1"""Strptime-related classes and functions. 2 3CLASSES: 4 LocaleTime -- Discovers and stores locale-specific time information 5 TimeRE -- Creates regexes for pattern matching a string of text containing 6 time information 7 8FUNCTIONS: 9 _getlang -- Figure out what language is being used for the locale 10 strptime -- Calculates the time struct represented by the passed-in string 11 12""" 13import time 14import locale 15import calendar 16from re import compile as re_compile 17from re import IGNORECASE 18from re import escape as re_escape 19from datetime import (date as datetime_date, 20 timedelta as datetime_timedelta, 21 timezone as datetime_timezone) 22from _thread import allocate_lock as _thread_allocate_lock 23 24__all__ = [] 25 26def _getlang(): 27 # Figure out what the current language is set to. 28 return locale.getlocale(locale.LC_TIME) 29 30class LocaleTime(object): 31 """Stores and handles locale-specific information related to time. 32 33 ATTRIBUTES: 34 f_weekday -- full weekday names (7-item list) 35 a_weekday -- abbreviated weekday names (7-item list) 36 f_month -- full month names (13-item list; dummy value in [0], which 37 is added by code) 38 a_month -- abbreviated month names (13-item list, dummy value in 39 [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) 44 timezone -- daylight- and non-daylight-savings timezone representation 45 (2-item list of sets) 46 lang -- Language used by instance (2-item tuple) 47 """ 48 49 def __init__(self): 50 """Set all attributes. 51 52 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. 63 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. 67 68 """ 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") 77 if time.tzname != self.tzname or time.daylight != self.daylight: 78 raise ValueError("timezone changed during initialization") 79 80 def __calc_weekday(self): 81 # Set self.a_weekday and self.f_weekday using the calendar 82 # module. 83 a_weekday = [calendar.day_abbr[i].lower() for i in range(7)] 84 f_weekday = [calendar.day_name[i].lower() for i in range(7)] 85 self.a_weekday = a_weekday 86 self.f_weekday = f_weekday 87 88 def __calc_month(self): 89 # Set self.f_month and self.a_month using the calendar module. 90 a_month = [calendar.month_abbr[i].lower() for i in range(13)] 91 f_month = [calendar.month_name[i].lower() for i in range(13)] 92 self.a_month = a_month 93 self.f_month = f_month 94 95 def __calc_am_pm(self): 96 # Set self.am_pm by using time.strftime(). 97 98 # The magic date (1999,3,17,hour,44,55,2,76,0) is not really that 99 # magical; just happened to have used it everywhere else where a 100 # static date was needed. 101 am_pm = [] 102 for hour in (1, 22): 103 time_tuple = time.struct_time((1999,3,17,hour,44,55,2,76,0)) 104 am_pm.append(time.strftime("%p", time_tuple).lower()) 105 self.am_pm = am_pm 106 107 def __calc_date_time(self): 108 # Set self.date_time, self.date, & self.time by using 109 # time.strftime(). 110 111 # Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of 112 # overloaded numbers is minimized. The order in which searches for 113 # values within the format string is very important; it eliminates 114 # possible ambiguity for what something represents. 