Wikipedia

Module:Wikt-lang: Difference between revisions

Browse history interactively
← Previous edit
Content deleted Content added
trying to make checking for validity less taxing
sync from sandbox: include explicitly given script subtag in language tagging
 
(78 intermediate revisions by 10 users not shown)
Line 1:
require('Module:No globalsstrict')
local m_data = mw.loadData("Module:Wikt-lang/data")
local langData = m_data.languages or m_data
 
local p = {}
 
local gsubfunction = mw.ustring.gsubifNotEmpty(value)
if value == "" then
local find = mw.ustring.find
return nil
local match = mw.ustring.match
else
local lower = mw.ustring.lower
return value
local upper = mw.ustring.upper
end
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)
return variable ~= "" and variable ~= nil
end
 
local function generatePrefix(languageCode)
local data = languages[languageCode]
local article = data["article"]
local name = data["name"]
return "[[" .. article .. "|" .. name .. "]]: "
end
 
local function makeEntryName(word, languageCode)
local data = languageslangData[languageCode]
local ugsub = mw.ustring.gsub
word = tostring(word)
if word == nil then
Line 192 ⟶ 23:
else
-- Remove bold and italics, so that words that contain bolding or emphasis can be linked without piping.
word = word:gsub(word, "\'\'\'", "")
word = word:gsub(word, "\'\'", "")
if data == nil then
return word
Line 201 ⟶ 32:
return word
else
-- Decompose so that the diacritics of characters such
for regex, replacement in pairs(replacements) do
-- as á can be removed in one go.
word = gsub(word, regex, replacement)
-- 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
Line 210 ⟶ 55:
end
 
local function tagfixScriptCode(textfirstLetter, languageCode, script, italicsthreeLetters)
return string.upper(firstLetter) .. string.lower(threeLetters)
local data = languages[languageCode]
end
local script = script or data and data["scripts"][1] or "unknown"
 
local italicize = ( (script == "unknown" and italics == "yes") or (script == "Latn" and italics ~= "no") ) and "yes" or "no"
local function getCodes(codes)
local textDirectionMarkers = data and data["direction"] == "rtl" and { ' dir="rtl"', '‎' } or { "", "" }
local languageCode, scriptCode, invalidCode
local out = italicize == "no" and "<span lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</span>" or italicize == "yes" and "<i lang=\"" .. languageCode .. "\" xml:lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</i>"
local errorText
if data and data["direction"] == "rtl" then
if codes == nil or codes == "" then
out = out .. textDirectionMarkers[2]
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 errorText then
errorText = ' <span style="font-size: smaller">[' .. errorText .. ']</span>'
else
errorText = ""
end
languageCode = m_data.redirects[languageCode] or languageCode
return out
return languageCode, scriptCode, errorText
end
 
local function tag(text, languageCode, script, italicize)
function p.lang(frame)
local data = langData[languageCode]
-- A three- or two-letter lowercase sequence at beginning of first parameter
-- Use Wikipedia code if it has been given: for instance,
local languageCode = match(frame.args[1], "^%s*(%l%l%l)") or match(frame.args[1], "^%s*(%l%l)") or error("Place a valid language code in the first parameter of {{\[\[Template:Lang/sandbox|lang/sandbox\]\]}}")
-- Proto-Indo-European has the Wiktionary code "ine-pro" but the Wikipedia
-- One uppercase and three lowercase letters at the end of the first parameter
-- code "ine-x-proto".
local scriptCode = match(frame.args[1], "(%u%l%l%l)%s*$") or nil
languageCode = data and data.Wikipedia_code or languageCode
local text = frame.args[2] or error("Provide text in the second parameter")
if script and script ~= "" then
local italics = frame.args["italics"] or frame.args["i"] or "auto"
languageCode = languageCode .. "-" .. script
end
 
if not text then text = "[text?]" end
local textDirectionMarkers = { "", "", "" }
return tag(text, languageCode, scriptCode, italics)
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 .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</i>")
else
table.insert(out, "<span lang=\"" .. languageCode .. "\"" .. textDirectionMarkers[1] .. ">" .. text .. "</span>")
end
table.insert(out, textDirectionMarkers[3])
return table.concat(out)
end
 
