blob: f5195af90c8a4d56ab6effeddf1465bfed2a18c3 [file] [log] [blame]
Guido van Rossum00efe7e2002-07-19 17:04:46 +00001"""Strptime-related classes and functions.
2
3CLASSES:
Brett Cannon474335c2003-08-05 04:02:49 +00004 LocaleTime -- Discovers and stores locale-specific time information
Barry Warsaw4d895fa2002-09-23 22:46:49 +00005 TimeRE -- Creates regexes for pattern matching a string of text containing
Brett Cannon474335c2003-08-05 04:02:49 +00006 time information
Guido van Rossum00efe7e2002-07-19 17:04:46 +00007
8FUNCTIONS:
Raymond Hettinger1fdb6332003-03-09 07:44:42 +00009 _getlang -- Figure out what language is being used for the locale
Guido van Rossum00efe7e2002-07-19 17:04:46 +000010 strptime -- Calculates the time struct represented by the passed-in string
11
Guido van Rossum00efe7e2002-07-19 17:04:46 +000012"""
13import time
14import locale
15import calendar
16from re import compile as re_compile
Victor Stinner7fa767e2014-03-20 09:16:38 +010017from re import IGNORECASE
Brett Cannon4f35c712004-10-06 02:11:37 +000018from re import escape as re_escape
Alexander Belopolskyca94f552010-06-17 18:30:34 +000019from datetime import (date as datetime_date,
Alexander Belopolskyca94f552010-06-17 18:30:34 +000020 timedelta as datetime_timedelta,
21 timezone as datetime_timezone)
Antoine Pitroua6a4dc82017-09-07 18:56:24 +020022from _thread import allocate_lock as _thread_allocate_lock
Guido van Rossum00efe7e2002-07-19 17:04:46 +000023
Christian Heimesdd15f6c2008-03-16 00:07:10 +000024__all__ = []
Guido van Rossum00efe7e2002-07-19 17:04:46 +000025
Tim Peters80cebc12003-01-19 04:40:44 +000026def _getlang():
27 # Figure out what the current language is set to.
Brett Cannon175ddb52003-07-24 06:27:17 +000028 return locale.getlocale(locale.LC_TIME)
Barry Warsaw35816e62002-08-29 16:24:50 +000029
Guido van Rossum00efe7e2002-07-19 17:04:46 +000030class LocaleTime(object):
31 """Stores and handles locale-specific information related to time.
32
Brett Cannon474335c2003-08-05 04:02:49 +000033 ATTRIBUTES:
Guido van Rossum00efe7e2002-07-19 17:04:46 +000034 f_weekday -- full weekday names (7-item list)
35 a_weekday -- abbreviated weekday names (7-item list)
Brett Cannonf5c96fb2003-08-08 01:53:05 +000036 f_month -- full month names (13-item list; dummy value in [0], which
Guido van Rossum00efe7e2002-07-19 17:04:46 +000037 is added by code)
Brett Cannonf5c96fb2003-08-08 01:53:05 +000038 a_month -- abbreviated month names (13-item list, dummy value in
Guido van Rossum00efe7e2002-07-19 17:04:46 +000039 [0], which is added by code)
40 am_pm -- AM/PM representation (2-item list)
41 LC_date_time -- format string for date/time representation (string)
42 LC_date -- format string for date representation (string)
43 LC_time -- format string for time representation (string)
Tim Peters469cdad2002-08-08 20:19:19 +000044 timezone -- daylight- and non-daylight-savings timezone representation
Brett Cannon474335c2003-08-05 04:02:49 +000045 (2-item list of sets)
46 lang -- Language used by instance (2-item tuple)
Guido van Rossum00efe7e2002-07-19 17:04:46 +000047 """
48
Brett Cannon474335c2003-08-05 04:02:49 +000049 def __init__(self):
50 """Set all attributes.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +000051
Brett Cannon474335c2003-08-05 04:02:49 +000052 Order of methods called matters for dependency reasons.
53
54 The locale language is set at the offset and then checked again before
55 exiting. This is to make sure that the attributes were not set with a
56 mix of information from more than one locale. This would most likely
57 happen when using threads where one thread calls a locale-dependent
58 function while another thread changes the locale while the function in
59 the other thread is still running. Proper coding would call for
60 locks to prevent changing the locale while locale-dependent code is
61 running. The check here is done in case someone does not think about
62 doing this.
