Wikipedia

Module:Delink/sandbox: Difference between revisions

Browse history interactively
← Previous edit
Content deleted Content added
Darklama (talk | contribs)
205 edits
try another way, most cases covered
Synced (rip one testcase but whatever), dramatically improved efficiency of one of the regex patterns
 
(34 intermediate revisions by 12 users not shown)
Line 1:
-- This module de-links most wikitext.
 
require("strict")
p = {};
 
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("%[%[|(.*)%]%]")
 
local function internalLink(s)
return (mw.ustring.gsub( s, "^([^|]*)(|?)(.*)$", function(l, p, t)
-- check for malformed links and text
if mw.ustring.match( l, "[<>{}%c\n]" ) or mw.ustring.match( t, "\n" ) then
return s;
end
-- remove whitespace from begining and end of link and text
l = mw.text.trim(l);
t = mw.text.trim(t);
local file = false;
if l == "" then
return s;
elseif mw.ustring.match( l, "^:" ) then
l = mw.ustring.gsub( l, "^:+", "" );
elseif mw.ustring.match( l, "^[Cc][Aa][Tt][Ee][Gg][Oo][Rr][Yy]:" ) then
return "";
elseif mw.ustring.match( l, "^[Ff][Ii][Ll][Ee]:" ) then
file = true;
elseif mw.ustring.match( l, "^[Ii][Mm][Aa][Gg][Ee]:" ) then
file = true;
end
if file == true then
return "";
elseif p == "|" and t ~= "" then
return t;
else
l = mw.ustring.gsub( l, "(#.+)$", function(h)
return mw.ustring.gsub( h, "%.([0-9A-Fa-f][0-9A-Fa-f])", "%%%1" );
end);
if p == "" and t == "" then
return mw.uri.decode(l, "PATH");
end
-- Need to transform text based on link per [[Help:Pipe trick]].
t = mw.ustring.match( l, "#(.+)$" );
if t ~= nil then
return mw.uri.decode(t, "PATH");
end
t = mw.ustring.gsub( l, "^([^#]*)#$", "%1", 1 );
t = mw.ustring.gsub( t, "(.-) ?%b()$", "%1", 1 );
t = mw.ustring.gsub( t, "^(.-),.*$", "%1", 1 );
t = mw.uri.decode(t, "PATH");
return t;
end
end, 1));
end
 
local function externalLinkdelinkPipeTrick(s)
-- We need to deal with colons, brackets, and commas, per [[Help:Pipe trick]].
-- If the text contains a line break it is not formatted as a URL, regardless of other content.
-- First, remove the text before the first colon, if any.
if mw.ustring.match(s, "\n") then
if s:match(":") then
return s;
s = s:match("%[%[.-:(.*)|%]%]")
end
-- If there are no colons, grab all of the text apart from the square brackets and the pipe.
else
return (mw.ustring.gsub(mw.text.trim(s), "^([^ \t]*)[ \t]*(.*)$", function(l, t)
s = s:match("%[%[(.*)|%]%]")
t = mw.text.trim(t);
end
if t == "" then
-- Next up, brackets and commas.
return s;
if s:match("%(.-%)$") then -- Brackets trump commas.
elseif mw.ustring.sub( l, 1, 2 ) == "//" then
s = s:match("(.-) ?%(.-%)$")
l = "http:" .. l;
elseif s:match(",") then -- If there are no brackets, display only the text before the first comma.
end
s = s:match("[^,]*")
end
l = mw.uri.new( l );
return s
if l.protocol == nil or l.host == nil or mw.ustring.match( l.host, "%." ) == nil then
return s;
else
return t;
end
end, 1));
end
 
