def testFind(self):
elem_list = [
Filler(' '),
DayOfMonth(),
Filler('/'),
MonthNum(),
Hour24(),
Year4()
]
t = ruleproc.Sequence.find
self.assertEqual(0, t([Filler(' ')], elem_list))
self.assertEqual(3, t([MonthNum], elem_list))
self.assertEqual(2, t([Filler('/'), MonthNum()], elem_list))
self.assertEqual(4, t([Hour24, Year4()], elem_list))
elem_list = [
WeekdayShort, MonthTextShort,
Filler(' '), Hour24,
Filler(':'), Minute,
Filler(':'), Second,
Filler(' '), Timezone,
Filler(' '), Year4
]
self.assertEqual(3, t([Hour24, Filler(':')], elem_list))
def testTagMostLikely(self):
examples = ['8/12/2004', '8/14/2004', '8/16/2004', '8/25/2004']
t = _tag_most_likely
actual = t(examples)
expected = [
MonthNum(),
Filler('/'),
DayOfMonth(),
Filler('/'),
Year4()
]
self.assertListEqual(actual, expected)
def testNext(self):
elem_list = [
Filler(' '),
DayOfMonth(),
Filler('/'),
MonthNum(),
Hour24(),
Year4()
]
next1 = ruleproc.Next(DayOfMonth, MonthNum)
self.assertTrue(next1.is_true(elem_list))
next2 = ruleproc.Next(MonthNum, Hour24)
self.assertTrue(next2.is_true(elem_list))
next3 = ruleproc.Next(Filler, Year4)
self.assertFalse(next3.is_true(elem_list))
def testMostRestrictive(self):
t = _most_restrictive
self.assertEqual(MonthNum(), t([DayOfMonth(), MonthNum, Year4()]))
self.assertEqual(Year2(), t([Year4(), Year2()]))
from dateinfer.date_elements import (AMPM, DayOfMonth, Filler, Hour12, Hour24,
Minute, MonthNum, MonthTextLong,
MonthTextShort, Second, Timezone,
UTCOffset, WeekdayLong, WeekdayShort,
Year2, Year4)
from dateinfer.ruleproc import (And, Contains, Duplicate, If, KeepOriginal,
Sequence, Swap, SwapDuplicateWhereSequenceNot,
SwapSequence)
# DATE_ELEMENTS is an ordered sequence of date elements, excluding the filler. It is ordered
# in descending "restrictivity".
# The order is a little loose since date element domains do not necessarily overlap (e.g., the
# range of Jan .. Dec is 12, but the domain is independent of hours 0 .. 23), but overall a lesser
# value should be preferred over a greater value.
# The RULES will be applied after the list is generated following these precedence rules.
DATE_ELEMENTS = (AMPM(), MonthNum(), Hour12(), Hour24(), DayOfMonth(),
Minute(), Second(), Year2(), Year4(), UTCOffset(),
MonthTextShort(), MonthTextLong(), WeekdayShort(),
WeekdayLong(), Timezone())
F = Filler # short-hand to clarify rules
RULES = [
If(Sequence(Year4, Year2), SwapSequence([Year4, Year2],
[Year4, MonthNum])),
If(
Sequence(MonthNum, F('/'), r'\d', F('/'), Year4),
SwapSequence(
[MonthNum, F('/'), r'\d', F('/'), Year4],
[MonthNum, F('/'), DayOfMonth,
F('/'), Year4])),
If(
Year2, Year4)
from dateinfer.ruleproc import (And, Contains, Duplicate, If, KeepOriginal,
Sequence, Swap, SwapDuplicateWhereSequenceNot,
SwapSequence)
# DATE_ELEMENTS is an ordered sequence of date elements, excluding the filler. It is ordered
# in descending "restrictivity".
# The order is a little loose since date element domains do not necessarily overlap (e.g., the
# range of Jan .. Dec is 12, but the domain is independent of hours 0 .. 23), but overall a lesser
# value should be preferred over a greater value.
# The RULES will be applied after the list is generated following these precedence rules.
DATE_ELEMENTS = (AMPM(),
MonthNum(),
Hour12(),
Hour24(),
DayOfMonth(),
Minute(),
Second(),
Year2(),
Year4(),
UTCOffset(),
MonthTextShort(),
MonthTextLong(),
WeekdayShort(),
WeekdayLong(),
Timezone())
F = Filler # short-hand to clarify rules
RULES = [
If(Sequence(MonthNum, F('/'), r'\d', F('/'), Year4),
SwapSequence([MonthNum, F('/'), r'\d', F('/'), Year4],