Module:ko-pron

local export = {}

local m_data = mw.loadData("Module:ko-pron/data")
local m_ko_utilities = require("Module:ko")
local m_str_utils = require("Module:string utilities")

local codepoint = m_str_utils.codepoint
local concat = table.concat
local floor = math.floor
local gmatch = m_str_utils.gmatch
local gsub = m_str_utils.gsub
local insert = table.insert
local len = m_str_utils.len
local match = m_str_utils.match
local pattern_escape = m_str_utils.pattern_escape
local sub = m_str_utils.sub
local toNFC = mw.ustring.toNFC
local u = m_str_utils.char
local upper = m_str_utils.upper

local PAGENAME = mw.loadData("Module:headword/data").pagename
local PAGENAME2 = gsub(PAGENAME, "-", "")

local system_lookup = {
	 = 1,  = 2,  = 3,
	 = 4,  = 5,  = 6,
}

local question_mark = "<sup><small>]</small></sup>"

local system_list = {
	{ 
		abbreviation = "ph", 
		display = "Phonetic hangul: ", 
		separator = "/",
	},
	{ 
		abbreviation = "rr", 
		display = "Revised Romanization" .. question_mark, 
		separator = "/",
	},
	{ 
		abbreviation = "rrr", 
		display = "Revised Romanization (translit.)" .. question_mark, 
		separator = "/"
	},
	{ 
		abbreviation = "mc", 
		display = "McCune–Reischauer" .. question_mark,
		separator = "/"
	},
	{ 
		abbreviation = "yr", 
		display = "Yale Romanization" .. question_mark,
		separator = "/"
	},
	{ 
		abbreviation = "ipa", 
		display = "(<i>]/]</i>) ]<sup>(])</sup>: ", 
		separator = " ~ "
	}
}

--[[

vowel_variation:
	rules for vowel transformation.
	key:
		the number of a syllable's vowel (vowel_id):
			floor(((codepoint('가') - 0xAC00) % 588) / 28) = 0
			floor(((codepoint('개') - 0xAC00) % 588) / 28) = 1
	value:
		an integer that is added to the decimal codepoint of the syllable
				u(codepoint('개') + 112) = '게'

allowed_vowel_scheme:
	a list of which systems vowel transformation is reflected in.
	key:
		vowel_id .. "-" .. system_index
		system_index: see system_list above. IPA is #6
	value:
		1, representing true

]]

local final_syllable_conversion = {  = "Ø",  = "" }
local com_mc = {  = "k",  = "t",  = "p",  = "ch",  = "s",  = "ss" }
local com_ph = {  = "ᄁ",  = "ᄄ",  = "ᄈ",  = "ᄊ",  = "ᄍ" }
local vowel_variation = {
	 = 112,  -- 개→게
	 = 112,  -- 걔→계
	 = 140, -- 괘→궤

	 = -56,  -- 계→게

	 = 112, -- 괴→궤
	 = 0,   -- 귀→귀
}
local allowed_vowel_scheme = {
	 = 1,
	 = 1,
	 = 1,
	 = 1,
	 = 1,
	 = 1,

	 = 1,
	 = 1,

	 = 1,
	 = 1,
	 = 1,
}
local ambiguous_intersyllabic_rr = {  = 1,  = 1,  = 1,  = 1,  = 1 }
local ambiguous_intersyllabic_mr = {  = 1,  = 1 }
local ambiguous_intersyllabic_yr = {  = 1,  = 1,  = 1,  = 1,  = 1,  = 1,  = 1,  = 1,  = 1 }

local function decompose_syllable(word)
	local decomposed_syllables = {}
	for syllable in mw.text.gsplit(word, "") do
		insert(decomposed_syllables, m_ko_utilities.decompose_jamo(syllable))
	end
	return decomposed_syllables
end

