Module:User:Suzukaze-c/zh-extract

local export = {}
local replace = mw.ustring.gsub
local match = mw.ustring.match
local itermatch = mw.ustring.gmatch
local split = mw.text.split

-- idea: +simplified? since we're pulling stuff from the page anyway

local default_set_separator = '//'

function export.extract_roman(word, combine, set_separator)
	local plaintext = false
	if type(word) == 'table' then
		plaintext = true
		word, combine, set_separator = word.args, word.args, word.args
	end

	mw.log('PROCESSING: ]')

	local content = mw.title.new(word):getContent() or error('the ] entry does not exist!?')
	local each = {}
	local roman_final = {}

	content = replace(content, "{{zh%-pron", "ⓐⓐⓐⓐⓐ")
	content = replace(content, "(|cat=*)\n?}}\n", "%1ⓩⓩⓩⓩⓩ") -- making assumptions about formatting

	if match(content, "ⓐ") and not match(content, "ⓩ") then
		error("please add the cat param to zh-pron at ]")
	end

	-- Convert each {{zh-pron}} instance to a table subsumed in $each
	local box_i = 1
	for innards in itermatch(content, "ⓐⓐⓐⓐⓐ(+)ⓩⓩⓩⓩⓩ") do
		each = {}

		innards = split(innards, "\n|")
		table.remove(innards, 1)
		for i, item in ipairs(innards) do
			local param, value = match(item, "^(+)=(.*)$")
			each = value
		end

		box_i = box_i + 1
	end

	-- If told to combine tables, then combine each $each sub-table into a mega-table,
	-- otherwise return the data of the first {{zh-pron}} instance
	if combine then
		-- make $roman_final a table containing every possible $value
		for i, etable in ipairs(each) do
			for param, value in pairs(etable) do
				if not roman_final then roman_final = {} end
				if value ~= '' then table.insert(roman_final, value) end
			end
		end

		-- flatten $roman_final into text
		for param, value in pairs(roman_final) do
			roman_final = table.concat(roman_final, (set_separator or default_set_separator))
		end
	else
		roman_final = each
	end

	if plaintext then
		return require('module:debug').dump(roman_final)
	else
		return roman_final
	end
end

return export