def test_parse_str(self):
# Parser should be able to handle bytestring and unicode
uni_str = '2014-05-01 08:00:00'
bytes_str = uni_str.encode()
res = parse(bytes_str)
expected = parse(uni_str)
assert res == expected
def testYearFirst(self):
dtstr = '090107'
# Should be MMDDYY
self.assertEqual(parse(dtstr, yearfirst=True),
datetime(2009, 1, 7))
self.assertEqual(parse(dtstr, yearfirst=True, dayfirst=False),
datetime(2009, 1, 7))
def testDayFirst(self):
dtstr = '090107'
# Should be DDMMYY
self.assertEqual(parse(dtstr, dayfirst=True),
datetime(2007, 1, 9))
self.assertEqual(parse(dtstr, yearfirst=False, dayfirst=True),
datetime(2007, 1, 9))
def test_includes_timestr(self):
timestr = "2020-13-97T44:61:83"
try:
parse(timestr)
except ParserError as e:
assert e.args[1] == timestr
else:
pytest.fail("Failed to raise ParserError")
def testNoYearFirstNoDayFirst(self):
dtstr = '090107'
# Should be MMDDYY
self.assertEqual(parse(dtstr),
datetime(2007, 9, 1))
self.assertEqual(parse(dtstr, yearfirst=False, dayfirst=False),
datetime(2007, 9, 1))
def testFuzzyWithTokens(self):
s1 = "Today is 25 of September of 2003, exactly " \
"at 10:49:41 with timezone -03:00."
self.assertEqual(parse(s1, fuzzy_with_tokens=True),
(datetime(2003, 9, 25, 10, 49, 41,
tzinfo=self.brsttz),
('Today is ', 'of ', ', exactly at ',
' with timezone ', '.')))
s2 = "http://biz.yahoo.com/ipo/p/600221.html"
self.assertEqual(parse(s2, fuzzy_with_tokens=True),
(datetime(2060, 2, 21, 0, 0, 0),
('http://biz.yahoo.com/ipo/p/', '.html')))
def testFuzzyAMPMProblem(self):
# Sometimes fuzzy parsing results in AM/PM flag being set without
# hours - if it's fuzzy it should ignore that.
s1 = "I have a meeting on March 1, 1974."
s2 = "On June 8th, 2020, I am going to be the first man on Mars"
# Also don't want any erroneous AM or PMs changing the parsed time
s3 = "Meet me at the AM/PM on Sunset at 3:00 AM on December 3rd, 2003"
s4 = "Meet me at 3:00AM on December 3rd, 2003 at the AM/PM on Sunset"
self.assertEqual(parse(s1, fuzzy=True), datetime(1974, 3, 1))
self.assertEqual(parse(s2, fuzzy=True), datetime(2020, 6, 8))
self.assertEqual(parse(s3, fuzzy=True), datetime(2003, 12, 3, 3))
self.assertEqual(parse(s4, fuzzy=True), datetime(2003, 12, 3, 3))
def testParseWithNulls(self):
# This relies on the from __future__ import unicode_literals, because
# explicitly specifying a unicode literal is a syntax error in Py 3.2
# May want to switch to u'...' if we ever drop Python 3.2 support.
pstring = '\x00\x00August 29, 1924'
assert parse(pstring) == datetime(1924, 8, 29)
def test_somewhat_ambiguous_string(self):
# Ref: github issue #487
# The parser is choosing the wrong part for hour
# causing datetime to raise an exception.
dtstr = '1237 PM BRST Mon Oct 30 2017'
res = parse(dtstr, tzinfo=self.tzinfos)
assert res == datetime(2017, 10, 30, 12, 37, tzinfo=self.tzinfos)
def test_on_era(self):
# This could be classified as an "eras" test, but the relevant part
# about this is the ` on `
dstr = '2:15 PM on January 2nd 1973 A.D.'
