Editing Module:Unicode data/age/make

local p = {}

function p.main(frame)
	local page = frame.args[1] or "User:Erutuon/Unicode/DerivedAge.txt"
	local text = assert(mw.title.new(page):getContent())
	local singles, ranges = {}, {}
	for code_point1, code_point2, age in text:gmatch("%f[^\n%z](%x+)%.?%.?(%x*)%s*;%s*([%d.]+)") 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 and last_range[3] == age then
			last_range[2] = code_point2 or code_point1
		elseif singles[code_point1 - 1] == age then
			singles[code_point1 - 1] = nil
			table.insert(ranges, { code_point1 - 1, code_point2 or code_point1, age })
		else
			if not code_point2 then
				singles[code_point1] = age
			else
				table.insert(ranges, { code_point1, code_point2, age })
			end
		end
	end
	
		local template = [[
return {
	singles = {
...
	},
	ranges = {
...
	},
	aliases = {
...
	},
}
]]

	table.sort(
		ranges,
		function(range1, range2)
			return range1[1] < range2[1]
		end)
	
	local function make_padding(age)
		return (" "):rep(#'"12.1"' - (#age + 2))
	end
	
	local Array = require "Module:array"
	local printed_ranges = Array()
	for _, range in ipairs(ranges) do
		local low, high, age = unpack(range)
		printed_ranges:insert(('\t\t{ 0x%06X, 0x%06X, %s%q },')
			:format(low, high, make_padding(age), age))
	end
	
	local printed_singles = Array()
	for codepoint, age in require 'Module:TableTools'.sortedPairs(singles) do
		printed_singles:insert(('\t\t[0x%06X] = %s%q,')
			:format(codepoint, make_padding(age), age))
	end
	
	local printed_aliases = Array()
	local property_value_aliases = mw.title.new "User:Erutuon/Unicode/PropertyValueAliases.txt":getContent()
	local age_aliases = property_value_aliases:match "# Age[^\n]+%s*(.-)%s*%f[^\n]#"
	for age, alias in age_aliases:gmatch "age%s*;%s*(%S+)%s*;%s*(%S+)" do
		printed_aliases:insert(('\t\t[%s"%s"] = "%s",')
			:format(make_padding(age), age, alias))
	end
	
	local data = template
		:gsub('%.%.%.', printed_singles:concat('\n'), 1)
		:gsub('%.%.%.', printed_ranges:concat('\n'), 1)
		:gsub('%.%.%.', printed_aliases:concat('\n'), 1)
	
	return mw.getCurrentFrame():extensionTag{
		name = "syntaxhighlight",
		content = data,
		args = { lang = "lua" }
	}
end

return p