115 time_tuple = time.struct_time((1999,3,17,22,44,55,2,76,0)) 116 date_time = [None, None, None] 117 date_time[0] = time.strftime("%c", time_tuple).lower() 118 date_time[1] = time.strftime("%x", time_tuple).lower() 119 date_time[2] = time.strftime("%X", time_tuple).lower() 120 replacement_pairs = [('%', '%%'), (self.f_weekday[2], '%A'), 121 (self.f_month[3], '%B'), (self.a_weekday[2], '%a'), 122 (self.a_month[3], '%b'), (self.am_pm[1], '%p'), 123 ('1999', '%Y'), ('99', '%y'), ('22', '%H'), 124 ('44', '%M'), ('55', '%S'), ('76', '%j'), 125 ('17', '%d'), ('03', '%m'), ('3', '%m'), 126 # '3' needed for when no leading zero. 127 ('2', '%w'), ('10', '%I')] 128 replacement_pairs.extend([(tz, "%Z") for tz_values in self.timezone 129 for tz in tz_values]) 130 for offset,directive in ((0,'%c'), (1,'%x'), (2,'%X')): 131 current_format = date_time[offset] 132 for old, new in replacement_pairs: 133 # Must deal with possible lack of locale info 134 # manifesting itself as the empty string (e.g., Swedish's 135 # lack of AM/PM info) or a platform returning a tuple of empty 136 # strings (e.g., MacOS 9 having timezone as ('','')). 137 if old: 138 current_format = current_format.replace(old, new) 139 # If %W is used, then Sunday, 2005-01-03 will fall on week 0 since 140 # 2005-01-03 occurs before the first Monday of the year. Otherwise 141 # %U is used. 142 time_tuple = time.struct_time((1999,1,3,1,1,1,6,3,0)) 143 if '00' in time.strftime(directive, time_tuple): 144 U_W = '%W' 145 else: 146 U_W = '%U' 147 date_time[offset] = current_format.replace('11', U_W) 148 self.LC_date_time = date_time[0] 149 self.LC_date = date_time[1] 150 self.LC_time = date_time[2] 151 152 def __calc_timezone(self): 153 # Set self.timezone by using time.tzname. 154 # Do not worry about possibility of time.tzname[0] == time.tzname[1] 155 # and time.daylight; handle that in strptime. 156 try: 157 time.tzset() 158 except AttributeError: 159 pass 160 self.tzname = time.tzname 161 self.daylight = time.daylight 162 no_saving = frozenset({"utc", "gmt", self.tzname[0].lower()}) 163 if self.daylight: 164 has_saving = frozenset({self.tzname[1].lower()}) 165 else: 166 has_saving = frozenset() 167 self.timezone = (no_saving, has_saving) 168 169 170class TimeRE(dict): 171 """Handle conversion from format directives to regexes.""" 172 173 def __init__(self, locale_time=None): 174 """Create keys/values. 175 176 Order of execution is important for dependency reasons. 177 178 """ 179 if locale_time: 180 self.locale_time = locale_time 181 else: 182 self.locale_time = LocaleTime() 183 base = super() 184 base.__init__({ 185 # The " [1-9]" part of the regex is to make %c from ANSI C work 186 'd': r"(?P<d>3[0-1]|[1-2]\d|0[1-9]|[1-9]| [1-9])", 187 'f': r"(?P<f>[0-9]{1,6})", 188 'H': r"(?P<H>2[0-3]|[0-1]\d|\d)", 189 'I': r"(?P<I>1[0-2]|0[1-9]|[1-9])", 190 'G': r"(?P<G>\d\d\d\d)", 191 '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])", 192 'm': r"(?P<m>1[0-2]|0[1-9]|[1-9])", 193 'M': r"(?P<M>[0-5]\d|\d)", 194 'S': r"(?P<S>6[0-1]|[0-5]\d|\d)", 195 'U': r"(?P<U>5[0-3]|[0-4]\d|\d)", 196 'w': r"(?P<w>[0-6])", 197 'u': r"(?P<u>[1-7])", 198 'V': r"(?P<V>5[0-3]|0[1-9]|[1-4]\d|\d)", 199 # W is set below by using 'U' 200 'y': r"(?P<y>\d\d)", 201 #XXX: Does 'Y' need to worry about having less or more than 202 # 4 digits? 203 'Y': r"(?P<Y>\d\d\d\d)", 204 'z': r"(?P<z>[+-]\d\d:?[0-5]\d(:?[0-5]\d(\.\d{1,6})?)?|Z)", 205 'A': self.__seqToRE(self.locale_time.f_weekday, 'A'), 206 'a': self.__seqToRE(self.locale_time.a_weekday, 'a'), 207 'B': self.__seqToRE(self.locale_time.f_month[1:], 'B'), 208 'b': self.__seqToRE(self.locale_time.a_month[1:], 'b'), 209 'p': self.__seqToRE(self.locale_time.am_pm, 'p'), 210 'Z': self.