Difference between revisions of "Module:Language/sandbox"

From blackwiki
Jump to navigation Jump to search
blackwiki>Erutuon
(return text as long as it's provided, but return distinct errors when codes are missing or invalid)
m (15 revisions imported)
 
(One intermediate revision by one other user not shown)
Line 1: Line 1:
 
require('Module:No globals')
 
require('Module:No globals')
 +
local m_data = mw.loadData("Module:Language/data")
 +
local langData = m_data.languages or m_data
  
 
local p = {}
 
local p = {}
 
local gsub = mw.ustring.gsub
 
local find = mw.ustring.find
 
local match = mw.ustring.match
 
local lower = mw.ustring.lower
 
local upper = mw.ustring.upper
 
local U = mw.ustring.char
 
 
--[[ Name is the "canonical name" used on Wiktionary. Article is the Wikipedia article. Script is the ISO 15924 code. ]]
 
 
local languages = {
 
["ar"] = {
 
["name"] = "Arabic",
 
["article"] = "Arabic language",
 
["scripts"] = { "Arab" },
 
--[[ Dagger ālif is replaced by full-size ālif;
 
fatḥatan, ḍammatan, kasratan, fatḥa, ḍamma, kasra, madda, and sukūn are removed. ]]
 
["direction"] = "rtl", -- Should be in the script data module.
 
["replacements"] = {
 
[U(0x0671)] = U(0x0627),
 
[U(0x064B)] = "",
 
[U(0x064C)] = "",
 
[U(0x064D)] = "",
 
[U(0x064E)] = "",
 
[U(0x064F)] = "",
 
[U(0x0650)] = "",
 
[U(0x0651)] = "",
 
[U(0x0652)] = "",
 
[U(0x0670)] = "",
 
[U(0x0640)] = "",
 
},
 
},
 
["bn"] = {
 
["name"] = "Bengali",
 
["article"] = "Bengali language",
 
["scripts"] = { "Beng" },
 
},
 
["de"] = {
 
["name"] = "German",
 
["article"] = "German language",
 
["scripts"] = { "Latn" },
 
--[[
 
["replacements"] = {
 
["ae"]    = "ä",
 
["oe"]    = "ö",
 
["ue"]    = "ü",
 
["A[Ee]"] = "Ä",
 
["O[Ee]"] = "Ö",
 
["U[Ee]"] = "Ü",
 
},
 
]]
 
},
 
["en"] = {
 
["name"] = "English",
 
["article"] = "English language",
 
["scripts"] = { "Latn" },
 
},
 
["es"] = {
 
["name"] = "Spanish",
 
["article"] = "Spanish language",
 
["scripts"] = { "Latn" },
 
},
 
["fr"] = {
 
["name"] = "French",
 
["article"] = "French language",
 
["scripts"] = { "Latn" },
 
},
 
["frm"] = {
 
["name"] = "Middle French",
 
["article"] = "Middle French",
 
["scripts"] = { "Latn" },
 
},
 
["grc"] = {
 
["name"] = "Ancient Greek",
 
["article"] = "Ancient Greek",
 
["scripts"] = { "Grek" },
 
["replacements"] = {
 
-- Vowels with macrons or breves are replaced with plain letters.
 
["[ᾱᾰ]"] = "α",
 
["[ᾹᾸ]"] = "Α",
 
["[ῑῐ]"] = "ι",
 
["[ῙῘ]"] = "Ι",
 
["[ῡῠ]"] = "υ",
 
["[ῩῨ]"] = "Υ",
 
["ϑ"]    = "θ",
 
["ϱ"]    = "ρ"
 
},
 
},
 
["hi"] = {
 
["name"] = "Hindi",
 
["article"] = "Hindi",
 
["scripts"] = { "Deva" },
 
},
 
["ja"] = {
 
["name"] = "Japanese",
 
["article"] = "Japanese language",
 
["scripts"] = { "Jpan" },
 
},
 
["la"] = {
 
["name"] = "Latin",
 
["article"] = "Latin",
 
["scripts"] = { "Latn" },
 
["replacements"] = {
 
-- Vowels with macrons, breves, or diaereses are replaced with plain letters.
 
