Kulemba Module:URL

--
-- This module implements {{URL}}
--
-- See unit tests at [[Module:URL/testcases]]

local p = {}
 
local function safeUri(s)
	local success, uri = pcall(function()
		return mw.uri.new(s)
	end)
	if success then
		return uri
	end
end

local function extractUrl(args)
	for name, val in pairs(args) do
		if name ~= 2 and name ~= "msg" then
			local url = name .. "=" .. val;
			url = mw.ustring.gsub(url, '^[Hh][Tt][Tt][Pp]([Ss]?):(/?)([^/])', 'http%1://%3')
			local uri = safeUri(url);
			if uri and uri.host then
				return url
			end
		end
	end
end

function p._url(url, text, msg)
	url = mw.text.trim(url or '')
	text = mw.text.trim(text or '')
	local nomsg = (msg or ''):sub(1,1):lower() == "n" or msg == 'false' -- boolean: true if msg is "false" or starts with n or N

	if url == '' then
		if text == '' then
			if nomsg then
				return nil
			else
				return mw.getCurrentFrame():expandTemplate{ title = 'tlx', args = { 'URL', "''example.com''", "''optional display text''" } }
			end
		else
			return text
		end
	end

	-- If the URL contains any unencoded spaces, encode them, because MediaWiki will otherwise interpret a space as the end of the URL.
	url = mw.ustring.gsub(url, '%s', function(s) return mw.uri.encode(s, 'PATH') end)

	-- If there is an empty query string or fragment id, remove it as it will cause mw.uri.new to throw an error
	url = mw.ustring.gsub(url, '#$', '')
	url = mw.ustring.gsub(url, '%?$', '')

	-- If it's an HTTP[S] URL without the double slash, fix it.
	url = mw.ustring.gsub(url, '^[Hh][Tt][Tt][Pp]([Ss]?):(/?)([^/])', 'http%1://%3')

	local uri = safeUri(url)

	-- Handle URL's without a protocol and URL's that are protocol-relative, 
	-- e.g. www.example.com/foo or www.example.com:8080/foo, and //www.example.com/foo
	if uri and (not uri.protocol or (uri.protocol and not uri.host)) and url:sub(1, 2) ~= '//' then
		url = 'http://' .. url
		uri = safeUri(url)
	end

	if text == '' then
		if uri then
			if uri.path == '/' then uri.path = '' end

			local port = ''
			if uri.port then port = ':' .. uri.port end

			text = mw.ustring.lower(uri.host or '') .. port .. (uri.relativePath or '')

			-- Add <wbr> before _/.-# sequences
			text = mw.ustring.gsub(text,"(/+)","<wbr/>%1")      -- This entry MUST be the first. "<wbr/>" has a "/" in it, you know.
			text = mw.ustring.gsub(text,"(%.+)","<wbr/>%1")
			-- text = mw.ustring.gsub(text,"(%-+)","<wbr/>%1") 	-- DISABLED for now
			text = mw.ustring.gsub(text,"(%#+)","<wbr/>%1")
			text = mw.ustring.gsub(text,"(_+)","<wbr/>%1")
		else -- URL is badly-formed, so just display whatever was passed in
			text = url
		end
	end

	return mw.ustring.format('<span class="url">[%s %s]</span>', url, text)
end

--[[
The main entry point for calling from Template:URL.
--]]
function p.url(frame)
	local templateArgs = frame.args
	local parentArgs = frame:getParent().args
	local url = templateArgs[1] or parentArgs[1]
	local text = templateArgs[2] or parentArgs[2] or ''
	local msg = templateArgs.msg or parentArgs.msg or ''
	url = url or extractUrl(templateArgs) or extractUrl(parentArgs) or ''
	return p._url(url, text, msg)
end

--[[
The entry point for calling from the forked Template:URL2.
This function returns no message by default.
It strips out wiki-link markup, html tags, and everything after a space.
--]]
function p.url2(frame)
	local templateArgs = frame.args
	local parentArgs = frame:getParent().args
	local url = templateArgs[1] or parentArgs[1]
	local text = templateArgs[2] or parentArgs[2] or ''
	-- default to no message
	local msg = templateArgs.msg or parentArgs.msg or 'no'
	url = url or extractUrl(templateArgs) or extractUrl(parentArgs) or ''
	-- if the url came from a Wikidata call, it might have a pen icon appended
	-- we want to keep that and add it back at the end.
	local u1, penicon = mw.ustring.match( url, "(.*)(&nbsp;<span class='penicon.*)" )
	if penicon then url = u1 end
	-- strip out html tags and [ ] from url
	url = (url or ''):gsub("<[^>]*>", ""):gsub("[%[%]]", "")
	-- truncate anything after a space
	url = url:gsub("%%20", " "):gsub(" .*", "")
	return (p._url(url, text, msg) or "") .. (penicon or "")
end

return p