Modul:scripts/data
Dokumentasi untuk modul ini dapat dibuat di Modul:scripts/data/doc
--[=[
When adding new scripts to this file, please don't forget to add
style definitons for the script in [[MediaWiki:Common.css]].
]=]
local concat = table.concat
local insert = table.insert
local ipairs = ipairs
local next = next
local remove = table.remove
local select = select
local sort = table.sort
local u = require("Module:string utilities").char
------------------------------------------------------------------------------------
--
-- Helper functions
--
------------------------------------------------------------------------------------
-- Note: a[2] > b[2] means opens are sorted before closes if otherwise equal.
local function sort_ranges(a, b)
return a[1] < b[1] or a[1] == b[1] and a[2] > b[2]
end
-- Returns the union of two or more range tables.
local function union(...)
local ranges = {}
for i = 1, select("#", ...) do
local argt = select(i, ...)
for j, v in ipairs(argt) do
insert(ranges, {v, j % 2 == 1 and 1 or -1})
end
end
sort(ranges, sort_ranges)
local ret, i = {}, 0
for _, range in ipairs(ranges) do
i = i + range[2]
if i == 0 and range[2] == -1 then -- close
insert(ret, range[1])
elseif i == 1 and range[2] == 1 then -- open
if ret[#ret] and range[1] <= ret[#ret] + 1 then
remove(ret) -- merge adjacent ranges
else
insert(ret, range[1])
end
end
end
return ret
end
-- Adds the `characters` key, which is determined by a script's `ranges` table.
local function process_ranges(sc)
local ranges, chars = sc.ranges, {}
for i = 2, #ranges, 2 do
if ranges[i] == ranges[i - 1] then
insert(chars, u(ranges[i]))
else
insert(chars, u(ranges[i - 1]))
if ranges[i] > ranges[i - 1] + 1 then
insert(chars, "-")
end
insert(chars, u(ranges[i]))
end
end
sc.characters = concat(chars)
ranges.n = #ranges
return sc
end
local function handle_normalization_fixes(fixes)
local combiningClasses = fixes.combiningClasses
if combiningClasses then
local chars, i = {}, 0
for char in next, combiningClasses do
i = i + 1
chars[i] = char
end
fixes.combiningClassCharacters = concat(chars)
end
return fixes
end
------------------------------------------------------------------------------------
--
-- Data
--
------------------------------------------------------------------------------------
local m = {}
m["Adlm"] = process_ranges{
"Adlam",
"alphabet",
ranges = {
0x061F, 0x061F,
0x0640, 0x0640,
0x1E900, 0x1E94B,
0x1E950, 0x1E959,
0x1E95E, 0x1E95F,
},
capitalized = true,
direction = "rtl",
}
m["Afak"] = {
"Afaka",
"syllabary",
-- Not in Unicode
}
m["Aghb"] = process_ranges{
"Caucasian Albanian",
"alphabet",
ranges = {
0x10530, 0x10563,
0x1056F, 0x1056F,
},
}
m["Ahom"] = process_ranges{
"Ahom",
"abugida",
ranges = {
0x11700, 0x1171A,
0x1171D, 0x1172B,
0x11730, 0x11746,
},
}
m["Arab"] = process_ranges{
"Arabic",
"abjad", -- more precisely, impure abjad
varieties = {"Jawi", {"Nastaliq", "Nastaleeq"}},
ranges = {
0x0600, 0x06FF,
0x0750, 0x077F,
0x0870, 0x088E,
0x0890, 0x0891,
0x0897, 0x08E1,
0x08E3, 0x08FF,
0xFB50, 0xFBC2,
0xFBD3, 0xFD8F,
0xFD92, 0xFDC7,
0xFDCF, 0xFDCF,
0xFDF0, 0xFDFF,
0xFE70, 0xFE74,
0xFE76, 0xFEFC,
0x102E0, 0x102FB,
0x10E60, 0x10E7E,
0x10EC2, 0x10EC4,
0x10EFC, 0x10EFF,
0x1EE00, 0x1EE03,
0x1EE05, 0x1EE1F,
0x1EE21, 0x1EE22,
0x1EE24, 0x1EE24,
0x1EE27, 0x1EE27,
0x1EE29, 0x1EE32,
0x1EE34, 0x1EE37,
0x1EE39, 0x1EE39,
0x1EE3B, 0x1EE3B,
0x1EE42, 0x1EE42,
0x1EE47, 0x1EE47,
0x1EE49, 0x1EE49,
0x1EE4B, 0x1EE4B,
0x1EE4D, 0x1EE4F,
0x1EE51, 0x1EE52,
0x1EE54, 0x1EE54,
0x1EE57, 0x1EE57,
0x1EE59, 0x1EE59,
0x1EE5B, 0x1EE5B,
0x1EE5D, 0x1EE5D,
0x1EE5F, 0x1EE5F,
0x1EE61, 0x1EE62,
0x1EE64, 0x1EE64,
0x1EE67, 0x1EE6A,
0x1EE6C, 0x1EE72,
0x1EE74, 0x1EE77,
0x1EE79, 0x1EE7C,
0x1EE7E, 0x1EE7E,
0x1EE80, 0x1EE89,
0x1EE8B, 0x1EE9B,
0x1EEA1, 0x1EEA3,
0x1EEA5, 0x1EEA9,
0x1EEAB, 0x1EEBB,
0x1EEF0, 0x1EEF1,
},
direction = "rtl",
normalizationFixes = handle_normalization_fixes{
from = {"ٳ"},
to = {"اٟ"}
},
}
m["fa-Arab"] = {
"Arabic",
m["Arab"][2],
ranges = m["Arab"].ranges,
characters = m["Arab"].characters,
otherNames = {"Perso-Arabic"},
direction = "rtl",
parent = "Arab",
normalizationFixes = m["Arab"].normalizationFixes,
}
m["kk-Arab"] = {
"Arabic",
m["Arab"][2],
ranges = m["Arab"].ranges,
characters = m["Arab"].characters,
direction = "rtl",
parent = "Arab",
normalizationFixes = m["Arab"].normalizationFixes,
}
m["ks-Arab"] = m["fa-Arab"]
m["ku-Arab"] = m["fa-Arab"]
m["ms-Arab"] = m["kk-Arab"]
m["mzn-Arab"] = m["fa-Arab"]
m["ota-Arab"] = m["fa-Arab"]
m["pa-Arab"] = {
"Shahmukhi",
m["Arab"][2],
ranges = m["Arab"].ranges,
characters = m["Arab"].characters,
otherNames = {"Arabic"},
direction = "rtl",
parent = "Arab",
normalizationFixes = m["Arab"].normalizationFixes,
}
m["ps-Arab"] = m["fa-Arab"]
m["sd-Arab"] = m["fa-Arab"]
m["tt-Arab"] = m["fa-Arab"]
m["ug-Arab"] = m["fa-Arab"]
m["ur-Arab"] = m["fa-Arab"]
-- Aran (Nastaliq) is subsumed into Arab
m["Armi"] = process_ranges{
"Imperial Aramaic",
"abjad",
ranges = {
0x10840, 0x10855,
0x10857, 0x1085F,
},
direction = "rtl",
}
m["Armn"] = process_ranges{
"Armenian",
"alphabet",
ranges = {
0x0531, 0x0556,
0x0559, 0x058A,
0x058D, 0x058F,
0xFB13, 0xFB17,
},
capitalized = true,
}
m["Avst"] = process_ranges{
"Avestan",
"alphabet",
ranges = {
0x10B00, 0x10B35,
0x10B39, 0x10B3F,
},
direction = "rtl",
}
m["pal-Avst"] = {
"Pazend",
m["Avst"][2],
ranges = m["Avst"].ranges,
characters = m["Avst"].characters,
direction = "rtl",
parent = "Avst",
}
m["Bali"] = process_ranges{
"Balinese",
"abugida",
ranges = {
0x1B00, 0x1B4C,
0x1B4E, 0x1B7F,
},
}
m["Bamu"] = process_ranges{
"Bamum",
"syllabary",
ranges = {
0xA6A0, 0xA6F7,
0x16800, 0x16A38,
},
}
m["Bass"] = process_ranges{
"Bassa",
"alphabet",
aliases = {"Bassa Vah", "Vah"},
ranges = {
0x16AD0, 0x16AED,
0x16AF0, 0x16AF5,
},
}
m["Batk"] = process_ranges{
"Batak",
"abugida",
ranges = {
0x1BC0, 0x1BF3,
0x1BFC, 0x1BFF,
},
}
m["Beng"] = process_ranges{
"Bengali",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0980, 0x0983,
0x0985, 0x098C,
0x098F, 0x0990,
0x0993, 0x09A8,
0x09AA, 0x09B0,
0x09B2, 0x09B2,
0x09B6, 0x09B9,
0x09BC, 0x09C4,
0x09C7, 0x09C8,
0x09CB, 0x09CE,
0x09D7, 0x09D7,
0x09DC, 0x09DD,
0x09DF, 0x09E3,
0x09E6, 0x09EF,
0x09F2, 0x09FE,
0x1CD0, 0x1CD0,
0x1CD2, 0x1CD2,