["[ĀĂ]"]  = "A",
 
["[āă]"]  = "a",
 
["[ĒĔ]"]  = "E",
 
["[ēĕë]"] = "e",
 
["[ĪĬÏ]"] = "I",
 
["[īĭï]"] = "i",
 
["[ŌŎ]"]  = "O",
 
["[ōŏ]"]  = "o",
 
["[ŪŬÜ]"] = "U",
 
["[ūŭü]"] = "u",
 
["Ȳ"]    = "Y",
 
["ȳ"]    = "y"
 
},
 
},
 
["mul"] = {
 
["name"] = "Translingual",
 
["article"] = "",
 
["script"] = { "" },
 
},
 
["pt"] = {
 
["name"] = "Portuguese",
 
["article"] = "Portuguese language",
 
["scripts"] = { "Latn" },
 
},
 
["pa"] = {
 
["name"] = "Punjabi",
 
["article"] = "Punjabi language",
 
["scripts"] = { "Guru", "Arab", }
 
},
 
["ru"] = {
 
["name"] = "Russian",
 
["article"] = "Russian language",
 
["scripts"] = { "Cyrl" },
 
-- Combining acute accent is removed.
 
["replacements"] = { [U(0x0301)] = "", }
 
},
 
["ur"] = {
 
["name"] = "Urdu",
 
["article"] = "Urdu",
 
["scripts"] = { "Arab" },
 
},
 
["zh"] = {
 
["name"] = "Chinese",
 
["article"] = "Chinese language",
 
["scripts"] = { "Hani" },
 
},
 
}
 
 
--[[
 
 
[""] = {
 
["name"] = "",
 
["article"] = "",
 
["script"] = { "" },
 
},
 
 
[""] = {
 
["name"] = "",
 
["article"] = "",
 
["script"] = { "" },
 
["replacements"] = {
 
},
 
},
 
 
]]
 
  
 
local function checkForString(variable)
 
local function checkForString(variable)
Line 177: Line 10:
  
 
local function makeLinkedName(languageCode)
 
local function makeLinkedName(languageCode)
local data = languages[languageCode]
+
local data = langData[languageCode]
 
local article = data["article"]
 
local article = data["article"]
local name = data["name"]
+
local name = data["Wikipedia_name"] or data["name"]
 
return "[[" .. article .. "|" .. name .. "]]: "
 
return "[[" .. article .. "|" .. name .. "]]: "
 
end
 
end
  
 
local function makeEntryName(word, languageCode)
 
local function makeEntryName(word, languageCode)
local data = languages[languageCode]
+
local data = langData[languageCode]
 +
local ugsub = mw.ustring.gsub
 
word = tostring(word)
 
word = tostring(word)
 
if word == nil then
 
if word == nil then
Line 192: Line 26:
 
else
 
else
 
-- Remove bold and italics, so that words that contain bolding or emphasis can be linked without piping.
 
-- Remove bold and italics, so that words that contain bolding or emphasis can be linked without piping.
word = gsub(word, "\'\'\'", "")
+
word = word:gsub("\'\'\'", "")
word = gsub(word, "\'\'", "")
+
word = word:gsub("\'\'", "")
 
if data == nil then
 
if data == nil then
 
return word
 
return word
Line 201: Line 35:
 
return word
 
return word
 
else
 
else
for regex, replacement in pairs(replacements) do
+
-- Decompose so that the diacritics of characters such
word = gsub(word, regex, replacement)
+
-- as á can be removed in one go.
 +
-- No need to compose at the end, because the MediaWiki software
 +
-- will handle that.
 +
if replacements.decompose then
 +
word = mw.ustring.toNFD(word)
 +
for i, from in ipairs(replacements.from) do
 +
word = ugsub(
 +
word,
 +
from,
 +
replacements.to and replacements.to[i] or "")
 +
end
 +
else
 +
for regex, replacement in pairs(replacements) do
 +
word = ugsub(word, regex, replacement)
 +
end
 
end
 
end
 
return word
 
return word
Line 208: Line 56:
 
