authorgravatar for hello@nektro.netMeghan Denny <hello@nektro.net> 2023-10-05 01:03:42 -07:00
committergravatar for hello@nektro.netMeghan Denny <hello@nektro.net> 2023-10-05 01:03:42 -07:00
log8a57b2c91d1554bdf5016adc7d8e81a4bf166fd1
treebdbcc2a2e8200f1ad7bc5838a0658bb56631ad08
parentf28cdca2832fcf3730b71b52669876f6c0d3728c

upgrade to unicode 15.1, closes #43


35 files changed, 474 insertions(+), 123 deletions(-)

README.md+1-1
......@@ -2,7 +2,7 @@
22
33Zig bindings for the Unicode Character Database
44
5Last updated as of Unicode 15.0.0
5Last updated as of Unicode 15.1.0
66
77http://www.unicode.org/reports/tr44/
88
scripts/CJKRadicals.zig+9-3
......@@ -10,7 +10,7 @@ pub usingnamespace common.Main(struct {
1010 \\pub const CJKRadical = struct {
1111 \\ number: u8,
1212 \\ simplified: bool,
13 \\ character: u21,
13 \\ character: ?u21,
1414 \\ ideograph: u21,
1515 \\};
1616 \\
......@@ -25,13 +25,19 @@ pub usingnamespace common.Main(struct {
2525
2626 pub fn exec(alloc: std.mem.Allocator, line: []const u8, writer: anytype) !void {
2727 _ = alloc;
28 var it = std.mem.tokenize(u8, line, "; ");
28 var it = std.mem.splitSequence(u8, line, "; ");
2929 var n = it.next().?;
3030 const s = std.mem.endsWith(u8, n, "'");
3131 const c = it.next().?;
3232 const i = it.next().?;
3333 n = std.mem.trimRight(u8, n, "'");
3434
35 try writer.print(" .{{ .number = {s}, .simplified = {}, .character = 0x{s}, .ideograph = 0x{s} }},\n", .{ n, s, c, i });
35 try writer.writeAll(" .{");
36 try writer.print(" .number = {s},", .{n});
37 try writer.print(" .simplified = {},", .{s});
38 try writer.writeAll(" .character =");
39 if (common.nullify(c)) |res| try common.printCodepoint(writer, res) else try writer.writeAll(" null,");
40 try writer.print(" .ideograph = 0x{s}", .{i});
41 try writer.writeAll(" },\n");
3642 }
3743});
scripts/DerivedCoreProperties.zig+1
......@@ -30,6 +30,7 @@ pub usingnamespace common.Main(struct {
3030 \\ Grapheme_Extend,
3131 \\ Grapheme_Base,
3232 \\ Grapheme_Link,
33 \\ InCB,
3334 \\ },
3435 \\};
3536 \\
scripts/LineBreak.zig+1
......@@ -16,6 +16,7 @@ pub usingnamespace common.Main(struct {
1616 \\ EM, EX, H2, H3, HL, HY, ID, IN, IS, JL,
1717 \\ JT, JV, NS, NU, OP, PO, PR, QU, RI, SA,
1818 \\ SG, SY, XX,
19 \\ AK, VI, AS, VF, AP,
1920 \\ },
2021 \\};
2122 \\
scripts/PropList.zig+3
......@@ -45,6 +45,9 @@ pub usingnamespace common.Main(struct {
4545 \\ Pattern_Syntax,
4646 \\ Prepended_Concatenation_Mark,
4747 \\ Regional_Indicator,
48 \\ IDS_Unary_Operator,
49 \\ ID_Compat_Math_Continue,
50 \\ ID_Compat_Math_Start,
4851 \\ },
4952 \\};
5053 \\
scripts/_common.zig+1-1
......@@ -3,7 +3,7 @@ const zfetch = @import("zfetch");
33const fmtValueLiteral = @import("fmt-valueliteral").fmtValueLiteral;
44const ansi = @import("ansi");
55
6pub const version = "15.0.0";
6pub const version = "15.1.0";
77
88pub fn Main(comptime T: type) type {
99 comptime std.debug.assert(@hasDecl(T, "source_file"));
src/arabic_shaping.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/ArabicShaping.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/ArabicShaping.txt
55//
66// zig fmt: off
77
src/bidi_brackets.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/BidiBrackets.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/BidiBrackets.txt
55//
66// zig fmt: off
77
src/bidi_mirroring.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/BidiMirroring.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/BidiMirroring.txt
55//
66// zig fmt: off
77
src/blocks.zig+2-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/Blocks.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/Blocks.txt
55//
66// zig fmt: off
77
......@@ -332,6 +332,7 @@ pub const data = [_]Block{
332332 .{ .from = 0x2B740, .to = 0x2B81F, .name = "CJK Unified Ideographs Extension D" },
333333 .{ .from = 0x2B820, .to = 0x2CEAF, .name = "CJK Unified Ideographs Extension E" },
334334 .{ .from = 0x2CEB0, .to = 0x2EBEF, .name = "CJK Unified Ideographs Extension F" },
335 .{ .from = 0x2EBF0, .to = 0x2EE5F, .name = "CJK Unified Ideographs Extension I" },
335336 .{ .from = 0x2F800, .to = 0x2FA1F, .name = "CJK Compatibility Ideographs Supplement" },
336337 .{ .from = 0x30000, .to = 0x3134F, .name = "CJK Unified Ideographs Extension G" },
337338 .{ .from = 0x31350, .to = 0x323AF, .name = "CJK Unified Ideographs Extension H" },
src/case_folding.zig+4-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/CaseFolding.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/CaseFolding.txt
55//
66// zig fmt: off
77
......@@ -895,6 +895,7 @@ pub const data = [_]CaseFolding{
895895 .{ .code = 0x1FCC, .status = .S, .mapping = .{ .S = 0x1FC3 } },
896896 .{ .code = 0x1FD2, .status = .F, .mapping = .{ .F = &[_]u21{0x03B9,0x0308,0x0300,} } },
897897 .{ .code = 0x1FD3, .status = .F, .mapping = .{ .F = &[_]u21{0x03B9,0x0308,0x0301,} } },
898 .{ .code = 0x1FD3, .status = .S, .mapping = .{ .S = 0x0390 } },
898899 .{ .code = 0x1FD6, .status = .F, .mapping = .{ .F = &[_]u21{0x03B9,0x0342,} } },
899900 .{ .code = 0x1FD7, .status = .F, .mapping = .{ .F = &[_]u21{0x03B9,0x0308,0x0342,} } },
900901 .{ .code = 0x1FD8, .status = .C, .mapping = .{ .C = 0x1FD0 } },
......@@ -903,6 +904,7 @@ pub const data = [_]CaseFolding{
903904 .{ .code = 0x1FDB, .status = .C, .mapping = .{ .C = 0x1F77 } },
904905 .{ .code = 0x1FE2, .status = .F, .mapping = .{ .F = &[_]u21{0x03C5,0x0308,0x0300,} } },
905906 .{ .code = 0x1FE3, .status = .F, .mapping = .{ .F = &[_]u21{0x03C5,0x0308,0x0301,} } },
907 .{ .code = 0x1FE3, .status = .S, .mapping = .{ .S = 0x03B0 } },
906908 .{ .code = 0x1FE4, .status = .F, .mapping = .{ .F = &[_]u21{0x03C1,0x0313,} } },
907909 .{ .code = 0x1FE6, .status = .F, .mapping = .{ .F = &[_]u21{0x03C5,0x0342,} } },
908910 .{ .code = 0x1FE7, .status = .F, .mapping = .{ .F = &[_]u21{0x03C5,0x0308,0x0342,} } },
......@@ -1294,6 +1296,7 @@ pub const data = [_]CaseFolding{
12941296 .{ .code = 0xFB03, .status = .F, .mapping = .{ .F = &[_]u21{0x0066,0x0066,0x0069,} } },
12951297 .{ .code = 0xFB04, .status = .F, .mapping = .{ .F = &[_]u21{0x0066,0x0066,0x006C,} } },
12961298 .{ .code = 0xFB05, .status = .F, .mapping = .{ .F = &[_]u21{0x0073,0x0074,} } },
1299 .{ .code = 0xFB05, .status = .S, .mapping = .{ .S = 0xFB06 } },
12971300 .{ .code = 0xFB06, .status = .F, .mapping = .{ .F = &[_]u21{0x0073,0x0074,} } },
12981301 .{ .code = 0xFB13, .status = .F, .mapping = .{ .F = &[_]u21{0x0574,0x0576,} } },
12991302 .{ .code = 0xFB14, .status = .F, .mapping = .{ .F = &[_]u21{0x0574,0x0565,} } },
src/cjk_radicals.zig+8-3
......@@ -1,14 +1,14 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/CJKRadicals.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/CJKRadicals.txt
55//
66// zig fmt: off
77
88pub const CJKRadical = struct {
99 number: u8,
1010 simplified: bool,
11 character: u21,
11 character: ?u21,
1212 ideograph: u21,
1313};
1414
......@@ -181,7 +181,6 @@ pub const data = [_]CJKRadical{
181181 .{ .number = 160, .simplified = false, .character = 0x2F9F, .ideograph = 0x8F9B },
182182 .{ .number = 161, .simplified = false, .character = 0x2FA0, .ideograph = 0x8FB0 },
183183 .{ .number = 162, .simplified = false, .character = 0x2FA1, .ideograph = 0x8FB5 },
184 .{ .number = 162, .simplified = true, .character = 0x2ECC, .ideograph = 0x8FB6 },
185184 .{ .number = 163, .simplified = false, .character = 0x2FA2, .ideograph = 0x9091 },
186185 .{ .number = 164, .simplified = false, .character = 0x2FA3, .ideograph = 0x9149 },
187186 .{ .number = 165, .simplified = false, .character = 0x2FA4, .ideograph = 0x91C6 },
......@@ -208,6 +207,7 @@ pub const data = [_]CJKRadical{
208207 .{ .number = 181, .simplified = true, .character = 0x2EDA, .ideograph = 0x9875 },
209208 .{ .number = 182, .simplified = false, .character = 0x2FB5, .ideograph = 0x98A8 },
210209 .{ .number = 182, .simplified = true, .character = 0x2EDB, .ideograph = 0x98CE },
210 .{ .number = 182, .simplified = true, .character = null, .ideograph = 0x322C4 },
211211 .{ .number = 183, .simplified = false, .character = 0x2FB6, .ideograph = 0x98DB },
212212 .{ .number = 183, .simplified = true, .character = 0x2EDC, .ideograph = 0x98DE },
213213 .{ .number = 184, .simplified = false, .character = 0x2FB7, .ideograph = 0x98DF },
......@@ -243,14 +243,19 @@ pub const data = [_]CJKRadical{
243243 .{ .number = 206, .simplified = false, .character = 0x2FCD, .ideograph = 0x9F0E },
244244 .{ .number = 207, .simplified = false, .character = 0x2FCE, .ideograph = 0x9F13 },
245245 .{ .number = 208, .simplified = false, .character = 0x2FCF, .ideograph = 0x9F20 },
246 .{ .number = 208, .simplified = true, .character = null, .ideograph = 0x9F21 },
246247 .{ .number = 209, .simplified = false, .character = 0x2FD0, .ideograph = 0x9F3B },
247248 .{ .number = 210, .simplified = false, .character = 0x2FD1, .ideograph = 0x9F4A },
248249 .{ .number = 210, .simplified = true, .character = 0x2EEC, .ideograph = 0x9F50 },
250 .{ .number = 210, .simplified = true, .character = 0x2EEB, .ideograph = 0x6589 },
249251 .{ .number = 211, .simplified = false, .character = 0x2FD2, .ideograph = 0x9F52 },
250252 .{ .number = 211, .simplified = true, .character = 0x2EEE, .ideograph = 0x9F7F },
253 .{ .number = 211, .simplified = true, .character = 0x2EED, .ideograph = 0x6B6F },
251254 .{ .number = 212, .simplified = false, .character = 0x2FD3, .ideograph = 0x9F8D },
252255 .{ .number = 212, .simplified = true, .character = 0x2EF0, .ideograph = 0x9F99 },
256 .{ .number = 212, .simplified = true, .character = 0x2EEF, .ideograph = 0x7ADC },
253257 .{ .number = 213, .simplified = false, .character = 0x2FD4, .ideograph = 0x9F9C },
254258 .{ .number = 213, .simplified = true, .character = 0x2EF3, .ideograph = 0x9F9F },
259 .{ .number = 213, .simplified = true, .character = 0x2EF2, .ideograph = 0x4E80 },
