Module:Wikidata

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search
Lua
CodeDiscussionEditHistoryLinksLink count Subpages:DocumentationTestsResultsSandboxLive code All modules

Module permitting basic data retrieval from Wikidata items, based on mw:Extension:Wikibase Client/Lua

Functions callable from Lua

p.getClaims returns claims in a particular item that match a particular query

  • item (required): its Qid
  • property (required) property that the claims should have
  • qualifier = qualifiers that the statement should have
  • withrank = rank of the statement ; 'preferred', 'normal', 'deprecated' or 'valid' (ie. normal and preferred). By default = preferred.
  • sourceproperty = this property should be used in the source
  • withsource = the source that should be provided in the statement (if sourceproperty is not provided, the property used is stated in (P248))
  • excludespecial = set to true if you do not want to get "novalue" and "somevalue".
  • numval = if you want to set a maximum number of values to be returned
  • sorttype = set to "chronological" to get the statements in chronological order using the start time (P580), end time (P582) and point in time (P585) qualifiers. Set to "inverted" for chronological order. From a Lua module, you can also define your own sorting criteria.
  • showsource = set to "true" if you want the source of the statement to be displayed.

p.formatStatements(args): returns a string containing the statements given in the table args. Same keys as getClaims, plus formatting arguments:

  • lang (required) for the desired language
  • displayformat = the format in which the args should be returned. For example, for a string-type property displayformat = "weblink" returns a formatted weblink.
  • conjtype = the conjunction separating the statements. For example, conj = '<br />' will make a new line between each statement.
  • showqualifiers = the qualifiers that should be shown along with the mainsnak value

p.getLabel get the label of an entity

  • entity = entity ID with its Q or P
  • lang

Functions callable from wikitext

  • p.formatStatementsE same as p._formatStatements, except that "lang" is not required. It is most conveniently used from {{Data}} that takes exactly the same arguments.

Examples

Code Render Comment
{{#invoke:Wikidata|pageId}} Q12069631 return wikidata q-code for pages connected to wikidata
{{#invoke:Wikidata|getLabel|entity=Q42|link=wikipedia}} Douglas Adams link to wikipedia
{{#invoke:Wikidata|getLabel|entity=Q42|link=wikipedia|format=lc}} Douglas Adams link to wikipedia but shown with lower case
{{#invoke:Wikidata|getLabel|entity=Q42|link=-}} Douglas Adams no links
{{#invoke:Wikidata|getLabel|entity=Q42|link=wikipedia|lang=ja}} ダグラス・アダムズ
{{#invoke:Wikidata|getLabel|entity=Q42|link=wikidata}} Douglas Adams link to wikidata
{{#invoke:Wikidata|getLabel|entity=Q42|link=wikidata|lang=ja}} ダグラス・アダムズ
{{#invoke:Wikidata|formatStatementsE|item=Q42|property=p31}} human
{{#invoke:Wikidata|formatStatementsE|item=Q42|property=p31|link=-}} human
{{#invoke:Wikidata|formatStatementsE|item=Q42|property=p31|lang=ja}} ヒト
{{#invoke:Wikidata|formatStatementsE|item=Q42|property=p569}}
{{#invoke:Wikidata|formatStatementsE|item=Q42|property=p569|lang=ja}}
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186}} oil paint and poplar wood
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|link=-}} oil paint and poplar wood
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|displayformat=raw}} Q296955 and Q291034
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|lang=ja}} 油絵具およびポプラ板
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|rank=valid}} oil paint, poplar wood and wood rank = "valid" accepts both "preferred" and "normal" values
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|showqualifiers=p518}} oil paint and poplar wood (painting surface) shows the value of the p518 qualifier (if any) in addition to the main value
{{#invoke:Wikidata|formatStatementsE|item=Q83259|property=p669|showqualifiers=p670|delimiter=&#32;}} rue d'Ulm 45 shows the value of the P670 qualifier (if any) in addition to the main value separated by a space
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|qualifier=p518}} should only display values that have a p518 qualifier
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=p186|numval=1}} oil paint returns the number of values numval (priority to those with "rank= preferred", if there are not enough of them, also accepts "rank = normal")
{{#invoke:Wikidata|formatStatementsE|item=Q12418|property=P276|sourceproperty=P854}} Louvre Museum
{{#invoke:Wikidata|formatStatementsE|item=Q11879536|property=P460|withsource=Q1645493}} Lisa del Giocondo
{{#invoke:Wikidata|formatStatementsE|item=Q153|property=P231|showsource=true}} 64-17-5[1][2]
{{#invoke:Wikidata|formatStatementsE|item=Q205309|property=P793|sorttype=inverted}} closure, demolition, renovation, renovation, first match and construction
{{#invoke:Wikidata|formatStatementsE|property=P625|item=Q90|displayformat=latitude}} 48.856944444444 Latitude of Paris

