/**
* Adds the given Statement's main Snak to the RDF graph.
*
* @todo share more of this code with FullStatementRdfBuilder
*
* @param EntityId $entityId
* @param Statement $statement
*
* @throws InvalidArgumentException
*/
private function addMainSnak(EntityId $entityId, Statement $statement)
{
$snak = $statement->getMainSnak();
$entityLName = $this->vocabulary->getEntityLName($entityId);
$this->writer->about(RdfVocabulary::NS_ENTITY, $entityLName);
$this->snakBuilder->addSnak($this->writer, $snak, RdfVocabulary::NSP_DIRECT_CLAIM);
}
/**
* Creates a value node for $value, and attaches it to the current subject of $writer.
* If a value node for $value was already created, null is returned. Otherwise, the
* value node's lname is returned, which should be used to generate detailed about the
* value into the writer returned by getValueNodeWriter().
*
* When this method returns a non-null lname, the current subject of the RdfWriter returned by
* getValueNodeWriter() will the be value node with that lname.
*
* @param RdfWriter $writer
* @param string $propertyValueNamespace Property value relation namespace
* @param string $propertyValueLName Property value relation name
* @param string $dataType Property data type (unused, passed here for symmetry
* with the signature of ValueSnakRdfBuilder::addValue).
* @param DataValue $value
*
* @return string|null The LName of the value node (in the RdfVocabulary::NS_VALUE namespace),
* or null if the value node should not be processed (generally, because it already has
* been processed).
*/
public function attachValueNode(RdfWriter $writer, $propertyValueNamespace, $propertyValueLName, $dataType, DataValue $value)
{
$valueLName = $value->getHash();
$writer->say(RdfVocabulary::$claimToValue[$propertyValueNamespace], $propertyValueLName)->is(RdfVocabulary::NS_VALUE, $valueLName);
if ($this->dedupeBag->alreadySeen($valueLName, 'V') !== false) {
return null;
}
$this->valueNodeWriter->about(RdfVocabulary::NS_VALUE, $valueLName)->a(RdfVocabulary::NS_ONTOLOGY, $this->vocabulary->getValueTypeName($value));
return $valueLName;
}
/**
* Adds the aliases of the given entity to the RDF graph.
*
* @param string $entityLName
* @param AliasGroupList $aliases
*/
public function addAliases($entityLName, AliasGroupList $aliases)
{
/** @var AliasGroup $aliasGroup */
foreach ($aliases as $aliasGroup) {
$languageCode = $aliasGroup->getLanguageCode();
if ($this->languages !== null && !isset($this->languages[$languageCode])) {
continue;
}
foreach ($aliasGroup->getAliases() as $alias) {
$this->writer->about(RdfVocabulary::NS_ENTITY, $entityLName)->say(RdfVocabulary::NS_SKOS, 'altLabel')->text($alias, $languageCode);
}
}
}
/**
* Adds the site links of the given item to the RDF graph.
*
* @param Item $item
*/
public function addSiteLinks(Item $item)
{
$entityLName = $this->vocabulary->getEntityLName($item->getId());
/** @var SiteLink $siteLink */
foreach ($item->getSiteLinkList() as $siteLink) {
if (!$this->isSiteIncluded($siteLink->getSiteId())) {
continue;
}
// FIXME: we should check the site exists using hasGlobalId here before asuming it does
$site = $this->siteLookup->getSite($siteLink->getSiteId());
// XXX: ideally, we'd use https if the target site supports it.
$baseUrl = str_replace('$1', rawurlencode($siteLink->getPageName()), $site->getLinkPath());
// $site->getPageUrl( $siteLink->getPageName() );
if (!parse_url($baseUrl, PHP_URL_SCHEME)) {
$url = "http:" . $baseUrl;
} else {
$url = $baseUrl;
}
$this->writer->about($url)->a(RdfVocabulary::NS_SCHEMA_ORG, 'Article')->say(RdfVocabulary::NS_SCHEMA_ORG, 'about')->is(RdfVocabulary::NS_ENTITY, $entityLName)->say(RdfVocabulary::NS_SCHEMA_ORG, 'inLanguage')->text($this->vocabulary->getCanonicalLanguageCode($site->getLanguageCode()));
foreach ($siteLink->getBadges() as $badge) {
$this->writer->say(RdfVocabulary::NS_ONTOLOGY, 'badge')->is(RdfVocabulary::NS_ENTITY, $this->vocabulary->getEntityLName($badge));
}
}
}
/**
* Adds the given Statement's main Snak to the RDF graph.
*
* @param EntityId $entityId
* @param string $statementLName
* @param Statement $statement
* @param bool $isBest Is this best ranked statement?
*/
private function addMainSnak(EntityId $entityId, $statementLName, Statement $statement, $isBest)
{
$snak = $statement->getMainSnak();
$entityLName = $this->vocabulary->getEntityLName($entityId);
$propertyLName = $this->vocabulary->getEntityLName($snak->getPropertyId());
$this->statementWriter->about(RdfVocabulary::NS_ENTITY, $entityLName)->say(RdfVocabulary::NSP_CLAIM, $propertyLName)->is(RdfVocabulary::NS_STATEMENT, $statementLName);
$this->statementWriter->about(RdfVocabulary::NS_STATEMENT, $statementLName)->a(RdfVocabulary::NS_ONTOLOGY, 'Statement');
$rank = $statement->getRank();
if (isset(RdfVocabulary::$rankMap[$rank])) {
if ($isBest) {
$this->statementWriter->a(RdfVocabulary::NS_ONTOLOGY, RdfVocabulary::WIKIBASE_RANK_BEST);
}
$this->statementWriter->about(RdfVocabulary::NS_STATEMENT, $statementLName)->say(RdfVocabulary::NS_ONTOLOGY, 'rank')->is(RdfVocabulary::NS_ONTOLOGY, RdfVocabulary::$rankMap[$rank]);
} else {
wfLogWarning("Unknown rank {$rank} encountered for {$entityId}:{$statement->getGuid()}");
}
$this->snakBuilder->addSnak($this->statementWriter, $snak, RdfVocabulary::NSP_CLAIM_STATEMENT);
}
/**
* Create header structure for the dump
*
* @param int $timestamp Timestamp (for testing)
*/
public function addDumpHeader($timestamp = 0)
{
// TODO: this should point to "this document"
$this->writer->about(RdfVocabulary::NS_ONTOLOGY, 'Dump')->a(RdfVocabulary::NS_SCHEMA_ORG, "Dataset")->a('owl', 'Ontology')->say(RdfVocabulary::NS_CC, 'license')->is(RdfVocabulary::LICENSE)->say(RdfVocabulary::NS_SCHEMA_ORG, 'softwareVersion')->value(RdfVocabulary::FORMAT_VERSION)->say(RdfVocabulary::NS_SCHEMA_ORG, 'dateModified')->value(wfTimestamp(TS_ISO_8601, $timestamp), 'xsd', 'dateTime')->say('owl', 'imports')->is(RdfVocabulary::getOntologyURI());
}