Brett Cannon5187a3b2003-08-11 07:24:05 +000063
64 Only other possible issue is if someone changed the timezone and did
65 not call tz.tzset . That is an issue for the programmer, though,
66 since changing the timezone is worthless without that call.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +000067
Brett Cannon474335c2003-08-05 04:02:49 +000068 """
69 self.lang = _getlang()
70 self.__calc_weekday()
71 self.__calc_month()
72 self.__calc_am_pm()
73 self.__calc_timezone()
74 self.__calc_date_time()
75 if _getlang() != self.lang:
76 raise ValueError("locale changed during initialization")
Serhiy Storchakac7217d72015-12-03 22:21:07 +020077 if time.tzname != self.tzname or time.daylight != self.daylight:
78 raise ValueError("timezone changed during initialization")
Guido van Rossum00efe7e2002-07-19 17:04:46 +000079
80 def __pad(self, seq, front):
Brett Cannon474335c2003-08-05 04:02:49 +000081 # Add '' to seq to either the front (is True), else the back.
Guido van Rossum00efe7e2002-07-19 17:04:46 +000082 seq = list(seq)
Barry Warsaw35816e62002-08-29 16:24:50 +000083 if front:
84 seq.insert(0, '')
85 else:
86 seq.append('')
Guido van Rossum00efe7e2002-07-19 17:04:46 +000087 return seq
88
Guido van Rossum00efe7e2002-07-19 17:04:46 +000089 def __calc_weekday(self):
Brett Cannon474335c2003-08-05 04:02:49 +000090 # Set self.a_weekday and self.f_weekday using the calendar
Barry Warsaw35816e62002-08-29 16:24:50 +000091 # module.
Brett Cannon474335c2003-08-05 04:02:49 +000092 a_weekday = [calendar.day_abbr[i].lower() for i in range(7)]
93 f_weekday = [calendar.day_name[i].lower() for i in range(7)]
94 self.a_weekday = a_weekday
95 self.f_weekday = f_weekday
Tim Peters469cdad2002-08-08 20:19:19 +000096
Guido van Rossum00efe7e2002-07-19 17:04:46 +000097 def __calc_month(self):
Brett Cannon474335c2003-08-05 04:02:49 +000098 # Set self.f_month and self.a_month using the calendar module.
99 a_month = [calendar.month_abbr[i].lower() for i in range(13)]
100 f_month = [calendar.month_name[i].lower() for i in range(13)]
101 self.a_month = a_month
102 self.f_month = f_month
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000103
104 def __calc_am_pm(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000105 # Set self.am_pm by using time.strftime().
Tim Peters469cdad2002-08-08 20:19:19 +0000106
Barry Warsaw35816e62002-08-29 16:24:50 +0000107 # The magic date (1999,3,17,hour,44,55,2,76,0) is not really that
108 # magical; just happened to have used it everywhere else where a
109 # static date was needed.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000110 am_pm = []
Guido van Rossumcd16bf62007-06-13 18:07:49 +0000111 for hour in (1, 22):
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000112 time_tuple = time.struct_time((1999,3,17,hour,44,55,2,76,0))
Brett Cannon474335c2003-08-05 04:02:49 +0000113 am_pm.append(time.strftime("%p", time_tuple).lower())
114 self.am_pm = am_pm
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000115
116 def __calc_date_time(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000117 # Set self.date_time, self.date, & self.time by using
Barry Warsaw35816e62002-08-29 16:24:50 +0000118 # time.strftime().
Tim Peters469cdad2002-08-08 20:19:19 +0000119
Barry Warsaw35816e62002-08-29 16:24:50 +0000120 # Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of
121 # overloaded numbers is minimized. The order in which searches for
122 # values within the format string is very important; it eliminates
123 # possible ambiguity for what something represents.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000124 time_tuple = time.struct_time((1999,3,17,22,44,55,2,76,0))
125 date_time = [None, None, None]
Brett Cannon474335c2003-08-05 04:02:49 +0000126 date_time[0] = time.strftime("%c", time_tuple).lower()
127 date_time[1] = time.strftime("%x", time_tuple).lower()
128 date_time[2] = time.strftime("%X", time_tuple).lower()
129 replacement_pairs = [('%', '%%'), (self.f_weekday[2], '%A'),
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000130 (self.f_month[3], '%B'), (self.a_weekday[2], '%a'),
131 (self.a_month[3], '%b'), (self.am_pm[1], '%p'),
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000132 ('1999', '%Y'), ('99', '%y'), ('22', '%H'),
133 ('44', '%M'), ('55', '%S'), ('76', '%j'),
134 ('17', '%d'), ('03', '%m'), ('3', '%m'),
135 # '3' needed for when no leading zero.
