blob: ce8525b7169ba39e1e1f981d7ccd738fe947dd57 [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
17from re import IGNORECASE
Brett Cannon4f35c712004-10-06 02:11:37 +000018from re import escape as re_escape
Raymond Hettinger1fdb6332003-03-09 07:44:42 +000019from datetime import date as datetime_date
Brett Cannon474335c2003-08-05 04:02:49 +000020try:
21 from thread import allocate_lock as _thread_allocate_lock
22except:
23 from dummy_thread import allocate_lock as _thread_allocate_lock
Guido van Rossum00efe7e2002-07-19 17:04:46 +000024
Guido van Rossum00efe7e2002-07-19 17:04:46 +000025__author__ = "Brett Cannon"
Raymond Hettinger1fdb6332003-03-09 07:44:42 +000026__email__ = "brett@python.org"
Guido van Rossum00efe7e2002-07-19 17:04:46 +000027
28__all__ = ['strptime']
29
Tim Peters80cebc12003-01-19 04:40:44 +000030def _getlang():
31 # Figure out what the current language is set to.
Brett Cannon175ddb52003-07-24 06:27:17 +000032 return locale.getlocale(locale.LC_TIME)
Barry Warsaw35816e62002-08-29 16:24:50 +000033
Guido van Rossum00efe7e2002-07-19 17:04:46 +000034class LocaleTime(object):
35 """Stores and handles locale-specific information related to time.
36
Brett Cannon474335c2003-08-05 04:02:49 +000037 ATTRIBUTES:
Guido van Rossum00efe7e2002-07-19 17:04:46 +000038 f_weekday -- full weekday names (7-item list)
39 a_weekday -- abbreviated weekday names (7-item list)
Brett Cannonf5c96fb2003-08-08 01:53:05 +000040 f_month -- full month names (13-item list; dummy value in [0], which
Guido van Rossum00efe7e2002-07-19 17:04:46 +000041 is added by code)
Brett Cannonf5c96fb2003-08-08 01:53:05 +000042 a_month -- abbreviated month names (13-item list, dummy value in
Guido van Rossum00efe7e2002-07-19 17:04:46 +000043 [0], which is added by code)
44 am_pm -- AM/PM representation (2-item list)
45 LC_date_time -- format string for date/time representation (string)
46 LC_date -- format string for date representation (string)
47 LC_time -- format string for time representation (string)
Tim Peters469cdad2002-08-08 20:19:19 +000048 timezone -- daylight- and non-daylight-savings timezone representation
Brett Cannon474335c2003-08-05 04:02:49 +000049 (2-item list of sets)
50 lang -- Language used by instance (2-item tuple)
Guido van Rossum00efe7e2002-07-19 17:04:46 +000051 """
52
Brett Cannon474335c2003-08-05 04:02:49 +000053 def __init__(self):
54 """Set all attributes.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +000055
Brett Cannon474335c2003-08-05 04:02:49 +000056 Order of methods called matters for dependency reasons.
57
58 The locale language is set at the offset and then checked again before
59 exiting. This is to make sure that the attributes were not set with a
60 mix of information from more than one locale. This would most likely
61 happen when using threads where one thread calls a locale-dependent
62 function while another thread changes the locale while the function in
63 the other thread is still running. Proper coding would call for
64 locks to prevent changing the locale while locale-dependent code is
65 running. The check here is done in case someone does not think about
66 doing this.
Brett Cannon5187a3b2003-08-11 07:24:05 +000067
68 Only other possible issue is if someone changed the timezone and did
69 not call tz.tzset . That is an issue for the programmer, though,
70 since changing the timezone is worthless without that call.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +000071
Brett Cannon474335c2003-08-05 04:02:49 +000072 """
73 self.lang = _getlang()
74 self.__calc_weekday()
75 self.__calc_month()
76 self.__calc_am_pm()
77 self.__calc_timezone()
78 self.__calc_date_time()
79 if _getlang() != self.lang:
80 raise ValueError("locale changed during initialization")
Guido van Rossum00efe7e2002-07-19 17:04:46 +000081
82 def __pad(self, seq, front):
Brett Cannon474335c2003-08-05 04:02:49 +000083 # Add '' to seq to either the front (is True), else the back.