expected = datetime(1973, 1, 2, 14, 15)
res = parse(dstr)
assert res == expected
def test_extraneous_year2(self):
# This was found in the wild at insidertrading.org
dstr = ("Berylson Amy Smith 1998 Grantor Retained Annuity Trust "
"u/d/t November 2, 1998 f/b/o Jennifer L Berylson")
res = parse(dstr, fuzzy_with_tokens=True)
expected = datetime(1998, 11, 2)
assert res == expected
def test_unambiguous_YYYYMM(self):
# 171206 can be parsed as YYMMDD. However, 201712 cannot be parsed
# as instance of YYMMDD and parser could fallback to YYYYMM format.
dstr = "201712"
res = parse(dstr)
expected = datetime(2017, 12, 1)
assert res == expected
def test_parse_unambiguous_nonexistent_local(self):
# When dates are specified "EST" even when they should be "EDT" in the
# local time zone, we should still assign the local time zone
with TZEnvContext('EST+5EDT,M3.2.0/2,M11.1.0/2'):
dt_exp = datetime(2011, 8, 1, 12, 30, tzinfo=tz.tzlocal())
dt = parse('2011-08-01T12:30 EST')
assert dt.tzname() == 'EDT'
assert dt == dt_exp
def test_tzlocal_in_gmt(self):
# GH #318
with TZEnvContext('GMT0BST,M3.5.0,M10.5.0'):
# This is an imaginary datetime in tz.tzlocal() but should still
# parse using the GMT-as-alias-for-UTC rule
dt = parse('2004-05-01T12:00 GMT')
dt_exp = datetime(2004, 5, 1, 12, tzinfo=tz.UTC)
assert dt == dt_exp
def test_strftime_formats_2003Sep25(self, fmt, dstr):
expected = datetime(2003, 9, 25)
# First check that the format strings behave as expected
# (not strictly necessary, but nice to have)
assert expected.strftime(fmt) == dstr
res = parse(dstr)
assert res == expected
def test_valid_tzinfo_callable_input(self):
dstr = "2014 January 19 09:00 UTC"
def tzinfos(*args, **kwargs):
return u"UTC+0"
expected = datetime(2014, 1, 19, 9, tzinfo=tz.tzstr("UTC+0"))
res = parse(dstr, tzinfos=tzinfos)
self.assert_equal_same_tz(res, expected)
def testMicrosecondPrecisionErrorReturns(self):
# One more precision issue, discovered by Eric Brown. This should
# be the last one, as we're no longer using floating points.
for ms in [100001, 100000, 99999, 99998,
10001, 10000, 9999, 9998,
1001, 1000, 999, 998,
101, 100, 99, 98]:
dt = datetime(2008, 2, 27, 21, 26, 1, ms)
assert parse(dt.isoformat()) == dt
def test_extraneous_year_tokens(self):
# This was found in the wild at insidertrading.org
# Unlike in the case above, identifying the first "2012" as the year
# would not be a problem, but inferring that the latter 2012 is hhmm
# is a problem.
dstr = "2012 MARTIN CHILDREN'S IRREVOCABLE TRUST u/a/d NOVEMBER 7, 2012"
expected = datetime(2012, 11, 7)
(res, tokens) = parse(dstr, fuzzy_with_tokens=True)
assert res == expected
assert tokens == ("2012 MARTIN CHILDREN'S IRREVOCABLE TRUST u/a/d ",)
def testIncreasingCTime(self):