end
 
end
 
end
 
end
 +
end
 +
 +
p.makeEntryName = makeEntryName
 +
 +
local function fixScriptCode(firstLetter, threeLetters)
 +
return string.upper(firstLetter) .. string.lower(threeLetters)
 
end
 
end
  
 
local function getCodes(codes, text)
 
local function getCodes(codes, text)
local languageCode, scriptCode
+
local languageCode, scriptCode, invalidCode
local errorText = ""
+
local errorText
if find(codes, "^%s*%a%a%a?%s*$") or find(codes, "^%s*%a%a%a?-%a%a%a%a%s*$") then
+
if codes == nil or codes == "" then
 +
errorText = 'no language or script code provided'
 +
elseif codes:find("^%a%a%a?$") or codes:find("^%a%a%a?%-%a%a%a%a$") then
 
-- A three- or two-letter lowercase sequence at beginning of first parameter
 
-- A three- or two-letter lowercase sequence at beginning of first parameter
languageCode = find(codes, "^%s*%a%a%a?") and (match(codes, "^%s*(%l%l%l?)") or gsub(match(codes, "^%s*(%a%a%a?)"), "(%a%a%a?)", function(a) return lower(a) end, 1) )
+
languageCode =
 +
codes:find("^%a%a%a?") and (
 +
codes:match("^(%l%l%l?)")
 +
or codes:match("^(%a%a%a?)")
 +
:gsub("(%a%a%a?)", string.lower, 1)
 +
)
 
-- One uppercase and three lowercase letters at the end of the first parameter
 
-- One uppercase and three lowercase letters at the end of the first parameter
scriptCode = find(codes, "%a%a%a%a%s*$") and (match(codes, "(%u%l%l%l)%s*$") or gsub(match(codes, "(%a%a%a%a)%s*$"), "(%a)(%a%a%a)", function(a, b) return upper(a) .. lower(b) end, 1) ) or require("Module:Language/scripts").isLatn(text) and "Latn" or "unknown"
+
scriptCode =
elseif find(codes, "^%s*%a%a%a?") then
+
codes:find("%a%a%a%a$") and (
languageCode = gsub(match(codes, "%a%a%a?"), "(%a%a%a?)", function(a) return lower(a) end, 1)
+
codes:match("(%u%l%l%l)$")
local invalidCode = gsub(codes, "^%s*%a%a%a?%-?", "")
+
or gsub(
errorText = ' <span style="font-size: smaller">[<code>'..invalidCode..'</code> is not a valid script code.]</span>'
+
codes:match("(%a%a%a%a)$"),
elseif find(codes, "%-?%a%a%a%a%s*$") then
+
"(%a)(%a%a%a)",
scriptCode = gsub(match(codes, "%a%a%a%a"), "%a%a%a%a", function(a) return lower(a) end, 1)
+
fixScriptCode,
local invalidCode = gsub(codes, "%s*%a%a%a%a%-?$", "")
+
1
errorText = ' <span style="font-size: smaller">[<code>'..invalidCode..'</code> is not a valid language code.]</span>'
+
)
elseif codes == nil or codes == "" then
+
)
errorText = ' <span style="font-size: smaller">[no language or script code provided]</span>'
+
elseif codes:find("^%a%a%a?%-%a%a%a?$")
 +
or codes:find("^%a%a%a%-%a%a%a%-%a%a%a$") then
 +
languageCode = codes
 +
 +
-- Private-use subtag: x followed by one or more sequences of 1-8 lowercase
 +
-- letters separated by hyphens. This only allows for one sequence, as it is
 +
-- needed for proto-languages such as ine-x-proto (Proto-Indo-European).
 +
elseif codes:find("^%a%a%a?%-x%-%a%a?%a?%a?%a?%a?%a?%a?$") then
 +
languageCode, scriptCode =
 +
codes:match("^(%a%a%a%-x%-%a%a?%a?%a?%a?%a?%a?%a?)%-?(.*)$")
 +
if not languageCode then
 +
errorText = '<code>'..codes..'</code> is not a valid language or script code.'
 +
elseif scriptCode ~= "" and not scriptCode:find("%a%a%a%a") then
 +
errorText = '<code>'..scriptCode..'</code> is not a valid script code.'
 +
else
 +
scriptCode = scriptCode:gsub(
 +
"(%a)(%a%a%a)",
 +
fixScriptCode,
 +
1
 +
)
 +
end
 +
elseif codes:find("^%a%a%a?") then
 +
languageCode, invalidCode = codes:match("^(%a%a%a?)%-?(.*)")
 +
languageCode = string.lower(languageCode)
 +
errorText = '<code>'..invalidCode..'</code> is not a valid script code.'
 +
elseif codes:find("%-?%a%a%a%a$") then
 +
invalidCode, scriptCode = codes:match("(.*)%-?(%a%a%a%a)$")
 +
scriptCode = gsub(
 +
scriptCode,
 +
"(%a)(%a%a%a)",
 +
fixScriptCode
 +
)
 +
errorText = '<code>'..invalidCode..'</code> is not a valid language code.'
 +
else
 +
errorText = '<code>'..codes..'</code> is not a valid language or script code.'
 +
end
 +
if not scriptCode or scriptCode == "" then
 +
scriptCode = require("Module:Unicode data").is_Latin(text) and "Latn" or "unknown"
 +
end
 +
if errorText then
 +
errorText = ' <span style="font-size: smaller">[' .. errorText .. ']</span>'
 