255260 .{ .number = 214, .simplified = false, .character = 0x2FD5, .ideograph = 0x9FA0 },
256261};
src/composition_exclusions.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/CompositionExclusions.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/CompositionExclusions.txt
55//
66// zig fmt: off
77
src/derived_age.zig+4-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/DerivedAge.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/DerivedAge.txt
55//
66// zig fmt: off
77
......@@ -1730,4 +1730,7 @@ pub const data = [_]Age{
17301730 .{ .from = 0x1FAF7, .to = 0x1FAF8, .since = .{ 15,0 } },
17311731 .{ .from = 0x2B739, .to = 0x2B739, .since = .{ 15,0 } },
17321732 .{ .from = 0x31350, .to = 0x323AF, .since = .{ 15,0 } },
1733 .{ .from = 0x2FFC, .to = 0x2FFF, .since = .{ 15,1 } },
1734 .{ .from = 0x31EF, .to = 0x31EF, .since = .{ 15,1 } },
1735 .{ .from = 0x2EBF0, .to = 0x2EE5D, .since = .{ 15,1 } },
17331736};
src/derived_core_properties.zig+218-2
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/DerivedCoreProperties.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/DerivedCoreProperties.txt
55//
66// zig fmt: off
77
......@@ -28,6 +28,7 @@ pub const CoreProperty = struct {
2828 Grapheme_Extend,
2929 Grapheme_Base,
3030 Grapheme_Link,
31 InCB,
3132 },
3233};
3334
......@@ -1409,6 +1410,7 @@ pub const data = [_]CoreProperty{
14091410 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .Alphabetic },
14101411 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .Alphabetic },
14111412 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .Alphabetic },
1413 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .Alphabetic },
14121414 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .Alphabetic },
14131415 .{ .from = 0x30000, .to = 0x3134A, .prop = .Alphabetic },
14141416 .{ .from = 0x31350, .to = 0x323AF, .prop = .Alphabetic },
......@@ -6765,6 +6767,7 @@ pub const data = [_]CoreProperty{
67656767 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .ID_Start },
67666768 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .ID_Start },
67676769 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .ID_Start },
6770 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .ID_Start },
67686771 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .ID_Start },
67696772 .{ .from = 0x30000, .to = 0x3134A, .prop = .ID_Start },
67706773 .{ .from = 0x31350, .to = 0x323AF, .prop = .ID_Start },
......@@ -7335,6 +7338,7 @@ pub const data = [_]CoreProperty{
73357338 .{ .from = 0x1FE0, .to = 0x1FEC, .prop = .ID_Continue },
73367339 .{ .from = 0x1FF2, .to = 0x1FF4, .prop = .ID_Continue },
73377340 .{ .from = 0x1FF6, .to = 0x1FFC, .prop = .ID_Continue },
7341 .{ .from = 0x200C, .to = 0x200D, .prop = .ID_Continue },
73387342 .{ .from = 0x203F, .to = 0x2040, .prop = .ID_Continue },
73397343 .{ .from = 0x2054, .to = 0x2054, .prop = .ID_Continue },
73407344 .{ .from = 0x2071, .to = 0x2071, .prop = .ID_Continue },
......@@ -7401,6 +7405,7 @@ pub const data = [_]CoreProperty{
74017405 .{ .from = 0x309D, .to = 0x309E, .prop = .ID_Continue },
74027406 .{ .from = 0x309F, .to = 0x309F, .prop = .ID_Continue },
74037407 .{ .from = 0x30A1, .to = 0x30FA, .prop = .ID_Continue },
7408 .{ .from = 0x30FB, .to = 0x30FB, .prop = .ID_Continue },
74047409 .{ .from = 0x30FC, .to = 0x30FE, .prop = .ID_Continue },
74057410 .{ .from = 0x30FF, .to = 0x30FF, .prop = .ID_Continue },
74067411 .{ .from = 0x3105, .to = 0x312F, .prop = .ID_Continue },
......@@ -7580,6 +7585,7 @@ pub const data = [_]CoreProperty{
75807585 .{ .from = 0xFF21, .to = 0xFF3A, .prop = .ID_Continue },
75817586 .{ .from = 0xFF3F, .to = 0xFF3F, .prop = .ID_Continue },
75827587 .{ .from = 0xFF41, .to = 0xFF5A, .prop = .ID_Continue },
7588 .{ .from = 0xFF65, .to = 0xFF65, .prop = .ID_Continue },
75837589 .{ .from = 0xFF66, .to = 0xFF6F, .prop = .ID_Continue },
75847590 .{ .from = 0xFF70, .to = 0xFF70, .prop = .ID_Continue },
75857591 .{ .from = 0xFF71, .to = 0xFF9D, .prop = .ID_Continue },
......@@ -8104,6 +8110,7 @@ pub const data = [_]CoreProperty{
81048110 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .ID_Continue },
81058111 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .ID_Continue },
81068112 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .ID_Continue },
8113 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .ID_Continue },
81078114 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .ID_Continue },
81088115 .{ .from = 0x30000, .to = 0x3134A, .prop = .ID_Continue },
81098116 .{ .from = 0x31350, .to = 0x323AF, .prop = .ID_Continue },
......@@ -8847,6 +8854,7 @@ pub const data = [_]CoreProperty{
88478854 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .XID_Start },
88488855 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .XID_Start },
88498856 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .XID_Start },
8857 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .XID_Start },
88508858 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .XID_Start },
88518859 .{ .from = 0x30000, .to = 0x3134A, .prop = .XID_Start },
88528860 .{ .from = 0x31350, .to = 0x323AF, .prop = .XID_Start },
......@@ -9416,6 +9424,7 @@ pub const data = [_]CoreProperty{
94169424 .{ .from = 0x1FE0, .to = 0x1FEC, .prop = .XID_Continue },
94179425 .{ .from = 0x1FF2, .to = 0x1FF4, .prop = .XID_Continue },
94189426 .{ .from = 0x1FF6, .to = 0x1FFC, .prop = .XID_Continue },
9427 .{ .from = 0x200C, .to = 0x200D, .prop = .XID_Continue },
94199428 .{ .from = 0x203F, .to = 0x2040, .prop = .XID_Continue },
94209429 .{ .from = 0x2054, .to = 0x2054, .prop = .XID_Continue },
94219430 .{ .from = 0x2071, .to = 0x2071, .prop = .XID_Continue },
......@@ -9481,6 +9490,7 @@ pub const data = [_]CoreProperty{
94819490 .{ .from = 0x309D, .to = 0x309E, .prop = .XID_Continue },
94829491 .{ .from = 0x309F, .to = 0x309F, .prop = .XID_Continue },
94839492 .{ .from = 0x30A1, .to = 0x30FA, .prop = .XID_Continue },
9493 .{ .from = 0x30FB, .to = 0x30FB, .prop = .XID_Continue },
94849494 .{ .from = 0x30FC, .to = 0x30FE, .prop = .XID_Continue },
94859495 .{ .from = 0x30FF, .to = 0x30FF, .prop = .XID_Continue },
94869496 .{ .from = 0x3105, .to = 0x312F, .prop = .XID_Continue },
......@@ -9666,6 +9676,7 @@ pub const data = [_]CoreProperty{
96669676 .{ .from = 0xFF21, .to = 0xFF3A, .prop = .XID_Continue },
96679677 .{ .from = 0xFF3F, .to = 0xFF3F, .prop = .XID_Continue },
96689678 .{ .from = 0xFF41, .to = 0xFF5A, .prop = .XID_Continue },
9679 .{ .from = 0xFF65, .to = 0xFF65, .prop = .XID_Continue },
96699680 .{ .from = 0xFF66, .to = 0xFF6F, .prop = .XID_Continue },
96709681 .{ .from = 0xFF70, .to = 0xFF70, .prop = .XID_Continue },
96719682 .{ .from = 0xFF71, .to = 0xFF9D, .prop = .XID_Continue },
......@@ -10190,6 +10201,7 @@ pub const data = [_]CoreProperty{
1019010201 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .XID_Continue },
1019110202 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .XID_Continue },
1019210203 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .XID_Continue },
10204 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .XID_Continue },
1019310205 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .XID_Continue },
1019410206 .{ .from = 0x30000, .to = 0x3134A, .prop = .XID_Continue },
1019510207 .{ .from = 0x31350, .to = 0x323AF, .prop = .XID_Continue },
......@@ -11440,7 +11452,7 @@ pub const data = [_]CoreProperty{
1144011452 .{ .from = 0x2E80, .to = 0x2E99, .prop = .Grapheme_Base },
1144111453 .{ .from = 0x2E9B, .to = 0x2EF3, .prop = .Grapheme_Base },
1144211454 .{ .from = 0x2F00, .to = 0x2FD5, .prop = .Grapheme_Base },
11443 .{ .from = 0x2FF0, .to = 0x2FFB, .prop = .Grapheme_Base },
11455 .{ .from = 0x2FF0, .to = 0x2FFF, .prop = .Grapheme_Base },
1144411456 .{ .from = 0x3000, .to = 0x3000, .prop = .Grapheme_Base },
1144511457 .{ .from = 0x3001, .to = 0x3003, .prop = .Grapheme_Base },
1144611458 .{ .from = 0x3004, .to = 0x3004, .prop = .Grapheme_Base },
......@@ -11495,6 +11507,7 @@ pub const data = [_]CoreProperty{
1149511507 .{ .from = 0x3196, .to = 0x319F, .prop = .Grapheme_Base },
1149611508 .{ .from = 0x31A0, .to = 0x31BF, .prop = .Grapheme_Base },
1149711509 .{ .from = 0x31C0, .to = 0x31E3, .prop = .Grapheme_Base },
11510 .{ .from = 0x31EF, .to = 0x31EF, .prop = .Grapheme_Base },
1149811511 .{ .from = 0x31F0, .to = 0x31FF, .prop = .Grapheme_Base },
1149911512 .{ .from = 0x3200, .to = 0x321E, .prop = .Grapheme_Base },
1150011513 .{ .from = 0x3220, .to = 0x3229, .prop = .Grapheme_Base },
......@@ -12335,6 +12348,7 @@ pub const data = [_]CoreProperty{
1233512348 .{ .from = 0x2B740, .to = 0x2B81D, .prop = .Grapheme_Base },
1233612349 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .Grapheme_Base },
1233712350 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .Grapheme_Base },
12351 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .Grapheme_Base },
1233812352 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .Grapheme_Base },
1233912353 .{ .from = 0x30000, .to = 0x3134A, .prop = .Grapheme_Base },
1234012354 .{ .from = 0x31350, .to = 0x323AF, .prop = .Grapheme_Base },
......@@ -12398,4 +12412,206 @@ pub const data = [_]CoreProperty{
1239812412 .{ .from = 0x11D97, .to = 0x11D97, .prop = .Grapheme_Link },
1239912413 .{ .from = 0x11F41, .to = 0x11F41, .prop = .Grapheme_Link },
1240012414 .{ .from = 0x11F42, .to = 0x11F42, .prop = .Grapheme_Link },
12415 .{ .from = 0x094D, .to = 0x094D, .prop = .InCB },
12416 .{ .from = 0x09CD, .to = 0x09CD, .prop = .InCB },
12417 .{ .from = 0x0ACD, .to = 0x0ACD, .prop = .InCB },
12418 .{ .from = 0x0B4D, .to = 0x0B4D, .prop = .InCB },
12419 .{ .from = 0x0C4D, .to = 0x0C4D, .prop = .InCB },
12420 .{ .from = 0x0D4D, .to = 0x0D4D, .prop = .InCB },
12421 .{ .from = 0x0915, .to = 0x0939, .prop = .InCB },
