Content deleted Content added
m fix module error for languages with no data |
sync from sandbox: include explicitly given script subtag in language tagging |
||
| (38 intermediate revisions by 10 users not shown) | |||
Line 1:
require('
local m_data = mw.loadData("Module:
local langData = m_data.languages or m_data
local p = {}
local function
if value == "" then
return nil
else
return value
end
end
Line 35 ⟶ 32:
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
Line 48 ⟶ 59:
end
local function getCodes(codes
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("^
languageCode, scriptCode =
codes:match("^
if not languageCode then
errorText = '<code>'..codes..'</code> is not a valid language or script code.'
Line 71 ⟶ 104:
)
end
elseif codes:find("^%a%a%a?") then
languageCode = string.lower(languageCode)
errorText = '<code>'..invalidCode..'</code> is not a valid script code.'
elseif codes:find("%-?%a%a%a%a
invalidCode, scriptCode = codes:match("(.*)%-?(%a%a%a%a)
scriptCode = gsub(
scriptCode,
Line 108 ⟶ 118:
else
errorText = '<code>'..codes..'</code> is not a valid language or script code.'
end
if errorText then
Line 121 ⟶ 128:
end
local function tag(text, languageCode, script,
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
if script and script ~= "" then
languageCode = languageCode .. "-" .. script
end
if not text then text = "[text?]" end
Line 136 ⟶ 147:
local out = { textDirectionMarkers[2] }
if italicize then
table.insert(out, "<i lang=\"" .. languageCode
else
table.insert(out, "<span
end
table.insert(out, textDirectionMarkers[3])
return table.concat(out)
end
Line 166 ⟶ 162:
if data and data.name then
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("
.. " 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) == "*"
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")
Line 193 ⟶ 213:
local args = parent.args[1] and parent.args or frame.args
local codes = args[1]
local word1 = ifNotEmpty(args[2]
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
local languageCode, scriptCode, errorText = getCodes(codes)
local italics = args.italics or args.i or args.italic
italics = not (italics == "n" or italics == "-" or italics == "no")
local entry, linkText
if
entry = makeEntryName(word1, languageCode)
linkText = word2
elseif
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,
elseif entry and linkText then
out = linkToWiktionary(entry, linkText)
Line 224 ⟶ 250:
else
return errorText or error("The function wiktlang generated nothing")
end
end
| |||