# This test will check 200 different years, every month, every day,
# every hour, every minute, every second, and every weekday, using
# a delta of more or less 1 year, 1 month, 1 day, 1 minute and
# 1 second.
delta = timedelta(days=365+31+1, seconds=1+60+60*60)
dt = datetime(1900, 1, 1, 0, 0, 0, 0)
for i in range(200):
assert parse(dt.ctime()) == dt
dt += delta
def test_parse_tzinfos_fold():
NYC = tz.gettz('America/New_York')
tzinfos = {'EST': NYC, 'EDT': NYC}
dt_exp = tz.enfold(datetime(2011, 11, 6, 1, 30, tzinfo=NYC), fold=1)
dt = parse('2011-11-06T01:30 EST', tzinfos=tzinfos)
assert dt == dt_exp
assert dt.tzinfo is dt_exp.tzinfo
assert getattr(dt, 'fold') == getattr(dt_exp, 'fold')
assert dt.astimezone(tz.UTC) == dt_exp.astimezone(tz.UTC)
def test_tzlocal_parse_fold(self):
# One manifestion of GH #318
with TZEnvContext('EST+5EDT,M3.2.0/2,M11.1.0/2'):
dt_exp = datetime(2011, 11, 6, 1, 30, tzinfo=tz.tzlocal())
dt_exp = tz.enfold(dt_exp, fold=1)
dt = parse('2011-11-06T01:30 EST')
# Because this is ambiguous, until `tz.tzlocal() is tz.tzlocal()`
# we'll just check the attributes we care about rather than
# dt == dt_exp
assert dt.tzname() == dt_exp.tzname()
assert dt.replace(tzinfo=None) == dt_exp.replace(tzinfo=None)
assert getattr(dt, 'fold') == getattr(dt_exp, 'fold')
assert dt.astimezone(tz.UTC) == dt_exp.astimezone(tz.UTC)
def test_duck_typing(self):
# We want to support arbitrary classes that implement the stream
# interface.
class StringPassThrough(object):
def __init__(self, stream):
self.stream = stream
def read(self, *args, **kwargs):
return self.stream.read(*args, **kwargs)
dstr = StringPassThrough(StringIO('2014 January 19'))
res = parse(dstr)
expected = datetime(2014, 1, 19)
assert res == expected
def testParseUnicodeWords(self):
class rus_parserinfo(parserinfo):
MONTHS = [("янв", "Январь"),
("фев", "Февраль"),
("мар", "Март"),
("апр", "Апрель"),
("май", "Май"),
("июн", "Июнь"),
("июл", "Июль"),
("авг", "Август"),
("сен", "Сентябрь"),
("окт", "Октябрь"),
("ноя", "Ноябрь"),
("дек", "Декабрь")]
expected = datetime(2015, 9, 10, 10, 20)
res = parse('10 Сентябрь 2015 10:20', parserinfo=rus_parserinfo())
assert res == expected
def test_ybd(self):
# If we have a 4-digit year, a non-numeric month (abbreviated or not),
# and a day (1 or 2 digits), then there is no ambiguity as to which
# token is a year/month/day. This holds regardless of what order the
# terms are in and for each of the separators below.
seps = ['-', ' ', '/', '.']
year_tokens = ['%Y']
month_tokens = ['%b', '%B']
day_tokens = ['%d']
if PLATFORM_HAS_DASH_D:
day_tokens.append('%-d')
prods = itertools.product(year_tokens, month_tokens, day_tokens)
perms = [y for x in prods for y in itertools.permutations(x)]
unambig_fmts = [sep.join(perm) for sep in seps for perm in perms]
actual = datetime(2003, 9, 25)
for fmt in unambig_fmts:
dstr = actual.strftime(fmt)
res = parse(dstr)
assert res == actual
def test_decimal_error(value):
# GH 632, GH 662 - decimal.Decimal raises some non-ParserError exception
# when constructed with an invalid value
with pytest.raises(ParserError):
parse(value)
def test_rounding_floatlike_strings(dtstr, dt):
assert parse(dtstr, default=datetime(2003, 9, 25)) == dt
def test_four_letter_day(self):
dstr = 'Frid Dec 30, 2016'
expected = datetime(2016, 12, 30)
res = parse(dstr)
assert res == expected
def test_non_date_number(self):
dstr = '1,700'
with pytest.raises(ParserError):
parse(dstr)
def test_extraneous_year(self):
# This was found in the wild at insidertrading.org
dstr = "2011 MARTIN CHILDREN'S IRREVOCABLE TRUST u/a/d NOVEMBER 7, 2012"
res = parse(dstr, fuzzy_with_tokens=True)
expected = datetime(2012, 11, 7)
assert res == expected
def test_extraneous_year3(self):
# This was found in the wild at insidertrading.org
dstr = "SMITH R & WEISS D 94 CHILD TR FBO M W SMITH UDT 12/1/1994"
res = parse(dstr, fuzzy_with_tokens=True)
expected = datetime(1994, 12, 1)
assert res == expected