else
 
else
errorText = ' <span style="font-size: smaller">[<code>'..codes..'</code> is not a valid language or script code.]</span>'
+
errorText = ""
 
end
 
end
 +
languageCode = m_data.redirects[languageCode] or languageCode
 
return languageCode, scriptCode, errorText
 
return languageCode, scriptCode, errorText
 
end
 
end
  
 
local function tag(text, languageCode, script, italics)
 
local function tag(text, languageCode, script, italics)
local data = languages[languageCode]
+
local data = langData[languageCode]
 +
-- Use Wikipedia code if it has been given: for instance,
 +
-- Proto-Indo-European has the Wiktionary code "ine-pro" but the Wikipedia
 +
-- code "ine-x-proto".
 +
languageCode = data and data.Wikipedia_code or languageCode
 +
 
local italicize = script == "Latn" and italics
 
local italicize = script == "Latn" and italics
 +
 
if not text then text = "[text?]" end
 
if not text then text = "[text?]" end
local textDirectionMarkers = data and data["direction"] == "rtl" and { ' dir="rtl"', '&lrm;' } or { "", "" }
+
local out = italicize and "<i lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode  .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</i>" or "<span lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</span>" .. textDirectionMarkers[2]
+
local textDirectionMarkers = { "", "", "" }
return out
+
if data and data["direction"] == "rtl" then
 +
textDirectionMarkers = { ' dir="rtl"', '&rlm;', '&lrm;' }
 +
end
 +
 +
local out = { textDirectionMarkers[2] }
 +
if italicize then
 +
table.insert(out, "<i lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode  .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</i>")
 +
else
 +
table.insert(out, "<span lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</span>")
 +
end
 +
table.insert(out, textDirectionMarkers[3])
 +
 +
return table.concat(out)
 
end
 
end
 +
 +
  
 
function p.lang(frame)
 
function p.lang(frame)
 
local parent = frame:getParent()
 
local parent = frame:getParent()
 
local args = parent.args[1] and parent.args or frame.args
 
local args = parent.args[1] and parent.args or frame.args
local codes = args[1]
+
 +
local codes = args[1] and mw.text.trim(args[1])
 
local text = args[2] or error("Provide text in the second parameter")
 
local text = args[2] or error("Provide text in the second parameter")
 +
 
local languageCode, scriptCode, errorText = getCodes(codes, text)
 
local languageCode, scriptCode, errorText = getCodes(codes, text)
local italics = args.italics or args.i
+
italics = not (italics == "n" or italics == "-")
+
local italics = args.italics or args.i or args.italic
 +
italics = not (italics == "n" or italics == "-" or italics == "no")
 
