timezone error in parse_rfc822

[shortwinds]

Parse datetime string error in parse_rfc822.
parsedate_to_datetime use datetime.timezone.utc in email.utils which depends local settings.

test_code:

from odps.utils import  parse_rfc822
rfc822_str = "Thu, 05 Oct 2023 14:30:00 +0000"
dt = parse_rfc822(rfc822_str)
print(dt)

mac output:

2023-10-05 22:30:00

docker output:

2023-10-06 06:30:00

[wjsi]

The internal utility parse_rfc822 converts string to local time, and this is by-design.

[shortwinds]

docker:

date +"%Z %z"
> CST +0800

 date
> Tue May 20 17:22:34 CST 2025

mac:

systemsetup -gettimezone
> Time Zone: Asia/Shanghai

 date
> Tue May 20 17:22:07 CST 2025

Mac timezone and docker timezone is same , but parse_rfc822 output is not same.

[shortwinds]

UTC debug output:
docker_error

zdump -v /usr/share/zoneinfo/UTC | tail
/usr/share/zoneinfo/UTC  Sat Apr 14 17:59:59 1990 UT = Sun Apr 15 01:59:59 1990 CST isdst=0 gmtoff=28800
/usr/share/zoneinfo/UTC  Sat Apr 14 18:00:00 1990 UT = Sun Apr 15 03:00:00 1990 CDT isdst=1 gmtoff=32400
/usr/share/zoneinfo/UTC  Sat Sep 15 16:59:59 1990 UT = Sun Sep 16 01:59:59 1990 CDT isdst=1 gmtoff=32400
/usr/share/zoneinfo/UTC  Sat Sep 15 17:00:00 1990 UT = Sun Sep 16 01:00:00 1990 CST isdst=0 gmtoff=28800
/usr/share/zoneinfo/UTC  Sat Apr 13 17:59:59 1991 UT = Sun Apr 14 01:59:59 1991 CST isdst=0 gmtoff=28800
/usr/share/zoneinfo/UTC  Sat Apr 13 18:00:00 1991 UT = Sun Apr 14 03:00:00 1991 CDT isdst=1 gmtoff=32400
/usr/share/zoneinfo/UTC  Sat Sep 14 16:59:59 1991 UT = Sun Sep 15 01:59:59 1991 CDT isdst=1 gmtoff=32400
/usr/share/zoneinfo/UTC  Sat Sep 14 17:00:00 1991 UT = Sun Sep 15 01:00:00 1991 CST isdst=0 gmtoff=28800
/usr/share/zoneinfo/UTC  9223372036854689407 = NULL
/usr/share/zoneinfo/UTC  9223372036854775807 = NULL

docker_good

usr/share/zoneinfo/UTC  -9223372036854775808 = NULL
/usr/share/zoneinfo/UTC  -9223372036854689408 = NULL
/usr/share/zoneinfo/UTC  9223372036854689407 = NULL
/usr/share/zoneinfo/UTC  9223372036854775807 = NULL

The UTC file is not work well from the above result.
Is there other UTC method which is not depend /etc/localtime? like datetime.timezone.utc?

[wjsi]

Please show results returned by email.utils.parsedate_to_datetime. We does not change what it returns by now.

[shortwinds]

test_code

from email.utils import parsedate_to_datetime
from datetime import timezone

s = "Thu, 05 Oct 2023 14:30:00 GMT"
dt = parsedate_to_datetime(s)
import zoneinfo

utc = zoneinfo.ZoneInfo("UTC")
dt = dt.astimezone(utc)

print(f"{dt} ({repr(dt.tzinfo)})") 

output

2023-10-05 22:30:00+08:00 (zoneinfo.ZoneInfo(key='UTC'))

utc file in os is not right, then python use the os utc produce the probem

resolve

from email.utils import parsedate_to_datetime
from datetime import timezone

s = "Thu, 05 Oct 2023 14:30:00 GMT"
dt = parsedate_to_datetime(s)
# import zoneinfo

# utc = zoneinfo.ZoneInfo("UTC")
# dt = dt.astimezone(utc)
if dt.tzinfo is not timezone.utc:
    if dt.utcoffset() == timezone.utc.utcoffset(None):
        dt = dt.replace(tzinfo=timezone.utc)

print(f"{dt} ({repr(dt.tzinfo)})") 

output

2023-10-05 14:30:00+00:00 (datetime.timezone.utc)

[wjsi]

Therefore replacing zoneinfo.ZoneInfo("UTC") with datetime.timezone.utc will work, OK?

[shortwinds]

yes.