local function tidy_phonetic(original, romanised)
	local j, k, w = 1, 1, {}
	for i = 1, len(romanised) do
		local romanised_syllable = sub(romanised, k, k)
		local original_syllable = sub(original, j, j)
		if romanised_syllable ~= original_syllable then
			insert(w, '<b>'..romanised_syllable..'</b>')
			local original_advance = match(original_syllable, "")
			local romanised_advance = match(romanised_syllable, "")
			if original_advance or not romanised_advance then
				k = k + 1
			end
			if romanised_advance or not original_advance then
				j = j + 1
			end
		else
			insert(w, '<span>'..romanised_syllable..'</span>')
			j, k = j + 1, k + 1
		end
	end
	return concat(w)
end

local function tidy_ipa(ipa)
	ipa = gsub(ipa, "ʌ̹%(ː%)", "ɘ(ː)") -- TODO: ] really should be  ~  instead of 
	ipa = gsub(ipa, "ɭɭi", "ʎʎi")
	ipa = gsub(ipa, "ɭɭj", "ʎʎ")
	ipa = gsub(ipa, "s()ɥi" ,"ʃ%1ɥi")
	ipa = gsub(ipa, "ss͈()" ,"ɕɕ͈%1")
	ipa = gsub(ipa, "s()()" ,"ɕ%1%2")
	ipa = gsub(ipa, "nj", "ɲ")
	ipa = gsub(ipa, "()(?)j", "%1%2")
	
	ipa = gsub(ipa, "kʰ", { 
		 = "cçi", 
		 = "cç", 
		 = "kxɯ" }
	)
	ipa = gsub(ipa, "", {
		 = "çi",
		 = "ç",
		 = "xɯ",
		 = "ɸʷo",
		 = "ɸʷu",
		 = "ɸw",
		 = "ʝi",
		 = "ʝ",
		 = "ɣɯ",
		 = "βo",
		 = "βu",
		 = "βw" }
	)
	
	if match(ipa, "ɥi") then
		local midpoint = floor(len(ipa) / 2)
		ipa = sub(ipa, 1, midpoint) .. gsub(sub(ipa, midpoint+1, -1), "ɥi", "y")
	end
	
	return ipa
end

function export.romanise(text_param, system_index, args)
	if type(text_param) == "table" then
		args = text_param:getParent().args
		system_index = args or 2
		text_param = args
	end
	local p, optional_params = {}, { "nn", "l", "com", "cap", "ni" }
	for _, pm in ipairs(optional_params) do
		p = { }
		if args then
			for pp in mw.text.gsplit(args, ",") do p = 1 end
		end
	end

	local categories = {}
		
	local vowel_ui_i, vowel_ui_e, no_batchim, batchim_reduce, s_variation, iotation = 
		args.ui, args.uie, args.nobc, args.bcred, args.svar, args.iot
	
	system_index = system_lookup or system_index
	text_param = gsub(text_param, '(.)', "%1")
	
	for primitive_word in gmatch(text_param, "+") do
		local the_original = primitive_word
		primitive_word = gsub(primitive_word, "'''", "ß")
		local formatting_position, formatting_count = {}, 0
		primitive_word = gsub(primitive_word, "()()", function(m1, m2)
			formatting_position = m2 == "ß" and "'''" or m2
			return ""
		end)
		
		local has_vowel = {}
		for ch in gmatch(primitive_word, ".") do
			local jungseong = floor(((codepoint(ch) - 0xAC00) % 588) / 28)
			if not match(ch, "") and match(ch, "") then has_vowel = true end
		end
		local word_set = { primitive_word }
		
		local function add_respelling(variable, modification, modification2)
			modification2 = modification2 or function(x) return x end
			if variable and match(system_index, "") then
				variable = tonumber(variable)
				local pre_length = #word_set
				for i = 1, pre_length do
					local item = mw.text.split(word_set, "")
					item = modification(item)
					item = modification2(item)
					word_set = concat(item)
				end
			end
		end
		add_respelling(vowel_ui_i, function(x) return "이" end)
		add_respelling(vowel_ui_e, function(x) return "에" end)
		
		add_respelling(no_batchim, 
			function(x) return u(codepoint(x) - (codepoint(x) - 0xAC00) % 28) end, 
			function(y) return u(codepoint(y) + 588) end)
		
