Lib/_strptime.py

   1 """Strptime-related classes and functions.
   2
   3 CLASSES:
   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
   8 FUNCTIONS:
   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 """
  13 import time
  14 import locale
  15 import calendar
  16 from re import compile as re_compile
  17 from re import IGNORECASE
  18 from re import escape as re_escape
  19 from datetime import date as datetime_date
  20 try:
  21     from thread import allocate_lock as _thread_allocate_lock
  22 except:
  23     from dummy_thread import allocate_lock as _thread_allocate_lock
  24
  25 __all__ = []
  26
  27 def _getlang():
  28     # Figure out what the current language is set to.
  29     return locale.getlocale(locale.LC_TIME)
  30
  31 class LocaleTime(object):
  32     """Stores and handles locale-specific information related to time.
  33
  34     ATTRIBUTES:
  35         f_weekday -- full weekday names (7-item list)
  36         a_weekday -- abbreviated weekday names (7-item list)
  37         f_month -- full month names (13-item list; dummy value in [0], which
  38                     is added by code)
  39         a_month -- abbreviated month names (13-item list, dummy value in
  40                     [0], which is added by code)
  41         am_pm -- AM/PM representation (2-item list)
  42         LC_date_time -- format string for date/time representation (string)
  43         LC_date -- format string for date representation (string)
  44         LC_time -- format string for time representation (string)
  45         timezone -- daylight- and non-daylight-savings timezone representation
  46                     (2-item list of sets)
  47         lang -- Language used by instance (2-item tuple)
  48     """
  49
  50     def __init__(self):
  51         """Set all attributes.
  52
  53         Order of methods called matters for dependency reasons.
  54
  55         The locale language is set at the offset and then checked again before
  56         exiting.  This is to make sure that the attributes were not set with a
  57         mix of information from more than one locale.  This would most likely
  58         happen when using threads where one thread calls a locale-dependent
  59         function while another thread changes the locale while the function in
  60         the other thread is still running.  Proper coding would call for
  61         locks to prevent changing the locale while locale-dependent code is
  62         running.  The check here is done in case someone does not think about
  63         doing this.
  64
  65         Only other possible issue is if someone changed the timezone and did
  66         not call tz.tzset .  That is an issue for the programmer, though,
  67         since changing the timezone is worthless without that call.
  68
  69         """
  70         self.lang = _getlang()
  71         self.__calc_weekday()
  72         self.__calc_month()
  73         self.__calc_am_pm()
  74         self.__calc_timezone()
  75         self.__calc_date_time()
  76         if _getlang() != self.lang:
  77             raise ValueError("locale changed during initialization")
  78
  79     def __pad(self, seq, front):
  80         # Add '' to seq to either the front (is True), else the back.
  81         seq = list(seq)
  82         if front:
  83             seq.insert(0, '')
  84         else:
  85             seq.append('')
  86         return seq
  87
  88     def __calc_weekday(self):
  89         # Set self.a_weekday and self.f_weekday using the calendar
  90         # module.
  91         a_weekday = [calendar.day_abbr[i].lower() for i in range(7)]
  92         f_weekday = [calendar.day_name[i].lower() for i in range(7)]
  93         self.a_weekday = a_weekday
  94         self.f_weekday = f_weekday
  95
  96     def __calc_month(self):
  97         # Set self.f_month and self.a_month using the calendar module.
  98         a_month = [calendar.month_abbr[i].lower() for i in range(13)]
  99         f_month = [calendar.month_name[i].lower() for i in range(13)]
 100         self.a_month = a_month
 101         self.f_month = f_month
 102
 103     def __calc_am_pm(self):
 104         # Set self.am_pm by using time.strftime().
 105
 106         # The magic date (1999,3,17,hour,44,55,2,76,0) is not really that
 107         # magical; just happened to have used it everywhere else where a
 108         # static date was needed.
