Módulo:LangInfo

Provides the functionality of {{langInfo}}.

The main and only function is go; the kind of information returned is determined by the second unnamed parameter. The table wplinks contains, for every language where it is necessary, the localized page title of the respective language's page about that language.

Example:

["en"] = "English language"

This is only necessary if that page title is not equal to the language's lname field in the data module.

---

--------------------------------------------------------------------------------
--
-- =============================================================================
--
-- Module:LangInfo
--
-- Miscellaneous information about languages
--
-- =============================================================================
--
-- Code annotations:
-- This module is documented according to LuaCATS (Lua Comment and Type System).
-- LuaCATS comments are prefixed with three dashes (---) and use Markdown syntax.
-- For a full list of annotations, see the following link:
-- https://luals.github.io/wiki/annotations/
--
--------------------------------------------------------------------------------


---The Wikipedia link to a language in that language is the same as the language's
---endonym by default. Adjust it manually here, if necessary.
local wplinks = {
	-- format: ["language code"] = "local Wikipedia article",
	-- example for English: ["en"] = "American English",
	["bg"] = "Български език",
	["da"] = "Dansk (sprog)",
	["de"] = "Deutsche Sprache",
	["es"] = "Idioma español",
	["fi"] = "Suomen kieli",
	["hi"] = "हिन्दी",
	["hu"] = "Magyar nyelv",
	["it"] = "Lingua italiana",
	["lt"] = "Lietuvių kalba",
	["lv"] = "Latviešu valoda",
	["pl"] = "Język polski",
	["pt"] = "Língua portuguesa",
	["ru"] = "Русский язык",
	["sk"] = "Slovenčina",
	["th"] = "ภาษาไทย",
	["uk"] = "Українська мова",
	["zh"] = "中文",
}

---The "countrycode" data of a language is the ISO 3166-1 code of the country
---that the language is commonly associated with.
---By default that is the same as the language's ISO 639-1/639-3 language code.
---Adjust it manually here, if necessary.
local countrycodes = {
	-- format: ["language code"] = "country code",
	-- example for English: ["en"] = "us",
	["ar"] = "eg",
	["be"] = "by",
	["cs"] = "cz",
	["da"] = "dk",
	["el"] = "gr",
	["en"] = "us",
	["hi"] = "in",
	["ja"] = "jp",
	["ko"] = "kr",
	["sv"] = "se",
	["uk"] = "ua",
	["vi"] = "vn",
	["yue"] = "cn",
	["zh"] = "cn",
}

local preciselanguagecodes = {
	-- format: ["language code"] = "precise language codes",
	-- example for Chinese Simplified: ["zh"] = "zh-hans",
	["zh"] = "zh-hans",
}

local trim = mw.text.trim

---Holds the `fullname`, `shortname`, and `lname` data for a lot of languages.
---@type table<string, table<string, string>>
local data = mw.loadData('Module:LangInfo/data')

---Holds the arguments from the template call.
---@type table<string, string>
local args_table

---Return a trimmed version of the value of the template parameter with the specified `key`.
---Return `nil` if the parameter is empty or unset.
---@param key string|integer
---@return string?
local getArg = function(key)
	local value = args_table[key]
	if not value then
		return nil
	end
	value = trim(value)
	if value == '' then
		return nil
	end
	return value
end

--------------------------------------------------------------------------------
---Main return object
local p = {}

---Return the queried piece of information about a language.
---@param frame table Interface to the parser (`mw.frame`)
---@return string
p.go = function(frame, args)
	args_table = args or frame.args  -- global input args cache

	local _arg1 = getArg(1) or ''
	local _arg2 = getArg(2) or ''

	if not data[_arg1] then
		return ''
	end

	if _arg2 == 'name' then
		if data[_arg1]['shortname'] ~= '' then
			return data[_arg1]['shortname']
		else
			return data[_arg1]['fullname']
		end

	elseif _arg2 == 'countrycode' then
		if countrycodes[_arg1] ~= nil then
			return countrycodes[_arg1]
		else
			-- default country code: same as language code
			return _arg1
		end

	elseif _arg2 == 'languagecode' then
		if preciselanguagecodes[_arg1] ~= nil then
			return preciselanguagecodes[_arg1]
		else
			-- default true language code: same as language code
			return _arg1
		end

	elseif _arg2 == 'lname' or _arg2 == 'wplink' then
		if _arg2 == 'wplink' and wplinks[_arg1] ~= nil then
			return wplinks[_arg1]
		else
			-- only get the first element of a potential comma-separated list
			-- (the parentheses are there to drop the second return value of `string.gsub`)
			return ( string.gsub(data[_arg1]['lname'], "(.-),.+", "%1") )
		end

	elseif data[_arg1][_arg2] ~= '' then
		return data[_arg1][_arg2]
	end

	return ''
end

return p