Editing Module:Sandbox/RexxS/GetDateValue

-- vim: set noexpandtab ft=lua ts=4 sw=4:
require('strict')

local p = {}
local debug = false

-- 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.",
		["unknown-datetime-format"] = "Unknown datetime format.",
		["local-article-not-found"] = "Article is not yet available in this wiki"
    },
    ["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 millennium",	 	-- 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
		-- the following are for function getDateValue() and getQualifierDateValue()
		["default-format"] = "dmy", -- default value of the #3 (getDateValue) or
								    -- #4 (getQualifierDateValue) argument
		["default-addon"] = "BC",   -- default value of the #4 (getDateValue) or
									-- #5 (getQualifierDateValue) argument
		["prefix-addon"] = false,   -- set to true for languages put "BC" in front of the
									-- datetime string; or the addon will be suffixed
		["addon-sep"] = " ",		-- separator between datetime string and addon (or inverse)
		["format"] =				-- options of the 3rd argument
		{
			["mdy"] = "F j, Y",
			["my"] = "F Y",
			["y"] = "Y",
			["dmy"] = "j F Y",
			["ymd"] = "Y-m-d",
			["ym"] = "Y-m"
		}
	},
	["monolingualtext"] = '<span lang="%language">%text</span>',
	["warnDump"] = "[[Category:Called function 'Dump' from module Wikidata]]"
}

-- Credit to http://stackoverflow.com/a/1283608/2644759
-- cc-by-sa 3.0
local function tableMerge(t1, t2)
	for k,v in pairs(t2) do
		if type(v) == "table" then
			if type(t1[k] or false) == "table" then
				tableMerge(t1[k] or {}, t2[k] or {})
			else
				t1[k] = v
			end
		else
			t1[k] = v
    	end
	end
	return t1
end

local function loadI18n()
	local exist, res = pcall(require, "Module:Wikidata-i18n")
	if exist then
		tableMerge(i18n, res.i18n)
	end
end

loadI18n()

local function printError(code)
	return '<span class="error">' .. (i18n.errors[code] or code) .. '</span>'
end

if debug then
	function p.inspectI18n(frame)
		local val = i18n
		for _, key in pairs(frame.args) do
			key = mw.text.trim(key)
			val = val[key]
		end
		return val
	end
end

local function parseDateFull(timestamp, precision, date_format, date_addon)
	local prefix_addon = i18n["datetime"]["prefix-addon"]
	local addon_sep = i18n["datetime"]["addon-sep"]
	local addon = ""

	-- check for negative date
	if string.sub(timestamp, 1, 1) == '-' then
		timestamp = '+' .. string.sub(timestamp, 2)
		addon = date_addon
	end
	
	-- get the next four characters after the + (should be the year now in all cases)
	-- ok, so this is dirty, but let's get it working first
	local intyear = tonumber(string.sub(timestamp, 2, 5))
	if intyear == 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(intyear) / factor)
		local relative = mw.ustring.gsub(i18n.datetime[precision], "$1", tostring(y2))
		if addon ~= "" then
			-- negative date
			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(intyear) - 1) / 1000) + 1)) end
	if precision == 7 then era = mw.ustring.gsub(i18n.datetime[7], "$1", tostring(math.floor((math.abs(intyear) - 1) / 100) + 1)) end
	if precision == 8 then era = mw.ustring.gsub(i18n.datetime[8], "$1", tostring(math.floor(math.abs(intyear) / 10) * 10)) end
	if era then
		if addon ~= "" then
			era = mw.ustring.gsub(mw.ustring.gsub(i18n.datetime.bc, '"', ""), "$1", era)
		else
			era = mw.ustring.gsub(mw.ustring.gsub(i18n.datetime.ad, '"', ""), "$1", era)
		end
		return era
	end
	