 109         am_pm = []
 110         for hour in (01,22):
 111             time_tuple = time.struct_time((1999,3,17,hour,44,55,2,76,0))
 112             am_pm.append(time.strftime("%p", time_tuple).lower())
 113         self.am_pm = am_pm
 114
 115     def __calc_date_time(self):
 116         # Set self.date_time, self.date, & self.time by using
 117         # time.strftime().
 118
 119         # Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of
 120         # overloaded numbers is minimized.  The order in which searches for
 121         # values within the format string is very important; it eliminates
 122         # possible ambiguity for what something represents.
 123         time_tuple = time.struct_time((1999,3,17,22,44,55,2,76,0))
 124         date_time = [None, None, None]
 125         date_time[0] = time.strftime("%c", time_tuple).lower()
 126         date_time[1] = time.strftime("%x", time_tuple).lower()
 127         date_time[2] = time.strftime("%X", time_tuple).lower()
 128         replacement_pairs = [('%', '%%'), (self.f_weekday[2], '%A'),
 129                     (self.f_month[3], '%B'), (self.a_weekday[2], '%a'),
 130                     (self.a_month[3], '%b'), (self.am_pm[1], '%p'),
 131                     ('1999', '%Y'), ('99', '%y'), ('22', '%H'),
 132                     ('44', '%M'), ('55', '%S'), ('76', '%j'),
 133                     ('17', '%d'), ('03', '%m'), ('3', '%m'),
 134                     # '3' needed for when no leading zero.
 135                     ('2', '%w'), ('10', '%I')]
 136         replacement_pairs.extend([(tz, "%Z") for tz_values in self.timezone
 137                                                 for tz in tz_values])
 138         for offset,directive in ((0,'%c'), (1,'%x'), (2,'%X')):
 139             current_format = date_time[offset]
 140             for old, new in replacement_pairs:
 141                 # Must deal with possible lack of locale info
 142                 # manifesting itself as the empty string (e.g., Swedish's
 143                 # lack of AM/PM info) or a platform returning a tuple of empty
 144                 # strings (e.g., MacOS 9 having timezone as ('','')).
 145                 if old:
 146                     current_format = current_format.replace(old, new)
 147             # If %W is used, then Sunday, 2005-01-03 will fall on week 0 since
 148             # 2005-01-03 occurs before the first Monday of the year.  Otherwise
 149             # %U is used.
 150             time_tuple = time.struct_time((1999,1,3,1,1,1,6,3,0))
 151             if '00' in time.strftime(directive, time_tuple):
 152                 U_W = '%W'
 153             else:
 154                 U_W = '%U'
 155             date_time[offset] = current_format.replace('11', U_W)
 156         self.LC_date_time = date_time[0]
 157         self.LC_date = date_time[1]
 158         self.LC_time = date_time[2]
 159
 160     def __calc_timezone(self):
 161         # Set self.timezone by using time.tzname.
 162         # Do not worry about possibility of time.tzname[0] == timetzname[1]
 163         # and time.daylight; handle that in strptime .
 164         try:
 165             time.tzset()
 166         except AttributeError:
 167             pass
 168         no_saving = frozenset(["utc", "gmt", time.tzname[0].lower()])
 169         if time.daylight:
 170             has_saving = frozenset([time.tzname[1].lower()])
 171         else:
 172             has_saving = frozenset()
 173         self.timezone = (no_saving, has_saving)
 174
 175
 176 class TimeRE(dict):
 177     """Handle conversion from format directives to regexes."""
 178
 179     def __init__(self, locale_time=None):
 180         """Create keys/values.