Brett Cannon474335c2003-08-05 04:02:49 +0000136 ('2', '%w'), ('10', '%I')]
137 replacement_pairs.extend([(tz, "%Z") for tz_values in self.timezone
138 for tz in tz_values])
139 for offset,directive in ((0,'%c'), (1,'%x'), (2,'%X')):
140 current_format = date_time[offset]
141 for old, new in replacement_pairs:
Jack Jansen62fe7552003-01-15 22:59:39 +0000142 # Must deal with possible lack of locale info
143 # manifesting itself as the empty string (e.g., Swedish's
144 # lack of AM/PM info) or a platform returning a tuple of empty
145 # strings (e.g., MacOS 9 having timezone as ('','')).
146 if old:
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000147 current_format = current_format.replace(old, new)
Brett Cannonf1b2ba62005-08-29 18:25:55 +0000148 # If %W is used, then Sunday, 2005-01-03 will fall on week 0 since
149 # 2005-01-03 occurs before the first Monday of the year. Otherwise
150 # %U is used.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000151 time_tuple = time.struct_time((1999,1,3,1,1,1,6,3,0))
Brett Cannon6e372d12005-08-27 19:25:59 +0000152 if '00' in time.strftime(directive, time_tuple):
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000153 U_W = '%W'
Brett Cannonf1b2ba62005-08-29 18:25:55 +0000154 else:
155 U_W = '%U'
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000156 date_time[offset] = current_format.replace('11', U_W)
Brett Cannon474335c2003-08-05 04:02:49 +0000157 self.LC_date_time = date_time[0]
158 self.LC_date = date_time[1]
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000159 self.LC_time = date_time[2]
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000160
161 def __calc_timezone(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000162 # Set self.timezone by using time.tzname.
Serhiy Storchakac7217d72015-12-03 22:21:07 +0200163 # Do not worry about possibility of time.tzname[0] == time.tzname[1]
164 # and time.daylight; handle that in strptime.
Brett Cannonabe8eb02003-05-13 20:28:15 +0000165 try:
166 time.tzset()
167 except AttributeError:
168 pass
Serhiy Storchakac7217d72015-12-03 22:21:07 +0200169 self.tzname = time.tzname
170 self.daylight = time.daylight
Serhiy Storchakab1f64e72015-12-03 22:26:36 +0200171 no_saving = frozenset({"utc", "gmt", self.tzname[0].lower()})
Serhiy Storchakac7217d72015-12-03 22:21:07 +0200172 if self.daylight:
Serhiy Storchakab1f64e72015-12-03 22:26:36 +0200173 has_saving = frozenset({self.tzname[1].lower()})
Brett Cannon172d9ef2003-05-11 06:23:36 +0000174 else:
Raymond Hettingera690a992003-11-16 16:17:49 +0000175 has_saving = frozenset()
Brett Cannon474335c2003-08-05 04:02:49 +0000176 self.timezone = (no_saving, has_saving)
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000177
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000178
179class TimeRE(dict):
180 """Handle conversion from format directives to regexes."""
181
Brett Cannon2c24d422003-07-24 20:02:28 +0000182 def __init__(self, locale_time=None):
Brett Cannon474335c2003-08-05 04:02:49 +0000183 """Create keys/values.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000184
Brett Cannon474335c2003-08-05 04:02:49 +0000185 Order of execution is important for dependency reasons.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000186
Brett Cannon474335c2003-08-05 04:02:49 +0000187 """
188 if locale_time:
189 self.locale_time = locale_time
190 else:
191 self.locale_time = LocaleTime()
Guido van Rossumcd16bf62007-06-13 18:07:49 +0000192 base = super()
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000193 base.__init__({
Brett Cannon474335c2003-08-05 04:02:49 +0000194 # The " \d" part of the regex is to make %c from ANSI C work
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000195 'd': r"(?P<d>3[0-1]|[1-2]\d|0[1-9]|[1-9]| [1-9])",
Christian Heimesdd15f6c2008-03-16 00:07:10 +0000196 'f': r"(?P<f>[0-9]{1,6})",
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000197 'H': r"(?P<H>2[0-3]|[0-1]\d|\d)",
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000198 'I': r"(?P<I>1[0-2]|0[1-9]|[1-9])",
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400199 'G': r"(?P<G>\d\d\d\d)",
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000200 'j': r"(?P<j>36[0-6]|3[0-5]\d|[1-2]\d\d|0[1-9]\d|00[1-9]|[1-9]\d|0[1-9]|[1-9])",
201 'm': r"(?P<m>1[0-2]|0[1-9]|[1-9])",
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000202 'M': r"(?P<M>[0-5]\d|\d)",
203 'S': r"(?P<S>6[0-1]|[0-5]\d|\d)",
204 'U': r"(?P<U>5[0-3]|[0-4]\d|\d)",
205 'w': r"(?P<w>[0-6])",
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400206 'u': r"(?P<u>[1-7])",
207 'V': r"(?P<V>5[0-3]|0[1-9]|[1-4]\d|\d)",
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000208 # W is set below by using 'U'
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000209 'y': r"(?P<y>\d\d)",
Brett Cannon474335c2003-08-05 04:02:49 +0000210 #XXX: Does 'Y' need to worry about having less or more than
211 # 4 digits?
