public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
$meta = $this->additionalInfo;
//generate dc modified
$date = date('c');
$datatype = XS_DATETIME;
$o = RDFtriple::Literal($date, $datatype, "");
$this->log('trace', $o->toString());
$result->addTriple($this->getPageURI(), RDFtriple::URI(DC_MODIFIED, false), $o);
//OAIIDENTIFIER
$this->log(DEBUG, $meta['oaiId']);
$result->addTriple($this->getPageURI(), RDFtriple::URI(DBM_OAIIDENTIFIER, false), new RDFLiteral($meta['oaiId'], XS_INTEGER));
//REVISION
$revisionURI = 'http://' . $meta['language'] . '.wikipedia.org/w/index.php?title=';
$revisionURI .= urlencode($pageID) . '&oldid=' . $meta['revision'];
//http://en.wikipedia.org/w/index.php?title=Robotics&oldid=293678514
$this->log(DEBUG, $revisionURI);
$result->addTriple($this->getPageURI(), RDFtriple::URI(DBM_REVISION, false), RDFtriple::URI($revisionURI));
$editLink = 'http://' . $meta['language'] . '.wikipedia.org/w/index.php?title=';
$editLink .= urlencode($pageID) . '&action=edit';
$this->log(DEBUG, $editLink);
$result->addTriple($this->getPageURI(), RDFtriple::URI(DBM_EDITLINK, false), RDFtriple::URI($editLink));
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$value = Util::replaceWikiLinks($value);
//TODO: WARUM NUR IN DIESEM FALL CITE RAUSNEHMEN?
preg_match_all("/{{2}cite.*?\\}{2}/i", $value, $matches);
foreach ($matches as $match) {
if (!array_key_exists(0, $match)) {
continue;
}
$value = str_replace($match[0], Util::replaceTemplates($match[0]), $value);
}
$value = Util::removeHtmlTags($value);
$value = Util::removeHtmlComments($value);
// get unit exact type
// Some arguments have a fixed type - e.g. weight_lb -> pounds
$unitExactType = null;
//UnitValueParser::parseValue($propvalue, $this->language, array($unit_type, $unit_exact_type, $propkey));
$parseResultArray = $this->parser->parse($value);
if (isset($parseResultArray)) {
foreach ($parseResultArray as $parseResults) {
$parsedDataType = $parseResults[1];
if ($parsedDataType == "") {
$parsedDataType = null;
}
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal((string) $parseResults[0], $parsedDataType, null));
}
} else {
//TODO: GENERATE LOGFILE WITH UNPARSED VALUES
$result[] = new RDFTriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($value));
}
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$parseResults = $this->parser->parse($value);
ob_start();
$str = "Date parser \n";
$str .= "value was: {$value} \n";
print_r($parseResults);
$str .= ob_get_contents();
ob_end_clean();
Logger::debug($str);
if (!isset($parseResults)) {
return $result;
}
$datePattern = "/\\d\\d\\d\\d-(0[1-9]|1[012])-(0[1-9]|[12][0-9]|3[01])/";
/*
if(preg_match($datePattern, $parseResults) != 1)
return $result;
$isValidDate = checkdate(
substr($parseResults, 5, 2),
substr($parseResults, 8, 2),
substr($parseResults, 0, 4));
*/
$isValidDate = true;
if ($isValidDate) {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($parseResults[0], $parseResults[1], null));
//print_r($result);
} else {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($parseResults, null, null));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
$category = Util::getMediaWikiNamespace($this->language, MW_CATEGORY_NAMESPACE);
if (preg_match_all("/" . $category . ":(.*)/", $pageID, $match)) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(SKOS_PREFLABEL, false), RDFtriple::Literal($this->decode_title($pageTitle), NULL, $this->language));
$result->addTriple($this->getPageURI(), RDFtriple::URI(RDF_TYPE, false), RDFtriple::URI(SKOS_CONCEPT, false));
if (preg_match_all("/\\[\\[" . $category . ":(.*)\\]\\]/", $pageSource, $matches, PREG_SET_ORDER)) {
foreach ($matches as $match) {
// split on | sign
if (strpos($match[1], '|') === false) {
$object = Util::getDBpediaCategoryPrefix($this->language) . URI::wikipediaEncode($match[1]);
} else {
$split = explode('|', $match[1]);
$object = Util::getDBpediaCategoryPrefix($this->language) . URI::wikipediaEncode($split[0]);
}
try {
$object = RDFtriple::URI($object);
} catch (Exception $e) {
echo 'Caught exception: ', $e->getMessage(), "\n";
continue;
}
$result->addTriple($this->getPageURI(), RDFtriple::URI(SKOS_BROADER, false), $object);
}
}
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$CleanSource = $this->remove_wikicode($pageSource);
$LongAbstract = $this->extract_abstract($CleanSource, 3000, false, false);
