Merge pull request #1740 from sebres/0.10-strptime-perf

strptime.py: small code review and performance optimization
pull/1742/head
Serg G. Brester 2017-03-29 11:33:57 +02:00 committed by GitHub
commit 1a59a5c5a7
2 changed files with 85 additions and 60 deletions

View File

@ -95,18 +95,10 @@ def reGroupDictStrptime(found_dict, msec=False):
Unix time stamp. Unix time stamp.
""" """
now = MyTime.now() now = \
year = month = day = hour = minute = None year = month = day = hour = minute = tzoffset = \
hour = minute = None weekday = julian = week_of_year = None
second = fraction = 0 second = fraction = 0
tzoffset = None
# Default to -1 to signify that values not known; not critical to have,
# though
week_of_year = -1
week_of_year_start = -1
# weekday and julian defaulted to -1 so as to signal need to calculate
# values
weekday = julian = -1
for key, val in found_dict.iteritems(): for key, val in found_dict.iteritems():
if val is None: continue if val is None: continue
# Directives not explicitly handled below: # Directives not explicitly handled below:
@ -116,13 +108,9 @@ def reGroupDictStrptime(found_dict, msec=False):
# worthless without day of the week # worthless without day of the week
if key == 'y': if key == 'y':
year = int(val) year = int(val)
# Open Group specification for strptime() states that a %y # Fail2ban year should be always in the current century (>= 2000)
#value in the range of [00, 68] is in the century 2000, while if year <= 2000:
#[69,99] is in the century 1900
if year <= 68:
year += 2000 year += 2000
else:
year += 1900
elif key == 'Y': elif key == 'Y':
year = int(val) year = int(val)
elif key == 'm': elif key == 'm':
@ -156,7 +144,7 @@ def reGroupDictStrptime(found_dict, msec=False):
elif key == 'S': elif key == 'S':
second = int(val) second = int(val)
elif key == 'f': elif key == 'f':
if msec: if msec: # pragma: no cover - currently unused
s = val s = val
# Pad to always return microseconds. # Pad to always return microseconds.
s += "0" * (6 - len(s)) s += "0" * (6 - len(s))
@ -166,21 +154,14 @@ def reGroupDictStrptime(found_dict, msec=False):
elif key == 'a': elif key == 'a':
weekday = locale_time.a_weekday.index(val.lower()) weekday = locale_time.a_weekday.index(val.lower())
elif key == 'w': elif key == 'w':
weekday = int(val) weekday = int(val) - 1
if weekday == 0: if weekday < 0: weekday = 6
weekday = 6
else:
weekday -= 1
elif key == 'j': elif key == 'j':
julian = int(val) julian = int(val)
elif key in ('U', 'W'): elif key in ('U', 'W'):
week_of_year = int(val) week_of_year = int(val)
if key == 'U': # U starts week on Sunday, W - on Monday
# U starts week on Sunday. week_of_year_start = 6 if key == 'U' else 0
week_of_year_start = 6
else:
# W starts week on Monday.
week_of_year_start = 0
elif key == 'z': elif key == 'z':
z = val z = val
if z in ("Z", "UTC", "GMT"): if z in ("Z", "UTC", "GMT"):
@ -199,31 +180,28 @@ def reGroupDictStrptime(found_dict, msec=False):
# Fail2Ban will assume it's this year # Fail2Ban will assume it's this year
assume_year = False assume_year = False
if year is None: if year is None:
if not now: now = MyTime.now()
year = now.year year = now.year
assume_year = True assume_year = True
# If we know the week of the year and what day of that week, we can figure if month is None or day is None:
# out the Julian day of the year. # If we know the week of the year and what day of that week, we can figure
if julian == -1 and week_of_year != -1 and weekday != -1: # out the Julian day of the year.
week_starts_Mon = True if week_of_year_start == 0 else False if julian is None and week_of_year is not None and weekday is not None:
julian = _calc_julian_from_U_or_W(year, week_of_year, weekday, julian = _calc_julian_from_U_or_W(year, week_of_year, weekday,
week_starts_Mon) (week_of_year_start == 0))
# Cannot pre-calculate datetime.datetime() since can change in Julian # Cannot pre-calculate datetime.datetime() since can change in Julian
# calculation and thus could have different value for the day of the week # calculation and thus could have different value for the day of the week
# calculation. # calculation.
if julian != -1 and (month is None or day is None): if julian is not None:
datetime_result = datetime.datetime.fromordinal((julian - 1) + datetime.datetime(year, 1, 1).toordinal()) datetime_result = datetime.datetime.fromordinal((julian - 1) + datetime.datetime(year, 1, 1).toordinal())
year = datetime_result.year year = datetime_result.year
month = datetime_result.month month = datetime_result.month
day = datetime_result.day day = datetime_result.day
# Add timezone info
if tzoffset is not None:
gmtoff = tzoffset * 60
else:
gmtoff = None
# Fail2Ban assume today # Fail2Ban assume today
assume_today = False assume_today = False
if month is None and day is None: if month is None and day is None:
if not now: now = MyTime.now()
month = now.month month = now.month
day = now.day day = now.day
assume_today = True assume_today = True
@ -231,22 +209,28 @@ def reGroupDictStrptime(found_dict, msec=False):
# Actully create date # Actully create date
date_result = datetime.datetime( date_result = datetime.datetime(
year, month, day, hour, minute, second, fraction) year, month, day, hour, minute, second, fraction)
if gmtoff is not None: # Add timezone info
date_result = date_result - datetime.timedelta(seconds=gmtoff) if tzoffset is not None:
date_result -= datetime.timedelta(seconds=tzoffset * 60)
if date_result > now and assume_today: if assume_today:
# Rollover at midnight, could mean it's yesterday... if not now: now = MyTime.now()
date_result = date_result - datetime.timedelta(days=1) if date_result > now:
if date_result > now and assume_year: # Rollover at midnight, could mean it's yesterday...
# Could be last year? date_result -= datetime.timedelta(days=1)
# also reset month and day as it's not yesterday... if assume_year:
date_result = date_result.replace( if not now: now = MyTime.now()
year=year-1, month=month, day=day) if date_result > now:
# Could be last year?
# also reset month and day as it's not yesterday...
date_result = date_result.replace(
year=year-1, month=month, day=day)
if gmtoff is not None: # make time:
if tzoffset is not None:
tm = calendar.timegm(date_result.utctimetuple()) tm = calendar.timegm(date_result.utctimetuple())
else: else:
tm = time.mktime(date_result.timetuple()) tm = time.mktime(date_result.timetuple())
if msec: if msec: # pragma: no cover - currently unused
tm += fraction/1000000.0 tm += fraction/1000000.0
return tm return tm