0x1CD5, 0x1CD6,
0x1CD8, 0x1CD8,
0x1CE1, 0x1CE1,
0x1CEA, 0x1CEA,
0x1CED, 0x1CED,
0x1CF2, 0x1CF2,
0x1CF5, 0x1CF7,
0xA8F1, 0xA8F1,
},
normalizationFixes = handle_normalization_fixes{
from = {"অা", "ঋৃ", "ঌৢ"},
to = {"আ", "ৠ", "ৡ"}
},
}
m["as-Beng"] = process_ranges{
"Assamese",
m["Beng"][2],
otherNames = {"Bengali-Assamese", "Eastern Nagari"},
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0980, 0x0983,
0x0985, 0x098C,
0x098F, 0x0990,
0x0993, 0x09A8,
0x09AA, 0x09AF,
0x09B2, 0x09B2,
0x09B6, 0x09B9,
0x09BC, 0x09C4,
0x09C7, 0x09C8,
0x09CB, 0x09CE,
0x09D7, 0x09D7,
0x09DC, 0x09DD,
0x09DF, 0x09E3,
0x09E6, 0x09FE,
0x1CD0, 0x1CD0,
0x1CD2, 0x1CD2,
0x1CD5, 0x1CD6,
0x1CD8, 0x1CD8,
0x1CE1, 0x1CE1,
0x1CEA, 0x1CEA,
0x1CED, 0x1CED,
0x1CF2, 0x1CF2,
0x1CF5, 0x1CF7,
0xA8F1, 0xA8F1,
},
normalizationFixes = m["Beng"].normalizationFixes,
}
m["Bhks"] = process_ranges{
"Bhaiksuki",
"abugida",
ranges = {
0x11C00, 0x11C08,
0x11C0A, 0x11C36,
0x11C38, 0x11C45,
0x11C50, 0x11C6C,
},
}
m["Blis"] = {
"Blissymbols",
"logography",
aliases = {"Blissymbolics"},
-- Not in Unicode
}
m["Bopo"] = process_ranges{
"Zhuyin",
"semisyllabary",
aliases = {"Zhuyin Fuhao", "Bopomofo"},
ranges = {
0x02EA, 0x02EB,
0x3001, 0x3003,
0x3008, 0x3011,
0x3013, 0x301F,
0x302A, 0x302D,
0x3030, 0x3030,
0x3037, 0x3037,
0x30FB, 0x30FB,
0x3105, 0x312F,
0x31A0, 0x31BF,
0xFE45, 0xFE46,
0xFF61, 0xFF65,
},
}
m["Brah"] = process_ranges{
"Brahmi",
"abugida",
ranges = {
0x11000, 0x1104D,
0x11052, 0x11075,
0x1107F, 0x1107F,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑀅𑀸", "𑀋𑀾", "𑀏𑁂"},
to = {"𑀆", "𑀌", "𑀐"}
},
}
m["Brai"] = process_ranges{
"Braille",
"alphabet",
ranges = {
0x2800, 0x28FF,
},
}
m["Bugi"] = process_ranges{
"Lontara",
"abugida",
aliases = {"Buginese"},
ranges = {
0x1A00, 0x1A1B,
0x1A1E, 0x1A1F,
0xA9CF, 0xA9CF,
},
}
m["Buhd"] = process_ranges{
"Buhid",
"abugida",
ranges = {
0x1735, 0x1736,
0x1740, 0x1751,
0x1752, 0x1753,
},
}
m["Cakm"] = process_ranges{
"Chakma",
"abugida",
ranges = {
0x09E6, 0x09EF,
0x1040, 0x1049,
0x11100, 0x11134,
0x11136, 0x11147,
},
}
m["Cans"] = process_ranges{
"Canadian syllabics",
"abugida",
ranges = {
0x1400, 0x167F,
0x18B0, 0x18F5,
0x11AB0, 0x11ABF,
},
}
m["Cari"] = process_ranges{
"Carian",
"alphabet",
ranges = {
0x102A0, 0x102D0,
},
}
m["Cham"] = process_ranges{
"Cham",
"abugida",
ranges = {
0xAA00, 0xAA36,
0xAA40, 0xAA4D,
0xAA50, 0xAA59,
0xAA5C, 0xAA5F,
},
}
m["Cher"] = process_ranges{
"Cherokee",
"syllabary",
ranges = {
0x13A0, 0x13F5,
0x13F8, 0x13FD,
0xAB70, 0xABBF,
},
}
m["Chis"] = {
"Chisoi",
"abugida",
-- Not in Unicode
}
m["Chrs"] = process_ranges{
"Chorasmian",
"abjad",
ranges = {
0x10FB0, 0x10FCB,
},
direction = "rtl",
}
m["Copt"] = process_ranges{
"Coptic",
"alphabet",
ranges = {
0x03E2, 0x03EF,
0x2C80, 0x2CF3,
0x2CF9, 0x2CFF,
0x102E0, 0x102FB,
},
capitalized = true,
}
m["Cpmn"] = process_ranges{
"Cypro-Minoan",
"syllabary",
aliases = {"Cypro Minoan"},
ranges = {
0x10100, 0x10101,
0x12F90, 0x12FF2,
},
}
m["Cprt"] = process_ranges{
"Cypriot",
"syllabary",
ranges = {
0x10100, 0x10102,
0x10107, 0x10133,
0x10137, 0x1013F,
0x10800, 0x10805,
0x10808, 0x10808,
0x1080A, 0x10835,
0x10837, 0x10838,
0x1083C, 0x1083C,
0x1083F, 0x1083F,
},
direction = "rtl",
}
m["Cyrl"] = process_ranges{
"Cyrillic",
"alphabet",
ranges = {
0x0400, 0x052F,
0x1C80, 0x1C8A,
0x1D2B, 0x1D2B,
0x1D78, 0x1D78,
0x1DF8, 0x1DF8,
0x2DE0, 0x2DFF,
0x2E43, 0x2E43,
0xA640, 0xA69F,
0xFE2E, 0xFE2F,
0x1E030, 0x1E06D,
0x1E08F, 0x1E08F,
},
capitalized = true,
}
m["Cyrs"] = {
"Old Cyrillic",
m["Cyrl"][2],
aliases = {"Early Cyrillic"},
ranges = m["Cyrl"].ranges,
characters = m["Cyrl"].characters,
capitalized = m["Cyrl"].capitalized,
wikipedia_article = "Early Cyrillic alphabet",
normalizationFixes = handle_normalization_fixes{
from = {"Ѹ", "ѹ"},
to = {"Ꙋ", "ꙋ"}
},
}
m["Deva"] = process_ranges{
"Devanagari",
"abugida",
ranges = {
0x0900, 0x097F,
0x1CD0, 0x1CF6,
0x1CF8, 0x1CF9,
0x20F0, 0x20F0,
0xA830, 0xA839,
0xA8E0, 0xA8FF,
0x11B00, 0x11B09,
},
normalizationFixes = handle_normalization_fixes{
from = {"ॆॆ", "ेे", "ाॅ", "ाॆ", "ाꣿ", "ॊॆ", "ाे", "ाै", "ोे", "ाऺ", "ॖॖ", "अॅ", "अॆ", "अा", "एॅ", "एॆ", "एे", "एꣿ", "ऎॆ", "अॉ", "आॅ", "अॊ", "आॆ", "अो", "आे", "अौ", "आै", "ओे", "अऺ", "अऻ", "आऺ", "अाꣿ", "आꣿ", "ऒॆ", "अॖ", "अॗ", "ॶॖ", "्?ा"},
to = {"ꣿ", "ै", "ॉ", "ॊ", "ॏ", "ॏ", "ो", "ौ", "ौ", "ऻ", "ॗ", "ॲ", "ऄ", "आ", "ऍ", "ऎ", "ऐ", "ꣾ", "ꣾ", "ऑ", "ऑ", "ऒ", "ऒ", "ओ", "ओ", "औ", "औ", "औ", "ॳ", "ॴ", "ॴ", "ॵ", "ॵ", "ॵ", "ॶ", "ॷ", "ॷ"}
},
}
m["Diak"] = process_ranges{
"Dhives Akuru",
"abugida",
aliases = {"Dhivehi Akuru", "Dives Akuru", "Divehi Akuru"},
ranges = {
0x11900, 0x11906,
0x11909, 0x11909,
0x1190C, 0x11913,
0x11915, 0x11916,
0x11918, 0x11935,
0x11937, 0x11938,
0x1193B, 0x11946,
0x11950, 0x11959,
},
}
m["Dogr"] = process_ranges{
"Dogra",
"abugida",
ranges = {
0x0964, 0x096F,
0xA830, 0xA839,
0x11800, 0x1183B,
},
}
m["Dsrt"] = process_ranges{
"Deseret",
"alphabet",
ranges = {
0x10400, 0x1044F,
},
capitalized = true,
}
m["Dupl"] = process_ranges{
"Duployan",
"alphabet",
ranges = {
0x1BC00, 0x1BC6A,
0x1BC70, 0x1BC7C,
0x1BC80, 0x1BC88,
0x1BC90, 0x1BC99,
0x1BC9C, 0x1BCA3,
},
}
m["Egyd"] = {
"Demotic",
"abjad, logography",
-- Not in Unicode
}
m["Egyh"] = {
"Hieratic",
"abjad, logography",
-- Unified with Egyptian hieroglyphic in Unicode
}
m["Egyp"] = process_ranges{
"Egyptian hieroglyphic",
"abjad, logography",
ranges = {
0x13000, 0x13455,
0x13460, 0x143FA,
},
varieties = {"Hieratic"},
wikipedia_article = "Egyptian hieroglyphs",
normalizationFixes = handle_normalization_fixes{
from = {"𓃁", "𓆖"},
to = {"𓃀𓂝", "𓆓𓏏𓇿"}
},
}
m["Elba"] = process_ranges{
"Elbasan",
"alphabet",
ranges = {
0x10500, 0x10527,
},
}
m["Elym"] = process_ranges{
"Elymaic",
"abjad",
ranges = {
0x10FE0, 0x10FF6,
},
direction = "rtl",
}
m["Ethi"] = process_ranges{
"Ethiopic",
"abugida",
aliases = {"Ge'ez"},
ranges = {
0x1200, 0x1248,
0x124A, 0x124D,
0x1250, 0x1256,
0x1258, 0x1258,
0x125A, 0x125D,
0x1260, 0x1288,
0x128A, 0x128D,
0x1290, 0x12B0,
0x12B2, 0x12B5,
0x12B8, 0x12BE,
0x12C0, 0x12C0,
0x12C2, 0x12C5,
0x12C8, 0x12D6,
0x12D8, 0x1310,
0x1312, 0x1315,
0x1318, 0x135A,
0x135D, 0x137C,
0x1380, 0x1399,
0x2D80, 0x2D96,
0x2DA0, 0x2DA6,
0x2DA8, 0x2DAE,
0x2DB0, 0x2DB6,
0x2DB8, 0x2DBE,
0x2DC0, 0x2DC6,