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate("abstract"), RDFtriple::Literal($LongAbstract, NULL, $this->language));
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$parseResults = $this->parser->parse($value);
foreach ($parseResults as $mystring) {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($mystring));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$CleanSource = $this->remove_wikicode($pageSource);
$Abstract = $this->extract_abstract($CleanSource);
//$LongAbstract = $this->extract_abstract($CleanSource, 3000, false, false);
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2000/01/rdf-schema#comment"), RDFtriple::Literal($Abstract, NULL, $this->language));
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
if ($this->decode_title($pageTitle) == NULL) {
return $result;
}
$result->addTriple($this->getPageURI(), RDFtriple::URI(RDFS_LABEL, false), RDFtriple::Literal($this->decode_title($pageTitle), NULL, $this->language));
return $result;
}
public function getPredicateTriples()
{
$predicateTriples = new ExtractionResult($this->pageID, $this->language, $this->extractorID);
foreach ($this->predicates as $subject => $bool) {
// array_push( $predicateTriples, new RDFtriple($subject, "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"));
$predicateTriples->addTriple(RDFtriple::URI($subject), RDFtriple::URI("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), RDFtriple::URI("http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"));
$predicateTriples->addTriple(RDFtriple::URI($subject), RDFtriple::URI("http://www.w3.org/2000/01/rdf-schema#label"), RDFtriple::Literal($this->getPredicateLabel($subject)));
}
return $predicateTriples;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$image = "image";
$parsedText = ActiveAbstractExtractor::stripMarkup($value, $image);
$parsedText = trim($parsedText);
if ($parsedText != "") {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($parsedText, null, $this->language));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
// Locate geo coordinates
if (preg_match('/<geo>([\\-0-9\\.]+);([\\-0-9\\.]+)[^0-9]*[^<]*<\\/geo>/', $pageSource, $match)) {
$lat = $match[1];
$long = $match[2];
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2003/01/geo/wgs84_pos#lat"), RDFtriple::Literal($lat, "http://www.w3.org/2001/XMLSchema#float", NULL));
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2003/01/geo/wgs84_pos#long"), RDFtriple::Literal($long, "http://www.w3.org/2001/XMLSchema#float", NULL));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$PersonData = $this->extractPersondata($pageSource, $this->language);
//var_dump($PersonData);
if ($PersonData != null) {
// preg_match("/\[\[en:(.*)\]\]/", $pageSource, $LangLinkmatch);
// $PersonData['enPageID'] = str_replace(" ","_",$LangLinkmatch[1]);
$WikiDB = new DatabaseWikipedia($this->language);
preg_match("/\\[\\[([^\\]]*)\\]\\]/", $PersonData['birthplace'], $Birthplacematch);
$Birthplacematch = $this->getLinkForLabeledLink($Birthplacematch);
$mySource = $WikiDB->getSource($Birthplacematch);
preg_match("/\\[\\[en:(.*)\\]\\]/", $mySource, $LangLinkmatch);
$BirthPlace = $LangLinkmatch[1];
preg_match("/\\[\\[([^\\]]*)\\]\\]/", $PersonData['deathplace'], $Deathplacematch);
$Deathplacematch = $this->getLinkForLabeledLink($Deathplacematch);
$mySource = $WikiDB->getSource($Deathplacematch);
preg_match("/\\[\\[en:(.*)\\]\\]/", $mySource, $LangLinkmatch);
$DeathPlace = $LangLinkmatch[1];
//var_dump($PersonData);
//var_dump($BirthPlace);
//var_dump($DeathPlace);
//var_dump($Deathplacematch);
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://xmlns.com/foaf/0.1/name"), RDFtriple::Literal($PersonData['name'], null, "de"));
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://xmlns.com/foaf/0.1/givenname"), RDFtriple::Literal($PersonData['givenname'], null, "de"));
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://xmlns.com/foaf/0.1/surname"), RDFtriple::Literal($PersonData['surname'], null, "de"));
if ($BirthPlace != "") {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate("birthPlace"), RDFtriple::page($BirthPlace));
// $result->addTriple(
// RDFtriple::page($pageID),
// RDFtriple::URI("http://purl.org/vocab/bio/0.1/event"),
// RDFtriple::URI("http://dbpedia.org/resource/" . URI::wikipediaEncode($pageID) . "/Birth"));
