Module:category tree/wiktionary maintenance

local raw_categories = {}
local raw_handlers = {}

local m_template_parser = require("Module:template parser")

local get_lang = require("Module:languages").getByCode
local insert = table.insert
local is_internal_title = require("Module:pages").is_internal_title
local new_title = mw.title.new
local split_lang_label = require("Module:category tree").split_lang_label
local php_trim = require("Module:Scribunto").php_trim
local uses_hidden_category = require("Module:maintenance category").uses_hidden_category

-----------------------------------------------------------------------------
--                                                                         --
--                              RAW CATEGORIES                             --
--                                                                         --
-----------------------------------------------------------------------------

raw_categories = {
	description = "Categories containing pages that are being tracked for attention and improvement by editors.",
	breadcrumb = "Maintenance",
	parents = "Wiktionary",
}

raw_categories = {
	topright = "{{shortcut|CAT:EC}}",
	description = "Categories with no members.",
	additional = ] when they contain no pages or subcategories.

Empty categories are not necessarily a problem, but they can clutter up their parent categories, or become orphaned if the structure of the category tree changes. This category therefore helps track down such cases, and allows them to be cleaned up.

Because of the way the wiki software works, categories will appear here for a while afterwards if they were empty at first but had entries added to them later. This can be fixed by simply performing a "null edit" on the category page: edit the page, and save without making any changes. (Alternatively, use the "null edit" option provided by the "purge tab" ].) This can be avoided by adding entries to categories before creating them. It also helps to create categories from the "bottom up": start at the lowest level that has entries, then create its parent categories, then the parent categories of that, and so on.]=],
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories with names that do not match the expected form within the category tree.",
	additional = [=[This usually happens when additional parameters have been given to {{tl|auto cat}} that don't match the name of the category, or when there is a problem with capitalization or spacing in the category name.

==See also==
* ]]=],
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories which use {{tl|auto cat}}, but which are not registered in the category tree data modules.",
	additional = [=[See the error box displayed on any of these categories for more info.

==See also==
* ]]=],
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories using the <code>]</code> behavior switch, which hides the category from the lists of categories in its members and subcategories.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories with more than 1 million members.",
	additional = "Such categories have the ] extension disabled, which is normally used to list the newest and oldest pages in a category. This is because categories above that size load very slowly when it is enabled, and in some cases become inaccessible due to timing-out.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages using the <code>]</code> behavior switch, which tells search engines to index the page.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages using the <code>]</code> behavior switch, which tells search engines not to index the page.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "This category contains entries, reconstruction pages, appendixes, sign glosses and citations pages using deprecated templates—templates that have failed our deletion process, and/or that have been replaced by superior templates.",
	additional = ]). The latter is wrapped around non-deprecated templates that accept the deprecated {{para|lang}} parameter; any use of that parameter will place the page in ]. Ideally, this category will be empty. 