Guido van Rossum00efe7e2002-07-19 17:04:46 +000084 seq = list(seq)
Barry Warsaw35816e62002-08-29 16:24:50 +000085 if front:
86 seq.insert(0, '')
87 else:
88 seq.append('')
Guido van Rossum00efe7e2002-07-19 17:04:46 +000089 return seq
90
Guido van Rossum00efe7e2002-07-19 17:04:46 +000091 def __calc_weekday(self):
Brett Cannon474335c2003-08-05 04:02:49 +000092 # Set self.a_weekday and self.f_weekday using the calendar
Barry Warsaw35816e62002-08-29 16:24:50 +000093 # module.
Brett Cannon474335c2003-08-05 04:02:49 +000094 a_weekday = [calendar.day_abbr[i].lower() for i in range(7)]
95 f_weekday = [calendar.day_name[i].lower() for i in range(7)]
96 self.a_weekday = a_weekday
97 self.f_weekday = f_weekday
Tim Peters469cdad2002-08-08 20:19:19 +000098
Guido van Rossum00efe7e2002-07-19 17:04:46 +000099 def __calc_month(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000100 # Set self.f_month and self.a_month using the calendar module.
101 a_month = [calendar.month_abbr[i].lower() for i in range(13)]
102 f_month = [calendar.month_name[i].lower() for i in range(13)]
103 self.a_month = a_month
104 self.f_month = f_month
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000105
106 def __calc_am_pm(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000107 # Set self.am_pm by using time.strftime().
Tim Peters469cdad2002-08-08 20:19:19 +0000108
Barry Warsaw35816e62002-08-29 16:24:50 +0000109 # The magic date (1999,3,17,hour,44,55,2,76,0) is not really that
110 # magical; just happened to have used it everywhere else where a
111 # static date was needed.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000112 am_pm = []
113 for hour in (01,22):
114 time_tuple = time.struct_time((1999,3,17,hour,44,55,2,76,0))
Brett Cannon474335c2003-08-05 04:02:49 +0000115 am_pm.append(time.strftime("%p", time_tuple).lower())
116 self.am_pm = am_pm
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000117
118 def __calc_date_time(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000119 # Set self.date_time, self.date, & self.time by using
Barry Warsaw35816e62002-08-29 16:24:50 +0000120 # time.strftime().
Tim Peters469cdad2002-08-08 20:19:19 +0000121
Barry Warsaw35816e62002-08-29 16:24:50 +0000122 # Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of
123 # overloaded numbers is minimized. The order in which searches for
124 # values within the format string is very important; it eliminates
125 # possible ambiguity for what something represents.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000126 time_tuple = time.struct_time((1999,3,17,22,44,55,2,76,0))
127 date_time = [None, None, None]
Brett Cannon474335c2003-08-05 04:02:49 +0000128 date_time[0] = time.strftime("%c", time_tuple).lower()
129 date_time[1] = time.strftime("%x", time_tuple).lower()
130 date_time[2] = time.strftime("%X", time_tuple).lower()
131 replacement_pairs = [('%', '%%'), (self.f_weekday[2], '%A'),
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000132 (self.f_month[3], '%B'), (self.a_weekday[2], '%a'),
133 (self.a_month[3], '%b'), (self.am_pm[1], '%p'),
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000134 ('1999', '%Y'), ('99', '%y'), ('22', '%H'),
135 ('44', '%M'), ('55', '%S'), ('76', '%j'),
136 ('17', '%d'), ('03', '%m'), ('3', '%m'),
137 # '3' needed for when no leading zero.
Brett Cannon474335c2003-08-05 04:02:49 +0000138 ('2', '%w'), ('10', '%I')]
139 replacement_pairs.extend([(tz, "%Z") for tz_values in self.timezone
140 for tz in tz_values])
141 for offset,directive in ((0,'%c'), (1,'%x'), (2,'%X')):
142 current_format = date_time[offset]
143 for old, new in replacement_pairs:
Jack Jansen62fe7552003-01-15 22:59:39 +0000144 # Must deal with possible lack of locale info
145 # manifesting itself as the empty string (e.g., Swedish's
146 # lack of AM/PM info) or a platform returning a tuple of empty
147 # strings (e.g., MacOS 9 having timezone as ('','')).
148 if old:
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000149 current_format = current_format.replace(old, new)
Brett Cannonf1b2ba62005-08-29 18:25:55 +0000150 # If %W is used, then Sunday, 2005-01-03 will fall on week 0 since
151 # 2005-01-03 occurs before the first Monday of the year. Otherwise
152 # %U is used.
