Module:Delink/sandbox explained

-- This module de-links most wikitext.

require("strict")

local p =

local getArgs

local function delinkReversePipeTrick(s) if s:match("^%[%[|.*[|\n]") then -- Check for newlines or multiple pipes. return s end return s:match("%[%[|(.*)%]%]")

end

local function delinkPipeTrick(s) -- We need to deal with colons, brackets, and commas, per . -- First, remove the text before the first colon, if any. if s:match(":") then s = s:match("%[%[.-:(.*)|%]%]") -- If there are no colons, grab all of the text apart from the square brackets and the pipe. else s = s:match("%[%[(.*)|%]%]") end -- Next up, brackets and commas. if s:match("%(.-%)$") then -- Brackets trump commas. s = s:match("(.-) ?%(.-%)$") elseif s:match(",") then -- If there are no brackets, display only the text before the first comma. s = s:match("[^,]*") end return send

-- Return wikilink target |wikilinks=targetlocal function getDelinkedTarget(s) local result = s -- Deal with the reverse pipe trick. if result:match("%[%[|") then return delinkReversePipeTrick(result) end result = mw.uri.decode(result, "PATH") -- decode percent-encoded entities. Leave underscores and plus signs. result = mw.text.decode(result, true) -- decode HTML entities. -- Check for bad titles. To do this we need to find the -- title area of the link, i.e. the part before any pipes. local target_area if result:match("|") then -- Find if we're dealing with a piped link. target_area = result:match("^%[%[(.-)|.*%]%]") else target_area = result:match("^%[%[(.-)%]%]") end

-- Check for bad characters. if mw.ustring.match(target_area, "[%[%]<>%%%c\n]") and mw.ustring.match(target_area, "[%[%]<>%%%c\n]") ~= "?" then return s end return target_areaend

local function getDelinkedLabel(s) local result = s -- Deal with the reverse pipe trick. if result:match("%[%[|") then return delinkReversePipeTrick(result) end result = mw.uri.decode(result, "PATH") -- decode percent-encoded entities. Leave underscores and plus signs. result = mw.text.decode(result, true) -- decode HTML entities. -- Check for bad titles. To do this we need to find the -- title area of the link, i.e. the part before any pipes. local target_area if result:match("|") then -- Find if we're dealing with a piped link. target_area = result:match("^%[%[(.-)|.*%]%]") else target_area = result:match("^%[%[(.-)%]%]") end

-- Check for bad characters. if mw.ustring.match(target_area, "[%[%]<>%%%c\n]") and mw.ustring.match(target_area, "[%[%]<>%%%c\n]") ~= "?" then return s end

-- Check for categories, interwikis, and files. local colon_prefix = result:match("%[%[(.-):.*%]%]") or "" -- Get the text before the first colon. local ns = mw.site.namespaces[colon_prefix] -- see if this is a known namespace if mw.language.isKnownLanguageTag(colon_prefix) or (ns and (ns.canonicalName

"File" or ns.canonicalName

"Category")) then return "" end

-- Remove the colon if the link is using the . if result:match("%[%[:") then result = "[[" .. result:match("%[%[:(.*%]%])") end

-- Deal with links using the . if mw.ustring.match(result, "^%[%[[^|]*|%]%]") then return delinkPipeTrick(result) end -- Find the display area of the wikilink if result:match("|") then -- Find if we're dealing with a piped link. result = result:match("^%[%[.-|(.+)%]%]") -- Remove new lines from the display of multiline piped links, -- where the pipe is before the first new line. result = result:gsub("\n", "") else result = result:match("^%[%[(.-)%]%]") end

return resultend

local function delinkURL(s) -- Assume we have already delinked internal wikilinks, and that -- we have been passed some text between two square brackets [foo]. -- If the text contains a line break it is not formatted as a URL, regardless of other content. if s:match("\n") then return s end -- Check if the text has a valid URL prefix and at least one valid URL character. local valid_url_prefixes = local url_prefix for _,v in ipairs(valid_url_prefixes) do if mw.ustring.match(s, '^%[' .. v ..'[^"%s].*%]') then url_prefix = v break end end -- Get display text if not url_prefix then return s end s = s:match("^%[" .. url_prefix .. "(.*)%]") -- Grab all of the text after the URL prefix and before the final square bracket. s = s:match('^.-(["<> ].*)') or "" -- Grab all of the text after the first URL separator character ("<>). s = mw.ustring.match(s, "^%s*(%S.*)$") or "" -- If the separating character was a space, trim it off. local s_decoded = mw.text.decode(s, true) if mw.ustring.match(s_decoded, "%c") then return s end

return s_decoded

end

local function delinkLinkClass(text, pattern, delinkFunction) if type(text) ~= "string" then error("Attempt to de-link non-string input.", 2) end if type(pattern) ~= "string" or mw.ustring.sub(pattern, 1, 1) ~= "^" then error('Invalid pattern detected. Patterns must begin with "^".', 2) end -- Iterate over the text string, and replace any matched text. using the -- delink function. We need to iterate character by character rather -- than just use gsub, otherwise nested links aren't detected properly. local result = "" while text ~= "" do -- Replace text using one iteration of gsub. text = mw.ustring.gsub(text, pattern, delinkFunction, 1) -- Append the left-most character to the result string. result = result .. mw.ustring.sub(text, 1, 1) text = mw.ustring.sub(text, 2, -1) end return resultend

function p._delink(args) local text = args[1] or "" if args.refs

"yes" then -- Remove any representing ref tags. In most situations -- this is not a good idea - only use it if you know what you are doing! text = mw.ustring.gsub(text, "UNIQ%w*%-ref%-%d*%-QINU", "") end if args.comments ~= "no" then text = text:gsub("", "") -- Remove html comments. end

if args.wikilinks ~= "no" and args.wikilinks ~= "target" then -- De-link wikilinks and return the label portion of the wikilink. text = delinkLinkClass(text, "^%[%[.-%]%]", getDelinkedLabel) elseif args.wikilinks

"target" then -- De-link wikilinks and return the target portions of the wikilink. text = delinkLinkClass(text, "^%[%[.-%]%]", getDelinkedTarget) end if args.urls ~= "no" then text = delinkLinkClass(text, "^%[.-%]", delinkURL) -- De-link URLs. end if args.whitespace ~= "no" then -- Replace single new lines with a single space, but leave double new lines -- and new lines only containing spaces or tabs before a second new line. text = mw.ustring.gsub(text, "([^\n \t][\t]*)\n([\t]*[^\n \t])", "%1 %2") text = text:gsub("[\t]+", " ") -- Remove extra tabs and spaces. end return textend

function p.delink(frame) if not getArgs then getArgs = require('Module:Arguments').getArgs end return p._delink(getArgs(frame,))end

return p