Module:Wikidata

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

local p = {}

-- module local variables
local wiki = 
{
	langcode = mw.language.getContentLanguage().code
}

-- internationalisation
local i18n = {
    ["errors"] = {
        ["property-not-found"] = "Property not found.",
        ["entity-not-found"] = "Wikidata entity not found.",
        ["unknown-claim-type"] = "Unknown claim type.",
        ["unknown-entity-type"] = "Unknown entity type.",
        ["qualifier-not-found"] = "Qualifier not found.",
        ["site-not-found"] = "Wikimedia project not found.",
    },
    ["datetime"] =
	{
		-- $1 is a placeholder for the actual number
		[0] = "$1 billion years",	-- precision: billion years
		[1] = "$100 million years",	-- precision: hundred million years
		[2] = "$10 million years",	-- precision: ten million years
		[3] = "$1 million years",	-- precision: million years
		[4] = "$100,000 years",		-- precision: hundred thousand years
		[5] = "$10,000 years",		-- precision: ten thousand years
		[6] = "$1  millenium",	 	-- precision: millennium
		[7] = "$1 century",			-- precision: century
		[8] = "$1s",				-- precision: decade
		-- the following use the format of #time parser function
		[9]  = "Y",					-- precision: year, 
		[10] = "F Y",				-- precision: month
		[11] = "F j, Y",			-- precision: day
		[12] = "F j, Y ga",			-- precision: hour
		[13] = "F j, Y g:ia",		-- precision: minute
		[14] = "F j, Y g:i:sa",		-- precision: second
		["beforenow"] = "$1 BCE",	-- how to format negative numbers for precisions 0 to 5
		["afternow"] = "$1 CE",		-- how to format positive numbers for precisions 0 to 5
		["bc"] = '$1 "BCE"',		-- how print negative years
		["ad"] = "$1"				-- how print positive years
	},
	["monolingualtext"] = '<span lang="%language">%text</span>'
}

function p.descriptionIn(frame)
	local langcode = frame.args[1]
	local id = frame.args[2]	-- "id" must be nil, as access to other Wikidata objects is disabled in Mediawiki configuration
	-- return description of a Wikidata entity in the given language or the default language of this Wikipedia site
	return mw.wikibase.getEntityObject(id).descriptions[langcode or wiki.langcode].value
end

function p.labelIn(frame)
	local langcode = frame.args[1]
	local id = frame.args[2]	-- "id" must be nil, as access to other Wikidata objects is disabled in Mediawiki configuration
	-- return label of a Wikidata entity in the given language or the default language of this Wikipedia site
	return mw.wikibase.getEntityObject(id).labels[langcode or wiki.langcode].value
end

-- This is used to get a value, or a comma separated list of them if multiple values exist
p.getValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local input_parm = mw.text.trim(frame.args[2] or "")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		local claims = entity.claims[propertyID]
		if claims then
			-- if wiki-linked value output as link if possible
			if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then
				local out = {}
				for k, v in pairs(claims) do
					local sitelink = mw.wikibase.sitelink("Q" .. v.mainsnak.datavalue.value["numeric-id"])
					local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])
					if label == nil then label = "Q" .. v.mainsnak.datavalue.value["numeric-id"] end
							
					if sitelink then
						out[#out + 1] = "[[" .. sitelink .. "|" .. label .. "]]"
					else
						out[#out + 1] = "[[:d:Q" .. v.mainsnak.datavalue.value["numeric-id"] .. "|" .. label .. "]]<abbr title='Article is not yet available in this wiki'>[*]</abbr>"
					end
				end
				return table.concat(out, ", ")
			else
				return entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
			end
		else
			return ""
		end
	else
		return input_parm
	end
end

p.getQualifierValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local qualifierID = mw.text.trim(frame.args[2] or "")
	local input_parm = mw.text.trim(frame.args[3] or "")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		if entity.claims[propertyID] ~= nil then
			local out = {}
			for k, v in pairs(entity.claims[propertyID]) do
				for k2, v2 in pairs(v.qualifiers[qualifierID]) do
					if v2.snaktype == 'value' then
						if (mw.wikibase.sitelink("Q" .. v2.datavalue.value["numeric-id"])) then
							out[#out + 1] = "[[" .. mw.wikibase.sitelink("Q" .. v2.datavalue.value["numeric-id"]) .. "]]"
						else
							out[#out + 1] = "[[:d:Q" .. v2.datavalue.value["numeric-id"] .. "|" .. mw.wikibase.label("Q" .. v2.datavalue.value["numeric-id"]) .. "]]<abbr title='Article is not yet available in this wiki'>[*]</abbr>"
						end
					end
				end
			end
			return table.concat(out, ", ")
		else
			return ""
		end
	else
		return input_parm
	end
end