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000153 time_tuple = time.struct_time((1999,1,3,1,1,1,6,3,0))
Brett Cannon6e372d12005-08-27 19:25:59 +0000154 if '00' in time.strftime(directive, time_tuple):
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000155 U_W = '%W'
Brett Cannonf1b2ba62005-08-29 18:25:55 +0000156 else:
157 U_W = '%U'
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000158 date_time[offset] = current_format.replace('11', U_W)
Brett Cannon474335c2003-08-05 04:02:49 +0000159 self.LC_date_time = date_time[0]
160 self.LC_date = date_time[1]
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000161 self.LC_time = date_time[2]
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000162
163 def __calc_timezone(self):
Brett Cannon474335c2003-08-05 04:02:49 +0000164 # Set self.timezone by using time.tzname.
Brett Cannon5187a3b2003-08-11 07:24:05 +0000165 # Do not worry about possibility of time.tzname[0] == timetzname[1]
166 # and time.daylight; handle that in strptime .
Brett Cannonabe8eb02003-05-13 20:28:15 +0000167 try:
168 time.tzset()
169 except AttributeError:
170 pass
Raymond Hettingera690a992003-11-16 16:17:49 +0000171 no_saving = frozenset(["utc", "gmt", time.tzname[0].lower()])
Brett Cannon172d9ef2003-05-11 06:23:36 +0000172 if time.daylight:
Raymond Hettingera690a992003-11-16 16:17:49 +0000173 has_saving = frozenset([time.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()
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000192 base = super(TimeRE, self)
193 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])",
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000196 'H': r"(?P<H>2[0-3]|[0-1]\d|\d)",
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000197 'I': r"(?P<I>1[0-2]|0[1-9]|[1-9])",
198 '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])",
199 'm': r"(?P<m>1[0-2]|0[1-9]|[1-9])",
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000200 'M': r"(?P<M>[0-5]\d|\d)",
201 'S': r"(?P<S>6[0-1]|[0-5]\d|\d)",
202 'U': r"(?P<U>5[0-3]|[0-4]\d|\d)",
203 'w': r"(?P<w>[0-6])",
Neal Norwitz5efc50d2002-12-30 22:23:12 +0000204 # W is set below by using 'U'
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000205 'y': r"(?P<y>\d\d)",
Brett Cannon474335c2003-08-05 04:02:49 +0000206 #XXX: Does 'Y' need to worry about having less or more than
207 # 4 digits?
208 'Y': r"(?P<Y>\d\d\d\d)",
209 'A': self.__seqToRE(self.locale_time.f_weekday, 'A'),
210 'a': self.__seqToRE(self.locale_time.a_weekday, 'a'),
211 'B': self.__seqToRE(self.locale_time.f_month[1:], 'B'),
212 'b': self.__seqToRE(self.locale_time.a_month[1:], 'b'),
213 'p': self.__seqToRE(self.locale_time.am_pm, 'p'),
Brett Cannonf7948c22004-10-06 02:23:14 +0000214 'Z': self.__seqToRE((tz for tz_names in self.locale_time.timezone
215 for tz in tz_names),
Brett Cannon474335c2003-08-05 04:02:49 +0000216 'Z'),
217 '%': '%'})
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000218 base.__setitem__('W', base.__getitem__('U').replace('U', 'W'))
Brett Cannon474335c2003-08-05 04:02:49 +0000219 base.__setitem__('c', self.pattern(self.locale_time.LC_date_time))
220 base.__setitem__('x', self.pattern(self.locale_time.LC_date))
221 base.__setitem__('X', self.pattern(self.locale_time.LC_time))
Tim Peters469cdad2002-08-08 20:19:19 +0000222
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000223 def __seqToRE(self, to_convert, directive):
Brett Cannon474335c2003-08-05 04:02:49 +0000224 """Convert a list to a regex string for matching a directive.
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000225
Brett Cannon474335c2003-08-05 04:02:49 +0000226 Want possible matching values to be from longest to shortest. This
227 prevents the possibility of a match occuring for a value that also
228 a substring of a larger value that should have matched (e.g., 'abc'
229 matching when 'abcdef' should have been the match).