12422 .{ .from = 0x0958, .to = 0x095F, .prop = .InCB },
12423 .{ .from = 0x0978, .to = 0x097F, .prop = .InCB },
12424 .{ .from = 0x0995, .to = 0x09A8, .prop = .InCB },
12425 .{ .from = 0x09AA, .to = 0x09B0, .prop = .InCB },
12426 .{ .from = 0x09B2, .to = 0x09B2, .prop = .InCB },
12427 .{ .from = 0x09B6, .to = 0x09B9, .prop = .InCB },
12428 .{ .from = 0x09DC, .to = 0x09DD, .prop = .InCB },
12429 .{ .from = 0x09DF, .to = 0x09DF, .prop = .InCB },
12430 .{ .from = 0x09F0, .to = 0x09F1, .prop = .InCB },
12431 .{ .from = 0x0A95, .to = 0x0AA8, .prop = .InCB },
12432 .{ .from = 0x0AAA, .to = 0x0AB0, .prop = .InCB },
12433 .{ .from = 0x0AB2, .to = 0x0AB3, .prop = .InCB },
12434 .{ .from = 0x0AB5, .to = 0x0AB9, .prop = .InCB },
12435 .{ .from = 0x0AF9, .to = 0x0AF9, .prop = .InCB },
12436 .{ .from = 0x0B15, .to = 0x0B28, .prop = .InCB },
12437 .{ .from = 0x0B2A, .to = 0x0B30, .prop = .InCB },
12438 .{ .from = 0x0B32, .to = 0x0B33, .prop = .InCB },
12439 .{ .from = 0x0B35, .to = 0x0B39, .prop = .InCB },
12440 .{ .from = 0x0B5C, .to = 0x0B5D, .prop = .InCB },
12441 .{ .from = 0x0B5F, .to = 0x0B5F, .prop = .InCB },
12442 .{ .from = 0x0B71, .to = 0x0B71, .prop = .InCB },
12443 .{ .from = 0x0C15, .to = 0x0C28, .prop = .InCB },
12444 .{ .from = 0x0C2A, .to = 0x0C39, .prop = .InCB },
12445 .{ .from = 0x0C58, .to = 0x0C5A, .prop = .InCB },
12446 .{ .from = 0x0D15, .to = 0x0D3A, .prop = .InCB },
12447 .{ .from = 0x0300, .to = 0x034E, .prop = .InCB },
12448 .{ .from = 0x0350, .to = 0x036F, .prop = .InCB },
12449 .{ .from = 0x0483, .to = 0x0487, .prop = .InCB },
12450 .{ .from = 0x0591, .to = 0x05BD, .prop = .InCB },
12451 .{ .from = 0x05BF, .to = 0x05BF, .prop = .InCB },
12452 .{ .from = 0x05C1, .to = 0x05C2, .prop = .InCB },
12453 .{ .from = 0x05C4, .to = 0x05C5, .prop = .InCB },
12454 .{ .from = 0x05C7, .to = 0x05C7, .prop = .InCB },
12455 .{ .from = 0x0610, .to = 0x061A, .prop = .InCB },
12456 .{ .from = 0x064B, .to = 0x065F, .prop = .InCB },
12457 .{ .from = 0x0670, .to = 0x0670, .prop = .InCB },
12458 .{ .from = 0x06D6, .to = 0x06DC, .prop = .InCB },
12459 .{ .from = 0x06DF, .to = 0x06E4, .prop = .InCB },
12460 .{ .from = 0x06E7, .to = 0x06E8, .prop = .InCB },
12461 .{ .from = 0x06EA, .to = 0x06ED, .prop = .InCB },
12462 .{ .from = 0x0711, .to = 0x0711, .prop = .InCB },
12463 .{ .from = 0x0730, .to = 0x074A, .prop = .InCB },
12464 .{ .from = 0x07EB, .to = 0x07F3, .prop = .InCB },
12465 .{ .from = 0x07FD, .to = 0x07FD, .prop = .InCB },
12466 .{ .from = 0x0816, .to = 0x0819, .prop = .InCB },
12467 .{ .from = 0x081B, .to = 0x0823, .prop = .InCB },
12468 .{ .from = 0x0825, .to = 0x0827, .prop = .InCB },
12469 .{ .from = 0x0829, .to = 0x082D, .prop = .InCB },
12470 .{ .from = 0x0859, .to = 0x085B, .prop = .InCB },
12471 .{ .from = 0x0898, .to = 0x089F, .prop = .InCB },
12472 .{ .from = 0x08CA, .to = 0x08E1, .prop = .InCB },
12473 .{ .from = 0x08E3, .to = 0x08FF, .prop = .InCB },
12474 .{ .from = 0x093C, .to = 0x093C, .prop = .InCB },
12475 .{ .from = 0x0951, .to = 0x0954, .prop = .InCB },
12476 .{ .from = 0x09BC, .to = 0x09BC, .prop = .InCB },
12477 .{ .from = 0x09FE, .to = 0x09FE, .prop = .InCB },
12478 .{ .from = 0x0A3C, .to = 0x0A3C, .prop = .InCB },
12479 .{ .from = 0x0ABC, .to = 0x0ABC, .prop = .InCB },
12480 .{ .from = 0x0B3C, .to = 0x0B3C, .prop = .InCB },
12481 .{ .from = 0x0C3C, .to = 0x0C3C, .prop = .InCB },
12482 .{ .from = 0x0C55, .to = 0x0C56, .prop = .InCB },
12483 .{ .from = 0x0CBC, .to = 0x0CBC, .prop = .InCB },
12484 .{ .from = 0x0D3B, .to = 0x0D3C, .prop = .InCB },
12485 .{ .from = 0x0E38, .to = 0x0E3A, .prop = .InCB },
12486 .{ .from = 0x0E48, .to = 0x0E4B, .prop = .InCB },
12487 .{ .from = 0x0EB8, .to = 0x0EBA, .prop = .InCB },
12488 .{ .from = 0x0EC8, .to = 0x0ECB, .prop = .InCB },
12489 .{ .from = 0x0F18, .to = 0x0F19, .prop = .InCB },
12490 .{ .from = 0x0F35, .to = 0x0F35, .prop = .InCB },
12491 .{ .from = 0x0F37, .to = 0x0F37, .prop = .InCB },
12492 .{ .from = 0x0F39, .to = 0x0F39, .prop = .InCB },
12493 .{ .from = 0x0F71, .to = 0x0F72, .prop = .InCB },
12494 .{ .from = 0x0F74, .to = 0x0F74, .prop = .InCB },
12495 .{ .from = 0x0F7A, .to = 0x0F7D, .prop = .InCB },
12496 .{ .from = 0x0F80, .to = 0x0F80, .prop = .InCB },
12497 .{ .from = 0x0F82, .to = 0x0F84, .prop = .InCB },
12498 .{ .from = 0x0F86, .to = 0x0F87, .prop = .InCB },
12499 .{ .from = 0x0FC6, .to = 0x0FC6, .prop = .InCB },
12500 .{ .from = 0x1037, .to = 0x1037, .prop = .InCB },
12501 .{ .from = 0x1039, .to = 0x103A, .prop = .InCB },
12502 .{ .from = 0x108D, .to = 0x108D, .prop = .InCB },
12503 .{ .from = 0x135D, .to = 0x135F, .prop = .InCB },
12504 .{ .from = 0x1714, .to = 0x1714, .prop = .InCB },
12505 .{ .from = 0x17D2, .to = 0x17D2, .prop = .InCB },
12506 .{ .from = 0x17DD, .to = 0x17DD, .prop = .InCB },
12507 .{ .from = 0x18A9, .to = 0x18A9, .prop = .InCB },
12508 .{ .from = 0x1939, .to = 0x193B, .prop = .InCB },
12509 .{ .from = 0x1A17, .to = 0x1A18, .prop = .InCB },
12510 .{ .from = 0x1A60, .to = 0x1A60, .prop = .InCB },
12511 .{ .from = 0x1A75, .to = 0x1A7C, .prop = .InCB },
12512 .{ .from = 0x1A7F, .to = 0x1A7F, .prop = .InCB },
12513 .{ .from = 0x1AB0, .to = 0x1ABD, .prop = .InCB },
12514 .{ .from = 0x1ABF, .to = 0x1ACE, .prop = .InCB },
12515 .{ .from = 0x1B34, .to = 0x1B34, .prop = .InCB },
12516 .{ .from = 0x1B6B, .to = 0x1B73, .prop = .InCB },
12517 .{ .from = 0x1BAB, .to = 0x1BAB, .prop = .InCB },
12518 .{ .from = 0x1BE6, .to = 0x1BE6, .prop = .InCB },
12519 .{ .from = 0x1C37, .to = 0x1C37, .prop = .InCB },
12520 .{ .from = 0x1CD0, .to = 0x1CD2, .prop = .InCB },
12521 .{ .from = 0x1CD4, .to = 0x1CE0, .prop = .InCB },
12522 .{ .from = 0x1CE2, .to = 0x1CE8, .prop = .InCB },
12523 .{ .from = 0x1CED, .to = 0x1CED, .prop = .InCB },
12524 .{ .from = 0x1CF4, .to = 0x1CF4, .prop = .InCB },
12525 .{ .from = 0x1CF8, .to = 0x1CF9, .prop = .InCB },
12526 .{ .from = 0x1DC0, .to = 0x1DFF, .prop = .InCB },
12527 .{ .from = 0x200D, .to = 0x200D, .prop = .InCB },
12528 .{ .from = 0x20D0, .to = 0x20DC, .prop = .InCB },
12529 .{ .from = 0x20E1, .to = 0x20E1, .prop = .InCB },
12530 .{ .from = 0x20E5, .to = 0x20F0, .prop = .InCB },
12531 .{ .from = 0x2CEF, .to = 0x2CF1, .prop = .InCB },
12532 .{ .from = 0x2D7F, .to = 0x2D7F, .prop = .InCB },
12533 .{ .from = 0x2DE0, .to = 0x2DFF, .prop = .InCB },
12534 .{ .from = 0x302A, .to = 0x302D, .prop = .InCB },
12535 .{ .from = 0x302E, .to = 0x302F, .prop = .InCB },
12536 .{ .from = 0x3099, .to = 0x309A, .prop = .InCB },
12537 .{ .from = 0xA66F, .to = 0xA66F, .prop = .InCB },
12538 .{ .from = 0xA674, .to = 0xA67D, .prop = .InCB },
12539 .{ .from = 0xA69E, .to = 0xA69F, .prop = .InCB },
12540 .{ .from = 0xA6F0, .to = 0xA6F1, .prop = .InCB },
12541 .{ .from = 0xA82C, .to = 0xA82C, .prop = .InCB },
12542 .{ .from = 0xA8E0, .to = 0xA8F1, .prop = .InCB },
12543 .{ .from = 0xA92B, .to = 0xA92D, .prop = .InCB },
12544 .{ .from = 0xA9B3, .to = 0xA9B3, .prop = .InCB },
12545 .{ .from = 0xAAB0, .to = 0xAAB0, .prop = .InCB },
12546 .{ .from = 0xAAB2, .to = 0xAAB4, .prop = .InCB },
12547 .{ .from = 0xAAB7, .to = 0xAAB8, .prop = .InCB },
12548 .{ .from = 0xAABE, .to = 0xAABF, .prop = .InCB },
12549 .{ .from = 0xAAC1, .to = 0xAAC1, .prop = .InCB },
12550 .{ .from = 0xAAF6, .to = 0xAAF6, .prop = .InCB },
12551 .{ .from = 0xABED, .to = 0xABED, .prop = .InCB },
12552 .{ .from = 0xFB1E, .to = 0xFB1E, .prop = .InCB },
12553 .{ .from = 0xFE20, .to = 0xFE2F, .prop = .InCB },
12554 .{ .from = 0x101FD, .to = 0x101FD, .prop = .InCB },
12555 .{ .from = 0x102E0, .to = 0x102E0, .prop = .InCB },
12556 .{ .from = 0x10376, .to = 0x1037A, .prop = .InCB },
12557 .{ .from = 0x10A0D, .to = 0x10A0D, .prop = .InCB },
12558 .{ .from = 0x10A0F, .to = 0x10A0F, .prop = .InCB },
12559 .{ .from = 0x10A38, .to = 0x10A3A, .prop = .InCB },
12560 .{ .from = 0x10A3F, .to = 0x10A3F, .prop = .InCB },
12561 .{ .from = 0x10AE5, .to = 0x10AE6, .prop = .InCB },
12562 .{ .from = 0x10D24, .to = 0x10D27, .prop = .InCB },
12563 .{ .from = 0x10EAB, .to = 0x10EAC, .prop = .InCB },
12564 .{ .from = 0x10EFD, .to = 0x10EFF, .prop = .InCB },
12565 .{ .from = 0x10F46, .to = 0x10F50, .prop = .InCB },
12566 .{ .from = 0x10F82, .to = 0x10F85, .prop = .InCB },
12567 .{ .from = 0x11070, .to = 0x11070, .prop = .InCB },
12568 .{ .from = 0x1107F, .to = 0x1107F, .prop = .InCB },
12569 .{ .from = 0x110BA, .to = 0x110BA, .prop = .InCB },
12570 .{ .from = 0x11100, .to = 0x11102, .prop = .InCB },
12571 .{ .from = 0x11133, .to = 0x11134, .prop = .InCB },
12572 .{ .from = 0x11173, .to = 0x11173, .prop = .InCB },
12573 .{ .from = 0x111CA, .to = 0x111CA, .prop = .InCB },
12574 .{ .from = 0x11236, .to = 0x11236, .prop = .InCB },
12575 .{ .from = 0x112E9, .to = 0x112EA, .prop = .InCB },
12576 .{ .from = 0x1133B, .to = 0x1133C, .prop = .InCB },
12577 .{ .from = 0x11366, .to = 0x1136C, .prop = .InCB },
12578 .{ .from = 0x11370, .to = 0x11374, .prop = .InCB },
12579 .{ .from = 0x11446, .to = 0x11446, .prop = .InCB },
12580 .{ .from = 0x1145E, .to = 0x1145E, .prop = .InCB },
12581 .{ .from = 0x114C3, .to = 0x114C3, .prop = .InCB },
12582 .{ .from = 0x115C0, .to = 0x115C0, .prop = .InCB },
12583 .{ .from = 0x116B7, .to = 0x116B7, .prop = .InCB },
12584 .{ .from = 0x1172B, .to = 0x1172B, .prop = .InCB },
12585 .{ .from = 0x1183A, .to = 0x1183A, .prop = .InCB },
12586 .{ .from = 0x1193E, .to = 0x1193E, .prop = .InCB },
12587 .{ .from = 0x11943, .to = 0x11943, .prop = .InCB },
12588 .{ .from = 0x11A34, .to = 0x11A34, .prop = .InCB },
12589 .{ .from = 0x11A47, .to = 0x11A47, .prop = .InCB },
12590 .{ .from = 0x11A99, .to = 0x11A99, .prop = .InCB },
12591 .{ .from = 0x11D42, .to = 0x11D42, .prop = .InCB },
12592 .{ .from = 0x11D44, .to = 0x11D45, .prop = .InCB },
12593 .{ .from = 0x11D97, .to = 0x11D97, .prop = .InCB },
12594 .{ .from = 0x11F42, .to = 0x11F42, .prop = .InCB },
12595 .{ .from = 0x16AF0, .to = 0x16AF4, .prop = .InCB },