 
 
return tag(text, languageCode, scriptCode, italics) .. errorText
 
return tag(text, languageCode, scriptCode, italics) .. errorText
Line 256: Line 181:
  
 
local function linkToWiktionary(entry, linkText, languageCode)
 
local function linkToWiktionary(entry, linkText, languageCode)
local data = languages[languageCode]
+
local data = langData[languageCode]
 
local name
 
local name
 
if languageCode then
 
if languageCode then
name = data and data.name or mw.language.fetchLanguageName(languageCode, 'en') -- On other languages' wikis, use mw.getContentLanguage():getCode(), or replace with that wiki's language code.
+
if data and data.name then
 +
name = data.name
 +
elseif mw.language.fetchLanguageName(languageCode, 'en') ~= "" then
 +
-- On other languages' wikis, use mw.getContentLanguage():getCode(), or replace with that wiki's language code.
 +
name = mw.language.fetchLanguageName(languageCode, 'en')
 +
else
 +
error("No name for the language " .. ("%q"):format(languageCode or nil) .. " could be found")
 +
end
 +
if entry:sub(1, 1) == "*" then
 +
if name ~= "" then
 +
entry = "Reconstruction:" .. name .. "/" .. entry:sub(2)
 +
else
 +
error("Language name is empty")
 +
end
 +
elseif data and data.type == "reconstructed" then
 +
mw.log("Reconstructed language without asterisk:", languageCode, name, entry)
 +
local frame = mw.getCurrentFrame()
 +
-- Track reconstructed entries with no asterisk by transcluding
 +
-- a nonexistent template. This technique is used in Wiktionary:
 +
-- see [[wikt:Module:debug]].
 +
-- [[Special:WhatLinksHere/tracking/wikt-lang/reconstructed with no asterisk]]
 +
pcall(frame.expandTemplate, frame,
 +
{ title = 'tracking/wikt-lang/reconstructed with no asterisk' })
 +
if name ~= "" then
 +
entry = "Reconstruction:" .. name .. "/" .. entry
 +
else
 +
error("Language name is empty")
 +
end
 +
elseif data and data.type == "appendix" then
 +
if name ~= "" then
 +
entry = "Appendix:" .. name .. "/" .. entry
 +
else
 +
error("Language name is empty")
 +
end
 +
end
 
if entry and linkText then
 
if entry and linkText then
 
return "[[wikt:" .. entry .. "#" .. name .. "|" .. linkText .. "]]"
 
return "[[wikt:" .. entry .. "#" .. name .. "|" .. linkText .. "]]"
Line 270: Line 229:
 
end
 
end
  
function p.wikt(frame)
+
function p.wiktlang(frame)
 
local parent = frame:getParent()
 
local parent = frame:getParent()
 
local args = parent.args[1] and parent.args or frame.args
 
local args = parent.args[1] and parent.args or frame.args
local codes = args[1] or nil
+
local word1 = args[2] or nil
+
local codes = args[1] and mw.text.trim(args[1])
local word2 = args[3] or nil
+
local word1 = args[2]
 +
local word2 = args[3]
 +
 +
if not args[2] then
 +
error("Parameter 2 is required")
 +
end
 +
 
local languageCode, scriptCode, errorText = getCodes(codes, word1)
 
local languageCode, scriptCode, errorText = getCodes(codes, word1)
local errorMessage = errorText
+
 
local italics = args.italics or args.i
 
local italics = args.italics or args.i
 
italics = not (italics == "n" or italics == "-")
 
italics = not (italics == "n" or italics == "-")
 +
 
local entry, linkText
 
local entry, linkText
 
if checkForString(word2) and checkForString(word1) then
 
if checkForString(word2) and checkForString(word1) then
Line 288: Line 254:
 
linkText = word1
 
linkText = word1
 
end
 
end
local out = (languageCode and entry and linkText and tag(linkToWiktionary(entry, linkText, languageCode), languageCode, scriptCode, italics) ) or entry and linkText and linkToWiktionary(entry, linkText) or  '<span style="font-size: smaller;">[text?]</span>'
+
return out and out .. errorMessage or errorMessage or error("The function wikt generated nothing")
+
local out
 +
if languageCode and entry and linkText then
 +
out = tag(linkToWiktionary(entry, linkText, languageCode), languageCode, scriptCode, italics)
 +
elseif entry and linkText then
 +
out = linkToWiktionary(entry, linkText)
 +
else
 +
out = '<span style="font-size: smaller;">[text?]</span>'
 +
end
 +
 +
if out and errorText then
 +
return out .. errorText
 +
else
 +
return errorText or error("The function wiktlang generated nothing")
 +
end
 +
end
 +
 