Raymond Hettinger6b59f5f2003-10-16 05:53:16 +0000230
Brett Cannon474335c2003-08-05 04:02:49 +0000231 """
Brett Cannonffa5cf92004-10-06 22:48:58 +0000232 to_convert = sorted(to_convert, key=len, reverse=True)
Jack Jansen62fe7552003-01-15 22:59:39 +0000233 for value in to_convert:
234 if value != '':
235 break
236 else:
237 return ''
Brett Cannon4f35c712004-10-06 02:11:37 +0000238 regex = '|'.join(re_escape(stuff) for stuff in to_convert)
Barry Warsaw4d895fa2002-09-23 22:46:49 +0000239 regex = '(?P<%s>%s' % (directive, regex)
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000240 return '%s)' % regex
241
242 def pattern(self, format):
Brett Cannon474335c2003-08-05 04:02:49 +0000243 """Return regex pattern for the format string.
Tim Peters0eadaac2003-04-24 16:02:54 +0000244
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000245 Need to make sure that any characters that might be interpreted as
Brett Cannon5187a3b2003-08-11 07:24:05 +0000246 regex syntax are escaped.
Tim Peters0eadaac2003-04-24 16:02:54 +0000247
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000248 """
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000249 processed_format = ''
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000250 # The sub() call escapes all characters that might be misconstrued
Brett Cannon4f35c712004-10-06 02:11:37 +0000251 # as regex syntax. Cannot use re.escape since we have to deal with
252 # format directives (%m, etc.).
Brett Cannon953c6f52003-08-29 02:28:54 +0000253 regex_chars = re_compile(r"([\\.^$*+?\(\){}\[\]|])")
Brett Cannon1e91d8e2003-04-19 04:00:56 +0000254 format = regex_chars.sub(r"\\\1", format)
Tim Peters80cebc12003-01-19 04:40:44 +0000255 whitespace_replacement = re_compile('\s+')
256 format = whitespace_replacement.sub('\s*', format)
Raymond Hettingerbac788a2004-05-04 09:21:43 +0000257 while '%' in format:
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000258 directive_index = format.index('%')+1
Tim Peters469cdad2002-08-08 20:19:19 +0000259 processed_format = "%s%s%s" % (processed_format,
Barry Warsaw35816e62002-08-29 16:24:50 +0000260 format[:directive_index-1],
261 self[format[directive_index]])
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000262 format = format[directive_index+1:]
263 return "%s%s" % (processed_format, format)
264
265 def compile(self, format):
266 """Return a compiled re object for the format string."""
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000267 return re_compile(self.pattern(format), IGNORECASE)
268
Brett Cannon474335c2003-08-05 04:02:49 +0000269_cache_lock = _thread_allocate_lock()
270# DO NOT modify _TimeRE_cache or _regex_cache without acquiring the cache lock
271# first!
272_TimeRE_cache = TimeRE()
Brett Cannon5187a3b2003-08-11 07:24:05 +0000273_CACHE_MAX_SIZE = 5 # Max number of regexes stored in _regex_cache
Brett Cannon474335c2003-08-05 04:02:49 +0000274_regex_cache = {}
Guido van Rossum00efe7e2002-07-19 17:04:46 +0000275
276def strptime(data_string, format="%a %b %d %H:%M:%S %Y"):
Brett Cannon5187a3b2003-08-11 07:24:05 +0000277 """Return a time struct based on the input string and the format string."""