212 'Y': r"(?P<Y>\d\d\d\d)",
Mario Corchero32318932017-10-26 01:35:41 +0100213 'z': r"(?P<z>[+-]\d\d:?[0-5]\d(:?[0-5]\d(\.\d{1,6})?)?|Z)",
Brett Cannon474335c2003-08-05 04:02:49 +0000214 'A': self.__seqToRE(self.locale_time.f_weekday, 'A'),
215 'a': self.__seqToRE(self.locale_time.a_weekday, 'a'),
216 'B': self.__seqToRE(self.locale_time.f_month[1:], 'B'),
217 'b': self.__seqToRE(self.locale_time.a_month[1:], 'b'),
218 'p': self.__seqToRE(self.locale_time.am_pm, 'p'),
Brett Cannonf7948c22004-10-06 02:23:14 +0000219 'Z': self.__seqToRE((tz for tz_names in self.locale_time.timezone
220 for tz in tz_names),
Brett Cannon474335c2003-08-05 04:02:49 +0000221 'Z'),
222 '%': '%'})
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000223 base.__setitem__('W', base.__getitem__('U').replace('U', 'W'))
Brett Cannon474335c2003-08-05 04:02:49 +0000224 base.__setitem__('c', self.pattern(self.locale_time.LC_date_time))
225 base.__setitem__('x', self.pattern(self.locale_time.LC_date))
226 base.__setitem__('X', self.pattern(self.locale_time.LC_time))
Tim Peters469cdad2002-08-08 20:19:19 +0000227
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000228 def __seqToRE(self, to_convert, directive):
Brett Cannon474335c2003-08-05 04:02:49 +0000229 """Convert a list to a regex string for matching a directive.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000230
Brett Cannon474335c2003-08-05 04:02:49 +0000231 Want possible matching values to be from longest to shortest. This
Ezio Melotti30b9d5d2013-08-17 15:50:46 +0300232 prevents the possibility of a match occurring for a value that also
Brett Cannon474335c2003-08-05 04:02:49 +0000233 a substring of a larger value that should have matched (e.g., 'abc'
234 matching when 'abcdef' should have been the match).
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000235
Brett Cannon474335c2003-08-05 04:02:49 +0000236 """
Brett Cannonffa5cf92004-10-06 22:48:58 +0000237 to_convert = sorted(to_convert, key=len, reverse=True)
Jack Jansen62fe7552003-01-15 22:59:39 +0000238 for value in to_convert:
239 if value != '':
240 break
241 else:
242 return ''
Brett Cannon4f35c712004-10-06 02:11:37 +0000243 regex = '|'.join(re_escape(stuff) for stuff in to_convert)
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000244 regex = '(?P<%s>%s' % (directive, regex)
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000245 return '%s)' % regex
246
247 def pattern(self, format):
Brett Cannon474335c2003-08-05 04:02:49 +0000248 """Return regex pattern for the format string.
Tim Peters0eadaac2003-04-24 16:02:54 +0000249
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000250 Need to make sure that any characters that might be interpreted as
Brett Cannon5187a3b2003-08-11 07:24:05 +0000251 regex syntax are escaped.
Tim Peters0eadaac2003-04-24 16:02:54 +0000252
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000253 """
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000254 processed_format = ''
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000255 # The sub() call escapes all characters that might be misconstrued
Brett Cannon4f35c712004-10-06 02:11:37 +0000256 # as regex syntax. Cannot use re.escape since we have to deal with
257 # format directives (%m, etc.).
Brett Cannon953c6f52003-08-29 02:28:54 +0000258 regex_chars = re_compile(r"([\\.^$*+?\(\){}\[\]|])")
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000259 format = regex_chars.sub(r"\\\1", format)
Serhiy Storchaka15fa1c42015-03-25 01:21:50 +0200260 whitespace_replacement = re_compile(r'\s+')
261 format = whitespace_replacement.sub(r'\\s+', format)
Raymond Hettingerbac788a2004-05-04 09:21:43 +0000262 while '%' in format:
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000263 directive_index = format.index('%')+1
Tim Peters469cdad2002-08-08 20:19:19 +0000264 processed_format = "%s%s%s" % (processed_format,
Barry Warsaw35816e62002-08-29 16:24:50 +0000265 format[:directive_index-1],
266 self[format[directive_index]])
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000267 format = format[directive_index+1:]
268 return "%s%s" % (processed_format, format)
269
270 def compile(self, format):
271 """Return a compiled re object for the format string."""