-- This is used to get a value like 'male' (for property p21) which won't be linked and numbers without the thousand separators
p.getRawValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local input_parm = mw.text.trim(frame.args[2] or "")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		local claims = entity.claims[propertyID]
		if claims then
			local result = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value
		
			-- if number type: remove thousand separators
			if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "quantity") then
				result = mw.ustring.gsub(result, "(%d),(%d)", "%1%2")
			end
			return result
		else
			return ""
		end
	else
		return input_parm
	end
end

p.getRawQualifierValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local qualifierID = mw.text.trim(frame.args[2] or "")
	local input_parm = mw.text.trim(frame.args[3] or "")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		if entity.claims[propertyID] ~= nil then
			local out = {}
			for k, v in pairs(entity.claims[propertyID]) do
				for k2, v2 in pairs(v.qualifiers[qualifierID]) do
					if v2.snaktype == 'value' then
						if v2.datavalue.value["numeric-id"] then
							out[#out + 1] = mw.wikibase.label("Q" .. v2.datavalue.value["numeric-id"])
						else
							out[#out + 1] = v2.datavalue.value
						end
					end
				end
			end
			local ret = table.concat(out, ", ")
			return string.upper(string.sub(ret, 1, 1)) .. string.sub(ret, 2)
		else
			return ""
		end
	else
		return input_parm
	end
end

-- This is used to get a date value for date_of_birth (p569), etc. which won't be linked -- consolidate by testing if entity.claims[propertyID].mainsnak.datavalue.type is "time"
-- Dates are stored as 28 characters if the year  >99 -- e.g. +00000002014-01-01T00:00:00Z for 2014
-- Dates are stored as 26 characters if the year =<99 -- e.g. +000000050-01-01T00:00:00Z   for 50 CE
p.getDateValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local input_parm = mw.text.trim(frame.args[2] or "")
	local date_format = mw.text.trim(frame.args[3] or "dmy")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		if entity.claims[propertyID] ~= nil then
			local out = {}
			local dt = {}
			for k, v in pairs(entity.claims[propertyID]) do
				if v.mainsnak.snaktype == 'value' then
					local d = v.mainsnak.datavalue.value.time
					if #d > 26 then
						dt.year = string.sub(d, 9, 12)
						dt.month = string.sub(d, 14, 15)
						dt.day = string.sub(d, 17, 18)
					else
						dt.year = string.sub(d, 9, 10)
						dt.month = string.sub(d, 12, 13)
						dt.day = string.sub(d, 15, 16)
					end
					if date_format == "mdy" then
						out[#out + 1] = os.date("%B %e, %Y", os.time(dt))
					elseif date_format == "my" then
						out[#out + 1] = os.date("%B %Y", os.time(dt))
					elseif date_format == "y" then
						out[#out + 1] = os.date("%Y", os.time(dt))
					else
						out[#out + 1] = os.date("%e %B %Y", os.time(dt))
					end
				end
			end
			return table.concat(out, ", ")
		else
			return ""
		end
	else
		return input_parm
	end
end

p.getQualifierDateValue = function(frame)
	local propertyID = mw.text.trim(frame.args[1] or "")
	local qualifierID = mw.text.trim(frame.args[2] or "")
	local input_parm = mw.text.trim(frame.args[3] or "")
	local date_format = mw.text.trim(frame.args[4] or "dmy")
	if input_parm == "FETCH_WIKIDATA" then
		local entity = mw.wikibase.getEntityObject()
		if entity.claims[propertyID] ~= nil then
			local out = {}
			local dt = {}
			for k, v in pairs(entity.claims[propertyID]) do
				for k2, v2 in pairs(v.qualifiers[qualifierID]) do
					if v2.snaktype == 'value' then
						local d = v2.datavalue.value.time
						if #d > 26 then
							dt.year = string.sub(d, 9, 12)
							dt.month = string.sub(d, 14, 15)
							dt.day = string.sub(d, 17, 18)
						else
							dt.year = string.sub(d, 9, 10)
							dt.month = string.sub(d, 12, 13)
							dt.day = string.sub(d, 15, 16)
						end
						if date_format == "mdy" then
							out[#out + 1] = os.date("%B %e, %Y", os.time(dt))
						elseif date_format == "my" then
							out[#out + 1] = os.date("%B %Y", os.time(dt))
						elseif date_format == "y" then
							out[#out + 1] = os.date("%Y", os.time(dt))
						else
							out[#out + 1] = os.date("%e %B %Y", os.time(dt))
						end
					end
				end
			end
			return table.concat(out, ", ")
		else
			return ""
		end
	else
		return input_parm
	end
end

