Modul:Citation/utilities

Die Dokumentation für dieses Modul kann unter Modul:Citation/utilities/doc erstellt werden

-- documentation
local citationUtilities = {
	suite  = 'Citation',
	sub    = 'utilities', 
	serial = '2021-09-01'
}

-- module variable and administration
local cu = {
	moduleInterface = citationUtilities
}

-- module import
require( 'Module:No globals' )
local ci = require( 'Module:Citation/i18n' )

-- global variable
local errorMsgs = {}

-- helper functions
function cu.isSet( param )
	return param and param ~= '';
end

-- math function round
function cu.round( num, decimalPlaces )
  local mult = 10^( decimalPlaces or 0 )
  return math.floor( num * mult + 0.5 ) / mult
end

-- converts string to number
function cu.getNumber( s )
	if not cu.isSet( s ) then
		return 0
	end
	if s:find( '[sic!]', 1, true ) then
		return 0
	end
	local kb = s:gsub( ci.texts.decimalPoint, '%.' )
	return tonumber( kb ) or 0
end

-- adds error message to array
function cu.addErrorMsg( msg )
	table.insert( errorMsgs, msg )
end

-- make complete message from message array
function cu.getErrorMsgs()
	-- remove duplicates
	for i = #errorMsgs, 1, -1 do
		for j = 1, #errorMsgs - 1, 1 do
			if errorMsgs[ i ] == errorMsgs[ j ] then
				table.remove( errorMsgs, i )
				break
			end
		end
	end

	local result = table.concat( errorMsgs, ' ' )
	if result ~= '' then
		result = result .. ' '
	end
	return result
end

-- get first item of a delimiter-separated list
function cu.getFirst( s, delimiter )
	local at = s:find( delimiter )
	if at then
		s = mw.text.trim( s:sub( 1, at - 1 ) )
	end
	return s
end

-- check if table contains the value
function cu.inArray( tab, val )
	if type( tab ) == 'string' then
		return tab == val
	end

	for index, value in ipairs( tab ) do
		if value == val then
			return true
		end
	end

	return false
end

-- convert values t from list if translated
function cu.getKey( t, list )
    local result = ''
    for key, tab in pairs( list ) do
        if cu.inArray( tab, t ) then
            result = key
            break
        end
    end
    return result
end

-- returns a single value from frame argument table
function cu.getArgValue( list, param, args )
	local value = ''
	if list[ param ] then
		for k, v in ipairs( list[ param ] ) do
			if cu.isSet( args[ v ] ) then
				value = args[ v ]
				break
			end
		end
	end
	return value
end

-- string cleanup
function cu.parameterCleanup( s )
	if not cu.isSet( s ) then
		return s -- nil or '' should be kept
	end

	local orig = s
	-- replace control characters
	s = s:gsub( '[\009\010\013]', ' ' ) -- horizontal tab, line feed, carriage return
	s = s:gsub( '[%z%c]', '' ) -- control characters

	-- remove tags
	s = mw.ustring.gsub( s, '</*br[^/>]*/*>', '' ) -- <br> tag
	s = mw.ustring.gsub( s, '</*p[^/>]*/*>', '' ) -- <p> tag
	s = mw.ustring.gsub( s, '</*div[^/>]*/*>', '' ) -- <div> tag

	if orig ~= s then
		cu.addErrorMsg( ci.texts.wrongChars )
	end

	-- replace character references and entities
	s = mw.text.decode( s, true )
	-- replace characters
	s = mw.ustring.gsub( s, '%.%.%.', '…' )
	s = mw.ustring.gsub( s, '%.%.', '‥' )
	return s:gsub( '%s%s+', ' ' ) -- multiple spaces
end

-- remove illegal chars from pages parameters
function cu.cleanupPageNumbers( pages )
	if not cu.isSet( pages ) then
		return ''
	end

	-- replace dashes with hyphens
	return mw.ustring.gsub( '' .. pages, '[–‒—]', '-' );
end

-- string cleanup before COinS creation
function cu.coinsCleanup( s )
	if not cu.isSet( s ) then
		return ''
	end