__seqToRE((tz for tz_names in self.locale_time.timezone 211 for tz in tz_names), 212 'Z'), 213 '%': '%'}) 214 base.__setitem__('W', base.__getitem__('U').replace('U', 'W')) 215 base.__setitem__('c', self.pattern(self.locale_time.LC_date_time)) 216 base.__setitem__('x', self.pattern(self.locale_time.LC_date)) 217 base.__setitem__('X', self.pattern(self.locale_time.LC_time)) 218 219 def __seqToRE(self, to_convert, directive): 220 """Convert a list to a regex string for matching a directive. 221 222 Want possible matching values to be from longest to shortest. This 223 prevents the possibility of a match occurring for a value that also 224 a substring of a larger value that should have matched (e.g., 'abc' 225 matching when 'abcdef' should have been the match). 226 227 """ 228 to_convert = sorted(to_convert, key=len, reverse=True) 229 for value in to_convert: 230 if value != '': 231 break 232 else: 233 return '' 234 regex = '|'.join(re_escape(stuff) for stuff in to_convert) 235 regex = '(?P<%s>%s' % (directive, regex) 236 return '%s)' % regex 237 238 def pattern(self, format): 239 """Return regex pattern for the format string. 240 241 Need to make sure that any characters that might be interpreted as 242 regex syntax are escaped. 243 244 """ 245 processed_format = '' 246 # The sub() call escapes all characters that might be misconstrued 247 # as regex syntax. Cannot use re.escape since we have to deal with 248 # format directives (%m, etc.). 249 regex_chars = re_compile(r"([\\.^$*+?\(\){}\[\]|])") 250 format = regex_chars.sub(r"\\\1", format) 251 whitespace_replacement = re_compile(r'\s+') 252 format = whitespace_replacement.sub(r'\\s+', format) 253 while '%' in format: 254 directive_index = format.index('%')+1 255 processed_format = "%s%s%s" % (processed_format, 256 format[:directive_index-1], 257 self[format[directive_index]]) 258 format = format[directive_index+1:] 259 return "%s%s" % (processed_format, format) 260 261 def compile(self, format): 262 """Return a compiled re object for the format string.""" 263 return re_compile(self.pattern(format), IGNORECASE) 264 265_cache_lock = _thread_allocate_lock() 266# DO NOT modify _TimeRE_cache or _regex_cache without acquiring the cache lock 267# first! 268_TimeRE_cache = TimeRE() 269_CACHE_MAX_SIZE = 5 # Max number of regexes stored in _regex_cache 270_regex_cache = {} 271 272def _calc_julian_from_U_or_W(year, week_of_year, day_of_week, week_starts_Mon): 273 """Calculate the Julian day based on the year, week of the year, and day of 274 the week, with week_start_day representing whether the week of the year 275 assumes the week starts on Sunday or Monday (6 or 0).""" 276 first_weekday = datetime_date(year, 1, 1).weekday() 277 # If we are dealing with the %U directive (week starts on Sunday), it's 278 # easier to just shift the view to Sunday being the first day of the 279 # week. 280 if not week_starts_Mon: 281 first_weekday = (first_weekday + 1) % 7 282 day_of_week = (day_of_week + 1) % 7 283 # Need to watch out for a week 0 (when the first day of the year is not 284 # the same as that specified by %U or %W). 285 week_0_length = (7 - first_weekday) % 7 286 if week_of_year == 0: 287 return 1 + day_of_week - first_weekday 288 else: 289 days_to_week = week_0_length + (7 * (week_of_year - 1)) 290 return 1 + days_to_week + day_of_week 291 292 293def _calc_julian_from_V(iso_year, iso_week, iso_weekday): 294 """Calculate the Julian day based on the ISO 8601 year, week, and weekday. 