0x2DC8, 0x2DCE,
0x2DD0, 0x2DD6,
0x2DD8, 0x2DDE,
0xAB01, 0xAB06,
0xAB09, 0xAB0E,
0xAB11, 0xAB16,
0xAB20, 0xAB26,
0xAB28, 0xAB2E,
0x1E7E0, 0x1E7E6,
0x1E7E8, 0x1E7EB,
0x1E7ED, 0x1E7EE,
0x1E7F0, 0x1E7FE,
},
}
m["Gara"] = process_ranges{
"Garay",
"alphabet",
capitalized = true,
direction = "rtl",
ranges = {
0x060C, 0x060C,
0x061B, 0x061B,
0x061F, 0x061F,
0x10D40, 0x10D65,
0x10D69, 0x10D85,
0x10D8E, 0x10D8F,
},
}
m["Geok"] = process_ranges{
"Khutsuri",
"alphabet",
ranges = { -- Ⴀ-Ⴭ is Asomtavruli, ⴀ-ⴭ is Nuskhuri
0x10A0, 0x10C5,
0x10C7, 0x10C7,
0x10CD, 0x10CD,
0x10FB, 0x10FB,
0x2D00, 0x2D25,
0x2D27, 0x2D27,
0x2D2D, 0x2D2D,
},
varieties = {"Nuskhuri", "Asomtavruli"},
capitalized = true,
}
m["Geor"] = process_ranges{
"Georgian",
"alphabet",
ranges = { -- ა-ჿ is lowercase Mkhedruli; Ა-Ჿ is uppercase Mkhedruli (Mtavruli)
0x0589, 0x0589,
0x10D0, 0x10FF,
0x1C90, 0x1CBA,
0x1CBD, 0x1CBF,
},
varieties = {"Mkhedruli", "Mtavruli"},
capitalized = true,
}
m["Glag"] = process_ranges{
"Glagolitic",
"alphabet",
ranges = {
0x0484, 0x0484,
0x0487, 0x0487,
0x0589, 0x0589,
0x10FB, 0x10FB,
0x2C00, 0x2C5F,
0x2E43, 0x2E43,
0xA66F, 0xA66F,
0x1E000, 0x1E006,
0x1E008, 0x1E018,
0x1E01B, 0x1E021,
0x1E023, 0x1E024,
0x1E026, 0x1E02A,
},
capitalized = true,
}
m["Gong"] = process_ranges{
"Gunjala Gondi",
"abugida",
ranges = {
0x0964, 0x0965,
0x11D60, 0x11D65,
0x11D67, 0x11D68,
0x11D6A, 0x11D8E,
0x11D90, 0x11D91,
0x11D93, 0x11D98,
0x11DA0, 0x11DA9,
},
}
m["Gonm"] = process_ranges{
"Masaram Gondi",
"abugida",
ranges = {
0x0964, 0x0965,
0x11D00, 0x11D06,
0x11D08, 0x11D09,
0x11D0B, 0x11D36,
0x11D3A, 0x11D3A,
0x11D3C, 0x11D3D,
0x11D3F, 0x11D47,
0x11D50, 0x11D59,
},
}
m["Goth"] = process_ranges{
"Gothic",
"alphabet",
ranges = {
0x10330, 0x1034A,
},
wikipedia_article = "Gothic alphabet",
}
m["Gran"] = process_ranges{
"Grantha",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0BE6, 0x0BF3,
0x1CD0, 0x1CD0,
0x1CD2, 0x1CD3,
0x1CF2, 0x1CF4,
0x1CF8, 0x1CF9,
0x20F0, 0x20F0,
0x11300, 0x11303,
0x11305, 0x1130C,
0x1130F, 0x11310,
0x11313, 0x11328,
0x1132A, 0x11330,
0x11332, 0x11333,
0x11335, 0x11339,
0x1133B, 0x11344,
0x11347, 0x11348,
0x1134B, 0x1134D,
0x11350, 0x11350,
0x11357, 0x11357,
0x1135D, 0x11363,
0x11366, 0x1136C,
0x11370, 0x11374,
0x11FD0, 0x11FD1,
0x11FD3, 0x11FD3,
},
}
m["Grek"] = process_ranges{
"Greek",
"alphabet",
ranges = {
0x0342, 0x0342,
0x0345, 0x0345,
0x0370, 0x0377,
0x037A, 0x037F,
0x0384, 0x038A,
0x038C, 0x038C,
0x038E, 0x03A1,
0x03A3, 0x03E1,
0x03F0, 0x03FF,
0x1D26, 0x1D2A,
0x1D5D, 0x1D61,
0x1D66, 0x1D6A,
0x1DBF, 0x1DBF,
0x2126, 0x2127,
0x2129, 0x2129,
0x213C, 0x2140,
0xAB65, 0xAB65,
0x10140, 0x1018E,
0x101A0, 0x101A0,
0x1D200, 0x1D245,
},
capitalized = true,
}
m["Polyt"] = process_ranges{
"Greek",
m["Grek"][2],
ranges = union(m["Grek"].ranges, {
0x1F00, 0x1F15,
0x1F18, 0x1F1D,
0x1F20, 0x1F45,
0x1F48, 0x1F4D,
0x1F50, 0x1F57,
0x1F59, 0x1F59,
0x1F5B, 0x1F5B,
0x1F5D, 0x1F5D,
0x1F5F, 0x1F7D,
0x1F80, 0x1FB4,
0x1FB6, 0x1FC4,
0x1FC6, 0x1FD3,
0x1FD6, 0x1FDB,
0x1FDD, 0x1FEF,
0x1FF2, 0x1FF4,
0x1FF6, 0x1FFE,
}),
ietf_subtag = "Grek",
capitalized = m["Grek"].capitalized,
parent = "Grek",
}
m["Gujr"] = process_ranges{
"Gujarati",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0A81, 0x0A83,
0x0A85, 0x0A8D,
0x0A8F, 0x0A91,
0x0A93, 0x0AA8,
0x0AAA, 0x0AB0,
0x0AB2, 0x0AB3,
0x0AB5, 0x0AB9,
0x0ABC, 0x0AC5,
0x0AC7, 0x0AC9,
0x0ACB, 0x0ACD,
0x0AD0, 0x0AD0,
0x0AE0, 0x0AE3,
0x0AE6, 0x0AF1,
0x0AF9, 0x0AFF,
0xA830, 0xA839,
},
normalizationFixes = handle_normalization_fixes{
from = {"ઓ", "અાૈ", "અા", "અૅ", "અે", "અૈ", "અૉ", "અો", "અૌ", "આૅ", "આૈ", "ૅા"},
to = {"અાૅ", "ઔ", "આ", "ઍ", "એ", "ઐ", "ઑ", "ઓ", "ઔ", "ઓ", "ઔ", "ૉ"}
},
}
m["Gukh"] = process_ranges{
"Khema",
"abugida",
aliases = {"Gurung Khema", "Khema Phri", "Khema Lipi"},
ranges = {
0x0965, 0x0965,
0x16100, 0x16139,
},
}
m["Guru"] = process_ranges{
"Gurmukhi",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0A01, 0x0A03,
0x0A05, 0x0A0A,
0x0A0F, 0x0A10,
0x0A13, 0x0A28,
0x0A2A, 0x0A30,
0x0A32, 0x0A33,
0x0A35, 0x0A36,
0x0A38, 0x0A39,
0x0A3C, 0x0A3C,
0x0A3E, 0x0A42,
0x0A47, 0x0A48,
0x0A4B, 0x0A4D,
0x0A51, 0x0A51,
0x0A59, 0x0A5C,
0x0A5E, 0x0A5E,
0x0A66, 0x0A76,
0xA830, 0xA839,
},
normalizationFixes = handle_normalization_fixes{
from = {"ਅਾ", "ਅੈ", "ਅੌ", "ੲਿ", "ੲੀ", "ੲੇ", "ੳੁ", "ੳੂ", "ੳੋ"},
to = {"ਆ", "ਐ", "ਔ", "ਇ", "ਈ", "ਏ", "ਉ", "ਊ", "ਓ"}
},
}
m["Hang"] = process_ranges{
"Hangul",
"syllabary",
aliases = {"Hangeul"},
ranges = {
0x1100, 0x11FF,
0x3001, 0x3003,
0x3008, 0x3011,
0x3013, 0x301F,
0x302E, 0x3030,
0x3037, 0x3037,
0x30FB, 0x30FB,
0x3131, 0x318E,
0x3200, 0x321E,
0x3260, 0x327E,
0xA960, 0xA97C,
0xAC00, 0xD7A3,
0xD7B0, 0xD7C6,
0xD7CB, 0xD7FB,
0xFE45, 0xFE46,
0xFF61, 0xFF65,
0xFFA0, 0xFFBE,
0xFFC2, 0xFFC7,
0xFFCA, 0xFFCF,
0xFFD2, 0xFFD7,
0xFFDA, 0xFFDC,
},
}
m["Hani"] = process_ranges{
"Han",
"logography",
ranges = {
0x2E80, 0x2E99,
0x2E9B, 0x2EF3,
0x2F00, 0x2FD5,
0x2FF0, 0x2FFF,
0x3001, 0x3003,
0x3005, 0x3011,
0x3013, 0x301F,
0x3021, 0x302D,
0x3030, 0x3030,
0x3037, 0x303F,
0x3190, 0x319F,
0x31C0, 0x31E5,
0x31EF, 0x31EF,
0x3220, 0x3247,
0x3280, 0x32B0,
0x32C0, 0x32CB,
0x30FB, 0x30FB,
0x32FF, 0x32FF,
0x3358, 0x3370,
0x337B, 0x337F,
0x33E0, 0x33FE,
0x3400, 0x4DBF,
0x4E00, 0x9FFF,
0xA700, 0xA707,
0xF900, 0xFA6D,
0xFA70, 0xFAD9,
0xFE45, 0xFE46,
0xFF61, 0xFF65,
0x16FE2, 0x16FE3,
0x16FF0, 0x16FF1,
0x1D360, 0x1D371,
0x1F250, 0x1F251,
0x20000, 0x2A6DF,
0x2A700, 0x2B739,
0x2B740, 0x2B81D,
0x2B820, 0x2CEA1,
0x2CEB0, 0x2EBE0,
0x2EBF0, 0x2EE5D,
0x2F800, 0x2FA1D,
0x30000, 0x3134A,
0x31350, 0x323AF,
},
varieties = {"Hanzi", "Kanji", "Hanja", "Chu Nom"},
spaces = false,
}
m["Hans"] = {
"Simplified Han",
m["Hani"][2],
ranges = m["Hani"].ranges,
characters = m["Hani"].characters,
spaces = m["Hani"].spaces,
parent = "Hani",
}
m["Hant"] = {
"Traditional Han",
m["Hani"][2],
ranges = m["Hani"].ranges,
characters = m["Hani"].characters,
spaces = m["Hani"].spaces,
parent = "Hani",
}
m["Hano"] = process_ranges{
"Hanunoo",
"abugida",
ranges = {
0x1720, 0x1736,
},
}
m["Hatr"] = process_ranges{
"Hatran",
"abjad",
ranges = {
0x108E0, 0x108F2,
0x108F4, 0x108F5,
0x108FB, 0x108FF,
},
direction = "rtl",
}
m["Hebr"] = process_ranges{
"Hebrew",
"abjad", -- more precisely, impure abjad
ranges = {
