Module:DYK queue formatting check

-- This module started out verifying punctuation rules unique to [[WP:DYKMOS]]
-- Then it expanded to include the remaining non-subjective part of
-- [[WP:DYKMOS]] and [[MOS:CQ]] to keep formatting errors from [[WP:MP/E]]

-- If there's a false positive with a hook and the warning is too annoying,
-- the current method is to just disableAll.

-- This module doesn't show up on the main page or prevent promotion,
-- but if a problem arises:
local disableAll = false
-- This module is not [[WP:EXPENSIVE]], but if performance becomes a problem:
local disableTransclusionBasedChecks = false
-- People don't seem to care about [[MOS:SOB]]. [[Special:Diff/1282220073]]
-- could be edited but [[Special:Diff/1281992817]] might be intentional
local disableSeaOfBlueCheck = true

local p = {}
local plainText = require("Module:Plain text")._main
local isDisambiguation = require('Module:Disambiguation').isDisambiguation
local getTargetFromText = require('Module:Redirect').getTargetFromText
local decode = require('Module:DecodeEncode')._decode

-- Checks that look at the destination page of a wikilink
local function check_link(page)
	if disableTransclusionBasedChecks then
		return
	end
	local title = mw.title.new(page)
	local content = title.content
	if not content then
		return "[[WP:DYKMOS]]: The hook must not contain redlinks: " .. page
	end
	local target = getTargetFromText(content)
	if target then
		return "[[WP:DYKMOS]]/[[WP:MPNOREDIRECT]]: The hook must not " ..
		"contain redirects: [[" .. page .. "]] → [[" .. target .. "]]"
	end
	if isDisambiguation(content) then
		return "[[WP:DYKMOS]]: The hook must not contain links to " ..
		"disambiguation pages: " .. page
	end
end

-- Per-hook checks that may have occasional exceptions
local function check_hook(wikitextLine, expandedLine, decodedLine, hookNum)
	if ((expandedLine:find("{{") and not wikitextLine:find("{{%("))
		or (expandedLine:find("}}") and not wikitextLine:find("{{%)"))
		or (expandedLine:find("%[%[") and not wikitextLine:find("{{!"))
		or (expandedLine:find("%]%]") and not wikitextLine:find("!}}"))) then
	    -- "code" is not ignored because some hooks use the word "codename"
	    -- Use templates or HTML alternatives from
	    -- [[Template:Escape template list]] if there's actually code in a hook.
	    return "Potentially broken wikitext"
	end
	local byteCount = -1 -- Counting ends at the question mark
	if wikitextLine:find("''%([^)]+%)''") then
		-- The eleven characters in a ''(pictured)'' tag do not count
		byteCount = byteCount - 11
	end
	local boldLinkFound = false
	for wikilink in wikitextLine:gmatch("'''%[%[([^]]+)%]%]") do
		-- Text in boldlinks after the first do not count toward the limit
		if boldLinkFound then
			wikilink = wikilink:gsub("[^|]+|", "", 1)
			byteCount = byteCount - #wikilink
		end
		boldLinkFound = true
	end
	byteCount = byteCount + #decodedLine
	if byteCount > 200 then
		return "[[WP:DYK200]]: The hook cannot exceed 200 prose characters. " ..
		"It is currently " .. byteCount .. " characters"
	elseif byteCount < 1 then
		if hookNum == 1 then
			return 'Empty first hooks are usually written as "... that ... ' ..
			"''(pictured)'' ...\""
		else
			return 'Empty hooks are usually written as "... that ..."'
		end
	end
	if hookNum > 1 then
		if wikitextLine:find("''%([^)]+%)''") then
			return "[[WP:DYKIMG]]/[[WP:DYKMOS]]: Extra ''(pictured)'' or " ..
			"alternative"
		end
	elseif hookNum < 0 then
		-- First hook might have been skipped by opt-out
	else
		if not decodedLine:find("%([^)]+%)") then
			return "[[WP:DYKIMG]]/[[WP:DYKMOS]]: Missing ''(pictured)'' or " ..
			"alternative"
		end
		-- Parentheses that are not the media marker are
		-- allowed if "absolutely unavoidable".
		-- Italic markup is not tested for because it might be complicated by
		-- adjacent bold markup.
	end
	if not disableSeaOfBlueCheck then
		if (string.find(wikitextLine, "%]%] *%[%[")
			or string.find(wikitextLine, "%]%] *'''+ *'''+ *%[%[")) then
			return "[[WP:DYKMOS]]/[[MOS:SEAOFBLUE]]: Two non-boldlinks or " ..
			"two boldlinks must be kept separate"
		end
	end
	if decodedLine:byte(-1) ~= 46 then -- Not an empty hook in prep area
		if not boldLinkFound then
			return "[[WP:DYKMOS]]: Every eligible article in the hook " ..
			"should be linked and wrapped in bold markup"
		end
		if decodedLine:byte(-1) ~= 63 then
			return "[[WP:DYKMOS]]: A hook should end in a question mark"
		end
	end
	if mw.ustring.find(decodedLine, ("[“”]")) then
		return '[[MOS:CURLY]]: Use "straight" quotation marks, not “curly” ones'
	end
	if mw.ustring.find(decodedLine, "[‘’]") then
		return "[[MOS:APOSTROPHE]]: Use straight apostrophes ('), " ..
		"not curly apostrophes (‘ or ’)"
	end
	-- Only humans can check if {{lang}} and {{transl}} are needed i.e. if
	-- the "non-English and transliterated text" is in "common English usage"
	-- [[WP:DYKMOS]] says that only ''most'' hooks begin with "that"
	-- [[MOS:CONTRACTION]] doesn't apply to quotes
end

