| 1 |
// SPDX-License-Identifier: MPL-2.0 |
| 2 |
// (c) Hare authors <https://harelang.org> |
| 3 |
|
| 4 |
use bufio; |
| 5 |
use bytes; |
| 6 |
use encoding::utf8; |
| 7 |
use endian; |
| 8 |
use errors; |
| 9 |
use fs; |
| 10 |
use io; |
| 11 |
use os; |
| 12 |
use path; |
| 13 |
use strings; |
| 14 |
use time; |
| 15 |
use time::chrono; |
| 16 |
|
| 17 |
// Error concerning the Timezone database. |
| 18 |
export type tzdberror = !(invalidtzif | fs::error | io::error); |
| 19 |
|
| 20 |
// Invalid TZif data. |
| 21 |
export type invalidtzif = !void; |
| 22 |
|
| 23 |
// Loads a [[timezone]] from the system's TZDB (Timezone Database) as a |
| 24 |
// [[locality]]. Each call returns a new instance. The caller must free the |
| 25 |
// return value. See [[timezone_free]]. |
| 26 |
// |
| 27 |
// The given "TZID" shall be a TZDB identifier and acts as the relative filepath |
| 28 |
// to a TZif (Time Zone Information Format) file under [[TZDB_PATH]]. |
| 29 |
// For example, "Europe/Amsterdam" for "/usr/share/zoneinfo/Europe/Amsterdam". |
| 30 |
// |
| 31 |
// The loaded timezones are assigned the TZID as its name, the |
| 32 |
// [[time::chrono::utc]] timescale, and the [[EARTH_DAY]] day-length |
| 33 |
export fn tzdb(tzid: str) (locality | tzdberror | nomem) = { |
| 34 |
const filepath = match (path::init(TZDB_PATH, tzid)) { |
| 35 |
case let buf: path::buffer => |
| 36 |
yield buf; |
| 37 |
case let err: path::error => |
| 38 |
assert(err is path::too_long); |
| 39 |
return errors::noentry: fs::error; |
| 40 |
}; |
| 41 |
const file = os::open(path::string(&filepath))?; |
| 42 |
|
| 43 |
static let buf: [os::BUFSZ]u8 = [0...]; |
| 44 |
const bufstrm = bufio::init(file, buf, []); |
| 45 |
|
| 46 |
let ok = false; |
| 47 |
let name = strings::dup(tzid)?; |
| 48 |
defer if (!ok) free(name); |
| 49 |
let loc = alloc(timezone { |
| 50 |
tsc = &chrono::utc, |
| 51 |
daylength = EARTH_DAY, |
| 52 |
petz = PETZ_EMPTY, |
| 53 |
name = name, |
| 54 |
... |
| 55 |
})?; |
| 56 |
defer if (!ok) free(loc); |
| 57 |
match (load_tzif(&bufstrm, loc)) { |
| 58 |
case void => |
| 59 |
io::close(&bufstrm)?; |
| 60 |
io::close(file)?; |
| 61 |
ok = true; |
| 62 |
return loc; |
| 63 |
case invalidtzif => |
| 64 |
_ = io::close(&bufstrm); |
| 65 |
_ = io::close(file); |
| 66 |
return invalidtzif; |
| 67 |
case let err: io::error => |
| 68 |
_ = io::close(&bufstrm); |
| 69 |
_ = io::close(file); |
| 70 |
return err; |
| 71 |
}; |
| 72 |
}; |
| 73 |
|
| 74 |
// Loads TZif (Time Zone Information Format) data onto a given [[timezone]], |
| 75 |
// initializing the fields ".phases", ".transitions", and ".petz". |
| 76 |
// |
| 77 |
// Refer to "RFC 8536" for the TZif format. |
| 78 |
