熊猫日期时间转换不一致

Question

A reproducible example:

import pandas as pd
now = pd.Timestamp('2018-04-09 09:10')
start_of_today = pd.datetime(now.year, now.month, now.day, 0)
print(pd.to_datetime(start_of_today.timestamp(), unit='s'))

The last line will return 2018-04-08 22:00:00 instead of 2018-04-09 00:00, why does this happen?

Answer 1

As per datetime.timestamp documentation :

For aware datetime instances, the return value is computed as:

(dt - datetime(1970, 1, 1, tzinfo=timezone.utc)).total_seconds()

Therefore, you need to align your pd.datetime object with UTC timezone:

from datetime import timezone
import pandas as pd

now = pd.Timestamp('2018-04-09 09:10')
start_of_today = pd.datetime(now.year, now.month, now.day, 0, 0, 0, 0, timezone.utc)
print(pd.to_datetime(start_of_today.timestamp(), unit='s'))

Answer 2

This is because of datetime.timestamp

it returns (dt - datetime(1970, 1, 1, tzinfo=timezone.utc)).total_seconds()

timezone is utc

You can see

In [65]: start_of_today.timestamp()
Out[65]: 1523203200.0

In [66]: pd.Timestamp('2018-04-09 00:00').timestamp()
Out[66]: 1523232000.0

Their timestamps are different.