Module:FileInfo/Reports

Generates the reports under Template:FileInfo/Reports.

---

-- This function is only used on [[Template:FileInfo/Reports]] 
-- so it lives in its own module instead of Module:File or Module:FileInfo

-- This way we can change it without triggering a massive job queue 
-- for all the pages linked by those high-usage modules
local p = {}

local utilsCargo = require("Module:UtilsCargo")
local utilsLayout = require("Module:UtilsLayout")
local utilsMarkup = require("Module:UtilsMarkup")
local utilsPage = require("Module:UtilsPage")
local utilsTable = require("Module:UtilsTable")

function p.LackingFileInfo(frame)
	local fileList = frame:preprocess("{{#dpl:"
		.."|namespace= File"
		.."|notuses= Template:FileInfo"
		.."|notuses= Template:Delete" -- to account for file pages marked for deletion after the files have been moved elsewhere
		.."|resultsheader= %TOTALPAGES% files are missing [[Template:FileInfo]]:\n"
		.."}}"
	)
	if fileList == "" then
		return frame:preprocess("[[File:TFH Green Link ok.png|32px|link=]] All files are using [[Template:FileInfo]].")
	else
		return fileList.."[[Category:Files needing attention]]"
	end
	return fileList
end

-- Queries Cargo for the 100 most-used subjects, uses DPL to determine which ones don't exist as categories yet.
-- Repeat for the next 100 subjects, and so on until there are ~100 table rows or no more subjects to process.
function p.MostWantedSubjectCategories(frame)
	local BATCH_SIZE = 100
	local MAX_ROWS = 100
	local offset = 0
	local rows = {}
	local listedCategories = {} 
	local cargoResults
	repeat 
		cargoResults = utilsCargo.query("Files, Files__subject", "Files._pageName=page, Files__subject._value=subject, COUNT(*)=count", {
			groupBy = "_value",
			orderBy = "COUNT(*) DESC",
			join = "Files._ID = Files__subject._rowID",
			where = " Files__subject._value != ''",
			limit = BATCH_SIZE,
			offset = offset
		})
		offset = offset + BATCH_SIZE
	
		local dplArgs = {
			namespace = "Category",
			titlematch = utilsTable.map(cargoResults, function(result)
				return "Images of "..result.subject
			end), 
			redirects = "include", -- some "Images of X" categories are redirects to synonymous terms (see Category:Synonymous Files)
		}
		dplArgs.namespace = "Category"
		dplArgs.redirects = "include" 
		local existingCategories = utilsPage.dpl(dplArgs)
		local existingCategoryLookup = utilsTable.invert(existingCategories)
		
		for _, result in ipairs(cargoResults) do
			local cat = "Category:Images of "..result.subject
			if not listedCategories[cat] and not existingCategoryLookup[cat] then
				table.insert(rows, { utilsMarkup.link(cat), result.count, utilsMarkup.link(result.page)})
				listedCategories[cat] = true -- because of the way we batch the cargo queries, we need keep track of results that were already added
			end
		end
	until #rows >= MAX_ROWS or #cargoResults == 0
	return utilsLayout.table({
		sortable = true,
		headers = {"Category", "Count", "Sample File"},
		rows = rows
	})
end

return p