export fn load_tzif(h: io::handle, tz: *timezone) (void | invalidtzif | io::error | nomem) = { |
| 79 |
const buf1: [1]u8 = [0...]; |
| 80 |
const buf4: [4]u8 = [0...]; |
| 81 |
const buf8: [8]u8 = [0...]; |
| 82 |
const buf15: [15]u8 = [0...]; |
| 83 |
let ok = false; |
| 84 |
|
| 85 |
// test for magic "TZif" |
| 86 |
mustread(h, buf4)?; |
| 87 |
if (!bytes::equal(buf4, ['T', 'Z', 'i', 'f'])) { |
| 88 |
return invalidtzif; |
| 89 |
}; |
| 90 |
|
| 91 |
// read version |
| 92 |
mustread(h, buf1)?; |
| 93 |
const version = switch (buf1[0]) { |
| 94 |
case 0 => |
| 95 |
yield 1; |
| 96 |
case '2' => |
| 97 |
yield 2; |
| 98 |
case '3' => |
| 99 |
yield 3; |
| 100 |
case => |
| 101 |
return invalidtzif; |
| 102 |
}; |
| 103 |
|
| 104 |
// skip padding |
| 105 |
mustread(h, buf15)?; |
| 106 |
|
| 107 |
// read counts |
| 108 |
mustread(h, buf4)?; let isutcnt = endian::begetu32(buf4); |
| 109 |
mustread(h, buf4)?; let isstdcnt = endian::begetu32(buf4); |
| 110 |
mustread(h, buf4)?; let leapcnt = endian::begetu32(buf4); |
| 111 |
mustread(h, buf4)?; let timecnt = endian::begetu32(buf4); |
| 112 |
mustread(h, buf4)?; let typecnt = endian::begetu32(buf4); |
| 113 |
mustread(h, buf4)?; let charcnt = endian::begetu32(buf4); |
| 114 |
|
| 115 |
let is64 = false; |
| 116 |
if (version > 1) { |
| 117 |
is64 = true; |
| 118 |
|
| 119 |
// skip to the version 2 data |
| 120 |
const skip = ( |
| 121 |
// size of version 1 data block |
| 122 |
timecnt * 4 |
| 123 |
+ timecnt |
| 124 |
+ typecnt * 6 |
| 125 |
+ charcnt |
| 126 |
+ leapcnt * 8 |
| 127 |
+ isstdcnt |
| 128 |
+ isutcnt |
| 129 |
// size of version 2 header |
| 130 |
+ 20 |
| 131 |
); |
| 132 |
for (let i = 0z; i < skip; i += 1) { |
| 133 |
mustread(h, buf1)?; |
| 134 |
}; |
| 135 |
|
| 136 |
// read version 2 counts |
| 137 |
mustread(h, buf4)?; isutcnt = endian::begetu32(buf4); |
| 138 |
mustread(h, buf4)?; isstdcnt = endian::begetu32(buf4); |
| 139 |
mustread(h, buf4)?; leapcnt = endian::begetu32(buf4); |
| 140 |
mustread(h, buf4)?; timecnt = endian::begetu32(buf4); |
| 141 |
mustread(h, buf4)?; typecnt = endian::begetu32(buf4); |
| 142 |
mustread(h, buf4)?; charcnt = endian::begetu32(buf4); |
| 143 |
}; |
| 144 |
|
| 145 |
if (typecnt == 0 || charcnt == 0) { |
| 146 |
return invalidtzif; |
| 147 |
}; |
| 148 |
|
| 149 |
if (!(isutcnt == 0 || isutcnt == typecnt) |
| 150 |
&& (isstdcnt == 0 && isstdcnt == typecnt)) { |
| 151 |
return invalidtzif; |
| 152 |
}; |
| 153 |
|
| 154 |
const timesz = if (is64) 8 else 4; |
| 155 |
|
| 156 |
// read data |
| 157 |
|
| 158 |
const transition_times: []i64 = []; |
| 159 |
if (is64) { |
| 160 |
readitems8(h, &transition_times, timecnt)?; |
| 161 |
} else { |
| 162 |
readitems4(h, &transition_times, timecnt)?; |
| 163 |
}; |
| 164 |