	local _date_format = i18n["datetime"]["format"][date_format]
	if _date_format ~= nil then
		 -- check for precision is year and override supplied date_format
 		if precision == 9 then
 			_date_format = i18n["datetime"][9]
 		end
		local year_suffix
		local tstr = ""
		local lang_obj = mw.language.new(wiki.langcode)
		local f_parts = mw.text.split(_date_format, 'Y', true)
		for idx, f_part in pairs(f_parts) do
			year_suffix = ''
			if string.match(f_part, "x[mijkot]$") then
				-- for non-Gregorian year
				f_part = f_part .. 'Y'
			elseif idx < #f_parts then
				-- supress leading zeros in year
				year_suffix = lang_obj:formatDate('Y', timestamp)
				year_suffix = string.gsub(year_suffix, '^0+', '', 1)
			end
			tstr = tstr .. lang_obj:formatDate(f_part, timestamp) .. year_suffix
		end
		local fdate
		if addon ~= "" and prefix_addon then
			fdate = addon .. addon_sep .. tstr
		elseif addon ~= "" then
			fdate = tstr .. addon_sep .. addon
		else
			fdate = tstr
		end
		
		return fdate
	else
		return printError("unknown-datetime-format")
	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

local function formatDate(date, precision, timezone)
	precision = precision or 11
	local 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 year
 	if precision == 9 then
 		return year
 	end
 
	-- precision is less than years
	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
	
	if data["entity-type"] == "item" then id = "Q" .. data["numeric-id"]
	elseif data["entity-type"] == "property" then id = "P" .. data["numeric-id"]
	else return printError("unknown-entity-type")
	end
	
	if parameter then
		if parameter == "link" then
			local linkTarget = mw.wikibase.sitelink(id)
			local linkName = mw.wikibase.label(id)
			if linkTarget then
				-- if there is a local Wikipedia article link to it using the label or the article title
				return "[[" .. linkTarget  .. "|" .. (linkName or linkTarget) .. "]]"
			else
				-- if there is no local Wikipedia article output the label or link to the Wikidata object to let the user input a proper label
				if linkName then return linkName else return "[[:d:" .. id .. "|" .. id .. "]]" end
			end
		else
			return data[parameter]
		end
	else
		return mw.wikibase.label(id) or id
	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

local 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
		property = mw.wikibase.resolvePropertyId(property)
		if not property then return end

		return entity.claims[property]
	end
end

local 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
 
local 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
 
local 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

local 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

-- look into entity object
function p.ViewSomething(frame)
	local f = (frame.args[1] or frame.args.id) and frame or frame:getParent()
	local id = f.args.id
	if id and (#id == 0) then
		id = nil
	end
	local data = mw.wikibase.getEntityObject(id)
	if not data then
		return nil
	end

	local i = 1
	while true do
		local index = f.args[i]
		if not index then
			if type(data) == "table" then
				return mw.text.jsonEncode(data, mw.text.JSON_PRESERVE_KEYS + mw.text.JSON_PRETTY)
			else
				return tostring(data)
			end
		end
		
		data = data[index] or data[tonumber(index)]
		if not data then
			return
		end
		
		i = i + 1
	end
end

-- getting sitelink of a given wiki
function p.getSiteLink(frame)
	local f = frame.args[1]
	local entity = mw.wikibase.getEntity()
	if not entity then
		return
	end
	local link = entity:getSitelink( f )
	if not link then
		return
	end
	return link
end

function p.Dump(frame)
	local f = (frame.args[1] or frame.args.id) and frame or frame:getParent()
	local data = mw.wikibase.getEntityObject(f.args.id)
	if not data then
		return i18n.warnDump
	end

	local i = 1
	while true do
		local index = f.args[i]
		if not index then
			return "<pre>"..mw.dumpObject(data).."</pre>".. i18n.warnDump
		end

		data = data[index] or data[tonumber(index)]
		if not data then
			return i18n.warnDump
		end

		i = i + 1
	end
end

-- This is used to get a date value for date_of_birth (P569), etc. which won't be linked
-- Dates and times are stored in ISO 8601 format (sort of).
-- At present the local formatDate(date, precision, timezone) function doesn't handle timezone
-- So I'll just supply "Z" in the call to formatDate below:
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 i18n["datetime"]["default-format"])
	local date_addon = mw.text.trim(frame.args[4] or i18n["datetime"]["default-addon"])
	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
				if v.mainsnak.datavalue.type == 'time' then
					local timestamp = v.mainsnak.datavalue.value.time
					local dateprecision = v.mainsnak.datavalue.value.precision
					out[#out + 1] = parseDateFull(timestamp, dateprecision, date_format, date_addon)
				end
			end
			return table.concat(out, ", ")
		else
			return ""
		end
	else
		return input_parm
	end
end


return p