Modul:Wikidata utilities

Die Dokumentation für dieses Modul kann unter Modul:Wikidata utilities/doc erstellt werden

-- documentation
local FastWikidata = {
	suite  = 'FastWikidata',
	serial = '2020-05-13',
	item   = 65439025
}

-- module variable
local fw = {}

-- table storing property ids used
local catTable = {
	P0 = ''
}

function fw.getEntity( id )
	local wrongQualifier = false
	local entity = nil
	
	if not id or id == '' then
		return '', entity, wrongQualifier
	end
	if mw.wikibase.isValidEntityId( id ) then
		-- expensive function call
		-- redirect ids marked false, too
		entity = mw.wikibase.getEntity( id )
	end
	if not entity then
		id = ''
		wrongQualifier = true
	end

	return id, entity, wrongQualifier
end

function fw.getEntityId( id )
	local wrongQualifier = false
	local entity = nil
	
	if not id or id == '' then
		return '', entity, wrongQualifier
	end
	if mw.wikibase.isValidEntityId( id ) and mw.wikibase.entityExists( id ) then
		-- expensive function call
		-- redirect ids marked false, too
		entity = id
	end
	if not entity then
		id = ''
		wrongQualifier = true
	end

	return id, entity, wrongQualifier
end

function fw.getLabel( entity, lang )
	local isString = type( entity ) == 'string'
	if not entity or ( isString and entity == '' ) then
		return nil
	end
	if isString then -- entity is id
		if lang and lang ~= '' then
			return mw.wikibase.getLabelByLang( entity, lang )
		else
			return mw.wikibase.getLabel( entity )
		end
	else -- entity is table
		if lang and lang ~= '' then
			return entity:getLabel( lang )
		else
			return entity:getLabel()
		end
	end
end

function fw.getSitelink( entity, globalSiteId )
	local isString = type( entity ) == 'string'
	if not entity or ( isString and entity == '' ) then
		return nil
	end
	if isString then -- entity is id
		if globalSiteId and globalSiteId ~= '' then
			return mw.wikibase.getSitelink( entity, globalSiteId )
		else
			return mw.wikibase.getSitelink( entity )
		end
	else -- entity is table
		if globalSiteId and globalSiteId ~= '' then
			return entity:getSitelink( globalSiteId )
		else
			return entity:getSitelink( )
		end
	end
end

local function getFirstValue( statements )
	if #statements == 0 then
		return nil
	end

	for i = 1, #statements, 1 do
		if statements[ i ].mainsnak.snaktype == 'value' then
			return statements[ i ].mainsnak.datavalue.value
		end
	end

	return nil
end

local function getNValues( statements, count )
	local ar = {}
	if count > #statements then
		count = #statements
	end
	if #statements == 0 or count <= 0 then
		return ar
	end

	local i = 0
	repeat
		i = i + 1
		if statements[ i ].mainsnak.snaktype == 'value' then
			table.insert( ar, statements[ i ].mainsnak.datavalue.value )
		end
	until i >= #statements or #ar >= count

	return ar
end

function fw.getBestStatements( entity, p )
	if type( entity ) == 'string' then
		return mw.wikibase.getBestStatements( entity, p )
	else
		return entity:getBestStatements( p )
	end
end

function fw.getStatements( entity, p, count )
	local ar = {}
	if not entity or entity == '' then
		return ar
	end

	local statements = fw.getBestStatements( entity, p )

	count = count or #statements
	if count > #statements then
		count = #statements
	end
	if #statements == 0 or count <= 0 then
		return ar
	end

	local i = 0
	repeat
		i = i + 1
		if statements[ i ].mainsnak.snaktype == 'value' then
			table.insert( ar, statements[ i ] )
		end
	until i >= #statements or #ar >= count

	return ar
end

function fw.getValue( entity, p )
	local value = ''
	if entity and entity ~= '' and p and p ~= '' then
		value = getFirstValue( fw.getBestStatements( entity, p ) )
		if value then
			catTable[ p ] = ''
		else
			value = ''
		end
	end
	return value
end

function fw.getId( entity, p )
	local value = ''
	if entity and entity ~= '' and p and p ~= '' then
		value = getFirstValue( fw.getBestStatements( entity, p ) )
		if value then
			catTable[ p ] = ''
			value = value.id
		else
			value = ''
		end
	end
	return value
end

function fw.getValues( entity, p, count )
	local values = ''
	if entity and entity ~= '' and p and p ~= '' then
		local statements = fw.getBestStatements( entity, p )
		values = getNValues( statements, count or #statements )
		if #values > 0 then
			catTable[ p ] = ''
		end
	end
	return values
end

function fw.getValuesByLang( entity, p, count, lang )
	local ar = ''
	if entity and entity ~= '' and p and p ~= '' then
		local statements = fw.getBestStatements( entity, p )
		ar = {}
		count = count or #statements
		if #statements > 0 and count > 0 then
			local i = 0
			local value
			repeat
				i = i + 1
				if statements[ i ].mainsnak.snaktype == 'value' then
					value = statements[ i ].mainsnak.datavalue.value
					if value.language and lang == value.language then
						table.insert( ar, statements[ i ].mainsnak.datavalue.value.text )
					end
				end
			until i >= #statements or #ar >= count
		end
		if #ar > 0 then
			catTable[ p ] = ''
		end
	end
	return ar
end

