Module:Parameter validation

This Lua module is used on approximately 132,000 pages.
To avoid major disruption and server load, any changes should be tested in the module's /sandbox or /testcases subpages, or in your own module sandbox. The tested changes can be added to this page in a single edit. Consider discussing changes on the talk page before implementing them.

This module depends on the following other modules:

Module:If preview

This module helps you find problematic usages of a template's parameters. Compared to Module:Check for unknown parameters, you do not need to create long lists of parameters which should be categorised; it will automatically read them from the template's TemplateData. This means you don't need to define the parameter configuration in multiple places, which will inevitably lead to it being out of sync.

As well as that, it supports a few other types of errors, such as defining the same parameter multiple times, or using deprecated parameters (useful to do so you can replace usages before removing support for a parameter entirely).

Basic usage

Add this to the bottom of a template:

{{#invoke:Parameter validation|validateparams|module_options = Module:Parameter validation/default config}}

This will use the default config defined in Module:Parameter validation/default config. The module will begin to populate the following categories:

Category:Pages using TEMPLATE_NAME with unknown parameters
Category:Pages using TEMPLATE_NAME with deprecated parameters
Category:Pages using TEMPLATE_NAME with duplicate parameters

For example, for {{Infobox station{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}{{---}}}}, it will populate:

Note that:

Unknown parameters means a page calling the template used a parameter that does not exist in the template documentation's TemplateData section
Deprecated parameters means a page calling the template used a parameter which, in the template's TemplateData, has the 'deprecated' option ticked
Duplicate parameters means a page calling the template and set a value for a parameter and one or more of its aliases

Use in protected templates

Note that because templates' TemplateData code typically lives in unprotected /doc pages, protected templates that invoke this module can be made to incorrectly categorize pages by editors who do not have user rights sufficient to edit the template page itself.

Full documentation

This module is based on idea and original code of User:IKhitron.

The source of this module, along with the original documentation, can be found at he:Module:ParamValidator.

local util = {
	empty = function( s ) 
		return s == nil  or type( s ) == 'string' and mw.text.trim( s ) == ''   
	end
	, 
	extract_options = function ( frame, optionsPrefix )
		optionsPrefix = optionsPrefix or 'options' 

		local options, n, more = {}
		if frame.args['module_options'] then
			local module_options = mw.loadData( frame.args['module_options'] ) 
			if type( module_options ) ~= 'table' then return {} end
			local title = mw.title.getCurrentTitle()
			local local_ptions = module_options[ title.namespace ] or module_options[ title.nsText ] or {} 
			for k, v in pairs( local_ptions ) do options[k] = v end
		end
		
		repeat
			ok, more = pcall( mw.text.jsonDecode, frame.args[optionsPrefix .. ( n or '' )] )
			if ok and type( more ) == 'table' then
				for k, v in pairs( more ) do options[k] = v end
			end
			n = ( n or 0 ) + 1
		until not ok

		return options
	end
	, 
	build_namelist = function ( template_name, sp )
		local res = { template_name }
		if sp then
			if type( sp ) == 'string' then sp = { sp } end
			for _, p in ipairs( sp ) do table.insert( res, template_name .. '/' .. p ) end
		end
		return res
	end
	,
	table_empty = function( t ) -- normally, test if next(t) is nil, but for some perverse reason, non-empty tables returned by loadData return nil...
		if type( t ) ~= 'table' then return true end
		for a, b in pairs( t ) do return false end
		return true
	end
	,
}

local function _readTemplateData( templateName ) 
	local title = mw.title.makeTitle( 0, templateName )  
	local templateContent = title and title.exists and title:getContent() -- template's raw content
	local capture =  templateContent and mw.ustring.match( templateContent, '<templatedata%s*>(.*)</templatedata%s*>' ) -- templatedata as text
--	capture = capture and mw.ustring.gsub( capture, '"(%d+)"', tonumber ) -- convert "1": {} to 1: {}. frame.args uses numerical indexes for order-based params.
	local trailingComma = capture and mw.ustring.find( capture, ',%s*[%]%}]' ) -- look for ,] or ,} : jsonDecode allows it, but it's verbotten in json
	if capture and not trailingComma then return pcall( mw.text.jsonDecode, capture ) end
	return false
end

local function readTemplateData( templateName )
	if type( templateName ) == 'string' then 
		templateName = { templateName, templateName .. '/' .. docSubPage }
	end
	if type( templateName ) == "table" then
		for _, name in ipairs( templateName ) do
			local td, result = _readTemplateData( name ) 
			if td then return result end
		end
	end
	return nil
end