Code

  1. ChEBI (title not provided in Wikidata), European Bioinformatics Institute, [www.ebi.ac.uk/chebi read online], Creative Commons Attribution 3.0 Unported
  2. Food and Drug Administration (eds.), Unique Ingredient Identifier
--script that retrieves basic data stored in Wikidata, for the datamodel, see https://www.mediawiki.org/wiki/Extension:Wikibase_Client/Lua

local p = {}

local linguistic = require('Module:Linguistic')
local selectClaims = require('Module:Wikidata/GetClaims')
local tools = require('Module:Wikidata/Tools')
local entities = require('Module:Wikidata/FormatEntity')
local dates = require('Module:Wikidata/Dates')
local weblink = require('Module:Weblink')

local langSwitch = require('Module:LangSwitch')._langSwitch
local fb = require('Module:Fallback')
local i18nmessages = mw.loadData('Module:i18n/wikidata')

-- Wiki-specific parameters
local defaultlang = mw.getCurrentFrame():callParserFunction('Int', 'Lang')
local defaultlink = 'wikidata'

local function i18n(str, lang)
	local message = i18nmessages[str]
	if type(message) == 'string' then
		return message
	end
	return langSwitch(message, lang or defaultlang)
end

local function formatError(key, text)
	return error(i18n(key) .. ' ' .. (text or ''), 2)
end

p.getClaims = selectClaims.getClaims

local function removeBlanks(args)
	for i = #args, 1, -1 do
		if (args[i] == '') or (args[i] == '-') then
			table.remove(args, i)
		end
	end
	return args
end

local function formatTheUnknown() -- See if we can grant/adapt the usage of 'unknown'
	return i18n('somevalue')
end

local function getQualifiers(statement, qualifs, params)
	if not statement.qualifiers then
		return nil
	end
	if type(qualifs) == 'string' then
		qualifs = mw.text.split(qualifs, ',')
	end
	local vals = {}
	for i, j in pairs(qualifs) do
		j = string.upper(j)
		if statement.qualifiers[j] then
			local inserted = false
			if statement.qualifiers[j][1].datatype == 'monolingualtext' then
				local in_preferred_lang
				for _, language in pairs(fb.fblist(params.lang or defaultlang)) do
					for _, snak in pairs(statement.qualifiers[j]) do
						if isInLanguage(snak, language) then
							in_preferred_lang = snak
							break
						end
					end
					if in_preferred_lang then
						break
					end
				end
				if in_preferred_lang then
					table.insert(vals, in_preferred_lang)
					inserted = true
				end
			end
			if not inserted then
				for _, snak in pairs(statement.qualifiers[j]) do
					table.insert(vals, snak)
				end
			end
		end
	end
	if #vals == 0 then
		return nil
	end
	return vals
end

