Module:ka-hyphen

Documentation for this module may be created at Module:ka-hyphen/doc

local export = {}

local gsub = mw.ustring.gsub
local gsplit = mw.text.gsplit
local format_hyphens = require('Module:hyphenation').format_hyphenations
local lang = require('Module:languages').getByCode('ka')
local scr = require('Module:scripts').getByCode('Geor')

local V = "აეიოუ"
local C = "ბგდვზთკლმნპჟრსტფქღყშჩცძწჭხჯჰ" .. "ABCDEFGHIJKL" -- Latin letters represent harmonic clusters

local harmonic = {
	["ფხ"] = "A",
	["თხ"] = "B",
	["ცხ"] = "C",
	["ჩხ"] = "D",
	["ბღ"] = "E",
	["დღ"] = "F",
	["ზღ"] = "G",
	["ჯღ"] = "H",
	["პყ"] = "I",
	["ტყ"] = "J",
	["წყ"] = "K",
	["ჭყ"] = "L"
}
local harmonic_inv = {}
for k, v in pairs(harmonic) do
	harmonic_inv[v] = k
end

function export.main(frame)
	local parent_args = frame:getParent().args
	local term = parent_args[1] or mw.title.getCurrentTitle().text  -- unnamed parameter or PAGENAME

	local syllables = {}

	-- check if there's numbered parameters (for custom syllables)
	for k, v in pairs(parent_args) do
		local num = tonumber(k)
		if num and v ~= '' then
			syllables[num] = v
		end
	end

	-- otherwise automatically process and syllabify term
	if next(syllables) == nil then
		local processed_term = export.hyphenateTerm(term)
		for syllable in gsplit(processed_term, "‧") do
			table.insert(syllables, syllable)
		end
	else
		-- convert numbered table to sequence
		local temp = {}
		for i = 1, #syllables do
			if syllables[i] then
				table.insert(temp, syllables[i])
			end
		end
		syllables = temp
	end

	-- data structure for displaying hyphenation
	local data = {
		lang = lang,
		hyphs = {
			{
				hyph = syllables,
				sc = scr
			}
		},
		sc = scr,
		nocaption = parent_args.nocaption,
		caption = parent_args.caption
	}

	return format_hyphens(data)
end

function export.hyphenateTerm(term)
	-- convert harmonic clusters
	for k, v in pairs(harmonic) do
		term = gsub(term, k, v)
	end

	-- only process actual Georgian text segments
	local hyphenated = gsub(term, "([" .. V .. C .. "]+)", hyphenateWord)

	-- restore harmonic clusters
	for k, v in pairs(harmonic_inv) do
		hyphenated = gsub(hyphenated, k, v)
	end

	return hyphenated
end

function hyphenateWord(word)
	-- თ‧ა‧ვქ‧უ‧დმ‧ო‧გლ‧ე‧ჯ‧ი‧ლ‧ი‧
	word = gsub(word, "([" .. V .. "])", "‧%1‧")

	-- remove leading and trailing hyphens
	word = gsub(word, "^‧+", "")
	word = gsub(word, "‧+$", "")
	word = gsub(word, "‧‧+", "‧")

	-- shift hyphens to the right of consonants
	word = gsub(word, "([" .. V .. "])‧([" .. C .. "])([" .. C .. "]+)‧([" .. V .. "])", "%1%2‧%3%4")  -- თ‧ა‧ვქ‧უ‧დმ‧ო‧გლ‧ე‧ჯ‧ი‧ლ‧ი
	word = gsub(word, "([" .. V .. "])‧([" .. C .. "])([" .. C .. "]+)‧([" .. V .. "])", "%1%2‧%3%4")  -- თ‧ავ‧ქუ‧დმ‧ოგ‧ლე‧ჯ‧ი‧ლ‧ი

	-- remove hyphens before and after consonants
	word = gsub(word, "([" .. C .. "]+)‧([" .. V .. "])", "%1%2")  -- თ‧ავ‧ქუდ‧მოგ‧ლე‧ჯ‧ი‧ლ‧ი
	word = gsub(word, "([" .. V .. "])‧([" .. C .. "]*)$", "%1%2")  -- თავ‧ქუდ‧მოგ‧ლე‧ჯი‧ლი

	-- remove hyphens before and after vowels
	word = gsub(word, "^([" .. V .. "])‧", "%1")  -- ე‧ნა -> ენა
	word = gsub(word, "‧([" .. V .. "])$", "%1")  -- აუ‧ღი‧ა -> აუ‧ღია

	return word
end

return export