		add_respelling(s_variation, function(x) return u(codepoint(x) - 12) end)
		add_respelling(iotation, function(x) return u(codepoint(x) + 56) end)
		
		for vowel_id, vowel_variation_increment in pairs(vowel_variation) do
			if has_vowel and allowed_vowel_scheme then
				local pre_length = #word_set
				for i = 1, pre_length do
					local item = mw.text.split(word_set, "")
					for num, it in ipairs(item) do
						if floor(((codepoint(it) - 0xAC00) % 588) / 28) == vowel_id then
							item = u(codepoint(it) + vowel_variation_increment)
						end
					end
					if vowel_id == 11 then
						insert(word_set, i, concat(item))
					else
						insert(word_set, concat(item))
					end
				end
			end
		end
		
		local word_set_romanisations = {}
		for _, respelling in ipairs(word_set) do
			local decomposed_syllables = decompose_syllable(respelling)
			local romanisation = {}
			local formatting_insert_count = 0
			for index = 0, #decomposed_syllables, 1 do
				local this_syllable_text = index ~= 0 and sub(respelling, index, index) or ""
				if this_syllable_text == "-" then
					-- skip it, it will be handled below
				else
					local syllable = decomposed_syllables or { initial = "Ø", vowel = "Ø", final = "X" }
					local next_index = index
					local next_syllable_text
					local saw_hyphen_after = false
					while true do
						next_index = next_index + 1
						next_syllable_text = next_index > #decomposed_syllables and "" or sub(respelling, next_index, next_index)
						if next_syllable_text ~= "-" then
							break
						end
						saw_hyphen_after = true
					end
					local next_syllable = decomposed_syllables or { initial = "Ø", vowel = "Ø", final = "Ø" }
					syllable.final = final_syllable_conversion or syllable.final
					
					if system_index == 5 and syllable.vowel == "ᅮ" and match(syllable.initial, "") then
						syllable.vowel = "ᅳ"
					end
					
					if match(system_index, "") then
						if match(syllable.initial, "") then
							if syllable.vowel == "ᅣ" then
								syllable.vowel = "ᅡ"
							elseif syllable.vowel == "ᅤ" then
								syllable.vowel = "ᅢ"
							elseif syllable.vowel == "ᅧ" then
								syllable.vowel = "ᅥ"
							elseif syllable.vowel == "ᅨ" then
								syllable.vowel = "ᅦ"
							elseif syllable.vowel == "ᅭ" then
								syllable.vowel = "ᅩ"
							elseif syllable.vowel == "ᅲ" then
								syllable.vowel = "ᅮ"
							end
						end
					end
					
					if match(system_index, "") then
						if syllable.vowel == "ᅴ" and this_syllable_text ~= "의" then
							syllable.vowel = "ᅵ"
						end
					end
					if match(system_index, "") then
						if this_syllable_text == "넓" then
							if match(next_syllable.initial, "") then
								syllable.final = "ᆸ"
								
							elseif next_syllable.initial == "ᄃ" then
								if match(next_syllable.vowel, "") then
									syllable.final = "ᆸ"
								end
							end
						end
					end
					
					local vowel = m_data.vowels

					if p.nn and match(system_index, "") then
						next_syllable.initial = "ᄂ"
					end
					if p.com and match(system_index, "") then
						next_syllable.initial = com_ph or next_syllable.initial
					end
					
					if p.ni and system_index ~= 3 then
						next_syllable.initial = (system_index == 5 and syllable.final == "ᆯ") and "ᄅ" or "ᄂ"
					end
					
					if match(system_index, "") then
						if tonumber(batchim_reduce or -1) == index then
							syllable.final = m_data.boundary
						end
					
						if index ~= 0 and this_syllable_text == "밟" and not
							match(next_syllable.initial, "") then
								syllable.final = "ᆸ"
						end
						
						if next_syllable_text == "없" then
							if match(syllable.final, "") then
								syllable.final = "ᆨ"
							elseif match(syllable.final, "") then
								syllable.final = "ᆫ"
							elseif match(syllable.final, "") then
								syllable.final = "ᆮ"
							elseif match(syllable.final, "") then
								syllable.final = "ᆯ"
							elseif syllable.final == "ᆱ" then
								syllable.final = "ᆷ"
							elseif match(syllable.final, "") then
								syllable.final = "ᆸ"
							end
						end
						
