| ... | @@ -38,70 +38,71 @@ pub const Tz = struct { | ... | @@ -38,70 +38,71 @@ pub const Tz = struct { |
| 38 | transitions: []const Transition, | 38 | transitions: []const Transition, |
| 39 | timetypes: []const Timetype, | 39 | timetypes: []const Timetype, |
| 40 | leapseconds: []const Leapsecond, | 40 | leapseconds: []const Leapsecond, |
| 41 | footer: []const u8, | 41 | footer: ?[]const u8, |
| 42 | | 42 | |
| 43 | pub fn parse(allocator: std.mem.Allocator, reader: anytype) !Tz { | 43 | const Header = extern struct { |
| 44 | const Header = extern struct { | 44 | magic: [4]u8, |
| 45 | magic: [4]u8, | 45 | version: u8, |
| 46 | version: u8, | 46 | reserved: [15]u8, |
| 47 | reserved: [15]u8, | 47 | counts: extern struct { |
| 48 | }; | | |
| 49 | | | |
| 50 | const Counts = extern struct { | | |
| 51 | isutcnt: u32, | 48 | isutcnt: u32, |
| 52 | isstdcnt: u32, | 49 | isstdcnt: u32, |
| 53 | leapcnt: u32, | 50 | leapcnt: u32, |
| 54 | timecnt: u32, | 51 | timecnt: u32, |
| 55 | typecnt: u32, | 52 | typecnt: u32, |
| 56 | charcnt: u32, | 53 | charcnt: u32, |
| 57 | }; | 54 | }, |
| | 55 | }; |
| 58 | | 56 | |
| 59 | // Parse and skip the legacy header and data | 57 | pub fn parse(allocator: std.mem.Allocator, reader: anytype) !Tz { |
| 60 | { | 58 | var legacy_header = try reader.readStruct(Header); |
| 61 | const header = try reader.readStruct(Header); | 59 | if (!std.mem.eql(u8, &legacy_header.magic, "TZif")) return error.BadHeader; |
| 62 | if (!std.mem.eql(u8, &header.magic, "TZif")) return error.BadHeader; | 60 | if (legacy_header.version != 0 and legacy_header.version != '2' and legacy_header.version != '3') return error.BadVersion; |
| 63 | if (header.version == 0) return error.UnsupportedLegacyFormat; | | |
| 64 | if (header.version != '2' and header.version != '3') return error.BadVersion; | | |
| 65 | | 61 | |
| 66 | var counts = try reader.readStruct(Counts); | 62 | if (builtin.target.cpu.arch.endian() != std.builtin.Endian.Big) { |
| 67 | if (builtin.target.cpu.arch.endian() != std.builtin.Endian.Big) { | 63 | std.mem.bswapAllFields(@TypeOf(legacy_header.counts), &legacy_header.counts); |
| 68 | std.mem.bswapAllFields(Counts, &counts); | 64 | } |
| 69 | } | | |
| 70 | | 65 | |
| 71 | const skipv = counts.timecnt * 5 + counts.typecnt * 6 + counts.charcnt + counts.leapcnt * 8 + counts.isstdcnt + counts.isutcnt; | 66 | if (legacy_header.version == 0) { |
| | 67 | return parseBlock(allocator, reader, legacy_header, true); |
| | 68 | } else { |
| | 69 | // If the format is modern, just skip over the legacy data |
| | 70 | const skipv = legacy_header.counts.timecnt * 5 + legacy_header.counts.typecnt * 6 + legacy_header.counts.charcnt + legacy_header.counts.leapcnt * 8 + legacy_header.counts.isstdcnt + legacy_header.counts.isutcnt; |
| 72 | try reader.skipBytes(skipv, .{}); | 71 | try reader.skipBytes(skipv, .{}); |
| 73 | } | | |
| 74 | | 72 | |
| 75 | const header = try reader.readStruct(Header); | 73 | var header = try reader.readStruct(Header); |
| 76 | if (!std.mem.eql(u8, &header.magic, "TZif")) return error.BadHeader; | 74 | if (!std.mem.eql(u8, &header.magic, "TZif")) return error.BadHeader; |
| 77 | if (header.version != '2' and header.version != '3') return error.BadVersion; | 75 | if (header.version != '2' and header.version != '3') return error.BadVersion; |
| | 76 | if (builtin.target.cpu.arch.endian() != std.builtin.Endian.Big) { |
| | 77 | std.mem.bswapAllFields(@TypeOf(header.counts), &header.counts); |
| | 78 | } |
| 78 | | 79 | |
| 79 | var counts = try reader.readStruct(Counts); | 80 | return parseBlock(allocator, reader, header, false); |
| 80 | if (builtin.target.cpu.arch.endian() != std.builtin.Endian.Big) { | | |
| 81 | std.mem.bswapAllFields(Counts, &counts); | | |
| 82 | } | 81 | } |
| | 82 | } |
| 83 | | 83 | |