-- this is the function to be called by other modules. it expects the frame, and then an optional list of subpages, e.g. { "Documentation" }.
-- if second parameter is nil, only tempalte page will be searched for templatedata.
function calculateViolations( frame, subpages )
-- used for parameter type validy test. keyed by TD 'type' string. values are function(val) returning bool.
	local type_validators = { 
		['number'] = function( s ) return mw.language.getContentLanguage():parseFormattedNumber( s ) end
	}
	function compatible( typ, val )
		local func = type_validators[typ]
		return type( func ) ~= 'function' or util.empty( val ) or func( val )
	end
	
	local t_frame = frame:getParent()
	local t_args, template_name = t_frame.args, t_frame:getTitle()
	template_name = mw.ustring.gsub( template_name, '/sandbox', '', 1 )
	local td_source = util.build_namelist( template_name, subpages )
	if frame.args['td_source'] then
		table.insert(td_source, frame.args['td_source'])
	end

	local templatedata = readTemplateData( td_source )
	local td_params = templatedata and templatedata.params
	local all_aliases, all_series = {}, {}

	if not td_params then return { ['no-templatedata'] = { [''] = '' } } end
	-- from this point on, we know templatedata is valid.

	local res = {} -- before returning to caller, we'll prune empty tables

	-- allow for aliases
	for x, p in pairs( td_params ) do for y, alias in ipairs( p.aliases or {} ) do
		p['primary'] = x
		td_params[x] = p
		all_aliases[alias] = p
		if tonumber(alias) then all_aliases[tonumber(alias)] = p end
	end end

	-- handle undeclared and deprecated
	local already_seen = {}
	local series = frame.args['series']
	for p_name, value in pairs( t_args ) do
		local tp_param, noval, numeric, table_name = td_params[p_name] or all_aliases[p_name], util.empty( value ), tonumber( p_name )
		local hasval = not noval

		if not tp_param and series then -- 2nd chance. check to see if series
			for s_name, p in pairs(td_params) do 
				if mw.ustring.match( p_name, '^' .. s_name .. '%d+' .. '$') then 
					-- mw.log('found p_name '.. p_name .. '  s_name:' .. s_name, ' p is:', p) debugging series support
					tp_param = p 
				end -- don't bother breaking. td always correct.
			end 				
		end

		if not tp_param then -- not in TD: this is called undeclared
			-- calculate the relevant table for this undeclared parameter, based on parameter and value types
			table_name = 
				noval and numeric and 'empty-undeclared-numeric' or
				noval and not numeric and 'empty-undeclared' or
				hasval and numeric and 'undeclared-numeric' or
				'undeclared' -- tzvototi nishar.
		else -- in td: test for deprecation and mistype. if deprecated, no further tests
			table_name = tp_param.deprecated and hasval and 'deprecated' 
				or tp_param.deprecated and noval and 'empty-deprecated' 
				or not compatible( tp_param.type, value ) and 'incompatible'
				or not series and already_seen[tp_param] and hasval and 'duplicate'

			if hasval and table_name ~= 'duplicate' then
				already_seen[tp_param] = p_name
			end
		end
		
		-- report it.
		if table_name then
			res[table_name] = res[table_name] or {}
			if table_name == 'duplicate' then
				local primary_param = tp_param['primary']
				local primaryData = res[table_name][primary_param]
				if not primaryData then
					primaryData = {}
					table.insert(primaryData, already_seen[tp_param])
				end
				table.insert(primaryData, p_name)
				res[table_name][primary_param] = primaryData
			else
				res[table_name][p_name] = value
			end
		end
	end