295 ISO weeks start on Mondays, with week 01 being the week containing 4 Jan. 296 ISO week days range from 1 (Monday) to 7 (Sunday). 297 """ 298 correction = datetime_date(iso_year, 1, 4).isoweekday() + 3 299 ordinal = (iso_week * 7) + iso_weekday - correction 300 # ordinal may be negative or 0 now, which means the date is in the previous 301 # calendar year 302 if ordinal < 1: 303 ordinal += datetime_date(iso_year, 1, 1).toordinal() 304 iso_year -= 1 305 ordinal -= datetime_date(iso_year, 1, 1).toordinal() 306 return iso_year, ordinal 307 308 309def _strptime(data_string, format="%a %b %d %H:%M:%S %Y"): 310 """Return a 2-tuple consisting of a time struct and an int containing 311 the number of microseconds based on the input string and the 312 format string.""" 313 314 for index, arg in enumerate([data_string, format]): 315 if not isinstance(arg, str): 316 msg = "strptime() argument {} must be str, not {}" 317 raise TypeError(msg.format(index, type(arg))) 318 319 global _TimeRE_cache, _regex_cache 320 with _cache_lock: 321 locale_time = _TimeRE_cache.locale_time 322 if (_getlang() != locale_time.lang or 323 time.tzname != locale_time.tzname or 324 time.daylight != locale_time.daylight): 325 _TimeRE_cache = TimeRE() 326 _regex_cache.clear() 327 locale_time = _TimeRE_cache.locale_time 328 if len(_regex_cache) > _CACHE_MAX_SIZE: 329 _regex_cache.clear() 330 format_regex = _regex_cache.get(format) 331 if not format_regex: 332 try: 333 format_regex = _TimeRE_cache.compile(format) 334 # KeyError raised when a bad format is found; can be specified as 335 # \\, in which case it was a stray % but with a space after it 336 except KeyError as err: 337 bad_directive = err.args[0] 338 if bad_directive == "\\": 339 bad_directive = "%" 340 del err 341 raise ValueError("'%s' is a bad directive in format '%s'" % 342 (bad_directive, format)) from None 343 # IndexError only occurs when the format string is "%" 344 except IndexError: 345 raise ValueError("stray %% in format '%s'" % format) from None 346 _regex_cache[format] = format_regex 347 found = format_regex.match(data_string) 348 if not found: 349 raise ValueError("time data %r does not match format %r" % 350 (data_string, format)) 351 if len(data_string) != found.end(): 352 raise ValueError("unconverted data remains: %s" % 353 data_string[found.end():]) 354 355 iso_year = year = None 356 month = day = 1 357 hour = minute = second = fraction = 0 358 tz = -1 359 gmtoff = None 360 gmtoff_fraction = 0 361 # Default to -1 to signify that values not known; not critical to have, 362 # though 363 iso_week = week_of_year = None 364 week_of_year_start = None 365 # weekday and julian defaulted to None so as to signal need to calculate 366 # values 367 weekday = julian = None 368 found_dict = found.groupdict() 369 for group_key in found_dict.keys(): 370 # Directives not explicitly handled below: 371 # c, x, X 372 # handled by making out of other directives 373 # U, W 374 # worthless without day of the week 375 if group_key == 'y': 376 year = int(found_dict['y']) 377 # Open Group specification for strptime() states that a %y 378 #value in the range of [00, 68] is in the century 2000, while 379 #[69,99] is in the century 1900 380 if year <= 68: 381 year += 2000 382 else: 383 year += 1900 384 elif group_key == 'Y': 385 year = int(found_dict['Y']) 386 elif group_key == 'G': 387 iso_year = int(found_dict['G']) 388 elif group_key == 'm': 389 month = int(found_dict['m']) 390 