Brett Cannon7f6b4f82009-03-30 21:30:26 +0000272 return re_compile(self.pattern(format), IGNORECASE)
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000273
Brett Cannon474335c2003-08-05 04:02:49 +0000274_cache_lock = _thread_allocate_lock()
275# DO NOT modify _TimeRE_cache or _regex_cache without acquiring the cache lock
276# first!
277_TimeRE_cache = TimeRE()
Brett Cannon5187a3b2003-08-11 07:24:05 +0000278_CACHE_MAX_SIZE = 5 # Max number of regexes stored in _regex_cache
Brett Cannon474335c2003-08-05 04:02:49 +0000279_regex_cache = {}
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000280
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000281def _calc_julian_from_U_or_W(year, week_of_year, day_of_week, week_starts_Mon):
282 """Calculate the Julian day based on the year, week of the year, and day of
283 the week, with week_start_day representing whether the week of the year
284 assumes the week starts on Sunday or Monday (6 or 0)."""
285 first_weekday = datetime_date(year, 1, 1).weekday()
286 # If we are dealing with the %U directive (week starts on Sunday), it's
287 # easier to just shift the view to Sunday being the first day of the
288 # week.
289 if not week_starts_Mon:
290 first_weekday = (first_weekday + 1) % 7
291 day_of_week = (day_of_week + 1) % 7
292 # Need to watch out for a week 0 (when the first day of the year is not
293 # the same as that specified by %U or %W).
294 week_0_length = (7 - first_weekday) % 7
295 if week_of_year == 0:
296 return 1 + day_of_week - first_weekday
297 else:
298 days_to_week = week_0_length + (7 * (week_of_year - 1))
299 return 1 + days_to_week + day_of_week
300
301
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400302def _calc_julian_from_V(iso_year, iso_week, iso_weekday):
303 """Calculate the Julian day based on the ISO 8601 year, week, and weekday.
304 ISO weeks start on Mondays, with week 01 being the week containing 4 Jan.
305 ISO week days range from 1 (Monday) to 7 (Sunday).
306 """
307 correction = datetime_date(iso_year, 1, 4).isoweekday() + 3
308 ordinal = (iso_week * 7) + iso_weekday - correction
309 # ordinal may be negative or 0 now, which means the date is in the previous
310 # calendar year
311 if ordinal < 1:
312 ordinal += datetime_date(iso_year, 1, 1).toordinal()
313 iso_year -= 1
314 ordinal -= datetime_date(iso_year, 1, 1).toordinal()
315 return iso_year, ordinal
316
317
Christian Heimesdd15f6c2008-03-16 00:07:10 +0000318def _strptime(data_string, format="%a %b %d %H:%M:%S %Y"):
Alexander Belopolskyf5682182010-06-18 18:44:37 +0000319 """Return a 2-tuple consisting of a time struct and an int containing
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000320 the number of microseconds based on the input string and the
321 format string."""
Brett Cannon7f6b4f82009-03-30 21:30:26 +0000322
323 for index, arg in enumerate([data_string, format]):
324 if not isinstance(arg, str):
325 msg = "strptime() argument {} must be str, not {}"
Brett Cannon71095ea2009-03-31 03:58:04 +0000326 raise TypeError(msg.format(index, type(arg)))
Brett Cannon7f6b4f82009-03-30 21:30:26 +0000327
Brett Cannona783d062005-09-15 02:34:56 +0000328 global _TimeRE_cache, _regex_cache
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000329 with _cache_lock:
Serhiy Storchakac7217d72015-12-03 22:21:07 +0200330 locale_time = _TimeRE_cache.locale_time
331 if (_getlang() != locale_time.lang or
332 time.tzname != locale_time.tzname or
333 time.daylight != locale_time.daylight):
Brett Cannon474335c2003-08-05 04:02:49 +0000334 _TimeRE_cache = TimeRE()
Guido van Rossumd8faa362007-04-27 19:54:29 +0000335 _regex_cache.clear()
Serhiy Storchakac7217d72015-12-03 22:21:07 +0200336 locale_time = _TimeRE_cache.locale_time
Brett Cannon474335c2003-08-05 04:02:49 +0000337 if len(_regex_cache) > _CACHE_MAX_SIZE:
338 _regex_cache.clear()
339 format_regex = _regex_cache.get(format)
340 if not format_regex:
Brett Cannon5d0bf942005-11-02 23:04:26 +0000341 try:
Guido van Rossumd8faa362007-04-27 19:54:29 +0000342 format_regex = _TimeRE_cache.compile(format)
Brett Cannon5d0bf942005-11-02 23:04:26 +0000343 # KeyError raised when a bad format is found; can be specified as
344 # \\, in which case it was a stray % but with a space after it
Guido van Rossumb940e112007-01-10 16:19:56 +0000345 except KeyError as err:
Brett Cannon5d0bf942005-11-02 23:04:26 +0000346 bad_directive = err.args[0]
347 if bad_directive == "\\":
348 bad_directive = "%"
349 del err
350 raise ValueError("'%s' is a bad directive in format '%s'" %
Ezio Melotti0f389082013-04-04 02:09:20 +0300351 (bad_directive, format)) from None
Brett Cannon5d0bf942005-11-02 23:04:26 +0000352 # IndexError only occurs when the format string is "%"
353 except IndexError:
Serhiy Storchakacdac3022013-11-24 18:15:37 +0200354 raise ValueError("stray %% in format '%s'" % format) from None
Brett Cannon474335c2003-08-05 04:02:49 +0000355 _regex_cache[format] = format_regex
Tim Peters80cebc12003-01-19 04:40:44 +0000356 found = format_regex.match(data_string)
Tim Peters08e54272003-01-18 03:53:49 +0000357 if not found:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000358 raise ValueError("time data %r does not match format %r" %
Raymond Hettinger4a6302b2003-07-13 01:31:38 +0000359 (data_string, format))
Brett Cannon2b6dfec2003-04-28 21:30:13 +0000360 if len(data_string) != found.end():
361 raise ValueError("unconverted data remains: %s" %
362 data_string[found.end():])
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000363
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400364 iso_year = year = None
Tim Peters08e54272003-01-18 03:53:49 +0000365 month = day = 1
Christian Heimesdd15f6c2008-03-16 00:07:10 +0000366 hour = minute = second = fraction = 0
Tim Peters08e54272003-01-18 03:53:49 +0000367 tz = -1
Mario Corchero32318932017-10-26 01:35:41 +0100368 gmtoff = None
369 gmtoff_fraction = 0
Brett Cannon8dc25ad2004-10-18 01:47:46 +0000370 # Default to -1 to signify that values not known; not critical to have,
371 # though
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400372 iso_week = week_of_year = None
373 week_of_year_start = None
Serhiy Storchaka423feea2015-03-19 19:13:37 +0200374 # weekday and julian defaulted to None so as to signal need to calculate
Brett Cannon8dc25ad2004-10-18 01:47:46 +0000375 # values
Serhiy Storchaka423feea2015-03-19 19:13:37 +0200376 weekday = julian = None
Tim Peters08e54272003-01-18 03:53:49 +0000377 found_dict = found.groupdict()
Guido van Rossumcc2b0162007-02-11 06:12:03 +0000378 for group_key in found_dict.keys():
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000379 # Directives not explicitly handled below:
380 # c, x, X
381 # handled by making out of other directives
382 # U, W
383 # worthless without day of the week
Tim Peters08e54272003-01-18 03:53:49 +0000384 if group_key == 'y':
385 year = int(found_dict['y'])
386 # Open Group specification for strptime() states that a %y
387 #value in the range of [00, 68] is in the century 2000, while
388 #[69,99] is in the century 1900
389 if year <= 68:
390 year += 2000
391 else:
392 year += 1900
393 elif group_key == 'Y':
394 year = int(found_dict['Y'])
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400395 elif group_key == 'G':
396 iso_year = int(found_dict['G'])
Tim Peters08e54272003-01-18 03:53:49 +0000397 elif group_key == 'm':
398 month = int(found_dict['m'])
399 elif group_key == 'B':
Brett Cannon474335c2003-08-05 04:02:49 +0000400 month = locale_time.f_month.index(found_dict['B'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000401 elif group_key == 'b':
Brett Cannon474335c2003-08-05 04:02:49 +0000402 month = locale_time.a_month.index(found_dict['b'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000403 elif group_key == 'd':
404 day = int(found_dict['d'])
Neal Norwitz77290f22003-06-29 04:16:49 +0000405 elif group_key == 'H':
Tim Peters08e54272003-01-18 03:53:49 +0000406 hour = int(found_dict['H'])
407 elif group_key == 'I':
408 hour = int(found_dict['I'])
409 ampm = found_dict.get('p', '').lower()
410 # If there was no AM/PM indicator, we'll treat this like AM
Brett Cannon474335c2003-08-05 04:02:49 +0000411 if ampm in ('', locale_time.am_pm[0]):
Tim Peters08e54272003-01-18 03:53:49 +0000412 # We're in AM so the hour is correct unless we're
413 # looking at 12 midnight.
414 # 12 midnight == 12 AM == hour 0
415 if hour == 12:
416 hour = 0
Brett Cannon474335c2003-08-05 04:02:49 +0000417 elif ampm == locale_time.am_pm[1]:
Tim Peters08e54272003-01-18 03:53:49 +0000418 # We're in PM so we need to add 12 to the hour unless
419 # we're looking at 12 noon.
