Croniter skips 2021-03-01 for "0 0 */10 * *" #178 (DST related bugs)

[cuu508]

(originally reported in taichino/croniter#178)

Test snippet:

from datetime import datetime
from croniter import croniter

it = croniter("0 0 */10 * *", datetime(2021, 1, 1))

for i in range(0, 15):
    print(it.get_next(datetime).isoformat())

Result:

2021-01-11T00:00:00
2021-01-21T00:00:00
2021-01-31T00:00:00
2021-02-01T00:00:00
2021-02-11T00:00:00
2021-02-21T00:00:00
2021-03-11T00:00:00
2021-03-21T00:00:00
2021-03-31T00:00:00
2021-04-01T00:00:00
2021-04-11T00:00:00
2021-04-21T00:00:00
2021-05-01T00:00:00
2021-05-11T00:00:00
2021-05-21T00:00:00

In the above output, 2021-03-01T00:00:00 is missing.

[kiorky]

dup taichino/croniter#147

[kiorky]

as said on taichino/croniter#147 (comment) and
taichino/croniter#178 (comment) and
taichino/croniter#141

i wont have time to fix dst changes, unfortunatly, if you can, you may submit a PR.
Thanks though for the report.

[kiorky]

The last work i didnt have time to finish was there : kiorky/croniter-fork#34

[evanpurkhiser]

Seems like some of these tickets are missing implementation detail differences about which tzinfo implementation is being used.

I've noticed for our case at Sentry, we use the newer zoneinfo implementation, and this test case is failing

@pytest.mark.skipif(
    sys.version_info < (3, 9),
    reason="Test only valid on python 3.9 and above"
)
def test_dst_zoneinfo(self):
    """
    Test DST crossover with zoneinfo
    """
    from zoneinfo import ZoneInfo

    # 5:00 AM EST before EDST crossover
    tz = ZoneInfo("America/New_York")
    ts = datetime(2024, 3, 9, 5, 0, 0, tzinfo=tz)

    # Next tick is after the EDST crossover. Closk move forward one hour,
    # but we would still expect the next cron iteration at 5:00 AM EDST.
    ts_after_dst = croniter("0 5 * * *", ts).get_next(datetime)
    self.assertEqual(ts_after_dst, datetime(2024, 3, 10, 5, 0, 0, tzinfo=tz))

This test case is producing 2024-3-10 4:00:00-04:00, but we would expect 2024-3-10 5:00:00-04:00.

The same test case using pytz

def test_dst_pytz(self):
    """
    Test DST crossover with zoneinfo
    """
    # 5:00 AM EST before EDST crossover
    tz = pytz.timezone("America/New_York")
    ts = tz.localize(datetime(2024, 3, 9, 5, 0, 0))

    # Next tick is after the EDST crossover. Closk move forward one hour,
    # but we would still expect the next cron iteration at 5:00 AM EDST.
    ts_after_dst = croniter("0 5 * * *", ts).get_next(datetime)
    self.assertEqual(ts_after_dst, tz.localize(datetime(2024, 3, 10, 5, 0, 0)))

This test passes.

[cuu508]

@evanpurkhiser have you looked into using redacted? It is designed to use zoneinfo, and handles DST.

[kiorky]

@cuu508 chiming in grabbing users to another library is not something i ll tolerate on this tracker related to croniter issues.

You have been warned.