/**
* Parse RDF/PHP into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return boolean true if parsing was successful
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'php') {
throw new EasyRdf_Exception("EasyRdf_Parser_RdfPhp does not support: {$format}");
}
// Reset the bnode mapping
$this->resetBnodeMap();
foreach ($data as $subject => $properties) {
if (substr($subject, 0, 2) === '_:') {
$subject = $this->remapBnode($graph, $subject);
} elseif (preg_match('/^\\w+$/', $subject)) {
# Cope with invalid RDF/JSON serialisations that
# put the node name in, without the _: prefix
# (such as net.fortytwo.sesametools.rdfjson)
$subject = $this->remapBnode($graph, $subject);
}
foreach ($properties as $property => $objects) {
foreach ($objects as $object) {
if ($object['type'] === 'bnode') {
$object['value'] = $this->remapBnode($graph, $object['value']);
}
$graph->add($subject, $property, $object);
}
}
}
return true;
}
/**
* 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 object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return integer The number of triples added to the graph
*/
public function parse($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 {
$quads = \ML\JsonLD\JsonLD::toRdf($data, array('base' => $baseUri));
} catch (\ML\JsonLD\Exception\JsonLdException $e) {
throw new EasyRdf_Parser_Exception($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 \ML\JsonLD\LanguageTaggedString) {
$object['lang'] = $quad->getObject()->getLanguage();
} else {
$object['datatype'] = $quad->getObject()->getType();
}
}
$this->addTriple($subject, $predicate, $object);
}
return $this->tripleCount;
}
/**
* Parse RDF/PHP into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return boolean true if parsing was successful
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'php') {
throw new EasyRdf_Exception("EasyRdf_Parser_RdfPhp does not support: {$format}");
}
// Reset the bnode mapping
$this->resetBnodeMap();
# FIXME: validate the data (?)
foreach ($data as $subject => $properties) {
if (substr($subject, 0, 2) === '_:') {
$subject = $this->remapBnode($graph, $subject);
}
foreach ($properties as $property => $objects) {
foreach ($objects as $object) {
if ($object['type'] === 'bnode') {
$object['value'] = $this->remapBnode($graph, $object['value']);
}
$graph->add($subject, $property, $object);
}
}
}
return true;
}
/**
* Parse an RDF document into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return boolean true if parsing was successful
*/
public function parse($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 EasyRdf_Exception("Failed to parse RDF stream for: {$rdfUri}");
}
do {
$statement = librdf_stream_get_object($stream);
if ($statement) {
$subject = EasyRdf_Parser_Redland::nodeUriString(librdf_statement_get_subject($statement));
$predicate = EasyRdf_Parser_Redland::nodeUriString(librdf_statement_get_predicate($statement));
$object = EasyRdf_Parser_Redland::nodeToArray(librdf_statement_get_object($statement));
$graph->add($subject, $predicate, $object);
}
} while (!librdf_stream_next($stream));
$errorCount = $this->parserErrorCount($parser);
if ($errorCount) {
throw new EasyRdf_Exception("{$errorCount} errors while parsing.");
}
librdf_free_uri($rdfUri);
librdf_free_stream($stream);
librdf_free_parser($parser);
// Success
return true;
}
/**
* Parse an N-Triples document into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return integer The number of triples added to the graph
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'ntriples') {
throw new EasyRdf_Exception("EasyRdf_Parser_Ntriples does not support: {$format}");
}
$lines = preg_split("/[\r\n]+/", strval($data));
foreach ($lines as $line) {
if (preg_match("/^\\s*#/", $line)) {
continue;
} else {
if (preg_match("/(.+)\\s+<([^<>]+)>\\s+(.+)\\s*\\./", $line, $matches)) {
$this->addTriple($this->parseNtriplesSubject($matches[1]), $this->unescapeString($matches[2]), $this->parseNtriplesObject($matches[3]));
}
}
}
return $this->_tripleCount;
}
/**
* Parse an RDF/XML document into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return integer The number of triples added to the graph
*/
public function parse($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, $data, false)) {
throw new EasyRdf_Exception('XML error: "' . xml_error_string(xml_get_error_code($this->_xmlParser)) . '" at line ' . xml_get_current_line_number($this->_xmlParser));
}
xml_parser_free($this->_xmlParser);
return $this->_tripleCount;
}
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
// Parsing goes here
return true;
}
/**
* Parse RDFa 1.1 into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return integer The number of triples added to the graph
*/
public function parse($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')) {
$this->_baseUri = new EasyRdf_ParsedUri($href);
}
$nodeList = $xpath->query('/html/head/base');
if ($node = $nodeList->item(0) and $href = $node->getAttribute('href')) {
$this->_baseUri = new EasyRdf_ParsedUri($href);
}
// Remove the fragment from the base URI
$this->_baseUri->setFragment(NULL);
// Recursively process XML nodes
$this->processNode($doc, $context);
return $this->_tripleCount;
}
public function testParseUndefined()
{
$this->setExpectedException('EasyRdf_Exception', 'This method should be overridden by sub-classes.');
$parser = new EasyRdf_Parser();
$parser->parse($this->graph, 'data', 'format', 'baseUri');
}
/**
* Parse RDF/PHP into an EasyRdf_Graph
*
* @param string $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 $baseUri the base URI of the data being parsed
* @return boolean true if parsing was successful
*/
public function parse($graph, $data, $format, $baseUri)
{
parent::checkParseParams($graph, $data, $format, $baseUri);
if ($format != 'php') {
throw new EasyRdf_Exception("EasyRdf_Parser_RdfPhp does not support: {$format}");
}
# Convert into an object graph
foreach ($data as $subject => $touple) {
$type = $this->getResourceType($data, $subject);
# Is this a bnode?
if (substr($subject, 0, 2) == '_:') {
if (!isset($this->_bnodeMap[$subject])) {
$this->_bnodeMap[$subject] = $graph->newBNode($type);
}
$res = $this->_bnodeMap[$subject];
} else {
$res = $graph->resource($subject, $type);
}
foreach ($touple as $property => $objects) {
$this->addProperty($graph, $data, $res, $property, $objects);
}
}
return true;
}
/**
* Parse an N-Triples document into an EasyRdf_Graph
*
* @param object EasyRdf_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 $baseUri the base URI of the data being parsed
* @return integer The number of triples added to the graph
*/
public function parse($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 EasyRdf_Parser_Exception("Failed to parse statement", $lineNum);
}
}
return $this->tripleCount;
}
