Difference between revisions of "Module:Delink/sandbox"
Jump to navigation
Jump to search
blackwiki>Mr. Stradivarius m (Dummy edit. That came out wrong - that should have been "start making this a bit more sane".) |
blackwiki>Mr. Stradivarius (use string library functions, and use single quotes rather than double) |
||
| Line 18: | Line 18: | ||
function p._delinkReversePipeTrick(s) | function p._delinkReversePipeTrick(s) | ||
| − | if | + | if s:find('^%[%[|.*[|\n]') or s == '[[|]]' then -- Check for newlines or multiple pipes. |
return s | return s | ||
else | else | ||
| − | return | + | return s:match('%[%[|(.*)%]%]') |
end | end | ||
end | end | ||
| Line 30: | Line 30: | ||
-- First, remove the text before the first colon, if any. | -- First, remove the text before the first colon, if any. | ||
| − | s = | + | s = s:gsub('^(.-:)', '') |
-- Next up, brackets and commas. | -- Next up, brackets and commas. | ||
| − | if | + | if s:find('%(.-%)$') then -- Brackets trump commas. |
| − | s = | + | s = s:match('(.-) ?%(.-%)$') |
| − | elseif | + | elseif s:find(',') then -- If there are no brackets, display only the text before the first comma. |
| − | s = | + | s = s:match('(.-),.*$') |
end | end | ||
return s | return s | ||
| Line 42: | Line 42: | ||
function p._delinkWikilink(s) | function p._delinkWikilink(s) | ||
| − | -- s should be a string starting | + | -- s should be a string starting with '[[' and ending with ']]'. It should not contain other ']]' |
-- Deal with nested links | -- Deal with nested links | ||
| − | local | + | local nested = '[[' .. s:sub(3):gsub('%[%[.-%]%]', p._delinkWikilink) |
| − | |||
if nested ~= s then | if nested ~= s then | ||
return nested | return nested | ||
| Line 52: | Line 51: | ||
-- Deal with the reverse pipe trick. | -- Deal with the reverse pipe trick. | ||
| − | if | + | if s:find('^%[%[|') then |
return p._delinkReversePipeTrick(s) | return p._delinkReversePipeTrick(s) | ||
end | end | ||
| − | local decoded = mw.uri.decode(s, | + | local decoded = mw.uri.decode(s, 'PATH') -- decode percent-encoded entities. Leave underscores and plus signs. |
decoded = mw.text.decode(decoded, true) -- decode HTML entities. | decoded = mw.text.decode(decoded, true) -- decode HTML entities. | ||
-- Check for bad titles. To do this we need to find the | -- Check for bad titles. To do this we need to find the | ||
-- title area of the link, i.e. the part before any pipes. | -- title area of the link, i.e. the part before any pipes. | ||
| − | local titlearea, display = | + | local titlearea, display = decoded:match('^%[%[([^|%]]*)|?(.*)%]%]') |
-- Check for bad characters. | -- Check for bad characters. | ||
| − | if | + | if titlearea:find('[%[%]<>{}%%%c\n]') then |
return s | return s | ||
end | end | ||
-- Check for categories, interwikis, and files. | -- Check for categories, interwikis, and files. | ||
| − | local colonprefix = | + | local colonprefix = titlearea:match('^(.-):') or '' -- Get the text before the first colon. |
local ns = mw.site.namespaces[colonprefix] -- see if this is a known namespace | local ns = mw.site.namespaces[colonprefix] -- see if this is a known namespace | ||
if mw.language.isKnownLanguageTag(colonprefix) | if mw.language.isKnownLanguageTag(colonprefix) | ||
| − | or ( ns and ( ns. | + | or (ns and (ns.id == 6 or ns.id == 14)) |
| − | return | + | then |
| + | return '' | ||
end | end | ||
-- Remove the colon if the link is using the [[Help:Colon trick]]. | -- Remove the colon if the link is using the [[Help:Colon trick]]. | ||
| − | if | + | if titlearea:sub(1, 1) == ':' then |
| − | titlearea | + | titlearea = titlearea:sub(2) |
end | end | ||
-- Deal with links using the [[Help:Pipe trick]]. | -- Deal with links using the [[Help:Pipe trick]]. | ||
| − | if display == | + | if display == '' then |
| − | return p._delinkPipeTrick(titlearea ) | + | return p._delinkPipeTrick(titlearea) |
end | end | ||
| Line 90: | Line 90: | ||
-- Remove new lines from the display of multiline piped links, | -- Remove new lines from the display of multiline piped links, | ||
-- where the pipe is before the first new line. | -- where the pipe is before the first new line. | ||
| − | titlearea = | + | titlearea = titlearea:gsub('\n', '') |
end | end | ||
| Line 101: | Line 101: | ||
-- If the text contains a line break it is not formatted as a URL, regardless of other content. | -- If the text contains a line break it is not formatted as a URL, regardless of other content. | ||
| − | if | + | if s:find('\n') then |
return s | return s | ||
end | end | ||
-- Check if the text has a valid URL prefix and at least one valid URL character. | -- Check if the text has a valid URL prefix and at least one valid URL character. | ||
| − | local valid_url_prefixes = { | + | local valid_url_prefixes = {'//', 'http://', 'https://', 'ftp://', 'gopher://', 'mailto:', 'news:', 'irc://'} |
local url_prefix | local url_prefix | ||
for i,v in ipairs(valid_url_prefixes) do | for i,v in ipairs(valid_url_prefixes) do | ||
| − | if | + | if s:find('^%[' .. v ..'[^"%s].*%]' ) then |
url_prefix = v | url_prefix = v | ||
break | break | ||
| Line 118: | Line 118: | ||
if not url_prefix then | if not url_prefix then | ||
-- Deal with nested links or send back original string. | -- Deal with nested links or send back original string. | ||
| − | return | + | return '[' .. s:sub(2):gsub('%[.-%]', p._delinkURL) |
end | end | ||
| − | s = | + | s = s:match('^%[' .. url_prefix .. '(.*)%]') -- Grab all of the text after the URL prefix and before the final square bracket. |
| − | s = | + | s = s:match('^.-(["<> [].*)') or '' -- Grab all of the text after the first URL separator character ("<> ). |
| − | s = | + | s = s:match('^%s*(%S.*)$') or '' -- If the separating character was a space, trim it off. |
local s_decoded = mw.text.decode(s, true) | local s_decoded = mw.text.decode(s, true) | ||
| − | if | + | if s_decoded:find('%c') then |
return s | return s | ||
else | else | ||
Revision as of 15:39, 17 April 2014
Documentation for this module may be created at Module:Delink/sandbox/doc
-- This module de-links most wikitext.