0x0591, 0x05C7,
0x05D0, 0x05EA,
0x05EF, 0x05F4,
0x2135, 0x2138,
0xFB1D, 0xFB36,
0xFB38, 0xFB3C,
0xFB3E, 0xFB3E,
0xFB40, 0xFB41,
0xFB43, 0xFB44,
0xFB46, 0xFB4F,
},
direction = "rtl",
}
m["Hira"] = process_ranges{
"Hiragana",
"syllabary",
ranges = {
0x3001, 0x3003,
0x3008, 0x3011,
0x3013, 0x301F,
0x3030, 0x3035,
0x3037, 0x3037,
0x303C, 0x303D,
0x3041, 0x3096,
0x3099, 0x30A0,
0x30FB, 0x30FC,
0xFE45, 0xFE46,
0xFF61, 0xFF65,
0xFF70, 0xFF70,
0xFF9E, 0xFF9F,
0x1B001, 0x1B11F,
0x1B132, 0x1B132,
0x1B150, 0x1B152,
0x1F200, 0x1F200,
},
varieties = {"Hentaigana"},
spaces = false,
}
m["Hluw"] = process_ranges{
"Anatolian Hieroglyphs",
"logography, syllabary",
ranges = {
0x14400, 0x14646,
},
wikipedia_article = "Anatolian hieroglyphs",
}
m["Hmng"] = process_ranges{
"Pahawh Hmong",
"semisyllabary",
aliases = {"Hmong"},
ranges = {
0x16B00, 0x16B45,
0x16B50, 0x16B59,
0x16B5B, 0x16B61,
0x16B63, 0x16B77,
0x16B7D, 0x16B8F,
},
}
m["Hmnp"] = process_ranges{
"Nyiakeng Puachue Hmong",
"alphabet",
ranges = {
0x1E100, 0x1E12C,
0x1E130, 0x1E13D,
0x1E140, 0x1E149,
0x1E14E, 0x1E14F,
},
}
m["Hung"] = process_ranges{
"Old Hungarian",
"alphabet",
aliases = {"Hungarian runic"},
ranges = {
0x10C80, 0x10CB2,
0x10CC0, 0x10CF2,
0x10CFA, 0x10CFF,
},
capitalized = true,
direction = "rtl",
}
m["Ibrnn"] = {
"Northeastern Iberian",
"semisyllabary",
ietf_subtag = "Zzzz",
-- Not in Unicode
}
m["Ibrns"] = {
"Southeastern Iberian",
"semisyllabary",
ietf_subtag = "Zzzz",
-- Not in Unicode
}
m["Image"] = {
-- To be used to avoid any formatting or link processing
"Image-rendered",
-- This should not have any characters listed
ietf_subtag = "Zyyy",
translit = false,
character_category = false,
}
m["Inds"] = {
"Indus",
aliases = {"Harappan", "Indus Valley"},
}
m["Ipach"] = {
"International Phonetic Alphabet",
aliases = {"IPA"},
ietf_subtag = "Latn",
}
m["Ital"] = process_ranges{
"Old Italic",
"alphabet",
ranges = {
0x10300, 0x10323,
0x1032D, 0x1032F,
},
}
m["Java"] = process_ranges{
"Javanese",
"abugida",
ranges = {
0xA980, 0xA9CD,
0xA9CF, 0xA9D9,
0xA9DE, 0xA9DF,
},
}
m["Jurc"] = {
"Jurchen",
"logography",
spaces = false,
}
m["Kali"] = process_ranges{
"Kayah Li",
"abugida",
ranges = {
0xA900, 0xA92F,
},
}
m["Kana"] = process_ranges{
"Katakana",
"syllabary",
ranges = {
0x3001, 0x3003,
0x3008, 0x3011,
0x3013, 0x301F,
0x3030, 0x3035,
0x3037, 0x3037,
0x303C, 0x303D,
0x3099, 0x309C,
0x30A0, 0x30FF,
0x31F0, 0x31FF,
0x32D0, 0x32FE,
0x3300, 0x3357,
0xFE45, 0xFE46,
0xFF61, 0xFF9F,
0x1AFF0, 0x1AFF3,
0x1AFF5, 0x1AFFB,
0x1AFFD, 0x1AFFE,
0x1B000, 0x1B000,
0x1B120, 0x1B122,
0x1B155, 0x1B155,
0x1B164, 0x1B167,
},
spaces = false,
}
m["Kawi"] = process_ranges{
"Kawi",
"abugida",
ranges = {
0x11F00, 0x11F10,
0x11F12, 0x11F3A,
0x11F3E, 0x11F5A,
},
}
m["Khar"] = process_ranges{
"Kharoshthi",
"abugida",
ranges = {
0x10A00, 0x10A03,
0x10A05, 0x10A06,
0x10A0C, 0x10A13,
0x10A15, 0x10A17,
0x10A19, 0x10A35,
0x10A38, 0x10A3A,
0x10A3F, 0x10A48,
0x10A50, 0x10A58,
},
direction = "rtl",
}
m["Khmr"] = process_ranges{
"Khmer",
"abugida",
ranges = {
0x1780, 0x17DD,
0x17E0, 0x17E9,
0x17F0, 0x17F9,
0x19E0, 0x19FF,
},
spaces = false,
normalizationFixes = handle_normalization_fixes{
from = {"ឣ", "ឤ"},
to = {"អ", "អា"}
},
}
m["Khoj"] = process_ranges{
"Khojki",
"abugida",
ranges = {
0x0AE6, 0x0AEF,
0xA830, 0xA839,
0x11200, 0x11211,
0x11213, 0x11241,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑈀𑈬𑈱", "𑈀𑈬", "𑈀𑈱", "𑈀𑈳", "𑈁𑈱", "𑈆𑈬", "𑈬𑈰", "𑈬𑈱", "𑉀𑈮"},
to = {"𑈇", "𑈁", "𑈅", "𑈇", "𑈇", "𑈃", "𑈲", "𑈳", "𑈂"}
},
}
m["Kitl"] = {
"Khitan Large",
"logography",
spaces = false,
}
m["Kits"] = process_ranges{
"Khitan Small",
"logography, syllabary",
ranges = {
0x16FE4, 0x16FE4,
0x18B00, 0x18CD5,
0x18CFF, 0x18CFF,
},
spaces = false,
}
m["Knda"] = process_ranges{
"Kannada",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0C80, 0x0C8C,
0x0C8E, 0x0C90,
0x0C92, 0x0CA8,
0x0CAA, 0x0CB3,
0x0CB5, 0x0CB9,
0x0CBC, 0x0CC4,
0x0CC6, 0x0CC8,
0x0CCA, 0x0CCD,
0x0CD5, 0x0CD6,
0x0CDD, 0x0CDE,
0x0CE0, 0x0CE3,
0x0CE6, 0x0CEF,
0x0CF1, 0x0CF3,
0x1CD0, 0x1CD0,
0x1CD2, 0x1CD3,
0x1CDA, 0x1CDA,
0x1CF2, 0x1CF2,
0x1CF4, 0x1CF4,
0xA830, 0xA835,
},
normalizationFixes = handle_normalization_fixes{
from = {"ಉಾ", "ಋಾ", "ಒೌ"},
to = {"ಊ", "ೠ", "ಔ"}
},
}
m["Kpel"] = {
"Kpelle",
"syllabary",
-- Not in Unicode
}
m["Krai"] = process_ranges{
"Kirat Rai",
"abugida",
aliases = {"Rai", "Khambu Rai", "Rai Barṇamālā", "Kirat Khambu Rai"},
ranges = {
0x16D40, 0x16D79,
},
}
m["Kthi"] = process_ranges{
"Kaithi",
"abugida",
ranges = {
0x0966, 0x096F,
0xA830, 0xA839,
0x11080, 0x110C2,
0x110CD, 0x110CD,
},
}
m["Lana"] = process_ranges{
"Tai Tham",
"abugida",
aliases = {"Tham", "Tua Mueang", "Lanna"},
ranges = {
0x1A20, 0x1A5E,
0x1A60, 0x1A7C,
0x1A7F, 0x1A89,
0x1A90, 0x1A99,
0x1AA0, 0x1AAD,
},
spaces = false,
}
m["Laoo"] = process_ranges{
"Lao",
"abugida",
ranges = {
0x0E81, 0x0E82,
0x0E84, 0x0E84,
0x0E86, 0x0E8A,
0x0E8C, 0x0EA3,
0x0EA5, 0x0EA5,
0x0EA7, 0x0EBD,
0x0EC0, 0x0EC4,
0x0EC6, 0x0EC6,
0x0EC8, 0x0ECE,
0x0ED0, 0x0ED9,
0x0EDC, 0x0EDF,
},
spaces = false,
}
m["Latn"] = process_ranges{
"Latin",
"alphabet",
aliases = {"Roman"},
ranges = {
0x0041, 0x005A,
0x0061, 0x007A,
0x00AA, 0x00AA,
0x00BA, 0x00BA,
0x00C0, 0x00D6,
0x00D8, 0x00F6,
0x00F8, 0x02B8,
0x02C0, 0x02C1,
0x02E0, 0x02E4,
0x0363, 0x036F,
0x0485, 0x0486,
0x0951, 0x0952,
0x10FB, 0x10FB,
0x1D00, 0x1D25,
0x1D2C, 0x1D5C,
0x1D62, 0x1D65,
0x1D6B, 0x1D77,
0x1D79, 0x1DBE,
0x1DF8, 0x1DF8,
0x1E00, 0x1EFF,
0x202F, 0x202F,
0x2071, 0x2071,
0x207F, 0x207F,
0x2090, 0x209C,
0x20F0, 0x20F0,
0x2100, 0x2125,
0x2128, 0x2128,
0x212A, 0x2134,
0x2139, 0x213B,
0x2141, 0x214E,
0x2160, 0x2188,
0x2C60, 0x2C7F,
0xA700, 0xA707,
0xA722, 0xA787,
0xA78B, 0xA7CD,
0xA7D0, 0xA7D1,
0xA7D3, 0xA7D3,
0xA7D5, 0xA7DC,
0xA7F2, 0xA7FF,
0xA92E, 0xA92E,
0xAB30, 0xAB5A,
0xAB5C, 0xAB64,
0xAB66, 0xAB69,
0xFB00, 0xFB06,
0xFF21, 0xFF3A,
0xFF41, 0xFF5A,
0x10780, 0x10785,
0x10787, 0x107B0,
0x107B2, 0x107BA,
0x1DF00, 0x1DF1E,
0x1DF25, 0x1DF2A,
},
varieties = {"Rumi", "Romaji", "Rōmaji", "Romaja"},
capitalized = true,
translit = false,
}
m["Latf"] = {
"Fraktur",
m["Latn"][2],
ranges = m["Latn"].ranges,
characters = m["Latn"].characters,
otherNames = {"Blackletter"}, -- Blackletter is actually the parent "script"