						if tonumber(batchim_reduce or -1) ~= index then
							if match(syllable.final .. next_syllable.initial, "ᇀᄋ") then
								if next_syllable.vowel == "ᅵ" then
									syllable.final = "ᆾ"
								elseif next_syllable.vowel == "ᅧ" then
									syllable.final = "ᆾ"
									next_syllable.vowel = "ᅥ"
								end
							
							elseif match(syllable.final .. next_syllable.initial, "ᆴᄋ") then
								if next_syllable.vowel == "ᅵ" then
									syllable.final = "ᆯ"
									next_syllable.initial = "ᄎ"
								elseif next_syllable.vowel == "ᅧ" then
									syllable.final = "ᆯ"
									next_syllable.initial = "ᄎ"
									next_syllable.vowel = "ᅥ"
								end
							
							elseif match(syllable.final .. next_syllable.initial, "ᆮᄋ") and tonumber(s_variation or -1) ~= index then
								if next_syllable.vowel == "ᅵ" then
									syllable.final = "ᆽ"
								elseif next_syllable.vowel == "ᅧ" then
									syllable.final = "ᆽ"
									next_syllable.vowel = "ᅥ"
								end
							
							elseif match(syllable.final .. next_syllable.initial, "ᆮᄒ") then
								if next_syllable.vowel == "ᅵ" then
									syllable.final = "ᆾ"
									next_syllable.initial = "ᄋ"
								elseif next_syllable.vowel == "ᅧ" then
									syllable.final = "ᆾ"
									next_syllable.initial = "ᄋ"
									next_syllable.vowel = "ᅥ"
								end
							
							elseif match(syllable.final .. next_syllable.initial .. next_syllable.vowel, "ᅧ") then
								next_syllable.vowel = "ᅥ"
							end
						end
						
						if syllable.final .. next_syllable.initial == "ᆺᄋ" and not
							match(next_syllable_text, "") then
								syllable.final = "ᆮ"
						end
					end
					
					local bound = syllable.final .. "-" .. next_syllable.initial
					if not m_data.boundary then
						require("Module:debug").track("ko-pron/no boundary data")
						mw.log("No boundary data for " .. bound .. ".")
						return nil
					end
					local junction = m_data.boundary
					
					if system_index == 2 then
						local pos_format_start = index + formatting_insert_count + 1
						local pos_format_end = pos_format_start
						while formatting_position do
							pos_format_end = pos_format_end + 1
							formatting_insert_count = formatting_insert_count + 1
						end
						if pos_format_end > pos_format_start then
							local a, b = match(junction, "^(ng%-?)(.?)$")
							if not a or not b then a, b = match(junction, "^(.?%-?)(.*)$") end
							junction = match(syllable.final .. next_syllable.initial, "^Ø?$")
								and concat(formatting_position, "", pos_format_start, pos_format_end - 1) .. (a or "") .. (b or "")
								or (a or "") .. concat(formatting_position, "", pos_format_start, pos_format_end - 1) .. (b or "")
						end
					end
					
					if p.l or (p.l and index == 1) then
						-- FIXME, verify this code still works with final/initial cons changes
						if system_index == 1 then
							if #junction == 0 then
								junction = junction .. "ː"
							else
								junction = gsub(junction, "^(.)(.?)$", function(a, b)
									return match(a, "") and a .. "ː" .. b or "ː" .. a .. b end)
							end
							
						elseif system_index == 5 then
							vowel = gsub(vowel, "()", "%1̄")
							
						elseif system_index == 6 then
							vowel = vowel .. "ː"
							if index == 1 then
								insert(categories, "Korean terms with long vowels in the first syllable")
							end
						end
					end
					
					if (p.l or p.l) and index == 0 and system_index == 6 and #decomposed_syllables > 1 then
						vowel = vowel .. "ˈ"
					end
				