Brett Cannona783d062005-09-15 02:34:56 +0000278 global _TimeRE_cache, _regex_cache
Brett Cannon474335c2003-08-05 04:02:49 +0000279 _cache_lock.acquire()
280 try:
281 time_re = _TimeRE_cache
282 locale_time = time_re.locale_time
283 if _getlang() != locale_time.lang:
284 _TimeRE_cache = TimeRE()
Brett Cannona783d062005-09-15 02:34:56 +0000285 _regex_cache = {}
Brett Cannon474335c2003-08-05 04:02:49 +0000286 if len(_regex_cache) > _CACHE_MAX_SIZE:
287 _regex_cache.clear()
288 format_regex = _regex_cache.get(format)
289 if not format_regex:
Brett Cannon5d0bf942005-11-02 23:04:26 +0000290 try:
291 format_regex = time_re.compile(format)
292 # KeyError raised when a bad format is found; can be specified as
293 # \\, in which case it was a stray % but with a space after it
294 except KeyError, err:
295 bad_directive = err.args[0]
296 if bad_directive == "\\":
297 bad_directive = "%"
298 del err
299 raise ValueError("'%s' is a bad directive in format '%s'" %
300 (bad_directive, format))
301 # IndexError only occurs when the format string is "%"
302 except IndexError:
303 raise ValueError("stray %% in format '%s'" % format)
Brett Cannon474335c2003-08-05 04:02:49 +0000304 _regex_cache[format] = format_regex
305 finally:
306 _cache_lock.release()
Tim Peters80cebc12003-01-19 04:40:44 +0000307 found = format_regex.match(data_string)
Tim Peters08e54272003-01-18 03:53:49 +0000308 if not found:
Raymond Hettinger4a6302b2003-07-13 01:31:38 +0000309 raise ValueError("time data did not match format: data=%s fmt=%s" %
310 (data_string, format))
Brett Cannon2b6dfec2003-04-28 21:30:13 +0000311 if len(data_string) != found.end():
312 raise ValueError("unconverted data remains: %s" %
313 data_string[found.end():])
Tim Peters08e54272003-01-18 03:53:49 +0000314 year = 1900
315 month = day = 1
316 hour = minute = second = 0
317 tz = -1
Brett Cannon8dc25ad2004-10-18 01:47:46 +0000318 # Default to -1 to signify that values not known; not critical to have,
319 # though
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000320 week_of_year = -1
321 week_of_year_start = -1
Brett Cannon8dc25ad2004-10-18 01:47:46 +0000322 # weekday and julian defaulted to -1 so as to signal need to calculate
323 # values
Tim Peters08e54272003-01-18 03:53:49 +0000324 weekday = julian = -1
325 found_dict = found.groupdict()
326 for group_key in found_dict.iterkeys():
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000327 # Directives not explicitly handled below:
328 # c, x, X
329 # handled by making out of other directives
330 # U, W
331 # worthless without day of the week
Tim Peters08e54272003-01-18 03:53:49 +0000332 if group_key == 'y':
333 year = int(found_dict['y'])
334 # Open Group specification for strptime() states that a %y
335 #value in the range of [00, 68] is in the century 2000, while
336 #[69,99] is in the century 1900
337 if year <= 68:
338 year += 2000
339 else:
340 year += 1900
341 elif group_key == 'Y':
342 year = int(found_dict['Y'])
343 elif group_key == 'm':
344 month = int(found_dict['m'])
345 elif group_key == 'B':
Brett Cannon474335c2003-08-05 04:02:49 +0000346 month = locale_time.f_month.index(found_dict['B'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000347 elif group_key == 'b':
Brett Cannon474335c2003-08-05 04:02:49 +0000348 month = locale_time.a_month.index(found_dict['b'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000349 elif group_key == 'd':
350 day = int(found_dict['d'])
Neal Norwitz77290f22003-06-29 04:16:49 +0000351 elif group_key == 'H':
Tim Peters08e54272003-01-18 03:53:49 +0000352 hour = int(found_dict['H'])
353 elif group_key == 'I':
354 hour = int(found_dict['I'])
355 ampm = found_dict.get('p', '').lower()
356 # If there was no AM/PM indicator, we'll treat this like AM
Brett Cannon474335c2003-08-05 04:02:49 +0000357 if ampm in ('', locale_time.am_pm[0]):
Tim Peters08e54272003-01-18 03:53:49 +0000358 # We're in AM so the hour is correct unless we're
359 # looking at 12 midnight.
360 # 12 midnight == 12 AM == hour 0
361 if hour == 12:
362 hour = 0
Brett Cannon474335c2003-08-05 04:02:49 +0000363 elif ampm == locale_time.am_pm[1]:
Tim Peters08e54272003-01-18 03:53:49 +0000364 # We're in PM so we need to add 12 to the hour unless
365 # we're looking at 12 noon.
366 # 12 noon == 12 PM == hour 12
367 if hour != 12:
368 hour += 12
369 elif group_key == 'M':
370 minute = int(found_dict['M'])
371 elif group_key == 'S':
372 second = int(found_dict['S'])
373 elif group_key == 'A':
Brett Cannon474335c2003-08-05 04:02:49 +0000374 weekday = locale_time.f_weekday.index(found_dict['A'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000375 elif group_key == 'a':
Brett Cannon474335c2003-08-05 04:02:49 +0000376 weekday = locale_time.a_weekday.index(found_dict['a'].lower())
Tim Peters08e54272003-01-18 03:53:49 +0000377 elif group_key == 'w':
378 weekday = int(found_dict['w'])
379 if weekday == 0:
380 weekday = 6
381 else:
382 weekday -= 1
383 elif group_key == 'j':
384 julian = int(found_dict['j'])
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000385 elif group_key in ('U', 'W'):
386 week_of_year = int(found_dict[group_key])
387 if group_key == 'U':
388 # U starts week on Sunday
389 week_of_year_start = 6
390 else:
391 # W starts week on Monday
392 week_of_year_start = 0
Tim Peters08e54272003-01-18 03:53:49 +0000393 elif group_key == 'Z':
Brett Cannon172d9ef2003-05-11 06:23:36 +0000394 # Since -1 is default value only need to worry about setting tz if
395 # it can be something other than -1.
