Module:languages/datax

From Linguifex
< Module:languages
Revision as of 17:25, 8 August 2021 by Sware (talk | contribs)
Jump to navigation Jump to search


local u = mw.ustring.char

-- UTF-8 encoded strings for some commonly-used diacritics
local GRAVE		= u(0x0300)
local ACUTE		= u(0x0301)
local CIRC		= u(0x0302)
local TILDE		= u(0x0303)
local MACRON	= u(0x0304)
local BREVE		= u(0x0306)
local DOTABOVE	= u(0x0307)
local DIAER		= u(0x0308)
local CARON		= u(0x030C)
local DGRAVE	= u(0x030F)
local INVBREVE	= u(0x0311)
local DOTBELOW	= u(0x0323)
local RINGBELOW	= u(0x0325)
local CEDILLA	= u(0x0327)
local TILDEBELOW = u(0x0330)
local Latn = {"Latn"}

local m = {}

m["alo-pro"] = {
	"Proto-Alopian",
	nil,
	"art",
	type = "appendix-constructed",
	scripts = {"Latinx"},
	sort_key = {
		from = {"[-*·]"},
		to =   {""},
		remove_diacritics = MACRON .. BREVE, 
	},
}m["art-blk"] = {
	"Bolak",
	"Q2909283",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-bsp"] = {
	"Black Speech",
	"Q686210",
	"art",
	type = "appendix-constructed",
	scripts = {"Latn", "Teng"},
}

m["art-com"] = {
	"Communicationssprache",
	"Q35227",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-dtk"] = {
	"Dothraki",
	"Q2914733",
	"art",
	scripts = Latn,
}

m["art-elo"] = {
	"Eloi",
	nil,
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-gld"] = {
	"Goa'uld",
	"Q19823",
	"art",
	type = "appendix-constructed",
	scripts = {"Latn", "Egyp", "Mero"},
}

m["art-lap"] = {
	"Lapine",
	"Q6488195",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-man"] = {
	"Mandalorian",
	"Q54289",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-mun"] = {
	"Mundolinco",
	"Q851355",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-nav"] = {
	"Na'vi",
	"Q316939",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-nox"] = {
	"Noxilo",
	nil,
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

local F = "[ɬsṣš]"

m["art-nui"] = {
	"ǂA Ṇṵĩ",
	nil,
	"art",
	scripts = {"Latinx"},
	sort_key = {
		from = {	
			"^N", "^Ɠ", "ṭr",
			F .. "([ʇǃǂǁ])qʼ",
			F .. "([ʇǃǂǁ])ʼ",
			F .. "([ʇǃǂǁ])",
			"ɴ([ʇǃǂǁ])x",
			"ɴ([ʇǃǂǁ])ʼ",
			"ɴ([ʇǃǂǁʘ])",
			"r([ǃǁ])", "(ʘ)r",
			"([ǃǁ])ʛ",
			"([ʇǃǂǁ])qʼ",
			"([ʇǃǂǁ])x",
			"([ʇǃǂǁ])ʼ",
			"^[aeiou]?(.)",
			},
		to = {
			"ɴ", "ʛ", "ṭᵣ",
			"%1₊",
			"%1₉",
			"%1₈",
			"%1₇",
			"%1₆",
			"%1₅",
			"%1₄", "%1₄",
			"%1₃",
			"%1₂",
			"%1₁",
			"%1₀",
			"%1",
			},
		remove_diacritics = TILDEBELOW .. GRAVE
	},
	aliases = {"ǂA"},
}

m["art-top"] = {
	"Toki Pona",
	"Q36846",
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["art-una"] = {
	"Unas",
	nil,
	"art",
	type = "appendix-constructed",
	scripts = Latn,
}

m["pls-pro"] = {
	"Proto-Plains",
	nil,
	"art",
	type = "appendix-constructed",
	scripts = {"Latinx"},
	sort_key = {
		from = {"*"},
		to =   {""},
	},
}

m["siwa"] = {
	"Siwa",
	nil,
	"art",
	aliases = {"Siųa"},
	scripts = {"Latinx"},
	ancestors = {"alo-pro"},
	entry_name = {
		from = {"‧"},
		to = {""},
	},
	sort_key = {
		from =  {"ạ", "ẹ", "ị", "ọ", "ụ", "ỵ",},
		to = {"a", "e", "i", "o", "u", "y",},
		remove_diacritics = u(0x0309), -- HOOK
	}
}

return m