12596 .{ .from = 0x16B30, .to = 0x16B36, .prop = .InCB },
12597 .{ .from = 0x1BC9E, .to = 0x1BC9E, .prop = .InCB },
12598 .{ .from = 0x1D165, .to = 0x1D165, .prop = .InCB },
12599 .{ .from = 0x1D167, .to = 0x1D169, .prop = .InCB },
12600 .{ .from = 0x1D16E, .to = 0x1D172, .prop = .InCB },
12601 .{ .from = 0x1D17B, .to = 0x1D182, .prop = .InCB },
12602 .{ .from = 0x1D185, .to = 0x1D18B, .prop = .InCB },
12603 .{ .from = 0x1D1AA, .to = 0x1D1AD, .prop = .InCB },
12604 .{ .from = 0x1D242, .to = 0x1D244, .prop = .InCB },
12605 .{ .from = 0x1E000, .to = 0x1E006, .prop = .InCB },
12606 .{ .from = 0x1E008, .to = 0x1E018, .prop = .InCB },
12607 .{ .from = 0x1E01B, .to = 0x1E021, .prop = .InCB },
12608 .{ .from = 0x1E023, .to = 0x1E024, .prop = .InCB },
12609 .{ .from = 0x1E026, .to = 0x1E02A, .prop = .InCB },
12610 .{ .from = 0x1E08F, .to = 0x1E08F, .prop = .InCB },
12611 .{ .from = 0x1E130, .to = 0x1E136, .prop = .InCB },
12612 .{ .from = 0x1E2AE, .to = 0x1E2AE, .prop = .InCB },
12613 .{ .from = 0x1E2EC, .to = 0x1E2EF, .prop = .InCB },
12614 .{ .from = 0x1E4EC, .to = 0x1E4EF, .prop = .InCB },
12615 .{ .from = 0x1E8D0, .to = 0x1E8D6, .prop = .InCB },
12616 .{ .from = 0x1E944, .to = 0x1E94A, .prop = .InCB },
1240112617};
src/east_asian_width.zig+6-3
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/EastAsianWidth.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/EastAsianWidth.txt
55//
66// zig fmt: off
77
......@@ -1402,7 +1402,7 @@ pub const data = [_]EastAsianWidth{
14021402 .{ .from = 0x2E80, .to = 0x2E99, .prop = .W },
14031403 .{ .from = 0x2E9B, .to = 0x2EF3, .prop = .W },
14041404 .{ .from = 0x2F00, .to = 0x2FD5, .prop = .W },
1405 .{ .from = 0x2FF0, .to = 0x2FFB, .prop = .W },
1405 .{ .from = 0x2FF0, .to = 0x2FFF, .prop = .W },
14061406 .{ .from = 0x3000, .to = 0x3000, .prop = .F },
14071407 .{ .from = 0x3001, .to = 0x3003, .prop = .W },
14081408 .{ .from = 0x3004, .to = 0x3004, .prop = .W },
......@@ -1461,6 +1461,7 @@ pub const data = [_]EastAsianWidth{
14611461 .{ .from = 0x3196, .to = 0x319F, .prop = .W },
14621462 .{ .from = 0x31A0, .to = 0x31BF, .prop = .W },
14631463 .{ .from = 0x31C0, .to = 0x31E3, .prop = .W },
1464 .{ .from = 0x31EF, .to = 0x31EF, .prop = .W },
14641465 .{ .from = 0x31F0, .to = 0x31FF, .prop = .W },
14651466 .{ .from = 0x3200, .to = 0x321E, .prop = .W },
14661467 .{ .from = 0x3220, .to = 0x3229, .prop = .W },
......@@ -2574,7 +2575,9 @@ pub const data = [_]EastAsianWidth{
25742575 .{ .from = 0x2B820, .to = 0x2CEA1, .prop = .W },
25752576 .{ .from = 0x2CEA2, .to = 0x2CEAF, .prop = .W },
25762577 .{ .from = 0x2CEB0, .to = 0x2EBE0, .prop = .W },
2577 .{ .from = 0x2EBE1, .to = 0x2F7FF, .prop = .W },
2578 .{ .from = 0x2EBE1, .to = 0x2EBEF, .prop = .W },
2579 .{ .from = 0x2EBF0, .to = 0x2EE5D, .prop = .W },
2580 .{ .from = 0x2EE5E, .to = 0x2F7FF, .prop = .W },
25782581 .{ .from = 0x2F800, .to = 0x2FA1D, .prop = .W },
25792582 .{ .from = 0x2FA1E, .to = 0x2FA1F, .prop = .W },
25802583 .{ .from = 0x2FA20, .to = 0x2FFFD, .prop = .W },
src/emoji.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/emoji/emoji-data.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/emoji/emoji-data.txt
55//
66// zig fmt: off
77
src/emoji_sources.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/EmojiSources.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/EmojiSources.txt
55//
66// zig fmt: off
77
src/equivalent_unified_ideograph.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/EquivalentUnifiedIdeograph.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/EquivalentUnifiedIdeograph.txt
55//
66// zig fmt: off
77
src/hangul_syllable_type.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/HangulSyllableType.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/HangulSyllableType.txt
55//
66// zig fmt: off
77
src/indic_positional_category.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/IndicPositionalCategory.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/IndicPositionalCategory.txt
55//
66// zig fmt: off
77
src/indic_syllabic_category.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/IndicSyllabicCategory.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/IndicSyllabicCategory.txt
55//
66// zig fmt: off
77
src/jamo.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/Jamo.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/Jamo.txt
55//
66// zig fmt: off
77
src/line_break.zig+78-65
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/LineBreak.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/LineBreak.txt
55//
66// zig fmt: off
77
......@@ -14,6 +14,7 @@ pub const LineBreak = struct {
1414 EM, EX, H2, H3, HL, HY, ID, IN, IS, JL,
1515 JT, JV, NS, NU, OP, PO, PR, QU, RI, SA,
1616 SG, SY, XX,
17 AK, VI, AS, VF, AP,
1718 },
1819};
1920
......@@ -172,7 +173,7 @@ pub const data = [_]LineBreak{
172173 .{ .from = 0x05D0, .to = 0x05EA, .category = .HL },
173174 .{ .from = 0x05EF, .to = 0x05F2, .category = .HL },
174175 .{ .from = 0x05F3, .to = 0x05F4, .category = .AL },
175 .{ .from = 0x0600, .to = 0x0605, .category = .AL },
176 .{ .from = 0x0600, .to = 0x0605, .category = .NU },
176177 .{ .from = 0x0606, .to = 0x0608, .category = .AL },
177178 .{ .from = 0x0609, .to = 0x060A, .category = .PO },
178179 .{ .from = 0x060B, .to = 0x060B, .category = .PO },
......@@ -196,7 +197,7 @@ pub const data = [_]LineBreak{
196197 .{ .from = 0x06D4, .to = 0x06D4, .category = .EX },
197198 .{ .from = 0x06D5, .to = 0x06D5, .category = .AL },
198199 .{ .from = 0x06D6, .to = 0x06DC, .category = .CM },
199 .{ .from = 0x06DD, .to = 0x06DD, .category = .AL },
200 .{ .from = 0x06DD, .to = 0x06DD, .category = .NU },
200201 .{ .from = 0x06DE, .to = 0x06DE, .category = .AL },
201202 .{ .from = 0x06DF, .to = 0x06E4, .category = .CM },
202203 .{ .from = 0x06E5, .to = 0x06E6, .category = .AL },
......@@ -246,12 +247,12 @@ pub const data = [_]LineBreak{
246247 .{ .from = 0x0870, .to = 0x0887, .category = .AL },
247248 .{ .from = 0x0888, .to = 0x0888, .category = .AL },
248249 .{ .from = 0x0889, .to = 0x088E, .category = .AL },
249 .{ .from = 0x0890, .to = 0x0891, .category = .AL },
250 .{ .from = 0x0890, .to = 0x0891, .category = .NU },
250251 .{ .from = 0x0898, .to = 0x089F, .category = .CM },
251252 .{ .from = 0x08A0, .to = 0x08C8, .category = .AL },
252253 .{ .from = 0x08C9, .to = 0x08C9, .category = .AL },
253254 .{ .from = 0x08CA, .to = 0x08E1, .category = .CM },
254 .{ .from = 0x08E2, .to = 0x08E2, .category = .AL },
255 .{ .from = 0x08E2, .to = 0x08E2, .category = .NU },
255256 .{ .from = 0x08E3, .to = 0x08FF, .category = .CM },
256257 .{ .from = 0x0900, .to = 0x0902, .category = .CM },
257258 .{ .from = 0x0903, .to = 0x0903, .category = .CM },
......@@ -746,7 +747,7 @@ pub const data = [_]LineBreak{
746747 .{ .from = 0x1ABF, .to = 0x1ACE, .category = .CM },
747748 .{ .from = 0x1B00, .to = 0x1B03, .category = .CM },
748749 .{ .from = 0x1B04, .to = 0x1B04, .category = .CM },
749 .{ .from = 0x1B05, .to = 0x1B33, .category = .AL },
750 .{ .from = 0x1B05, .to = 0x1B33, .category = .AK },
750751 .{ .from = 0x1B34, .to = 0x1B34, .category = .CM },
751752 .{ .from = 0x1B35, .to = 0x1B35, .category = .CM },
752753 .{ .from = 0x1B36, .to = 0x1B3A, .category = .CM },
......@@ -754,15 +755,16 @@ pub const data = [_]LineBreak{
754755 .{ .from = 0x1B3C, .to = 0x1B3C, .category = .CM },
755756 .{ .from = 0x1B3D, .to = 0x1B41, .category = .CM },
756757 .{ .from = 0x1B42, .to = 0x1B42, .category = .CM },
757 .{ .from = 0x1B43, .to = 0x1B44, .category = .CM },
758 .{ .from = 0x1B45, .to = 0x1B4C, .category = .AL },
759 .{ .from = 0x1B50, .to = 0x1B59, .category = .NU },
758 .{ .from = 0x1B43, .to = 0x1B43, .category = .CM },
759 .{ .from = 0x1B44, .to = 0x1B44, .category = .VI },
760 .{ .from = 0x1B45, .to = 0x1B4C, .category = .AK },
761 .{ .from = 0x1B50, .to = 0x1B59, .category = .ID },
760762 .{ .from = 0x1B5A, .to = 0x1B5B, .category = .BA },
761 .{ .from = 0x1B5C, .to = 0x1B5C, .category = .AL },
763 .{ .from = 0x1B5C, .to = 0x1B5C, .category = .ID },
762764 .{ .from = 0x1B5D, .to = 0x1B60, .category = .BA },
763 .{ .from = 0x1B61, .to = 0x1B6A, .category = .AL },
765 .{ .from = 0x1B61, .to = 0x1B6A, .category = .ID },
764766 .{ .from = 0x1B6B, .to = 0x1B73, .category = .CM },
765 .{ .from = 0x1B74, .to = 0x1B7C, .category = .AL },
767 .{ .from = 0x1B74, .to = 0x1B7C, .category = .ID },
766768 .{ .from = 0x1B7D, .to = 0x1B7E, .category = .BA },
767769 .{ .from = 0x1B80, .to = 0x1B81, .category = .CM },
768770 .{ .from = 0x1B82, .to = 0x1B82, .category = .CM },
......@@ -776,7 +778,7 @@ pub const data = [_]LineBreak{
776778 .{ .from = 0x1BAE, .to = 0x1BAF, .category = .AL },
777779 .{ .from = 0x1BB0, .to = 0x1BB9, .category = .NU },
778780 .{ .from = 0x1BBA, .to = 0x1BBF, .category = .AL },
779 .{ .from = 0x1BC0, .to = 0x1BE5, .category = .AL },
781 .{ .from = 0x1BC0, .to = 0x1BE5, .category = .AS },
780782 .{ .from = 0x1BE6, .to = 0x1BE6, .category = .CM },
781783 .{ .from = 0x1BE7, .to = 0x1BE7, .category = .CM },
782784 .{ .from = 0x1BE8, .to = 0x1BE9, .category = .CM },
......@@ -784,7 +786,7 @@ pub const data = [_]LineBreak{
784786 .{ .from = 0x1BED, .to = 0x1BED, .category = .CM },
785787 .{ .from = 0x1BEE, .to = 0x1BEE, .category = .CM },
786788 .{ .from = 0x1BEF, .to = 0x1BF1, .category = .CM },
787 .{ .from = 0x1BF2, .to = 0x1BF3, .category = .CM },
789 .{ .from = 0x1BF2, .to = 0x1BF3, .category = .VF },
788790 .{ .from = 0x1BFC, .to = 0x1BFF, .category = .AL },
789791 .{ .from = 0x1C00, .to = 0x1C23, .category = .AL },
790792 .{ .from = 0x1C24, .to = 0x1C2B, .category = .CM },
......@@ -1384,7 +1386,7 @@ pub const data = [_]LineBreak{
13841386 .{ .from = 0x2E80, .to = 0x2E99, .category = .ID },
13851387 .{ .from = 0x2E9B, .to = 0x2EF3, .category = .ID },