View File

@ -298,6 +298,16 @@ iso8601 = DatePatternRegex("%Y-%m-%d[T ]%H:%M:%S(?:\.%f)?%z")
class CustomDateFormatsTest(unittest.TestCase): class CustomDateFormatsTest(unittest.TestCase):
def setUp(self):
"""Call before every test case."""
unittest.TestCase.setUp(self)
setUpMyTime()
def tearDown(self):
"""Call after every test case."""
unittest.TestCase.tearDown(self)
tearDownMyTime()
def testIso8601(self): def testIso8601(self):
date = datetime.datetime.utcfromtimestamp( date = datetime.datetime.utcfromtimestamp(
iso8601.getDate("2007-01-25T12:00:00Z")[0]) iso8601.getDate("2007-01-25T12:00:00Z")[0])
@ -411,6 +421,37 @@ class CustomDateFormatsTest(unittest.TestCase):
else: else:
self.assertEqual(date, None) self.assertEqual(date, None)
def testVariousFormatSpecs(self):
for (matched, dp, line) in (
# cover %B (full-month-name) and %I (as 12 == 0):
(1106438399.0, "^%B %Exd %I:%ExM:%ExS**", 'January 23 12:59:59'),
# cover %U (week of year starts on sunday) and %A (weekday):
(985208399.0, "^%y %U %A %ExH:%ExM:%ExS**", '01 11 Wednesday 21:59:59'),
# cover %W (week of year starts on monday) and %A (weekday):
(984603599.0, "^%y %W %A %ExH:%ExM:%ExS**", '01 11 Wednesday 21:59:59'),
# cover %W (week of year starts on monday) and %w (weekday, 0 - sunday):
(984949199.0, "^%y %W %w %ExH:%ExM:%ExS**", '01 11 0 21:59:59'),
# cover %W (week of year starts on monday) and %w (weekday, 6 - saturday):
(984862799.0, "^%y %W %w %ExH:%ExM:%ExS**", '01 11 6 21:59:59'),
# cover time only, current date, in test cases now == 14 Aug 2005 12:00 -> back to yesterday (13 Aug):
(1123963199.0, "^%ExH:%ExM:%ExS**", '21:59:59'),
# cover time only, current date, in test cases now == 14 Aug 2005 12:00 -> today (14 Aug):
(1123970401.0, "^%ExH:%ExM:%ExS**", '00:00:01'),
# cover date with current year, in test cases now == Aug 2005 -> back to last year (Sep 2004):
(1094068799.0, "^%m/%d %ExH:%ExM:%ExS**", '09/01 21:59:59'),
):
logSys.debug('== test: %r', (matched, dp, line))
dd = DateDetector()
dd.appendTemplate(dp)
date = dd.getTime(line)
if matched:
self.assertTrue(date)
if isinstance(matched, basestring): # pragma: no cover
self.assertEqual(matched, date[1].group(1))
else:
self.assertEqual(matched, date[0])
else: # pragma: no cover
self.assertEqual(date, None)
# def testDefaultTempate(self): # def testDefaultTempate(self):
# self.__datedetector.setDefaultRegex("^\S{3}\s{1,2}\d{1,2} \d{2}:\d{2}:\d{2}") # self.__datedetector.setDefaultRegex("^\S{3}\s{1,2}\d{1,2} \d{2}:\d{2}:\d{2}")