Module:Delink/sandbox: Difference between revisions

Revision as of 11:47, 30 September 2021

Latest revision as of 18:29, 22 April 2023

(8 intermediate revisions by 3 users not shown)

Line 1:

-- This module de-links most wikitext.

require('Module:No globals'"strict")

local p = {}

Line 10:

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

return s

else

return s:match("%[%[|(.*)%]%]")

end

return s:match("%[%[|(.*)%]%]")

end

local function delinkPipeTrick(s)

local linkarea, display = "", ""

-- We need to deal with colons, brackets, and commas, per [[Help:Pipe trick]].

-- First, remove the text before the first colon, if any.

if s:match(":") then

Line 31 ⟶ 29:

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 s

end

local function delinkWikilink_tgt(s) -- returnReturn wikilink target |wikilinks=target

local function getDelinkedTarget(s)

local result = s

-- Deal with the reverse pipe trick.

Line 54 ⟶ 53:

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

Line 62:

end

local function delinkWikilinkgetDelinkedLabel(s)

local result = s

-- Deal with the reverse pipe trick.

Line 68:

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.

Line 80:

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 colonprefixcolon_prefix = result:match("%[%[(.-):.*%]%]") or "" -- Get the text before the first colon.

local ns = mw.site.namespaces[colonprefixcolon_prefix] -- see if this is a known namespace

if mw.language.isKnownLanguageTag(colonprefixcolon_prefix) or (ns and (ns.canonicalName == "File" or ns.canonicalName == "Category")) then

or ( ns and ( ns.canonicalName == "File" or ns.canonicalName == "Category" ) ) then

return ""

end

-- Remove the colon if the link is using the [[Help:Colon trick]].

if result:match("%[%[:") then

result = "[[" .. result:match("%[%[:(.*%]%])")

end

-- Deal with links using the [[Help:Pipe trick]].

if mw.ustring.match(result, "^%[%[[^|]*|%]%]") then

Line 128:

local valid_url_prefixes = {"//", "http://", "https://", "ftp://", "gopher://", "mailto:", "news:", "irc://"}

local url_prefix

for i_ ,v in ipairs(valid_url_prefixes) do

if mw.ustring.match(s, '^%[' .. v ..'[^"%s].*%]' ) then

url_prefix = v

Line 146:

if mw.ustring.match(s_decoded, "%c") then

return s

else

return s_decoded

end

return s_decoded

end

local function delinkLinkClass(stext, pattern, delinkFunction)

if not type(stext) =~= "string" then

error("Attempt to de-link non-string input.", 2)

end

if not ( type(pattern) =~= "string" andor mw.ustring.sub(pattern, 1, 1) =~= "^" ) then

error('Invalid pattern detected. Patterns must begin with "^".', 2)

end

Line 162 ⟶ 163:

-- than just use gsub, otherwise nested links aren't detected properly.

local result = ""

while stext ~= ''"" do

-- Replace text using one iteration of gsub.

stext = mw.ustring.gsub(stext, pattern, delinkFunction, 1)

-- Append the left-most character to the result string.

result = result .. mw.ustring.sub(stext, 1, 1)

stext = mw.ustring.sub(stext, 2, -1)

end

return result

Line 179 ⟶ 180:

text = mw.ustring.gsub(text, "UNIQ%w*%-ref%-%d*%-QINU", "")

end

if not (args.comments =~= "no") then

text = text:gsub("<!%-%-.-%-%->", "") -- Remove html comments.

end

if not (args.wikilinks =~= "no") and 'target'args.wikilinks ~= args.wikilinks"target" then

text = delinkLinkClass(text, "^%[%[.-%]%]", delinkWikilink) -- De-link wikilinks and return the label portion of the wikilink.

elseif 'target' == args.wikilinks then

text = delinkLinkClass(text, "^%[%[.-%]%]", delinkWikilink_tgtgetDelinkedLabel)
elseif args.wikilinks == "target" then
-- De-link wikilinks and return the target portions of the wikilink.

text = delinkLinkClass(text, "^%[%[.-%]%]", getDelinkedTarget)

end

if not (args.urls =~= "no") then

text = delinkLinkClass(text, "^%[.-%]", delinkURL) -- De-link URLs.

end

if not (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.