Editing Module:Unicode data/derived core properties/make

local p = {}

function p.main(frame)
	local page = frame.args[1] or "User:Erutuon/Unicode/DerivedCoreProperties.txt"
	local text = assert(mw.title.new(page):getContent())
	local default_ignorable = text
		:match("Derived Property: Default_Ignorable_Code_Point.-(%f[^\n]%x%x%x%x.-)%s*\n# Total code points")
	local singles, ranges = {}, {}
	for code_point1, code_point2 in default_ignorable:gmatch("%f[^\n%z](%x+)%.?%.?(%x*)") do
		code_point1, code_point2 = tonumber(code_point1, 16), tonumber(code_point2, 16)
		local last_range = ranges[#ranges]
		if last_range and last_range[2] == code_point1 - 1 then
			last_range[2] = code_point2 or code_point1
		elseif singles[code_point1 - 1] then
			singles[code_point1 - 1] = nil
			table.insert(ranges, { code_point1 - 1, code_point2 or code_point1 })
		else
			if not code_point2 then
				singles[code_point1] = true
			else
				table.insert(ranges, { code_point1, code_point2 })
			end
		end
	end
	
		local template = [[
local data = {}

data.default_ignorable = {
	singles = {
...
	},
	ranges = {
...
	},
}

return data
]]

	local Array = require "Module:array"
	local printed_ranges = Array()
	for _, range in ipairs(ranges) do
		local low, high, script_code = unpack(range)
		printed_ranges:insert(('\t\t{ 0x%05X, 0x%05X },'):format(low, high))
	end
	
	local printed_singles = Array()
	for codepoint in require 'Module:TableTools'.sortedPairs(singles) do
		printed_singles:insert(('\t\t[0x%05X] = true,'):format(codepoint))
	end
	
	local data = template
		:gsub('%.%.%.', printed_singles:concat('\n'), 1)
		:gsub('%.%.%.', printed_ranges:concat('\n'), 1)
	
	return mw.getCurrentFrame():extensionTag{
		name = "syntaxhighlight",
		content = data,
		args = { lang = "lua" }
	}
end

return p