-- Return wikilink target |wikilinks=target
local function delinkLinkClass(s, pattern, delinker)
local function getDelinkedTarget(s)
if type(s) ~= "string" then
local result = s
error("Attempt to de-link non-string input.", 2);
-- Deal with the reverse pipe trick.
elseif type(pattern) ~= "string" then
if result:match("%[%[|") then
error('Patterns must begin a string.', 2);
return delinkReversePipeTrick(result)
end
end
-- Iterate over the text string, and replace any matched text. using the
-- delink function. We need to iterate one by one otherwise nested links
result = mw.uri.decode(result, "PATH") -- decode percent-encoded entities. Leave underscores and plus signs.
-- aren't detected properly.
result = local mw.text.decode(result, f,true) e,-- b,decode a =HTML {};entities.
-- Check for bad titles. To do this we need to find the
repeat
-- title area of the link, i.e. the part before any pipes.
f, e = mw.ustring.find(s, pattern);
local target_area
if result:match("|") then -- Find if we're dealing with a piped link.
if f == nil then
target_area = result:match("^%[%[(.-)|.*%]%]")
table.insert( result, s );
else
break;
target_area = result:match("^%[%[(.-)%]%]")
end
end
 
if f > 1 then
-- Check for bad characters.
b = mw.ustring.gsub( mw.ustring.sub(s, 1, f-1), "(%[+)$", mw.text.nowiki, 1 );
if mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") and mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") ~= "?" then
end
return s
end
if e ~= nil then
a = mw.ustring.gsub( mw.ustring.sub(s, e+1), "^(%]+)", mw.text.nowiki, 1 );
return target_area
else
a = "";
end
-- Replace text using one iteration.
s = mw.ustring.gsub( mw.ustring.sub(s, f, e), pattern, delinker, 1 );
-- Append to result and start after last result.
table.insert( result, b );
table.insert( result, s );
s = a;
until f == nil or s == "";
return table.concat( result );
end
 
local function _delinkgetDelinkedLabel(argss)
local textresult = args[1] or ""s
-- Deal with the reverse pipe trick.
if args.refs == "yes" then
if result:match("%[%[|") then
-- Remove any [[Help:Strip markers]] representing ref tags. In most situations
return delinkReversePipeTrick(result)
-- this is not a good idea - only use it if you know what you are doing!
end
text = mw.ustring.gsub(text, "UNIQ%w*%-ref%-%d*%-QINU", "")
 
end
result = mw.uri.decode(result, "PATH") -- decode percent-encoded entities. Leave underscores and plus signs.
if args.comments ~= "no" then
text result = mw.ustringtext.gsubdecode(text, "<!%-%-.-%-%->"result, ""true) -- Removedecode htmlHTML commentsentities.
 
end
-- Check for bad titles. To do this we need to find the
if args.wikilinks ~= "no" then
-- title area of the link, i.e. the part before any pipes.
text = delinkLinkClass(text, "%[%[([^%[%]]*)%]%]", internalLink) -- De-link wikilinks.
local target_area
end
if result:match("|") then -- Find if we're dealing with a piped link.
if args.urls ~= "no" then
target_area = result:match("^%[%[(.-)|.*%]%]")
text = delinkLinkClass(text, "%[([^%[%]]*)%]", externalLink) -- De-link URLs.
else
end
target_area = result:match("^%[%[(.-)%]%]")
if args.whitespace ~= "no" then
end
-- 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.
-- Check for bad characters.
text = mw.ustring.gsub(text, "([^\n \t][ \t]*)\n([ \t]*[^\n \t])", "%1 %2")
if mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") and mw.ustring.match(target_area, "[%[%]<>{}%%%c\n]") ~= "?" then
text = mw.ustring.gsub(text, "[ \t]+", " ") -- Remove extra tabs and spaces.
return s
end
end
return text
 
-- 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 [[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
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 result
end
 
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 = {"//", "http://", "https://", "ftp://", "gopher://", "mailto:", "news:", "irc://"}
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 result
end
 
function p._delink(args)
local text = args[1] or ""
if 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 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 text
end
 
function p.delink(frame)
if not getArgs then
local args
getArgs = require('Module:Arguments').getArgs
if frame == mw.getCurrentFrame() then
end
-- We're being called via #invoke. If the invoking template passed any args, use
return p._delink(getArgs(frame, {wrappers = 'Template:Delink'}))
-- them. Otherwise, use the args that were passed into the template.
for k, v in pairs(frame.args) do
args = frame.args;
break;
end
if args == nil then args = frame:getParent().args; 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 _delink(args);
end
 
Retrieved from "https://en.wikipedia.org/wiki/Module:Delink/sandbox"