Module:Th-rm

Module documentation

[view] [edit] [history] [purge]

Main

Documentation

Usage

This module implements {{th-rm}}.

Text above can be found here (edit)
Editors can experiment in this module's draft (create |  mirror) page.
Please add categories to the documentation subpage. Subpages of this module.

local p = {}

local arguments = require('Module:Arguments')
local tt = require('Module:Tt')

local AUTO_TOOLTIP = 'This romanization was generated automatically. Please add a user-curated romanization as soon as possible.'

local THAI = '[ก-๛]'

local consonants = {
	['ก'] = 'k',
	['ข'] = 'kh',
	['ฃ'] = 'kh',
	['ค'] = 'kh',
	['ฅ'] = 'kh',
	['ฆ'] = 'kh',
	['ง'] = 'ng',
	['จ'] = 'ch',
	['ฉ'] = 'ch',
	['ช'] = 'ch',
	['ซ'] = 's',
	['ฌ'] = 'ch',
	['ญ'] = 'y',
	['ฎ'] = 'd',
	['ฏ'] = 't',
	['ฐ'] = 'th',
	['ฑ'] = 'th',
	['ฒ'] = 'th',
	['ณ'] = 'n',
	['ด'] = 'd',
	['ต'] = 't',
	['ถ'] = 'th',
	['ท'] = 'th',
	['ธ'] = 'th',
	['น'] = 'n',
	['บ'] = 'b',
	['ป'] = 'p',
	['ผ'] = 'ph',
	['ฝ'] = 'f',
	['พ'] = 'ph',
	['ฟ'] = 'f',
	['ภ'] = 'ph',
	['ม'] = 'm',
	['ย'] = 'y',
	['ร'] = 'r',
	['ล'] = 'l',
	['ว'] = 'w',
	['ศ'] = 's',
	['ษ'] = 's',
	['ส'] = 's',
	['ห'] = 'h',
	['ฬ'] = 'l',
	['ฮ'] = 'h',
	['อ'] = ''
}

local finals = {
	['ก'] = 'k',
	['ข'] = 'k',
	['ฃ'] = 'k',
	['ค'] = 'k',
	['ฅ'] = 'k',
	['ฆ'] = 'k',

	['จ'] = 't',
	['ฉ'] = 't',
	['ช'] = 't',
	['ซ'] = 't',
	['ฌ'] = 't',
	['ฎ'] = 't',
	['ฏ'] = 't',
	['ฐ'] = 't',
	['ฑ'] = 't',
	['ฒ'] = 't',
	['ด'] = 't',
	['ต'] = 't',
	['ถ'] = 't',
	['ท'] = 't',
	['ธ'] = 't',
	['ศ'] = 't',
	['ษ'] = 't',
	['ส'] = 't',

	['บ'] = 'p',
	['ป'] = 'p',
	['ผ'] = 'p',
	['ฝ'] = 'p',
	['พ'] = 'p',
	['ฟ'] = 'p',
	['ภ'] = 'p',

	['ง'] = 'ng',

	['ญ'] = 'n',
	['ณ'] = 'n',
	['น'] = 'n',
	['ร'] = 'n',
	['ล'] = 'n',
	['ฬ'] = 'n',

	['ม'] = 'm'
}

local vowels = {
	-- Longest forms first.
	['เียะ'] = 'ia',
	['เีย'] = 'ia',
	['เือะ'] = 'uea',
	['เือ'] = 'uea',
	['เือย'] = 'ueai',
	['เียว'] = 'iao',

	['ัวะ'] = 'ua',
	['ัว'] = 'ua',

	['เิ็ย'] = 'oei',
	['เิย'] = 'oei',
	['เย'] = 'oei',
	['เอย'] = 'oei',

	['็อย'] = 'oi',
	['อย'] = 'oi',
	['โย'] = 'oi',

	['ุย'] = 'ui',
	['วย'] = 'uai',

	['ัย'] = 'ai',
	['ไย'] = 'ai',
	['าย'] = 'ai',
	['ใ'] = 'ai',
	['ไ'] = 'ai',

	['เ็ว'] = 'eo',
	['แ็ว'] = 'aeo',
	['เา'] = 'ao',
	['าว'] = 'ao',
	['เอว'] = 'oeu',
	['แว'] = 'aeo',
	['เว'] = 'eo',
	['โว'] = 'ou',