 181
 182         Order of execution is important for dependency reasons.
 183
 184         """
 185         if locale_time:
 186             self.locale_time = locale_time
 187         else:
 188             self.locale_time = LocaleTime()
 189         base = super(TimeRE, self)
 190         base.__init__({
 191             # The " \d" part of the regex is to make %c from ANSI C work
 192             'd': r"(?P<d>3[0-1]|[1-2]\d|0[1-9]|[1-9]| [1-9])",
 193             'f': r"(?P<f>[0-9]{1,6})",
 194             'H': r"(?P<H>2[0-3]|[0-1]\d|\d)",
 195             'I': r"(?P<I>1[0-2]|0[1-9]|[1-9])",
 196             '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])",
 197             'm': r"(?P<m>1[0-2]|0[1-9]|[1-9])",
 198             'M': r"(?P<M>[0-5]\d|\d)",
 199             'S': r"(?P<S>6[0-1]|[0-5]\d|\d)",
 200             'U': r"(?P<U>5[0-3]|[0-4]\d|\d)",
 201             'w': r"(?P<w>[0-6])",
 202             # W is set below by using 'U'
 203             'y': r"(?P<y>\d\d)",
 204             #XXX: Does 'Y' need to worry about having less or more than
 205             #     4 digits?
 206             'Y': r"(?P<Y>\d\d\d\d)",
 207             'A': self.__seqToRE(self.locale_time.f_weekday, 'A'),
 208             'a': self.__seqToRE(self.locale_time.a_weekday, 'a'),
 209             'B': self.__seqToRE(self.locale_time.f_month[1:], 'B'),
 210             'b': self.__seqToRE(self.locale_time.a_month[1:], 'b'),
 211             'p': self.__seqToRE(self.locale_time.am_pm, 'p'),
 212             'Z': self.__seqToRE((tz for tz_names in self.locale_time.timezone
 213                                         for tz in tz_names),
 214                                 'Z'),
 215             '%': '%'})
 216         base.__setitem__('W', base.__getitem__('U').replace('U', 'W'))
 217         base.__setitem__('c', self.pattern(self.locale_time.LC_date_time))
 218         base.__setitem__('x', self.pattern(self.locale_time.LC_date))
 219         base.__setitem__('X', self.pattern(self.locale_time.LC_time))
 220
 221     def __seqToRE(self, to_convert, directive):
 222         """Convert a list to a regex string for matching a directive.
 223
 224         Want possible matching values to be from longest to shortest.  This
 225         prevents the possibility of a match occuring for a value that also
 226         a substring of a larger value that should have matched (e.g., 'abc'
 227         matching when 'abcdef' should have been the match).
 228
 229         """
 230         to_convert = sorted(to_convert, key=len, reverse=True)
 231         for value in to_convert:
 232             if value != '':
 233                 break
 234         else:
 235             return ''
 236         regex = '|'.join(re_escape(stuff) for stuff in to_convert)
 237         regex = '(?P<%s>%s' % (directive, regex)
 238         return '%s)' % regex
 239
 240     def pattern(self, format):
 241         """Return regex pattern for the format string.
 242
 243         Need to make sure that any characters that might be interpreted as
 244         regex syntax are escaped.
 245
 246         """
 247         processed_format = ''
 248         # The sub() call escapes all characters that might be misconstrued
 249         # as regex syntax.  Cannot use re.escape since we have to deal with
 250         # format directives (%m, etc.).
 251         regex_chars = re_compile(r"([\\.^$*+?\(\){}\[\]|])")
 252         format = regex_chars.sub(r"\\\1", format)
 253         whitespace_replacement = re_compile('\s+')
 254         format = whitespace_replacement.sub('\s+', format)
 255         while '%' in format:
 256             directive_index = format.index('%')+1
 257             processed_format = "%s%s%s" % (processed_format,
 258                                            format[:directive_index-1],
 259                                            self[format[directive_index]])
 260             format = format[directive_index+1:]
 261         return "%s%s" % (processed_format, format)
 262
 263     def compile(self, format):
 264         """Return a compiled re object for the format string."""
 265         return re_compile(self.pattern(format), IGNORECASE)
 266
 267 _cache_lock = _thread_allocate_lock()
 268 # DO NOT modify _TimeRE_cache or _regex_cache without acquiring the cache lock
 269 # first!
 270 _TimeRE_cache = TimeRE()
 271 _CACHE_MAX_SIZE = 5 # Max number of regexes stored in _regex_cache
 272 _regex_cache = {}
 273
 274 def _calc_julian_from_U_or_W(year, week_of_year, day_of_week, week_starts_Mon):
 275     """Calculate the Julian day based on the year, week of the year, and day of
 276     the week, with week_start_day representing whether the week of the year
 277     assumes the week starts on Sunday or Monday (6 or 0)."""
