jpayne@69: '''
jpayne@69: $Id: tzfile.py,v 1.8 2004/06/03 00:15:24 zenzen Exp $
jpayne@69: '''
jpayne@69: 
jpayne@69: from datetime import datetime
jpayne@69: from struct import unpack, calcsize
jpayne@69: 
jpayne@69: from pytz.tzinfo import StaticTzInfo, DstTzInfo, memorized_ttinfo
jpayne@69: from pytz.tzinfo import memorized_datetime, memorized_timedelta
jpayne@69: 
jpayne@69: 
jpayne@69: def _byte_string(s):
jpayne@69:     """Cast a string or byte string to an ASCII byte string."""
jpayne@69:     return s.encode('ASCII')
jpayne@69: 
jpayne@69: _NULL = _byte_string('\0')
jpayne@69: 
jpayne@69: 
jpayne@69: def _std_string(s):
jpayne@69:     """Cast a string or byte string to an ASCII string."""
jpayne@69:     return str(s.decode('ASCII'))
jpayne@69: 
jpayne@69: 
jpayne@69: def build_tzinfo(zone, fp):
jpayne@69:     head_fmt = '>4s c 15x 6l'
jpayne@69:     head_size = calcsize(head_fmt)
jpayne@69:     (magic, format, ttisgmtcnt, ttisstdcnt, leapcnt, timecnt,
jpayne@69:         typecnt, charcnt) = unpack(head_fmt, fp.read(head_size))
jpayne@69: 
jpayne@69:     # Make sure it is a tzfile(5) file
jpayne@69:     assert magic == _byte_string('TZif'), 'Got magic %s' % repr(magic)
jpayne@69: 
jpayne@69:     # Read out the transition times, localtime indices and ttinfo structures.
jpayne@69:     data_fmt = '>%(timecnt)dl %(timecnt)dB %(ttinfo)s %(charcnt)ds' % dict(
jpayne@69:         timecnt=timecnt, ttinfo='lBB' * typecnt, charcnt=charcnt)
jpayne@69:     data_size = calcsize(data_fmt)
jpayne@69:     data = unpack(data_fmt, fp.read(data_size))
jpayne@69: 
jpayne@69:     # make sure we unpacked the right number of values
jpayne@69:     assert len(data) == 2 * timecnt + 3 * typecnt + 1
jpayne@69:     transitions = [memorized_datetime(trans)
jpayne@69:                    for trans in data[:timecnt]]
jpayne@69:     lindexes = list(data[timecnt:2 * timecnt])
jpayne@69:     ttinfo_raw = data[2 * timecnt:-1]
jpayne@69:     tznames_raw = data[-1]
jpayne@69:     del data
jpayne@69: 
jpayne@69:     # Process ttinfo into separate structs
jpayne@69:     ttinfo = []
jpayne@69:     tznames = {}
jpayne@69:     i = 0
jpayne@69:     while i < len(ttinfo_raw):
jpayne@69:         # have we looked up this timezone name yet?
jpayne@69:         tzname_offset = ttinfo_raw[i + 2]
jpayne@69:         if tzname_offset not in tznames:
jpayne@69:             nul = tznames_raw.find(_NULL, tzname_offset)
jpayne@69:             if nul < 0:
jpayne@69:                 nul = len(tznames_raw)
jpayne@69:             tznames[tzname_offset] = _std_string(
jpayne@69:                 tznames_raw[tzname_offset:nul])
jpayne@69:         ttinfo.append((ttinfo_raw[i],
jpayne@69:                        bool(ttinfo_raw[i + 1]),
jpayne@69:                        tznames[tzname_offset]))
jpayne@69:         i += 3
jpayne@69: 
jpayne@69:     # Now build the timezone object
jpayne@69:     if len(ttinfo) == 1 or len(transitions) == 0:
jpayne@69:         ttinfo[0][0], ttinfo[0][2]
jpayne@69:         cls = type(zone, (StaticTzInfo,), dict(
jpayne@69:             zone=zone,
jpayne@69:             _utcoffset=memorized_timedelta(ttinfo[0][0]),
jpayne@69:             _tzname=ttinfo[0][2]))
jpayne@69:     else:
jpayne@69:         # Early dates use the first standard time ttinfo
jpayne@69:         i = 0
jpayne@69:         while ttinfo[i][1]:
jpayne@69:             i += 1
jpayne@69:         if ttinfo[i] == ttinfo[lindexes[0]]:
jpayne@69:             transitions[0] = datetime.min
jpayne@69:         else:
jpayne@69:             transitions.insert(0, datetime.min)
jpayne@69:             lindexes.insert(0, i)
jpayne@69: 
jpayne@69:         # calculate transition info
jpayne@69:         transition_info = []
jpayne@69:         for i in range(len(transitions)):
jpayne@69:             inf = ttinfo[lindexes[i]]
jpayne@69:             utcoffset = inf[0]
jpayne@69:             if not inf[1]:
jpayne@69:                 dst = 0
jpayne@69:             else:
jpayne@69:                 for j in range(i - 1, -1, -1):
jpayne@69:                     prev_inf = ttinfo[lindexes[j]]
jpayne@69:                     if not prev_inf[1]:
jpayne@69:                         break
jpayne@69:                 dst = inf[0] - prev_inf[0]  # dst offset
jpayne@69: 
jpayne@69:                 # Bad dst? Look further. DST > 24 hours happens when
jpayne@69:                 # a timzone has moved across the international dateline.
jpayne@69:                 if dst <= 0 or dst > 3600 * 3:
jpayne@69:                     for j in range(i + 1, len(transitions)):
jpayne@69:                         stdinf = ttinfo[lindexes[j]]
jpayne@69:                         if not stdinf[1]:
jpayne@69:                             dst = inf[0] - stdinf[0]
jpayne@69:                             if dst > 0:
jpayne@69:                                 break  # Found a useful std time.
jpayne@69: 
jpayne@69:             tzname = inf[2]
jpayne@69: 
jpayne@69:             # Round utcoffset and dst to the nearest minute or the
jpayne@69:             # datetime library will complain. Conversions to these timezones
jpayne@69:             # might be up to plus or minus 30 seconds out, but it is
jpayne@69:             # the best we can do.
jpayne@69:             utcoffset = int((utcoffset + 30) // 60) * 60
jpayne@69:             dst = int((dst + 30) // 60) * 60
jpayne@69:             transition_info.append(memorized_ttinfo(utcoffset, dst, tzname))
jpayne@69: 
jpayne@69:         cls = type(zone, (DstTzInfo,), dict(
jpayne@69:             zone=zone,
jpayne@69:             _utc_transition_times=transitions,
jpayne@69:             _transition_info=transition_info))
jpayne@69: 
jpayne@69:     return cls()
jpayne@69: 
jpayne@69: if __name__ == '__main__':
jpayne@69:     import os.path
jpayne@69:     from pprint import pprint
jpayne@69:     base = os.path.join(os.path.dirname(__file__), 'zoneinfo')
jpayne@69:     tz = build_tzinfo('Australia/Melbourne',
jpayne@69:                       open(os.path.join(base, 'Australia', 'Melbourne'), 'rb'))
jpayne@69:     tz = build_tzinfo('US/Eastern',
jpayne@69:                       open(os.path.join(base, 'US', 'Eastern'), 'rb'))
jpayne@69:     pprint(tz._utc_transition_times)