Module:ja-pron

local m_str_utils = require("Module:string utilities")

local concat = table.concat
local gsplit = m_str_utils.gsplit
local gsub = m_str_utils.gsub
local insert = table.insert
local len = m_str_utils.len
local match = m_str_utils.match
local min = math.min
local split = m_str_utils.split
local sub = m_str_utils.sub
local toNFC = mw.ustring.toNFC

local lang = require("Module:languages").getByCode("ja")
local kana_to_romaji = require("Module:Hrkt-translit").tr
local m_accent = require("Module:accent qualifier")
-- also [[Module:qualifier]]
local PAGENAME = mw.loadData("Module:headword/data").pagename

local range = mw.loadData("Module:ja/data/range")
local a_kana = range.vowels.a
local i_kana = range.vowels.i
local u_kana = range.vowels.u
local e_kana = range.vowels.e
local o_kana = range.vowels.o
local n_kana = range.vowels.n
local submoraic_kana = range.submoraic_kana

local export = {}

local ref_template_name_data = {
	["DJR"] = "R:Daijirin",
	["DJR4"] = "R:Daijirin4",
	["DJS"] = "R:Daijisen",
	["KDJ"] = "R:Kokugo Dai Jiten",
	["NHK"] = "R:NHK Hatsuon",
	["NHK16"] = "R:NHK2016",
	["NKD2"] = "R:Nihon Kokugo Daijiten 2 Online",
	["SMK2"] = "R:Shinmeikai2",
	["SMK5"] = "R:Shinmeikai5",
	["SMK7"] = "R:Shinmeikai7",
	["SMK8"] = "R:Shinmeikai8",
	["SKK8"] = "R:Sankoku8",
	["ZAJ"] = "R:Zenkoku Akusento Jiten",
	["JEL"] = "R:Kenkyusha JEL Pocket",
	["JAC"] = "R:ja:JAccent",
}

local function add_acc_refs(frame, text)
	local output = {}
	for ref_name in gsplit(text, ",") do
		mw.log(ref_name)
		local ref_template_name = ref_template_name_data[ref_name]
		if ref_template_name then
			insert(output, frame:extensionTag("ref", "{{" .. ref_template_name .. "}}", {name = ref_name}))
		elseif match(ref_name, "ref") then
			insert(output, frame:preprocess(ref_name))
		else
			-- [[Special:WhatLinksHere/Wiktionary:Tracking/ja-pron/unrecognized ref]]
			require("Module:debug").track("ja-pron/unrecognized ref")
		end
	end
	return concat(output)
end

function export.show(frame)
	local params = {
		[1] = {default = PAGENAME, list = true},
		
		["accent"] = {list = true},
		["accent\1_loc"] = {list = true, allow_holes = true},
		["accent\1_ref"] = {list = true, allow_holes = true},
		["accent\1_note"] = {list = true, allow_holes = true},
		
		["acc"] = {alias_of = "accent", list = true},
		["acc\1_loc"] = {alias_of = "accent\1_loc", list = true},
		["acc\1_ref"] = {alias_of = "accent\1_ref", list = true},
		["acc\1_note"] = {alias_of = "accent\1_note", list = true},
		
		["dev"] = {},
		["dev2"] = {},
		["devm"] = {},
		
		["a"] = {alias_of = "audio"},
		["audio"] = {}
	}
	
	local args = require("Module:parameters").process(frame:getParent().args, params)
	
	local au = args.audio
	local dev = args.dev or args.devm
	local dev2 = args.dev2
	local maxindex = table.getn(args[1])
	local html_list_main = mw.html.create("ul")
	local categories = {}
	
	-- Deals with the accents
	local a, al, ar, an = args.accent, args.accent_loc, args.accent_ref, args.accent_note
	for i, position in ipairs(a) do
		local text = args[1][min(maxindex,i)]
		if not al[i] then
			al[i] = "[[w:Tokyo dialect|Tokyo]]"
		end
		local result = m_accent.format_qualifiers(lang, {al[i]}) .. " "
		
		result = result .. export.accent(text, position, dev, dev2)
		if ar[i] then
			result = result .. add_acc_refs(frame, ar[i])
		else
			require("Module:debug").track("ja-pron/unsourced accent")
		end
		result = result .. (an[i] and (" " .. an[i]) or "")

		html_list_main:tag("li"):wikitext(
			result
		)
	end
	
	-- Deals with the IPA
	local m_IPA, no_acc = require("Module:IPA"), false
	for i, text in ipairs(args[1]) do
		html_list_main:tag("li"):wikitext(
			m_IPA.format_IPA_full {
				lang = lang,
				items = {{ pron = "[" .. export.ipa(text, dev, dev2) .. "]" }},
			}
		)
		if not a[i] then
			no_acc = true
		end
	end
	
	if no_acc then
		insert(categories, lang:getCanonicalName() .. " terms with IPA pronunciation missing pitch accent")
	end
	