-- This is used to get the TA98 (Terminologia Anatomica first edition 1998) values like 'A01.1.00.005' (property P1323)
-- which are then linked to http://www.unifr.ch/ifaa/Public/EntryPage/TA98%20Tree/Entity%20TA98%20EN/01.1.00.005%20Entity%20TA98%20EN.htm
-- uses the newer mw.wikibase calls instead of directly using the snaks
-- formatPropertyValues returns a table with the P1323 values concatenated with ", " so we have to split them out into a table in order to construct the return string
p.getTAValue = function(frame)
	local ent = mw.wikibase.getEntityObject()
	local props = ent:formatPropertyValues('P1323')
	local out = {}
	local t = {}
	for k, v in pairs(props) do
		if k == 'value' then
			t = mw.text.split( v, ", ")
			for k2, v2 in pairs(t) do
				out[#out + 1] = "[http://www.unifr.ch/ifaa/Public/EntryPage/TA98%20Tree/Entity%20TA98%20EN/" .. string.sub(v2, 2) .. "%20Entity%20TA98%20EN.htm " .. v2 .. "]"
			end
		end
	end
	ret = table.concat(out, "<br> ")
	if #ret == 0 then
		ret = "Invalid TA"
	end
	return ret
end

-- returns the page id (Q...) of the current page or nothing of the page is not connected to Wikidata
function p.pageId(frame)
	local entity = mw.wikibase.getEntityObject()
	if not entity then return nil else return entity.id end
end

-- the "qualifiers" and "snaks" field have a respective "qualifiers-order" and "snaks-order" field
-- use these as the second parameter and this function instead of the built-in "pairs" function
-- to iterate over all qualifiers and snaks in the intended order.
local function orderedpairs(array, order)
	if not order then return pairs(array) end
 
	-- return iterator function
    local i = 0
    return function()
        i = i + 1
        if order[i] then
            return order[i], array[order[i]]
        end
    end	
end

-- precision: 0 - billion years, 1 - hundred million years, ..., 6 - millennia, 7 - century, 8 - decade, 9 - year, 10 - month, 11 - day, 12 - hour, 13 - minute, 14 - second
local function normalizeDate(date)
	date = mw.text.trim(date, "+")
	-- extract year
	local yearstr = mw.ustring.match(date, "^\-?%d+")
	local year = tonumber(yearstr)
	-- remove leading zeros of year
	return year .. mw.ustring.sub(date, #yearstr + 1), year
end

function formatDate(date, precision, timezone)
	precision = precision or 11
	date, year = normalizeDate(date)
	if year == 0 and precision <= 9 then return "" end
 
 	-- precision is 10000 years or more
	if precision <= 5 then
		local factor = 10 ^ ((5 - precision) + 4)
		local y2 = math.ceil(math.abs(year) / factor)
		local relative = mw.ustring.gsub(i18n.datetime[precision], "$1", tostring(y2))
		if year < 0 then
			relative = mw.ustring.gsub(i18n.datetime.beforenow, "$1", relative)
		else
			relative = mw.ustring.gsub(i18n.datetime.afternow, "$1", relative)
		end			
		return relative
	end
 
 	-- precision is decades, centuries and millennia
	local era
	if precision == 6 then era = mw.ustring.gsub(i18n.datetime[6], "$1", tostring(math.floor((math.abs(year) - 1) / 1000) + 1)) end
	if precision == 7 then era = mw.ustring.gsub(i18n.datetime[7], "$1", tostring(math.floor((math.abs(year) - 1) / 100) + 1)) end
	if precision == 8 then era = mw.ustring.gsub(i18n.datetime[8], "$1", tostring(math.floor(math.abs(year) / 10) * 10)) end
	if era then
		if year < 0 then era = mw.ustring.gsub(mw.ustring.gsub(i18n.datetime.bc, '"', ""), "$1", era)
		elseif year > 0 then era = mw.ustring.gsub(mw.ustring.gsub(i18n.datetime.ad, '"', ""), "$1", era) end
		return era
	end
 