	['ิว'] = 'io',
	['ีว'] = 'io',

	['ำ'] = 'am',

	['ือ'] = 'ue',
	['ื'] = 'ue',

	['ึ'] = 'ue',

	['ิ'] = 'i',
	['ี'] = 'i',

	['ุ'] = 'u',
	['ู'] = 'u',

	['เ็'] = 'e',
	['เ'] = 'e',

	['แ็'] = 'ae',
	['แ'] = 'ae',

	['โะ'] = 'o',
	['โ'] = 'o',

	['เาะ'] = 'o',
	['อ'] = 'o',

	['เอะ'] = 'oe',
	['เอ'] = 'oe',
	['เะ'] = 'e',

	['ะ'] = 'a',
	['ั'] = 'a',
	['า'] = 'a'
}

local vowel_order = {
	'เือย',
	'เียว',
	'เือะ',
	'เียะ',
	'เีย',
	'เือ',
	'ัวะ',
	'ัว',
	'เิ็ย',
	'เิย',
	'็อย',
	'เอว',
	'แ็ว',
	'เ็ว',
	'เอย',
	'เา',
	'าว',
	'เว',
	'แว',
	'โว',
	'าย',
	'ไย',
	'ัย',
	'ุย',
	'วย',
	'อย',
	'โย',
	'เย',
	'ำ',
	'ิ',
	'ี',
	'ึ',
	'ื',
	'ุ',
	'ู',
	'เ็',
	'แ็',
	'เาะ',
	'เอะ',
	'เะ',
	'ะ',
	'ั',
	'า',
	'เ',
	'แ',
	'โ',
	'อ',
	'ใ',
	'ไ'
}

local vowels_with_front = {
	['เ'] = true,
	['แ'] = true,
	['โ'] = true,
	['ใ'] = true,
	['ไ'] = true
}

local consonant_set = {}
for c in pairs(consonants) do
	consonant_set[c] = true
end

local valid_clusters = {
	['กร'] = true,
	['กล'] = true,
	['กว'] = true,
	['ขร'] = true,
	['ขล'] = true,
	['ขว'] = true,
	['คร'] = true,
	['คล'] = true,
	['คว'] = true,
	['ฆร'] = true,
	['ฆล'] = true,
	['จร'] = true,
	['ตร'] = true,
	['ปร'] = true,
	['ปล'] = true,
	['พร'] = true,
	['พล'] = true,
	['ผล'] = true,
	['ทร'] = true,
	['ธร'] = true,
	['ศร'] = true,
	['สร'] = true,
	['สล'] = true,
	['สว'] = true,
	['ซร'] = true,
	['ฝร'] = true,
	['ฟร'] = true,
	['หน'] = true,
	['หม'] = true,
	['หย'] = true,
	['หร'] = true,
	['หล'] = true,
	['หว'] = true,
	['หญ'] = true
}

local h_silent_clusters = {
	['หง'] = true,
	['หน'] = true,
	['หม'] = true,
	['หย'] = true,
	['หร'] = true,
	['หล'] = true,
	['หว'] = true,
	['หญ'] = true
}

local r_silent_clusters = {
	['จร'] = true,
	['ซร'] = true,
	['ศร'] = true,
	['สร'] = true
}

local function chars(text)
	local result = {}

	for c in mw.text.gsplit(text, '') do
		table.insert(result, c)
	end

	return result
end

local function isConsonant(c)
	return c and consonant_set[c] == true
end

local function isThai(c)
	return c and mw.ustring.find(c, THAI) ~= nil
end

local function isTone(c)
	return c == '่' or c == '้' or c == '๊' or c == '๋' or c == '์'
end

