Module:Plain text/sandbox

Documentation for this module may be created at Module:Plain text/sandbox/doc

--converts text with wikilinks to plain text, e.g "[[foo|gah]] is [[bar]]" to "gah is bar"
--removes anything enclosed in tags that isn't nested, mediawiki strip markers (references etc), files, italic and bold markup
local p = {}

function p.main(frame)
	local text = frame.args[1]
	local encode = require('Module:yesno')(frame.args.encode)
	return p._main(text, encode)
end

function p._main(text, encode)
	if not text then return end
	text = mw.text.killMarkers(text)
		:gsub(' ', ' ') --replace nbsp spaces with regular spaces
		:gsub('<br ?/?>', ', ') --replace br with commas
		:gsub('<(%a+)[^>]+>(.-)</%1>', function(tag, contents)
			if tag:lower() == 'span' then
				return contents
			else
				return ''
			end
		end)
		:gsub('<i[^>]+>([^<]+)</i>', '%1') --remove italics while keeping text inside
		:gsub('<[^>]+>[^<]+<[^>]+>', '') --strip out remaining tags and the text inside
		:gsub('%b<>', '') --remove any other tag markup
		:gsub('__[^_]+__', '') --remove __ markups
		:gsub('^=+[^=]+=+', ''):gsub('\n=+[^=]+=+', '') --remove section titles		
		:gsub('%b[]',
			function(bracketed)
				return bracketed:gsub('^%[%[%s*(%a+):.-%]%]$',
					function(link_prefix)
						link_prefix = link_prefix:lower()
						if link_prefix == 'image' or link_prefix == 'file'
						or link_prefix == 'media' or link_prefix == 'category' then
							return ""
						end -- otherwise leave it alone
					end)
			end)
		:gsub('%[%[[^%]|]+|', '') --strip out piped link text
		:gsub('([^%[])%[[^%[%]][^%]]-%s', '%1') --strip out external link text
		:gsub('^%[[^%[%]][^%]]-%s', '') --strip out external link text
		:gsub('[%[%]]', '') --then strip out remaining [ and ]
		:gsub("'''''", "") --strip out bold italic markup
		:gsub("'''?", "") --not stripping out '''' gives correct output for bolded text in quotes
		:gsub('----', '') --remove ---- lines
		:gsub('^%s+', ''):gsub('\n%s+', '\n') --strip leading
		:gsub('%s+$', ''):gsub('%s+\n', '\n') --and trailing spaces
		:gsub('(%s)%s+', '%1') --strip redundant spaces
	if encode then
		return mw.text.encode(text)
	else
		return text
	end
end

return p