def test_filter_with_datetime_timezone(self):
dt1 = datetime.datetime(2015, 4, 17, 23, 1, 2, 3000, tzinfo=UTCOffsetTimezone(0))
dt2 = datetime.datetime(2015, 4, 17, 23, 1, 2, 3000, tzinfo=UTCOffsetTimezone(1))
row = Row(date=dt1)
df = self.spark.createDataFrame([row])
self.assertEqual(0, df.filter(df.date == dt2).count())
self.assertEqual(1, df.filter(df.date > dt2).count())
self.assertEqual(0, df.filter(df.date < dt2).count())
def test_time_with_timezone(self):
day = datetime.date.today()
now = datetime.datetime.now()
ts = time.mktime(now.timetuple())
# class in __main__ is not serializable
from pyspark.testing.sqlutils import UTCOffsetTimezone
utc = UTCOffsetTimezone()
utcnow = datetime.datetime.utcfromtimestamp(ts) # without microseconds
# add microseconds to utcnow (keeping year,month,day,hour,minute,second)
utcnow = datetime.datetime(*(utcnow.timetuple()[:6] + (now.microsecond, utc)))
df = self.spark.createDataFrame([(day, now, utcnow)])
day1, now1, utcnow1 = df.first()
self.assertEqual(day1, day)
self.assertEqual(now, now1)
self.assertEqual(now, utcnow1)