Module:yue-pron

local export = {}
local m_string_utils = require("Module:string utilities")

local gsplit = m_string_utils.gsplit
local gsub = m_string_utils.gsub
local len = m_string_utils.len
local lower = m_string_utils.lower
local split = m_string_utils.split

local entering = {
	p = 1, t = 1, k = 1
}

local entering_tones = {
	 = "7",  = "8",  = "9"
}

local ipa_allophones = {
	ei = "eri",
	eoi = "eoy",
	ing = "irng",
	ik = "irk",
	ou = "oru",
	ung = "urng",
	uk = "urk",
}

local ipa_initial = {
	 = "p",  = "pʰ",  = "m",  = "f", 
	 = "t",  = "tʰ",  = "n",  = "l", 
	 = "k",  = "kʰ",  = "ŋ",  = "kʷ",  = "kʷʰ",
--  = "t͡ʃ",  = "t͡ʃʰ",  = "ʃ",
	 = "t͡s",  = "t͡sʰ",  = "s", 
	 = "h",  = "w",  = "j", 
	 = ""
}

-- vowels with "r" only appear as allophones and should not appear in the input
local ipa_nucleus = {
	 = "aː",  = "ɐ", 
	 = "ɛː",  = "e", 
	 = "iː",  = "ɪ", 
	 = "ɔː",  = "o", 
	 = "œː",  = "ɵ", 
	 = "uː",  = "ʊ", 
	 = "yː"
}

local ipa_coda = {
	 = "i̯",  = "u̯",  = "y̯", 
	 = "m",  = "n",  = "ŋ", 
	 = "p̚",  = "t̚",  = "k̚", 
	 = ""
}

local ipa_tone = {
	 = "<span style=\"cursor:help\" title=\"or 53\">⁵⁵</span>", 
	 = "³⁵", 
	 = "³³", 
	 = "²¹", 
	 = "¹³", 
	 = "²²", 
	 = "⁵", 
	 = "³", 
	 = "²", 
	 = ""
}

local ipa_tone_sandhi = {
	 = "⁻",  = ""
}

local ipa_syllabic = {
	 = "m̩",  = "ŋ̍"
}


-- display `main`, but show `option` on hovering
local function alt(main,option)
	return '<span style="cursor:help" title="or ' .. option .. '">' .. main .. '</span>'
end

local acute = {
	a="á", e="é", i="í", o="ó", u="ú", m="ḿ", ng="ńg"
}

local grave = {
	a="à", e="è", i="ì", o="ò", u="ù", m="m̀", ng="ǹg"
}

local macron = {
	a="ā", e="ē", i="ī", o="ō", u="ū", m="m̄", ng="n̄g"
}

-- "?" indicates finals that are not supported by Yale
local yale_final = {
	 = "?",  = "a",
	 = "?",  = "?",  = "?",  = "?",  = "?",
	 = "?",  = "?",
	 = "?",  = "?",
	 = "eu",  = "?",  = "?",  = "eung",  = "?",  = "?",  = "euk", 
	 = "eui",  = "eun",  = "eut",
}

-- The core function to handle conversion to Yale.
-- For non-syllabics, there is exactly one vowel cluster in the syllable:
-- the first vowel is inputted as `a`, and the rest of the vowels is `b`, and `t` is the tone.
--- (e.g. "keui" -> a="e",b="ui"; "keung" -> a="e",b="u")
-- Conversion rule:
--- if `t` is 4, 5, or 6, then "h" is inserted after `b`.
--- if `t` is 1, 2, 4, or 5, then the corresponding accent is put on `a`.
-- (finally, for syllabics, the whole syllabic is `a`, and `b` is empty)
local function yale_tone(a,b,t)
	local h = ""
	if t == "4" or t == "5" or t == "6" then
		h = "h"
	end
	if t == "1" then a = alt(macron, grave) end
	if t == "4" then a = grave end
	if t == "2" or t == "5" then a = acute end
	return a..b..h
end

function export.jyutping_to_ipa(text)
	if type(text) == "table" then text = text.args end
	