 278     first_weekday = datetime_date(year, 1, 1).weekday()
 279     # If we are dealing with the %U directive (week starts on Sunday), it's
 280     # easier to just shift the view to Sunday being the first day of the
 281     # week.
 282     if not week_starts_Mon:
 283         first_weekday = (first_weekday + 1) % 7
 284         day_of_week = (day_of_week + 1) % 7
 285     # Need to watch out for a week 0 (when the first day of the year is not
 286     # the same as that specified by %U or %W).
 287     week_0_length = (7 - first_weekday) % 7
 288     if week_of_year == 0:
 289         return 1 + day_of_week - first_weekday
 290     else:
 291         days_to_week = week_0_length + (7 * (week_of_year - 1))
 292         return 1 + days_to_week + day_of_week
 293
 294
 295 def _strptime(data_string, format="%a %b %d %H:%M:%S %Y"):
 296     """Return a time struct based on the input string and the format string."""
 297     global _TimeRE_cache, _regex_cache
 298     with _cache_lock:
 299         if _getlang() != _TimeRE_cache.locale_time.lang:
 300             _TimeRE_cache = TimeRE()
 301             _regex_cache.clear()
 302         if len(_regex_cache) > _CACHE_MAX_SIZE:
 303             _regex_cache.clear()
 304         locale_time = _TimeRE_cache.locale_time
 305         format_regex = _regex_cache.get(format)
 306         if not format_regex:
 307             try:
 308                 format_regex = _TimeRE_cache.compile(format)
 309             # KeyError raised when a bad format is found; can be specified as
 310             # \\, in which case it was a stray % but with a space after it
 311             except KeyError, err:
 312                 bad_directive = err.args[0]
 313                 if bad_directive == "\\":
 314                     bad_directive = "%"
 315                 del err
 316                 raise ValueError("'%s' is a bad directive in format '%s'" %
 317                                     (bad_directive, format))
 318             # IndexError only occurs when the format string is "%"
 319             except IndexError:
 320                 raise ValueError("stray %% in format '%s'" % format)
 321             _regex_cache[format] = format_regex
 322     found = format_regex.match(data_string)
 323     if not found:
 324         raise ValueError("time data %r does not match format %r" %
 325                          (data_string, format))
 326     if len(data_string) != found.end():
 327         raise ValueError("unconverted data remains: %s" %
 328                           data_string[found.end():])
 329     year = 1900
 330     month = day = 1
 331     hour = minute = second = fraction = 0
 332     tz = -1
 333     # Default to -1 to signify that values not known; not critical to have,
 334     # though
 335     week_of_year = -1
 336     week_of_year_start = -1
 337     # weekday and julian defaulted to -1 so as to signal need to calculate
 338     # values
 339     weekday = julian = -1
 340     found_dict = found.groupdict()
 341     for group_key in found_dict.iterkeys():
 342         # Directives not explicitly handled below:
 343         #   c, x, X
 344         #      handled by making out of other directives
 345         #   U, W
 346         #      worthless without day of the week
 347         if group_key == 'y':
 348             year = int(found_dict['y'])
 349             # Open Group specification for strptime() states that a %y
 350             #value in the range of [00, 68] is in the century 2000, while
 351             #[69,99] is in the century 1900
 352             if year <= 68:
 353                 year += 2000
 354             else:
 355                 year += 1900
 356         elif group_key == 'Y':
 357             year = int(found_dict['Y'])
 358         elif group_key == 'm':
 359             month = int(found_dict['m'])
 360         elif group_key == 'B':
 361             month = locale_time.f_month.index(found_dict['B'].lower())
 362         elif group_key == 'b':
 363             month = locale_time.a_month.index(found_dict['b'].lower())
 364         elif group_key == 'd':
 365             day = int(found_dict['d'])
 366         elif group_key == 'H':
 367             hour = int(found_dict['H'])
 368         elif group_key == 'I':
 369             hour = int(found_dict['I'])
 370             ampm = found_dict.get('p', '').lower()
 371             # If there was no AM/PM indicator, we'll treat this like AM
 372             if ampm in ('', locale_time.am_pm[0]):
 373                 # We're in AM so the hour is correct unless we're
 374                 # looking at 12 midnight.
