Moduli:Unicode data/category/make

local p = {}

local function read_page(title)
	return mw.title.new(title):getContent()
end

function p.make_singles_and_ranges(derived_general_category_page, property_value_aliases_page)
	local text = read_page(derived_general_category_page)
	
	local singles, ranges = {}, {}
	for code_point1, code_point2, general_category in text:gmatch "%f[^\n%z](%x+)%.?%.?(%x*)%s*;%s*(%u%l)" do
		if general_category ~= "Cn" then
			code_point1 = tonumber(code_point1, 16)
			
			-- XXXX..XXXX ; gc
			if code_point2 ~= "" then
				code_point2 = tonumber(code_point2, 16)
				table.insert(ranges, { code_point1, code_point2, general_category })
			
			-- XXXX ; gc
			else
				singles[code_point1] = general_category
			end
		end
	end
	
	local property_value_aliases = read_page(property_value_aliases_page)
	local general_category_aliases = property_value_aliases:match "# General_Category[^\n]+%s*(.-)%s*%f[^\n]#"
	local long_names = {}
	for abbr, long_name in general_category_aliases:gmatch "gc%s*;%s*(%S+)%s*;%s*([%a_]+)" do
		long_names[abbr] = long_name
	end
	
	return singles, ranges, long_names
end

function p.print_data(data)
	local output = require "Module:array"()
	local function writef(...)
		output:insert(string.format(...))
	end
	
	writef [[
return {
	singles = {
]]
	
	-- Check that maximum "singles" codepoint is less than 0x100000?
	for codepoint, category in require "Module:TableTools".sortedPairs(data.singles) do
		writef('\t\t[0x%05X] = "%s",\n', codepoint, category)
	end
	
	writef [[
	},
	ranges = {
]]
	
	local function compare_ranges(range1, range2)
		return range1[1] < range2[1]
	end
	
	table.sort(data.ranges, compare_ranges)
	for _, range in ipairs(data.ranges) do
		writef('\t\t{ 0x%06X, 0x%06X, "%s" },\n', unpack(range))
	end
	
	writef [[
	},
	long_names = {
]]
	for abbr, long_name in require "Module:TableTools".sortedPairs(data.long_names) do
		writef('\t\t%-2s = "%s",\n', abbr, long_name)
	end
	
	writef [[
	},
}]]

	return output:concat()
end

function p.main(frame)
	local data = {}
	data.singles, data.ranges, data.long_names = p.make_singles_and_ranges(
		"User:Erutuon/Unicode/DerivedGeneralCategory.txt",
		"User:Erutuon/Unicode/PropertyValueAliases.txt")
	return frame:extensionTag{
		name = "syntaxhighlight",
		content = p.print_data(data),
		args = { lang = "lua" }
	}
end

return p