| 84 | if (counts.isstdcnt != 0 and counts.isstdcnt != counts.typecnt) return error.Malformed; // rfc8536: isstdcnt [...] MUST either be zero or equal to "typecnt" | 84 | fn parseBlock(allocator: std.mem.Allocator, reader: anytype, header: Header, legacy: bool) !Tz { |
| 85 | if (counts.isutcnt != 0 and counts.isutcnt != counts.typecnt) return error.Malformed; // rfc8536: isutcnt [...] MUST either be zero or equal to "typecnt" | 85 | if (header.counts.isstdcnt != 0 and header.counts.isstdcnt != header.counts.typecnt) return error.Malformed; // rfc8536: isstdcnt [...] MUST either be zero or equal to "typecnt" |
| 86 | if (counts.typecnt == 0) return error.Malformed; // rfc8536: typecnt [...] MUST NOT be zero | 86 | if (header.counts.isutcnt != 0 and header.counts.isutcnt != header.counts.typecnt) return error.Malformed; // rfc8536: isutcnt [...] MUST either be zero or equal to "typecnt" |
| 87 | if (counts.charcnt == 0) return error.Malformed; // rfc8536: charcnt [...] MUST NOT be zero | 87 | if (header.counts.typecnt == 0) return error.Malformed; // rfc8536: typecnt [...] MUST NOT be zero |
| 88 | if (counts.charcnt > 256 + 6) return error.Malformed; // Not explicitly banned by rfc8536 but nonsensical | 88 | if (header.counts.charcnt == 0) return error.Malformed; // rfc8536: charcnt [...] MUST NOT be zero |
| | 89 | if (header.counts.charcnt > 256 + 6) return error.Malformed; // Not explicitly banned by rfc8536 but nonsensical |
| 89 | | 90 | |
| 90 | var leapseconds = try allocator.alloc(Leapsecond, counts.leapcnt); | 91 | var leapseconds = try allocator.alloc(Leapsecond, header.counts.leapcnt); |
| 91 | errdefer allocator.free(leapseconds); | 92 | errdefer allocator.free(leapseconds); |
| 92 | var transitions = try allocator.alloc(Transition, counts.timecnt); | 93 | var transitions = try allocator.alloc(Transition, header.counts.timecnt); |
| 93 | errdefer allocator.free(transitions); | 94 | errdefer allocator.free(transitions); |
| 94 | var timetypes = try allocator.alloc(Timetype, counts.typecnt); | 95 | var timetypes = try allocator.alloc(Timetype, header.counts.typecnt); |
| 95 | errdefer allocator.free(timetypes); | 96 | errdefer allocator.free(timetypes); |
| 96 | | 97 | |
| 97 | // Parse transition types | 98 | // Parse transition types |
| 98 | var i: usize = 0; | 99 | var i: usize = 0; |
| 99 | while (i < counts.timecnt) : (i += 1) { | 100 | while (i < header.counts.timecnt) : (i += 1) { |
| 100 | transitions[i].ts = try reader.readIntBig(i64); | 101 | transitions[i].ts = if (legacy) try reader.readIntBig(i32) else try reader.readIntBig(i64); |
| 101 | } | 102 | } |
| 102 | | 103 | |
| 103 | i = 0; | 104 | i = 0; |
| 104 | while (i < counts.timecnt) : (i += 1) { | 105 | while (i < header.counts.timecnt) : (i += 1) { |
| 105 | const tt = try reader.readByte(); | 106 | const tt = try reader.readByte(); |
| 106 | if (tt >= timetypes.len) return error.Malformed; // rfc8536: Each type index MUST be in the range [0, "typecnt" - 1] | 107 | if (tt >= timetypes.len) return error.Malformed; // rfc8536: Each type index MUST be in the range [0, "typecnt" - 1] |
| 107 | transitions[i].timetype = &timetypes[tt]; | 108 | transitions[i].timetype = &timetypes[tt]; |
| ... | @@ -109,13 +110,13 @@ pub const Tz = struct { | ... | @@ -109,13 +110,13 @@ pub const Tz = struct { |
| 109 | | 110 | |
| 110 | // Parse time types | 111 | // Parse time types |
| 111 | i = 0; | 112 | i = 0; |
| 112 | while (i < counts.typecnt) : (i += 1) { | 113 | while (i < header.counts.typecnt) : (i += 1) { |
| 113 | const offset = try reader.readIntBig(i32); | 114 | const offset = try reader.readIntBig(i32); |
| 114 | if (offset < -2147483648) return error.Malformed; // rfc8536: utoff [...] MUST NOT be -2**31 | 115 | if (offset < -2147483648) return error.Malformed; // rfc8536: utoff [...] MUST NOT be -2**31 |
| 115 | const dst = try reader.readByte(); | 116 | const dst = try reader.readByte(); |