	-- Deals with the audio
	if au then
		html_list_main:tag("li"):wikitext(
			require("Module:audio").format_audio {
				lang = lang,
				file = au,
			}
		)
	end
	
	return "\n" .. tostring(html_list_main) ..
		(#categories > 0 and require("Module:utilities").format_categories(categories, lang) or "")
end

function export.ipa(text, dev, dev2)
	if type(text) == "table" then
		text, dev, dev2 = text.args[1], text.args["dev"], text.args["dev2"] end
	dev = dev or ""
	dev2 = dev2 or ""
	
	if dev2 ~= "" then error("Please remove parameter dev2 and change parameter dev to \"dev=" .. dev .. "," .. dev2 .. "\"") end
	
	-- Convert 〜 and 〰 to a regular ー.
	text = gsub(text, "[〜〰]", "ー")
	
	local position_mora = {}
	
	for i = 1, len(text) do
		if not match(sub(text, i, i), "[ " .. submoraic_kana .. "%.]") then
			local nxt = sub(text, i + 1,i + 1)
			if nxt and match(nxt, "[" .. submoraic_kana .. "]") then
				insert(position_mora, i + 1)
			else
				insert(position_mora, i)
			end
		end
	end
	
	-- insert @ to stand for devoicing
	if dev ~= "" then
		for position in gsplit(dev, ",") do
			position = tonumber(position)
			if #position_mora == position then
				text = text .. "@"
			else
				local position_devspace = position_mora[position]
				text = sub(text, 1, position_devspace) .. "@" .. sub(text, position_devspace+1, -1)
			end
			for i = position + 1, #position_mora do
				position_mora[i] = position_mora[i] + 1
			end
		end
	end
	
	text = toNFC(kana_to_romaji(text, "ja", nil, {keep_dot = true, disambig = true}))
	
	text = gsub(text, "[lv'@]", {
		["l"] = "r", ["v"] = "b", ["'"] = "ʔ", ["@"] = "̥"
	})
	
	-- Hyphens which have been geminated over are removed; otherwise converted to dots.
	text = gsub(text, "([bcdfghjkmnprstvw])%-%1", "%1%1")
		:gsub("-", ".")
	
	text = text:gsub("([kprt])(%s*)%1", "%1̚%2%1")
		:gsub("t(%s*)ch", "t̚%1ch")
		:gsub("([bd])(%s*)%1", "%1̚%2%1̥")
		:gsub("g(%s*)g", "g̚%1g̊")
		:gsub("([jz])(%s*)%1", "d̚%2%1")
		:gsub("s(%s*)sh", "ɕ%1ɕ")
	
	text = gsub(text, "ei", "ē")
	text = gsub(text, "[āēīōūfvjryz]", {
		["ā"] = "aː", ["ē"] = "eː", ["ī"] = "iː", ["ō"] = "oː", ["ū"] = "uː", 
		["f"] = "ɸ", ["j"] = "d͡ʑ", ["r"] = "ɾ", ["y"] = "j", ["z"] = "d͡z" })
	
	text = gsub(text, "[sct][hs]", {
		["sh"] = "ɕ", 
		["ch"] = "t͡ɕ", 
		["ts"] = "t͡s" })
	
	text = gsub(text, "([aeiouː̥])d͡([zʑ])", "%1%2")

