def testMonthRange(self):
text = "2010-01/2012-05"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 2010)
self.assertEqual(result["start"]["month"], 1)
self.assertEqual(result["end"]["year"], 2012)
self.assertEqual(result["end"]["month"], 5)
def check_date(date_string):
""" for checking validity of any date expressed in the DwCA format
It does not check if date is in the future, just that it is
parseable and that min is smaller than max. Returns QCs as
described.
"""
qc_mask = 0
try:
parser = ISODateParser(date_string)
qc_mask |= qc_mask_7
ms_start = misc.date_to_millis(parser.dates["start"])
ms_end = misc.date_to_millis(parser.dates["end"])
# End before start
if ms_start <= ms_end:
qc_mask |= qc_mask_12
# No timezone
if parser.components["start"]["timezone"] is not None:
qc_mask |= qc_mask_13
# No hour:minute
if parser.components["start"]["hours"] is not None:
if parser.components["start"]["minutes"] is not None:
qc_mask |= qc_mask_13
except (ValueError, TypeError):
return qc_mask
return qc_mask
def testImpossibleTimeValues(self):
texts = [
"2018-01-01T25:01:01", "2018-01-01T01:65:01", "2018-01-01T01:01:65"
]
for text in texts:
with self.assertRaises(ValueError):
ISODateParser(text).dates
def testDateOmitted2(self):
text = "2018-03-01T05:06/T07:08"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 2018)
self.assertEqual(result["start"]["month"], 3)
self.assertEqual(result["start"]["day"], 1)
self.assertEqual(result["end"]["hours"], 7)
self.assertEqual(result["end"]["minutes"], 8)
def testYearOmitted(self):
text = "2008-02-15/03-14"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 2008)
self.assertEqual(result["start"]["month"], 2)
self.assertEqual(result["start"]["day"], 15)
self.assertEqual(result["end"]["month"], 3)
self.assertEqual(result["end"]["day"], 14)
def testTimeZoneWithoutSeparator(self):
text = "2010-01-02T03:04+0630"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 2010)
self.assertEqual(result["start"]["month"], 1)
self.assertEqual(result["start"]["day"], 2)
self.assertEqual(result["start"]["hours"], 3)
self.assertEqual(result["start"]["minutes"], 4)
self.assertEqual(result["start"]["timezone"], 6.5)
def testSimpleDateTimeSpace(self):
text = "1990-01-02 12:13:14"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 1990)
self.assertEqual(result["start"]["month"], 1)
self.assertEqual(result["start"]["day"], 2)
self.assertEqual(result["start"]["hours"], 12)
self.assertEqual(result["start"]["minutes"], 13)
self.assertEqual(result["start"]["seconds"], 14)
self.assertEqual(result["start"]["timezone"], None)
def testDateTimeUTC(self):
text = "2017-08-24T14:51:57Z"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 2017)
self.assertEqual(result["start"]["month"], 8)
self.assertEqual(result["start"]["day"], 24)
self.assertEqual(result["start"]["hours"], 14)
self.assertEqual(result["start"]["minutes"], 51)
self.assertEqual(result["start"]["seconds"], 57)
self.assertEqual(result["start"]["timezone"], 0)
def testEndDay(self):
text = "1973-06-18/26"
parser = ISODateParser(text)
result = parser.components
self.assertEqual(result["start"]["year"], 1973)
self.assertEqual(result["start"]["month"], 6)
self.assertEqual(result["start"]["day"], 18)
self.assertEqual(result["end"]["day"], 26)
dates = parser.dates
self.assertEqual(dates["start"], datetime.date(1973, 6, 18))
self.assertEqual(dates["end"], datetime.date(1973, 6, 26))
def testDateRangeTimeZone(self):
text = "1990-01-02T03:04:05/2014-05-06 07:08+0900"
result = ISODateParser(text).components
self.assertEqual(result["start"]["year"], 1990)
self.assertEqual(result["start"]["month"], 1)
self.assertEqual(result["start"]["day"], 2)
self.assertEqual(result["start"]["hours"], 3)
self.assertEqual(result["start"]["minutes"], 4)
self.assertEqual(result["start"]["seconds"], 5)
self.assertEqual(result["start"]["timezone"], None)
self.assertEqual(result["end"]["year"], 2014)
self.assertEqual(result["end"]["month"], 5)
self.assertEqual(result["end"]["day"], 6)
self.assertEqual(result["end"]["hours"], 7)
self.assertEqual(result["end"]["minutes"], 8)
self.assertEqual(result["end"]["seconds"], None)
self.assertEqual(result["end"]["timezone"], 9)
def check_record(record, min_year=0):
"""Check the eventDate."""