 +
function p.wikt(frame)
 +
local parent = frame:getParent()
 +
local args = parent.args[1] and parent.args or frame.args
 +
 +
local codes = args[1] and mw.text.trim(args[1])
 +
local word1 = args[2]
 +
local word2 = args[3]
 +
 +
if not word1 then
 +
error("Provide a word in parameter 2.")
 +
end
 +
 +
local languageCode, scriptCode, errorText = getCodes(codes, word1)
 +
 +
local entry, linkText
 +
if checkForString(word2) and checkForString(word1) then
 +
entry = makeEntryName(word1, languageCode)
 +
linkText = word2
 +
elseif checkForString(word1) then
 +
entry = makeEntryName(word1, languageCode)
 +
linkText = word1
 +
end
 +
 +
local out
 +
if languageCode and entry and linkText then
 +
out = linkToWiktionary(entry, linkText, languageCode)
 +
elseif entry and linkText then
 +
out = linkToWiktionary(entry, linkText)
 +
else
 +
out = '<span style="font-size: smaller;">[text?]</span>'
 +
end
 +
 +
if out and errorText then
 +
return out and out .. errorText
 +
else
 +
return errorText or error("The function wikt generated nothing")
 +
end
 
end
 
end
  
 
return p
 
return p

Latest revision as of 05:55, 27 September 2020

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

require('Module:No globals')
local m_data = mw.loadData("Module:Language/data")
local langData = m_data.languages or m_data

local p = {}

local function checkForString(variable)
	return variable ~= "" and variable ~= nil
end

local function makeLinkedName(languageCode)
	local data = langData[languageCode]
	local article = data["article"]
	local name = data["Wikipedia_name"] or data["name"]
	return "[[" .. article .. "|" .. name .. "]]:&nbsp;"
end

local function makeEntryName(word, languageCode)
	local data = langData[languageCode]
	local ugsub = mw.ustring.gsub
	word = tostring(word)
	if word == nil then
		error("The function makeEntryName requires a string argument")
	elseif word == "" then
		return ""
	else
		-- Remove bold and italics, so that words that contain bolding or emphasis can be linked without piping.
		word = word:gsub("\'\'\'", "")
		word = word:gsub("\'\'", "")
		if data == nil then
			return word
		else
			local replacements = data and data["replacements"]
			if replacements == nil then
				return word
			else
				-- Decompose so that the diacritics of characters such
				-- as á can be removed in one go.
				-- No need to compose at the end, because the MediaWiki software
				-- will handle that.
				if replacements.decompose then
					word = mw.ustring.toNFD(word)
					for i, from in ipairs(replacements.from) do
						word = ugsub(
							word,
							from,
							replacements.to and replacements.to[i] or "")
					end
				else
					for regex, replacement in pairs(replacements) do
						word = ugsub(word, regex, replacement)
					end
				end
				return word
			end
		end
	end
end

p.makeEntryName = makeEntryName

local function fixScriptCode(firstLetter, threeLetters)
	return string.upper(firstLetter) .. string.lower(threeLetters)
end

local function getCodes(codes, text)
	local languageCode, scriptCode, invalidCode
	local errorText
	if codes == nil or codes == "" then
		errorText = 'no language or script code provided'
	elseif codes:find("^%a%a%a?$") or codes:find("^%a%a%a?%-%a%a%a%a$") then
		-- A three- or two-letter lowercase sequence at beginning of first parameter
		languageCode =
			codes:find("^%a%a%a?") and (
				codes:match("^(%l%l%l?)")
				or codes:match("^(%a%a%a?)")
					:gsub("(%a%a%a?)", string.lower, 1)
			)
		-- One uppercase and three lowercase letters at the end of the first parameter
		scriptCode =
			codes:find("%a%a%a%a$") and (
				codes:match("(%u%l%l%l)$")
				or gsub(
					codes:match("(%a%a%a%a)$"),
					"(%a)(%a%a%a)",
					fixScriptCode,
					1
				)
			)
	elseif codes:find("^%a%a%a?%-%a%a%a?$")
	or codes:find("^%a%a%a%-%a%a%a%-%a%a%a$") then
		languageCode = codes
	
