Module:UrlToWiki
Documentation for this module may be created at Module:UrlToWiki/doc
-- This module takes a URL from a Wikimedia project and returns the equivalent wikitext.
-- Any actions such as edit, history, etc., are stripped, and percent-encoded characters
-- are converted to normal text.
p = {}
local interwiki_table = mw.loadData("Module:InterwikiTable")
local function getHostId(host)
if type(host) ~= "string" then
error("Non-string value for host name detected.")
end
-- Find the entry for the host in the interwiki table.
local id
for k,v in pairs(interwiki_table) do
if mw.ustring.match(host, interwiki_table[k].domain) then
id = k
break
end
end
return id
end
local function getInterwiki(host)
-- This function returns the value of the interwiki prefix.
if type(host) ~= "string" then
error("Non-string value for host name detected.")
end
host = mw.language.getContentLanguage():lc(host)
local host_id = getHostId(host)
if not host_id then return end -- If the host isn't valid then the language and the interwiki are also invalid.
-- Find the language in the interwiki prefix, if applicable.
local lang = mw.ustring.match(host, "^(.-)%.") -- Find the text before the first period.
if not lang or not mw.language.isSupportedLanguage(lang) then -- Check if lang is a valid language code.
lang = false
end
-- A language prefix is not necessary if there is already a language prefix for the host in the interwiki table.
local domain_lang = mw.ustring.match(interwiki_table[host_id].domain, "^(.-)%.") -- Find the text before the first period.
if mw.language.isSupportedLanguage(domain_lang) then
lang = false
end
-- No need for an interwiki link if we are on the same site as the URL.
local current_host = mw.uri.new(mw.title.getCurrentTitle():fullUrl()).host -- Get the host portion of the current page URL.
if host == current_host then
return nil, lang, host_id
end
-- Check if the URL language is the same as the current language.
local same_lang
if lang and lang == mw.ustring.match(current_host, "^(.-)%.") then
same_lang = true
end
-- Check if the project is the same as the current project (but a different language).
local current_host_id = getHostId(current_host)
local same_project
if current_host_id == host_id then
same_project = true
end
-- Return the interwiki prefix, omitting the language or the project code if
-- it is not necessary.
local project = interwiki_table[host_id].iw_prefix[1]
if same_lang or ( not lang and interwiki_table[host_id].takes_lang_prefix == false ) then
return project, lang, host_id
elseif same_project then
return lang, lang, host_id
elseif not lang then -- If the language code is bad but the rest of the host name is ok.
return nil, nil, host_id
else
return project .. ":" .. lang, lang, host_id
end
end
local function _urlToWiki(args)
local url = args[1] or error("No URL specified")
url = mw.text.trim(url) -- Trim whitespace.
url = mw.uri.new(url)
local host = url.host -- Get the host name.
-- Get the interwiki prefix. Lang and host_id are passed through so we don't have to work them out again.
local interwiki, lang, host_id
if host then
interwiki, lang, host_id = getInterwiki(host)
end
local link = true -- This decides whether the resulting wikitext will be linked or not. Default is yes.
if args.link == "no" then
link = false -- Let users override the default.
end
-- Get the page title.
local pagetitle, title_prefix
if host_id and not ( interwiki_table[host_id].takes_lang_prefix == true and not lang ) then
title_prefix = interwiki_table[host_id].title_prefix
end
-- If the URL path starts with the title prefix in the interwiki table, use that to get the title.
if title_prefix and mw.ustring.sub(url.path, 1, mw.ustring.len(title_prefix)) == title_prefix then
pagetitle = mw.ustring.sub(url.path, mw.ustring.len(title_prefix) + 1, -1)
-- Else, if the URL is a history "index.php", use url.query.title. Check for host_id
-- in case the URL isn't of a Wikimedia site.
elseif host_id and mw.ustring.match(url.path, "index%.php") and url.query.title then
pagetitle = url.query.title
-- Special case for Bugzilla.
