public function extractPage($pageId, $pageTitle, $pageSource)
{
$this->setPageURI($pageTitle);
// Set up the result object
$result = new ExtractionResult($pageId, $this->language, $this->getExtractorID());
// Return empty result if there is no title
if ($this->decode_title($pageTitle) == NULL) {
return $result;
}
try {
$breadCrumb = new Breadcrumb($pageId);
//echo "\n\n\n\n\nPAGE = $pageId --- $pageTitle\n\n";
$triples = $this->rootTripleGenerator->generate($breadCrumb, $pageSource);
//print_r($triples);
// Add annotation to all triples
//$rootSubject = $this->getPageURI();//RDFTriple::page($pageId);
//$rootSubject = RDFTriple::page($pageId);
//$this->log(DEBUG, "diff root to getPageURI: $rootSubject | {$this->getPageURI()}");
$this->log(TRACE, "generated triples:");
$logmsg = "";
foreach ($triples as $triple) {
$triple->addOnDeleteCascadeAnnotation($this->getPageURI());
$result->addTripleObject($triple);
$logmsg .= $triple->toNTriples();
}
$this->log(TRACE, "\n" . $logmsg);
} catch (Exception $e) {
$this->log(WARN, "Caught exception: " . $e->getMessage());
}
//$this->log('info','LiveMappingBasedExtractor: Count of generated triples ' + sizeof($result));
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
global $parseResult;
// Contains the Extraction result
$parseResult = null;
$this->parsePage($pageID, $pageSource, $this->language);
if (count($parseResult) < 1) {
return $result;
}
$knownProperties = array($parseResult[0][1]);
foreach ($parseResult as $myTriple) {
try {
$subject = RDFtriple::URI($myTriple[0]);
} catch (Exception $e) {
echo 'Caught exception: ', $e->getMessage(), "\n";
continue;
}
// Rename Properties like LeaderName1, LeaderName2, ... to LeaderName
if (preg_match("/(.*[^0-9_]+)([0-9])\$/", $myTriple[1], $matches)) {
// if property consist of letters from another writing system then latin, e.g. Korean,
// the words are decoded as e.g. _percent_B1, this must not be changed
// if language.use_percent_encoding = false, it looks like e.g. %B1
if (substr(substr($myTriple[1], -11), 0, 9) != "_percent_" && !ereg("%([A-F0-9]{2})", substr($myTriple[1], -3))) {
$key = array_search($matches[1], $knownProperties);
if ($key) {
$myTriple[1] = $knownProperties[$key];
} else {
array_push($knownProperties, $matches[1]);
$myTriple[1] = $matches[1];
}
}
} else {
if (!array_search($myTriple[1], $knownProperties)) {
array_push($knownProperties, $myTriple[1]);
}
}
// if a property is longer than the maximum configured length, we do
// do not write the triple
if (strlen($myTriple[1]) > $GLOBALS['W2RCFG']['maximumPropertyLength']) {
continue;
}
try {
$predicate = RDFtriple::URI($myTriple[1]);
} catch (Exception $e) {
echo 'Caught exception: ', $e->getMessage(), "\n";
continue;
}
if ($myTriple[3] == "r") {
try {
$object = RDFtriple::URI($myTriple[2]);
} catch (Exception $e) {
echo 'Caught exception: ', $e->getMessage(), "\n";
continue;
}
} else {
if ($myTriple[5] == null) {
$myTriple[5] = $this->language;
}
$object = RDFtriple::literal($myTriple[2], $myTriple[4], $myTriple[5]);
}
//this is for the db:London/rating
//subtemplate problem
$triple = new RDFtriple($subject, $predicate, $object);
$currentSubject = RDFtriple::page($pageID);
$small = $currentSubject->getURI();
$big = $subject->getURI();
if (strpos($big, $small) === 0 && strlen($big) > strlen($small)) {
$triple->addOnDeleteCascadeAnnotation($currentSubject);
}
$result->addTripleObject($triple);
$this->allPredicates->addPredicate($myTriple[1]);
}
return $result;
}