13861388 .{ .from = 0x2F00, .to = 0x2FD5, .category = .ID },
1387 .{ .from = 0x2FF0, .to = 0x2FFB, .category = .ID },
1389 .{ .from = 0x2FF0, .to = 0x2FFF, .category = .ID },
13881390 .{ .from = 0x3000, .to = 0x3000, .category = .BA },
13891391 .{ .from = 0x3001, .to = 0x3002, .category = .CL },
13901392 .{ .from = 0x3003, .to = 0x3003, .category = .ID },
......@@ -1486,6 +1488,7 @@ pub const data = [_]LineBreak{
14861488 .{ .from = 0x3196, .to = 0x319F, .category = .ID },
14871489 .{ .from = 0x31A0, .to = 0x31BF, .category = .ID },
14881490 .{ .from = 0x31C0, .to = 0x31E3, .category = .ID },
1491 .{ .from = 0x31EF, .to = 0x31EF, .category = .ID },
14891492 .{ .from = 0x31F0, .to = 0x31FF, .category = .CJ },
14901493 .{ .from = 0x3200, .to = 0x321E, .category = .ID },
14911494 .{ .from = 0x3220, .to = 0x3229, .category = .ID },
......@@ -1596,38 +1599,39 @@ pub const data = [_]LineBreak{
15961599 .{ .from = 0xA960, .to = 0xA97C, .category = .JL },
15971600 .{ .from = 0xA980, .to = 0xA982, .category = .CM },
15981601 .{ .from = 0xA983, .to = 0xA983, .category = .CM },
1599 .{ .from = 0xA984, .to = 0xA9B2, .category = .AL },
1602 .{ .from = 0xA984, .to = 0xA9B2, .category = .AK },
16001603 .{ .from = 0xA9B3, .to = 0xA9B3, .category = .CM },
16011604 .{ .from = 0xA9B4, .to = 0xA9B5, .category = .CM },
16021605 .{ .from = 0xA9B6, .to = 0xA9B9, .category = .CM },
16031606 .{ .from = 0xA9BA, .to = 0xA9BB, .category = .CM },
16041607 .{ .from = 0xA9BC, .to = 0xA9BD, .category = .CM },
1605 .{ .from = 0xA9BE, .to = 0xA9C0, .category = .CM },
1606 .{ .from = 0xA9C1, .to = 0xA9C6, .category = .AL },
1608 .{ .from = 0xA9BE, .to = 0xA9BF, .category = .CM },
1609 .{ .from = 0xA9C0, .to = 0xA9C0, .category = .VI },
1610 .{ .from = 0xA9C1, .to = 0xA9C6, .category = .ID },
16071611 .{ .from = 0xA9C7, .to = 0xA9C9, .category = .BA },
1608 .{ .from = 0xA9CA, .to = 0xA9CD, .category = .AL },
1609 .{ .from = 0xA9CF, .to = 0xA9CF, .category = .AL },
1610 .{ .from = 0xA9D0, .to = 0xA9D9, .category = .NU },
1611 .{ .from = 0xA9DE, .to = 0xA9DF, .category = .AL },
1612 .{ .from = 0xA9CA, .to = 0xA9CD, .category = .ID },
1613 .{ .from = 0xA9CF, .to = 0xA9CF, .category = .BA },
1614 .{ .from = 0xA9D0, .to = 0xA9D9, .category = .ID },
1615 .{ .from = 0xA9DE, .to = 0xA9DF, .category = .ID },
16121616 .{ .from = 0xA9E0, .to = 0xA9E4, .category = .SA },
16131617 .{ .from = 0xA9E5, .to = 0xA9E5, .category = .SA },
16141618 .{ .from = 0xA9E6, .to = 0xA9E6, .category = .SA },
16151619 .{ .from = 0xA9E7, .to = 0xA9EF, .category = .SA },
16161620 .{ .from = 0xA9F0, .to = 0xA9F9, .category = .NU },
16171621 .{ .from = 0xA9FA, .to = 0xA9FE, .category = .SA },
1618 .{ .from = 0xAA00, .to = 0xAA28, .category = .AL },
1622 .{ .from = 0xAA00, .to = 0xAA28, .category = .AS },
16191623 .{ .from = 0xAA29, .to = 0xAA2E, .category = .CM },
16201624 .{ .from = 0xAA2F, .to = 0xAA30, .category = .CM },
16211625 .{ .from = 0xAA31, .to = 0xAA32, .category = .CM },
16221626 .{ .from = 0xAA33, .to = 0xAA34, .category = .CM },
16231627 .{ .from = 0xAA35, .to = 0xAA36, .category = .CM },
1624 .{ .from = 0xAA40, .to = 0xAA42, .category = .AL },
1628 .{ .from = 0xAA40, .to = 0xAA42, .category = .BA },
16251629 .{ .from = 0xAA43, .to = 0xAA43, .category = .CM },
1626 .{ .from = 0xAA44, .to = 0xAA4B, .category = .AL },
1630 .{ .from = 0xAA44, .to = 0xAA4B, .category = .BA },
16271631 .{ .from = 0xAA4C, .to = 0xAA4C, .category = .CM },
16281632 .{ .from = 0xAA4D, .to = 0xAA4D, .category = .CM },
1629 .{ .from = 0xAA50, .to = 0xAA59, .category = .NU },
1630 .{ .from = 0xAA5C, .to = 0xAA5C, .category = .AL },
1633 .{ .from = 0xAA50, .to = 0xAA59, .category = .ID },
1634 .{ .from = 0xAA5C, .to = 0xAA5C, .category = .ID },
16311635 .{ .from = 0xAA5D, .to = 0xAA5F, .category = .BA },
16321636 .{ .from = 0xAA60, .to = 0xAA6F, .category = .SA },
16331637 .{ .from = 0xAA70, .to = 0xAA70, .category = .SA },
......@@ -2781,17 +2785,19 @@ pub const data = [_]LineBreak{
27812785 .{ .from = 0x11000, .to = 0x11000, .category = .CM },
27822786 .{ .from = 0x11001, .to = 0x11001, .category = .CM },
27832787 .{ .from = 0x11002, .to = 0x11002, .category = .CM },
2784 .{ .from = 0x11003, .to = 0x11037, .category = .AL },
2785 .{ .from = 0x11038, .to = 0x11046, .category = .CM },
2788 .{ .from = 0x11003, .to = 0x11004, .category = .AP },
2789 .{ .from = 0x11005, .to = 0x11037, .category = .AK },
2790 .{ .from = 0x11038, .to = 0x11045, .category = .CM },
2791 .{ .from = 0x11046, .to = 0x11046, .category = .VI },
27862792 .{ .from = 0x11047, .to = 0x11048, .category = .BA },
2787 .{ .from = 0x11049, .to = 0x1104D, .category = .AL },
2788 .{ .from = 0x11052, .to = 0x11065, .category = .AL },
2789 .{ .from = 0x11066, .to = 0x1106F, .category = .NU },
2793 .{ .from = 0x11049, .to = 0x1104D, .category = .ID },
2794 .{ .from = 0x11052, .to = 0x11065, .category = .ID },
2795 .{ .from = 0x11066, .to = 0x1106F, .category = .AS },
27902796 .{ .from = 0x11070, .to = 0x11070, .category = .CM },
2791 .{ .from = 0x11071, .to = 0x11072, .category = .AL },
2797 .{ .from = 0x11071, .to = 0x11072, .category = .AK },
27922798 .{ .from = 0x11073, .to = 0x11074, .category = .CM },
2793 .{ .from = 0x11075, .to = 0x11075, .category = .AL },
2794 .{ .from = 0x1107F, .to = 0x1107F, .category = .CM },
2799 .{ .from = 0x11075, .to = 0x11075, .category = .AK },
2800 .{ .from = 0x1107F, .to = 0x1107F, .category = .GL },
27952801 .{ .from = 0x11080, .to = 0x11081, .category = .CM },
27962802 .{ .from = 0x11082, .to = 0x11082, .category = .CM },
27972803 .{ .from = 0x11083, .to = 0x110AF, .category = .AL },
......@@ -2800,10 +2806,10 @@ pub const data = [_]LineBreak{
28002806 .{ .from = 0x110B7, .to = 0x110B8, .category = .CM },
28012807 .{ .from = 0x110B9, .to = 0x110BA, .category = .CM },
28022808 .{ .from = 0x110BB, .to = 0x110BC, .category = .AL },
2803 .{ .from = 0x110BD, .to = 0x110BD, .category = .AL },
2809 .{ .from = 0x110BD, .to = 0x110BD, .category = .NU },
28042810 .{ .from = 0x110BE, .to = 0x110C1, .category = .BA },
28052811 .{ .from = 0x110C2, .to = 0x110C2, .category = .CM },
2806 .{ .from = 0x110CD, .to = 0x110CD, .category = .AL },
2812 .{ .from = 0x110CD, .to = 0x110CD, .category = .NU },
28072813 .{ .from = 0x110D0, .to = 0x110E8, .category = .AL },
28082814 .{ .from = 0x110F0, .to = 0x110F9, .category = .NU },
28092815 .{ .from = 0x11100, .to = 0x11102, .category = .CM },
......@@ -2869,22 +2875,25 @@ pub const data = [_]LineBreak{
28692875 .{ .from = 0x112F0, .to = 0x112F9, .category = .NU },
28702876 .{ .from = 0x11300, .to = 0x11301, .category = .CM },
28712877 .{ .from = 0x11302, .to = 0x11303, .category = .CM },
2872 .{ .from = 0x11305, .to = 0x1130C, .category = .AL },
2873 .{ .from = 0x1130F, .to = 0x11310, .category = .AL },
2874 .{ .from = 0x11313, .to = 0x11328, .category = .AL },
2875 .{ .from = 0x1132A, .to = 0x11330, .category = .AL },
2876 .{ .from = 0x11332, .to = 0x11333, .category = .AL },
2877 .{ .from = 0x11335, .to = 0x11339, .category = .AL },
2878 .{ .from = 0x11305, .to = 0x1130C, .category = .AK },
2879 .{ .from = 0x1130F, .to = 0x11310, .category = .AK },
2880 .{ .from = 0x11313, .to = 0x11328, .category = .AK },
2881 .{ .from = 0x1132A, .to = 0x11330, .category = .AK },
2882 .{ .from = 0x11332, .to = 0x11333, .category = .AK },
2883 .{ .from = 0x11335, .to = 0x11339, .category = .AK },
28782884 .{ .from = 0x1133B, .to = 0x1133C, .category = .CM },
2879 .{ .from = 0x1133D, .to = 0x1133D, .category = .AL },
2885 .{ .from = 0x1133D, .to = 0x1133D, .category = .BA },
28802886 .{ .from = 0x1133E, .to = 0x1133F, .category = .CM },
28812887 .{ .from = 0x11340, .to = 0x11340, .category = .CM },
28822888 .{ .from = 0x11341, .to = 0x11344, .category = .CM },
28832889 .{ .from = 0x11347, .to = 0x11348, .category = .CM },
2884 .{ .from = 0x1134B, .to = 0x1134D, .category = .CM },
2885 .{ .from = 0x11350, .to = 0x11350, .category = .AL },
2890 .{ .from = 0x1134B, .to = 0x1134C, .category = .CM },
2891 .{ .from = 0x1134D, .to = 0x1134D, .category = .VI },
2892 .{ .from = 0x11350, .to = 0x11350, .category = .AS },
28862893 .{ .from = 0x11357, .to = 0x11357, .category = .CM },
2887 .{ .from = 0x1135D, .to = 0x11361, .category = .AL },
2894 .{ .from = 0x1135D, .to = 0x1135D, .category = .BA },
2895 .{ .from = 0x1135E, .to = 0x1135F, .category = .AS },
2896 .{ .from = 0x11360, .to = 0x11361, .category = .AK },
28882897 .{ .from = 0x11362, .to = 0x11363, .category = .CM },
28892898 .{ .from = 0x11366, .to = 0x1136C, .category = .CM },
28902899 .{ .from = 0x11370, .to = 0x11374, .category = .CM },
......@@ -2974,23 +2983,23 @@ pub const data = [_]LineBreak{
29742983 .{ .from = 0x118E0, .to = 0x118E9, .category = .NU },
29752984 .{ .from = 0x118EA, .to = 0x118F2, .category = .AL },
29762985 .{ .from = 0x118FF, .to = 0x118FF, .category = .AL },
2977 .{ .from = 0x11900, .to = 0x11906, .category = .AL },
2978 .{ .from = 0x11909, .to = 0x11909, .category = .AL },
2979 .{ .from = 0x1190C, .to = 0x11913, .category = .AL },
2980 .{ .from = 0x11915, .to = 0x11916, .category = .AL },
2981 .{ .from = 0x11918, .to = 0x1192F, .category = .AL },
2986 .{ .from = 0x11900, .to = 0x11906, .category = .AK },
2987 .{ .from = 0x11909, .to = 0x11909, .category = .AK },
2988 .{ .from = 0x1190C, .to = 0x11913, .category = .AK },
2989 .{ .from = 0x11915, .to = 0x11916, .category = .AK },
2990 .{ .from = 0x11918, .to = 0x1192F, .category = .AK },
29822991 .{ .from = 0x11930, .to = 0x11935, .category = .CM },
29832992 .{ .from = 0x11937, .to = 0x11938, .category = .CM },
29842993 .{ .from = 0x1193B, .to = 0x1193C, .category = .CM },
29852994 .{ .from = 0x1193D, .to = 0x1193D, .category = .CM },
2986 .{ .from = 0x1193E, .to = 0x1193E, .category = .CM },
2987 .{ .from = 0x1193F, .to = 0x1193F, .category = .AL },