	text = gsub(text, "([bdɸgkmnpɾstz][̥̊]*)i", "%1ʲi")
	text = gsub(text, "([bdɸgkmnpɾstwz][̥̊]*)j", "%1ʲ")
	text = gsub(text, "([bɕdɸghjkmpɾstzʑʲ][̥̊]*)w", "%1ᵝ")
	text = gsub(text, "nʲ", "ɲ̟")
	text = gsub(text, "n+$", function(n)
		return ("ɴ"):rep(#n)
	end)
	text = gsub(text, "([^ ː_nɴʔ])(ː*)ɴ", "%1̃%2ɴ")
	text = gsub(text, "([^ ː_nʔ])(ː*)n([^aeou%s])", "%1̃%2n%3")
	text = gsub(text, "n[n ]*[bmp]", function(m)
		return m:gsub("n", "m")
	end)
	text = gsub(text, "n[n ]*.͡[ɕʑ]", function(m)
		return m:gsub("n", "ɲ̟")
	end)
	text = gsub(text, "n[n ]*ɲ̟", function(m)
		return m:gsub("n", "ɲ̟")
	end)
	text = gsub(text, "n[n ]*ɾ", function(m)
		return m:gsub("n", "n̺")
	end)
	text = gsub(text, "_ng", "ŋ")
	text = gsub(text, "(n[n ]*)([kg])([ʲᵝ]*)", function(m1, m2, m3)
		return m1:gsub("n", "ŋ" .. m3) .. m2 .. m3
	end)
	text = gsub(text, "_nw", "nᵝ")
	text = gsub(text, "n[n ]*[ɸszɕhjw]", function(m)
		return m:gsub("n", "ɰ̃")
	end)
	text = gsub(text, "([n ]*n)[ʔ_]", function(m) -- ʔ to be removed once Hrkt-translit `disambig` flag is implemented, as ʔ will always represent a glottal
		return m:gsub("n", "ɰ̃")
	end)
	text = gsub(text, "n[n ]* [aeiou]", function(m)
		return m:gsub("n", "ɰ̃")
	end)
	
	text = gsub(text, "h[iju]", {
		["hi"] = "çi", ["hj"] = "ç",
		["hu"] = "ɸu"
	})
	
	text = gsub(text, "h([çɸ])", "%1%1")
	for _, cons in ipairs{"ç", "ɕ", "ɸ", "h", "j", "m", "n", "ɴ", "ŋ", "ɾ", "s", "w", "z", "ʑ"} do
		text = gsub(text, "(" .. cons .. ")()" .. cons .. "+()", function(cons, i, j)
			return cons .. ("ː"):rep(j - i)
		end)
	end
	for _, cons in ipairs{"n̺", "nᵝ", "ɲ̟", "ŋʲ", "ŋᵝ", "ɰ̃"} do
		local char1, char2 = cons:match("(.[\128-\191]*)(.*)")
		text = gsub(text, "(" .. cons .. ")()" .. char1 .. "[" .. char1 .. char2 .. "]*" .. char2 .. "()", function(cons, i, j)
			return cons .. ("ː"):rep((j - i) / 2)
		end)
	end
	text = gsub(text, "(ː+)([ʲᵝ]+)", "%2%1")
	text = gsub(text, "̚(.[̥̊]*)([ʲᵝ]+)", "̚%2%1%2")
	text = gsub(text, "[aeiouw]", {
		["a"] = "a̠", 
		["e"] = "e̞", 
		["o"] = "o̞", 
		["u"] = "ɯ̟", 
		["w"] = "β̞"
	})
	
	text = gsub(text, "([szɕʑɲçʲ][̟̥̊]*ː*)ɯ̟", "%1ɨ")
	
	text = gsub(text, "̠[̥̃][̥̃]", "̥̃˗")
	text = gsub(text, "̞[̥̃][̥̃]", "̥̃˕")
	text = gsub(text, "̟[̥̃][̥̃]", "̥̃˖")
	text = gsub(text, "([̠̞̟])̥", "%1̊")
		:gsub("[%._]", "")
		:gsub("g", "ɡ")
	
	return text
end

function export.rise_and_fall(word, rftype)
	word = gsub(word, "([" .. o_kana .. "][゙゚]?)([うウ])", "%1.%2")
	word = gsub(word, "([" .. e_kana .. "][゙゚]?)([いイ])", "%1.%2")
	word = kana_to_romaji(word, "ja")

	if rftype == "rise" then
		word = gsub(word, ".", {
			["a"] = "á", ["e"] = "é", ["i"] = "í", ["o"] = "ó", ["u"] = "ú", 
			["ā"] = "áá", ["ē"] = "éé", ["ī"] = "íí", ["ō"] = "óó", ["ū"] = "úú" })
	
		word = gsub(gsub(word, "n([bcdfghjkmnprstvw%'z ])", "ń%1"), "n$", "ń")
		
	elseif rftype == "fall" then
		word = gsub(word, ".", {
			["a"] = "à", ["e"] = "è", ["i"] = "ì", ["o"] = "ò", ["u"] = "ù", 
			["ā"] = "àà", ["ē"] = "èè", ["ī"] = "ìì", ["ō"] = "òò", ["ū"] = "ùù" })
		