	-- Private-use subtag: x followed by one or more sequences of 1-8 lowercase
	-- letters separated by hyphens. This only allows for one sequence, as it is
	-- needed for proto-languages such as ine-x-proto (Proto-Indo-European).
	elseif codes:find("^%a%a%a?%-x%-%a%a?%a?%a?%a?%a?%a?%a?$") then
		languageCode, scriptCode =
			codes:match("^(%a%a%a%-x%-%a%a?%a?%a?%a?%a?%a?%a?)%-?(.*)$")
		if not languageCode then
			errorText = '<code>'..codes..'</code> is not a valid language or script code.'
		elseif scriptCode ~= "" and not scriptCode:find("%a%a%a%a") then
			errorText = '<code>'..scriptCode..'</code> is not a valid script code.'
		else
			scriptCode = scriptCode:gsub(
				"(%a)(%a%a%a)",
				fixScriptCode,
				1
			)
		end
	elseif codes:find("^%a%a%a?") then
		languageCode, invalidCode = codes:match("^(%a%a%a?)%-?(.*)")
		languageCode = string.lower(languageCode)
		errorText = '<code>'..invalidCode..'</code> is not a valid script code.'
	elseif codes:find("%-?%a%a%a%a$") then
		invalidCode, scriptCode = codes:match("(.*)%-?(%a%a%a%a)$")
		scriptCode = gsub(
			scriptCode,
			"(%a)(%a%a%a)",
			fixScriptCode
		)
		errorText = '<code>'..invalidCode..'</code> is not a valid language code.'
	else
		errorText = '<code>'..codes..'</code> is not a valid language or script code.'
	end
	if not scriptCode or scriptCode == "" then
		scriptCode = require("Module:Unicode data").is_Latin(text) and "Latn" or "unknown"
	end
	if errorText then
		errorText = ' <span style="font-size: smaller">[' .. errorText .. ']</span>'
	else
		errorText = ""
	end
	languageCode = m_data.redirects[languageCode] or languageCode
	return languageCode, scriptCode, errorText
end

local function tag(text, languageCode, script, italics)
	local data = langData[languageCode]
	-- Use Wikipedia code if it has been given: for instance,
	-- Proto-Indo-European has the Wiktionary code "ine-pro" but the Wikipedia
	-- code "ine-x-proto".
	languageCode = data and data.Wikipedia_code or languageCode
	
	local italicize = script == "Latn" and italics
	
	if not text then text = "[text?]" end
	
	local textDirectionMarkers = { "", "", "" }
	if data and data["direction"] == "rtl" then
		textDirectionMarkers = { ' dir="rtl"', '&rlm;', '&lrm;' }
	end
	
	local out = { textDirectionMarkers[2] }
	if italicize then
		table.insert(out, "<i lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode  .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</i>")
	else
		table.insert(out, "<span lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</span>")
	end
	table.insert(out, textDirectionMarkers[3])
	
	return table.concat(out)
end



function p.lang(frame)
	local parent = frame:getParent()
	local args = parent.args[1] and parent.args or frame.args
	
	local codes = args[1] and mw.text.trim(args[1])
	local text = args[2] or error("Provide text in the second parameter")
	
	local languageCode, scriptCode, errorText = getCodes(codes, text)
	
	local italics = args.italics or args.i or args.italic
	italics = not (italics == "n" or italics == "-" or italics == "no")
	