local function isFrontVowel(c)
	return c and vowels_with_front[c] == true
end

local function getVowel(chars_, pos)
	local remaining = table.concat(chars_, '', pos)

	for _, form in ipairs(vowel_order) do
		if mw.ustring.sub(remaining, 1, mw.ustring.len(form)) == form then
			return form, vowels[form]
		end
	end

	return nil, nil
end

local function skipMarks(chars_, pos)
	while chars_[pos] and (
		chars_[pos] == '็' or
		chars_[pos] == '่' or
		chars_[pos] == '้' or
		chars_[pos] == '๊' or
		chars_[pos] == '๋' or
		chars_[pos] == '์' or
		chars_[pos] == 'ํ' or
		chars_[pos] == '๎'
	) do
		pos = pos + 1
	end

	return pos
end

local function findNextVowel(chars_, pos)
	local i = pos

	while chars_[i] do
		if isFrontVowel(chars_[i]) then
			return i
		end

		for _, form in ipairs(vowel_order) do
			if mw.ustring.sub(table.concat(chars_, '', i), 1, mw.ustring.len(form)) == form then
				return i
			end
		end

		if chars_[i] == '์' then
			i = i + 2
		else
			i = i + 1
		end
	end

	return nil
end

local function hasVowelAt(chars_, pos)
	local form = getVowel(chars_, pos)

	return form ~= nil
end

local function getInitial(chars_, pos)
	local c1 = chars_[pos]

	if not isConsonant(c1) then
		return '', pos
	end

	local c2 = chars_[pos + 1]

	if c1 == 'ห' and c2 and h_silent_clusters[c1 .. c2] then
		return consonants[c2], pos + 2, true, c1 .. c2
	end

	if c2 and valid_clusters[c1 .. c2] then
		if c1 == 'ท' and c2 == 'ร' then
			return 's', pos + 2, false, c1 .. c2
		end

		if r_silent_clusters[c1 .. c2] then
			return consonants[c1], pos + 2, false, c1 .. c2
		end

		return consonants[c1] .. consonants[c2], pos + 2, false, c1 .. c2
	end

	return consonants[c1], pos + 1, false, c1
end

local function getCoda(chars_, pos)
	local c = chars_[pos]

	if not c or not isConsonant(c) then
		return '', pos
	end

	if c == 'อ' or c == 'ห' or c == 'ฮ' then
		return '', pos
	end

	if c == 'ย' or c == 'ว' then
		return '', pos
	end

	if finals[c] then
		return finals[c], pos + 1
	end

	return '', pos + 1
end

local function removeSilent(chars_)
	local result = {}
	local i = 1

	while chars_[i] do
		if chars_[i] == '์' then
			i = i + 1

			while chars_[i] and (
				isConsonant(chars_[i]) or
				chars_[i] == '็' or
				chars_[i] == '่' or
				chars_[i] == '้' or
				chars_[i] == '๊' or
				chars_[i] == '๋'
			) do
				i = i + 1
			end
		else
			table.insert(result, chars_[i])
			i = i + 1
		end
	end

	return result
end

local function cleanWord(word)
	word = mw.ustring.gsub(word, '์', '')

	return word
end

local function parseSyllable(chars_, start)
	local n = #chars_
	local pos = start

	if not chars_[pos] then
		return nil, pos
	end

	local front = ''

	if isFrontVowel(chars_[pos]) then
		front = chars_[pos]
		pos = pos + 1
	end

	local initial, afterInitial, silentH, cluster =
		getInitial(chars_, pos)

	if not chars_[pos] or not isConsonant(chars_[pos]) then
		return nil, start + 1
	end

	pos = afterInitial

	local vowelForm
	local vowel

	if front ~= '' then
		local rest = front

		for _, form in ipairs(vowel_order) do
			local candidate = front .. mw.ustring.sub(
				table.concat(chars_, '', pos),
				1,
				mw.ustring.len(form)
			)

			if vowels[candidate] then
				vowelForm = candidate
				vowel = vowels[candidate]
				pos = pos + mw.ustring.len(form)
				break
			end
		end

