Module:grc-utilities/data

---

This module lacks a documentation subpage. Please create it.

Useful links: root page • root page’s subpages • links • transclusions • testcases • sandbox

local data = {}

local concat = table.concat
local insert = table.insert

local U = require("Module:string/char")
local macron = U(0x304)
local spacing_macron = U(0xAF)
local modifier_macron = U(0x2C9)
local breve = U(0x306)
local spacing_breve = U(0x2D8)
local rough = U(0x314)
local smooth = U(0x313)
local diaeresis = U(0x308)
local acute = U(0x301)
local grave = U(0x300)
local circum = U(0x342)
local Latin_circum = U(0x302)
local coronis = U(0x343)
local subscript = U(0x345)
local undertie = U(0x35C) -- actually "combining double breve below"

data["diacritics"] = {
	["macron"] = macron,
	["spacing_macron"] = spacing_macron,
	["modifier_macron"] = modifier_macron,
	["breve"] = breve,
	["spacing_breve"] = spacing_breve,
	["rough"] = rough,
	["smooth"] = smooth,
	["diaeresis"] = diaeresis,
	["acute"] = acute,
	["grave"] = grave,
	["circum"] = circum,
	["Latin_circum"] = Latin_circum,
	["coronis"] = coronis,
	["subscript"] = subscript,
}

local diacritics_all, diacritics_combining = {}, {}
for _, diacritic in pairs(data.diacritics) do
	insert(diacritics_all, diacritic)
	if not (diacritic == spacing_macron or diacritic == modifier_macron or diacritic == spacing_breve) then
		insert(diacritics_combining, diacritic)
	end
end
diacritics_all = concat(diacritics_all)
data.diacritics.all = diacritics_all
diacritics_combining = concat(diacritics_combining)
data.diacritics.combining = diacritics_combining

data["named"] = data["diacritics"]

data["diacritic"] = "[" .. diacritics_all .. "]"
data["combining_diacritic"] = "[" .. diacritics_combining .. "]"
data["all"] = data["diacritic"]

data["diacritic_groups"] = {
	[1] = "[".. macron .. breve .."]",
	[2] = "[".. diaeresis .. smooth .. rough .."]",
	[3] = "[".. acute .. grave .. circum .. "]",
	[4] = subscript,
}
data["groups"] = data["diacritic_groups"]
data["diacritic_groups"]["accents"] = data["groups"][3]

data["length"] = {
	optional = macron .. "?" .. breve .. "?"
}
data["length"]["mandatory"] = "%f[" .. macron .. breve .. "]" .. data["length"]["optional"]

data["diacritic_order"] = {
	[macron] = 1,
	[breve] = 2,
	[rough] = 3,
	[smooth] = 3,
	[diaeresis] = 3,
	[acute] = 4,
	[grave] = 4,
	[circum] = 4,
	[subscript] = 5,
}

data["diacritical_conversions"] = {
	-- Convert spacing to combining diacritics
	[spacing_macron] = macron, -- macron
	[modifier_macron] = macron,
	[spacing_breve] = breve, -- breve
	["῾"] = rough, -- rough breathing, modifier letter reversed comma
	["ʽ"] = rough,
	["᾿"] = smooth, -- smooth breathing, modifier letter apostrophe, coronis, combining coronis
	["ʼ"] = smooth,
	["´"] = acute, -- acute
	["`"] = grave, -- grave
	["῀"] = circum, -- Greek circumflex (perispomeni), circumflex, combining circumflex
	["ˆ"] = circum,
	[Latin_circum] = circum,
	["¨"] = diaeresis,
}

-- Canonical letter forms
data["canonical"] = {
	["ϴ"] = "Θ",
	["Ϗ"] = "Καί",
	["Ϗ̀"] = "Καὶ",
	["Ϟ"] = "Ϙ",
	["Ϲ"] = "Σ",
	["ϒ"] = "Υ",
	["ϓ"] = "Ύ",
	["ϔ"] = "Ϋ",
	["Ϡ"] = "Ͳ",

	["ϐ"] = "β",
	["ϵ"] = "ε",
	["ϑ"] = "θ",
	["ϰ"] = "κ",
	["ϗ"] = "καί",
	["ϗ̀"] = "καὶ",
	["ϖ"] = "π",
	["ϟ"] = "ϙ",
	["ϱ"] = "ρ",
	["ς"] = "σ",
	["ϲ"] = "σ",
	["ϕ"] = "φ",
	["ϡ"] = "ͳ",
}

data["consonants"] = "ΒβΓγΔδϜϝͶͷϚϛΖζͰͱΘθͿϳΚκΛλΜμΝνΞξΠπϺϻϘϙϞϟΡρΣσςϹϲΤτΦφΧχΨψͲͳϠϡϷϸ"
data["consonant"] = "[" .. data.consonants .. "]"
data["vowels"] = "ΑαΕεΗηΙιΟοΥυΩω"
data["vowel"] = "[" .. data.vowels .. "]"

-- Basic letters with and without diacritics
data.word_characters = "%*'’" .. require("Module:scripts").getByCode("Polyt"):getCharacters() .. diacritics_combining .. undertie
data.word_character = "[" .. data.word_characters .. "]"

return data