function p.formatSnak(snak, params)
	-- special values: 'novalue' and 'somevalue'
	local datatype = snak.datatype
	if datatype == 'somevalue' then
		return formatTheUnknown()
	elseif datatype == 'novalue' then
		return i18n('novalue') --todo
	end
	local value = snak.datavalue.value
	-- user-defined displayformat
	local displayformat = params.displayformat
	if type(displayformat) == 'function' then
		return displayformat(snak, params)
 	end
	if datatype == 'wikibase-item' or datatype == 'wikibase-property' then
		return entities.formatEntity(tools.getId(snak), params)
	elseif datatype == 'url' then
		return weblink.makelink(value, params.text, params.displayformat)
	elseif datatype == 'math' then
		return mw.getCurrentFrame():extensionTag('math', value)
	elseif datatype == 'string' or datatype == 'external-id' or datatype == 'commonsMedia' then
		if params.urlpattern then
			local urlpattern = params.urlpattern
			if type(urlpattern) == 'function' then
				urlpattern = urlpattern(value)
			end
			value = '[' .. mw.ustring.gsub(urlpattern, '$1', value) .. ' ' .. (params.text or value) .. ']'
		end
		return value
	elseif datatype == 'time' then -- format example: +00000001809-02-12T00:00:00Z
		return dates.formatTimeSnak(snak, params) 
	elseif datatype == 'globe-coordinate' then
		-- default return a table with latitude, longitude, precision and globe that can be used by another module
		if displayformat == 'latitude' then
			return value.latitude
		elseif displayformat == 'longitude' then
			return value.longitude
		elseif displayformat == 'qualifier' then
			local coord = require('Module:Coordinates')
			value.globe = require('Module:Wikidata/Globes')[value.globe]
			value.precision = nil
			return coord._coord(value)
		else
			value.globe = require('Module:Wikidata/Globes')[value.globe] -- get English name for geohack
			return value
		end
	elseif datatype == 'quantity' then
		-- TODO: handle precision parameters
		if displayformat == 'raw' then
			return tonumber(value.amount)
		else
			local formatNum = require('Module:Formatnum')
			local number = formatNum.formatNum(value.amount)
			local unit = mw.ustring.match(value.unit, '(Q%d+)')
			if unit then
				number = number .. '&nbsp;' .. entities.formatEntity(unit, params)
			end
			return number
		end
	elseif datatype == 'monolingualtext' then
		if displayformat == 'raw' then
			-- Don't use HTML
			local byte, char = string.byte, mw.ustring.char
			local lang = value.language:lower()
			local tag = {}
			table.insert(tag, char(0x2068)) -- U+2068: First Strong Isolate (FSI)
			table.insert(tag, char(0xE0001)) -- U+E0001: Language tag
			for i = 1, #lang do
				local b = byte(lang, i)
				if b >= 0x61 and b <= 0x7A or b == 0x2D or b == 0x5F then -- 'a'..'z', '-', '_'
					table.insert(tag, char(0xE0000 + b)) -- U+E0020..U+E007E: Tag characters (remaps ASCII only)
				end
			end
			table.insert(tag, value.text)
			table.insert(tag, char(0xE007F)) -- U+E007F: Cancel Tag
			table.insert(tag, char(0x2069)) -- U+2069: Pop Directional Isolate (PDI)
			return table.concat(tag)
		else
			return '<bdi lang="' .. value.language .. '">' .. value.text .. '</bdi>'
		end
	else
		return formatError('unknown-datavalue-type', datatype )
	end
end

function p.formatStatementQualifiers(statement, qualifs, params)
	if not params then params = {} end
	local qualiftable = getQualifiers(statement, qualifs, params)
	if not qualiftable then
		return nil
	end
	for i, j in pairs(qualiftable) do
		local params = params
		if j.datatype == 'globe-coordinate' then
			params.displayformat = 'qualifier'
		end
		qualiftable[i] = p.formatSnak(j, params)
	end
	return linguistic.conj(qualiftable, params.lang or defaultlang)
end

function p.formatStatement(statement, args)
	if not statement.type or statement.type ~= 'statement' then
		return formatError('unknown-claim-type', statement.type)
	end
	if not args then args = {} end
	local lang = args.lang or defaultlang
	local str = p.formatSnak(statement.mainsnak, args)

	local qualifs = args.showqualifiers
	if qualifs then
		local qualifStr = p.formatStatementQualifiers(statement, qualifs, args)
		if qualifStr then
			if args.delimiter then
				str = str .. args.delimiter .. qualifStr
			else
				-- str = str .. linguistic.inparentheses(qualifStr, lang)
				str = str .. ' (' .. qualifStr .. ')'
			end
		end
	end

	if args.showdate then -- when `showdate` and `p.chronosort` are both set, date retrieval is performed twice
		local params
		local date = dates.getFormattedDate(statement, params)
		if date then
			-- str = str .. ' <small>' .. linguistic.inparentheses(date, lang) .. '</small>'
			str = str .. ' <small>(' .. date ..')</small>'
		end
	end

	if args.showsource and statement.references then
		local cite = require('Module:Cite')
		local frame = mw.getCurrentFrame()
		local sourcestring = ''
		for i, ref in pairs(statement.references) do
			if ref.snaks.P248 then
				for j, source in pairs(ref.snaks.P248) do
					if not tools.isSpecial(source) then
						local page
						if ref.snaks.P304 and not tools.isSpecial(ref.snaks.P304[1]) then
							page = ref.snaks.P304[1].datavalue.value
						end
						local s = cite.citeitem('Q' .. source.datavalue.value['numeric-id'], lang, page)
						s = frame:extensionTag('ref', s)
						sourcestring = sourcestring .. s
					end
				end
			elseif ref.snaks.P854 and not tools.isSpecial(ref.snaks.P854[1]) then
				s = frame:extensionTag('ref', p.getDatavalue(ref.snaks.P854[1]))
				sourcestring = sourcestring .. s
			end
		end
		str = str .. sourcestring
	end
	return str
end

