Module:Pornbasedata/Places

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

local categorizeByPlace = true;

local WDS = require( 'Module:PornBaseDataSelectors' );
local Flags = require( 'Module:Pornbasedata/Flags' );
local p = {
	config = {
		hideSameLabels = false,
		hidePartOfLabels = false,
		hideUnitsForCapitals = true,
		reverseOrder = false,
		catAmbiguousGeoChains = '[[Category:Pages with ambiguous geo-chains]]',
		catLoopInGeoChains = '[[Category:Pages with loop geo-chains]]',
		catWikibaseError = '[[Category:Script error pages using PornBaseData]]'
	}
};

local function min( prev, next )
	if prev == nil then
		return next;
	elseif prev > next then
		return next;
	else
		return prev;
	end
end

local function max( prev, next )
	if prev == nil then
		return next;
	elseif prev < next then
		return next;
	else
		return prev;
	end
end

local function getTimeBoundariesFromProperty( context, propertyId )
	local dateClaims = WDS.filter( context.entity.claims, propertyId );
	if not dateClaims or #dateClaims == 0 then
		return nil;
	end

	-- only support exact date so far, but need improvment
	local left = nil;
	local right = nil;
	for _, claim in pairs( dateClaims ) do
		if not claim.mainsnak then return nil; end
		local boundaries = context.parseTimeBoundariesFromSnak( claim.mainsnak );
		if not boundaries then return nil; end
		left = min( left, boundaries[ 1 ] );
		right = max( right, boundaries[ 2 ] );
	end

	if not left or not right then return nil; end

	return { left, right };
end

local function getTimeBoundariesFromProperties( context, propertyIds )
	for _, propertyId in ipairs( propertyIds ) do
		local result = getTimeBoundariesFromProperty( context, propertyId );
		if result then
			return result;
		end
	end

	return nil;
end

local function getTimeBoundariesFromQualifiers( context, statement, qualifierId )
	-- only support exact date so far, but need improvment
	local left = nil;
	local right = nil;
	if statement.qualifiers and statement.qualifiers[qualifierId] then
		for _, qualifier in pairs( statement.qualifiers[qualifierId] ) do
			local boundaries = context.parseTimeBoundariesFromSnak( qualifier );
			if not boundaries then return nil; end
			left = min( left, boundaries[1] );
			right = max( right, boundaries[2] );
		end
	end

	if not left or not right then
		return nil;
	end

	return { left, right };
end

local function getParentsInBoundariesSnakImpl( context, entityId, boundaries, propertyIds, selectors )
	local results = {};

	if not propertyIds or #propertyIds == 0 then
		return results;
	end
	
	for _, propertyId in ipairs( propertyIds ) do
		if (not string.match( propertyId, '^P%d+$' )) then error('Incorrect propertyId: ' + propertyId); end
		local selector;
		if (selectors ~= nil) then
			selector = selectors[propertyId] or propertyId;
		else
			selector = propertyId;
		end

		local entityClaims = {};
		entityClaims[propertyId] = mw.wikibase.getAllStatements( entityId, propertyId );

		local filteredClaims = WDS.filter( entityClaims, selector .. '[rank:preferred, rank:normal]' );
		if filteredClaims then
			for _, claim in pairs( filteredClaims ) do
				if not boundaries or not propertyIds or #propertyIds == 0 then
					table.insert( results, claim.mainsnak );
				else
					local startBoundaries = getTimeBoundariesFromQualifiers( context, claim, 'P580' );
					local endBoundaries = getTimeBoundariesFromQualifiers( context, claim, 'P582' );
		
					if ( startBoundaries == nil or startBoundaries[2] <= boundaries[1] ) and
						( endBoundaries == nil or endBoundaries[1] >= boundaries[2] )
					then
						table.insert( results, claim.mainsnak );
					end 
				end
			end
		end

		if #results > 0 then
			break;
		end
	end

	return results;
end

local function getParentsInBoundariesSnak( context, entityId, boundaries )
	if not entityId then error('entityId must be specified'); end
	if type(entityId) ~= 'string' then error('entityId must be string'); end
	if not boundaries then error('boundaries must be specified'); end
	if type(boundaries) ~= 'table' then error('boundaries must be table'); end

	local results = getParentsInBoundariesSnakImpl( context, entityId, boundaries, {'P49'} ) -- located in
	if not results or #results == 0 then
		results = getParentsInBoundariesSnakImpl( context, entityId, boundaries, {'P40'} ) -- country
	end