// $result->addTriple(
// RDFtriple::URI("http://dbpedia.org/resource/" . URI::wikipediaEncode($pageID) . "/Birth");
// RDFtriple::URI("http://purl.org/vocab/bio/0.1/place"),
// RDFtriple::page($BirthPlace));
}
if ($PersonData['birthdate'] != "") {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate("birth"), RDFtriple::Literal($PersonData['birthdate'], "http://www.w3.org/2001/XMLSchema#date", null));
}
if ($DeathPlace != "") {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate("deathPlace"), RDFtriple::page($DeathPlace));
}
if ($PersonData['deathdate'] != "") {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate("death"), RDFtriple::Literal($PersonData['deathdate'], "http://www.w3.org/2001/XMLSchema#date", null));
}
if ($PersonData['description'] != "") {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://purl.org/dc/elements/1.1/description"), RDFtriple::Literal($PersonData['description'], null, "de"));
}
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), RDFtriple::URI("http://xmlns.com/foaf/0.1/Person"));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
//create a new Extraction Result to hold all extrated Triples
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
//Look for {{chembox header}} in PageSource
if (preg_match("/{{chembox header}}/", $pageSource, $match)) {
//DO SOME PARSING
//Add a Triple for each Property
$result->addTriple($this->getPageURI(), RDFtriple::URI(DB_MY_CHEM_PROPERTY, false), RDFtriple::Literal("my_value"));
//Add each Predicate to the Predicate Collection
$this->allPredicates->addPredicate("my_chem_property");
}
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
if (strpos($value, "{{") !== false) {
$value = Util::replaceTemplates($value);
}
$value = Util::removeHtmlTags($value);
$value = Util::removeHtmlComments($value);
$value = Util::removeWikiEmphasis($value);
$parseResults = $this->parser->parse($value);
if (isset($parseResults)) {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($parseResults));
} else {
//TODO: ADD DEGUB LOGFILE FOR UN-PARSED TRIPLES
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($value));
}
return $result;
}
public function findPND($text, $pageTitle, &$result)
{
$templates = Util::getTemplates($text);
foreach ($templates as $template) {
if ($template["name"] == "Normdaten") {
preg_match('/\\|\\s*PND\\s*=\\s*([0-9X]*)/i', $template["content"], $match);
if (isset($match[1])) {
// add individualised PND
$result->addTriple(RDFtriple::page($this->pageID), RDFtriple::URI(DBO_INDIVIDUALISED_PND, false), RDFtriple::Literal($match[1], NULL, NULL));
}
} else {
if ($template["name"] == "PND") {
preg_match('/\\s*PND\\s*\\|\\s*([0-9X]*)(.*)/i', $template["content"], $match);
if (isset($match)) {
if (isset($match[1]) && strlen($match[1]) >= 9) {
// add individualised PND
$result->addTriple(RDFtriple::page($this->pageID), RDFtriple::URI(DBO_INDIVIDUALISED_PND, false), RDFtriple::Literal($match[1], NULL, NULL));
}
if (isset($match[2])) {
preg_match('/\\|\\s*([0-9X]*)/i', $match[2], $match1);
if (isset($match1[1]) && strlen($match1[1]) >= 9) {
// add non-individualised PND
$result->addTriple(RDFtriple::page($this->pageID), RDFtriple::URI(DBO_NON_INDIVIDUALISED_PND, false), RDFtriple::Literal($match1[1], NULL, NULL));
}
}
}
} else {
if ($template["name"] == "PNDfehlt") {
preg_match('/\\s*PNDfehlt\\s*(\\|\\s*.*)/i', $template["content"], $match);
if (isset($match[1])) {
preg_match('/\\|\\s*([0-9X]*)/i', $match[1], $match1);
if (isset($match1[1]) && strlen($match1[1]) >= 9) {
// add non-individualised PND
$result->addTriple(RDFtriple::page($this->pageID), RDFtriple::URI(DBO_NON_INDIVIDUALISED_PND, false), RDFtriple::Literal($match1[1], NULL, NULL));
}
}
}
}
}
}
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
//TODO image namespace
$text = $pageSource;
//TODO not sure what to take as magic number here:
// 4096 was to short, e.g. inappropriate for london
$text = substr($text, 0, 8192);
$text = self::stripMarkup($text);
//TODO REMOVE THIS LINE FOR DEBUGGING:
$text = $this->_exceptions($text);
//2 is probalby perfect, since it guarantuees a certain lentgh
$firstTwoSentences = $this->_extractStart($text, 2);
//better than nothing
$fullabstract = $firstTwoSentences;
//this is crazy code as it could also be 0
//it is a heuristical approach to nicen the abstracts.