420 # 12 noon == 12 PM == hour 12
421 if hour != 12:
422 hour += 12
423 elif group_key == 'M':
424 minute = int(found_dict['M'])
425 elif group_key == 'S':
426 second = int(found_dict['S'])
Christian Heimesdd15f6c2008-03-16 00:07:10 +0000427 elif group_key == 'f':
428 s = found_dict['f']
429 # Pad to always return microseconds.
430 s += "0" * (6 - len(s))
431 fraction = int(s)
Tim Peters08e54272003-01-18 03:53:49 +0000432 elif group_key == 'A':
Brett Cannon474335c2003-08-05 04:02:49 +0000433 weekday = locale_time.f_weekday.index(found_dict['A'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000434 elif group_key == 'a':
Brett Cannon474335c2003-08-05 04:02:49 +0000435 weekday = locale_time.a_weekday.index(found_dict['a'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000436 elif group_key == 'w':
437 weekday = int(found_dict['w'])
438 if weekday == 0:
439 weekday = 6
440 else:
441 weekday -= 1
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400442 elif group_key == 'u':
443 weekday = int(found_dict['u'])
444 weekday -= 1
Tim Peters08e54272003-01-18 03:53:49 +0000445 elif group_key == 'j':
446 julian = int(found_dict['j'])
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000447 elif group_key in ('U', 'W'):
448 week_of_year = int(found_dict[group_key])
449 if group_key == 'U':
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000450 # U starts week on Sunday.
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000451 week_of_year_start = 6
452 else:
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000453 # W starts week on Monday.
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000454 week_of_year_start = 0
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400455 elif group_key == 'V':
456 iso_week = int(found_dict['V'])
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000457 elif group_key == 'z':
458 z = found_dict['z']
Mario Corchero32318932017-10-26 01:35:41 +0100459 if z == 'Z':
460 gmtoff = 0
461 else:
462 if z[3] == ':':
463 z = z[:3] + z[4:]
464 if len(z) > 5:
465 if z[5] != ':':
466 msg = f"Unconsistent use of : in {found_dict['z']}"
467 raise ValueError(msg)
468 z = z[:5] + z[6:]
469 hours = int(z[1:3])
470 minutes = int(z[3:5])
471 seconds = int(z[5:7] or 0)
472 gmtoff = (hours * 60 * 60) + (minutes * 60) + seconds
473 gmtoff_fraction = int(z[8:] or 0)
474 if z.startswith("-"):
475 gmtoff = -gmtoff
476 gmtoff_fraction = -gmtoff_fraction
Tim Peters08e54272003-01-18 03:53:49 +0000477 elif group_key == 'Z':
Brett Cannon172d9ef2003-05-11 06:23:36 +0000478 # Since -1 is default value only need to worry about setting tz if
479 # it can be something other than -1.
Tim Peters08e54272003-01-18 03:53:49 +0000480 found_zone = found_dict['Z'].lower()
Brett Cannon5187a3b2003-08-11 07:24:05 +0000481 for value, tz_values in enumerate(locale_time.timezone):
482 if found_zone in tz_values:
483 # Deal with bad locale setup where timezone names are the
484 # same and yet time.daylight is true; too ambiguous to
485 # be able to tell what timezone has daylight savings
Brett Cannon8172ac32004-03-07 23:16:27 +0000486 if (time.tzname[0] == time.tzname[1] and
487 time.daylight and found_zone not in ("utc", "gmt")):
Tim Peters58eb11c2004-01-18 20:29:55 +0000488 break
Brett Cannon5187a3b2003-08-11 07:24:05 +0000489 else:
Brett Cannon474335c2003-08-05 04:02:49 +0000490 tz = value
Brett Cannon5187a3b2003-08-11 07:24:05 +0000491 break
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400492 # Deal with the cases where ambiguities arize
493 # don't assume default values for ISO week/year
494 if year is None and iso_year is not None:
495 if iso_week is None or weekday is None:
496 raise ValueError("ISO year directive '%G' must be used with "
497 "the ISO week directive '%V' and a weekday "
498 "directive ('%A', '%a', '%w', or '%u').")
499 if julian is not None:
500 raise ValueError("Day of the year directive '%j' is not "
501 "compatible with ISO year directive '%G'. "
502 "Use '%Y' instead.")
503 elif week_of_year is None and iso_week is not None:
504 if weekday is None:
505 raise ValueError("ISO week directive '%V' must be used with "
506 "the ISO year directive '%G' and a weekday "
507 "directive ('%A', '%a', '%w', or '%u').")