defer free(transition_times); |
| 165 |
const zone_indicies: []u8 = []; |
| 166 |
defer free(zone_indicies); |
| 167 |
readbytes(h, &zone_indicies, timecnt)?; |
| 168 |
const zonedata: []u8 = []; |
| 169 |
defer free(zonedata); |
| 170 |
readbytes(h, &zonedata, typecnt * 6)?; |
| 171 |
const abbrdata: []u8 = []; |
| 172 |
defer free(abbrdata); |
| 173 |
readbytes(h, &abbrdata, charcnt)?; |
| 174 |
const leapdata: []u8 = []; |
| 175 |
defer free(leapdata); |
| 176 |
readbytes(h, &leapdata, leapcnt * (timesz: u32 + 4))?; |
| 177 |
const stdwalldata: []u8 = []; |
| 178 |
defer free(stdwalldata); |
| 179 |
readbytes(h, &stdwalldata, isstdcnt)?; |
| 180 |
const normlocaldata: []u8 = []; |
| 181 |
defer free(normlocaldata); |
| 182 |
readbytes(h, &normlocaldata, isutcnt)?; |
| 183 |
// read footer |
| 184 |
|
| 185 |
let footerdata: []u8 = []; |
| 186 |
defer free(footerdata); |
| 187 |
mustread(h, buf1)?; |
| 188 |
if (buf1[0] != 0x0A) { // '\n' newline |
| 189 |
return invalidtzif; |
| 190 |
}; |
| 191 |
for (true) { |
| 192 |
mustread(h, buf1)?; |
| 193 |
if (buf1[0] == 0x0A) { // '\n' newline |
| 194 |
break; |
| 195 |
}; |
| 196 |
if (buf1[0] == 0x0) { // cannot contain NUL |
| 197 |
return invalidtzif; |
| 198 |
}; |
| 199 |
append(footerdata, buf1...)?; |
| 200 |
}; |
| 201 |
const petz = match (strings::fromutf8(footerdata)) { |
| 202 |
case let s: str => |
| 203 |
yield s; |
| 204 |
case utf8::invalid => |
| 205 |
return invalidtzif; |
| 206 |
}; |
| 207 |
|
| 208 |
// assemble structured data |
| 209 |
|
| 210 |
// assemble zones |
| 211 |
let zones: []zonephase = []; |
| 212 |
defer if (!ok) free(zones); |
| 213 |
for (let i = 0z; i < typecnt; i += 1) { |
| 214 |
const idx = i * 6; |
| 215 |
const zone = zonephase { ... }; |
| 216 |
|
| 217 |
// offset |
| 218 |
const zoff = endian::begetu32(zonedata[idx..idx + 4]): i32; |
| 219 |
if (zoff == -2147483648) { // -2^31 |
| 220 |
return invalidtzif; |
| 221 |
}; |
| 222 |
zone.zoff = zoff * time::SECOND; |
| 223 |
|
| 224 |
// daylight saving time indicator |
| 225 |
zone.dst = switch (zonedata[idx + 4]) { |
| 226 |
case 1u8 => |
| 227 |
yield true; |
| 228 |
case 0u8 => |
| 229 |
yield false; |
| 230 |
case => |
| 231 |
return invalidtzif; |
| 232 |
}; |
| 233 |
|
| 234 |
// abbreviation |
| 235 |
const abbridx = zonedata[idx + 5]; |
| 236 |
if (abbridx < 0 || abbridx > (charcnt - 1)) { |
| 237 |
return invalidtzif; |
| 238 |
}; |
| 239 |
let bytes: []u8 = []; |
| 240 |
for (let j = abbridx; j < len(abbrdata); j += 1) { |
| 241 |
if (abbrdata[j] == 0x0) { |
| 242 |
bytes = abbrdata[abbridx..j]; |
| 243 |
break; |
| 244 |
}; |
| 245 |
}; |
| 246 |
if (len(bytes) == 0) { // no NUL encountered |
| 247 |
return invalidtzif; |
| 248 |
}; |
| 249 |