 375                 # 12 midnight == 12 AM == hour 0
 376                 if hour == 12:
 377                     hour = 0
 378             elif ampm == locale_time.am_pm[1]:
 379                 # We're in PM so we need to add 12 to the hour unless
 380                 # we're looking at 12 noon.
 381                 # 12 noon == 12 PM == hour 12
 382                 if hour != 12:
 383                     hour += 12
 384         elif group_key == 'M':
 385             minute = int(found_dict['M'])
 386         elif group_key == 'S':
 387             second = int(found_dict['S'])
 388         elif group_key == 'f':
 389             s = found_dict['f']
 390             # Pad to always return microseconds.
 391             s += "0" * (6 - len(s))
 392             fraction = int(s)
 393         elif group_key == 'A':
 394             weekday = locale_time.f_weekday.index(found_dict['A'].lower())
 395         elif group_key == 'a':
 396             weekday = locale_time.a_weekday.index(found_dict['a'].lower())
 397         elif group_key == 'w':
 398             weekday = int(found_dict['w'])
 399             if weekday == 0:
 400                 weekday = 6
 401             else:
 402                 weekday -= 1
 403         elif group_key == 'j':
 404             julian = int(found_dict['j'])
 405         elif group_key in ('U', 'W'):
 406             week_of_year = int(found_dict[group_key])
 407             if group_key == 'U':
 408                 # U starts week on Sunday.
 409                 week_of_year_start = 6
 410             else:
 411                 # W starts week on Monday.
 412                 week_of_year_start = 0
 413         elif group_key == 'Z':
 414             # Since -1 is default value only need to worry about setting tz if
 415             # it can be something other than -1.
 416             found_zone = found_dict['Z'].lower()
 417             for value, tz_values in enumerate(locale_time.timezone):
 418                 if found_zone in tz_values:
 419                     # Deal with bad locale setup where timezone names are the
 420                     # same and yet time.daylight is true; too ambiguous to
 421                     # be able to tell what timezone has daylight savings
 422                     if (time.tzname[0] == time.tzname[1] and
 423                        time.daylight and found_zone not in ("utc", "gmt")):
 424                         break
 425                     else:
 426                         tz = value
 427                         break
 428     # If we know the week of the year and what day of that week, we can figure
 429     # out the Julian day of the year.
 430     if julian == -1 and week_of_year != -1 and weekday != -1:
 431         week_starts_Mon = True if week_of_year_start == 0 else False
 432         julian = _calc_julian_from_U_or_W(year, week_of_year, weekday,
 433                                             week_starts_Mon)
 434     # Cannot pre-calculate datetime_date() since can change in Julian
 435     # calculation and thus could have different value for the day of the week
 436     # calculation.
 437     if julian == -1:
 438         # Need to add 1 to result since first day of the year is 1, not 0.
 439         julian = datetime_date(year, month, day).toordinal() - \
 440                   datetime_date(year, 1, 1).toordinal() + 1
 441     else:  # Assume that if they bothered to include Julian day it will
 442            # be accurate.
 443         datetime_result = datetime_date.fromordinal((julian - 1) + datetime_date(year, 1, 1).toordinal())
 444         year = datetime_result.year
 445         month = datetime_result.month
 446         day = datetime_result.day
 447     if weekday == -1:
 448         weekday = datetime_date(year, month, day).weekday()
 449     return (time.struct_time((year, month, day,
 450                               hour, minute, second,
 451                               weekday, julian, tz)), fraction)
 452
 453 def _strptime_time(data_string, format="%a %b %d %H:%M:%S %Y"):
 454     return _strptime(data_string, format)[0]