		if not vowel then
			if front == 'เ' then
				if chars_[pos] == 'า' then
					vowelForm = 'เา'
					vowel = 'ao'
					pos = pos + 1
				else
					vowelForm = 'เ'
					vowel = 'e'
				end
			elseif front == 'แ' then
				vowelForm = 'แ'
				vowel = 'ae'
			elseif front == 'โ' then
				vowelForm = 'โ'
				vowel = 'o'
			elseif front == 'ใ' or front == 'ไ' then
				vowelForm = front
				vowel = 'ai'
			end
		end
	else
		vowelForm, vowel = getVowel(chars_, pos)

		if vowelForm then
			pos = pos + mw.ustring.len(vowelForm)
		end
	end

	-- No written vowel: determine an implied vowel.
	if not vowel then
		local nextC = chars_[pos]

		if nextC and isConsonant(nextC) then
			local nextNext = chars_[pos + 1]

			if nextNext and (
				isFrontVowel(nextNext) or
				hasVowelAt(chars_, pos + 1)
			) then
				-- Open syllable with an implied a.
				vowelForm = ''
				vowel = 'a'
			else
				-- Closed syllable with an implied o.
				vowelForm = ''
				vowel = 'o'
			end
		else
			vowelForm = ''
			vowel = 'a'
		end
	end

	pos = skipMarks(chars_, pos)

	local coda = ''
	local codaPos = pos

	if chars_[pos] and isConsonant(chars_[pos]) then
		local nextC = chars_[pos]
		local nextNext = chars_[pos + 1]

		-- If the following consonant is followed by a vowel,
		-- it normally starts the next syllable rather than being a coda.
		if nextNext and (
			isFrontVowel(nextNext) or
			hasVowelAt(chars_, pos + 1)
		) then
			coda = ''
		else
			coda, codaPos = getCoda(chars_, pos)
		end
	end

	pos = codaPos
	pos = skipMarks(chars_, pos)

	return {
		initial = initial,
		vowel = vowel,
		vowelForm = vowelForm or '',
		coda = coda,
		next = pos,
		cluster = cluster,
		silentH = silentH
	}, pos
end

local function shouldHyphen(previous, current)
	if not previous or not current then
		return false
	end

	-- A hyphen is used before a syllable beginning with a vowel.
	if current.initial == '' and current.vowel ~= '' then
		return true
	end

	-- A syllable ending in a vowel followed by initial ng.
	if previous.coda == '' and current.initial == 'ng' then
		return true
	end

	return false
end

local function romanizeWord(word)
	word = cleanWord(word)

	local c = chars(word)
	c = removeSilent(c)

	local syllables = {}
	local pos = 1
	local safety = 0

	while pos <= #c and safety < 1000 do
		safety = safety + 1

		if not c[pos] then
			break
		end

		if c[pos] == 'ๆ' then
			table.insert(syllables, {
				raw = 'ๆ',
				initial = '',
				vowel = '',
				coda = ''
			})
			pos = pos + 1
		elseif isConsonant(c[pos]) or isFrontVowel(c[pos]) then
			local syllable, nextPos = parseSyllable(c, pos)

			if syllable then
				table.insert(syllables, syllable)

				if nextPos <= pos then
					pos = pos + 1
				else
					pos = nextPos
				end
			else
				-- Unknown Thai structure: consume one character so that
				-- the parser can never become stuck.
				table.insert(syllables, {
					raw = c[pos],
					initial = consonants[c[pos]] or '',
					vowel = 'a',
					coda = ''
				})

				pos = pos + 1
			end
		else
			pos = pos + 1
		end
	end

	local output = {}

	for i, s in ipairs(syllables) do
		if s.raw == 'ๆ' then
			table.insert(output, 'ๆ')
		else
			local result = (s.initial or '') .. (s.vowel or '') .. (s.coda or '')