// take anything until you find '=='
if (($pos = strpos($text, '==')) !== false) {
$fullabstract = trim(substr($text, 0, $pos));
} else {
$fullabstract = trim($text);
}
if (!empty($firstTwoSentences)) {
$s = $this->getPageURI();
$p = RDFtriple::URI(DBCOMM_COMMENT, false);
$o = RDFtriple::Literal($firstTwoSentences, NULL, $this->language);
$this->log('debug', 'Found: ' . $s->toString() . " " . $p->toString() . " " . $o->toString());
$result->addTriple($s, $p, $o);
}
if (!empty($fullabstract)) {
$s = $this->getPageURI();
$p = RDFtriple::URI(DBCOMM_ABSTRACT, false);
$o = RDFtriple::Literal($fullabstract, NULL, $this->language);
$this->log('debug', 'Found: ' . $s->toString() . " " . $p->toString() . " " . $o->toString());
$result->addTriple($s, $p, $o);
}
//TODO $clipped = substr( $extract, 0, 1024 );
//TODO UtfNormal::cleanUp( $clipped ); in include/normal/UtfNormal
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$value = Util::removeHtmlTags($value);
$value = Util::removeHtmlComments($value);
$parseResultArray = $this->parser->parse($value);
if (isset($parseResultArray)) {
foreach ($parseResultArray as $parseResults) {
$parsedDataType = $parseResults[1];
if ($parsedDataType == "") {
$parsedDataType = null;
}
if ($parseResults[0] != "") {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal((string) $parseResults[0], $parsedDataType, null));
}
}
} else {
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::Literal($value));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
$foundCoordinates = array();
/* Main title */
if ($geoInfo = $this->extractGeoInfo($pageSource, true)) {
$pageG = new geo_param($geoInfo);
array_push($foundCoordinates, $pageG);
$this->log('debug', "Found title entry '" . implode("|", $geoInfo) . "'");
}
/* Coordinates provided in infobox formats */
$infoboxes = $this->getInfoboxes($pageSource);
foreach ($infoboxes[1] as $box) {
$boxProperties = $this->getBoxProperties($box, true);
foreach (GeoExtractor::$knownInfoboxFormats as $format) {
/* Initialize global defaults */
$pieces = array(null, 0, 0, 'N', null, 0, 0, 'E');
/* Apply template-specific NS/EW defaults */
if (isset($format[8])) {
/* NS default */
$pieces[3] = $format[8];
}
if (isset($format[9])) {
/* EW default */
$pieces[7] = $format[9];
}
/* Copy from template */
for ($i = 0; $i < count($format); $i++) {
$formatString = urldecode($format[$i]);
if ($formatString != "" && isset($boxProperties[strtolower($formatString)])) {
/* German coordinates: Treat 'O' (Ost) as 'E' (East) */
if ($i == 7 && $boxProperties[strtolower($formatString)] == 'O') {
$pieces[$i] = 'E';
} else {
$pieces[$i] = $boxProperties[strtolower($formatString)];
}
}
}
if (geo_param::is_lat($pieces[0]) && geo_param::is_long($pieces[4])) {
$g = new geo_param($pieces);
array_push($foundCoordinates, $g);
$this->log('debug', "Found format '" . $format[0] . "'");
}
}
/* $this->knownInfoboxFormats */
/*
* Look for coordinate tags inside the infobox
* Used widely, e.g. airports (=> Los_Angeles_International_Airport)
* These don't have to set the title, as they're in a first-level infobox
* We could simply take all references, but the idea here is to filter out
* cases where multiple locations are specified
*/
foreach (GeoExtractor::$knownTemplatePredicates as $coord) {
if (isset($boxProperties[strtolower(urldecode($coord))]) && ($geoInfo = $this->extractGeoInfo($boxProperties[strtolower(urldecode($coord))]))) {
$pageG = new geo_param($geoInfo);
array_push($foundCoordinates, $pageG);
$this->log('debug', "Found infobox entry '" . implode("|", $geoInfo) . "'");