function p._getDescription(entity, lang)
	if not entity then
		return i18n('no description')
	end
	if type(entity) == 'string' and (not lang or lang == defaultlang) then
		return mw.wikibase.description(entity)
	end
	entity = mw.wikibase.getEntity(entity)
	local descriptions = entity.descriptions
	if not descriptions then
		return i18n('no description')
	end
	local langlist = fb.fblist(lang or defaultlang) -- list of fallback languages if no label in the desired language
	for i, lg in pairs(langlist) do
		if descriptions[lg] then
			return descriptions[lg].value
		end
	end
	return i18n('no description')
end

-- == FUNCTIONS HANDLING SEVERAL STATEMENTS ===

function p.stringTable(args) -- find the relevant claims, and formats them as a list of strings
	-- Get claims
	local claims = p.getClaims(args)
	if not claims then
		return nil
	end
	-- Define the formatter function
	local formatterFun = p.formatStatement
	if args.type == 'date' then
		formatterFun = dates.getFormattedDate
	elseif args.type == 'qualifiers' then
		formatterFun = formatStatementQualifiers
	end
	-- Format claims
	for i = #claims, 1, -1 do
		claims[i] = formatterFun(claims[i], args)
		if not claims[i] then
			table.remove(claims, i)
		end
	end
	return claims
end

function p.formatStatements(args)--Format statements and concat them cleanly
	-- If a value is already set, use it
	if args.value == '-' then
		return nil
	end
	if args.value then
		return args.value
	end
	-- Obsolete parameters 
	if args.item then
		args.entity = args.item
	end
	local values = p.stringTable(args) -- gets statements, and format each of them
	if not values then
		return nil
	end
	return linguistic.conj(values, args.lang or defaultlang, args.conjtype) -- concatenate them
end

-- == FRAME FUNCTIONS ==

function p.getDescription(frame) -- simple for simple templates like {{Q|}}}
	local entity = frame.args.entity
	local lang = frame.args.lang
	return p._getDescription(entity, lang)
end

function p.formatStatementsE(frame)
	local args = {}
	if frame == mw.getCurrentFrame() then
		args = frame:getParent().args -- paramètres du modèle appelant (est-ce vraiment une bonne idée ?)
		for k, v in pairs(frame.args) do
			args[k] = v
		end
	else
		args = frame
	end
	args = removeBlanks(args)
	return p.formatStatements(args)
end

function p._getLabel(entity, args) 
	return entities.formatEntity(entity, args)
end
function p.getLabel(frame) -- simple for simple templates like {{Q|}}}
	local args = removeBlanks(frame.args)
	local entity = args.entity or args[1]
	if string.sub(entity, 1, 10) == 'Property:P' then
		entity = string.sub(entity, 10)
	elseif string.sub(entity, 1, 1) ~= 'P' and string.sub(entity, 1, 1) ~= 'Q'
	or not tonumber(string.sub(entity, 2)) then
		return i18n('invalid-id')
	end
	return entities.formatEntity(entity, args)
end

function p._wikidataDate(prop, item, params)
	local claims = p.getClaims{entity = item, property = prop}
	if not claims then
		return nil
	end
	params = params or {}
	local vals = {}
	for i, j in pairs(claims) do
		local v = dates.getFormattedDate(j, params)
		if v then
			table.insert(vals, v)
		end
	end
	return linguistic.conj(vals, params.conjtype or 'or')
end

function p.wikidataDate(frame)
	 p._wikidataDate(frame.args[property], frame.args[params], frame.args)
end

function p._main_date(entity)	
	-- First try with P580/P582
	local startpoint = p._wikidataDate('P580', entity)
	local endpoint = p._wikidataDate('P582', entity)
	if startpoint or endpoint then
		return (startpoint or '') .. '&#x202F;–&#x2009;' .. (endpoint or '')
	end
	-- Else use P585
	return p._wikidataDate('P585', entity)
end

function p.main_date(frame)
	return p._main_date(frame.args[1])
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()
	return entity and entity.id
end

return p