Module:DPLlua

From MediaWiki
Jump to navigation Jump to search

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

-- <nowiki>
local dpl = {}
local libraryUtil = require( 'libraryUtil' )
local checkType = libraryUtil.checkType
local checkTypeForNamedArg = libraryUtil.checkTypeForNamedArg

dpl.pipe = '¦'
local dataContentMarker = '`#@@#`'
local allIncludedParamNames = {}

-- Custom function for splitting a string because mw.text.split() is waaay too slow
local function split( str, pattern, plain )
	local res = {}
	local continue = true
	local startIndex = 1

	while continue do
		local i, j = string.find( str, pattern, startIndex, plain )
		if i then
			table.insert( res, string.sub( str, startIndex, i-1 ) )
			startIndex = j + 1
		else
			table.insert( res, string.sub( str, startIndex ) )
			continue = false
		end
	end

	return res
end

-- Also custom function for speed
local function trim( str )
	return string.match( str, '^%s*(.-)%s*$' )
end

local escapeChars = {
	['{'] = '&#123;',
	['❴'] = '&#123;', -- Wtf dpl...
	['}'] = '&#125;',
	['❵'] = '&#125;',
	['['] = '&#91;',
	[']'] = '&#93;',
	['|'] = '&#124;',
	['-'] = '&#8208;'
}
local function escape( str )
	-- the \226\157\180\181 are used to match ❴ (U+2774) and ❵ (U+2775) wich are 3 bytes long (UTF-8) so
	-- we can't use them directly inside [] patterns. Ustring would fix this but it's way too slow.
	str = string.gsub( str, '[{}%[%]|%-]', escapeChars ):gsub( '\226\157[\180\181]', escapeChars )
	return str
end

local unEscapeChars = {
	['&#123;'] = '{',
	['&#125;'] = '}',
	['&#91;'] = '[',
	['&#93;'] = ']',
	['&#124;'] = '|',
	['&#8208;'] = '-'
}
local function unEscape( str )
	str = string.gsub( str, '&#%d+;', unEscapeChars )
	return str
end

local function removeFormattingSettings( query )
	local toRemove = {
		'mode',
		'table',
		'tablerow',
		'tablesortcol',
		'headingmode',
		'headingcount',
		'listattr',
		'itemattr',
		'hlistattr',
		'hitemattr',
		'userdateformat',
		'shownamespace',
		'escapelinks',
		'titlemaxlength',
		'replaceintitle',
		'columns',
		'rows',
		'rowsize',
		'rowcolformat',
		'resultsheader',
		'resultsfooter',
		'oneresultheader',
		'oneresultfooter',
		'noresultsheader',
		'suppresserrors',
		'noresultsfooter',
		'format'
	}

	for _, k in ipairs( toRemove ) do
		query[k] = nil
	end
end

local function formatInclude( query )
	checkTypeForNamedArg( 'include', 'Module:DPLlua.ask', query, 'string' )
	query = split( query, ',', true )
	local includedParamNames = {}

	for i = 1, #query do
		if query[i]:match( '%b{}' ) then -- Check if we are including a template
			local templateName, params = query[i]:match( '{(.-)[¦|}]([^,]*)' )
			if params:find( '%S' ) then
				params:gsub( '^:%-', '' )
				query[i] = string.format( '{%s}%s', templateName, params )

				for param in params:gmatch( ':([^:]*)' ) do
					param = trim( param )
					table.insert( includedParamNames, { name=templateName, isTemplate=true, param=param } )
				end

			else
				query[i] = string.format( '{%s¦DPLlua helper}', templateName ) -- Use a helper template to get all the parameters of our included template
				table.insert( includedParamNames, { name=templateName, isTemplate=true, includeAll=true } )
			end
		else
			table.insert( includedParamNames, { name=trim( query[i] ) } )
		end
	end

	return table.concat( query, ',' ), includedParamNames
end

local function formatDpl( query )
	local queries = {}
	local count = query.count or 500
	local offset = query.offset or 0
	local usesInclude = false
	local includedParamNames = {}
	query.count = nil
	query.offset = nil

	-- We use table format when the include parameter is used to make sure we can
	-- differentiate between the results in case more than one item is included
	local dplStringInclude =
[=[
{{#dpl:
|noresultsheader=@@
|count=%s
|offset=%s
|%s
|table=,
|tablerow=%s
}}]=]

	-- Table format requires an include statement so we use format instead.
	-- This is also a lot faster than adding an empty include statement
	local dplStringNoInclude =
[=[
{{#dpl:
|noresultsheader=@@
|count=%s
|offset=%s
|%s
|format=,¦-¦[[%%PAGE%%¦]],,
}}]=]

