Module:File

Documentation for this module may be created at Module:File/doc

-- Module for handling Media files (Origin: Wikimedia Commons)

-- Helpers
local h = {}
h.expandPage = function(fullPageName, args)
	-- This expansion may fail, use it with pcall and inspect error status
	return mw.getCurrentFrame():expandTemplate{title = ':' .. fullPageName, args}
end
h.expandPageNoFail = function(fullPageName, args)
	local ok, wikiText = pcall(h.expandPage, fullPageName, args)
	if ok then
		return wikiText
	end
	return ''
end
-- Limits in megapixels are currently stored in Commons templates, they could be in this module.
-- There may be more limits for other supported mime types (djvu, flac, ogv, pdf, svg, webm, xcf).
h.maxthumbMap = {
--	['image/gif'] = h.expandPageNoFail('Template:LargeTIFF/limit', {}), -- current value is '50' but higher now
	['image/png' ] = h.expandPageNoFail('Template:LargePNG/limit', {}), -- current value is '2500'
	['image/tiff'] = h.expandPageNoFail('Template:LargeTIFF/limit', {}), -- current value is '50' but higher now
}
h.maxthumbMap['image/gif' ] = h.maxthumbMap['image/tiff'] -- current value is '50' but higher now
h.extensionMap = {
	-- File types with full support in Commons (See [[c:Commons:File types]]).
	DJVU = 'image/vnd.djvu',
	FLAC = 'audio/x-flac',
	GIF  = 'image/gif',
	JPEG = 'image/jpeg',
	JPG  = 'image/jpeg',
	MID  = 'audio/midi',
	MP3  = 'audio/mpeg',
	OGA  = 'audio/ogg',
	OGG  = 'audio/ogg',
	OGV  = 'video/ogg',
	PDF  = 'application/pdf',
	PNG  = 'image/png',
	SVG  = 'image/svg+xml',
	TIF  = 'image/tiff',
	TIFF = 'image/tiff',
	WEBM = 'video/webm',
	WAV  = 'audio/x-wav',
	XCF  = 'image/xcf',

	-- Other file types with restrictions (not accepted in standard uploads on Commons but in 'User:' namespace).
	-- They could be supported in Wiki pages by embedding their content in an <pre> or <source> elements.
	CSS  = 'text/css',
	CSV  = 'text/csv',
	JS   = 'application/javascript',
	JSON = 'application/json',
	TXT  = 'text/plain',
	XML  = 'application/xml',

	-- Only generated by MediaWiki on output of some queries, restricted in all uploads.
	GZ   = 'application/gzip', -- delivered only only for some wiki results
	ZIP  = 'application/zip', -- delivered only for some wiki data exports

	-- Other file types not supported and to convert (a few of them may be in special administration namespaces).
	DOC  = 'application/msword', -- please convert to PDF, DJVU, or Wiki
	F4V  = 'video/mpeg', -- (deprecated, replaced by MP4) please convert to OGV or WEBM
	FLV  = 'video/x-flv', -- (deprecated, replaced by MP4) please convert to OGV or WEBM
	ICO  = 'image/vnd.microsoft.icon', -- used in MediaWiki resources for 'website icons'
	MP4  = 'video/mp4', -- please convert to OGV or WEBM
	QT   = 'video/quicktime', -- (deprecated, replaced by MP4) please convert to OGV or WEBM
	RA   = 'audio/vnd.rn-realaudio', -- (deprecated, replaced by MP3) please convert to OGA
	SWF  = 'video/x-flv', -- (deprecated, replaced by MP4) please convert to OGV or WEBM
	WMA  = 'audio/x-ms-wma', -- please convert to OGA
	WMV  = 'video/x-ms-wmv', -- please convert to OGV or WEBM
	XLS  = 'application/vnd.ms-excel', -- please convert to PDF, DJVU, or Wiki
}
h.parse = require("Module:HTMLParser").parse

local File = function(title)
	local funcs = {}
	local titleInstance, metadataInstance

	function getTitle()
		if titleInstance == nil then
			titleInstance = mw.title.new(title, 6)
		end
		return titleInstance
	end

	function getFullName()
		return getTitle().prefixedText
	end

	-- =p.File("Foo.bar.svg").extension()
	-- @return "svg"
	funcs.extension = function()
		local parts = mw.text.split(title, '.', true)
		return parts[#parts]
	end

	-- =p.File("Foo.bar.svg").woExtension()
	-- @return "Foo.bar"
	-- Original author: Bawolff at [[Module:FileName]]
	funcs.woExtension = function()
		local parts = mw.text.split(title , '.', true)
		local upTo = #parts - 1
		if upTo == 0 then upTo = 1 end
		return table.concat(parts, '.', 1, upTo)
	end
	
	function getMetadata()
		if metadataInstance == nil then
			metadataInstance = getTitle().file
		end
		return metadataInstance
	end
	funcs.metadata = function()
		return getMetadata()
	end