	for r, result in pairs( results ) do
		if result.snaktype ~= 'value' then
			return nil;
		end
		local resultId = result.datavalue.value.id;
		if resultId == entityId then
			return nil;
		end
	end
	return results;
end

local unions = { 
	Q3127 = true, -- political union
	Q3128 = true, -- Экономический союз
	Q3129 = true -- региональная организация
}

local countries = {
	Q210 = true, -- страна
	Q3130 = true, -- государство
	Q3124 = true, -- историческое государство
	Q1041 = true -- суверенное государство
}

local function isSkipTopLevel( entity )
	local isCountry = false;
	local isUnion = false;
	if entity and
		entity.claims and
		entity.claims.P1
	then
		for c, claim in pairs( entity.claims.P1 ) do
			if claim and
				claim.mainsnak and
				claim.mainsnak.datavalue and
				claim.mainsnak.datavalue.value and
				claim.mainsnak.datavalue.value.id
			then
				local typeId = claim.mainsnak.datavalue.value.id;
				isCountry = isCountry or countries[ typeId ];
				isUnion = isUnion or unions[ typeId ];
			end
		end
	end
	return isUnion and not isCountry;
end

local function isPartOfNext( prevLabel, nextLabel )
	return ( mw.ustring.len( prevLabel ) > mw.ustring.len( nextLabel ) )
		and ( mw.ustring.sub( prevLabel, mw.ustring.len( prevLabel ) - mw.ustring.len( nextLabel ) + 1 ) == nextLabel );
end

--Property:P20, Property:P23, Property:P26
function p.formatPlaceWithQualifiers( context, options, statement )
	local property = mw.ustring.upper( options.property );

	local actualDateBoundariesProperties = nil;
	if property == 'P20' then actualDateBoundariesProperties = {'P9','P247'}; end
	if property == 'P23' then actualDateBoundariesProperties = {'P247','P9'}; end
	if property == 'P26' then actualDateBoundariesProperties = {'P247','P9'}; end
	if property == 'P273' then actualDateBoundariesProperties = {'P572'}; end

	local boundaries = nil;
	if actualDateBoundariesProperties ~= nil then
		boundaries = getTimeBoundariesFromProperties( context, actualDateBoundariesProperties );
		if (boundaries == nil) and (property == 'P273') then
			boundaries = {os.time() * 1000, os.time() * 1000};
		end
	end

	local entriesToLookupCategory = {};

	local circumstances = context.getSourcingCircumstances( statement );
	local result = '';
	local baseResult = context.formatSnak( options, statement.mainsnak, circumstances );
	if not baseResult then
		return nil;
	end
	
	insertFromSnak( statement.mainsnak, entriesToLookupCategory )

	local hasAdditionalQualifiers = false;
	if statement.qualifiers then
		--parent divisions
		if statement.qualifiers.P49 then
			for i, qualifier in ipairs( statement.qualifiers.P49 ) do
				if qualifier.datavalue then
					local parentOptions = context.cloneOptions( options );
					local qualifierEntityId = qualifier.datavalue.value.id;
					parentOptions['text'] = getLabel( context, qualifierEntityId, boundaries );
					local link = context.formatSnak( parentOptions, qualifier );
	
					if p.config.reverseOrder then
						result = link .. ', ' .. result;
					else
						result = result .. ', ' .. link;
					end
					insertFromSnak( qualifier, entriesToLookupCategory )
					hasAdditionalQualifiers = true;
				end
			end
		end

		--country
		if statement.qualifiers.P40 then
			for i, qualifier in ipairs( statement.qualifiers.P40 ) do
				if qualifier.datavalue then
					local parentOptions = context.cloneOptions( options );
					local qualifierEntityId = qualifier.datavalue.value.id;
					parentOptions[ 'text' ] = getLabel( context, qualifierEntityId, boundaries );
					local link = context.formatSnak( parentOptions, qualifier );
	
					if p.config.reverseOrder then
						result = link .. ', ' .. result;
					else
						result = result .. ', ' .. link;
					end
					insertFromSnak( qualifier, entriesToLookupCategory )
					hasAdditionalQualifiers = true;
				end
			end
		end
	end

	if statement.mainsnak and
		statement.mainsnak.datavalue and
		statement.mainsnak.datavalue.value and
		statement.mainsnak.datavalue.value.id
	then
		local entityId = statement.mainsnak.datavalue.value.id;
		local parentSnaks = { statement.mainsnak };
		local parentEntityIds = { entityId };

