Module:Replace

Documentation for this module may be created at Module:Replace/doc

-- <pre>
-- A module to allow substitution of regex replace functions

local p = {}

function p.main(frame)
	local args = frame:getParent().args
	return p._main(args)
end

-- a : string
-- b : search
-- c : replace
function p._main(args)
	local a =  mw.text.decode(args[1])
	local b = mw.text.decode(args[2] or '')
	local c = mw.text.decode(args[3] or '')

	-- let us use real regex stuff
	local b = mw.ustring.gsub(b,'\\','%')
				:gsub('%*%?','-')
	local c = mw.ustring.gsub(c,'$(%d)','%%%1')

	-- get just the new string
	local ret = mw.ustring.gsub(a,b,c)

	-- trim whitespace
	ret = mw.text.trim(ret)

	-- condense whitespace
	ret = mw.ustring.gsub(ret,'  +',' ')

	return ret
end

-- trying to emulate regex's alteration in pattern matches
function p.test(reg)
	-- create 2 sets of captures to use
	-- one for storage, the other for manipulation
	local captures,_captures = {},{}

	-- string to perfom the search on
	local _s = '55(aa)66(bb)77 7aaaa'

	-- string to hold pattern match
	local s = reg or '55%(aa%)66%((bb)%)..%s.(aaaa|asas).(aa|45|4444)'
	mw.log(s)
	mw.log('---')

	-- convert parentheses into full width for temp parsing
	s = mw.ustring.gsub(s,'%%%(','（')
	s = mw.ustring.gsub(s,'%%%)','）')

	-- string to use for string.format in tests
	-- convert %s to %$ for temp parsing
	local _sform = mw.ustring.gsub(s,'%%s','%%$')

	-- matches any set of parentheses that isn't started with a %
	_sform = string.gsub(_sform,'%f[(%%](%b())','(%%s)')

	-- convert full width back to escaped
	_sform = _sform:gsub('（','%%('):gsub('）','%%)')

	-- double up % since string.format is bitchy
	_sform = _sform:gsub('%%([^s])','%%%%%1')

	-- turn $ back to s
	_sform = _sform:gsub('%%%%%$','%%%%s')

	-- finds all parenthetical groups that aren't begun with a %
	-- add to table of manipulate-able capture
	for v in string.gmatch(s,'%f[(%%](%b())') do
		-- match to remove parentheses
		table.insert(captures,string.match(v,'^%((.+)%)$'))
	end

	-- convert each capture into a table
	-- split by alteration character
	for _, v in ipairs(captures) do
		table.insert(_captures,mw.text.split(v,'|'))
	end

	-- table of all possible combinations used for the formatting
	local groupstouse = {}

	-- recursive function 
	local function addtogroups(x,stor)
		-- for all in the set
		for i, v in ipairs(_captures[x]) do
			-- temporary storage
			local _stor = {}

			-- deep copy because fuck lua
			for _, u in ipairs(stor or {}) do
				table.insert(_stor,u)
			end

			-- add current pattern to storage
			table.insert(_stor,v)

			-- if there's a next group, run func on those
			if _captures[x+1] then
				addtogroups(x+1,_stor)

			-- otherwise just add this to the master table
			else
				table.insert(groupstouse,_stor)
			end
		end
	end

	-- run recursive func
	addtogroups(1)

	for _, v in ipairs(groupstouse) do

		-- if a match is found, use those groups
		local formatted = mw.ustring.format(_sform,unpack(v))

		-- logging results
		mw.log(table.concat(v,' -- ')..' : '..formatted)
	end
end

return p