2020-01-01 00:00:00+00:00') == pd.Timestamp('2020-01-01 05:00:00+05:00')  # True
pd.Timestamp('2020-01-01 00:00:00.000Z') == pd.Timestamp('2020-01-01 00:00:00')  # False

Besides, the 2nd comparison above should throw an exception (does it make sens to compare tz-aware with tz-naive?!), but that is another subject.