elif group_key == 'B': 391 month = locale_time.f_month.index(found_dict['B'].lower()) 392 elif group_key == 'b': 393 month = locale_time.a_month.index(found_dict['b'].lower()) 394 elif group_key == 'd': 395 day = int(found_dict['d']) 396 elif group_key == 'H': 397 hour = int(found_dict['H']) 398 elif group_key == 'I': 399 hour = int(found_dict['I']) 400 ampm = found_dict.get('p', '').lower() 401 # If there was no AM/PM indicator, we'll treat this like AM 402 if ampm in ('', locale_time.am_pm[0]): 403 # We're in AM so the hour is correct unless we're 404 # looking at 12 midnight. 405 # 12 midnight == 12 AM == hour 0 406 if hour == 12: 407 hour = 0 408 elif ampm == locale_time.am_pm[1]: 409 # We're in PM so we need to add 12 to the hour unless 410 # we're looking at 12 noon. 411 # 12 noon == 12 PM == hour 12 412 if hour != 12: 413 hour += 12 414 elif group_key == 'M': 415 minute = int(found_dict['M']) 416 elif group_key == 'S': 417 second = int(found_dict['S']) 418 elif group_key == 'f': 419 s = found_dict['f'] 420 # Pad to always return microseconds. 421 s += "0" * (6 - len(s)) 422 fraction = int(s) 423 elif group_key == 'A': 424 weekday = locale_time.f_weekday.index(found_dict['A'].lower()) 425 elif group_key == 'a': 426 weekday = locale_time.a_weekday.index(found_dict['a'].lower()) 427 elif group_key == 'w': 428 weekday = int(found_dict['w']) 429 if weekday == 0: 430 weekday = 6 431 else: 432 weekday -= 1 433 elif group_key == 'u': 434 weekday = int(found_dict['u']) 435 weekday -= 1 436 elif group_key == 'j': 437 julian = int(found_dict['j']) 438 elif group_key in ('U', 'W'): 439 week_of_year = int(found_dict[group_key]) 440 if group_key == 'U': 441 # U starts week on Sunday. 442 week_of_year_start = 6 443 else: 444 # W starts week on Monday. 445 week_of_year_start = 0 446 elif group_key == 'V': 447 iso_week = int(found_dict['V']) 448 elif group_key == 'z': 449 z = found_dict['z'] 450 if z == 'Z': 451 gmtoff = 0 452 else: 453 if z[3] == ':': 454 z = z[:3] + z[4:] 455 if len(z) > 5: 456 if z[5] != ':': 457 msg = f"Inconsistent use of : in {found_dict['z']}" 458 raise ValueError(msg) 459 z = z[:5] + z[6:] 460 hours = int(z[1:3]) 461 minutes = int(z[3:5]) 462 seconds = int(z[5:7] or 0) 463 gmtoff = (hours * 60 * 60) + (minutes * 60) + seconds 464 gmtoff_remainder = z[8:] 465 # Pad to always return microseconds. 466 gmtoff_remainder_padding = "0" * (6 - len(gmtoff_remainder)) 467 gmtoff_fraction = int(gmtoff_remainder + gmtoff_remainder_padding) 468 if z.startswith("-"): 469 gmtoff = -gmtoff 470 gmtoff_fraction = -gmtoff_fraction 471 elif group_key == 'Z': 472 # Since -1 is default value only need to worry about setting tz if 473 # it can be something other than -1. 474 found_zone = found_dict['Z'].lower() 475 for value, tz_values in enumerate(locale_time.timezone): 476 if found_zone in tz_values: 477 # Deal with bad locale setup where timezone names are the 478 # same and yet time.daylight is true; too ambiguous to 479 # be able to tell what timezone has daylight savings 480 if (time.tzname[0] == time.tzname[1] and 481 time.daylight and found_zone not in ("utc", "gmt")): 482 break 483 else: 484 tz = value 485 break 486 # Deal with the cases where ambiguities arize 487 # don't assume default values for ISO week/year 488 if year is None and iso_year is not None: 489 if iso_week is None or weekday is None: 490 raise ValueError("ISO year directive '%G' must be used with " 491 "the ISO week directive '%V' and a weekday " 492 "directive ('%A', '%a', '%w', or '%u').") 