	-- replace characters
	s = s:gsub( '\226\128\138', ' ' ); -- hair space
	s = mw.ustring.gsub( s, '[\226\128\141\226\128\139\194\173]', '' );
		-- zero-width joiner, zero-width space, soft hyphen

	-- remove characters: soft hyphen, LTR mark, RTL mark
	for i, value in ipairs( { '­', '‎', '‏' } ) do
		s = mw.ustring.gsub( s, value, '' )
	end

	-- replace Wiki syntax
	s = s:gsub( "''+", '' ) -- multiple apostrophes
	s = mw.ustring.gsub( s, '</*span[^/>]*/*>', '' ) -- span tags
	s = mw.ustring.gsub( s, '%[%[[^%[%]]*|([^%[%]]*)%]%]', '%1' ) -- MediaWiki links
	s = mw.ustring.gsub( s, '%[%[([^%[%]]*)%]%]', '%1' )
	s = mw.ustring.gsub( s, '%[%a*:?//[^ ]+%s+([^%]]+)%]', '%1' ) -- web links
	s = mw.ustring.gsub( s, '%[mailto:[^ ]+%s+([^%]]+)%]', '%1' )
	s = mw.ustring.gsub( s, '%[%a*:?//([^%]]+)%]', '%1' )
	s = mw.ustring.gsub( s, '%[mailto:([^%]]+)%]', '%1' )

	return s:gsub( '%s%s+', ' ' ) -- multiple spaces
end

-- remove adjoining punctuation marks etc.
function cu.finalCleanup( s )
	s = s:gsub(	'%.+%.', '.' ):gsub( '%s%s+', ' ' ):gsub( '([,;:])(%s%.+)', '.' )
	for _, replacement in ipairs( ci.replacements ) do
		s = mw.ustring.gsub( s, replacement.s, replacement.r )
	end
	return s
end

function cu.makeLink( url, text )
	return mw.ustring.format( '[%s %s]', url, text )
end

function cu.makeDoiLink( doi )
	doi = mw.ustring.gsub( doi, 'https?://doi.org/', '' )
	doi = doi:gsub( '%[', '%%5B' ):gsub( '%]', '%%5D' )
	return cu.makeLink( 'https://doi.org/' .. mw.uri.encode( doi ), doi )
end

function cu.getDate( aDate, aFormat, errorMsg )
	local function formatDate( aDate, aFormat )
		return mw.getContentLanguage():formatDate( aFormat, aDate, true )
	end

	if aFormat == 'R' then
		return aDate
	end

	if aDate ~='' then
		local success, t = pcall( formatDate, aDate, aFormat )
		if success then
			return t
		else
			cu.addErrorMsg( errorMsg )
			return ''
		end
	else
		return ''
	end
end

-- check if url contains a file extension
function cu.hasExtension( url, ext )
	ext = '%.' .. ext:upper()
	url = url:upper()
	return url:match( ext .. '$' ) or url:match( ext .. '[%?#]' ) or
		url:match( ext .. '&#035' ); -- &#035 is #
end

function cu.insertItem( tab, s, formatStr )
	if cu.isSet( s ) then
		if formatStr then
			table.insert( tab,
				mw.ustring.format( formatStr, s )
			)
		else
			table.insert( tab, s )
		end
	end
end

function cu.insertItem2( tab, s1, s2, formatStr )
	if cu.isSet( s1 ) and cu.isSet( s2 ) then
		if formatStr then
			table.insert( tab,
				mw.ustring.format( formatStr, s1, s2 )
			)
		else
			table.insert( tab, s1 )
			table.insert( tab, s2 )
		end
	end
end

function cu.formatItem( s, formatStr )
	if cu.isSet( s ) then
		if formatStr then
			return mw.ustring.format( formatStr, s )
		else
			return s
		end
	else
		return ''
	end
end

function cu.formatItem2( s1, s2, formatStr )
	if cu.isSet( s1 ) and cu.isSet( s2 ) then
		if formatStr then
			 return mw.ustring.format( formatStr, s1, s2 )
		else
			return s1 .. ' ' .. s2
		end
	else
		return ''
	end
end

return cu