Any pages in this category, particularly those in the mainspace, need to have their deprecated template usages corrected. ]=],
	breadcrumb = "Using deprecated templates",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages which use at least one of the lite templates.",
	additional = "See ].",
	breadcrumb = "Using lite templates",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages which contain a tab character in their wikitext.",
	additional = "These should either be removed or replaced with spaces, because they go against ].",
	breadcrumb = "Tab characters",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages in which the headings for each language's entry are in the wrong order.",
	additional = "Level 2 language headings should be in alphabetical order, except for Translingual and English, which go at the top (in that order).",
	breadcrumb = "Language headings in the wrong order",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages which contain a level 2 heading which does not match any language's canonical name.",
	additional = "The level 2 language heading for each language should always be that language's canonical name.",
	breadcrumb = "Nonstandard language headings",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages which contain an unwanted level 1 heading.",
	additional = "Level 1 headings are not used in Wiktionary content pages, and only occur due to user error or vandalism.",
	breadcrumb = "Unwanted L1 headings",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages which contain raw template parameters in the form of triple braces.",
	additional = "Triple-brace template parameters (e.g. {{param|param}}) are intended for use in templates, as they are substituted with the relevant template argument when the page is transcluded. Although they can theoretically be used on any page, there are currently no legitimate uses for them in content namespaces.\n\nTemplate parameters usually occur due to typos, or when {{tl|subst:}} has been used with a template that isn't supposed to be substed.",
	breadcrumb = "Raw template parameters",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	topright = "{{shortcut|CAT:DEFAULTSORT}}",
	description = "Pages on which the {{tl|DEFAULTSORT:}} magic word has been used multiple times with different values.",
	additional = "In some (but not all) cases, this causes a warning to display on the page. In the vast majority of instances, an explicit use of {{tl|DEFAULTSORT:}} in wikitext should be <u>removed</u>.This is because the {{tl|head}} template handles it automatically. The only instances where it should be used in wikitext is outside of entries (i.e. outside of mainspace or the Reconstruction namespace)." ..
		"\n\nSee also ].",
	breadcrumb = "DEFAULTSORT conflicts",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	topright = "{{shortcut|CAT:DISPLAYTITLE}}",
	description = "Pages on which the {{tl|DISPLAYTITLE:}} magic word has been used multiple times with different values.",
	additional = "In some (but not all) cases, this causes a warning to display on the page. In the vast majority of instances, an explicit use of {{tl|DISPLAYTITLE:}} in wikitext should be <u>removed</u>.This is because the {{tl|head}} template handles it automatically. The only instances where it should be used in wikitext is outside of entries (i.e. outside of mainspace or the Reconstruction namespace)." ..
		"\n\nSee also ].",
	breadcrumb = "DISPLAYTITLE conflicts",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages on which a sortkey has been used with a raw category.",
	additional = "For example, {{code|]}}." ..
	"\n\nThese are a priority to replace with category templates, since they are hard-coded and override the {{tl|DEFAULTSORT:}} value for the page. This causes problems if there are any changes to the sorting scheme for the category, because there is no way of changing them centrally.\n\n" ..
	"By comparison, raw categories which have no sortkey are less of a problem, because they will use the {{tl|DEFAULTSORT:}} value; this can be centrally controlled and is designed to be language-neutral, so avoids the issue of different editors using multiple different sorting schemes for the same category. However, they should still be replaced with category templates, since there may be additional language-specific sorting rules which cannot otherwise be applied.",
	breadcrumb = "Raw sortkeys",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	topright = "{{shortcut|CAT:E|CAT:ERR|CAT:ERROR}}",
	description = "Pages that have errors in a ] module.",
	additional = "If entries are listed here for more than a day or two, the error should probably be reported at ]. Memory errors are a common source of these errors; see the discussion at ]." ..
		"\n\nBecause the software does not immediately update pages when a change occurs in a template or module, errors listed here may have already been fixed. Therefore, please ensure that the error is still present before reporting problems. You can do this by performing a \"]\" (editing the page and saving without making changes). If the error goes away then, it has already been fixed." ..
		"\n\n<u>You can use  and press \"Make request\" to purge the cache of up to 20 pages from this category in one click.</u> This number can be adjusted up to 5,000, but anything above 30–100 will likely cause time-outs (depending on the size of the pages)." ..
		"\n\nThe contents of this category is controlled by ]. It is currently set to place talk pages, user pages{{,}} and user sandbox modules and templates in a separate category." ..
		"\n\nSee also ].",
	breadcrumb = "Module errors",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	topright = "{{shortcut|CAT:PFE}}",
	description = "Pages that have errors in a ] magic word.",
	additional = "Examples of these magic words are {{tl|#expr:}} and {{tl|#time:}}. If entries are listed here for more than a day or two, the error should probably be reported at ]." ..
		"\n\nBecause the software does not immediately update pages when a change occurs in a template or module, errors listed here may have already been fixed. Therefore, please ensure that the error is still present before reporting problems. You can do this by performing a \"]\" (editing the page and saving without making changes). If the error goes away then, it has already been fixed." ..
		"\n\n<u>You can use  and press \"Make request\" to purge the cache of up to 20 pages from this category in one click.</u> This number can be adjusted up to 5,000, but anything above 30–100 will likely cause time-outs (depending on the size of the pages)." ..
		"\n\nThe contents of this category is controlled by ]. It is currently set to place talk pages, user pages{{,}} and user sandbox modules and templates in a separate category." ..
		"\n\nSee also ].",
	breadcrumb = "ParserFunction errors",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages and categories which have been tagged with a request for them to be moved, merged or split.",
	breadcrumb = "Moves, mergers and splits",
	parents = {"Wiktionary maintenance", "Requests"},
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages tagged to be merged by the {{tl|merge}} template.",
	parents = "Requests for moves, mergers and splits",
	can_be_empty = true,
}

raw_categories = {
	description = "Pages tagged to be moved by the {{tl|move}} template.",
	parents = "Requests for moves, mergers and splits",
	can_be_empty = true,
}

raw_categories = {
	description = "Pages tagged to be split by the {{tl|split}} template.",
	parents = "Requests for moves, mergers and splits",
	can_be_empty = true,
}

raw_categories = {
	description = "Pages that use unrecognized parameters when calling a template.",
	breadcrumb = "Invalid template parameters",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
}

raw_categories = {
	description = "Pages that use the <code>]</code> gadget.",
	additional = "This processes links to entries in language-specific categories by adding language-specific formatting, and points them to the language's section of the entry.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages that display a mini table of contents by calling {{tl|minitoc}}.",
	additional = "This is is used on very large pages with many entries, to assist with navigation.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories that have been placed in another category by calling {{tl|auto cat}}.",
	additional = "This is the preferred way for categories to be subcategorized. The chief reason for this category is to facilitate the finding of categories which are not using {{tl|auto cat}} through the use of negative searches (e.g. qualifying a search with {{code|-incategory:\"{{PAGENAME}}\"}}).",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Categories that have been placed in another category using raw wiki markup (e.g. {{cl|Wiktionary}}). They should be added to the ] data instead.",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