}
}
}
/* infoboxes */
$numResults = 0;
foreach ($foundCoordinates as $g) {
if ($g->is_valid()) {
if (GeoExtractor::enablePreview) {
?>
<iframe src="http://maps.google.com/?q=<?php
echo $g->latdeg;
?>
,<?php
echo $g->londeg;
?>
&z=5" style="width: 1000px; height: 600px; border: none;" scrolling="no">
</iframe>
<?php
}
/* Only process the first result */
if (++$numResults == 1) {
$sqlParams = array("resource" => $pageID, "lang" => $this->language, "point" => (string) $g->latdeg . " " . (string) $g->londeg);
/* Check whether this exact entry is present */
if ($this->batchExtraction) {
$results = mysql_query($this->sqlParameterize(GeoExtractor::sqlCheckExact, $sqlParams), $this->dbSharedConnection);
$row = mysql_fetch_assoc($results);
$hasExact = $row['count'] != '0';
if ($hasExact) {
$hasAny = true;
} else {
/* Check whether any entry is present */
$results = mysql_query($this->sqlParameterize(GeoExtractor::sqlCheckAny, $sqlParams), $this->dbSharedConnection);
$row = mysql_fetch_assoc($results);
$hasAny = $row['count'] != '0';
}
} else {
$hasExact = $hasAny = false;
}
if ($hasAny) {
$this->log('debug', "Not generating geocoordinates because coordinates were previously generated for this resource");
} else {
/* Store in results table for duplicate detection */
if ($this->batchExtraction) {
mysql_query($this->sqlParameterize(GeoExtractor::sqlInsertPoint, $sqlParams), $this->dbSharedConnection);
}
/* Triple generation */
/* W3C Geospatial Vocabulary (GeoRSS) */
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEORSS_POINT, false), RDFtriple::Literal((string) $g->latdeg . " " . (string) $g->londeg));
/* Basic Geo Vocabulary - only add them if there are no points for the resource so far; it'd be ambigous otherwise! */
if ($hasAny) {
$this->log('debug', "Not generating W3C Basic geocoordinates because basic coordinates were previously generated for this resource");
} else {
$result->addTriple($this->getPageURI(), RDFtriple::URI(WGS_LAT, false), RDFtriple::Literal((string) $g->latdeg, XS_FLOAT, NULL));
$result->addTriple($this->getPageURI(), RDFtriple::URI(WGS_LONG, false), RDFtriple::Literal((string) $g->londeg, XS_FLOAT, NULL));
}
/*
* Process additional attributes
* See http://en.wikipedia.org/wiki/Wikipedia:WikiProject_Geographical_coordinates#Parameters
* and http://de.wikipedia.org/wiki/Wikipedia:WikiProjekt_Georeferenzierung/Neue_Koordinatenvorlage#Parameter
*/
if (isset($g->attr['type'])) {
if (isset(GeoExtractor::$typeToGeoNames[$g->attr['type']])) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEO_FEATURECLASS, false), RDFtriple::URI(GEONAMES_NS . GeoExtractor::$typeToGeoNames[$g->attr['type']][0]));
if (isset(GeoExtractor::$typeToGeoNames[$g->attr['type']][1])) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEO_FEATURECODE, false), RDFtriple::URI(GEONAMES_NS . GeoExtractor::$typeToGeoNames[$g->attr['type']][0] . '.' . GeoExtractor::$typeToGeoNames[$g->attr['type']][1]));
}
}
/* city(pop): City, town or village with specified population */
if (strtolower($g->attr['type'] == 'city') && isset($g->attr['arg:type'])) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEO_POPULATION, false), RDFtriple::Literal((string) $g->attr['arg:type'], XS_INTEGER, NULL));
}
/*
* landmark: Cultural landmark, building of special interest, tourist attration and other points of interest
* Mapped to YAGO; could in theory use http://www.eionet.europa.eu/gemet/concept/8525 ("tourist facility"),
* but it seems as if any POI not matching the above categories is tagged as a landmark
* (=> "Google" etc.)