capitalized = m["Latn"].capitalized,
translit = m["Latn"].translit,
parent = "Latn",
}
m["Latg"] = {
"Gaelic",
m["Latn"][2],
ranges = m["Latn"].ranges,
characters = m["Latn"].characters,
otherNames = {"Irish"},
capitalized = m["Latn"].capitalized,
translit = m["Latn"].translit,
parent = "Latn",
}
m["pjt-Latn"] = {
"Latin",
m["Latn"][2],
ranges = m["Latn"].ranges,
characters = m["Latn"].characters,
capitalized = m["Latn"].capitalized,
translit = m["Latn"].translit,
parent = "Latn",
}
m["Leke"] = {
"Leke",
"abugida",
-- Not in Unicode
}
m["Lepc"] = process_ranges{
"Lepcha",
"abugida",
ranges = {
0x1C00, 0x1C37,
0x1C3B, 0x1C49,
0x1C4D, 0x1C4F,
},
}
m["Limb"] = process_ranges{
"Limbu",
"abugida",
ranges = {
0x0965, 0x0965,
0x1900, 0x191E,
0x1920, 0x192B,
0x1930, 0x193B,
0x1940, 0x1940,
0x1944, 0x194F,
},
}
m["Lina"] = process_ranges{
"Linear A",
ranges = {
0x10107, 0x10133,
0x10600, 0x10736,
0x10740, 0x10755,
0x10760, 0x10767,
},
}
m["Linb"] = process_ranges{
"Linear B",
ranges = {
0x10000, 0x1000B,
0x1000D, 0x10026,
0x10028, 0x1003A,
0x1003C, 0x1003D,
0x1003F, 0x1004D,
0x10050, 0x1005D,
0x10080, 0x100FA,
0x10100, 0x10102,
0x10107, 0x10133,
0x10137, 0x1013F,
},
}
m["Lisu"] = process_ranges{
"Fraser",
"alphabet",
aliases = {"Old Lisu", "Lisu"},
ranges = {
0x300A, 0x300B,
0xA4D0, 0xA4FF,
0x11FB0, 0x11FB0,
},
normalizationFixes = handle_normalization_fixes{
from = {"['’]", "[.ꓸ][.ꓸ]", "[.ꓸ][,ꓹ]"},
to = {"ʼ", "ꓺ", "ꓻ"}
},
}
m["Loma"] = {
"Loma",
"syllabary",
-- Not in Unicode
}
m["Lyci"] = process_ranges{
"Lycian",
"alphabet",
ranges = {
0x10280, 0x1029C,
},
}
m["Lydi"] = process_ranges{
"Lydian",
"alphabet",
ranges = {
0x10920, 0x10939,
0x1093F, 0x1093F,
},
direction = "rtl",
}
m["Mahj"] = process_ranges{
"Mahajani",
"abugida",
ranges = {
0x0964, 0x096F,
0xA830, 0xA839,
0x11150, 0x11176,
},
}
m["Maka"] = process_ranges{
"Makasar",
"abugida",
aliases = {"Old Makasar"},
ranges = {
0x11EE0, 0x11EF8,
},
}
m["Mand"] = process_ranges{
"Mandaic",
aliases = {"Mandaean"},
ranges = {
0x0640, 0x0640,
0x0840, 0x085B,
0x085E, 0x085E,
},
direction = "rtl",
}
m["Mani"] = process_ranges{
"Manichaean",
"abjad",
ranges = {
0x0640, 0x0640,
0x10AC0, 0x10AE6,
0x10AEB, 0x10AF6,
},
direction = "rtl",
}
m["Marc"] = process_ranges{
"Marchen",
"abugida",
ranges = {
0x11C70, 0x11C8F,
0x11C92, 0x11CA7,
0x11CA9, 0x11CB6,
},
}
m["Maya"] = process_ranges{
"Maya",
aliases = {"Maya hieroglyphic", "Mayan", "Mayan hieroglyphic"},
ranges = {
0x1D2E0, 0x1D2F3,
},
}
m["Medf"] = process_ranges{
"Medefaidrin",
aliases = {"Oberi Okaime", "Oberi Ɔkaimɛ"},
ranges = {
0x16E40, 0x16E9A,
},
capitalized = true,
}
m["Mend"] = process_ranges{
"Mende",
aliases = {"Mende Kikakui"},
ranges = {
0x1E800, 0x1E8C4,
0x1E8C7, 0x1E8D6,
},
direction = "rtl",
}
m["Merc"] = process_ranges{
"Meroitic cursive",
"abugida",
ranges = {
0x109A0, 0x109B7,
0x109BC, 0x109CF,
0x109D2, 0x109FF,
},
direction = "rtl",
}
m["Mero"] = process_ranges{
"Meroitic hieroglyphic",
"abugida",
ranges = {
0x10980, 0x1099F,
},
direction = "rtl",
}
m["Mlym"] = process_ranges{
"Malayalam",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0D00, 0x0D0C,
0x0D0E, 0x0D10,
0x0D12, 0x0D44,
0x0D46, 0x0D48,
0x0D4A, 0x0D4F,
0x0D54, 0x0D63,
0x0D66, 0x0D7F,
0x1CDA, 0x1CDA,
0x1CF2, 0x1CF2,
0xA830, 0xA832,
},
normalizationFixes = handle_normalization_fixes{
from = {"ഇൗ", "ഉൗ", "എെ", "ഒാ", "ഒൗ", "ക്", "ണ്", "ന്റ", "ന്", "മ്", "യ്", "ര്", "ല്", "ള്", "ഴ്", "െെ", "ൻ്റ"},
to = {"ഈ", "ഊ", "ഐ", "ഓ", "ഔ", "ൿ", "ൺ", "ൻറ", "ൻ", "ൔ", "ൕ", "ർ", "ൽ", "ൾ", "ൖ", "ൈ", "ന്റ"}
},
}
m["Modi"] = process_ranges{
"Modi",
"abugida",
ranges = {
0xA830, 0xA839,
0x11600, 0x11644,
0x11650, 0x11659,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑘀𑘹", "𑘀𑘺", "𑘁𑘹", "𑘁𑘺"},
to = {"𑘊", "𑘋", "𑘌", "𑘍"}
},
}
m["Mong"] = process_ranges{
"Mongolian",
"alphabet",
aliases = {"Mongol bichig", "Hudum Mongol bichig"},
ranges = {
0x1800, 0x1805,
0x180A, 0x1819,
0x1820, 0x1842,
0x1878, 0x1878,
0x1880, 0x1897,
0x18A6, 0x18A6,
0x18A9, 0x18A9,
0x200C, 0x200D,
0x202F, 0x202F,
0x3001, 0x3002,
0x3008, 0x300B,
0x11660, 0x11668,
},
direction = "vertical-ltr",
}
m["mnc-Mong"] = process_ranges{
"Manchu",
m["Mong"][2],
ranges = {
0x1801, 0x1801,
0x1804, 0x1804,
0x1808, 0x180F,
0x1820, 0x1820,
0x1823, 0x1823,
0x1828, 0x182A,
0x182E, 0x1830,
0x1834, 0x1838,
0x183A, 0x183A,
0x185D, 0x185D,
0x185F, 0x1861,
0x1864, 0x1869,
0x186C, 0x1871,
0x1873, 0x1877,
0x1880, 0x1888,
0x188F, 0x188F,
0x189A, 0x18A5,
0x18A8, 0x18A8,
0x18AA, 0x18AA,
0x200C, 0x200D,
0x202F, 0x202F,
},
direction = "vertical-ltr",
parent = "Mong",
}
m["sjo-Mong"] = process_ranges{
"Xibe",
m["Mong"][2],
aliases = {"Sibe"},
ranges = {
0x1804, 0x1804,
0x1807, 0x1807,
0x180A, 0x180F,
0x1820, 0x1820,
0x1823, 0x1823,
0x1828, 0x1828,
0x182A, 0x182A,
0x182E, 0x1830,
0x1834, 0x1838,
0x183A, 0x183A,
0x185D, 0x1872,
0x200C, 0x200D,
0x202F, 0x202F,
},
direction = "vertical-ltr",
parent = "mnc-Mong",
}
m["xwo-Mong"] = process_ranges{
"Clear Script",
m["Mong"][2],
aliases = {"Todo", "Todo bichig"},
ranges = {
0x1800, 0x1801,
0x1804, 0x1806,
0x180A, 0x1820,
0x1828, 0x1828,
0x182F, 0x1831,
0x1834, 0x1834,
0x1837, 0x1838,
0x183A, 0x183B,
0x1840, 0x1840,
0x1843, 0x185C,
0x1880, 0x1887,
0x1889, 0x188F,
0x1894, 0x1894,
0x1896, 0x1899,
0x18A7, 0x18A7,
0x200C, 0x200D,
0x202F, 0x202F,
0x11669, 0x1166C,
},
direction = "vertical-ltr",
parent = "Mong",
}
m["Moon"] = {
"Moon",
"alphabet",
aliases = {"Moon System of Embossed Reading", "Moon type", "Moon writing", "Moon alphabet", "Moon code"},
-- Not in Unicode
}
m["Morse"] = {
"Morse code",
ietf_subtag = "Zsym",
}
m["Mroo"] = process_ranges{
"Mro",
ranges = {
0x16A40, 0x16A5E,
0x16A60, 0x16A69,
0x16A6E, 0x16A6F,
},
}
m["Mtei"] = process_ranges{
"Meitei Mayek",
"abugida",
aliases = {"Meetei Mayek", "Manipuri"},
ranges = {
0xAAE0, 0xAAF6,
0xABC0, 0xABED,
0xABF0, 0xABF9,
},
}
m["Mult"] = process_ranges{
"Multani",
"abugida",
ranges = {
0x0A66, 0x0A6F,
0x11280, 0x11286,
0x11288, 0x11288,
0x1128A, 0x1128D,
0x1128F, 0x1129D,
0x1129F, 0x112A9,
},
}
m["Music"] = process_ranges{
"Musical notation",
"pictography",
ranges = {
0x2669, 0x266F,
0x1D100, 0x1D126,
0x1D129, 0x1D1EA,
},
ietf_subtag = "Zsym",
translit = false,
}
m["Mymr"] = process_ranges{
"Burmese",
"abugida",