	-- Auto generate more than one dpl if count > 500
	-- The results of these are later combined
	for i = 1, math.ceil( count / 500 ) do
		local params = {}

		for k, v in pairs( query ) do
			if k == 'include' then
				v, includedParamNames = formatInclude( v )
				usesInclude =  true
			end

			if type( v ) == 'table' then
				for _, x in ipairs( v ) do
					table.insert( params, k .. '=' .. tostring( x ):gsub( '|', '¦' ) )
				end
			else
				table.insert( params, k .. '=' .. tostring( v ):gsub( '|', '¦' ) )
			end
		end

		if usesInclude then
			table.insert( queries, string.format(
				dplStringInclude,
				count > 500 and 500 or count,
				offset,
				table.concat( params, '\n|' ),
				string.rep( dataContentMarker..'%%'..dataContentMarker..',', #includedParamNames )
			) )
		else
			table.insert( queries, string.format(
				dplStringNoInclude,
				count > 500 and 500 or count,
				offset,
				table.concat( params, '\n|' )
			) )
		end

		count = count - 500
		offset = offset + 500
	end

	table.insert( allIncludedParamNames, includedParamNames )

	return table.concat( queries )
end

local function toTable( query )
	local includedParamNames = table.remove( allIncludedParamNames, 1 )
	local usesInclude = #includedParamNames > 0
	local res = {}

	query = query:gsub( '<p>Extension:DynamicPageList .-</p>', function(item) res.error = item; return '' end )

	if query:find( '^@@' ) then -- @@ is used when no result is found
		return res
	end

	if usesInclude then
		query = query:gsub( '\127\'"`UNIQ%-%-nowiki%-%x+%-QINU`"\'\127', function(item) return '<nowiki>' .. item .. '</nowiki>' end ) -- Unstrip nowiki so we can clean their content
		query = mw.text.unstripNoWiki( query )
		query = query:gsub( dataContentMarker..'(.-)'..dataContentMarker, escape )
		query = query:gsub( '{|.-|%-', '' ) -- Remove the header of the table
		-- Replace the footer of the table width a row indicator. This effectively
		-- combines the output of multiple dpl queries when count > 500
		query = query:gsub( '|}', '|-' )
	end

	query = trim( query )
	query = split( query, '|-', true ) -- Results of the returned pages are separated by |-

	for _, v in ipairs( query ) do
		if v:find( '%S' ) and not v:find( '^@@' ) then
			v = trim( v )
			local title = v:match( '^|%[%[(.-)|' )
			local rawDataList = v:match( '^|.-|.-|(.*)' ) -- This is everything after the title

			if not usesInclude then
				if title and title ~= '' then
					table.insert( res, title )
				end
			else
				-- When multiple includes are used (e.g. include={Template1},{Template2} or include={Template}:1:2) its results are separated by a pipe
				rawDataList = split( rawDataList, '|', true )
				local cleanedDataList = {}

				for incIndex, dataItem in ipairs( rawDataList ) do
					dataItem = unEscape( dataItem )
					-- When we include an entire template we use the %ARGS% parameter supplied by dpl.
					-- However all | characters are repaced with §, e.g.:
					-- §nameLessParam
					-- §param = text [[wowee§link text]]
					-- §param2 = text {{something§something else}}
					dataItem = dataItem:gsub( '%b{}', function(x) return x:gsub( '§', '|' ) end ) -- Restore pipe characters inside links and templates
					dataItem = dataItem:gsub( '%b[]', function(x) return x:gsub( '§', '|' ) end )
					dataItem = dataItem:gsub( '<nowiki>(.-)</nowiki>', function(x) return mw.getCurrentFrame():extensionTag( 'nowiki', x ) end ) -- Restrip nowiki
					dataItem = trim( dataItem )

					if includedParamNames[ incIndex ].isTemplate and includedParamNames[ incIndex ].includeAll then -- Check if we included a full template
						local _dataItem = {}

						if dataItem ~= '' then
							dataItem = split( dataItem:sub( 3 ), '§' ) -- The sub(3) removes the first § at the start. § is 2 bytes wide so start at index 3

							for i, item in ipairs( dataItem ) do
								if item:find( '=' ) then -- Check if the parameter is named or unnamed
									local param, value = item:match( '^%s*(.-)%s*=%s*(.-)%s*$' )
									_dataItem[ param ] = value
								else
									table.insert( _dataItem, trim( item ) )
								end
							end
						end

						dataItem = _dataItem
					end

					if includedParamNames[ incIndex ].isTemplate and not includedParamNames[ incIndex ].includeAll then -- This means there was an include in the form 'include = {template}:param'
						local templateName = includedParamNames[ incIndex ].name
						local paramName = includedParamNames[ incIndex ].param
						paramName = tonumber( paramName ) or paramName -- Keep as string if tonumber fails
						cleanedDataList[ templateName ] = cleanedDataList[ templateName ] or {}
						cleanedDataList[ templateName ][ paramName ] = dataItem
					else
						cleanedDataList[ includedParamNames[ incIndex ].name ] = dataItem
					end
				end

				if title and title ~= '' then
					table.insert( res, { title=title, include=cleanedDataList } )
				end
			end
		end
	end

	return res
end

-- Accepts a series of tables each containig the settings for a dpl query.
-- Combinig multiple dpl queries yields better performance than doing them sequentially
function dpl.ask( ... )
	local queries = { ... }

	for i = 1, #queries do
		checkType( 'Module:DPLlua.ask', i, queries[i], 'table' )
		removeFormattingSettings( queries[i] )
		queries[i] = formatDpl( queries[i] )
	end

	queries = table.concat( queries, '$@µ@$' )
	local time = os.clock()
	queries = mw.getCurrentFrame():preprocess( queries )
	time = os.clock() - time
	queries = split( queries, '$@µ@$', true )

	for i = 1, #queries do
		queries[i] = toTable( queries[i] )
		queries[i].time = time
	end

	return unpack( queries )
end

return dpl
-- </nowiki>