2995 .{ .from = 0x1193E, .to = 0x1193E, .category = .VI },
2996 .{ .from = 0x1193F, .to = 0x1193F, .category = .AP },
29882997 .{ .from = 0x11940, .to = 0x11940, .category = .CM },
2989 .{ .from = 0x11941, .to = 0x11941, .category = .AL },
2998 .{ .from = 0x11941, .to = 0x11941, .category = .AP },
29902999 .{ .from = 0x11942, .to = 0x11942, .category = .CM },
29913000 .{ .from = 0x11943, .to = 0x11943, .category = .CM },
29923001 .{ .from = 0x11944, .to = 0x11946, .category = .BA },
2993 .{ .from = 0x11950, .to = 0x11959, .category = .NU },
3002 .{ .from = 0x11950, .to = 0x11959, .category = .ID },
29943003 .{ .from = 0x119A0, .to = 0x119A7, .category = .AL },
29953004 .{ .from = 0x119AA, .to = 0x119D0, .category = .AL },
29963005 .{ .from = 0x119D1, .to = 0x119D3, .category = .CM },
......@@ -3072,24 +3081,25 @@ pub const data = [_]LineBreak{
30723081 .{ .from = 0x11D97, .to = 0x11D97, .category = .CM },
30733082 .{ .from = 0x11D98, .to = 0x11D98, .category = .AL },
30743083 .{ .from = 0x11DA0, .to = 0x11DA9, .category = .NU },
3075 .{ .from = 0x11EE0, .to = 0x11EF2, .category = .AL },
3084 .{ .from = 0x11EE0, .to = 0x11EF1, .category = .AS },
3085 .{ .from = 0x11EF2, .to = 0x11EF2, .category = .BA },
30763086 .{ .from = 0x11EF3, .to = 0x11EF4, .category = .CM },
30773087 .{ .from = 0x11EF5, .to = 0x11EF6, .category = .CM },
3078 .{ .from = 0x11EF7, .to = 0x11EF8, .category = .AL },
3088 .{ .from = 0x11EF7, .to = 0x11EF8, .category = .BA },
30793089 .{ .from = 0x11F00, .to = 0x11F01, .category = .CM },
3080 .{ .from = 0x11F02, .to = 0x11F02, .category = .AL },
3090 .{ .from = 0x11F02, .to = 0x11F02, .category = .AP },
30813091 .{ .from = 0x11F03, .to = 0x11F03, .category = .CM },
3082 .{ .from = 0x11F04, .to = 0x11F10, .category = .AL },
3083 .{ .from = 0x11F12, .to = 0x11F33, .category = .AL },
3092 .{ .from = 0x11F04, .to = 0x11F10, .category = .AK },
3093 .{ .from = 0x11F12, .to = 0x11F33, .category = .AK },
30843094 .{ .from = 0x11F34, .to = 0x11F35, .category = .CM },
30853095 .{ .from = 0x11F36, .to = 0x11F3A, .category = .CM },
30863096 .{ .from = 0x11F3E, .to = 0x11F3F, .category = .CM },
30873097 .{ .from = 0x11F40, .to = 0x11F40, .category = .CM },
30883098 .{ .from = 0x11F41, .to = 0x11F41, .category = .CM },
3089 .{ .from = 0x11F42, .to = 0x11F42, .category = .CM },
3099 .{ .from = 0x11F42, .to = 0x11F42, .category = .VI },
30903100 .{ .from = 0x11F43, .to = 0x11F44, .category = .BA },
30913101 .{ .from = 0x11F45, .to = 0x11F4F, .category = .ID },
3092 .{ .from = 0x11F50, .to = 0x11F59, .category = .NU },
3102 .{ .from = 0x11F50, .to = 0x11F59, .category = .AS },
30933103 .{ .from = 0x11FB0, .to = 0x11FB0, .category = .AL },
30943104 .{ .from = 0x11FC0, .to = 0x11FD4, .category = .AL },
30953105 .{ .from = 0x11FD5, .to = 0x11FDC, .category = .AL },
......@@ -3115,7 +3125,8 @@ pub const data = [_]LineBreak{
31153125 .{ .from = 0x1328A, .to = 0x13378, .category = .AL },
31163126 .{ .from = 0x13379, .to = 0x13379, .category = .OP },
31173127 .{ .from = 0x1337A, .to = 0x1337B, .category = .CL },
3118 .{ .from = 0x1337C, .to = 0x1342F, .category = .AL },
3128 .{ .from = 0x1337C, .to = 0x1342E, .category = .AL },
3129 .{ .from = 0x1342F, .to = 0x1342F, .category = .OP },
31193130 .{ .from = 0x13430, .to = 0x13436, .category = .GL },
31203131 .{ .from = 0x13437, .to = 0x13437, .category = .OP },
31213132 .{ .from = 0x13438, .to = 0x13438, .category = .CL },
......@@ -3546,7 +3557,9 @@ pub const data = [_]LineBreak{
35463557 .{ .from = 0x2B820, .to = 0x2CEA1, .category = .ID },
35473558 .{ .from = 0x2CEA2, .to = 0x2CEAF, .category = .ID },
35483559 .{ .from = 0x2CEB0, .to = 0x2EBE0, .category = .ID },
3549 .{ .from = 0x2EBE1, .to = 0x2F7FF, .category = .ID },
3560 .{ .from = 0x2EBE1, .to = 0x2EBEF, .category = .ID },
3561 .{ .from = 0x2EBF0, .to = 0x2EE5D, .category = .ID },
3562 .{ .from = 0x2EE5E, .to = 0x2F7FF, .category = .ID },
35503563 .{ .from = 0x2F800, .to = 0x2FA1D, .category = .ID },
35513564 .{ .from = 0x2FA1E, .to = 0x2FA1F, .category = .ID },
35523565 .{ .from = 0x2FA20, .to = 0x2FFFD, .category = .ID },
src/name_aliases.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/NameAliases.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/NameAliases.txt
55//
66// zig fmt: off
77
src/named_sequences.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/NamedSequences.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/NamedSequences.txt
55//
66// zig fmt: off
77
src/named_sequences_prov.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/NamedSequencesProv.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/NamedSequencesProv.txt
55//
66// zig fmt: off
77
src/prop_list.zig+50-2
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/PropList.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/PropList.txt
55//
66// zig fmt: off
77
......@@ -43,6 +43,9 @@ pub const PropList = struct {
4343 Pattern_Syntax,
4444 Prepended_Concatenation_Mark,
4545 Regional_Indicator,
46 IDS_Unary_Operator,
47 ID_Compat_Math_Continue,
48 ID_Compat_Math_Start,
4649 },
4750};
4851
......@@ -839,6 +842,7 @@ pub const data = [_]PropList{
839842 .{ .from = 0x2B740, .to = 0x2B81D, .property = .Ideographic },
840843 .{ .from = 0x2B820, .to = 0x2CEA1, .property = .Ideographic },
841844 .{ .from = 0x2CEB0, .to = 0x2EBE0, .property = .Ideographic },
845 .{ .from = 0x2EBF0, .to = 0x2EE5D, .property = .Ideographic },
842846 .{ .from = 0x2F800, .to = 0x2FA1D, .property = .Ideographic },
843847 .{ .from = 0x30000, .to = 0x3134A, .property = .Ideographic },
844848 .{ .from = 0x31350, .to = 0x323AF, .property = .Ideographic },
......@@ -1189,8 +1193,10 @@ pub const data = [_]PropList{
11891193 .{ .from = 0x1D16E, .to = 0x1D172, .property = .Other_Grapheme_Extend },
11901194 .{ .from = 0xE0020, .to = 0xE007F, .property = .Other_Grapheme_Extend },
11911195 .{ .from = 0x2FF0, .to = 0x2FF1, .property = .IDS_Binary_Operator },
1192 .{ .from = 0x2FF4, .to = 0x2FFB, .property = .IDS_Binary_Operator },
1196 .{ .from = 0x2FF4, .to = 0x2FFD, .property = .IDS_Binary_Operator },
1197 .{ .from = 0x31EF, .to = 0x31EF, .property = .IDS_Binary_Operator },
11931198 .{ .from = 0x2FF2, .to = 0x2FF3, .property = .IDS_Trinary_Operator },
1199 .{ .from = 0x2FFE, .to = 0x2FFF, .property = .IDS_Unary_Operator },
11941200 .{ .from = 0x2E80, .to = 0x2E99, .property = .Radical },
11951201 .{ .from = 0x2E9B, .to = 0x2EF3, .property = .Radical },
11961202 .{ .from = 0x2F00, .to = 0x2FD5, .property = .Radical },
......@@ -1208,6 +1214,7 @@ pub const data = [_]PropList{
12081214 .{ .from = 0x2B740, .to = 0x2B81D, .property = .Unified_Ideograph },
12091215 .{ .from = 0x2B820, .to = 0x2CEA1, .property = .Unified_Ideograph },
12101216 .{ .from = 0x2CEB0, .to = 0x2EBE0, .property = .Unified_Ideograph },
1217 .{ .from = 0x2EBF0, .to = 0x2EE5D, .property = .Unified_Ideograph },
12111218 .{ .from = 0x30000, .to = 0x3134A, .property = .Unified_Ideograph },
12121219 .{ .from = 0x31350, .to = 0x323AF, .property = .Unified_Ideograph },
12131220 .{ .from = 0x034F, .to = 0x034F, .property = .Other_Default_Ignorable_Code_Point },
......@@ -1279,6 +1286,46 @@ pub const data = [_]PropList{
12791286 .{ .from = 0x0387, .to = 0x0387, .property = .Other_ID_Continue },
12801287 .{ .from = 0x1369, .to = 0x1371, .property = .Other_ID_Continue },
12811288 .{ .from = 0x19DA, .to = 0x19DA, .property = .Other_ID_Continue },
1289 .{ .from = 0x200C, .to = 0x200D, .property = .Other_ID_Continue },
1290 .{ .from = 0x30FB, .to = 0x30FB, .property = .Other_ID_Continue },
1291 .{ .from = 0xFF65, .to = 0xFF65, .property = .Other_ID_Continue },
1292 .{ .from = 0x00B2, .to = 0x00B3, .property = .ID_Compat_Math_Continue },
1293 .{ .from = 0x00B9, .to = 0x00B9, .property = .ID_Compat_Math_Continue },
1294 .{ .from = 0x2070, .to = 0x2070, .property = .ID_Compat_Math_Continue },
1295 .{ .from = 0x2074, .to = 0x2079, .property = .ID_Compat_Math_Continue },
1296 .{ .from = 0x207A, .to = 0x207C, .property = .ID_Compat_Math_Continue },
1297 .{ .from = 0x207D, .to = 0x207D, .property = .ID_Compat_Math_Continue },
1298 .{ .from = 0x207E, .to = 0x207E, .property = .ID_Compat_Math_Continue },
1299 .{ .from = 0x2080, .to = 0x2089, .property = .ID_Compat_Math_Continue },
1300 .{ .from = 0x208A, .to = 0x208C, .property = .ID_Compat_Math_Continue },
1301 .{ .from = 0x208D, .to = 0x208D, .property = .ID_Compat_Math_Continue },
1302 .{ .from = 0x208E, .to = 0x208E, .property = .ID_Compat_Math_Continue },
1303 .{ .from = 0x2202, .to = 0x2202, .property = .ID_Compat_Math_Continue },
1304 .{ .from = 0x2207, .to = 0x2207, .property = .ID_Compat_Math_Continue },
1305 .{ .from = 0x221E, .to = 0x221E, .property = .ID_Compat_Math_Continue },
1306 .{ .from = 0x1D6C1, .to = 0x1D6C1, .property = .ID_Compat_Math_Continue },
1307 .{ .from = 0x1D6DB, .to = 0x1D6DB, .property = .ID_Compat_Math_Continue },
1308 .{ .from = 0x1D6FB, .to = 0x1D6FB, .property = .ID_Compat_Math_Continue },
1309 .{ .from = 0x1D715, .to = 0x1D715, .property = .ID_Compat_Math_Continue },
1310 .{ .from = 0x1D735, .to = 0x1D735, .property = .ID_Compat_Math_Continue },
1311 .{ .from = 0x1D74F, .to = 0x1D74F, .property = .ID_Compat_Math_Continue },
1312 .{ .from = 0x1D76F, .to = 0x1D76F, .property = .ID_Compat_Math_Continue },
1313 .{ .from = 0x1D789, .to = 0x1D789, .property = .ID_Compat_Math_Continue },
1314 .{ .from = 0x1D7A9, .to = 0x1D7A9, .property = .ID_Compat_Math_Continue },
1315 .{ .from = 0x1D7C3, .to = 0x1D7C3, .property = .ID_Compat_Math_Continue },
1316 .{ .from = 0x2202, .to = 0x2202, .property = .ID_Compat_Math_Start },
1317 .{ .from = 0x2207, .to = 0x2207, .property = .ID_Compat_Math_Start },
1318 .{ .from = 0x221E, .to = 0x221E, .property = .ID_Compat_Math_Start },
1319 .{ .from = 0x1D6C1, .to = 0x1D6C1, .property = .ID_Compat_Math_Start },