	text = text:gsub(", "," "):gsub("%.%.%.", " "):gsub(" $",""):gsub(" / ","/, /")
		-- :gsub("()yu", "%1hyu")
		-- :gsub("()oe", "%1hoe")
		-- :gsub("()eo", "%1heo")
		:gsub("(%l+)(%d)(%-?)(%d?)", function(main,tone,symbol,tone2)
			-- try initial+final
			local initial, final = main:match("^(??)(%l*)$")
			if not initial then
				-- otherwise try initial+syllabic
				local syllabic
				initial, syllabic = main:match("^(h?)(g?)$")
				if not initial then
					error("Invalid Jyutping syllable: " .. main)
				end
				main = ipa_initial
					.. (ipa_syllabic or error("Unrecognised syllabic: " .. syllabic)) -- really?
			else
				-- e.g. convert <ei> (which would be */ɛːi̯/) to <eri> (/ei̯/)
				final = ipa_allophones or final
				local nucleus, coda = final:match("^(y??)(?g?)$")
				if not nucleus then
					error("Invalid Jyutping final: " .. final)
				end
				if entering then
					tone = entering_tones or tone
					tone2 = entering_tones or tone2
				end
				main = (ipa_initial or error(("Unrecognised initial: \"%s\""):format(initial)))
					.. (ipa_nucleus or error(("Unrecognised nucleus: \"%s\""):format(nucleus)))
					.. (ipa_coda or error(("Unrecognised coda: \"%s\""):format(coda)))
			end
			return main .. ipa_tone .. ipa_tone_sandhi .. ipa_tone
		end)
	return text
end

function export.jyutping_to_yale(text)
	if type(text) == "table" then text = text.args end
	
	text = text:gsub("jy?","y")
		:gsub("",{z="j",c="ch"})
		--:gsub("%-","")
		:gsub("(%l+)(%d)(%-?)(%d?)", function(main,tone,symbol,tone2)
			if tone2 ~= "" then
				tone = tone2
			end
			-- find the first vowel letter
			local initial,final = main:match("^(*)(.*)$")
			local a,b,c
			if initial then
				final = yale_final or final
				if final == "?" then
					return "?"
				end
				a,b,c = final:match("^()(*)(?g?)$")
			else -- otherwise it is a syllabic
				initial,a = main:match("(h?)(g?)")
				b,c = "",""
			end
			return initial..yale_tone(a,b,tone)..c
		end)
	
	if text:find("?") then
		return false
	end
	return text
end

function export.jyutping_to_cantonese_pinyin(text)
	if type(text) == "table" then text = text.args end
	
	if text:find("oe") then -- unsupported finals
		return false
	end
	
	text = text:gsub("yu","y")
		:gsub("eo",{eoi="oey",eon="oen",eot="oet"})
		:gsub("",{z="dz",c="ts"})
		:gsub("()()(%-?)(?)",function(a,b,c,d)
			return a .. (entering_tones or b) .. c .. (entering_tones or d)
		end)
	
	return text
end

function export.jyutping_to_guangdong(text)
	if type(text) == 'table' then text = text.args end
	
	-- unsupported finals
	if text:find("%fa%d") or text:find("oe")
			or text:find("") or text:find("e") then
		return false
	end
	
	text = text:gsub("yu","ü")
		:gsub("j","y")
		:gsub("%f",{z="j",c="q",s="x"})  -- ü=\xC3\xBC
		:gsub("()ü","%1u")
		:gsub("eoi","êu")
		:gsub("?",{aa="a",a="e",e="é",oe="ê",eo="ê"})
		:gsub("()u","%1o")
		:gsub("()w","%1u")
		:gsub("%f",{p="b",t="d",k="g"})
	
	return text
end

-- substitute changed tones for finding homophones
function export.jyutping_format(text)
	text = text:gsub("%-()", "%1")
	return split(text," / ")
end

local function Consolas(text)
	return '<span style="font-family: Consolas, monospace;">' .. text .. "</span>"
end

local function format_IPA(text)
	return '<span class="IPA">' .. text .. "</span>"
end

local function make_superscript(text)
	return (text:gsub("%d*%f", "<sup>%0</sup>"))
end

-- the only allowed punctuations are:
--- ", ": represents a comma (or a break of any sort)
--- "...": represents a slot where a text can go to (e.g. ])
--- ",": separates alternate readings
local function validate(c_rom)
	if c_rom:find("") then error("Invalid tone in Jyutping.") end
	if c_rom:find("") then error("Please do not capitalize the Jyutping.") end
	if c_rom:find("%-") then error("Please do not hyphenate the Jyutping.") end
	if c_rom:find("") then error("Error in Jyutping: please use spaces to separate syllables.") end
	if c_rom:find("h") then error("'zh'/'ch'/'sh' are non-valid Jyutping, use 'z'/'c'/'s' instead.") end
	if c_rom:find("y") then error("Wrong usage of 'y' in Jyutping.") end
	if c_rom:find("oei") then error("Invalid rime oei in Jyutping. Did you mean eoi?") end
	if c_rom:find("eong") then error("Invalid rime eong in Jyutping. Did you mean oeng?") end
	if c_rom:find("eok") then error("Invalid rime eok in Jyutping. Did you mean oek?") end
	if c_rom:find("r") then error("Invalid letter \"r\" in Jyutping.") end
	if c_rom:find("%d%d") then error("Invalid Jyutping: please use a hyphen to indicate a changed tone.") end
	
	c_rom = c_rom:gsub("^%.%.%.",""):gsub("%.%.%.%f",""):gsub("%.%.%."," "):gsub(", ?"," ")
	if c_rom:find("^ ") or c_rom:find("  ") or c_rom:find(" $") then
		error("Empty syllable detected.")
	end
	if c_rom:find("") then
		error("Invalid character found.")
	end
	
