Modul:sort

local lang = require('Modul:lang')
local diacritic_objects = mw.loadData("Modul:sort/data-diacritics")

local export = {}

local function excludeSpecificCharsFromString(str)
	--settings for which characters that should always be excluded from the sort key, fetched from Wiktionary:Användare/Robotar/Tecken som exkluderas för sortering 2017-05-12
	local excluded_chars = {" ", "%.", ",", "%&", "!", "%-", "/", ":", "\"", "\'", "%$", "%(", "%)", "%*", "%+", "\\", ";", "=", "%?", "~", "%%"}

	for k,v in pairs(excluded_chars) do
		str = mw.ustring.gsub(str, v, "")
	end

	return str;
end

local function undiacritic(str, language_specific_rules)
	local diacritic_exceptions = {} --don't remove diacritics for letters put in this table
	for k,v in pairs(language_specific_rules) do
	    local replace_pair = mw.text.split(v, ">", true)
		str = mw.ustring.gsub(str, replace_pair, replace_pair)
		table.insert(diacritic_exceptions, replace_pair)
	end

	for k,v in pairs(diacritic_objects) do --i.e. v == {"α", {"ά", "ὰ", "ἀ", "ἅ", "ᾶ", "ἆ", "ἇ"}}
		local char_without_diacritics = v
		local chars_with_diacritics = v
		for k2,v2 in pairs(chars_with_diacritics) do 
			local is_diacritic_exception = false
			for k3,v3 in pairs(diacritic_exceptions) do
		        if v3 == v2 then 
		        	is_diacritic_exception = true
		        end
		    end
		    if is_diacritic_exception ~= true then
				str = mw.ustring.gsub(str, v2, char_without_diacritics)
			end
		end
	end

	return str;
end

local function applyNumberSorting(str)
	return mw.ustring.gsub(str, "0*(%d+)", function(d) return mw.ustring.rep(":", mw.ustring.len(d)) .. d end) --add a ":" for each numeral in each number in front of the number
end

local function getKey(str, language_specific_rules)
	str = excludeSpecificCharsFromString(str)
	str = mw.ustring.lower(str)
	str = undiacritic(str, language_specific_rules)
	str = applyNumberSorting(str)
	
	return str
end

function export.getSortKey(arg1, arg2)
	local lang_code, pagename

	if arg1 == mw.getCurrentFrame() then
		local frame = arg1
		lang_code = frame.args or "xx"
		pagename = frame.args or mw.title.getCurrentTitle().text
	else
		lang_code = arg1 or "xx"
		pagename = arg2 or ""
	end

	local language_specific_rules = lang.getSortRules(lang_code) or {}

	return getKey(pagename, language_specific_rules)
end

return export