1320 .{ .from = 0x1D6DB, .to = 0x1D6DB, .property = .ID_Compat_Math_Start },
1321 .{ .from = 0x1D6FB, .to = 0x1D6FB, .property = .ID_Compat_Math_Start },
1322 .{ .from = 0x1D715, .to = 0x1D715, .property = .ID_Compat_Math_Start },
1323 .{ .from = 0x1D735, .to = 0x1D735, .property = .ID_Compat_Math_Start },
1324 .{ .from = 0x1D74F, .to = 0x1D74F, .property = .ID_Compat_Math_Start },
1325 .{ .from = 0x1D76F, .to = 0x1D76F, .property = .ID_Compat_Math_Start },
1326 .{ .from = 0x1D789, .to = 0x1D789, .property = .ID_Compat_Math_Start },
1327 .{ .from = 0x1D7A9, .to = 0x1D7A9, .property = .ID_Compat_Math_Start },
1328 .{ .from = 0x1D7C3, .to = 0x1D7C3, .property = .ID_Compat_Math_Start },
12821329 .{ .from = 0x0021, .to = 0x0021, .property = .Sentence_Terminal },
12831330 .{ .from = 0x002E, .to = 0x002E, .property = .Sentence_Terminal },
12841331 .{ .from = 0x003F, .to = 0x003F, .property = .Sentence_Terminal },
......@@ -1296,6 +1343,7 @@ pub const data = [_]PropList{
12961343 .{ .from = 0x1367, .to = 0x1368, .property = .Sentence_Terminal },
12971344 .{ .from = 0x166E, .to = 0x166E, .property = .Sentence_Terminal },
12981345 .{ .from = 0x1735, .to = 0x1736, .property = .Sentence_Terminal },
1346 .{ .from = 0x17D4, .to = 0x17D5, .property = .Sentence_Terminal },
12991347 .{ .from = 0x1803, .to = 0x1803, .property = .Sentence_Terminal },
13001348 .{ .from = 0x1809, .to = 0x1809, .property = .Sentence_Terminal },
13011349 .{ .from = 0x1944, .to = 0x1945, .property = .Sentence_Terminal },
src/property_aliases.zig+9-4
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/PropertyAliases.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/PropertyAliases.txt
55//
66// zig fmt: off
77
......@@ -12,12 +12,16 @@ pub const data = [_][2][]const u8{
1212 .{ "cjkOtherNumeric", "kOtherNumeric" },
1313 .{ "cjkPrimaryNumeric", "kPrimaryNumeric" },
1414 .{ "nv", "Numeric_Value" },
15 .{ "bmg", "Bidi_Mirroring_Glyph" },
16 .{ "bpb", "Bidi_Paired_Bracket" },
1517 .{ "cf", "Case_Folding" },
1618 .{ "cjkCompatibilityVariant", "kCompatibilityVariant" },
1719 .{ "dm", "Decomposition_Mapping" },
20 .{ "EqUIdeo", "Equivalent_Unified_Ideograph" },
1821 .{ "FC_NFKC", "FC_NFKC_Closure" },
1922 .{ "lc", "Lowercase_Mapping" },
2023 .{ "NFKC_CF", "NFKC_Casefold" },
24 .{ "NFKC_SCF", "NFKC_Simple_Casefold" },
2125 .{ "scf", "Simple_Case_Folding" },
2226 .{ "sfc", "Simple_Case_Folding" },
2327 .{ "slc", "Simple_Lowercase_Mapping" },
......@@ -25,8 +29,6 @@ pub const data = [_][2][]const u8{
2529 .{ "suc", "Simple_Uppercase_Mapping" },
2630 .{ "tc", "Titlecase_Mapping" },
2731 .{ "uc", "Uppercase_Mapping" },
28 .{ "bmg", "Bidi_Mirroring_Glyph" },
29 .{ "bpb", "Bidi_Paired_Bracket" },
3032 .{ "cjkIICore", "kIICore" },
3133 .{ "cjkIRG_GSource", "kIRG_GSource" },
3234 .{ "cjkIRG_HSource", "kIRG_HSource" },
......@@ -42,7 +44,6 @@ pub const data = [_][2][]const u8{
4244 .{ "cjkRSUnicode", "kRSUnicode" },
4345 .{ "Unicode_Radical_Stroke", "kRSUnicode" },
4446 .{ "URS", "kRSUnicode" },
45 .{ "EqUIdeo", "Equivalent_Unified_Ideograph" },
4647 .{ "isc", "ISO_Comment" },
4748 .{ "JSN", "Jamo_Short_Name" },
4849 .{ "na", "Name" },
......@@ -60,6 +61,7 @@ pub const data = [_][2][]const u8{
6061 .{ "gc", "General_Category" },
6162 .{ "GCB", "Grapheme_Cluster_Break" },
6263 .{ "hst", "Hangul_Syllable_Type" },
64 .{ "InCB", "Indic_Conjunct_Break" },
6365 .{ "InPC", "Indic_Positional_Category" },
6466 .{ "InSC", "Indic_Syllabic_Category" },
6567 .{ "jg", "Joining_Group" },
......@@ -103,11 +105,14 @@ pub const data = [_][2][]const u8{
103105 .{ "Gr_Link", "Grapheme_Link" },
104106 .{ "Hex", "Hex_Digit" },
105107 .{ "Hyphen", "Hyphen" },
108 .{ "ID_Compat_Math_Continue", "ID_Compat_Math_Continue" },
109 .{ "ID_Compat_Math_Start", "ID_Compat_Math_Start" },
106110 .{ "IDC", "ID_Continue" },
107111 .{ "Ideo", "Ideographic" },
108112 .{ "IDS", "ID_Start" },
109113 .{ "IDSB", "IDS_Binary_Operator" },
110114 .{ "IDST", "IDS_Trinary_Operator" },
115 .{ "IDSU", "IDS_Unary_Operator" },
111116 .{ "Join_C", "Join_Control" },
112117 .{ "LOE", "Logical_Order_Exception" },
113118 .{ "Lower", "Lowercase" },
src/property_value_aliases.zig+30-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/PropertyValueAliases.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/PropertyValueAliases.txt
55//
66// zig fmt: off
77
......@@ -39,6 +39,7 @@ pub const data = [_][3][]const u8{
3939 .{ "age", "13.0", "V13_0" },
4040 .{ "age", "14.0", "V14_0" },
4141 .{ "age", "15.0", "V15_0" },
42 .{ "age", "15.1", "V15_1" },
4243 .{ "age", "NA", "Unassigned" },
4344 .{ "Alpha", "N", "No" },
4445 .{ "Alpha", "F", "No" },
......@@ -143,6 +144,7 @@ pub const data = [_][3][]const u8{
143144 .{ "blk", "CJK_Ext_F", "CJK_Unified_Ideographs_Extension_F" },
144145 .{ "blk", "CJK_Ext_G", "CJK_Unified_Ideographs_Extension_G" },
145146 .{ "blk", "CJK_Ext_H", "CJK_Unified_Ideographs_Extension_H" },
147 .{ "blk", "CJK_Ext_I", "CJK_Unified_Ideographs_Extension_I" },
146148 .{ "blk", "CJK_Radicals_Sup", "CJK_Radicals_Supplement" },
147149 .{ "blk", "CJK_Strokes", "CJK_Strokes" },
148150 .{ "blk", "CJK_Symbols", "CJK_Symbols_And_Punctuation" },
......@@ -776,6 +778,24 @@ pub const data = [_][3][]const u8{
776778 .{ "IDST", "Y", "Yes" },
777779 .{ "IDST", "T", "Yes" },
778780 .{ "IDST", "True", "Yes" },
781 .{ "IDSU", "N", "No" },
782 .{ "IDSU", "F", "No" },
783 .{ "IDSU", "False", "No" },
784 .{ "IDSU", "Y", "Yes" },
785 .{ "IDSU", "T", "Yes" },
786 .{ "IDSU", "True", "Yes" },
787 .{ "ID_Compat_Math_Continue", "N", "No" },
788 .{ "ID_Compat_Math_Continue", "F", "No" },
789 .{ "ID_Compat_Math_Continue", "False", "No" },
790 .{ "ID_Compat_Math_Continue", "Y", "Yes" },
791 .{ "ID_Compat_Math_Continue", "T", "Yes" },
792 .{ "ID_Compat_Math_Continue", "True", "Yes" },
793 .{ "ID_Compat_Math_Start", "N", "No" },
794 .{ "ID_Compat_Math_Start", "F", "No" },
795 .{ "ID_Compat_Math_Start", "False", "No" },
796 .{ "ID_Compat_Math_Start", "Y", "Yes" },
797 .{ "ID_Compat_Math_Start", "T", "Yes" },
798 .{ "ID_Compat_Math_Start", "True", "Yes" },
779799 .{ "IDC", "N", "No" },
780800 .{ "IDC", "F", "No" },
781801 .{ "IDC", "False", "No" },
......@@ -794,6 +814,10 @@ pub const data = [_][3][]const u8{
794814 .{ "Ideo", "Y", "Yes" },
795815 .{ "Ideo", "T", "Yes" },
796816 .{ "Ideo", "True", "Yes" },
817 .{ "InCB", "Consonant", "Consonant" },
818 .{ "InCB", "Extend", "Extend" },
819 .{ "InCB", "Linker", "Linker" },
820 .{ "InCB", "None", "None" },
797821 .{ "InPC", "Bottom", "Bottom" },
798822 .{ "InPC", "Bottom_And_Left", "Bottom_And_Left" },
799823 .{ "InPC", "Bottom_And_Right", "Bottom_And_Right" },
......@@ -1015,7 +1039,10 @@ pub const data = [_][3][]const u8{
10151039 .{ "jt", "T", "Transparent" },
10161040 .{ "jt", "U", "Non_Joining" },
10171041 .{ "lb", "AI", "Ambiguous" },
1042 .{ "lb", "AK", "Aksara" },
10181043 .{ "lb", "AL", "Alphabetic" },
1044 .{ "lb", "AP", "Aksara_Prebase" },
1045 .{ "lb", "AS", "Aksara_Start" },
10191046 .{ "lb", "B2", "Break_Both" },
10201047 .{ "lb", "BA", "Break_After" },
10211048 .{ "lb", "BB", "Break_Before" },
......@@ -1054,6 +1081,8 @@ pub const data = [_][3][]const u8{
10541081 .{ "lb", "SG", "Surrogate" },
10551082 .{ "lb", "SP", "Space" },
10561083 .{ "lb", "SY", "Break_Symbols" },
1084 .{ "lb", "VF", "Virama_Final" },
1085 .{ "lb", "VI", "Virama" },
10571086 .{ "lb", "WJ", "Word_Joiner" },
10581087 .{ "lb", "XX", "Unknown" },
10591088 .{ "lb", "ZW", "ZWSpace" },
src/script_extensions.zig+10-10
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/ScriptExtensions.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/ScriptExtensions.txt
55//
66// zig fmt: off
77
......@@ -388,9 +388,9 @@ pub const data = [_]ScriptExtension{
388388 .{ .code = 0x102F9, .scripts = &.{ .Arab, .Copt, } },
389389 .{ .code = 0x102FA, .scripts = &.{ .Arab, .Copt, } },
390390 .{ .code = 0x102FB, .scripts = &.{ .Arab, .Copt, } },
391 .{ .code = 0x06D4, .scripts = &.{ .Arab, .Rohg, } },
392391 .{ .code = 0xFD3E, .scripts = &.{ .Arab, .Nkoo, } },
393392 .{ .code = 0xFD3F, .scripts = &.{ .Arab, .Nkoo, } },
393 .{ .code = 0x06D4, .scripts = &.{ .Arab, .Rohg, } },
394394 .{ .code = 0x64B, .scripts = &.{ .Arab, .Syrc, } },
395395 .{ .code = 0x64C, .scripts = &.{ .Arab, .Syrc, } },
396396 .{ .code = 0x64D, .scripts = &.{ .Arab, .Syrc, } },
......@@ -664,20 +664,20 @@ pub const data = [_]ScriptExtension{
664664 .{ .code = 0xFF65, .scripts = &.{ .Bopo, .Hang, .Hani, .Hira, .Kana, .Yiii, } },
665665 .{ .code = 0x1CDA, .scripts = &.{ .Deva, .Knda, .Mlym, .Orya, .Taml, .Telu, } },
666666 .{ .code = 0x061F, .scripts = &.{ .Adlm, .Arab, .Nkoo, .Rohg, .Syrc, .Thaa, .Yezi, } },
667 .{ .code = 0x1CF2, .scripts = &.{ .Beng, .Deva, .Gran, .Knda, .Nand, .Orya, .Telu, .Tirh, } },
668667 .{ .code = 0x0640, .scripts = &.{ .Adlm, .Arab, .Mand, .Mani, .Ougr, .Phlp, .Rohg, .Sogd, .Syrc, } },
668 .{ .code = 0x1CF2, .scripts = &.{ .Beng, .Deva, .Gran, .Knda, .Mlym, .Nand, .Orya, .Sinh, .Telu, .Tirh, } },
669669 .{ .code = 0xA836, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Kthi, .Mahj, .Modi, .Sind, .Takr, .Tirh, } },
670670 .{ .code = 0xA837, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Kthi, .Mahj, .Modi, .Sind, .Takr, .Tirh, } },
671 .{ .code = 0xA838, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Kthi, .Mahj, .Modi, .Sind, .Takr, .Tirh, } },
672671 .{ .code = 0xA839, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Kthi, .Mahj, .Modi, .Sind, .Takr, .Tirh, } },
673672 .{ .code = 0x0952, .scripts = &.{ .Beng, .Deva, .Gran, .Gujr, .Guru, .Knda, .Latn, .Mlym, .Orya, .Taml, .Telu, .Tirh, } },
673 .{ .code = 0xA838, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Kthi, .Mahj, .Modi, .Shrd, .Sind, .Takr, .Tirh, } },
674674 .{ .code = 0x0951, .scripts = &.{ .Beng, .Deva, .Gran, .Gujr, .Guru, .Knda, .Latn, .Mlym, .Orya, .Shrd, .Taml, .Telu, .Tirh, } },