aliases = {"Myanmar"},
ranges = {
0x1000, 0x109F,
0xA92E, 0xA92E,
0xA9E0, 0xA9FE,
0xAA60, 0xAA7F,
0x116D0, 0x116E3,
},
spaces = false,
}
m["Nagm"] = process_ranges{
"Nag Mundari",
"alphabet",
ranges = {
0x1E4D0, 0x1E4F9,
},
}
m["Nand"] = process_ranges{
"Nandinagari",
"abugida",
ranges = {
0x0964, 0x0965,
0x0CE6, 0x0CEF,
0x1CE9, 0x1CE9,
0x1CF2, 0x1CF2,
0x1CFA, 0x1CFA,
0xA830, 0xA835,
0x119A0, 0x119A7,
0x119AA, 0x119D7,
0x119DA, 0x119E4,
},
}
m["Narb"] = process_ranges{
"Old North Arabian",
"abjad",
ranges = {
0x10A80, 0x10A9F,
},
direction = "rtl",
}
m["Nbat"] = process_ranges{
"Nabataean",
"abjad",
aliases = {"Nabatean"},
ranges = {
0x10880, 0x1089E,
0x108A7, 0x108AF,
},
direction = "rtl",
}
m["Newa"] = process_ranges{
"Newa",
"abugida",
aliases = {"Newar", "Newari", "Prachalit Nepal"},
ranges = {
0x11400, 0x1145B,
0x1145D, 0x11461,
},
}
m["Nkdb"] = {
"Dongba",
"pictography",
aliases = {"Naxi Dongba", "Nakhi Dongba", "Tomba", "Tompa", "Mo-so"},
spaces = false,
-- Not in Unicode
}
m["Nkgb"] = {
"Geba",
"syllabary",
aliases = {"Nakhi Geba", "Naxi Geba"},
spaces = false,
-- Not in Unicode
}
m["Nkoo"] = process_ranges{
"N'Ko",
"alphabet",
ranges = {
0x060C, 0x060C,
0x061B, 0x061B,
0x061F, 0x061F,
0x07C0, 0x07FA,
0x07FD, 0x07FF,
0xFD3E, 0xFD3F,
},
direction = "rtl",
}
m["None"] = {
"Unspecified", -- renders as 'unspecified script'
-- This should not have any characters listed
ietf_subtag = "Zyyy",
translit = false,
character_category = false,
}
m["Nshu"] = process_ranges{
"Nüshu",
"syllabary",
aliases = {"Nushu"},
ranges = {
0x16FE1, 0x16FE1,
0x1B170, 0x1B2FB,
},
spaces = false,
}
m["Ogam"] = process_ranges{
"Ogham",
ranges = {
0x1680, 0x169C,
},
}
m["Olck"] = process_ranges{
"Ol Chiki",
aliases = {"Ol Chemetʼ", "Ol", "Santali"},
ranges = {
0x1C50, 0x1C7F,
},
}
m["Onao"] = process_ranges{
"Ol Onal",
"alphabet",
ranges = {
0x0964, 0x0965,
0x1E5D0, 0x1E5FA,
0x1E5FF, 0x1E5FF,
},
}
m["Orkh"] = process_ranges{
"Orkhon runes",
aliases = {"Old Turkic"},
ranges = {
0x10C00, 0x10C48,
},
direction = "rtl",
}
m["Orya"] = process_ranges{
"Odia",
"abugida",
aliases = {"Oriya"},
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0B01, 0x0B03,
0x0B05, 0x0B0C,
0x0B0F, 0x0B10,
0x0B13, 0x0B28,
0x0B2A, 0x0B30,
0x0B32, 0x0B33,
0x0B35, 0x0B39,
0x0B3C, 0x0B44,
0x0B47, 0x0B48,
0x0B4B, 0x0B4D,
0x0B55, 0x0B57,
0x0B5C, 0x0B5D,
0x0B5F, 0x0B63,
0x0B66, 0x0B77,
0x1CDA, 0x1CDA,
0x1CF2, 0x1CF2,
},
normalizationFixes = handle_normalization_fixes{
from = {"ଅା", "ଏୗ", "ଓୗ"},
to = {"ଆ", "ଐ", "ଔ"}
},
}
m["Osge"] = process_ranges{
"Osage",
ranges = {
0x104B0, 0x104D3,
0x104D8, 0x104FB,
},
capitalized = true,
}
m["Osma"] = process_ranges{
"Osmanya",
ranges = {
0x10480, 0x1049D,
0x104A0, 0x104A9,
},
}
m["Ougr"] = process_ranges{
"Old Uyghur",
"abjad, alphabet",
ranges = {
0x0640, 0x0640,
0x10AF2, 0x10AF2,
0x10F70, 0x10F89,
},
-- This should ideally be "vertical-ltr", but getting the CSS right is tricky because it's right-to-left horizontally, but left-to-right vertically. Currently, displaying it vertically causes it to display bottom-to-top.
direction = "rtl",
}
m["Palm"] = process_ranges{
"Palmyrene",
ranges = {
0x10860, 0x1087F,
},
direction = "rtl",
}
m["Pauc"] = process_ranges{
"Pau Cin Hau",
ranges = {
0x11AC0, 0x11AF8,
},
}
m["Perm"] = process_ranges{
"Old Permic",
ranges = {
0x0483, 0x0483,
0x10350, 0x1037A,
},
}
m["Phag"] = process_ranges{
"Phags-pa",
"abugida",
ranges = {
0x1802, 0x1803,
0x1805, 0x1805,
0x200C, 0x200D,
0x202F, 0x202F,
0x3002, 0x3002,
0xA840, 0xA877,
},
direction = "vertical-ltr",
}
m["Phli"] = process_ranges{
"Inscriptional Pahlavi",
"abjad",
ranges = {
0x10B60, 0x10B72,
0x10B78, 0x10B7F,
},
direction = "rtl",
}
m["Phlp"] = process_ranges{
"Psalter Pahlavi",
"abjad",
ranges = {
0x0640, 0x0640,
0x10B80, 0x10B91,
0x10B99, 0x10B9C,
0x10BA9, 0x10BAF,
},
direction = "rtl",
}
m["Phlv"] = {
"Book Pahlavi",
"abjad",
direction = "rtl",
-- Not in Unicode
}
m["Phnx"] = process_ranges{
"Phoenician",
"abjad",
ranges = {
0x10900, 0x1091B,
0x1091F, 0x1091F,
},
direction = "rtl",
}
m["Plrd"] = process_ranges{
"Pollard",
"abugida",
aliases = {"Miao"},
ranges = {
0x16F00, 0x16F4A,
0x16F4F, 0x16F87,
0x16F8F, 0x16F9F,
},
}
m["Prti"] = process_ranges{
"Inscriptional Parthian",
ranges = {
0x10B40, 0x10B55,
0x10B58, 0x10B5F,
},
direction = "rtl",
}
m["Ranj"] = {
"Ranjana",
"abugida",
-- Not in Unicode
}
m["Rjng"] = process_ranges{
"Rejang",
"abugida",
ranges = {
0xA930, 0xA953,
0xA95F, 0xA95F,
},
}
m["Rohg"] = process_ranges{
"Hanifi Rohingya",
"alphabet",
ranges = {
0x060C, 0x060C,
0x061B, 0x061B,
0x061F, 0x061F,
0x0640, 0x0640,
0x06D4, 0x06D4,
0x10D00, 0x10D27,
0x10D30, 0x10D39,
},
direction = "rtl",
}
m["Roro"] = {
"Rongorongo",
-- Not in Unicode
}
m["Rumin"] = process_ranges{
"Rumi numerals",
ranges = {
0x10E60, 0x10E7E,
},
ietf_subtag = "Arab",
character_category = "Rumi numerals",
}
m["Runr"] = process_ranges{
"Runic",
"alphabet",
ranges = {
0x16A0, 0x16EA,
0x16EE, 0x16F8,
},
}
m["Samr"] = process_ranges{
"Samaritan",
"abjad",
ranges = {
0x0800, 0x082D,
0x0830, 0x083E,
},
direction = "rtl",
}
m["Sarb"] = process_ranges{
"Old South Arabian",
"abjad",
ranges = {
0x10A60, 0x10A7F,
},
direction = "rtl",
}
m["Saur"] = process_ranges{
"Saurashtra",
"abugida",
ranges = {
0xA880, 0xA8C5,
0xA8CE, 0xA8D9,
},
}
m["Semap"] = {
"flag semaphore",
"pictography",
ietf_subtag = "Zsym",
}
m["Sgnw"] = process_ranges{
"SignWriting",
"pictography",
aliases = {"Sutton SignWriting"},
ranges = {
0x1D800, 0x1DA8B,
0x1DA9B, 0x1DA9F,
0x1DAA1, 0x1DAAF,
},
translit = false,
}
m["Shaw"] = process_ranges{
"Shavian",
aliases = {"Shaw"},
ranges = {
0x10450, 0x1047F,
},
}
m["Shrd"] = process_ranges{
"Sharada",
"abugida",
ranges = {
0x0951, 0x0951,
0x1CD7, 0x1CD7,
0x1CD9, 0x1CD9,
0x1CDC, 0x1CDD,
0x1CE0, 0x1CE0,
0xA830, 0xA835,
0xA838, 0xA838,
0x11180, 0x111DF,
},
}
m["Shui"] = {
"Sui",
"logography",
spaces = false,
-- Not in Unicode
}
m["Sidd"] = process_ranges{
"Siddham",
"abugida",
ranges = {
0x11580, 0x115B5,
0x115B8, 0x115DD,
},
}
m["Sidt"] = {
"Sidetic",
"alphabet",
direction = "rtl",
-- Not in Unicode
}
m["Sind"] = process_ranges{
"Khudawadi",
"abugida",
aliases = {"Khudabadi"},
ranges = {
0x0964, 0x0965,
0xA830, 0xA839,
0x112B0, 0x112EA,