result = {
"missing": [],
"invalid": [],
"flags": [],
"annotations": {},
"dropped": False
}
if "eventDate" in record and record["eventDate"] is not None:
try:
parser = ISODateParser(record["eventDate"])
if parser.dates["start"].year < min_year:
# year precedes minimum year in settings
result["flags"].append(Flag.DATE_BEFORE_MIN.value)
raise ValueError
ms_start = date_to_millis(parser.dates["start"])
ms_mid = date_to_millis(parser.dates["mid"])
ms_end = date_to_millis(parser.dates["end"])
year = datetime.datetime.fromtimestamp(ms_mid / 1000).year
if ms_end > date_to_millis(datetime.date.today()):
# date in the future
result["flags"].append(Flag.DATE_IN_FUTURE.value)
raise ValueError
result["annotations"]["date_start"] = ms_start
result["annotations"]["date_mid"] = ms_mid
result["annotations"]["date_end"] = ms_end
result["annotations"]["date_year"] = year
except ValueError:
result["invalid"].append("eventDate")
except:
logger.error("Error processing date " + record["eventDate"])
raise
else:
result["missing"].append("eventDate")
return result
def check_record(record, min_year=0):
"""Check the event date """
qc_mask = 0
if "eventDate" in record and record["eventDate"] is not None:
try:
parser = ISODateParser(record["eventDate"])
# Date parsed, means format is OK
qc_mask |= qc_mask_7
if parser.dates["mid"].year >= min_year:
# year not precede minimum year in settings
qc_mask |= qc_mask_11
ms_start = misc.date_to_millis(parser.dates["start"])
ms_end = misc.date_to_millis(parser.dates["end"])
if ms_start >= misc.date_to_millis(datetime.date.today()):
# date not in the future
qc_mask |= qc_mask_11
if ms_start <= ms_end:
# Min and max consistent
qc_mask |= qc_mask_12
# is timezone filled (does it need to be checked for both ends)
if parser.components["start"]["timezone"] is not None:
qc_mask |= qc_mask_13
# is a time filled (limiting to hours:minutes)
if parser.components["start"]["hours"] is not None:
if parser.components["start"]["minutes"] is not None:
qc_mask |= qc_mask_13
except (ValueError, TypeError):
return qc_mask
return qc_mask
def testTimeMissing(self):
text = "1981-06-01+00:00"
with self.assertRaises(ValueError):
ISODateParser(text).dates
def testDates(self):
text = "1990-01/2014-05"
result = ISODateParser(text).dates
self.assertEqual(result["start"], datetime.date(1990, 1, 1))
self.assertEqual(result["end"], datetime.date(2014, 5, 31))
def testDatesOnlyStart(self):
text = "1990-01-02"
result = ISODateParser(text).dates
self.assertEqual(result["start"], datetime.date(1990, 1, 2))
self.assertEqual(result["end"], datetime.date(1990, 1, 2))
self.assertEqual(result["mid"], datetime.date(1990, 1, 2))
def testDashes(self):
text = "--"
with self.assertRaises(ValueError):
ISODateParser(text).dates
def testBackSlash(self):
text = "2003-04-30T12:00\\2003-04-30T17:30"
with self.assertRaises(ValueError):
ISODateParser(text).dates
def testDatesOnlyMonth(self):
text = "1990-01"
result = ISODateParser(text).dates
self.assertEqual(result["start"], datetime.date(1990, 1, 1))
self.assertEqual(result["end"], datetime.date(1990, 1, 31))