Module:Hangul

local p = {}

local data = mw.loadData 'Module:Hangul/data'

---- From [[wikt:Module:ko-hangul]
 
-- Given the "syllable index" of a precomposed Hangul syllable (see
-- above), returns "indices" representing the three constituent jamo
-- ("lead", i.e. initial consonant; "vowel"; and "tail", i.e. final
-- consonant, except that zero denotes the absence of a final consonant).
local function syllableIndex2JamoIndices(syllableIndex)
    local lIndex = math.floor(syllableIndex / 588)
    local vIndex = math.floor((syllableIndex % 588) / 28)
    local tIndex = syllableIndex % 28
 
    return lIndex, vIndex, tIndex
end

---- From [[wikt:Module:Unicode data]].

local tocodepoint = mw.ustring.codepoint
-- Cheaper to derive names from decomposed form of syllable?
local function get_name(syllable)
	local li, vi, ti = syllableIndex2JamoIndices(tocodepoint(syllable) - 0xAC00)

	return ("HANGUL SYLLABLE %s%s%s"):format(
		data.leads[li], data.vowels[vi], data.trails[ti])
end

----

local function tag(text)
	return '<span lang="ko">' .. text .. '</span>'
end

function p.show(frame)
	local initial = frame.args[1] or 'ᄀ'
	local codepoint = mw.ustring.codepoint(initial)
	if not (0x1100 <= codepoint and codepoint <= 0x1112) then
		error('Incorrect initial ' .. initial .. '. Should be between U+1100 and U+1112.')
	end
	
	local output = {}
	local i = 0
	function output.add(text)
		i = i + 1
		output[i] = text
	end
	function output.row()
		output.add('|-\n')
	end
	
	output.add(
([[
{| class="wikitable collapsible collapsed" style="width: 96px; height: 96px;"
! colspan="29" | Initial&nbsp;%s
|-
! Final→<br>Medial↓]]):format(initial))

	for _, final in ipairs(data.finals) do
		output.add(('! %s'):format(final))
	end
	
	for _, medial in ipairs(data.medials) do
		output.row()
		output.add(('! scope="row" | %s'):format(medial))
		for _, final in ipairs(data.finals) do
			output.add(('| %s%s%s'):format(initial, medial, final))
		end
	end
	
	output.add('|}')
	
	output = table.concat(output, '\n')
	output = mw.ustring.toNFC(output)
	output = mw.ustring.gsub( -- Add names of syllable codepoints.
		output,
		'[가-힣]', -- [[Hangul Syllables]] block (U+AC00-D7AF)
		function (syllable)
			return ('title="%s" | %s'):format(get_name(syllable), syllable)
		end)
	
	-- Tag all Korean characters.
	-- The pattern matches the UTF-8 encoding of the codepoint ranges
	-- U+1000-1FFF, U+3000-3FFF, and U+A000-DFFF.
	-- (See [[UTF-8#Codepage layout]].)
	output = output:gsub('[\225\227\234-\237][\128-\191]+', tag)
	
	-- Check for consecutive span tags.
	-- output:gsub('<span[^>]+>[^<]*</span><span[^>]+>[^<]*</span>', mw.log)
	
	return output
end

return p
Usage