elseif host_id == "bugzilla" and url.query.id then
pagetitle = url.query.id
elseif host_id == "bugzilla" and not url.query.id then
interwiki = false -- disable the interwiki prefix as we are returning a full URL.
link = false -- don't use double square brackets for URLs.
pagetitle = tostring(url)
-- If the URL is valid but not a recognised interwiki, return the URL and don't link it.
elseif host and not host_id then
link = false -- Don't use double square brackets for URLs.
pagetitle = tostring(url)
-- Otherwise, return whatever text we were passed and link it.
else
pagetitle = tostring(url)
end
-- Get the fragment and pre-process percent-encoded characters.
local fragment = url.fragment
if fragment then
fragment = mw.ustring.gsub(fragment, "%.([0-9A-F][0-9A-F])", "%%%1")
end
-- Assemble the wikilink.
local wikitext = pagetitle
if interwiki then
wikitext = interwiki .. ":" .. wikitext
end
if fragment and not (args.section == "no") then
wikitext = wikitext .. "#" .. fragment
end
-- Decode percent-encoded characters and convert underscores to spaces.
wikitext = mw.uri.decode(wikitext, "WIKI")
-- If the wikitext is to be linked, re-encode illegal characters. Don't re-encode
-- characters from invalid URLs to make the default [[{{{1}}}]] display correctly.
if link and host then
wikitext = mw.ustring.gsub(wikitext, "[<>%[%]|{}%c\n]", mw.uri.encode)
end
-- Use the [[Help:Colon trick]] with categories, interwikis, and files.
local colon_prefix = mw.ustring.match(wikitext, "^(.-):.*$") or "" -- Get the text before the first colon.
local current_lang = mw.language.getContentLanguage()
local ns = mw.site.namespaces
local need_colon_trick
if mw.language.isSupportedLanguage(colon_prefix) -- Check for interwiki links.
or current_lang:lc(ns[6].name) == current_lang:lc(colon_prefix) -- Check for files.
or current_lang:lc(ns[14].name) == current_lang:lc(colon_prefix) then -- Check for categories.
need_colon_trick = true
end
for i,v in ipairs(ns[6].aliases) do -- Check for file namespace aliases.
if current_lang:lc(v) == current_lang:lc(colon_prefix) then
need_colon_trick = true
break
end
end
for i,v in ipairs(ns[14].aliases) do -- Check for category namespace aliases.
if current_lang:lc(v) == current_lang:lc(colon_prefix) then
need_colon_trick = true
break
end
end
-- Don't use the colon trick if the user says so or if we are not linking
-- (due to [[bugzilla:12974]]).
if need_colon_trick and link and not (args.colontrick == "no") then
wikitext = ":" .. wikitext
end
-- Add the link
if link then
local display = args[2] -- The display text in piped links.
if display then
if type(display) ~= "string" then
error("Non-string display value detected")
end
display = mw.text.trim(display) -- Trim whitespace.
wikitext = wikitext .. "|" .. display
end
wikitext = "[[" .. wikitext .. "]]"
end
return wikitext
end
function p.urlToWiki(frame)
local origArgs
if frame == mw.getCurrentFrame() then
-- We're being called via #invoke. If the invoking template passed any args, use
-- them. Otherwise, use the args that were passed into the template.
origArgs = frame:getParent().args
for k, v in pairs(frame.args) do
origArgs = frame.args
break
end
else
-- We're being called from another module or from the debug console, so assume
-- the args are passed in directly.
origArgs = frame
end
-- ParserFunctions considers the empty string to be false, so to preserve the previous
-- behavior of the template, change any empty arguments to nil, so Lua will consider
-- them false too.
local args = {}
for k, v in pairs(origArgs) do
if mw.ustring.match(v, '%S') then
args[k] = v
end
end
return _urlToWiki(args)
end
return p