	-- precision is years or less
	if precision >= 9 then
		--[[ the following code replaces the UTC suffix with the given negated timezone to convert the global time to the given local time
		timezone = tonumber(timezone)
		if timezone and timezone ~= 0 then
			timezone = -timezone
			timezone = string.format("%.2d%.2d", timezone / 60, timezone % 60)
			if timezone[1] ~= '-' then timezone = "+" .. timezone end
			date = mw.text.trim(date, "Z") .. " " .. timezone
		end
		]]--
 
		local formatstr = i18n.datetime[precision]
		if year == 0 then formatstr = mw.ustring.gsub(formatstr, i18n.datetime[9], "")
		elseif year < 0 then
			-- Mediawiki formatDate doesn't support negative years
			date = mw.ustring.sub(date, 2)
			formatstr = mw.ustring.gsub(formatstr, i18n.datetime[9], mw.ustring.gsub(i18n.datetime.bc, "$1", i18n.datetime[9]))
		elseif year > 0 and i18n.datetime.ad ~= "$1" then
			formatstr = mw.ustring.gsub(formatstr, i18n.datetime[9], mw.ustring.gsub(i18n.datetime.ad, "$1", i18n.datetime[9]))
		end
		return mw.language.new(wiki.langcode):formatDate(formatstr, date)
	end
end

local function printDatavalueEntity(data, parameter)
	-- data fields: entity-type [string], numeric-id [int, Wikidata id]
	local id = "Q" .. data["numeric-id"]
	if parameter then
		if parameter == "link" then
			return "[[" .. (mw.wikibase.sitelink(id) or (":d:" .. id))  .. "|" ..  (mw.wikibase.label(id) or id)  .. "]]"
		else
			return data[parameter]
		end
	else
		if data["entity-type"] == "item" then return mw.wikibase.label("Q" .. data["numeric-id"]) or id else printError("unknown-entity-type") end
	end
end

local function printDatavalueTime(data, parameter)
	-- data fields: time [ISO 8601 time], timezone [int in minutes], before [int], after [int], precision [int], calendarmodel [wikidata URI]
	--   precision: 0 - billion years, 1 - hundred million years, ..., 6 - millennia, 7 - century, 8 - decade, 9 - year, 10 - month, 11 - day, 12 - hour, 13 - minute, 14 - second
	--   calendarmodel: e.g. http://www.wikidata.org/entity/Q1985727 for the proleptic Gregorian calendar or http://www.wikidata.org/wiki/Q11184 for the Julian calendar]
	if parameter then
		if parameter == "calendarmodel" then data.calendarmodel = mw.ustring.match(data.calendarmodel, "Q%d+") -- extract entity id from the calendar model URI
		elseif parameter == "time" then data.time = normalizeDate(data.time) end
		return data[parameter]
	else
		return formatDate(data.time, data.precision, data.timezone)
	end
end

local function printDatavalueMonolingualText(data, parameter)
	-- data fields: language [string], text [string]
	if parameter then
		return data[parameter]
	else
		local result = mw.ustring.gsub(mw.ustring.gsub(i18n.monolingualtext, "%%language", data["language"]), "%%text", data["text"])
		return result
	end
end

function findClaims(entity, property)
	if not property or not entity or not entity.claims then return end
 
	if mw.ustring.match(property, "^P%d+$") then
		-- if the property is given by an id (P..) access the claim list by this id
		return entity.claims[property]
	else
		-- otherwise, iterate over all properties, fetch their labels and compare this to the given property name
		for k, v in pairs(entity.claims) do
			if mw.wikibase.label(k) == property then return v end
		end
		return
	end
end