raw_categories = {
	description = "Pages that use the ] extension with legacy {{wt|source}} tags instead of {{wt|syntaxhighlight}}.",
	breadcrumb = "Deprecated source tags",
	parents = "Wiktionary maintenance",
	can_be_empty = true,
	hidden = true,
}

insert(raw_handlers, function(data)
	local template_type = data.category:match("^Pages using invalid parameters when calling (.+) templates$")
	
	if not template_type then
		return
	end
	
	local parents = {{
		name = "Pages using invalid parameters when calling templates",
		sort = template_type == "general use" and "*" or template_type,
	}}
	
	local lang = require("Module:languages").getByCanonicalName(template_type, nil, true)
	if lang then
		insert(parents, {name = "entry maintenance", is_label = true, lang = lang:getCode()})
	end
	
	return {
		lang = lang and lang:getCode() or nil,
		description = "Pages that use unrecognized parameters when calling " .. template_type .. " templates.",
		parents = parents,
		breadcrumb = template_type,
	}
end)

do
	local prefixes = require("Module:table").listToSet{"list", "P", "R", "RQ", "table", "U"}
	
	local function add_parent(parents, seen, cat_type, sortkey)
		if seen then
			return
		end
		insert(parents, {
			name = ("Pages using invalid parameters when calling %s templates"):format(cat_type),
			sort = sortkey,
		})
		seen = true
	end
	
	insert(raw_handlers, function(data)
		local template = data.category:match("^Pages using invalid parameters when calling (.+)$")
		
		if not template then
			return
		end
		
		-- Resolve any redirects.
		template = new_title(template)
		while template do
			local redirect = template.redirectTarget
			if not (redirect and is_internal_title(redirect)) then
				break
			end
			template = redirect
		end
		
		-- Disallow templates which would always hidden maintennace categories (e.g. sandboxes).
		if not (template and not uses_hidden_category(template)) then
			return
		end
		
		local prefixed_text, lang = template.prefixedText
		if template.namespace == 10 then
			local name = template.text
			-- Remove the prefix if present (e.g. "R:" or "RQ:").
			local prefix, text = name:match("^(.-):(.+)")
			if not (prefix and prefixes) then
				text = name
			end
			-- Check the initial language code, chopping off hyphenated sections until there's a match or they run out.
			local code = mw.ustring.match(text, "^*%f")
			while code do
				lang = get_lang(code)
				if lang then
					break
				end
				code = code:match("(.+)%-%a*$")
			end
			-- If no match and it's a list: or table: template, check if the template name ends "/CODE".
			if not lang and (prefix == "list" or prefix == "table") then
				code = text:match("%f%l*%a$")
				if code then
					lang = get_lang(code)
				end
			end
		end
		
		local sortkey = template.text
		local parents, seen = {}, {}
		
		-- Categorize as language-specific if a language was found.
		if lang then
			add_parent(parents, seen, lang:getCanonicalName(), sortkey)
		end
		
		-- Also grab any language categories from the template page.
		for _, cat in ipairs(template.categories) do
			if cat:sub(-10) == " templates" or cat:sub(-13) == " subtemplates" then
				local cat_lang = split_lang_label(new_title(cat).text)
				if cat_lang then
					add_parent(parents, seen, cat_lang:getCanonicalName(), sortkey)
				end
			end
		end
		
		-- If none were found, categorize as general use.
		if #parents == 0 then
			add_parent(parents, seen, "general use", sortkey)
		end
		
		-- Only add can_be_empty if the template exists and contains checkparams.
		local content, can_be_empty = template:getContent()
		if content then
			-- Check for {{#invoke:checkparams|warn|...}}.
			-- args is the module and args is the function name, so #INVOKE: will throw an error if either is not present.
			for template in require("Module:template parser").find_templates(content) do
				if template:get_name() == "#INVOKE:" then
					local args = template:get_arguments()
					local arg_2 = args
					if arg_2 and php_trim(args) == "checkparams" and php_trim(arg_2) == "warn" then
						can_be_empty = true
						break
					end
				end
			end
		end
		
		return {
			canonical_name = "Pages using invalid parameters when calling " .. prefixed_text,
			lang = lang and lang:getCode() or nil,
			description = ("Pages that use unrecognized parameters when calling {{tl|%s}}.")
				:format(m_template_parser.getTemplateInvocationName(template)),
			additional = "These template calls should be reviewed and the invalid parameter(s) should be corrected or removed.",
			breadcrumb = prefixed_text,
			parents = parents,
			can_be_empty = can_be_empty,
			hidden = true,
		}
	end)
end

return {RAW_CATEGORIES = raw_categories, RAW_HANDLERS = raw_handlers}