	-- check for empty/missing parameters declared "required" 
	for p_name, param in pairs( td_params ) do 
		if param.required and util.empty( t_args[p_name] ) then
			local is_alias
			for _, alias in ipairs( param.aliases or {} ) do is_alias = is_alias or not util.empty( t_args[alias] ) end
			if not is_alias then
				res['empty-required'] = res['empty-required'] or {} 
				res['empty-required'][p_name] = '' 
			end
		end
	end
	
	mw.logObject(res)
	
	return res
end

-- wraps report in hidden frame
function wrapReport(report, template_name, options)
	mw.logObject(report)
	if util.empty( report ) then return '' end
	local naked = mw.title.new( template_name )['text']
	naked = mw.ustring.gsub(naked, 'Infobox', 'infobox', 1)
	
	report = ( options['wrapper-prefix'] or "<div class = 'paramvalidator-wrapper'><span class='paramvalidator-error'>" )
			.. report
			.. ( options['wrapper-suffix'] or "</span></div>" )
	
	report = mw.ustring.gsub( report, 'tname_naked', naked )
	report = mw.ustring.gsub( report, 'templatename', template_name )

	return report
end

-- this is the "user" version, called with {{#invoke:}} returns a string, as defined by the options parameter
function validateParams( frame )
	local options, report, template_name = util.extract_options( frame ), '', frame:getParent():getTitle()

	local ignore = function( p_name )
		for _, pattern in ipairs( options['ignore'] or {} ) do
			if mw.ustring.match( p_name, '^' .. pattern .. '$' ) then return true end
		end
		return false
	end

	local replace_macros = function( error_type, s, param_names )
		function concat_and_escape( t , sep )
			sep = sep or ', '
			local s = table.concat( t, sep )
			return ( mw.ustring.gsub( s, '%%', '%%%%' ) )
		end
		
		if s and ( type( param_names ) == 'table' ) then
			local k_ar, kv_ar = {}, {}
			for k, v in pairs( param_names ) do
				table.insert( k_ar, k )
				if type(v) == 'table' then
					v = table.concat(v, ', ')
				end
					
				if error_type == 'duplicate' then
					table.insert( kv_ar, v)
				else
					table.insert( kv_ar, k .. ': ' .. v)
				end
			end
			
			s = mw.ustring.gsub( s, 'paramname', concat_and_escape( k_ar ) )
			s = mw.ustring.gsub( s, 'paramandvalue', concat_and_escape( kv_ar, ' AND ' ) )

			if mw.getCurrentFrame():preprocess( "{{REVISIONID}}" ) ~= "" then
				s = mw.ustring.gsub( s, "<div.*<%/div>", "", 1 )
			end
		end
		return s
	end

	local report_params = function( key, param_names )
		local res = replace_macros( key, options[key], param_names )
		res = frame:preprocess(res or '')
		report = report ..  ( res or '' )
		return res
	end

	-- no option no work.
	if util.table_empty( options ) then return '' end

	-- get the errors.
	local violations = calculateViolations( frame, options['doc-subpage'] )
	-- special request of bora: use skip_empty_numeric
	if violations['empty-undeclared-numeric'] then 
		for i = 1, tonumber( options['skip-empty-numeric'] ) or 0 do 
			violations['empty-undeclared-numeric'][i] = nil 
		end
	end
	
	-- handle ignore list, and prune empty violations - in that order!
	local offenders = 0
	for name, tab in pairs( violations ) do 
		-- remove ignored parameters from all violations
		for pname in pairs( tab ) do if ignore( pname ) then tab[pname] = nil end end
		-- prune empty violations
		if util.table_empty( tab ) then violations[name] = nil end
	-- WORK IS DONE. report the errors.
	-- if report then count it.
		if violations[name] and report_params( name, tab ) then offenders = offenders + 1 end 
	end

	if offenders > 1 then report_params( 'multiple' ) end
	if offenders ~= 0 then report_params( 'any' ) end -- could have tested for empty( report ), but since we count them anyway...
	return wrapReport(report, template_name, options)
end

return {
	['validateparams'] = validateParams,
	['calculateViolations'] = calculateViolations,
	['wrapReport'] = wrapReport
}