		if actualDateBoundariesProperties ~= nil then
			local filterCapitalOf = {
				[ entityId ] = getParentsInBoundariesSnakImpl( context, entityId, boundaries, {'P70'} )
			};
			if boundaries then
				local entityOptions = context.cloneOptions( options );
				entityOptions['text'] = getLabel( context, entityId, boundaries );
				baseResult = context.formatSnak( entityOptions, statement.mainsnak, circumstances );

				local parentId = entityId;
				while parentId ~= nil do
					-- get parent
					local newParentSnaks = getParentsInBoundariesSnak( context, parentId, boundaries );
					if not newParentSnaks or #newParentSnaks == 0 then
						parentId = nil;
					elseif #newParentSnaks == 1 then
						local parentSnak = newParentSnaks[ 1 ];
						parentId = parentSnak.datavalue.value.id;
						
						local hasLoop = false
						for _, parentEntityId in pairs(parentEntityIds) do
							if parentEntityId == parentId then
								hasLoop = true
							end
						end
						
						if hasLoop then
							if p.config and p.config.catLoopInGeoChains then
								result = result .. p.config.catLoopInGeoChains;
							end
							break -- while parentId ~= nil do
						end
								
						table.insert( parentSnaks, parentSnak );
						table.insert( parentEntityIds, parentId );
						filterCapitalOf[ parentId ] = getParentsInBoundariesSnakImpl( context, parentId, boundaries, { 'P70' } );
					else
						parentId = nil;
						if p.config and p.config.catAmbiguousGeoChains then
							result = result .. p.config.catAmbiguousGeoChains;
						end
					end
				end

				if not hasAdditionalQualifiers then
					for i = 2, #parentSnaks, 1 do
						local parentSnak = parentSnaks[ i ];
						insertFromSnak( parentSnak, entriesToLookupCategory )
					end
				end

				--do not output similar countries like "Denmark, the Kingdom of Denmark"
				if (#parentSnaks > 1) then
					if ((parentSnaks[ #parentSnaks ].datavalue.value['numeric-id'] == 756617) and (parentSnaks[ #parentSnaks-1 ].datavalue.value['numeric-id'] == 35)) or ((parentSnaks[ #parentSnaks ].datavalue.value['numeric-id'] == 29999) and (parentSnaks[ #parentSnaks-1 ].datavalue.value['numeric-id'] == 55)) or ((parentSnaks[ #parentSnaks ].datavalue.value['numeric-id'] == 865) and (parentSnaks[ #parentSnaks-1 ].datavalue.value['numeric-id'] == 32081)) then
						table.remove( parentSnaks, #parentSnaks );
						table.remove( parentEntityIds, #parentEntityIds );
					end
				end

				--do not output (maternity) hospitals and houses
				if (#parentSnaks > 1) then
					local p31 = mw.wikibase.getAllStatements(parentEntityIds[ 1 ], 'P1');
					for _, iOf in pairs (p1) do
						if ((iOf.mainsnak.datavalue.value['numeric-id'] == 3947) or (iOf.mainsnak.datavalue.value['numeric-id'] == 16917) or (iOf.mainsnak.datavalue.value['numeric-id'] == 958822) or (iOf.mainsnak.datavalue.value['numeric-id'] == 2087181)) then
							baseResult = '';
						end
					end
				end

				do
					local capofstate = false;
					local i = #parentSnaks;
					while i > 1 do
						local prevEntityId = parentEntityIds[ i - 1 ];
						-- TODO: use English labels, if there is no current language labels
						local prevLabel = getLabel( context, prevEntityId, boundaries ) or '';
						local nextEntityId = parentEntityIds[ i ];
						local nextLabel = getLabel( context, nextEntityId, boundaries ) or '';
						if p.config and p.config.hideSameLabels == true and prevLabel == nextLabel then
							-- do not output same label twice (NY, NY, USA)
							table.remove( parentSnaks, i );
							table.remove( parentEntityIds, i );
						elseif p.config and p.config.hidePartOfLabels == true and isPartOfNext( prevLabel, ' ' .. nextLabel ) then
							-- do not output same label if it's part of previos
							table.remove( parentSnaks, i - 1 );
							table.remove( parentEntityIds, i - 1 );
						elseif p.config and p.config.hideUnitsForCapitals == true then
							-- do not ouput items whose capital is the first item
							local capitalId = nil;
							for _capitalId, capitalSnaks in pairs( filterCapitalOf ) do
								if #capitalSnaks > 0 then
									for __, capitalSnak in pairs( capitalSnaks ) do
										if parentSnaks[ i ].datavalue.value.id == capitalSnak.datavalue.value.id then
											capitalId = _capitalId;
											if (i == #parentSnaks) then
												capofstate = true;
											end
											break;
										end
									end
								end
							end
							if capitalId ~= nil then
								if i == #parentSnaks then
									i = i - 1;
								end
								-- always ouput constituent countries like England or Russian SFSR
								if (i == (#parentSnaks-1)) and (capofstate == false) then
									local p31 = mw.wikibase.getAllStatements(parentEntityIds[ i ], 'P1');
										for _, iOf in pairs (p1) do
											if (iOf.mainsnak.datavalue.value['numeric-id'] == 236036) or (iOf.mainsnak.datavalue.value['numeric-id'] == 3336843) or (iOf.mainsnak.datavalue.value['numeric-id'] == 12959600) or (iOf.mainsnak.datavalue.value['numeric-id'] == 56219758) or (iOf.mainsnak.datavalue.value['numeric-id'] == 15304003) or (iOf.mainsnak.datavalue.value['numeric-id'] == 66724388) then
												i = i - 1;
										end
									end
								end
								while i > 1 and parentEntityIds[ i ] ~= capitalId do
									table.remove( parentSnaks, i );
									table.remove( parentEntityIds, i );
									i = i - 1;
								end
							end
						end
						i = i - 1;
					end
				end

