Module:ko-appendix-junggan-nogeoldae

The following documentation is located at Module:ko-appendix-junggan-nogeoldae/documentation.
Useful links: subpage list • links • transclusions • testcases • sandbox
Generates the table in Appendix:Early Mandarin transcription using Old Hangeul in Junggan Nogeoldae Eonhae.
local export = {}

local memoize = require("Module:memoize")

local function string_rep(str, n, sep)
	assert(n >= 0)
	local list = {}
	for i = 1, n do
		list = str
	end
	return table.concat(list, sep)
end

local function tag(sc, lang_code)
	return function(text)
		return '<span class="' .. sc .. '" lang="' .. lang_code .. '">'
			.. text .. '</span>'
	end
end

local function link_and_tag(sc, lang_code, lang_name)
	local tagger = tag(sc, lang_code)
	return function(term)
		return tagger(']')
	end
end

local tag_Korean = tag("Hang", "ko")
local link_Korean = link_and_tag("Hang", "ko", "Korean")
local tag_Middle_Korean = tag("Hang", "okm")
local link_Chinese = link_and_tag("Hani", "zh", "Chinese")
local link_Mandarin = link_and_tag("Latn", "cmn", "Mandarin")

local function format_Han(traditional_and_simplified)
	if traditional_and_simplified:find("(", 1, true) then
		return ((traditional_and_simplified):gsub(
			"(+)%s*%((+)%)",
			function(traditional, simplified)
				return link_Chinese(traditional) .. "(" .. link_Chinese(simplified) .. ")"
			end))
	else
		local traditional = traditional_and_simplified
		return link_Chinese(traditional)
	end
end

local conversions = {
	 = "k",
	 = "kk",
	 = "n",
	 = "t",
	 = "tt",
	 = "l",
	 = "m",
	 = "p",
	 = "pp",
	 = "s",
	 = "ss",
	 = "",
	 = "c",
	 = "cc",
	 = "ch",
	 = "kh",
	 = "th",
	 = "ph",
	 = "h",
	 = "W",
	 = "f",
	 = "v",
	 = "z",
	 = "ng",
	 = "hh",
	 = "q",
	 = "a",
	 = "ay",
	 = "ya",
	 = "yay",
	 = "e",
	 = "ey",
	 = "ye",
	 = "yey",
	 = "wo",
	 = "wa",
	 = "way",
	 = "woy",
	 = "yo",
	 = "wu",
	 = "we",
	 = "wuy",
	 = "yu",
	 = "u",
	 = "uy",
	 = "i",
	 = "awo",
	 = "yawo",
	 = "yoya",
	 = "yue",
	 = "yuye",
	 = "yuy",
	 = "uwu",
	 = "iwu",
	 = "iu",
	 = "n",
	 = "l",
	 = "m",
	 = "ng",
	 = "W",
	 = "f",
	 = "z",
	 = "q",
}
local vowels = require "Module:array".keys {
	 = "a",
	 = "ay",
	 = "ya",
	 = "yay",
	 = "e",
	 = "ey",
	 = "ye",
	 = "yey",
	 = "wo",
	 = "wa",
	 = "way",
	 = "woy",
	 = "yo",
	 = "wu",
	 = "we",
	 = "wuy",
	 = "yu",
	 = "u",
	 = "uy",
	 = "i",
	 = "awo",
	 = "yawo",
	 = "yoya",
	 = "yue",
	 = "yuye",
	 = "yuy",
	 = "uwu",
	 = "iwu",
	 = "iu",
}:sort():concat()
local vowel = ""
local Yale_romanization = memoize(function(syllable)
	return (mw.ustring.gsub(syllable, ".", conversions))
end)

local to_full_letter = {
   = "ㄱ",
   = "ㄲ",
   = "ㄴ",
   = "ㄷ",
   = "ㄸ",
   = "ㄹ",
   = "ㅁ",
   = "ㅂ",
   = "ㅃ",
   = "ㅅ",
   = "ㅆ",
   = "ㅇ",
   = "ㅈ",
   = "ㅉ",
   = "ㅊ",
   = "ㅋ",
   = "ㅌ",
   = "ㅍ",
   = "ㅎ",
   = "ㅱ",
   = "ㅸ",
   = "ㅹ",
   = "ㅿ",
   = "ㆁ",
   = "ㆅ",
   = "ㆆ",
}

local function format_initial_or_final(Korean, link)
	local format_Korean = link and link_Korean or tag_Korean
	local romanization = Yale_romanization(Korean)
	local displayed_Korean = to_full_letter or Korean
	if romanization == "" then
		romanization = "-"
	end
	return romanization .. " (" .. format_Korean(displayed_Korean) .. ")"
end

local split_syllable = memoize(function(syllable)
	local initial, final = mw.ustring.match(syllable, "^(.)(.*)$")
	if mw.ustring.find(final, "^" .. vowel) then
		final = "ᄋ" .. final
	end
	return { initial = initial, final = final }
end)

local format_transcription_romanization_initial_final = memoize(function (tr)
	local rom = Yale_romanization(tr)
	local split = split_syllable(tr)
	return ('| %s || %s || %s || %s'):format(
		tag_Middle_Korean(tr),
		rom,
		format_initial_or_final(split.initial, true),
		format_initial_or_final(split.final, false))
end)

local Revised_Romanization = memoize(function(Korean)
	return require "Module:ko-translit".tr(mw.ustring.toNFC(Korean))
end)

local format_modern = memoize(function(Korean)
	return Korean:gsub("+", function(Korean)
		return link_Korean(Korean) .. " (" .. Revised_Romanization(Korean) .. ")"
	end)
end)

-- for ]
-- format:
--ā	啊	아	a	- (ㅇ)	a (아)	ᅙᅥ	qe	q (ㆆ)	e (어)	아 (a)	6
function export.main(frame)
	local text_format = mw.ustring.toNFD(frame.args)
	local Array = require "Module:array"
	local output = Array()
	local function insert(val)
		output:insert(val)
	end
	
	local soft_hyphen = "&shy;"
	local hyphenate = {}
	for _, word in ipairs {
		"Fi-nal", "Fre-quen-cy", "I-ni-tial", "Ko-re-an", "Mo-dern", "Pin-yin",
		"Ro-ma-ni-za-tion", "Tran-scrip-tion"
	} do
		hyphenate = word:gsub("%-", soft_hyphen)
	end
	
	insert(([[
{| class="wikitable sortable"
! Pinyin !! Han character !! Transcription 1 !! Romanization 1 !! Initial 1 !! Final 1 !! Transcription 2 !! Romanization 2 !! Initial 2 !! Final 2 !! Modern Sino-Korean !! Frequency]]):gsub("%a+", hyphenate))

	local pattern = string_rep("(+)", 6, "\t")
	for line in text_format:gmatch "+" do
		local pinyin, han, tr1, tr2, mod, freq =
			line:match(pattern)
		if not pinyin then
			error("The line " .. line .. " did not match the pattern " .. pattern)
		end
		insert(([[
|-
| %s || %s]]):format(link_Mandarin(pinyin), format_Han(han)))
		insert(format_transcription_romanization_initial_final(tr1))
		
		insert(format_transcription_romanization_initial_final(tr2))
		insert(([[
| %s || %d]]):format(format_modern(mod), freq))
	end
	insert "|}"
	
	return output:concat "\n"
end

return export
Module:ko-appendix-junggan-nogeoldae

Wikious

Boobota

Sagapedia