function getSnakValue(snak, parameter)
	if snak.snaktype == "value" then
		-- call the respective snak parser
		if snak.datavalue.type == "string" then return snak.datavalue.value
		elseif snak.datavalue.type == "globecoordinate" then return printDatavalueCoordinate(snak.datavalue.value, parameter)
		elseif snak.datavalue.type == "quantity" then return printDatavalueQuantity(snak.datavalue.value, parameter)
		elseif snak.datavalue.type == "time" then return printDatavalueTime(snak.datavalue.value, parameter)
		elseif snak.datavalue.type == "wikibase-entityid" then return printDatavalueEntity(snak.datavalue.value, parameter)
		elseif snak.datavalue.type == "monolingualtext" then return printDatavalueMonolingualText(snak.datavalue.value, parameter)
		end
	end
	return mw.wikibase.renderSnak(snak)
end
 
function getQualifierSnak(claim, qualifierId)
	-- a "snak" is Wikidata terminology for a typed key/value pair
	-- a claim consists of a main snak holding the main information of this claim,
	-- as well as a list of attribute snaks and a list of references snaks
	if qualifierId then
		-- search the attribute snak with the given qualifier as key
		if claim.qualifiers then
			local qualifier = claim.qualifiers[qualifierId]
			if qualifier then return qualifier[1] end
		end
		return nil, printError("qualifier-not-found")
	else
		-- otherwise return the main snak
		return claim.mainsnak
	end
end
 
function getValueOfClaim(claim, qualifierId, parameter)
	local error
	local snak
	snak, error = getQualifierSnak(claim, qualifierId)
	if snak then
		return getSnakValue(snak, parameter)
	else
		return nil, error
	end
end

function getReferences(frame, claim)
	local result = ""
	-- traverse through all references
	for ref in pairs(claim.references or {}) do
		local refparts
		-- traverse through all parts of the current reference
		for snakkey, snakval in orderedpairs(claim.references[ref].snaks or {}, claim.references[ref]["snaks-order"]) do
			if refparts then refparts = refparts .. ", " else refparts = "" end
			-- output the label of the property of the reference part, e.g. "imported from" for P143
			refparts = refparts .. tostring(mw.wikibase.label(snakkey)) .. ": " 
			-- output all values of this reference part, e.g. "German Wikipedia" and "English Wikipedia" if the referenced claim was imported from both sites
			for snakidx = 1, #snakval do
				if snakidx > 1 then refparts = refparts .. ", " end
				refparts = refparts .. getSnakValue(snakval[snakidx])
			end
		end
		if refparts then result = result .. frame:extensionTag("ref", refparts) end
	end
	return result
end

function p.claim(frame)
	local property = frame.args[1] or ""
	local id = frame.args["id"]	-- "id" must be nil, as access to other Wikidata objects is disabled in Mediawiki configuration
	local qualifierId = frame.args["qualifier"]
	local parameter = frame.args["parameter"]
	local list = frame.args["list"]
	local references = frame.args["references"]
	local showerrors = frame.args["showerrors"]
	local default = frame.args["default"]
	if default then showerrors = nil end
 
	-- get wikidata entity
	local entity = mw.wikibase.getEntityObject(id)
	if not entity then
		if showerrors then return printError("entity-not-found") else return default end
	end
	-- fetch the first claim of satisfying the given property
	local claims = findClaims(entity, property)
	if not claims or not claims[1] then
		if showerrors then return printError("property-not-found") else return default end
	end
 
	-- get initial sort indices
	local sortindices = {}
	for idx in pairs(claims) do
		sortindices[#sortindices + 1] = idx
	end
	-- sort by claim rank
	local comparator = function(a, b)
		local rankmap = { deprecated = 2, normal = 1, preferred = 0 }
		local ranka = rankmap[claims[a].rank or "normal"] ..  string.format("%08d", a)
		local rankb = rankmap[claims[b].rank or "normal"] ..  string.format("%08d", b)
		return ranka < rankb
 	end
	table.sort(sortindices, comparator)
 
	local result
	local error
	if list then
		local value
		-- iterate over all elements and return their value (if existing)
		result = {}
		for idx in pairs(claims) do
			local claim = claims[sortindices[idx]]
			value, error =  getValueOfClaim(claim, qualifierId, parameter)
			if not value and showerrors then value = error end
			if value and references then value = value .. getReferences(frame, claim) end
			result[#result + 1] = value
		end
		result = table.concat(result, list)
	else
		-- return first element	
		local claim = claims[sortindices[1]]
		result, error = getValueOfClaim(claim, qualifierId, parameter)
		if result and references then result = result .. getReferences(frame, claim) end
	end
 
	if result then return result else
		if showerrors then return error else return default end
	end
end

return p