Tim Peters08e54272003-01-18 03:53:49 +0000396 found_zone = found_dict['Z'].lower()
Brett Cannon5187a3b2003-08-11 07:24:05 +0000397 for value, tz_values in enumerate(locale_time.timezone):
398 if found_zone in tz_values:
399 # Deal with bad locale setup where timezone names are the
400 # same and yet time.daylight is true; too ambiguous to
401 # be able to tell what timezone has daylight savings
Brett Cannon8172ac32004-03-07 23:16:27 +0000402 if (time.tzname[0] == time.tzname[1] and
403 time.daylight and found_zone not in ("utc", "gmt")):
Tim Peters58eb11c2004-01-18 20:29:55 +0000404 break
Brett Cannon5187a3b2003-08-11 07:24:05 +0000405 else:
Brett Cannon474335c2003-08-05 04:02:49 +0000406 tz = value
Brett Cannon5187a3b2003-08-11 07:24:05 +0000407 break
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000408 # If we know the week of the year and what day of that week, we can figure
409 # out the Julian day of the year
410 # Calculations below assume 0 is a Monday
Brett Cannon14adbe72004-10-28 04:49:21 +0000411 if julian == -1 and week_of_year != -1 and weekday != -1:
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000412 # Calculate how many days in week 0
413 first_weekday = datetime_date(year, 1, 1).weekday()
414 preceeding_days = 7 - first_weekday
415 if preceeding_days == 7:
416 preceeding_days = 0
Brett Cannon14adbe72004-10-28 04:49:21 +0000417 # Adjust for U directive so that calculations are not dependent on
418 # directive used to figure out week of year
419 if weekday == 6 and week_of_year_start == 6:
420 week_of_year -= 1
421 # If a year starts and ends on a Monday but a week is specified to
422 # start on a Sunday we need to up the week to counter-balance the fact
423 # that with %W that first Monday starts week 1 while with %U that is
424 # week 0 and thus shifts everything by a week
425 if weekday == 0 and first_weekday == 0 and week_of_year_start == 6:
426 week_of_year += 1
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000427 # If in week 0, then just figure out how many days from Jan 1 to day of
428 # week specified, else calculate by multiplying week of year by 7,
429 # adding in days in week 0, and the number of days from Monday to the
430 # day of the week
Brett Cannon14adbe72004-10-28 04:49:21 +0000431 if week_of_year == 0:
Brett Cannon8abcc5d2004-10-18 01:37:57 +0000432 julian = 1 + weekday - first_weekday
433 else:
434 days_to_week = preceeding_days + (7 * (week_of_year - 1))
435 julian = 1 + days_to_week + weekday
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000436 # Cannot pre-calculate datetime_date() since can change in Julian
437 #calculation and thus could have different value for the day of the week
438 #calculation
Tim Peters08e54272003-01-18 03:53:49 +0000439 if julian == -1:
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000440 # Need to add 1 to result since first day of the year is 1, not 0.
441 julian = datetime_date(year, month, day).toordinal() - \
442 datetime_date(year, 1, 1).toordinal() + 1
443 else: # Assume that if they bothered to include Julian day it will
Tim Peters08e54272003-01-18 03:53:49 +0000444 #be accurate
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000445 datetime_result = datetime_date.fromordinal((julian - 1) + datetime_date(year, 1, 1).toordinal())
446 year = datetime_result.year
447 month = datetime_result.month
448 day = datetime_result.day
Tim Peters08e54272003-01-18 03:53:49 +0000449 if weekday == -1:
Raymond Hettinger1fdb6332003-03-09 07:44:42 +0000450 weekday = datetime_date(year, month, day).weekday()
Tim Peters08e54272003-01-18 03:53:49 +0000451 return time.struct_time((year, month, day,
452 hour, minute, second,
453 weekday, julian, tz))