					if p.com then
						-- FIXME, verify this code still works with final/initial cons changes
						junction = gsub(junction, "(.)$", function(next_letter)
							return 
								(system_index == 5 and "q" or "") .. 
								(system_index == 4
									and (com_mc or "")] or com_mc or next_letter)
									or next_letter) end)
					end
					
					if p.ni and system_index == 5 then
						-- FIXME, verify this code still works with final/initial cons changes
						junction = gsub(junction, "()$", "<sup>%1</sup>")
					end

					local final_cons, initial_cons = match(junction, "^(.*);(.*)$")
					if not final_cons then
						if system_index == 2 then
							error("Need a semicolon in the boundary value for " .. bound)
						end
						-- FIXME, throw an error for all systems once we've added semicolons everywhere
						final_cons = junction
						initial_cons = ""
					end

					if system_index == 2 then
						insert(romanisation, vowel .. final_cons .. (saw_hyphen_after and "-" or "") .. initial_cons)
					else
						insert(romanisation, vowel .. junction)
					end
				end
			end
			
			local temp_romanisation = concat(romanisation)
			if system_index == 1 then
				temp_romanisation = tidy_phonetic(primitive_word, toNFC(temp_romanisation))
			
			elseif match(system_index, "") then
				for i = 1, 2 do
					temp_romanisation = gsub(temp_romanisation, "(.)…(.)", function(a, b)
						return a .. (ambiguous_intersyllabic_rr and "'" or "") .. b end)
					temp_romanisation = gsub(temp_romanisation, "wo'e", "woe")
					temp_romanisation = gsub(temp_romanisation, "yo'e", "yoe")
					temp_romanisation = gsub(temp_romanisation, "we'o", "weo")
					temp_romanisation = gsub(temp_romanisation, "we'u", "weu")
					temp_romanisation = gsub(temp_romanisation, "ye'u", "yeu")
					temp_romanisation = gsub(temp_romanisation, "yu'i", "yui")
				end
			
			elseif system_index == 4 then
				for i = 1, 2 do
					temp_romanisation = gsub(temp_romanisation, "(.)…(.)", function(a, b)
						return a .. (ambiguous_intersyllabic_mr and "'" or "") .. b end)
					temp_romanisation = gsub(temp_romanisation, "yo'e", "yoe")
					temp_romanisation = gsub(temp_romanisation, "a'e", "aë")
					temp_romanisation = gsub(temp_romanisation, "o'e", "oë")
					temp_romanisation = gsub(temp_romanisation, "n'k", "nk")
					temp_romanisation = gsub(temp_romanisation, "swi", "shwi")
				end
			
			elseif system_index == 5 then
				for i = 1, 2 do
					temp_romanisation = gsub(temp_romanisation, "(.)…(.)", function(a, b)
						return a .. (ambiguous_intersyllabic_yr and "." or "") .. b end)
					temp_romanisation = gsub(temp_romanisation, "q", "q")
				end

			elseif system_index == 6 then
				temp_romanisation = ""
			end

			if match(system_index, "") then
				temp_romanisation = gsub(temp_romanisation, "ː", "(ː)")
			end
			
			if p.cap and match(system_index, "") then
				temp_romanisation = upper(sub(temp_romanisation, 1, 1)) .. sub(temp_romanisation, 2, -1)
			end

			insert(word_set_romanisations, temp_romanisation)
		end

		text_param = gsub(
			text_param,
			pattern_escape(the_original),
			concat(word_set_romanisations, system_list.separator),
			1
		)
	end

	if system_index == 6 then
		text_param = tidy_ipa(text_param)
	end

	if #categories > 0 then
		text_param = text_param .. require("Module:utilities").format_categories(categories, m_ko_utilities.lang)
	end
	
	return text_param
end

function export.make(frame, scheme)
	local params = {
		 = { default = PAGENAME2, list = true },
		