-- General checks that should be uncontroversial for the entire page
local function general_checks(content)
	if content:find("[^[]%[//") or content:find("https?://") then
		-- Assuming nobody inserts a [[mw:Help:Links|mailto]] link
		return "[[WP:DYKMOS]]: The hook must not contain external links"
	end
	if content:find("|''[^]]+''%]") then
		return "[[WP:DYKMOS]]: Markup should go on the outside of the link " ..
		"if possible"
	end
	if content:find("%(''[^)]*pictured[^)]*''%)") then
		-- Media markers should be ''(pictured)'' like at
		-- [[Special:Diff/1287459353]] or similar,
		-- usually just with an extra word.
		-- But this shouldn't affect species, like
		-- (''Sitta pusilla'') at [[Special:Diff/1290843555]]
		return "[[WP:DYKMOS]]: Note that the italics should sit outside the "
		.. "parentheses"
	end
	if content:find("%(disambiguation%)") then
		return "[[WP:DYKMOS]]: The hook must not contain links to " ..
		"disambiguation pages"
	end
	local delinkedContent = content:gsub("%]", "")
	if (delinkedContent:find("[^']''%?")
		or delinkedContent:find("''''%?")
		or delinkedContent:find("[ ;?]%?")) then
		return "[[WP:DYKMOS]]: There should not be a space before the " ..
		"question mark, but if the text directly preceding it is " ..
		"italicized, the {{-?}} tag can offset it."
	end
	if delinkedContent:find("[^']'''{{%-%?}}") then
		return "[[WP:DYKMOS]]: (Only) if the text directly preceding it is " ..
		"italicized, the {{-?}} tag can offset it."
	end
	if delinkedContent:find("''s ") then
		return "[[WP:DYKMOS]]: Keep the bold / italic (or bold italic) text " ..
		"and the apostrophe distinct using {{`s}}/{{'s}} respectively"
	end
	if (delinkedContent:find("''''{{`") -- Use {{'}} with bold italic
		or delinkedContent:find("[^']''{{`") -- Use {{'}} with italic
		or delinkedContent:find("[^']'''{{'")) then -- Use {{`}} with bold
		-- Adapted from [[Template:Quotation mark templates]]
		return "[[WP:DYKMOS]]: {{`}} (or {{`s}}) is for adjacent bold " ..
		"markup; {{'}} (or {{'s}}) is for adjacent italic (or bold italic)"
	end
	if (delinkedContent:find("[^'] *{{'s}}")) then
		return "[[WP:DYKMOS]]: {{'s}} is only for italics"
	end
	if content:find("\n%*%.%.%.") then
		return "[[WP:DYKMOS]]: The three dots should be preceded by a space"
	end
	if content:find("\n%* %.%.%.that") then
		return "[[WP:DYKMOS]]: The three dots should be followed by a space"
	end
end

-- Checks to use for queues and complete prep areas
local function check_queue(content)
	if content:find("example%-serious%.jpg") then
		return "Image is still \"example-serious.jpg\""
	end
	if content:find("Caption goes here") then
		return "Caption is still \"Caption goes here\""
	end
	for article in content:gmatch("\n%* *%{%{DYK[^\n|]+|([^\n|]+)|") do
		article = article:lower()
		if article == "example" or article == "articlename" then
			return "[[WP:DYKPBI]]: DYKmake or DYKnom article is still " ..
			'"Example"/"ArticleName"'
		end
	end
end

-- Run all checks
-- Useful for the debug console:
-- =p._check("\n* ... that '''''[[Main Page]]''''' test ''(pictured)'' test?\n")
function p._check(content)
	local generalProblem = general_checks(content)
	if generalProblem then
		return generalProblem
	end

	local hookNum = 1
	local frame = mw.getCurrentFrame()
	-- Counting starts from after the space following the three dots
	for wikitextLine in content:gmatch("\n%* *%.%.%. *([^\n]*)") do
		wikitextLine = wikitextLine:gsub(" *%[%[File:Symbol confirmed%.svg.+$",
			"") -- {{subst:dyktick}}
		local expandedLine = frame:preprocess(wikitextLine)
		expandedLine = plainText(expandedLine, false)
		local decodedLine = decode(expandedLine)
		local hookProblem = check_hook(wikitextLine, expandedLine,
									   decodedLine, hookNum)
		if hookProblem then
			return hookProblem .. ": " .. decodedLine
		end

		for page in wikitextLine:gmatch("%[%[([^]|]+)|?[^]]*%]%]") do
			local linkProblem = check_link(page)
			if linkProblem then
				return linkProblem
			end
		end

		hookNum = hookNum + 1
	end

	-- No empty hooks ("... that ...") = queue or completed prep area
	if not content:find("\n%* %.%.%. that[^\n]*%.%.%.\n") then
		local queueProblem = check_queue(content)
		if queueProblem then
			return queueProblem
		end
	end
end

function p.main(frame)
	if disableAll then
		return
	end

	local message = p._check(mw.title.getCurrentTitle().content)
	if message then
		return frame:expandTemplate{title = "Ombox", args = {type = "style",
			text = '<p style="color:var(--color-error,#bf3c2c);font-weight:' ..
				'bold">' .. message .. "</p>" .. frame:expandTemplate{
					title = "Navbar",
					args = {
						"Module:DYK queue formatting check",
						text = "This warning module:"
					}
				}
		}}
	end
end

return p
This page was redesigned at the edge.

Module:DYK queue formatting check

Usage
