Modul:Month translator/data

--[[--------------------------< L A N G U A G E S >------------------------------------------------------------

Table of ISO 639 codes for languages supported by this module.    List of MediaWiki supported languages and their
codes can be found at: Template:Citation_Style_documentation/language/doc

]]

local langs = {'ca', 'da', 'de', 'es', 'fi', 'fr', 'it', 'nb', 'pl', 'pt', 'ru', 'sl', 'sv', 'tr'};


--[[--------------------------< E N G L I S H   M O N T H   N A M E S >----------------------------------------

Table of local language month names filled by month_names_get()

]]

local en_months_t = {};


--[[--------------------------< P A T T E R N S >--------------------------------------------------------------

table of tables that hold pattern and other information used to extract date parts for translation and reconstruction

Tables have the form:
	[1] - pattern for a particular date format; must include captures for the requiste date components:
			day, month, year which ever are included in the date format
	[2] - a letter 'd', 'm', or 'y' that identified the content of the first (left-most) capture
	[3] - same as [2] for the second capture
	[4] - same as [2] for the last (right-most) capture

ymd numeric dates have no hames so are not translated; use |df= parameter in the cs1|2 template for that.

]]

local patterns = {
	{'^(%d%d?) +(%a+) +(%d%d%d%d%a?)$', 'd', 'm', 'y'},							-- dd Mmm yyyy
	{'^(%d%d?)%. +(%a+) +(%d%d%d%d%a?)$', 'd', 'm', 'y'},						-- dd. Mmm yyyy – for Danish
	{'^(%d%d?) +de +(%a+) +de +(%d%d%d%d%a?)$', 'd', 'm', 'y'},					-- dd de Mmm de yyyy
	{'^(%a+) +(%d%d%d%d%a?)$', 'm', 'y'},										-- Mmm yyyy
	{'^(%a+) +de +(%d%d%d%d%a?)$', 'm', 'y'},									-- Mmm de yyyy
	{'^(%a+) +(%d%d?) *, +(%d%d%d%d%a?)$', 'm', 'd', 'y'},						-- Mmm dd, yyyy
	{'^(%a+)$', 'm'},															-- month only; mostly for debug purposes
	};


--[[--------------------------< O V E R R I D E _ N A M E S >--------------------------------------------------

table of non-English month names that are valid but are not in the list of names taken from MediaWiki with 
month_names_get().  Items in this list have the form:
	['<non-English month name>'] = 'English month name',
		'<non-English month name>' must be lowercase
]]

local override_names = {
	['januar'] = 'January',
	['februar'] = 'February',
	['marec'] = 'March',
	['april'] = 'April',
	['maj'] = 'May',
	['junij'] = 'June',
	['julij'] = 'July',
	['avgust'] = 'August',
	['september'] = 'September',
	['oktober'] = 'October',
	['november'] = 'November',
	['december'] = 'December',	
	}


--[[--------------------------< M O N T H _ N A M E S _ G E T >------------------------------------------------

creates a translation table of non-English month names listed in lang{} mapped to English month names listed in
<en_months_t> (which this function also fills).

lang_obj:formatDate() must be given a day in addition to <en_month> else it will use the current day so, if today
is 31 August 2021, without the expicit '1' preceding the month from <en_month>, when creating a translation for
February, April, June, September, or October lang_obj:formatDate() will return the 'next' month.

To dump this table: =mw.dumpObject (p.month_names_t)

]]

local function month_names_get ()
	local month_names_t = {};

	local lang_obj = mw.language.getContentLanguage();							-- make a language object for the local language 
	for i=1, 12 do																-- loop 12x and 
		en_months_t[i] = lang_obj:formatDate('F', '2018-' .. i);				-- get month names for each i
	end

	for _, lang in ipairs (langs) do											-- spin through the languages table
		lang_obj = mw.getLanguage (lang);										-- make a language object for the current language
		for i, en_month in ipairs (en_months_t) do								-- spin through the English month-names table
			local en_date = '1 ' .. en_month;
			local month_name = mw.ustring.lower (lang_obj:formatDate('F', en_date));	-- translate the English date '1 <en month>' to <month_name>; ustring requred for tr February (Şubat)
			month_names_t[month_name] = en_month;								-- add to translations table; first day of month required

																				-- for pl and other languages that have nominative and genitive forms
			month_name = mw.ustring.lower (lang_obj:formatDate('xg', en_date));	-- translate the English date '1 <en month>' to <month_name> (genitive form)
			month_names_t[month_name] = en_month;								-- add to translations table; when 'xg' is same as 'F', this overwrites the 'F' month name
		end
	end

	return month_names_t;
end


--[[--------------------------< E X P O R T E D   T A B L E S >------------------------------------------------
]]

return {
	month_names_t = month_names_get (),
	override_names = override_names,
	patterns = patterns,
	}