		word = gsub(gsub(word, "n([bcdfghjkmnprstvw%'z ])", "ǹ%1"), "n$", "ǹ")
		
	else
		return error("Type not recognised.")
	end
	
	return word
end

-- [[Module:ja-ojad]] and [[Module:ja-infl-demo]] rely on the output format of this function
function export.accent(text, class, dev, dev2)
	local result

	if(type(text)) == "table" then text, class, dev, dev2 = text.args[1], text.args[2], text.args["dev"], text.args["dev2"] end
	text = gsub(text, "([" .. o_kana .. "][゙゚]?)[うウ]", "%1ー")
	text = gsub(text, "([" .. e_kana .. "][゙゚]?)[いイ]", "%1ー")
	text = gsub(text, "%.", "")
	if dev == "" then dev = false end
	if dev2 == "" then dev2 = false end
	
	local down_first = "<span style=\"border-top:1px solid;position:relative;padding:1px;\">"
	local down_last = "<span style=\"position:absolute;top:0;bottom:67%;right:0%;border-right:1px solid;\"></span></span>"
	local high_first = "<span style=\"border-top:1px solid\">"
	local start = "<span lang=\"ja\" class=\"Jpan\">"
	local romaji_start = " <span class=\"Latn\"><samp>["
	local romaji_last = "]</samp></span> "
	local last = "</span>"
	
	local position_kana = {}       --position of each kana (ぁ counted), text without space
	local position_mora = {}       --position of each mora (ぁ not counted), text without space
	local position_mora_space = {} --position of each mora (ぁ not counted), text with space
	
	for i=1, len(text) do
		if not match(sub(text,i,i), "[ " .. submoraic_kana .. "]") then
			local extra = len(match(sub(text,i+1), "^[" .. submoraic_kana .. "]*"))
			insert(position_mora_space, i+extra)
		end
	end
	local space_removed = gsub(text," ","")
	for i=1, len(space_removed) do
		insert(position_kana, i)
		if not match(sub(space_removed,i,i), "[" .. submoraic_kana .. "]") then
			local extra = len(match(sub(space_removed,i+1), "^[" .. submoraic_kana .. "]*"))
			insert(position_mora, i+extra)
		end
	end
	
	local acc_type, acc_number
	if match(class, "^[h0]$") then
		acc_type, acc_number = "h", 0
		
	elseif match(class, "^[a1]$") then
		acc_type, acc_number = "a", 1
	
	elseif match(class, "^o$") then
		acc_type = "o"
		acc_number = len(gsub(text, "[ " .. submoraic_kana .. "]", ""))
	end
	
	if match(class, "^[0-9]+$") and not match(class,"^[01]$") then
		class = gsub(class, "[on]", "")
		acc_number = tonumber(class)
		
		local morae_count = len(gsub(text, "[ " .. submoraic_kana .. "]", ""))
		
		if morae_count == acc_number then
			acc_type = "o"
		elseif morae_count < acc_number then
			return error(("Mora count (%d) is smaller than position of downstep mora (%d).")
				:format(morae_count, acc_number))
		else
			acc_type = "n"
		end
	elseif not acc_number then
		acc_number = class
	end
	
	local start_index = 1
	while match(sub(text, start_index+1, start_index+1), "[" .. submoraic_kana .. "]") do
		start_index = start_index + 1
	end
	
	local kanas = {}
	local single_mora
	for i=1, len(text) do
		if not match(sub(text,i,i), "[ " .. submoraic_kana .. "]") then
			single_mora = gsub(sub(text, i, -1), "^(.[" .. submoraic_kana .. "]*).*", "%1")
			insert(kanas, single_mora)
		end
	end
	
	local function kana_devoice(text)
		return "<span style=\"border:1px dotted gray; border-radius:50%;\">" .. text .. "</span>"
	end
	
	if dev then
		for position in gsplit(dev, ",") do
			position = tonumber(position)
			kanas[position] = kana_devoice(kanas[position])
		end
	end
	