		 = {},
		 = { alias_of = "a" },
		
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
	}
	
	local args = require("Module:parameters").process(frame:getParent().args, params)
	
	local results = {}
	for _, text_param in ipairs(args) do
		local current_word_dataset = {}
		for system_index, system in pairs(system_list) do
			local romanised = export.romanise(text_param, system_index, args)
			insert(current_word_dataset, romanised)
		end
		insert(results, current_word_dataset)
	end
	
	local output_result = {  = {},  = {},  = {},  = {},  = {},  = {} }
	for _, result in ipairs(results) do
		for result_index, value in ipairs(result) do
			insert(output_result, value)
		end
	end

	local html_ul = mw.html.create( "ul" )
	:done()
	local html_li_ipa = mw.html.create( "li" )
		:wikitext( system_list.display )
		:tag( "span" )
			:addClass( "IPA" )
			:wikitext( concat(output_result, system_list.separator) )
		:done()
	:done()
	local html_li_ph = mw.html.create( "li" )
		:addClass( "ko-pron__ph" )
		:wikitext( system_list.display )
		:tag( "span" )
			:addClass( "Kore" )
			:attr( "lang", "ko" )
			:wikitext( ", system_list.separator) .. "]" )
		:done()
	:done()

	if args.a then
		html_li_ipa
			:tag( "ul" )
				:tag( "li" )
					:wikitext( require("Module:audio").format_audio {
						lang = m_ko_utilities.lang,
						file = args.a == "y" and "Ko-" .. PAGENAME .. ".ogg" or args.a,
					})
				:done()
			:done()
		:done()
	end

	if args.l then
		html_li_ph
			:tag( "ul" )
				:tag( "li" )
					:addClass( "ko-pron__note-vowel-length" )
					:wikitext( 'Though still prescribed in Standard Korean, most speakers in both Koreas no longer distinguish vowel length.' )
				:done()
			:done()
		:done()
	end

	html_ul
		:node( html_li_ipa )
		:node( html_li_ph )
	:done()

	local html_table = mw.html.create( "table" )
		:addClass( "ko-pron" )
		:addClass( "mw-collapsible" )
		:addClass( "mw-collapsed" )
		:tag( "tr" )
			:tag( "th" )
				:attr( "colspan", 2 )
				:wikitext( "Romanizations" )
			:done()
		:done()
	:done()
		
	for roman_index = 2, 5 do
		html_table
			:tag( "tr" )
				:tag( "th" )
					:wikitext( system_list.display )
				:done()
				:tag( "td" )
					:addClass( "IPA" )
					:wikitext( concat(output_result, system_list.separator) )
				:done()
			:done()
		:done()
	end
		
	return tostring(html_ul) .. tostring(html_table) .. require("Module:TemplateStyles")("Template:ko-IPA/style.css")
end

function export.make_hanja(frame, scheme)
	local params = {
		 = { list = true },

		 = {},
	}

	local args = require("Module:parameters").process(frame:getParent().args, params)

	local results = {
		 = {},
		 = {},
	}
	for _, text_param in ipairs(args) do
		for _, system_index in pairs({1, 6}) do
			local romanised = export.romanise(text_param, system_index, args)
			insert(results, romanised)
		end
	end

	local html_ul = mw.html.create( "ul" )
	:done()
	local html_li_ipa = mw.html.create( "li" )
		:wikitext( system_list.display )
		:tag( "span" )
			:addClass( "IPA" )
			:wikitext( concat(results, system_list.separator) )
		:done()
	:done()
	local html_li_ph = mw.html.create( "li" )
		:addClass( "ko-pron__ph" )
		:wikitext( system_list.display )
		:tag( "span" )
			:addClass( "Kore" )
			:attr( "lang", "ko" )
			:wikitext( ", system_list.separator) .. "]" )
		:done()
	:done()

	if args.l then
		html_li_ph
			:tag( "ul" )
				:tag( "li" )
					:addClass( "ko-pron__note-vowel-length" )
					:wikitext( 'Though still prescribed in Standard Korean, most speakers in both Koreas no longer distinguish vowel length.' )
				:done()
			:done()
		:done()
	end

	html_ul
		:node( html_li_ipa )
		:node( html_li_ph )
	:done()

	return tostring(html_ul) .. require("Module:TemplateStyles")("Template:ko-IPA/style.css")
end

return export