luatz/tzfile.lua

   1 local tz_info_mt = require "luatz.tzinfo".tz_info_mt
   2 local tt_info_mt = require "luatz.tzinfo".tt_info_mt
   3
   4 local read_int32be, read_int64be
   5
   6 -- luacheck: push std max
   7 if string.unpack then
   8         -- Only available in Lua 5.3+
   9         function read_int32be(fd)
  10                 local data, err = fd:read(4)
  11                 if data == nil then return nil, err end
  12                 return string.unpack(">i4", data)
  13         end
  14
  15         function read_int64be(fd)
  16                 local data, err = fd:read(8)
  17                 if data == nil then return nil, err end
  18                 return string.unpack(">i8", data)
  19         end
  20 else -- luacheck: pop
  21         function read_int32be(fd)
  22                 local data, err = fd:read(4)
  23                 if data == nil then return nil, err end
  24                 local o1, o2, o3, o4 = data:byte(1, 4)
  25
  26                 local unsigned = o4 + o3*2^8 + o2*2^16 + o1*2^24
  27                 if unsigned >= 2^31 then
  28                         return unsigned - 2^32
  29                 else
  30                         return unsigned
  31                 end
  32         end
  33
  34         function read_int64be(fd)
  35                 local data, err = fd:read(8)
  36                 if data == nil then return nil, err end
  37                 local o1, o2, o3, o4, o5, o6, o7, o8 = data:byte(1, 8)
  38
  39                 local unsigned = o8 + o7*2^8 + o6*2^16 + o5*2^24 + o4*2^32 + o3*2^40 + o2*2^48 + o1*2^56
  40                 if unsigned >= 2^63 then
  41                         return unsigned - 2^64
  42                 else
  43                         return unsigned
  44                 end
  45         end
  46 end
  47
  48 local function read_flags(fd, n)
  49         local data, err = fd:read(n)
  50         if data == nil then return nil, err end
  51
  52         local res = {}
  53         for i=1, n do
  54                 res[i] = data:byte(i,i) ~= 0
  55         end
  56         return res
  57 end
  58
  59 local fifteen_nulls = ("\0"):rep(15)
  60 local function read_tz(fd)
  61         assert(fd:read(4) == "TZif", "Invalid TZ file")
  62         local version = assert(fd:read(1))
  63         if version == "\0" or version == "2" or version == "3" then
  64                 local MIN_TIME = -2^32+1
  65
  66                 assert(assert(fd:read(15)) == fifteen_nulls, "Expected 15 nulls")
  67
  68                 -- The number of UTC/local indicators stored in the file.
  69                 local tzh_ttisgmtcnt = assert(read_int32be(fd))
  70
  71                 -- The number of standard/wall indicators stored in the file.
  72                 local tzh_ttisstdcnt = assert(read_int32be(fd))
  73
  74                 -- The number of leap seconds for which data is stored in the file.
  75                 local tzh_leapcnt = assert(read_int32be(fd))
  76
  77                 -- The number of "transition times" for which data is stored in the file.
  78                 local tzh_timecnt = assert(read_int32be(fd))
  79
  80                 -- The number of "local time types" for which data is stored in the file (must not be zero).
  81                 local tzh_typecnt = assert(read_int32be(fd))
  82
  83                 -- The number of characters of "timezone abbreviation strings" stored in the file.
  84                 local tzh_charcnt = assert(read_int32be(fd))
  85
  86                 local transition_times = {}
  87                 for i=1, tzh_timecnt do
  88                         transition_times[i] = assert(read_int32be(fd))
  89                 end
  90                 local transition_time_ind = {assert(fd:read(tzh_timecnt)):byte(1, -1)}
  91
  92                 local ttinfos = {}
  93                 for i=1, tzh_typecnt do
  94                         ttinfos[i] = {
  95                                 gmtoff = assert(read_int32be(fd));
  96                                 isdst  = assert(fd:read(1)) ~= "\0";
  97                                 abbrind = assert(fd:read(1)):byte();
  98                         }
  99                 end
 100
 101                 local abbreviations = assert(fd:read(tzh_charcnt))
 102
 103                 local leap_seconds = {} -- luacheck: ignore 241
 104                 for i=1, tzh_leapcnt do
 105                         leap_seconds[i] = {
 106                                 offset = assert(read_int32be(fd));
 107                                 n = assert(read_int32be(fd));
 108                         }
 109                 end
 110
 111                 local isstd = assert(read_flags(fd, tzh_ttisstdcnt))
 112
 113                 local isgmt = assert(read_flags(fd, tzh_ttisgmtcnt))
 114
 115                 local TZ
 116
 117                 if version == "2" or version == "3" then
 118                         --[[
 119                         For version-2-format timezone files, the above header and data is followed by a second header and data,
 120                         identical in format except that eight bytes are used for each transition time or leap-second time.