675 .{ .code = 0xA833, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
676 .{ .code = 0xA834, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
677 .{ .code = 0xA835, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
678 .{ .code = 0xA830, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
679 .{ .code = 0xA831, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
680 .{ .code = 0xA832, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Sind, .Takr, .Tirh, } },
675 .{ .code = 0xA833, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
676 .{ .code = 0xA834, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
677 .{ .code = 0xA835, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
678 .{ .code = 0xA830, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
679 .{ .code = 0xA831, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
680 .{ .code = 0xA832, .scripts = &.{ .Deva, .Dogr, .Gujr, .Guru, .Khoj, .Knda, .Kthi, .Mahj, .Mlym, .Modi, .Nand, .Shrd, .Sind, .Takr, .Tirh, } },
681681 .{ .code = 0x0964, .scripts = &.{ .Beng, .Deva, .Dogr, .Gong, .Gonm, .Gran, .Gujr, .Guru, .Knda, .Mahj, .Mlym, .Nand, .Orya, .Sind, .Sinh, .Sylo, .Takr, .Taml, .Telu, .Tirh, } },
682682 .{ .code = 0x0965, .scripts = &.{ .Beng, .Deva, .Dogr, .Gong, .Gonm, .Gran, .Gujr, .Guru, .Knda, .Limb, .Mahj, .Mlym, .Nand, .Orya, .Sind, .Sinh, .Sylo, .Takr, .Taml, .Telu, .Tirh, } },
683683};
src/scripts.zig+4-2
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/Scripts.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/Scripts.txt
55//
66// zig fmt: off
77
......@@ -509,7 +509,7 @@ pub const data = [_]Scripts{
509509 .{ .from = 0x2E5B, .to = 0x2E5B, .script = .Common },
510510 .{ .from = 0x2E5C, .to = 0x2E5C, .script = .Common },
511511 .{ .from = 0x2E5D, .to = 0x2E5D, .script = .Common },
512 .{ .from = 0x2FF0, .to = 0x2FFB, .script = .Common },
512 .{ .from = 0x2FF0, .to = 0x2FFF, .script = .Common },
513513 .{ .from = 0x3000, .to = 0x3000, .script = .Common },
514514 .{ .from = 0x3001, .to = 0x3003, .script = .Common },
515515 .{ .from = 0x3004, .to = 0x3004, .script = .Common },
......@@ -551,6 +551,7 @@ pub const data = [_]Scripts{
551551 .{ .from = 0x3192, .to = 0x3195, .script = .Common },
552552 .{ .from = 0x3196, .to = 0x319F, .script = .Common },
553553 .{ .from = 0x31C0, .to = 0x31E3, .script = .Common },
554 .{ .from = 0x31EF, .to = 0x31EF, .script = .Common },
554555 .{ .from = 0x3220, .to = 0x3229, .script = .Common },
555556 .{ .from = 0x322A, .to = 0x3247, .script = .Common },
556557 .{ .from = 0x3248, .to = 0x324F, .script = .Common },
......@@ -1570,6 +1571,7 @@ pub const data = [_]Scripts{
15701571 .{ .from = 0x2B740, .to = 0x2B81D, .script = .Han },
15711572 .{ .from = 0x2B820, .to = 0x2CEA1, .script = .Han },
15721573 .{ .from = 0x2CEB0, .to = 0x2EBE0, .script = .Han },
1574 .{ .from = 0x2EBF0, .to = 0x2EE5D, .script = .Han },
15731575 .{ .from = 0x2F800, .to = 0x2FA1D, .script = .Han },
15741576 .{ .from = 0x30000, .to = 0x3134A, .script = .Han },
15751577 .{ .from = 0x31350, .to = 0x323AF, .script = .Han },
src/special_casing.zig+1-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/SpecialCasing.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/SpecialCasing.txt
55//
66// zig fmt: off
77
src/unicode_data.zig+8-1
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/UnicodeData.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/UnicodeData.txt
55//
66// zig fmt: off
77
......@@ -11257,6 +11257,10 @@ pub const data = [_]Codepoint{
1125711257 .{ 0x2FF9, "IDEOGRAPHIC DESCRIPTION CHARACTER SURROUND FROM UPPER RIGHT", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1125811258 .{ 0x2FFA, "IDEOGRAPHIC DESCRIPTION CHARACTER SURROUND FROM LOWER LEFT", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1125911259 .{ 0x2FFB, "IDEOGRAPHIC DESCRIPTION CHARACTER OVERLAID", .So, 0, .ON, false, "", "", "", false, null, null, null, },
11260 .{ 0x2FFC, "IDEOGRAPHIC DESCRIPTION CHARACTER SURROUND FROM RIGHT", .So, 0, .ON, false, "", "", "", false, null, null, null, },
11261 .{ 0x2FFD, "IDEOGRAPHIC DESCRIPTION CHARACTER SURROUND FROM LOWER RIGHT", .So, 0, .ON, false, "", "", "", false, null, null, null, },
11262 .{ 0x2FFE, "IDEOGRAPHIC DESCRIPTION CHARACTER HORIZONTAL REFLECTION", .So, 0, .ON, false, "", "", "", false, null, null, null, },
11263 .{ 0x2FFF, "IDEOGRAPHIC DESCRIPTION CHARACTER ROTATION", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1126011264 .{ 0x3000, "IDEOGRAPHIC SPACE", .Zs, 0, .WS, true, "", "", "", false, null, null, null, },
1126111265 .{ 0x3001, "IDEOGRAPHIC COMMA", .Po, 0, .ON, false, "", "", "", false, null, null, null, },
1126211266 .{ 0x3002, "IDEOGRAPHIC FULL STOP", .Po, 0, .ON, false, "", "", "", false, null, null, null, },
......@@ -11731,6 +11735,7 @@ pub const data = [_]Codepoint{
1173111735 .{ 0x31E1, "CJK STROKE HZZZG", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1173211736 .{ 0x31E2, "CJK STROKE PG", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1173311737 .{ 0x31E3, "CJK STROKE Q", .So, 0, .ON, false, "", "", "", false, null, null, null, },
11738 .{ 0x31EF, "IDEOGRAPHIC DESCRIPTION CHARACTER SUBTRACTION", .So, 0, .ON, false, "", "", "", false, null, null, null, },
1173411739 .{ 0x31F0, "KATAKANA LETTER SMALL KU", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
1173511740 .{ 0x31F1, "KATAKANA LETTER SMALL SI", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
1173611741 .{ 0x31F2, "KATAKANA LETTER SMALL SU", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
......@@ -34061,6 +34066,8 @@ pub const data = [_]Codepoint{
3406134066 .{ 0x2CEA1, "<CJK Ideograph Extension E, Last>", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
3406234067 .{ 0x2CEB0, "<CJK Ideograph Extension F, First>", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
3406334068 .{ 0x2EBE0, "<CJK Ideograph Extension F, Last>", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
34069 .{ 0x2EBF0, "<CJK Ideograph Extension I, First>", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
34070 .{ 0x2EE5D, "<CJK Ideograph Extension I, Last>", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
3406434071 .{ 0x2F800, "CJK COMPATIBILITY IDEOGRAPH-2F800", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
3406534072 .{ 0x2F801, "CJK COMPATIBILITY IDEOGRAPH-2F801", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
3406634073 .{ 0x2F802, "CJK COMPATIBILITY IDEOGRAPH-2F802", .Lo, 0, .L, false, "", "", "", false, null, null, null, },
src/vertical_orientation.zig+12-7
......@@ -1,7 +1,7 @@
11// This file is part of the Unicode Character Database
22// For documentation, see http://www.unicode.org/reports/tr44/
33//
4// Based on the source file: https://unicode.org/Public/15.0.0/ucd/VerticalOrientation.txt
4// Based on the source file: https://unicode.org/Public/15.1.0/ucd/VerticalOrientation.txt
55//
66// zig fmt: off
77
......@@ -1128,8 +1128,7 @@ pub const data = [_]VerticalOrientation{
11281128 .{ .from = 0x2F00, .to = 0x2FD5, .orientation = .U },
11291129 .{ .from = 0x2FD6, .to = 0x2FDF, .orientation = .U },
11301130 .{ .from = 0x2FE0, .to = 0x2FEF, .orientation = .U },
1131 .{ .from = 0x2FF0, .to = 0x2FFB, .orientation = .U },
1132 .{ .from = 0x2FFC, .to = 0x2FFF, .orientation = .U },
1131 .{ .from = 0x2FF0, .to = 0x2FFF, .orientation = .U },
11331132 .{ .from = 0x3000, .to = 0x3000, .orientation = .U },
11341133 .{ .from = 0x3001, .to = 0x3002, .orientation = .Tu },
11351134 .{ .from = 0x3003, .to = 0x3003, .orientation = .U },
......@@ -1237,7 +1236,8 @@ pub const data = [_]VerticalOrientation{
12371236 .{ .from = 0x3196, .to = 0x319F, .orientation = .U },
12381237 .{ .from = 0x31A0, .to = 0x31BF, .orientation = .U },
12391238 .{ .from = 0x31C0, .to = 0x31E3, .orientation = .U },
1240 .{ .from = 0x31E4, .to = 0x31EF, .orientation = .U },
1239 .{ .from = 0x31E4, .to = 0x31EE, .orientation = .U },
1240 .{ .from = 0x31EF, .to = 0x31EF, .orientation = .U },
12411241 .{ .from = 0x31F0, .to = 0x31FF, .orientation = .Tu },
12421242 .{ .from = 0x3200, .to = 0x321E, .orientation = .U },
12431243 .{ .from = 0x321F, .to = 0x321F, .orientation = .U },
......@@ -1442,7 +1442,9 @@ pub const data = [_]VerticalOrientation{
14421442 .{ .from = 0xD7C7, .to = 0xD7CA, .orientation = .U },
14431443 .{ .from = 0xD7CB, .to = 0xD7FB, .orientation = .U },
14441444 .{ .from = 0xD7FC, .to = 0xD7FF, .orientation = .U },
1445 .{ .from = 0xD800, .to = 0xDFFF, .orientation = .R },
1445 .{ .from = 0xD800, .to = 0xDB7F, .orientation = .R },
1446 .{ .from = 0xDB80, .to = 0xDBFF, .orientation = .R },
1447 .{ .from = 0xDC00, .to = 0xDFFF, .orientation = .R },
14461448 .{ .from = 0xE000, .to = 0xF8FF, .orientation = .U },
14471449 .{ .from = 0xF900, .to = 0xFA6D, .orientation = .U },
14481450 .{ .from = 0xFA6E, .to = 0xFA6F, .orientation = .U },
......@@ -2367,9 +2369,12 @@ pub const data = [_]VerticalOrientation{
23672369 .{ .from = 0x2B820, .to = 0x2CEA1, .orientation = .U },
23682370 .{ .from = 0x2CEA2, .to = 0x2CEAF, .orientation = .U },
23692371 .{ .from = 0x2CEB0, .to = 0x2EBE0, .orientation = .U },
2370 .{ .from = 0x2EBE1, .to = 0x2F7FF, .orientation = .U },
2372 .{ .from = 0x2EBE1, .to = 0x2EBEF, .orientation = .U },
2373 .{ .from = 0x2EBF0, .to = 0x2EE5D, .orientation = .U },
2374 .{ .from = 0x2EE5E, .to = 0x2F7FF, .orientation = .U },
23712375 .{ .from = 0x2F800, .to = 0x2FA1D, .orientation = .U },
2372 .{ .from = 0x2FA1E, .to = 0x2FFFD, .orientation = .U },
2376 .{ .from = 0x2FA1E, .to = 0x2FA1F, .orientation = .U },
2377 .{ .from = 0x2FA20, .to = 0x2FFFD, .orientation = .U },
23732378 .{ .from = 0x30000, .to = 0x3134A, .orientation = .U },
23742379 .{ .from = 0x3134B, .to = 0x3134F, .orientation = .U },
23752380 .{ .from = 0x31350, .to = 0x323AF, .orientation = .U },