| 116 | if (dst != 0 and dst != 1) return error.Malformed; // rfc8536: (is)dst [...] The value MUST be 0 or 1. | 117 | if (dst != 0 and dst != 1) return error.Malformed; // rfc8536: (is)dst [...] The value MUST be 0 or 1. |
| 117 | const idx = try reader.readByte(); | 118 | const idx = try reader.readByte(); |
| 118 | if (idx > counts.charcnt - 1) return error.Malformed; // rfc8536: (desig)idx [...] Each index MUST be in the range [0, "charcnt" - 1] | 119 | if (idx > header.counts.charcnt - 1) return error.Malformed; // rfc8536: (desig)idx [...] Each index MUST be in the range [0, "charcnt" - 1] |
| 119 | timetypes[i] = .{ | 120 | timetypes[i] = .{ |
| 120 | .offset = offset, | 121 | .offset = offset, |
| 121 | .flags = dst, | 122 | .flags = dst, |
| ... | @@ -127,8 +128,8 @@ pub const Tz = struct { | ... | @@ -127,8 +128,8 @@ pub const Tz = struct { |
| 127 | } | 128 | } |
| 128 | | 129 | |
| 129 | var designators_data: [256 + 6]u8 = undefined; | 130 | var designators_data: [256 + 6]u8 = undefined; |
| 130 | try reader.readNoEof(designators_data[0..counts.charcnt]); | 131 | try reader.readNoEof(designators_data[0..header.counts.charcnt]); |
| 131 | const designators = designators_data[0..counts.charcnt]; | 132 | const designators = designators_data[0..header.counts.charcnt]; |
| 132 | if (designators[designators.len - 1] != 0) return error.Malformed; // rfc8536: charcnt [...] includes the trailing NUL (0x00) octet | 133 | if (designators[designators.len - 1] != 0) return error.Malformed; // rfc8536: charcnt [...] includes the trailing NUL (0x00) octet |
| 133 | | 134 | |
| 134 | // Iterate through the timetypes again, setting the designator names | 135 | // Iterate through the timetypes again, setting the designator names |
| ... | @@ -142,8 +143,8 @@ pub const Tz = struct { | ... | @@ -142,8 +143,8 @@ pub const Tz = struct { |
| 142 | | 143 | |
| 143 | // Parse leap seconds | 144 | // Parse leap seconds |
| 144 | i = 0; | 145 | i = 0; |
| 145 | while (i < counts.leapcnt) : (i += 1) { | 146 | while (i < header.counts.leapcnt) : (i += 1) { |
| 146 | const occur = try reader.readIntBig(i64); | 147 | const occur: i64 = if (legacy) try reader.readIntBig(i32) else try reader.readIntBig(i64); |
| 147 | if (occur < 0) return error.Malformed; // rfc8536: occur [...] MUST be nonnegative | 148 | if (occur < 0) return error.Malformed; // rfc8536: occur [...] MUST be nonnegative |
| 148 | if (i > 0 and leapseconds[i - 1].occurrence + 2419199 > occur) return error.Malformed; // rfc8536: occur [...] each later value MUST be at least 2419199 greater than the previous value | 149 | if (i > 0 and leapseconds[i - 1].occurrence + 2419199 > occur) return error.Malformed; // rfc8536: occur [...] each later value MUST be at least 2419199 greater than the previous value |
| 149 | if (occur > std.math.maxInt(i48)) return error.Malformed; // Unreasonably far into the future | 150 | if (occur > std.math.maxInt(i48)) return error.Malformed; // Unreasonably far into the future |
| ... | @@ -161,7 +162,7 @@ pub const Tz = struct { | ... | @@ -161,7 +162,7 @@ pub const Tz = struct { |
| 161 | | 162 | |
| 162 | // Parse standard/wall indicators | 163 | // Parse standard/wall indicators |
| 163 | i = 0; | 164 | i = 0; |
| 164 | while (i < counts.isstdcnt) : (i += 1) { | 165 | while (i < header.counts.isstdcnt) : (i += 1) { |
| 165 | const stdtime = try reader.readByte(); | 166 | const stdtime = try reader.readByte(); |
| 166 | if (stdtime == 1) { | 167 | if (stdtime == 1) { |
| 167 | timetypes[i].flags |= 0x02; | 168 | timetypes[i].flags |= 0x02; |
| ... | @@ -170,7 +171,7 @@ pub const Tz = struct { | ... | @@ -170,7 +171,7 @@ pub const Tz = struct { |
| 170 | | 171 | |
| 171 | // Parse UT/local indicators | 172 | // Parse UT/local indicators |
| 172 | i = 0; | 173 | i = 0; |