 121                         ]]
 122                         assert(fd:read(4) == "TZif")
 123                         assert(fd:read(1) == version)
 124                         assert(assert(fd:read(15)) == fifteen_nulls, "Expected 15 nulls")
 125
 126                         MIN_TIME = -2^64+1
 127
 128                         -- The number of UTC/local indicators stored in the file.
 129                         tzh_ttisgmtcnt = assert(read_int32be(fd))
 130
 131                         -- The number of standard/wall indicators stored in the file.
 132                         tzh_ttisstdcnt = assert(read_int32be(fd))
 133
 134                         -- The number of leap seconds for which data is stored in the file.
 135                         tzh_leapcnt = assert(read_int32be(fd))
 136
 137                         -- The number of "transition times" for which data is stored in the file.
 138                         tzh_timecnt = assert(read_int32be(fd))
 139
 140                         -- The number of "local time types" for which data is stored in the file (must not be zero).
 141                         tzh_typecnt = assert(read_int32be(fd))
 142
 143                         -- The number of characters of "timezone abbreviation strings" stored in the file.
 144                         tzh_charcnt = assert(read_int32be(fd))
 145
 146                         transition_times = {}
 147                         for i=1, tzh_timecnt do
 148                                 transition_times[i] = assert(read_int64be(fd))
 149                         end
 150                         transition_time_ind = {assert(fd:read(tzh_timecnt)):byte(1, -1)}
 151
 152                         ttinfos = {}
 153                         for i=1, tzh_typecnt do
 154                                 ttinfos[i] = {
 155                                         gmtoff = assert(read_int32be(fd));
 156                                         isdst  = assert(fd:read(1)) ~= "\0";
 157                                         abbrind = assert(fd:read(1)):byte();
 158                                 }
 159                         end
 160
 161                         abbreviations = assert(fd:read(tzh_charcnt))
 162
 163                         leap_seconds = {}
 164                         for i=1, tzh_leapcnt do
 165                                 leap_seconds[i] = {
 166                                         offset = assert(read_int64be(fd));
 167                                         n = assert(read_int32be(fd));
 168                                 }
 169                         end
 170
 171                         isstd = assert(read_flags(fd, tzh_ttisstdcnt))
 172
 173                         isgmt = assert(read_flags(fd, tzh_ttisgmtcnt))
 174
 175                         --[[
 176                         After the second header and data comes a newline-enclosed, POSIX-TZ-environment-variable-style string
 177                         for use in handling instants after the last transition time stored in the file
 178                         (with nothing between the newlines if there is no POSIX representation for such instants).
 179                         ]]
 180
 181                         --[[
 182                         For version-3-format time zone files, the POSIX-TZ-style string may
 183                         use two minor extensions to the POSIX TZ format, as described in newtzset (3).
 184                         First, the hours part of its transition times may be signed and range from
 185                         -167 through 167 instead of the POSIX-required unsigned values
 186                         from 0 through 24.  Second, DST is in effect all year if it starts
 187                         January 1 at 00:00 and ends December 31 at 24:00 plus the difference
 188                         between daylight saving and standard time.
 189                         ]]
 190
 191                         assert(assert(fd:read(1)) == "\n", "Expected newline at end of version 2 header")
 192
 193                         TZ = assert(fd:read("*l"))
 194                         if #TZ == 0 then
 195                                 TZ = nil
 196                         end
 197                 end
 198
 199                 for i=1, tzh_typecnt do
 200                         local v = ttinfos[i]
 201                         v.abbr = abbreviations:sub(v.abbrind+1, v.abbrind+3)
 202                         v.isstd = isstd[i] or false
 203                         v.isgmt = isgmt[i] or false
 204                         setmetatable(v, tt_info_mt)
 205                 end
 206
 207                 --[[
 208                 Use the first standard-time ttinfo structure in the file
 209                 (or simply the first ttinfo structure in the absence of a standard-time structure)
 210                 if either tzh_timecnt is zero or the time argument is less than the first transition time recorded in the file.
 211                 ]]
 212                 local first = 1
 213                 do
 214                         for i=1, tzh_ttisstdcnt do
 215                                 if isstd[i] then
 216                                         first = i
 217                                         break
 218                                 end
 219                         end
 220                 end
 221
 222                 local res = {
 223                         future = TZ;
 224                         [0] = {
 225                                 transition_time = MIN_TIME;
 226                                 info = ttinfos[first];
 227                         }
 228                 }
 229                 for i=1, tzh_timecnt do
 230                         res[i] = {
 231                                 transition_time = transition_times[i];
 232                                 info = ttinfos[transition_time_ind[i]+1];
 233                         }
 234                 end
 235                 return setmetatable(res, tz_info_mt)
 236         else
 237                 error("Unsupported version")
 238         end
 239 end
 240
 241 local function read_tzfile(path)
 242         local fd = assert(io.open(path, "rb"))
 243         local tzinfo = read_tz(fd)
 244         fd:close()
 245         return tzinfo
 246 end
 247
 248 return {
 249         read_tz = read_tz;
 250         read_tzfile = read_tzfile;
 251 }