0x112F0, 0x112F9,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑊰𑋠", "𑊰𑋥", "𑊰𑋦", "𑊰𑋧", "𑊰𑋨"},
to = {"𑊱", "𑊶", "𑊷", "𑊸", "𑊹"}
},
}
m["Sinh"] = process_ranges{
"Sinhalese",
"abugida",
aliases = {"Sinhala"},
ranges = {
0x0964, 0x0965,
0x0D81, 0x0D83,
0x0D85, 0x0D96,
0x0D9A, 0x0DB1,
0x0DB3, 0x0DBB,
0x0DBD, 0x0DBD,
0x0DC0, 0x0DC6,
0x0DCA, 0x0DCA,
0x0DCF, 0x0DD4,
0x0DD6, 0x0DD6,
0x0DD8, 0x0DDF,
0x0DE6, 0x0DEF,
0x0DF2, 0x0DF4,
0x1CF2, 0x1CF2,
0x111E1, 0x111F4,
},
normalizationFixes = handle_normalization_fixes{
from = {"අා", "අැ", "අෑ", "උෟ", "ඍෘ", "ඏෟ", "එ්", "එෙ", "ඔෟ", "ෘෘ"},
to = {"ආ", "ඇ", "ඈ", "ඌ", "ඎ", "ඐ", "ඒ", "ඓ", "ඖ", "ෲ"}
},
}
m["Sogd"] = process_ranges{
"Sogdian",
"abjad",
ranges = {
0x0640, 0x0640,
0x10F30, 0x10F59,
},
direction = "rtl",
}
m["Sogo"] = process_ranges{
"Old Sogdian",
"abjad",
ranges = {
0x10F00, 0x10F27,
},
direction = "rtl",
}
m["Sora"] = process_ranges{
"Sorang Sompeng",
aliases = {"Sora Sompeng"},
ranges = {
0x110D0, 0x110E8,
0x110F0, 0x110F9,
},
}
m["Soyo"] = process_ranges{
"Soyombo",
"abugida",
ranges = {
0x11A50, 0x11AA2,
},
}
m["Sund"] = process_ranges{
"Sundanese",
"abugida",
ranges = {
0x1B80, 0x1BBF,
0x1CC0, 0x1CC7,
},
}
m["Sunu"] = process_ranges{
"Sunuwar",
"alphabet",
ranges = {
0x11BC0, 0x11BE1,
0x11BF0, 0x11BF9,
},
}
m["Sylo"] = process_ranges{
"Syloti Nagri",
"abugida",
aliases = {"Sylheti Nagari"},
ranges = {
0x0964, 0x0965,
0x09E6, 0x09EF,
0xA800, 0xA82C,
},
}
m["Syrc"] = process_ranges{
"Syriac",
"abjad", -- more precisely, impure abjad
ranges = {
0x060C, 0x060C,
0x061B, 0x061C,
0x061F, 0x061F,
0x0640, 0x0640,
0x064B, 0x0655,
0x0670, 0x0670,
0x0700, 0x070D,
0x070F, 0x074A,
0x074D, 0x074F,
0x0860, 0x086A,
0x1DF8, 0x1DF8,
0x1DFA, 0x1DFA,
},
direction = "rtl",
}
-- Syre, Syrj, Syrn are apparently subsumed into Syrc; discuss if this causes issues
m["Tagb"] = process_ranges{
"Tagbanwa",
"abugida",
ranges = {
0x1735, 0x1736,
0x1760, 0x176C,
0x176E, 0x1770,
0x1772, 0x1773,
},
}
m["Takr"] = process_ranges{
"Takri",
"abugida",
ranges = {
0x0964, 0x0965,
0xA830, 0xA839,
0x11680, 0x116B9,
0x116C0, 0x116C9,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑚀𑚭", "𑚀𑚴", "𑚀𑚵", "𑚆𑚲"},
to = {"𑚁", "𑚈", "𑚉", "𑚇"}
},
}
m["Tale"] = process_ranges{
"Tai Nüa",
"abugida",
aliases = {"Tai Nuea", "New Tai Nüa", "New Tai Nuea", "Dehong Dai", "Tai Dehong", "Tai Le"},
ranges = {
0x1040, 0x1049,
0x1950, 0x196D,
0x1970, 0x1974,
},
spaces = false,
}
m["Talu"] = process_ranges{
"New Tai Lue",
"abugida",
ranges = {
0x1980, 0x19AB,
0x19B0, 0x19C9,
0x19D0, 0x19DA,
0x19DE, 0x19DF,
},
spaces = false,
}
m["Taml"] = process_ranges{
"Tamil",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0B82, 0x0B83,
0x0B85, 0x0B8A,
0x0B8E, 0x0B90,
0x0B92, 0x0B95,
0x0B99, 0x0B9A,
0x0B9C, 0x0B9C,
0x0B9E, 0x0B9F,
0x0BA3, 0x0BA4,
0x0BA8, 0x0BAA,
0x0BAE, 0x0BB9,
0x0BBE, 0x0BC2,
0x0BC6, 0x0BC8,
0x0BCA, 0x0BCD,
0x0BD0, 0x0BD0,
0x0BD7, 0x0BD7,
0x0BE6, 0x0BFA,
0x1CDA, 0x1CDA,
0xA8F3, 0xA8F3,
0x11301, 0x11301,
0x11303, 0x11303,
0x1133B, 0x1133C,
0x11FC0, 0x11FF1,
0x11FFF, 0x11FFF,
},
normalizationFixes = handle_normalization_fixes{
from = {"அூ", "ஸ்ரீ"},
to = {"ஆ", "ஶ்ரீ"}
},
}
m["Tang"] = process_ranges{
"Tangut",
"logography, syllabary",
ranges = {
0x31EF, 0x31EF,
0x16FE0, 0x16FE0,
0x17000, 0x187F7,
0x18800, 0x18AFF,
0x18D00, 0x18D08,
},
spaces = false,
}
m["Tavt"] = process_ranges{
"Tai Viet",
"abugida",
ranges = {
0xAA80, 0xAAC2,
0xAADB, 0xAADF,
},
spaces = false,
}
m["Tayo"] = {
"Lai Tay",
"abugida",
aliases = {"Tai Yo"},
direction = "vertical-rtl",
-- Not in Unicode
}
m["Telu"] = process_ranges{
"Telugu",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x0C00, 0x0C0C,
0x0C0E, 0x0C10,
0x0C12, 0x0C28,
0x0C2A, 0x0C39,
0x0C3C, 0x0C44,
0x0C46, 0x0C48,
0x0C4A, 0x0C4D,
0x0C55, 0x0C56,
0x0C58, 0x0C5A,
0x0C5D, 0x0C5D,
0x0C60, 0x0C63,
0x0C66, 0x0C6F,
0x0C77, 0x0C7F,
0x1CDA, 0x1CDA,
0x1CF2, 0x1CF2,
},
normalizationFixes = handle_normalization_fixes{
from = {"ఒౌ", "ఒౕ", "ిౕ", "ెౕ", "ొౕ"},
to = {"ఔ", "ఓ", "ీ", "ే", "ో"}
},
}
m["Teng"] = {
"Tengwar",
}
m["Tfng"] = process_ranges{
"Tifinagh",
"abjad, alphabet",
ranges = {
0x2D30, 0x2D67,
0x2D6F, 0x2D70,
0x2D7F, 0x2D7F,
},
otherNames = {"Libyco-Berber", "Berber"}, -- per Wikipedia, Libyco-Berber is the parent
}
m["Tglg"] = process_ranges{
"Baybayin",
"abugida",
aliases = {"Tagalog"},
ranges = {
0x1700, 0x1715,
0x171F, 0x171F,
0x1735, 0x1736,
},
}
m["Thaa"] = process_ranges{
"Thaana",
"abugida",
ranges = {
0x060C, 0x060C,
0x061B, 0x061C,
0x061F, 0x061F,
0x0660, 0x0669,
0x0780, 0x07B1,
0xFDF2, 0xFDF2,
0xFDFD, 0xFDFD,
},
direction = "rtl",
}
m["Thai"] = process_ranges{
"Thai",
"abugida",
ranges = {
0x0E01, 0x0E3A,
0x0E40, 0x0E5B,
},
spaces = false,
}
m["Tibt"] = process_ranges{
"Tibetan",
"abugida",
ranges = {
0x0F00, 0x0F47,
0x0F49, 0x0F6C,
0x0F71, 0x0F97,
0x0F99, 0x0FBC,
0x0FBE, 0x0FCC,
0x0FCE, 0x0FD4,
0x0FD9, 0x0FDA,
0x3008, 0x300B,
},
normalizationFixes = handle_normalization_fixes{
combiningClasses = {["༹"] = 1},
from = {"ཷ", "ཹ"},
to = {"ྲཱྀ", "ླཱྀ"}
},
}
m["sit-tam-Tibt"] = {
"Tamyig",
m["Tibt"][2],
ranges = m["Tibt"].ranges,
characters = m["Tibt"].characters,
parent = "Tibt",
normalizationFixes = m["Tibt"].normalizationFixes,
}
m["Tirh"] = process_ranges{
"Tirhuta",
"abugida",
ranges = {
0x0951, 0x0952,
0x0964, 0x0965,
0x1CF2, 0x1CF2,
0xA830, 0xA839,
0x11480, 0x114C7,
0x114D0, 0x114D9,
},
normalizationFixes = handle_normalization_fixes{
from = {"𑒁𑒰", "𑒋𑒺", "𑒍𑒺", "𑒪𑒵", "𑒪𑒶"},
to = {"𑒂", "𑒌", "𑒎", "𑒉", "𑒊"}
},
}
m["Tnsa"] = process_ranges{
"Tangsa",
"alphabet",
ranges = {
0x16A70, 0x16ABE,
0x16AC0, 0x16AC9,
},
}
m["Todr"] = process_ranges{
"Todhri",
"alphabet",
direction = "rtl",
ranges = {
0x105C0, 0x105F3,
},
}
m["Tols"] = {
"Tolong Siki",
"alphabet",
-- Not in Unicode
}
m["Toto"] = process_ranges{
"Toto",
"abugida",
ranges = {
0x1E290, 0x1E2AE,
},
}
m["Tutg"] = process_ranges{
"Tigalari",
"abugida",
aliases = {"Tulu"},
ranges = {
0x1CF2, 0x1CF2,
0x1CF4, 0x1CF4,
0xA8F1, 0xA8F1,
0x11380, 0x11389,
0x1138B, 0x1138B,
0x1138E, 0x1138E,
0x11390, 0x113B5,
0x113B7, 0x113C0,
0x113C2, 0x113C2,
0x113C5, 0x113C5,
0x113C7, 0x113CA,
0x113CC, 0x113D5,
0x113D7, 0x113D8,
0x113E1, 0x113E2,
},
}
m["Ugar"] = process_ranges{
"Ugaritic",
"abjad",
ranges = {
0x10380, 0x1039D,