508 else:
509 raise ValueError("ISO week directive '%V' is incompatible with "
510 "the year directive '%Y'. Use the ISO year '%G' "
511 "instead.")
512
Antoine Pitrou072e4a32012-05-14 19:44:59 +0200513 leap_year_fix = False
Antoine Pitrou1682e5d2012-05-10 20:17:46 +0200514 if year is None and month == 2 and day == 29:
515 year = 1904 # 1904 is first leap year of 20th century
Antoine Pitrou072e4a32012-05-14 19:44:59 +0200516 leap_year_fix = True
Antoine Pitrou1682e5d2012-05-10 20:17:46 +0200517 elif year is None:
518 year = 1900
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400519
520
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000521 # If we know the week of the year and what day of that week, we can figure
Thomas Wouters9fe394c2007-02-05 01:24:16 +0000522 # out the Julian day of the year.
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400523 if julian is None and weekday is not None:
524 if week_of_year is not None:
525 week_starts_Mon = True if week_of_year_start == 0 else False
526 julian = _calc_julian_from_U_or_W(year, week_of_year, weekday,
527 week_starts_Mon)
528 elif iso_year is not None and iso_week is not None:
529 year, julian = _calc_julian_from_V(iso_year, iso_week, weekday + 1)
Serhiy Storchaka6e4150f2016-03-12 10:53:09 +0200530 if julian is not None and julian <= 0:
Serhiy Storchaka8a7240e2016-03-12 10:51:16 +0200531 year -= 1
532 yday = 366 if calendar.isleap(year) else 365
533 julian += yday
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400534
Serhiy Storchaka423feea2015-03-19 19:13:37 +0200535 if julian is None:
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400536 # Cannot pre-calculate datetime_date() since can change in Julian
537 # calculation and thus could have different value for the day of
538 # the week calculation.
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000539 # Need to add 1 to result since first day of the year is 1, not 0.
540 julian = datetime_date(year, month, day).toordinal() - \
541 datetime_date(year, 1, 1).toordinal() + 1
Alexander Belopolsky68713e42015-10-06 13:29:56 -0400542 else: # Assume that if they bothered to include Julian day (or if it was
543 # calculated above with year/week/weekday) it will be accurate.
544 datetime_result = datetime_date.fromordinal(
545 (julian - 1) +
546 datetime_date(year, 1, 1).toordinal())
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000547 year = datetime_result.year
548 month = datetime_result.month
549 day = datetime_result.day
Serhiy Storchaka423feea2015-03-19 19:13:37 +0200550 if weekday is None:
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000551 weekday = datetime_date(year, month, day).weekday()
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000552 # Add timezone info
553 tzname = found_dict.get("Z")
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000554
Antoine Pitrou072e4a32012-05-14 19:44:59 +0200555 if leap_year_fix:
556 # the caller didn't supply a year but asked for Feb 29th. We couldn't
557 # use the default of 1900 for computations. We set it back to ensure
558 # that February 29th is smaller than March 1st.
559 year = 1900
560
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000561 return (year, month, day,
562 hour, minute, second,
Mario Corchero32318932017-10-26 01:35:41 +0100563 weekday, julian, tz, tzname, gmtoff), fraction, gmtoff_fraction
Christian Heimesdd15f6c2008-03-16 00:07:10 +0000564
565def _strptime_time(data_string, format="%a %b %d %H:%M:%S %Y"):
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000566 """Return a time struct based on the input string and the
567 format string."""
568 tt = _strptime(data_string, format)[0]
Alexander Belopolskyc142bba2012-06-13 22:15:26 -0400569 return time.struct_time(tt[:time._STRUCT_TM_ITEMS])
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000570
Alexander Belopolsky4988d7a2010-07-14 13:46:57 +0000571def _strptime_datetime(cls, data_string, format="%a %b %d %H:%M:%S %Y"):
572 """Return a class cls instance based on the input string and the
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000573 format string."""
Mario Corchero32318932017-10-26 01:35:41 +0100574 tt, fraction, gmtoff_fraction = _strptime(data_string, format)
Alexander Belopolskyc142bba2012-06-13 22:15:26 -0400575 tzname, gmtoff = tt[-2:]
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000576 args = tt[:6] + (fraction,)
577 if gmtoff is not None:
Mario Corchero32318932017-10-26 01:35:41 +0100578 tzdelta = datetime_timedelta(seconds=gmtoff, microseconds=gmtoff_fraction)
Alexander Belopolskyca94f552010-06-17 18:30:34 +0000579 if tzname:
580 tz = datetime_timezone(tzdelta, tzname)
581 else:
582 tz = datetime_timezone(tzdelta)
583 args += (tz,)
584
Alexander Belopolsky4988d7a2010-07-14 13:46:57 +0000585 return cls(*args)