			if result == '' then
				result = 'a'
			end

			if shouldHyphen(syllables[i - 1], s) then
				table.insert(output, '-')
			end

			table.insert(output, result)
		end
	end

	return table.concat(output)
end

local function romanizeThaiText(text)
	local result = {}

	local parts = mw.text.split(
		text,
		'(\1MANUALHYPHEN\1|\1MANUALSPACE\1)',
		true
	)

	for _, part in ipairs(parts) do
		if part == '\1MANUALHYPHEN\1' then
			table.insert(result, '-')
		elseif part == '\1MANUALSPACE\1' then
			table.insert(result, ' ')
		else
			local last = 1

			while true do
				local startPos, endPos =
					mw.ustring.find(part, THAI .. '+', last)

				if not startPos then
					table.insert(result, mw.ustring.sub(part, last))
					break
				end

				if startPos > last then
					table.insert(
						result,
						mw.ustring.sub(part, last, startPos - 1)
					)
				end

				local word =
					mw.ustring.sub(part, startPos, endPos)

				table.insert(result, romanizeWord(word))

				last = endPos + 1
			end
		end
	end

	return table.concat(result)
end

local function hasThai(text)
	return mw.ustring.find(text, THAI) ~= nil
end

local function containsThai(text)
	return mw.ustring.find(text, '[ก-๛]') ~= nil
end

local function hasOnlyThaiOrFormatting(text)
	return not mw.ustring.find(text, '[ก-๛]')
end

local function validateRomanization(text)
	-- The automatic result must never leak Thai characters.
	if containsThai(text) then
		return false
	end

	-- Thai marks should never survive either.
	if mw.ustring.find(text, '[ะ-๛]') then
		return false
	end

	return true
end

local function capitalizeWords(text)
	return mw.ustring.gsub(
		text,
		'(%a)([%a]*)',
		function(first, rest)
			return mw.ustring.upper(first) .. rest
		end
	)
end

local function capitalizeSentence(text)
	local first = mw.ustring.match(text, '^[%s%p]*([a-z])')

	if not first then
		return text
	end

	local upper = mw.ustring.upper(first)

	return mw.ustring.gsub(
		text,
		'^([%s%p]*)' .. first,
		'%1' .. upper,
		1
	)
end

local function convertThaiDigits(text)
	local digits = {
		['๐'] = '0',
		['๑'] = '1',
		['๒'] = '2',
		['๓'] = '3',
		['๔'] = '4',
		['๕'] = '5',
		['๖'] = '6',
		['๗'] = '7',
		['๘'] = '8',
		['๙'] = '9'
	}

	for thai, latin in pairs(digits) do
		text = mw.ustring.gsub(text, thai, latin)
	end

	return text
end

function p.romanize(text)
	if not text or mw.text.trim(text) == '' then
		return ''
	end

	if not hasThai(text) then
		return ''
	end

	-- Manual spacing/hyphen controls.
	text = text
		:gsub('<->', '\1')
		:gsub('<>', '\2')

	text = convertThaiDigits(text)

	local result = romanizeThaiText(text)

	-- Restore manual separators.
	result = result
		:gsub('\1', '-')
		:gsub('\2', ' ')

	result = mw.ustring.gsub(result, '%s+', ' ')
	result = mw.text.trim(result)

	if not validateRomanization(result) then
		return ''
	end

	return result
end

function p.main(frame)
	local args = arguments.getArgs(frame, {
		parentFirst = true,
		wrappers = {'Template:Th-rm'}
	})

	local text = args[1]

	if not text or mw.text.trim(text) == '' then
		return ''
	end

	-- If there is no Thai at all, this module should not create
	-- an automatic romanization.
	if not hasThai(text) then
		return ''
	end

	local result = p.romanize(text)

	if result == '' then
		return ''
	end

	if args.capi then
		result = capitalizeWords(result)
	end

	if args.sent then
		result = capitalizeSentence(result)
	end

	-- Automatically generated romanizations are deliberately marked
	-- with the tooltip from Module:Tt.
	return tt._main({
		[1] = result,
		[2] = AUTO_TOOLTIP,
		['class'] = args.class
	})
end

return p