function fw.getValuesWithQualifierIds( entity, p, qualifierP, defaultId )
	local result = {}
	if entity and entity ~= '' and p and p ~= '' and qualifierP and qualifierP ~= '' then
		local statements = fw.getStatements( entity, p, nil )
		if #statements > 0 then
			-- defaultId is used if a qualifier is missing
			if not defaultId or defaultId == '' or type( defaultId ) ~= 'string' then
				defaultId = 'unknown'
			end

			catTable[ p ] = ''
			local id, value
			for i = 1, #statements, 1 do
				value = statements[ i ].mainsnak.datavalue.value
				id = defaultId
				if statements[ i ].qualifiers and statements[ i ].qualifiers[ qualifierP ]
					and ( #statements[ i ].qualifiers[ qualifierP ] > 0 ) then
					for j = 1, #statements[ i ].qualifiers[ qualifierP ], 1 do
						if statements[ i ].qualifiers[ qualifierP ][ j ].snaktype == 'value' then
							id = statements[ i ].qualifiers[ qualifierP ][ j ].datavalue.value.id
							break
						end
					end
				end
				result[ id ] = value
			end
		end
	end
	return result
end

-- get values array for monolingual text
function fw.getValuesWithLanguages( entity, p )
	local result = {}
	if entity and entity ~= '' and p and p ~= '' then
		local statements = fw.getStatements( entity, p, nil )
		local hyphen, lng, value
		if #statements > 0 and statements[ 1 ].mainsnak.datatype == 'monolingualtext' then
			catTable[ p ] = ''
			for i = 1, #statements, 1 do
				value = statements[ i ].mainsnak.datavalue.value
				lng = value.language
				hyphen = lng:find( '-' )
				if hyphen then
					lng = lng:sub( 1, hyphen - 1 )
				end
				if not result[ lng ] then
					result[ lng ] = value.text
				end
			end
		end
	end
	return result
end

local function getValueFromDatavalue( datavalue )
	local v = datavalue.value
	local t = datavalue.type
	if type( v ) == 'table' then
		-- items which can be reduced to a string
		if t == 'wikibase-entityid' then
			v = v.id
		elseif t == 'time' then
			v = v.time
		end
	end
	return v, t
end

-- The following function is an experimental one, not for extensive use
function fw.getValuesWithQualifiers( entity, p, qualifiers, count )
	local result = {}
	local statements = fw.getStatements( entity, p, count )
	if #statements == 0 then
		return result
	end

	if qualifiers and type( qualifiers ) == 'string' then
		qualifiers = { qualifiers }
	end

	local array
	for i = 1, #statements, 1 do
		array = { value = statements[ i ].mainsnak.datavalue.value,
			[ 'value-type' ] = statements[ i ].mainsnak.datavalue.type }
		if statements[ i ].qualifiers then
			if not qualifiers then -- all qualifier properties
				for key, value in pairs( statements[ i ].qualifiers ) do
					if #value > 0 then
						for j = 1, #value, 1 do
							if value[ j ].snaktype == 'value' then
								array[ key ], array[ key .. '-type' ] =
									getValueFromDatavalue( value[ j ].datavalue )
								break
							end
						end
					end
				end
			else -- table of selected qualifier properties
				for key, value in pairs( qualifiers ) do
					if statements[ i ].qualifiers[ value ] and
						#statements[ i ].qualifiers[ value ] > 0 then
						for j = 1, #statements[ i ].qualifiers[ value ], 1 do
							if statements[ i ].qualifiers[ value ][ j ].snaktype == 'value' then
								array[ value ], array[ value .. '-type' ] =
									getValueFromgetValueFromDatavalue( statements[ i ].qualifiers[ value ][ j ].datavalue )
								break
							end
						end
					end
				end
			end
		end
		table.insert( result, array )
	end
	return result
end

function fw.typeSearch( p31, list, limit )
	-- p31: array of Wikidata values
	-- list: indexed array of q id - types relations
	-- limit: maximum levels to analyse
	if not list or not p31 or #p31 == 0 then
		return 'error'
	end

	local function compareIds( ar )
		local t
		for i = 1, #ar, 1 do
			t = list[ ar[ i ].id ]
			if t then
				return t
			end
		end
		return nil
	end

	local aType = compareIds( p31 ) -- check p31 ids first, maybe step 2 is not nessary
	if aType then
		return aType
	end

	-- now functions becomes expensive because of multiple fw.getValues calls
	local id, ids
	for i = 1, #p31, 1 do -- step 2: analyse P279 chains of first ids
		id = p31[ i ].id -- start id
		local j = 0
		repeat
			ids = fw.getValues( id, 'P279', nil )
			if #ids > 0 then
				id = ids[ 1 ].id
				aType = compareIds( ids )
				if aType then
					return aType
				end
			end
			j = j + 1
		until j >= limit or #ids == 0
	end

	return 'error'
end

function fw.getCategories( formatStr )
	local result = ''

	if not formatStr or formatStr == '' then
		formatStr = '[[Category:%s]]'
	end

	catTable.P0 = nil
	for key, value in pairs( catTable ) do
		result = result .. string.format( formatStr, key )
	end
	return result
end

return fw