	local romaji_text = gsub(text, "([" .. o_kana .. "][゙゚]?)ー", "%1お")
	romaji_text = gsub(romaji_text, "([" .. e_kana .. "][゙゚]?)ー", "%1え")
	romaji_text = gsub(romaji_text, "([" .. u_kana .. "][゙゚]?)ー", "%1う")
	romaji_text = gsub(romaji_text, "([" .. i_kana .. "][゙゚]?)ー", "%1い")
	romaji_text = gsub(romaji_text, "([" .. a_kana .. "][゙゚]?)ー", "%1あ")
	romaji_text = gsub(romaji_text, "([" .. n_kana .. "][゙゚]?)ー", "%1%1")
	
	local romajis = split(romaji_text, "")
	
	local function count_nspaces(text, index)
		local i, sample, nspaces = 0, "", 0
		while len(sample) < index do
			i = i + 1
			sample, nspaces = gsub(sub(text, 1, i), " ", "")
		end
		return nspaces
	end
	
	local function romaji_devoice(text)
		-- use @ instead of  ̥
		return text .. "@"
	end
	
	if dev then
		for position in gsplit(dev,",") do
			position = position_mora_space[tonumber(position)]
			romajis[position] = romaji_devoice(romajis[position])
		end
	end
	
	if acc_type == "n" then
		local r_start_index = start_index + count_nspaces(romaji_text, start_index)
		local r_index = position_mora_space[acc_number]
		local k_index = acc_number
		
		local r_parts = {
			[1] = concat(romajis, "", 1, r_start_index),
			[2] = concat(romajis, "", r_start_index + 1, r_index),
			[3] = concat(romajis, "", r_index + 1, #romajis)
		}
		
		local k_parts = {
			[1] = concat(kanas, "", 1, 1),
			[2] = concat(kanas, "", 2, k_index),
			[3] = concat(kanas, "", k_index + 1, #kanas)
		}
		
		local space2 = ""
		local space3 = ""
		if sub(r_parts[2], 1, 1) == " " then
			space2 = " "
		end
		if sub(r_parts[3], 1, 1) == " " then
			space3 = " "
		end
		
		result = start .. 
			k_parts[1] .. 
			down_first .. 
			k_parts[2] .. 
			down_last .. 
			k_parts[3] .. 
			last .. 
			romaji_start .. 
			export.rise_and_fall(r_parts[1], "fall") .. 
			space2 ..
			export.rise_and_fall(r_parts[2], "rise") .. 
			"ꜜ" .. 
			space3 ..
			export.rise_and_fall(r_parts[3], "fall") .. 
			romaji_last .. 
			"([[中高型|Nakadaka]] – [" .. acc_number .. "])"
		
	else
		local r_start_index = start_index + count_nspaces(romaji_text, start_index)
		local r_parts = {
			[1] = concat(romajis, "", 1, r_start_index),
			[2] = concat(romajis, "", r_start_index + 1, #romajis)
		}
		
		local k_parts = {
			[1] = concat(kanas, "", 1, 1),
			[2] = concat(kanas, "", 2, #kanas)
		}
		
		local space2 = ""
		if sub(r_parts[2], 1, 1) == " " then
			space2 = " "
		end
			
		if acc_type == "h" then
			result = start .. 
				k_parts[1] .. 
				high_first .. 
				k_parts[2] .. 
				last .. 
				last .. 
				romaji_start .. 
				export.rise_and_fall(r_parts[1], "fall") .. 
				space2 ..
				export.rise_and_fall(r_parts[2], "rise") .. 
				romaji_last .. 
				"([[平板型|Heiban]] – [" .. acc_number .. "])"
				
		elseif acc_type == "a" then
			result = start .. 
				down_first .. 
				k_parts[1] .. 
				down_last .. 
				k_parts[2] .. 
				last .. 
				romaji_start .. 
				export.rise_and_fall(r_parts[1], "rise") .. 
				"ꜜ" .. 
				space2 ..
				export.rise_and_fall(r_parts[2], "fall") .. 
				romaji_last .. 
				"([[頭高型|Atamadaka]] – [" .. acc_number .. "])"
				
		elseif acc_type == "o" then
			result = start .. 
				k_parts[1] .. 
				down_first .. 
				k_parts[2] .. 
				down_last .. 
				last .. 
				romaji_start .. 
				export.rise_and_fall(r_parts[1], "fall") .. 
				space2 ..
				export.rise_and_fall(r_parts[2], "rise") .. 
				"ꜜ" .. 
				romaji_last .. 
				"([[尾高型|Odaka]] – [" .. acc_number .. "])"
				
		else
			return error("Accent type not recognised.")
		end
		
	end
	
	result = gsub(result, "(.)@", "<del>%1</del>")
	
	return result
end

return export