local function linkToWiktionary(entry, linkText, languageCode, script, italics)
local data = languageslangData[languageCode]
local name
if languageCode then
if data and data.name 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.
name = data.name
else
-- On other languages' wikis, use mw.getContentLanguage():getCode(),
-- or replace 'en' with that wiki's language code.
name = mw.language.fetchLanguageName(languageCode, 'en')
if name == "" then
error("Name for the language code " .. ("%q"):format(languageCode or nil)
.. " could not be retrieved with mw.language.fetchLanguageName, "
.. "so it should be added to [[Module:Wikt-lang/data]]")
end
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 tag("[[wikt:" .. entry .. "#" .. name .. "|" .. linkText .. "]]", languageCode, script, italics)
else
error("linkToWiktionary needs a Wiktionary entry or link text, or both")
end
else
return "[[wikt:" .. entry .. "|" .. linkText .. "]]"
error("linkToWiktionary needs a language code")
end
end
 
function p.wiktwiktlang(frame)
local codesparent = frame.args[1]:getParent()
local args = parent.args[1] and parent.args or frame.args
local languageCode, scriptCode
local validity = (find(codes, "^%s*%l%l%l?%s*$") or find(codes, "^%s*%l%l%l?-%u%l%l%l%s*$") ) and "correct capitalization" or (find(codes, "^%s*%a%a%a?%s*$") or find(codes, "^%s*%a%a%a?-%a%a%a%a%s*$") ) and "incorrect capitalization" or "invalid"
local codes = args[1] and mw.text.trim(args[1])
if validity == "invalid" then
local word1 = ifNotEmpty(args[2])
error("Language or script code or codes in first parameter not recognized")
local word2 = ifNotEmpty(args[3])
if not args[2] or '' == args[2] then
return '<span style="color:#d33">[text?] Parameter 2 is required</span>';
end
if codes then
local languageCode, scriptCode, errorText = getCodes(codes)
-- A three- or two-letter lowercase sequence at beginning of first parameter
languageCode = match(codes, "^%s*(%l%l%l)") or match(codes, "^%s*(%l%l)") --[[ or gsub(match(codes, "^%s*(%a%a%a?)"), "^%s*(%a%a%a?)", function(a) return lower(a) end)]] or error("No language code was found in the first parameter")
local italics = args.italics or args.i or args.italic
-- One uppercase and three lowercase letters at the end of the first parameter
italics = not (italics == "n" or italics == "-" or italics == "no")
scriptCode = match(codes, "(%u%l%l%l)%s*$") --[[ or gsub(match(codes, "(%a%a%a%a)%s*$"), "(%a)(%a%a%a)%s*$", function(a, b) return upper(a) .. lower(b) end)]] or nil
local entry, linkText
if word2 and word1 then
entry = makeEntryName(word1, languageCode)
linkText = word2
elseif word1 then
entry = makeEntryName(word1, languageCode)
linkText = word1
end
 
local italicize = italics and (scriptCode == "Latn" or require("Module:Unicode data").is_Latin(linkText))
 
local out
if languageCode and entry and linkText then
out = tag(linkToWiktionary(entry, linkText, languageCode), languageCode, scriptCode, italicize)
elseif entry and linkText then
out = linkToWiktionary(entry, linkText)
else
out = '<span style="font-size: smaller;">[text?]</span>'
error("No content in the first parameter")
end
local word1 = frame.args[2]
if out and errorText then
local word2 = frame.args[3]
return out .. errorText
local italics = frame.args.italics or frame.args.i or "auto"
italics = (italics == "n" or italics == "-") and "no" or italics
local entry, linkText
if languageCode then
if checkForString(word2) and checkForString(word1) then
entry = makeEntryName(word1, languageCode)
linkText = word2
elseif checkForString(word1) then
entry = makeEntryName(word1, languageCode)
linkText = word1
else
error("Please provide a word in the second parameter")
end
else
return errorText or error("The function wiktlang generated nothing")
error("Please provide a language code in the first parameter")
end
return linkToWiktionary(entry, linkText, languageCode, scriptCode, italics)
end
 
Retrieved from "https://en.wikipedia.org/wiki/Module:Wikt-lang"