493 if julian is not None: 494 raise ValueError("Day of the year directive '%j' is not " 495 "compatible with ISO year directive '%G'. " 496 "Use '%Y' instead.") 497 elif week_of_year is None and iso_week is not None: 498 if weekday is None: 499 raise ValueError("ISO week directive '%V' must be used with " 500 "the ISO year directive '%G' and a weekday " 501 "directive ('%A', '%a', '%w', or '%u').") 502 else: 503 raise ValueError("ISO week directive '%V' is incompatible with " 504 "the year directive '%Y'. Use the ISO year '%G' " 505 "instead.") 506 507 leap_year_fix = False 508 if year is None and month == 2 and day == 29: 509 year = 1904 # 1904 is first leap year of 20th century 510 leap_year_fix = True 511 elif year is None: 512 year = 1900 513 514 515 # If we know the week of the year and what day of that week, we can figure 516 # out the Julian day of the year. 517 if julian is None and weekday is not None: 518 if week_of_year is not None: 519 week_starts_Mon = True if week_of_year_start == 0 else False 520 julian = _calc_julian_from_U_or_W(year, week_of_year, weekday, 521 week_starts_Mon) 522 elif iso_year is not None and iso_week is not None: 523 year, julian = _calc_julian_from_V(iso_year, iso_week, weekday + 1) 524 if julian is not None and julian <= 0: 525 year -= 1 526 yday = 366 if calendar.isleap(year) else 365 527 julian += yday 528 529 if julian is None: 530 # Cannot pre-calculate datetime_date() since can change in Julian 531 # calculation and thus could have different value for the day of 532 # the week calculation. 533 # Need to add 1 to result since first day of the year is 1, not 0. 534 julian = datetime_date(year, month, day).toordinal() - \ 535 datetime_date(year, 1, 1).toordinal() + 1 536 else: # Assume that if they bothered to include Julian day (or if it was 537 # calculated above with year/week/weekday) it will be accurate. 538 datetime_result = datetime_date.fromordinal( 539 (julian - 1) + 540 datetime_date(year, 1, 1).toordinal()) 541 year = datetime_result.year 542 month = datetime_result.month 543 day = datetime_result.day 544 if weekday is None: 545 weekday = datetime_date(year, month, day).weekday() 546 # Add timezone info 547 tzname = found_dict.get("Z") 548 549 if leap_year_fix: 550 # the caller didn't supply a year but asked for Feb 29th. We couldn't 551 # use the default of 1900 for computations. We set it back to ensure 552 # that February 29th is smaller than March 1st. 553 year = 1900 554 555 return (year, month, day, 556 hour, minute, second, 557 weekday, julian, tz, tzname, gmtoff), fraction, gmtoff_fraction 558 559def _strptime_time(data_string, format="%a %b %d %H:%M:%S %Y"): 560 """Return a time struct based on the input string and the 561 format string.""" 562 tt = _strptime(data_string, format)[0] 563 return time.struct_time(tt[:time._STRUCT_TM_ITEMS]) 564 565def _strptime_datetime(cls, data_string, format="%a %b %d %H:%M:%S %Y"): 566 """Return a class cls instance based on the input string and the 567 format string.""" 568 tt, fraction, gmtoff_fraction = _strptime(data_string, format) 569 tzname, gmtoff = tt[-2:] 570 args = tt[:6] + (fraction,) 571 if gmtoff is not None: 572 tzdelta = datetime_timedelta(seconds=gmtoff, microseconds=gmtoff_fraction) 573 if tzname: 574 tz = datetime_timezone(tzdelta, tzname) 575 else: 576 tz = datetime_timezone(tzdelta) 577 args += (tz,) 578 579 return cls(*args) 580