Module:Delink/sandbox: Difference between revisions

Content deleted Content added
a few tweaks
Synced (rip one testcase but whatever), dramatically improved efficiency of one of the regex patterns
 
(25 intermediate revisions by 11 users not shown)
Line 1:
-- This module de-links most wikitext.
 
require("strict")
local getArgs = require('Module:Arguments').getArgs
local yesno = require('Module:Yesno')
 
local p = {}
 
local getArgs
local function whitespaceYesno(val)
-- Like yesno, but trims whitespace from vals and removes blank strings.
if type(val) == 'string' then
val = val:match('^%s*(.-)%s*$')
if val == '' then
return nil
end
end
return yesno(val)
end
 
local function p._delinkReversePipeTrickdelinkReversePipeTrick(s)
if s:findmatch('"^%[%[|.*[|\n]'") or s == '[[|]]' then -- Check for newlines or multiple pipes.
return s
else
return s:match('%[%[|(.*)%]%]')
end
return s:match("%[%[|(.*)%]%]")
 
end
 
local function p._delinkPipeTrickdelinkPipeTrick(s)
-- s the tile area, without bracket nor pipe
-- We need to deal with colons, brackets, and commas, per [[Help:Pipe trick]].
 
-- First, remove the text before the first colon, if any.
s =if s:gsubmatch('^(.-":")', '')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:findmatch('"%(.-%)$'") then -- Brackets trump commas.
s = s:match('"(.-) ?%(.-%)$'")
elseif s:findmatch('",'") then -- If there are no brackets, display only the text before the first comma.
s = s:match('(.-)"[^,.]*$'")
end
return s
end
 
-- Return wikilink target |wikilinks=target
function p._delinkWikilink(s)
local function getDelinkedTarget(s)
-- s is a string starting with '[[' and ending with ']]'. It does not contain any other ']]' strings.
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
 
-- DealCheck withfor nestedbad linkscharacters.
if mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") and mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") ~= "?" then
local nested = '[[' .. s:sub(3):gsub('%[%[.-%]%]$', p._delinkWikilink)
return s
if nested ~= s then
return nested
end
return target_area
end
 
local function getDelinkedLabel(s)
local result = s
-- Deal with the reverse pipe trick.
if sresult:findmatch('^"%[%[|'") then
return p._delinkReversePipeTrickdelinkReversePipeTrick(sresult)
end
 
local decodedresult = mw.uri.decode(sresult, '"PATH'") -- decode percent-encoded entities. Leave underscores and plus signs.
decodedresult = mw.text.decode(decodedresult, 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
local titlearea, display = decoded:match('^%[%[([^|%]]*)|?(.*)%]%]$')
if result:match("|") then -- Find if we're dealing with a piped link.
local temptitlearea, fragment = titlearea:match('^(.-)#(.*)$')
target_area = result:match("^%[%[(.-)|.*%]%]")
titlearea = temptitlearea or titlearea
else
target_area = result:match("^%[%[(.-)%]%]")
end
 
-- Check for bad characters.
if titlearea:findmw.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 = titlearearesult: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
return ""
or ns and (ns.id == 6 or ns.id == 14)
then
return ''
end
 
-- Remove the colon if the link is using the [[Help:Colon trick]].
if titlearearesult:submatch(1, 1) == '"%[%[:'") then
result = "[[" .. result:match("%[%[:(.*%]%])")
titlearea = titlearea:sub(2)
end
 
-- Deal with links using the [[Help:Pipe trick]].
if mw.ustring.match(result, "^%[%[[^|]*|%]%]") then
if display == '' then
return p._delinkPipeTrickdelinkPipeTrick(titlearearesult)
end
 
-- Find the display area of the wikilink
if not displayresult: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.
titlearea result = titlearearesult:gsub('"\n'", ''"")
else
result = result:match("^%[%[(.-)%]%]")
end
 
return display or titlearea result
end
 
local function p._delinkURLdelinkURL(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:findmatch('"\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 = {'"//'", '"http://'", '"https://'", '"ftp://'", '"gopher://'", '"mailto:'", '"news:'", '"irc://'"}
local url_prefix
for i_ ,v in ipairs(valid_url_prefixes) do
if s:findmw.ustring.match(s, '^%[' .. v ..'[^"%s].*%]' ) then
url_prefix = v
break
end
end
 
-- Get display text
if not url_prefix then
return s
-- Deal with nested links or send back original string.
return '[' .. s:sub(2):gsub('%[.-%]', p._delinkURL)
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 = 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 s_decoded:findmw.ustring.match('s_decoded, "%c'") then
return s
else
return s_decoded
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 result
end
 
function p._delink(args)
local text = args[1] or ''""
if whitespaceYesno(args.refs) == "yes" then
-- Remove any [[Help:Strip markers]] 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 whitespaceYesno(args.comments) ~= false"no" then
text = text:gsub('"<!%-%-.-%-%->'", ''"") -- Remove html comments.
end
 
if whitespaceYesno(args.wikilinks) ~= false then
if args.wikilinks ~= "no" and args.wikilinks ~= "target" then
text = text:gsub('%[%[.-%]%]', p._delinkWikilink) -- De-link wikilinks.
-- 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 whitespaceYesno(args.urls) ~= false"no" then
text = text:gsubdelinkLinkClass('text, "^%[.-%]'", p._delinkURLdelinkURL) -- De-link URLs.
end
if whitespaceYesno(args.whitespace) ~= false"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 = 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 text
Line 160 ⟶ 204:
 
function p.delink(frame)
if not getArgs then
local args = frame:getParent().args
getArgs = require('Module:Arguments').getArgs
return p._delink(args)
end
return p._delink(getArgs(frame, {wrappers = 'Template:Delink'}))
end