/**
* Parse a JSON-LD document into an EasyRdf\Graph
*
* Attention: Since JSON-LD supports datasets, a document may contain
* multiple graphs and not just one. This parser returns only the
* default graph. An alternative would be to merge all graphs.
*
* @param Graph $graph the graph to load the data into
* @param string $data the RDF document data
* @param string $format the format of the input data
* @param string $baseUri the base URI of the data being parsed
*
* @throws Exception
* @throws \EasyRdf\Exception
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'jsonld') {
throw new \EasyRdf\Exception("EasyRdf\\Parser\\JsonLd does not support {$format}");
}
try {
$quads = LD\JsonLD::toRdf($data, array('base' => $baseUri));
} catch (LD\Exception\JsonLdException $e) {
throw new Exception($e->getMessage());
}
foreach ($quads as $quad) {
// Ignore named graphs
if (null !== $quad->getGraph()) {
continue;
}
$subject = (string) $quad->getSubject();
if ('_:' === substr($subject, 0, 2)) {
$subject = $this->remapBnode($subject);
}
$predicate = (string) $quad->getProperty();
if ($quad->getObject() instanceof \ML\IRI\IRI) {
$object = array('type' => 'uri', 'value' => (string) $quad->getObject());
if ('_:' === substr($object['value'], 0, 2)) {
$object = array('type' => 'bnode', 'value' => $this->remapBnode($object['value']));
}
} else {
$object = array('type' => 'literal', 'value' => $quad->getObject()->getValue());
if ($quad->getObject() instanceof LD\LanguageTaggedString) {
$object['lang'] = $quad->getObject()->getLanguage();
} else {
$object['datatype'] = $quad->getObject()->getType();
}
}
$this->addTriple($subject, $predicate, $object);
}
return $this->tripleCount;
}
/**
* Parse an N-Triples document into an EasyRdf\Graph
*
* @param Graph $graph the graph to load the data into
* @param string $data the RDF document data
* @param string $format the format of the input data
* @param string $baseUri the base URI of the data being parsed
*
* @throws Exception
* @throws \EasyRdf\Exception
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'ntriples') {
throw new \EasyRdf\Exception("EasyRdf\\Parser\\Ntriples does not support: {$format}");
}
$lines = preg_split('/\\x0D?\\x0A/', strval($data));
foreach ($lines as $index => $line) {
$lineNum = $index + 1;
if (preg_match('/^\\s*#/', $line)) {
# Comment
continue;
} elseif (preg_match('/^\\s*(.+?)\\s+<([^<>]+?)>\\s+(.+?)\\s*\\.\\s*$/', $line, $matches)) {
$this->addTriple($this->parseNtriplesSubject($matches[1], $lineNum), $this->unescapeString($matches[2]), $this->parseNtriplesObject($matches[3], $lineNum));
} elseif (preg_match('/^\\s*$/', $line)) {
# Blank line
continue;
} else {
throw new Exception("Failed to parse statement", $lineNum);
}
}
return $this->tripleCount;
}
/**
* Parse RDFa 1.1 into an EasyRdf\Graph
*
* @param Graph $graph the graph to load the data into
* @param string $data the RDF document data
* @param string $format the format of the input data
* @param string $baseUri the base URI of the data being parsed
*
* @throws \EasyRdf\Exception
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'rdfa') {
throw new \EasyRdf\Exception("EasyRdf\\Parser\\Rdfa does not support: {$format}");
}
// Initialise evaluation context.
$context = $this->initialContext();
libxml_use_internal_errors(true);
// Parse the document into DOM
$doc = new \DOMDocument();
// Attempt to parse the document as strict XML, and fall back to HTML
// if XML parsing fails.
if ($doc->loadXML($data, LIBXML_NONET)) {
if ($this->debug) {
print "Document was parsed as XML.";
}
// Collect all xmlns namespaces defined throughout the document.
$sxe = simplexml_import_dom($doc);
$context['xmlns'] = $sxe->getDocNamespaces(true);
unset($context['xmlns']['']);
} else {
$doc->loadHTML($data);
if ($this->debug) {
print "Document was parsed as HTML.";
}
}
// Establish the base for both XHTML and HTML documents.
$xpath = new \DOMXPath($doc);
$xpath->registerNamespace('xh', "http://www.w3.org/1999/xhtml");
$nodeList = $xpath->query('/xh:html/xh:head/xh:base');
if ($node = $nodeList->item(0) and $href = $node->getAttribute('href')) {
$this->baseUri = new ParsedUri($href);
}
$nodeList = $xpath->query('/html/head/base');
if ($node = $nodeList->item(0) and $href = $node->getAttribute('href')) {
$this->baseUri = new ParsedUri($href);
}
// Remove the fragment from the base URI
$this->baseUri->setFragment(null);
// Recursively process XML nodes
$this->processNode($doc, $context);
return $this->tripleCount;
}
/**
* Parse an RDF document into an EasyRdf\Graph
*
* @param Graph $graph the graph to load the data into
* @param string $data the RDF document data
* @param string $format the format of the input data
* @param string $baseUri the base URI of the data being parsed
*
* @throws Exception
* @throws \EasyRdf\Exception
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
$parser = librdf_new_parser($this->world, $format, null, null);
if (!$parser) {
throw new \EasyRdf\Exception("Failed to create librdf_parser of type: {$format}");
}
$rdfUri = librdf_new_uri($this->world, $baseUri);
if (!$rdfUri) {
throw new \EasyRdf\Exception("Failed to create librdf_uri from: {$baseUri}");
}
$stream = librdf_parser_parse_string_as_stream($parser, $data, $rdfUri);
if (!$stream) {
throw new Exception("Failed to parse RDF stream");
}
do {
$statement = librdf_stream_get_object($stream);
if ($statement) {
$subject = self::nodeUriString(librdf_statement_get_subject($statement));
$predicate = self::nodeUriString(librdf_statement_get_predicate($statement));
$object = self::nodeToRdfPhp(librdf_statement_get_object($statement));
$this->addTriple($subject, $predicate, $object);
}
} while (!librdf_stream_next($stream));
$errorCount = $this->parserErrorCount($parser);
if ($errorCount) {
throw new Exception("{$errorCount} errors while parsing.");
}
librdf_free_uri($rdfUri);
librdf_free_stream($stream);
librdf_free_parser($parser);
return $this->tripleCount;
}
/**
* Parse an RDF/XML document into an EasyRdf\Graph
*
* @param Graph $graph the graph to load the data into
* @param string $data the RDF document data
* @param string $format the format of the input data
* @param string $baseUri the base URI of the data being parsed
*
* @throws Exception
* @throws \EasyRdf\Exception
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'rdfxml') {
throw new \EasyRdf\Exception("EasyRdf\\Parser\\RdfXml does not support: {$format}");
}
$this->init($graph, $baseUri);
$this->resetBnodeMap();
/* xml parser */
$this->initXMLParser();
/* parse */
if (!xml_parse($this->xmlParser, $data, false)) {
$message = xml_error_string(xml_get_error_code($this->xmlParser));
throw new Exception('XML error: "' . $message . '"', xml_get_current_line_number($this->xmlParser), xml_get_current_column_number($this->xmlParser));
}
xml_parser_free($this->xmlParser);
return $this->tripleCount;
}