Editing Module:Sandbox/S-1-5-7/versiontests

local module = {};     --All lua modules on Wikipedia must begin by defining a variable 
                    --that will hold their externally accessible functions.
                    --Such variables can have whatever name you want and may 
                    --also contain various data as well as functions.
                    
function get_raw_text(str)
	local builder = ""
	for index=1,string.len(str),1
	do
		local ch = string.sub(str,index,index)
		if ( ch == ' ' or
				ch == ',' or
				ch == '.' or
				ch == '-' or 
				ch == '_' or 
				ch == '&' or
				ch == '#' or
				ch == ';' or
				ch == '(' or 
				ch == ')' or
				(ch >= '0' and ch <= '9') or
				(ch >= 'A' and ch < 'Z') or 
				( ch >= 'a' and ch < 'z') ) then
			builder = builder .. ch
		else
			builder = builder .. "\\" .. ch .. " (" .. string.byte(ch,1) .. ")"
		end
	end
	return builder
end

function module.extract_version(frame)
	local base = frame.args[1]
	local stripAdditional = frame.args[2]
	
	-- remove the markers
	base = mw.text.killMarkers(base)
	
	-- the version is in the version beginning and is separated by '('
	local version = string.match(base, "^%s*([^(]+)%s*%(", 1, true)
	if version == nil then
		return "No version?"
	end
	
	-- it seems like match() doesn't support non-greedy captures, so just
	-- strip the whitespaces
	version = mw.text.trim(version)
	
	-- sometimes there is text that is put in front of it like
	--   Version 2019 24.2.2
	-- and we optionally want to remove that
	if stripAdditional == "true" then
		test = string.find(version, " ")
		while ( test ~= nil ) do
			version = string.sub(version, test + 1, string.len(version))
			test = string.find(version, " ") 
		end
	end
	
	-- done
	return version
end

function module.extract_date(frame)
	local base = frame.args[1]
	local format = frame.args[2]
	
	-- remove the markers
	base = mw.text.killMarkers(base)
	
	-- remove HTML entities
	base = mw.text.decode(base)
	
	-- if they request it in raw format, just extract it
	if format == nil or format == "raw" then
		-- remove spans
		--base = base:gsub("<span[^>]*>", "")
		--base = base:gsub("</span>", "")
	
		-- look for "ago" to get the date part
		local datepart = base:match("%(([^)]*);[^)]*ago")	
		if datepart == nil then
			return "No date?"
		end

		-- remove any extra space
		datepart = mw.text.trim(datepart)
		return datepart
	else
		
		-- extract the "dtstart" part
		local datepart = base:match("%(([^)]*);[^)]*ago")	

		-- parse it; dates usually are in the form:
		--			Month Year
		--			Month Day, Year
		--			Day Month Year
		-- TODO
		
		-- and convert it to the 
	
		-- done
		return "TODO"
	end
end

function module.debug_get_raw_text(frame)
	local base = frame.args[1]
	base = mw.text.killMarkers(base)
	local builder = get_raw_text(base)
	return "<pre>" .. builder .. "</pre>"
end

return module