				if isSkipTopLevel( parentEntityIds[ #parentEntityIds ] ) then
					table.remove( parentSnaks, #parentEntityIds );
					table.remove( parentEntityIds, #parentEntityIds );
				end

				if not hasAdditionalQualifiers then
					for i = 2, #parentSnaks, 1 do
						local parentSnak = parentSnaks[ i ];

						local parentOptions = context.cloneOptions( options );
						parentOptions['text'] = getLabel( context, parentEntityIds[ i ], boundaries );

						local comma;
						if ((baseResult == '') and (i == 2)) then
							comma = '';
						else
							comma = ', ';
						end

						if p.config.reverseOrder then
							result = context.formatSnak( parentOptions, parentSnak ) .. comma .. result;
						else
							result = result .. comma .. context.formatSnak( parentOptions, parentSnak );
						end
					end
				end
			end
		end
	end

	if options[ 'thisLocationOnly' ] then
		result = baseResult;
	elseif p.config.reverseOrder then
		result = result .. baseResult;
	else
		result = baseResult .. result;
	end
	if options.references then
		result = result .. context.formatRefs( options, statement );
	end

	if categorizeByPlace then
		if property == 'P20' then result = result .. getCategory( 'P22', entriesToLookupCategory ); end
		if property == 'P23' then result = result .. getCategory( 'P65', entriesToLookupCategory ); end
		if property == 'P26' then result = result .. getCategory( 'P95', entriesToLookupCategory ); end
	end

	return result;
end

-- append entity id from snak to result
function insertFromSnak( snak, result )
	if not categorizeByPlace then
		return;
	end
	if snak and
		snak.datavalue and
		snak.datavalue.type == 'wikibase-entityid' and
		snak.datavalue.value and
		snak.datavalue.value[ 'entity-type' ] == 'item'
	then
		table.insert( result, snak.datavalue.value.id );
	end
end

function getCategory( propertyId, entriesToLookupCategoryFor )
	if mw.title.getCurrentTitle().namespace == 0 then
		for _, placeId in pairs( entriesToLookupCategoryFor ) do
			local claims = mw.wikibase.getBestStatements(placeId, propertyId);
	
			if claims then
				for _, claim in pairs( claims ) do
					if claim.mainsnak and
						claim.mainsnak and
						claim.mainsnak.datavalue and
						claim.mainsnak.datavalue.type == 'wikibase-entityid'
					then
						local catEntityId = claim.mainsnak.datavalue.value.id;
						local catSitelink = mw.wikibase.getSitelink(catEntityId);
						if (catSitelink) then
							return '[[' .. catSitelink .. ']]';
						end
					end
				end
			end
		end
	end
	return '';
end


local historicNamesProperties = { 'P7', 'P203', 'P77', 'P79' };

local langCode = mw.language.getContentLanguage():getCode();
local historicNamesPropertySelectors = {
	    P7 = 'P7[language:' .. langCode .. '][!P25,P25:Q105690470]',
		P203 = 'P203[language:' .. langCode .. '][!P25,P25:Q105690470]',
		P77 = 'P77[language:' .. langCode .. '][!P25,P25:Q105690470]',
		P79 = 'P79[language:' .. langCode .. '][!P25,P25:Q105690470]'
};