local getArgs = require('Module:Arguments').getArgs
local yesno = require('Module:Yesno')
p = {}
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*$')
if val == '' then
return nil
end
end
return yesno(val)
end
function p._delinkReversePipeTrick(s)
if s:find('^%[%[|.*[|\n]') or s == '[[|]]' then -- Check for newlines or multiple pipes.
return s
else
return s:match('%[%[|(.*)%]%]')
end
end
function p._delinkPipeTrick(s)
-- s the tile area, without bracket nor pipe
-- We need to deal with colons, brackets, and commas, per [[Help:Pipe trick]].
-- First, remove the text before the first colon, if any.
s = s:gsub('^(.-:)', '')
-- Next up, brackets and commas.
if s:find('%(.-%)$') then -- Brackets trump commas.
s = s:match('(.-) ?%(.-%)$')
elseif s:find(',') then -- If there are no brackets, display only the text before the first comma.
s = s:match('(.-),.*$')
end
return s
end
function p._delinkWikilink(s)
-- s should be a string starting with '[[' and ending with ']]'. It should not contain other ']]'
-- Deal with nested links
local nested = '[[' .. s:sub(3):gsub('%[%[.-%]%]', p._delinkWikilink)
if nested ~= s then
return nested
end
-- Deal with the reverse pipe trick.
if s:find('^%[%[|') then
return p._delinkReversePipeTrick(s)
end
local decoded = mw.uri.decode(s, 'PATH') -- decode percent-encoded entities. Leave underscores and plus signs.
decoded = mw.text.decode(decoded, 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 titlearea, display = decoded:match('^%[%[([^|%]]*)|?(.*)%]%]')
-- Check for bad characters.
if titlearea:find('[%[%]<>{}%%%c\n]') then
return s
end
-- Check for categories, interwikis, and files.
local colonprefix = titlearea:match('^(.-):') or '' -- Get the text before the first colon.
local ns = mw.site.namespaces[colonprefix] -- see if this is a known namespace
if mw.language.isKnownLanguageTag(colonprefix)
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 titlearea:sub(1, 1) == ':' then
titlearea = titlearea:sub(2)
end
-- Deal with links using the [[Help:Pipe trick]].
if display == '' then
return p._delinkPipeTrick(titlearea)
end
-- Find the display area of the wikilink
if not display then -- Find if we're dealing with a piped link.
-- Remove new lines from the display of multiline piped links,
-- where the pipe is before the first new line.
titlearea = titlearea:gsub('\n', '')
end
return display or titlearea
end
function p._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:find('\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:find('^%[' .. v ..'[^"%s].*%]' ) then
url_prefix = v
break
end
end
-- Get display text
if not url_prefix then
-- 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:match('^%s*(%S.*)$') or '' -- If the separating character was a space, trim it off.
local s_decoded = mw.text.decode(s, true)
if s_decoded:find('%c') then
return s
else
return s_decoded
end
end
function p._delink(args)
local text = args[1] or ''
if whitespaceYesno(args.refs) 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 then
text = mw.ustring.gsub(text, '<!%-%-.-%-%->', '') -- Remove html comments.
end
if whitespaceYesno(args.wikilinks) == false then
text = mw.ustring.gsub(text, '%[%[.-%]%]', p._delinkWikilink) -- De-link wikilinks.
end
if whitespaceYesno(args.urls) == false then
text = mw.ustring.gsub(text, '%[.-%]', p._delinkURL) -- De-link URLs.
end
if whitespaceYesno(args.whitespace) == false 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 = mw.ustring.gsub(text, '[ \t]+', ' ') -- Remove extra tabs and spaces.
end
return text
end
function p.delink(frame)
local args = frame:getParent().args
return p._delink(args)
end
return p