	-- ensure that each syllable matches `^%l+%d%-?%d?$`
	for syllable in c_rom:gmatch("%S+") do
		if not syllable:match("^%l+%d%-?%d?$") then
			error("Invalid Jyutping syllable: " .. syllable)
		end
	end
end

-- generate the shown text of Standard Cantonese
-- if the pagename is one character long, then generate links to all the readings
function export.show_standard(c_rom, is_single_hanzi)
	validate(c_rom)
	c_rom = c_rom:gsub(",%f"," / ")
	if is_single_hanzi then
		for reading in c_rom:gmatch("+") do
			require('Module:debug').track('yue-pron/'..reading)
		end
		c_rom = c_rom:gsub("(%l+)(%d%-?%d?)","]")
	else
		c_rom = make_superscript(c_rom)
	end
	return c_rom
end

-- generate the collapsed text of Standard Cantonese
-- generate all the different romanisations, as well as homophones
function export.hide_standard(c_rom, is_single_hanzi)
	local res = ""
	c_rom = c_rom:gsub(",%f"," / ")
	
	-- generate IPA first because the error-catching is located there
	local c_ipa = export.jyutping_to_ipa(c_rom)
	local c_yale = export.jyutping_to_yale(c_rom)
	local c_cp = export.jyutping_to_cantonese_pinyin(c_rom)
	local c_gd = export.jyutping_to_guangdong(c_rom)
	
	local c_hom = mw.loadData("Module:yue-pron/hom")
	local c_hom_exists = false
	for _,c_first in ipairs(export.jyutping_format(c_rom)) do
		if c_hom then
			c_hom_exists = c_first
			break
		end
	end
	
	res = res .. "\n** <small>(<i>], ]–]</i>)</small>"
	if not c_hom_exists and not is_single_hanzi then
		res = res .. '<sup><small><abbr title="Add Cantonese homophones"><span class="plainlinks">['
		res = res .. tostring(mw.uri.fullUrl("Module:yue-pron/hom",{="edit"}))
		res = res .. " +]</span></abbr></small></sup>"
	end
	res = res .. "\n*** <small><i>]</i></small>: "
	res = res .. Consolas(make_superscript(c_rom))
	if c_yale then
		res = res .. "\n*** <small><i>]</i></small>: "
		res = res .. Consolas(c_yale)
	end
	if c_cp then
		res = res .. "\n*** <small><i>]</i></small>: "
		res = res .. Consolas(make_superscript(c_cp))
	end
	if c_gd then
		res = res .. "\n*** <small><i>]</i></small>: "
		res = res .. Consolas(make_superscript(c_gd))
	end
	res = res .. "\n*** <small>Sinological ] <sup>(])</sup></small>: "
	res = res .. format_IPA("/" .. c_ipa .. "/")
	if c_hom_exists then
		res = res .. '\n*** <small>Homophones</small>: <table class="wikitable mw-collapsible mw-collapsed" style="width:15em;margin:0;'
		res = res .. 'position:left; text-align:center"><tr><th></th></tr><tr><td><div style="float: right; clear: right;"><sup>'
		res = res .. '<span class="plainlinks">['
		res = res .. tostring(mw.uri.fullUrl("Module:yue-pron/hom",{="edit"}))
		res = res .. ' edit]</span></sup></div><div style="visibility:hidden; float:left"><sup><span style="color:#FFF">edit</span></sup></div>'
		local hom_text = {}
		local yue = require("Module:languages").getByCode("yue")
		for _,hom in ipairs(c_hom) do
			table.insert(hom_text, require("Module:links").full_link( { term = hom, lang = yue, tr = "-" } ))
		end
		res = res .. table.concat(hom_text, "<br>")
		res = res .. "</td></tr></table>"
		res = res .. "]"
	end
	return res
end

function export.jyutping_headword(frame)
	local args = require("Module:parameters").process(frame:getParent().args, {
		 = {},
	})
	
	return require("Module:headword").full_headword{
		lang = require("Module:languages").getByCode("yue"),
		sc = require("Module:scripts").getByCode("Latn"),
		heads = {make_superscript(args.head or mw.loadData("Module:headword/data").pagename)},
		pos_category = "jyutping"
	}
end

return export