	return tag(text, languageCode, scriptCode, italics) .. errorText
end

local function linkToWiktionary(entry, linkText, languageCode)
	local data = langData[languageCode]
	local name
	if languageCode then
		if data and data.name then
			name = data.name
		elseif mw.language.fetchLanguageName(languageCode, 'en') ~= "" then
			-- On other languages' wikis, use mw.getContentLanguage():getCode(), or replace with that wiki's language code.
			name = mw.language.fetchLanguageName(languageCode, 'en')
		else
			error("No name for the language " .. ("%q"):format(languageCode or nil) .. " could be found")
		end
		if entry:sub(1, 1) == "*" then
			if name ~= "" then
				entry = "Reconstruction:" .. name .. "/" .. entry:sub(2)
			else
				error("Language name is empty")
			end
		elseif data and data.type == "reconstructed" then
			mw.log("Reconstructed language without asterisk:", languageCode, name, entry)
			local frame = mw.getCurrentFrame()
			-- Track reconstructed entries with no asterisk by transcluding
			-- a nonexistent template. This technique is used in Wiktionary:
			-- see [[wikt:Module:debug]].
			-- [[Special:WhatLinksHere/tracking/wikt-lang/reconstructed with no asterisk]]
			pcall(frame.expandTemplate, frame,
				{ title = 'tracking/wikt-lang/reconstructed with no asterisk' })
			if name ~= "" then
				entry = "Reconstruction:" .. name .. "/" .. entry
			else
				error("Language name is empty")
			end
		elseif data and data.type == "appendix" then
			if name ~= "" then
				entry = "Appendix:" .. name .. "/" .. entry
			else
				error("Language name is empty")
			end
		end
		if entry and linkText then
			return "[[wikt:" .. entry .. "#" .. name .. "|" .. linkText .. "]]"
		else
			error("linkToWiktionary needs a Wiktionary entry or link text, or both")
		end
	else
		return "[[wikt:" .. entry .. "|" .. linkText .. "]]"
	end
end

function p.wiktlang(frame)
	local parent = frame:getParent()
	local args = parent.args[1] and parent.args or frame.args
	
	local codes = args[1] and mw.text.trim(args[1])
	local word1 = args[2]
	local word2 = args[3]
	
	if not args[2] then
		error("Parameter 2 is required")
	end
	
	local languageCode, scriptCode, errorText = getCodes(codes, word1)
	
	local italics = args.italics or args.i
	italics = not (italics == "n" or italics == "-")
	
	local entry, linkText
	if checkForString(word2) and checkForString(word1) then
		entry = makeEntryName(word1, languageCode)
		linkText = word2
	elseif checkForString(word1) then
		entry = makeEntryName(word1, languageCode)
		linkText = word1
	end
	
	local out
	if languageCode and entry and linkText then
		out = tag(linkToWiktionary(entry, linkText, languageCode), languageCode, scriptCode, italics)
	elseif entry and linkText then
		out = linkToWiktionary(entry, linkText)
	else
		out = '<span style="font-size: smaller;">[text?]</span>'
	end
	
	if out and errorText then
		return out .. errorText
	else
		return errorText or error("The function wiktlang generated nothing")
	end
end

function p.wikt(frame)
	local parent = frame:getParent()
	local args = parent.args[1] and parent.args or frame.args
	
	local codes = args[1] and mw.text.trim(args[1])
	local word1 = args[2]
	local word2 = args[3]
	
	if not word1 then
		error("Provide a word in parameter 2.")
	end
	
	local languageCode, scriptCode, errorText = getCodes(codes, word1)
	
	local entry, linkText
	if checkForString(word2) and checkForString(word1) then
		entry = makeEntryName(word1, languageCode)
		linkText = word2
	elseif checkForString(word1) then
		entry = makeEntryName(word1, languageCode)
		linkText = word1
	end
	
	local out
	if languageCode and entry and linkText then
		out = linkToWiktionary(entry, linkText, languageCode) 
	elseif entry and linkText then
		out = linkToWiktionary(entry, linkText)
	else
		out = '<span style="font-size: smaller;">[text?]</span>'
	end
	
	if out and errorText then
		return out and out .. errorText
	else
		return errorText or error("The function wikt generated nothing")
	end
end

return p