-- get current of historic name of place
function getLabel( context, entityId, boundaries )
	if not entityId then
		return nil;
	end
	if (type(entityId) ~= 'string') then error('incorrect type of entityId argument'); end;

    local label = nil;

	-- name from properties
	local results = getParentsInBoundariesSnakImpl( context, entityId, boundaries,
		historicNamesProperties, historicNamesPropertySelectors);

	for r, result in pairs( results ) do
		if result.datavalue and
			result.datavalue.value and
			result.datavalue.value.text
		then
			label = result.datavalue.value.text;
			break;
		end
	end

    -- name from label
    if label == nil then
		label = mw.wikibase.getLabel( entityId );
    end

    return label;
end

p.getLabel = getLabel;

local function calculateEndDateTimestamp( context, options, statement )
	if not context then error( 'context not specified' ) end;
	if not options then error( 'options not specified' ) end;
	if not options.entity then error( 'options.entity missing' ) end;
	if not statement then error( 'statement not specified' ) end;

	if statement.qualifiers and statement.qualifiers.P582 then
		for i, qualifier in ipairs( statement.qualifiers.P582 ) do
			local parsedTime = context.parseTimeFromSnak( qualifier );
			if parsedTime then
				return parsedTime;
			end
		end
	end

	-- check death day... do we have it at all?
	for h, propertyId in pairs( { "P247", "P19", "P572" } ) do
		local dateClaims = context.selectClaims( options, propertyId );
		if dateClaims then
			for i, statement in ipairs( dateClaims ) do
				local parsedTime = context.parseTimeFromSnak( statement.mainsnak );
				if parsedTime then
					return parsedTime;
				end
			end
		end
	end

	-- TODO: check other "end" properties

	-- no death day
	return os.time() * 1000;
end

local function deleteTwinAncestors( countryEntityId, propertyId ) --do not display countries which have twin ancestors
	local badTwins
	if ( countryEntityId == 'Q3123' ) then				--Great Britain and Ireland
		badTwins = {'Q380'}									--Great Brirani & NI
	elseif ( countryEntityId == 'Q3125' ) then			--FR of Yugoslavia
		badTwins = {'Q3126'}								--Serbia & Montenegro
	else
		return true;
	end

	local listforcheck
	if propertyId == 'P299' then
		listforcheck = mw.wikibase.getAllStatements( mw.wikibase.getEntityIdForCurrentPage(), propertyId );
	else
		listforcheck = mw.wikibase.getBestStatements( mw.wikibase.getEntityIdForCurrentPage(), propertyId );
	end

	for _, claim in pairs( listforcheck ) do
		if ( claim and claim.mainsnak
			and claim.mainsnak.datavalue
			and claim.mainsnak.datavalue.value
			and claim.mainsnak.datavalue.value.id ) then
			local actualId = claim.mainsnak.datavalue.value.id;
			for index, value in ipairs(badTwins) do
				if ( value == actualId ) then
            		return false;
        		end
			end
		end
	end
	return true; 
end

function p.formatCountryClaimWithFlag( context, options, statement )
	if not context then error('context not specified') end;
	if not options then error('options not specified') end;
	if not options.entity then error('options.entity is missing') end;
	if not statement then error('statement not specified') end;

	if not statement.mainsnak or
		not statement.mainsnak.datavalue or
		not statement.mainsnak.datavalue.value or
		not statement.mainsnak.datavalue.value.id
	then
		local result = context.formatStatementDefault( context, options, statement );
		if not result then
			return '';
		end
		return '<span class="country-name">' .. result .. '</span>';
	end
	
	local countryEntityId = statement.mainsnak.datavalue.value.id;
	local endDateTimestamp = calculateEndDateTimestamp( context, options, statement );
	local boundaries = getTimeBoundariesFromProperties( context, {'P247', 'P19', 'P85'} );

	if deleteTwinAncestors( countryEntityId, string.upper(options.property) ) then

		local countryOptions = context.cloneOptions( options );
		if not countryOptions['text'] or countryOptions['text'] == '' then
			countryOptions['text'] = getLabel( context, countryEntityId, boundaries );
		end

		local flag = Flags.getFlag( context, countryEntityId, endDateTimestamp );	
		if flag then
			return flag .. '&nbsp;<span class="country-name">' ..
				context.formatStatementDefault( context, countryOptions, statement ) ..
				'</span>';
		end

		return '<span class="country-name">' ..
			context.formatStatementDefault( context, countryOptions, statement ) ..
			'</span>';
	else
		return nil;
	end
end

return p;