	-- Mapping file extensions to MIME-types (only MIME types accepted for files in Commons).
	-- Works even if file still does not exist.
	-- =p.File('Foo.bar.svg').extension()
	-- @return 'image/svg+xml'
	funcs.mime = function()
		local ok, metadata = pcall(funcs.metadata())
		if ok and metadata.exists then
			-- Note: does not work if file does not exist, where metadata == {'exists': false} only
			return metadata.mimeType
		end
		-- mw.log('mime() is deprecated. Use metadata().mimeType.')
		return h.extensionMap[funcs.extension():upper()] or 'unknown'
	end

	-- =p.File('Foo.bar.tiff').maxthumb()
	funcs.maxthumb = function()
		return h.maxthumbMap[funcs.mime()] or 'unknown @Module:File'
	end

	funcs.dateWorkCreated = function()
		-- Parse the expanded wiki text into an html root node and select a child node by specific ID.
		local root = h.parse('<html>' ..
			h.expandPageNoFail(getFullName(), {})
				:gsub('<nowiki>.*?</nowiki>', '')
			.. '</html>')
		local tdElem = root('#fileinfotpl_date')
		for td in pairs(tdElem) do -- We queried an ID so there should be only one result in this loop.
			-- We need the next sibling, which doesn't seem to be directly supported by HTMLParser.
			-- ... so ask him for the parent <tr> node and find the first <time> element in it.
			local timeElem = td.parent('time')
			for t in pairs(timeElem) do
				return t.attributes['datetime']
			end
		end
	end

	return funcs
end
h.getFile = function(frame)
	return File(frame.args[1] or frame.args["file"] or frame.args["title"] or mw.title.getCurrentTitle().text)
end

-- @exports
local p = {}
p.File = File
p.extension = function(frame)
	return h.getFile(frame).extension():lower()
end
p.extensionUpper = function(frame)
	return h.getFile(frame).extension():upper()
end
p.csExtension = function(frame)
	return h.getFile(frame).extension()
end
p.woExtension = function(frame)
	return h.getFile(frame).woExtension()
end
p.mime = function(frame)
	return h.getFile(frame).mime()
end
p.fileExists = function(frame)
	return h.getFile(frame).metadata().exists or ''
end
-- This one won't throw errors at you
p.fileExistsRelaxed = function(frame)
	local ok, metadata = pcall(h.getFile(frame).metadata)
	if ok then
		return metadata.exists or ''
	else
		return ''
	end
end
p.mimeType = function(frame)
	return h.getFile(frame).metadata().mimeType
end
p.maxthumb = function(frame)
	return h.getFile(frame).maxthumb()
end
p.dateWorkCreated = function(frame)
	return h.getFile(frame).dateWorkCreated(frame) or ''
end
p.width = function(frame)
	return h.getFile(frame).metadata().width
end
p.height = function(frame)
	return h.getFile(frame).metadata().height
end
p.dimensions = function(frame)
	local d = h.getFile(frame).metadata()
	return d.width .. ' × ' .. d.height
end
p.size = function(frame)
	return h.getFile(frame).metadata().size
end
p.pageCount = function(frame)
	local pages = h.getFile(frame).metadata().pages or {1}
	return #pages
end
-- Module autotest
p.runTests = function()
	local toTest = require('Module:File/tests/all')
	local result = true
	for i, t in ipairs(toTest) do
		local f = File(t.fileName)
		local stringResult = ''
		local ret = true
		local results = {
			extension = (t.extension == f.extension()),
			extensionLower = (t.extensionLower == f.extension():lower()),
			extensionUpper = (t.extensionUpper == f.extension():upper()),
			woExtension = (t.woExtension == f.woExtension()),
			mime = (t.mime == f.mime()),
			maxthumb = (not (tonumber(f.maxthumb()) == nil) == t.maxthumbIsNumber),
			dateWorkCreated = t.dateWorkCreated == f.dateWorkCreated()
		}
		for k, v in pairs(results) do
			stringResult = stringResult .. k .. ': ' .. (v and 'ok    ' or 'failed') .. ' '
			ret = ret and v
		end
		mw.log(i, ret and 'passed' or 'FAILED', t.typeOfFileName, (not ret) and ('\n\t>>\t' .. t.fileName .. '\n\t>>\t' .. stringResult) or '')
		result = result and ret
	end
	return result
end
-- p.h = h -- if needed for running some tests

return p