1.2.0 - PREFER_DATES_FROM & Timezone bug

[Keiron-Beadle]

It's currently 08/10/2024 23:37 pm Europe/London time. Observe the following code,

time_zone = 'Europe/London'
tz_result = dateparser.parse('at 12am', settings={'PREFER_DATES_FROM' : 'future', 'TIMEZONE': time_zone})
normal_result = dateparser.parse('at 12am', settings={'PREFER_DATES_FROM': 'future'})
print(tz_result)
print(normal_result)

I'd be expecting both results to print, 2024-10-09 00:00:00. The actual prints are,

2024-10-08 00:00:00
2024-10-09 00:00:00

this may be intended behaviour but it doesn't seem like it from my perspective.

Edit: Possibly this is user error, as using 'TO_TIMEZONE' : time_zone instead of TIMEZONE : time_zone seems to return the correct 12am for the tz_result variable. Although I'm a still little confused on why "TIMEZONE" was returning me the previous day's 12am since I've specified PREFER_DATES_FROM_FUTURE. If anyone knows why I'd love to hear it, thanks

[gutsytechster]

Hi @Keiron-Beadle

I was trying to reproduce the behavior observed by you, i.e. different output with the parameter provided & not provided, however, I couldn't do that.

At the same time, I think I discovered one of the possible bug. When we enter the value at 12 am , it is parsed with RELATIVE_TIME to be current time in UTC, while I think it should be RELATIVE TIME in the given TIMEZONE or the local timezone.

Example

>>> from datetime import datetime
>>> datetime.now()
datetime.datetime(2024, 10, 27, 1, 19, 34, 921587)
>>> dateparser.parse('at 12 am')
datetime.datetime(2024, 10, 26, 0, 0)
>>> dateparser.parse('at 12 am', settings={'PREFER_DATES_FROM': 'future'})
datetime.datetime(2024, 10, 27, 0, 0)
>>> dateparser.parse('at 12 am', settings={'PREFER_DATES_FROM': 'future', 'TIMEZONE': 'Asia/Kolkata'})
datetime.datetime(2024, 10, 27, 0, 0)
>>> dateparser.parse('at 12 am', settings={'PREFER_DATES_FROM': 'future', 'TO_TIMEZONE': 'Asia/Kolkata'})
datetime.datetime(2024, 10, 27, 0, 0)

Here, the current time is 27/10/2024 1:19 AM Asia/Kolkata. The default parsing use the current time in UTC and hence date is picked as 26/10/2024.

If I apply the PREFER_DATES_FROM to future , the date is moved one day ahead, which is correct w.r.t the UTC timezone, but incorrect if I address my local timezone. It also doesn't respect the timezone provided in the settings and only apply it after parsing.

A solution to this would be to provide explicity RELATIVE_BASE to the settings with the datetime.now() value that would pick your local timezone value.