| 173 | while (i < counts.isutcnt) : (i += 1) { | 174 | while (i < header.counts.isutcnt) : (i += 1) { |
| 174 | const ut = try reader.readByte(); | 175 | const ut = try reader.readByte(); |
| 175 | if (ut == 1) { | 176 | if (ut == 1) { |
| 176 | timetypes[i].flags |= 0x04; | 177 | timetypes[i].flags |= 0x04; |
| ... | @@ -178,29 +179,34 @@ pub const Tz = struct { | ... | @@ -178,29 +179,34 @@ pub const Tz = struct { |
| 178 | } | 179 | } |
| 179 | } | 180 | } |
| 180 | | 181 | |
| 181 | if ((try reader.readByte()) != '\n') return error.Malformed; // An rfc8536 footer must start with a newline | | |
| 182 | | | |
| 183 | // Footer | 182 | // Footer |
| 184 | var footerdata_buf: [128]u8 = undefined; | 183 | var footer: ?[]u8 = null; |
| 185 | const footer = reader.readUntilDelimiter(&footerdata_buf, '\n') catch |err| switch (err) { | 184 | if (!legacy) { |
| 186 | error.StreamTooLong => return error.OverlargeFooter, // Read more than 128 bytes, much larger than any reasonable POSIX TZ string | 185 | if ((try reader.readByte()) != '\n') return error.Malformed; // An rfc8536 footer must start with a newline |
| 187 | else => return err, | 186 | var footerdata_buf: [128]u8 = undefined; |
| 188 | }; | 187 | const footer_mem = reader.readUntilDelimiter(&footerdata_buf, '\n') catch |err| switch (err) { |
| 189 | | 188 | error.StreamTooLong => return error.OverlargeFooter, // Read more than 128 bytes, much larger than any reasonable POSIX TZ string |
| 190 | const footer_dup = try allocator.dupe(u8, footer); | 189 | else => return err, |
| 191 | errdefer allocator.free(footer_dup); | 190 | }; |
| | 191 | if (footer_mem.len != 0) { |
| | 192 | footer = try allocator.dupe(u8, footer_mem); |
| | 193 | } |
| | 194 | } |
| | 195 | errdefer if (footer) |ft| allocator.free(ft); |
| 192 | | 196 | |
| 193 | return Tz{ | 197 | return Tz{ |
| 194 | .allocator = allocator, | 198 | .allocator = allocator, |
| 195 | .transitions = transitions, | 199 | .transitions = transitions, |
| 196 | .timetypes = timetypes, | 200 | .timetypes = timetypes, |
| 197 | .leapseconds = leapseconds, | 201 | .leapseconds = leapseconds, |
| 198 | .footer = footer_dup, | 202 | .footer = footer, |
| 199 | }; | 203 | }; |
| 200 | } | 204 | } |
| 201 | | 205 | |
| 202 | pub fn deinit(self: *Tz) void { | 206 | pub fn deinit(self: *Tz) void { |
| 203 | self.allocator.free(self.footer); | 207 | if (self.footer) |footer| { |
| | 208 | self.allocator.free(footer); |
| | 209 | } |
| 204 | self.allocator.free(self.leapseconds); | 210 | self.allocator.free(self.leapseconds); |
| 205 | self.allocator.free(self.transitions); | 211 | self.allocator.free(self.transitions); |
| 206 | self.allocator.free(self.timetypes); | 212 | self.allocator.free(self.timetypes); |
| ... | @@ -231,3 +237,16 @@ test "fat" { | ... | @@ -231,3 +237,16 @@ test "fat" { |
| 231 | try std.testing.expect(std.mem.eql(u8, tz.transitions[3].timetype.name(), "+05")); | 237 | try std.testing.expect(std.mem.eql(u8, tz.transitions[3].timetype.name(), "+05")); |
| 232 | try std.testing.expectEqual(tz.transitions[4].ts, 1268251224); // 2010-03-10 20:00:00 UTC | 238 | try std.testing.expectEqual(tz.transitions[4].ts, 1268251224); // 2010-03-10 20:00:00 UTC |
| 233 | } | 239 | } |
| | 240 | |
| | 241 | test "legacy" { |
| | 242 | // Taken from Slackware 8.0, from 2001 |
| | 243 | const data = @embedFile("tz/europe_vatican.tzif"); |
| | 244 | var in_stream = std.io.fixedBufferStream(data); |
| | 245 | |
| | 246 | var tz = try std.Tz.parse(std.testing.allocator, in_stream.reader()); |
| | 247 | defer tz.deinit(); |
| | 248 | |
| | 249 | try std.testing.expectEqual(tz.transitions.len, 170); |
| | 250 | try std.testing.expect(std.mem.eql(u8, tz.transitions[69].timetype.name(), "CET")); |
| | 251 | try std.testing.expectEqual(tz.transitions[123].ts, 1414285200); // 2014-10-26 01:00:00 UTC |
| | 252 | } |