Module:siwa-noun

syll_count

function syll_count(word)

export.show

function export.show(frame)

---

local export = {}

local m_u = require('Module:utilities')
local m_data = require('Module:siwa-noun/data')

local sub = mw.ustring.sub
local find = mw.ustring.find
local match = mw.ustring.match
local gsub = mw.ustring.gsub
local u = mw.ustring.char
local split = mw.text.split
local gsplit = mw.text.gsplit

local pagename = mw.title.getCurrentTitle().text
local vowels = "[aeiouyůõảẻỉỏủỷę̊]"
local consonants = "[mpbvntdsṡʦʨʥŋɲcɟħðrṁṅḥkgġhłƛɬḍ]"

local stressed_vowels = {
	["ả"] = "a", ["a[ui]"] = "a", ["oa"] = "a", ["[eẻę]"] = "e", ["e[iu]"] = "e",
	["ay"] = "e", ["[iỉ]"] = "i", ["i[aeou]"] = "i", ["[oỏõ]"] = "o", ["[oõ]u"] = "o",
	["oi"] = "o", ["[uủ]"] = "u", ["u[oi]"] = "u", ["[yỷ]"] = "y", ["ů[ai]?"] = "y", ["ẻu"] = "y", ["ey"] = "y",
}

local digraphs_to_single = {
	["ts"] = "ʦ", ["tṡ"] = "ʨ", ["dį"] = "ʥ", ["ng"] = "ŋ", ["nį"] = "ɲ", ["kį"] = "c", ["gį"] = "ɟ", ["hh"] = "ħ", ["ḍḍ"] = "ð", ["dl"] = "ɬ",
}

local lenition_patterns = {
	["bb"] = "b", ["dd"] = "d", ["gg"] = "g", ["ɟ"] = "į", ["mm"] = "m", ["ll"] = "l", ["nn"] = "n",
	["rr"] = "r", ["bġ"] = "p", ["pr"] = "p", ["dġ"] = "t", ["tr"] = "t", ["ḍb"] = "p", ["ð"] = "hh",
	["ḍg"] = "k", ["bm"] = "m", ["dn"] = "n", ["kn"] = "ng", ["([lr])pp"] = "%1p", ["([lr])tt"] = "%1t", ["([lrms])kk"] = "%1k",
	["k([lvs])"] = "g%1", ["ps"] = "bs", ["[vųbhḥg]"] = "", ["d[aou]"] = "l", ["ġ[aou]"] = "vv", ["[dġ][eůy]"] = "", ["[rġ]i"] = "ṡi",
	["di"] = "", ["ɲi"] = "ɟi", ["ħį"] = "ṡ", ["[ou]ų"] = "ů", ["ɬ"] = "l",
}

local function dedigraphize(word)
	for digraph, repl in pairs(digraphs_to_single) do
		word = gsub(word, digraph, repl)
	end
	return word
end	

function syll_count(word)
	local pattern = "(" .. consonants .. "?" .. vowels .. "+ː?" .. consonants .. "*)"
	local syllable, n = gsub(word, pattern, "%1")
	syllable = match(syllable, pattern)
	
	return syllable, n
end

local function stressed_components(word)
	local pattern = consonants .. "?(" .. vowels .. "+ː?)(" .. consonants .. "*)"
	local pattern2 = consonants .. "?" .. vowels .. "+ː?" .. consonants .. "*(.)"
	local v, c = match(word, pattern)
	if c == "d" or c == "ġ" or c == "r" or c == "ɲ" or c == "ħ" then
		c = c .. match(word, pattern2)
	end
	return v, c
end	

local function detect_quality(word)
	local stressed, n = syll_count(word)
	if match(stressed, vowels .. vowels .. vowels .. "?") or match(stressed, "ː") or n>=3 then
		return "w" -- weak nouns
	elseif (match(stressed, vowels .. vowels .. vowels .. "?") or match(stressed, "ː")) and n<3 then
		return "l" -- long nouns 
	else return "s" -- strong nouns
	end 
end

local function detect_decl(word, gender, quality)
	local stressed = syll_count(word)
	local tonic_vowel = stressed_components(word)
	for vowel, d in pairs(stressed_vowels) do
		tonic_vowel = gsub(tonic_vowel, vowel, d)	
	end	
	if gender and quality then
		if find(word, vowels .. "$") then
			local decl = gender .. "-" .. tonic_vowel .. "-" .. quality
			return decl
		else
			return gender .. "-" .. sub(word, -1) .. "-" .. tonic_vowel
		end	
	end	
end

local function lenition(word)
	local lenited = ""
	local _, c = stressed_components(word)
	for regex, repl in pairs(lenition_patterns) do
		lenited = gsub(c, regex, repl, 1)
	end
	local i, j = find(word, c)
	return sub(word, 1, i-1) .. lenited .. sub(word, j+1)
end

-- The main entry point.
-- This is the only function that can be invoked from a template.
function export.show(frame)
	local parent_args = frame:getParent().args
	local args = {}
	local args = require("Module:parameters").process(parent_args, m_data.params, true)
	local decl_type = {}
	local gender, word = parent_args[1], parent_args[2]~=nil and dedigraphize(parent_args[2]) or dedigraphize(pagename)
	local quality = parent_args[3]~=nil and args[3] or detect_quality(word)
	
	local prefix = sub(pagename, 1, -(#word+1))
	
	return stressed_components(word) .. " " .. lenition(word)
end

return export