0x1039F, 0x1039F,
},
}
m["Vaii"] = process_ranges{
"Vai",
"syllabary",
ranges = {
0xA500, 0xA62B,
},
}
m["Visp"] = {
"Visible Speech",
"alphabet",
-- Not in Unicode
}
m["Vith"] = process_ranges{
"Vithkuqi",
"alphabet",
ranges = {
0x10570, 0x1057A,
0x1057C, 0x1058A,
0x1058C, 0x10592,
0x10594, 0x10595,
0x10597, 0x105A1,
0x105A3, 0x105B1,
0x105B3, 0x105B9,
0x105BB, 0x105BC,
},
capitalized = true,
}
m["Wara"] = process_ranges{
"Varang Kshiti",
aliases = {"Warang Citi"},
ranges = {
0x118A0, 0x118F2,
0x118FF, 0x118FF,
},
capitalized = true,
}
m["Wcho"] = process_ranges{
"Wancho",
"alphabet",
ranges = {
0x1E2C0, 0x1E2F9,
0x1E2FF, 0x1E2FF,
},
}
m["Wole"] = {
"Woleai",
"syllabary",
-- Not in Unicode
}
m["Xpeo"] = process_ranges{
"Old Persian",
ranges = {
0x103A0, 0x103C3,
0x103C8, 0x103D5,
},
}
m["Xsux"] = process_ranges{
"Cuneiform",
aliases = {"Sumero-Akkadian Cuneiform"},
ranges = {
0x12000, 0x12399,
0x12400, 0x1246E,
0x12470, 0x12474,
0x12480, 0x12543,
},
}
m["Yezi"] = process_ranges{
"Yezidi",
"alphabet",
ranges = {
0x060C, 0x060C,
0x061B, 0x061B,
0x061F, 0x061F,
0x0660, 0x0669,
0x10E80, 0x10EA9,
0x10EAB, 0x10EAD,
0x10EB0, 0x10EB1,
},
direction = "rtl",
}
m["Yiii"] = process_ranges{
"Yi",
"syllabary",
ranges = {
0x3001, 0x3002,
0x3008, 0x3011,
0x3014, 0x301B,
0x30FB, 0x30FB,
0xA000, 0xA48C,
0xA490, 0xA4C6,
0xFF61, 0xFF65,
},
}
m["Zanb"] = process_ranges{
"Zanabazar Square",
ranges = {
0x11A00, 0x11A47,
},
}
m["Zmth"] = process_ranges{
"mathematical notation",
ranges = {
0x00AC, 0x00AC,
0x00B1, 0x00B1,
0x00D7, 0x00D7,
0x00F7, 0x00F7,
0x03D0, 0x03D2,
0x03D5, 0x03D5,
0x03F0, 0x03F1,
0x03F4, 0x03F6,
0x0606, 0x0608,
0x2016, 0x2016,
0x2032, 0x2034,
0x2040, 0x2040,
0x2044, 0x2044,
0x2052, 0x2052,
0x205F, 0x205F,
0x2061, 0x2064,
0x207A, 0x207E,
0x208A, 0x208E,
0x20D0, 0x20DC,
0x20E1, 0x20E1,
0x20E5, 0x20E6,
0x20EB, 0x20EF,
0x2102, 0x2102,
0x2107, 0x2107,
0x210A, 0x2113,
0x2115, 0x2115,
0x2118, 0x211D,
0x2124, 0x2124,
0x2128, 0x2129,
0x212C, 0x212D,
0x212F, 0x2131,
0x2133, 0x2138,
0x213C, 0x2149,
0x214B, 0x214B,
0x2190, 0x21A7,
0x21A9, 0x21AE,
0x21B0, 0x21B1,
0x21B6, 0x21B7,
0x21BC, 0x21DB,
0x21DD, 0x21DD,
0x21E4, 0x21E5,
0x21F4, 0x22FF,
0x2308, 0x230B,
0x2320, 0x2321,
0x237C, 0x237C,
0x239B, 0x23B5,
0x23B7, 0x23B7,
0x23D0, 0x23D0,
0x23DC, 0x23E2,
0x25A0, 0x25A1,
0x25AE, 0x25B7,
0x25BC, 0x25C1,
0x25C6, 0x25C7,
0x25CA, 0x25CB,
0x25CF, 0x25D3,
0x25E2, 0x25E2,
0x25E4, 0x25E4,
0x25E7, 0x25EC,
0x25F8, 0x25FF,
0x2605, 0x2606,
0x2640, 0x2640,
0x2642, 0x2642,
0x2660, 0x2663,
0x266D, 0x266F,
0x27C0, 0x27FF,
0x2900, 0x2AFF,
0x2B30, 0x2B44,
0x2B47, 0x2B4C,
0xFB29, 0xFB29,
0xFE61, 0xFE66,
0xFE68, 0xFE68,
0xFF0B, 0xFF0B,
0xFF1C, 0xFF1E,
0xFF3C, 0xFF3C,
0xFF3E, 0xFF3E,
0xFF5C, 0xFF5C,
0xFF5E, 0xFF5E,
0xFFE2, 0xFFE2,
0xFFE9, 0xFFEC,
0x1D400, 0x1D454,
0x1D456, 0x1D49C,
0x1D49E, 0x1D49F,
0x1D4A2, 0x1D4A2,
0x1D4A5, 0x1D4A6,
0x1D4A9, 0x1D4AC,
0x1D4AE, 0x1D4B9,
0x1D4BB, 0x1D4BB,
0x1D4BD, 0x1D4C3,
0x1D4C5, 0x1D505,
0x1D507, 0x1D50A,
0x1D50D, 0x1D514,
0x1D516, 0x1D51C,
0x1D51E, 0x1D539,
0x1D53B, 0x1D53E,
0x1D540, 0x1D544,
0x1D546, 0x1D546,
0x1D54A, 0x1D550,
0x1D552, 0x1D6A5,
0x1D6A8, 0x1D7CB,
0x1D7CE, 0x1D7FF,
0x1EE00, 0x1EE03,
0x1EE05, 0x1EE1F,
0x1EE21, 0x1EE22,
0x1EE24, 0x1EE24,
0x1EE27, 0x1EE27,
0x1EE29, 0x1EE32,
0x1EE34, 0x1EE37,
0x1EE39, 0x1EE39,
0x1EE3B, 0x1EE3B,
0x1EE42, 0x1EE42,
0x1EE47, 0x1EE47,
0x1EE49, 0x1EE49,
0x1EE4B, 0x1EE4B,
0x1EE4D, 0x1EE4F,
0x1EE51, 0x1EE52,
0x1EE54, 0x1EE54,
0x1EE57, 0x1EE57,
0x1EE59, 0x1EE59,
0x1EE5B, 0x1EE5B,
0x1EE5D, 0x1EE5D,
0x1EE5F, 0x1EE5F,
0x1EE61, 0x1EE62,
0x1EE64, 0x1EE64,
0x1EE67, 0x1EE6A,
0x1EE6C, 0x1EE72,
0x1EE74, 0x1EE77,
0x1EE79, 0x1EE7C,
0x1EE7E, 0x1EE7E,
0x1EE80, 0x1EE89,
0x1EE8B, 0x1EE9B,
0x1EEA1, 0x1EEA3,
0x1EEA5, 0x1EEA9,
0x1EEAB, 0x1EEBB,
0x1EEF0, 0x1EEF1,
},
translit = false,
character_category = "Mathematical notation symbols", -- ?
}
m["Zsym"] = process_ranges{
"symbol",
"pictography",
ranges = {
0x20DD, 0x20E0,
0x20E2, 0x20E4,
0x20E7, 0x20EA,
0x20F0, 0x20F0,
0x2100, 0x2101,
0x2103, 0x2106,
0x2108, 0x2109,
0x2114, 0x2114,
0x2116, 0x2117,
0x211E, 0x2123,
0x2125, 0x2127,
0x212A, 0x212B,
0x212E, 0x212E,
0x2132, 0x2132,
0x2139, 0x213B,
0x214A, 0x214A,
0x214C, 0x214F,
0x21A8, 0x21A8,
0x21AF, 0x21AF,
0x21B2, 0x21B5,
0x21B8, 0x21BB,
0x21DC, 0x21DC,
0x21DE, 0x21E3,
0x21E6, 0x21F3,
0x2300, 0x2307,
0x230C, 0x231F,
0x2322, 0x237B,
0x237D, 0x239A,
0x23B6, 0x23B6,
0x23B8, 0x23CF,
0x23D1, 0x23DB,
0x23E3, 0x23FF,
0x2500, 0x259F,
0x25A2, 0x25AD,
0x25B8, 0x25BB,
0x25C2, 0x25C5,
0x25C8, 0x25C9,
0x25CC, 0x25CE,
0x25D4, 0x25E1,
0x25E3, 0x25E3,
0x25E5, 0x25E6,
0x25ED, 0x25F7,
0x2600, 0x2604,
0x2607, 0x263F,
0x2641, 0x2641,
0x2643, 0x265F,
0x2664, 0x266C,
0x2670, 0x27BF,
0x2B00, 0x2B2F,
0x2B45, 0x2B46,
0x2B4D, 0x2B73,
0x2B76, 0x2B95,
0x2B97, 0x2BFF,
0x4DC0, 0x4DFF,
0x1F000, 0x1F02B,
0x1F030, 0x1F093,
0x1F0A0, 0x1F0AE,
0x1F0B1, 0x1F0BF,
0x1F0C1, 0x1F0CF,
0x1F0D1, 0x1F0F5,
0x1F300, 0x1F6D7,
0x1F6DC, 0x1F6EC,
0x1F6F0, 0x1F6FC,
0x1F700, 0x1F776,
0x1F77B, 0x1F7D9,
0x1F7E0, 0x1F7EB,
0x1F7F0, 0x1F7F0,
0x1F800, 0x1F80B,
0x1F810, 0x1F847,
0x1F850, 0x1F859,
0x1F860, 0x1F887,
0x1F890, 0x1F8AD,
0x1F8B0, 0x1F8B1,
0x1F900, 0x1FA53,
0x1FA60, 0x1FA6D,
0x1FA70, 0x1FA7C,
0x1FA80, 0x1FA88,
0x1FA90, 0x1FABD,
0x1FABF, 0x1FAC5,
0x1FACE, 0x1FADB,
0x1FAE0, 0x1FAE8,
0x1FAF0, 0x1FAF8,
0x1FB00, 0x1FB92,
0x1FB94, 0x1FBCA,
0x1FBF0, 0x1FBF9,
},
translit = false,
character_category = false, -- none
}
m["Zyyy"] = {
"undetermined",
-- This should not have any characters listed, probably
translit = false,
character_category = false, -- none
}
m["Zzzz"] = {
"uncoded",
-- This should not have any characters listed
translit = false,
character_category = false, -- none
}
-- These should be defined after the scripts they are composed of.
m["Hrkt"] = process_ranges{
"Kana",
"syllabary",
aliases = {"Japanese syllabaries"},
ranges = union(
m["Hira"].ranges,
m["Kana"].ranges
),
spaces = false,
}
m["Jpan"] = process_ranges{
"Japanese",
"logography, syllabary",
ranges = union(
m["Hrkt"].ranges,
m["Hani"].ranges,
m["Latn"].ranges
),
spaces = false,
sort_by_scraping = true,
}
m["Kore"] = process_ranges{
"Korean",
"logography, syllabary",
ranges = union(
m["Hang"].ranges,
m["Hani"].ranges,
m["Latn"].ranges
),
}
return m