-- 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 = {
-- Projects.
{ domain = "wikipedia.org" , iw_prefix = "w" , title_prefix = "/wiki/" },
{ domain = "wiktionary.org" , iw_prefix = "wikt" , title_prefix = "/wiki/" },
{ domain = "wikinews.org" , iw_prefix = "n" , title_prefix = "/wiki/" },
{ domain = "wikibooks.org" , iw_prefix = "b" , title_prefix = "/wiki/" },
{ domain = "wikiquote.org" , iw_prefix = "q" , title_prefix = "/wiki/" },
{ domain = "wikisource.org" , iw_prefix = "s" , title_prefix = "/wiki/" },
{ domain = "species.wikimedia.org" , iw_prefix = "species" , title_prefix = "/wiki/" },
{ domain = "wikiversity.org" , iw_prefix = "v" , title_prefix = "/wiki/" },
{ domain = "wikivoyage.org" , iw_prefix = "voy" , title_prefix = "/wiki/" },
{ domain = "wikimediafoundation.org" , iw_prefix = "wmf" , title_prefix = "/wiki/" },
{ domain = "commons.wikimedia.org" , iw_prefix = "commons" , title_prefix = "/wiki/" },
{ domain = "wikidata.org" , iw_prefix = "d" , title_prefix = "/wiki/" },
{ domain = "meta.wikimedia.org" , iw_prefix = "m" , title_prefix = "/wiki/" },
{ domain = "incubator.wikimedia.org" , iw_prefix = "incubator" , title_prefix = "/wiki/" },
{ domain = "strategy.wikimedia.org" , iw_prefix = "strategy" , title_prefix = "/wiki/" },
{ domain = "mediawiki.org" , iw_prefix = "mw" , title_prefix = "/wiki/" },
{ domain = "bugzilla.wikimedia.org" , iw_prefix = "bugzilla" , title_prefix = "/show_bug.cgi?id=" },
{ domain = "test.wikipedia.org" , iw_prefix = "testwiki" , title_prefix = "/wiki/" },
-- Chapters.
{ domain = "wikimedia.org.ar" , iw_prefix = "wmar" , title_prefix = "/wiki/" },
{ domain = "wikimedia.org.au" , iw_prefix = "wmau" , title_prefix = "/wiki/" },
{ domain = "be.wikimedia.org" , iw_prefix = "wmbe" , title_prefix = "/wiki/" },
{ domain = "wikimedia.ca" , iw_prefix = "wmca" , title_prefix = "/wiki/" },
{ domain = "wikimedia.de" , iw_prefix = "wmde" , title_prefix = "/wiki/" },
{ domain = "fi.wikimedia.org" , iw_prefix = "wmfi" , title_prefix = "/wiki/" },
{ domain = "wikimedia.hk" , iw_prefix = "wmhk" , title_prefix = "/index.php/" },
{ domain = "wiki.media.hu" , iw_prefix = "wmhu" , title_prefix = "/wiki/" },
{ domain = "wiki.wikimedia.in" , iw_prefix = "wmin" , title_prefix = "/" },
{ domain = "wikimedia.org.id" , iw_prefix = "wmid" , title_prefix = "/wiki/" },
{ domain = "wikimedia.org.il" , iw_prefix = "wmil" , title_prefix = "/" },
{ domain = "wikimedia.it" , iw_prefix = "wmit" , title_prefix = "/wiki/" },
{ domain = "nl.wikimedia.org" , iw_prefix = "wmnl" , title_prefix = "/wiki/" },
{ domain = "no.wikimedia.org" , iw_prefix = "wmno" , title_prefix = "/wiki/" },
{ domain = "pl.wikimedia.org" , iw_prefix = "wmpl" , title_prefix = "/wiki/" },
{ domain = "no.wikimedia.org" , iw_prefix = "wmno" , title_prefix = "/wiki/" },
{ domain = "ru.wikimedia.org" , iw_prefix = "wmru" , title_prefix = "/wiki/" },
{ domain = "rs.wikimedia.org" , iw_prefix = "wmrs" , title_prefix = "/wiki/" },
{ domain = "se.wikimedia.org" , iw_prefix = "wmse" , title_prefix = "/wiki/" },
{ domain = "wikimedia.ch" , iw_prefix = "wmch" , title_prefix = "/" },
{ domain = "tw.wikimedia.org" , iw_prefix = "wmtw" , title_prefix = "/wiki/index.php5/" },
{ domain = "uk.wikimedia.org" , iw_prefix = "wmuk" , title_prefix = "/wiki/" },
}
local function convertInterwiki(url)
local host = url.host
-- If host is nil then the text passed to the module is probably a URL fragment,
-- not a full URL. Return blank so that it can get percent-de-encoded as it is.
if not host then
return false
end
-- Find the entry for the host in the interwiki table.
local host_id
for i,v in ipairs(interwiki_table) do
if mw.ustring.match(host, interwiki_table[i].domain) then
host_id = i
break
end
end
-- If no valid Wikimedia site is found, return blank so that whatever the input
-- was can be percent-de-encoded.
if not host_id then
return false
end
-- Find the language.
local lang = mw.ustring.match(host, "^(.-)%.") -- Find the text before the first period.
local domain_lang = mw.ustring.match(interwiki_table[host_id].domain, "^(.-)%.")
-- Return false if the text isn't a valid language code or if there's already a language code in the host name.
if not ( lang and mw.language.isSupportedLanguage(lang) ) or mw.language.isSupportedLanguage(domain_lang) then
lang = false
end
-- Check if the language is the same as the current language.
local current_host = mw.uri.new(mw.title.getCurrentTitle():fullUrl()).host
local same_lang
if lang == mw.ustring.match(current_host, "^(.-)%.") then
same_lang = true
end
-- If the url points to a page on the same site as the module is invoked from,
-- return false, as there is no need for an interwiki prefix in this case.
if host == current_host then
return false, host_id
end
-- Check if the project is the same as the current project (but in a different language).
local current_host_id, same_host
for i,v in ipairs(interwiki_table) do
if mw.ustring.match(current_host, interwiki_table[i].domain) then
current_host_id = i
break
end
end
if current_host_id == host_id then
same_host = 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
if same_lang then
return project, host_id
elseif same_host then
return lang, host_id
elseif not lang then -- If the language code is bad but the rest of the host name is ok.
return false
else
return project .. ":" .. lang, host_id
end
end
local function _urlToWiki(args)
url = args[1] or error("No URL specified")
url = mw.ustring.match(url, "^%s*(.*%S)") or "" -- Trim whitespace.
url = mw.uri.new(url)
-- Get the interwiki prefix.
local interwiki, host_id = convertInterwiki(url)
-- Get the page title.
local pagetitle, title_prefix
if host_id 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 title_prefix
-- in case the URL isn't of a Wikimedia site.
elseif title_prefix and mw.ustring.match(url.path, "index%.php") and url.query.title then
pagetitle = url.query.title
-- Otherwise, use the whole URL as the title.
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")
return wikitext
end
function p.urlToWiki(frame)
local args
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.
args = frame:getParent().args
for k, v in pairs(frame.args) do
args = 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.
args = frame
end
return _urlToWiki(args)
end
return p