const abbr = match (strings::fromutf8(bytes)) { |
| 250 |
case let s: str => |
| 251 |
yield s; |
| 252 |
case utf8::invalid => |
| 253 |
return invalidtzif; |
| 254 |
}; |
| 255 |
zone.abbr = strings::dup(abbr)?; |
| 256 |
|
| 257 |
if (append(zones, zone) is nomem) { |
| 258 |
free(zone.abbr); |
| 259 |
return nomem; |
| 260 |
}; |
| 261 |
}; |
| 262 |
|
| 263 |
// assemble transitions |
| 264 |
let transitions: []zonetransition = []; |
| 265 |
defer if (!ok) free(transitions); |
| 266 |
|
| 267 |
for (let i = 0z; i < timecnt; i += 1) { |
| 268 |
const zoneindex = zone_indicies[i]; |
| 269 |
if (zoneindex < 0 || zoneindex > (typecnt - 1)) { |
| 270 |
return invalidtzif; |
| 271 |
}; |
| 272 |
|
| 273 |
const tx = zonetransition { |
| 274 |
when = time::instant { |
| 275 |
sec = transition_times[i], |
| 276 |
... |
| 277 |
}, |
| 278 |
zoneindex = zoneindex, |
| 279 |
}; |
| 280 |
|
| 281 |
// stdwalldata and normlocaldata have been omitted, |
| 282 |
// until they show their utility. |
| 283 |
|
| 284 |
append(transitions, tx)?; |
| 285 |
}; |
| 286 |
|
| 287 |
// save and return data |
| 288 |
tz.phases = zones; |
| 289 |
tz.transitions = transitions; |
| 290 |
|
| 291 |
if (parse_petz(&tz.petz, petz) is invalidpetzstr) { |
| 292 |
return invalidtzif; |
| 293 |
}; |
| 294 |
|
| 295 |
ok = true; |
| 296 |
}; |
| 297 |
|
| 298 |
fn mustread(h: io::handle, buf: []u8) (void | invalidtzif | io::error) = { |
| 299 |
match (io::readall(h, buf)) { |
| 300 |
case let err: io::error => |
| 301 |
return err; |
| 302 |
case io::EOF => |
| 303 |
return invalidtzif; |
| 304 |
case size => |
| 305 |
return; |
| 306 |
}; |
| 307 |
}; |
| 308 |
|
| 309 |
fn readbytes( |
| 310 |
h: io::handle, |
| 311 |
items: *[]u8, |
| 312 |
n: size, |
| 313 |
) (void | invalidtzif | io::error | nomem) = { |
| 314 |
const buf: [1]u8 = [0]; |
| 315 |
for (let i = 0z; i < n; i += 1) { |
| 316 |
mustread(h, buf)?; |
| 317 |
const it = buf[0]; |
| 318 |
append(items, it)?; |
| 319 |
}; |
| 320 |
}; |
| 321 |
|
| 322 |
fn readitems8( |
| 323 |
h: io::handle, |
| 324 |
items: *[]i64, |
| 325 |
n: size, |
| 326 |
) (void | invalidtzif | io::error | nomem) = { |
| 327 |
const buf: [8]u8 = [0...]; |
| 328 |
for (let i = 0z; i < n; i += 1) { |
| 329 |
mustread(h, buf)?; |
| 330 |
const it = endian::begetu64(buf): i64; |
| 331 |
append(items, it)?; |
| 332 |
}; |
| 333 |
}; |
| 334 |
|
| 335 |
fn readitems4( |
| 336 |
h: io::handle, |
| 337 |
items: *[]i64, |
| 338 |
n: size, |
| 339 |
) (void | invalidtzif | io::error | nomem) = { |
| 340 |
const buf: [4]u8 = [0...]; |
| 341 |
for (let i = 0z; i < n; i += 1) { |
| 342 |
mustread(h, buf)?; |
| 343 |
const it = endian::begetu32(buf): i64; |
| 344 |
append(items, it)?; |
| 345 |
}; |
| 346 |
}; |