*/
if (strtolower($g->attr['type'] == 'landmark')) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(RDF_TYPE, false), RDFtriple::URI(YAGO_LANDMARK, false));
}
}
/* type */
/* population */
if (isset($g->attr['pop'])) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEO_POPULATION, false), RDFtriple::Literal((string) $g->attr['pop'], XS_INTEGER, NULL));
}
/* elevation in meters above sea level
* TODO must be converted from sea level *to* wgs84 ellipsoid!
*/
/*if (isset($g->attr['elevation'])) {
$result->addTriple(
$this->getPageURI(),
RDFtriple::URI("http://www.georss.org/georss/elev"),
RDFtriple::Literal((string) $g->attr['elevation']);
}*/
/* Diameter (m) to GeoRSS radius (m) */
if (isset($g->attr['dim'])) {
$result->addTriple($this->getPageURI(), RDFtriple::URI(GEORSS_RADIUS, false), RDFtriple::Literal((string) ($g->attr['dim'] / 2), XS_DECIMAL, NULL));
}
/*
* region: ISO 3166-1 alpha-2 country code or ISO 3166-2 region code to GeoNames inCountry
* Very unreliable, as it just sets a preferred map view,
* i.e. Germany's and The Czech Republic's regions are set to "EN"....
*/
/*if (isset($g->attr['region']))
{
$result->addTriple(
$this->getPageURI(),
RDFtriple::URI("http://www.geonames.org/ontology#inCountry"),
RDFtriple::URI("http://www.geonames.org/countries/#".
substr($g->attr['region'], 0, 2)));
} */
}
/* !$hasExact */
}
/* first entry */
}
/* is_valid */
}
/* $foundCoordinates */
return $result;
}
/**
* Write given triple to given destination.
* @param $destination must not be null
* @param $pageID will be turned into a RDF URI using RDFtriple::page()
* @param $predicate must be a RDF URI
* @param $text will be turned into a RDF literal using RDFtriple::Literal()
* @return void
*/
private function writeTriple($destination, $pageID, $predicate, $text)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
$subject = RDFtriple::page($pageID);
$object = RDFtriple::Literal($text, NULL, $this->language);
// $this->log('warn','Found: '.$subject->toString()." ".$predicate->toString()." ".$object->toString());
$result->addTriple($subject, $predicate, $object);
Timer::start('destination:accept');
$destination->accept($result);
Timer::stop('destination:accept');
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, $this->getExtractorID());
$PersonData = $this->extractPersondata($pageSource, $this->language);
//var_dump($PersonData);
if ($PersonData != null) {
// preg_match("/\[\[en:(.*)\]\]/", $pageSource, $LangLinkmatch);
// $PersonData['enPageID'] = str_replace(" ","_",$LangLinkmatch[1]);
if (Options::getOption('Persondata.usedb')) {
$WikiDB = new DatabaseWikipediaCollection($this->language);
}
$mysource = "";
if (isset($PersonData['birthplace'])) {
preg_match("/\\[\\[([^\\]]*)\\]\\]/", $PersonData['birthplace'], $Birthplacematch);
if (isset($Birthplacematch[0])) {
$Birthplacematch = $this->getLinkForLabeledLink($Birthplacematch);
if (Options::getOption('Persondata.usedb')) {
$mySource = $WikiDB->getSource($Birthplacematch);
}
preg_match("/\\[\\[en:(.*)\\]\\]/", $mySource, $LangLinkmatch);
if (isset($LangLinkmatch[1])) {
$BirthPlace = $LangLinkmatch[1];
}
}
}
if (isset($PersonData['deathplace'])) {
preg_match("/\\[\\[([^\\]]*)\\]\\]/", $PersonData['deathplace'], $Deathplacematch);
if (isset($Deathplacematch[0])) {
$Deathplacematch = $this->getLinkForLabeledLink($Deathplacematch);
if (Options::getOption('Persondata.usedb')) {
$mySource = $WikiDB->getSource($Deathplacematch);
}
preg_match("/\\[\\[en:(.*)\\]\\]/", $mySource, $LangLinkmatch);
if (isset($LangLinkmatch[1])) {
$DeathPlace = $LangLinkmatch[1];
}
}
}
//var_dump($PersonData);
//var_dump($BirthPlace);
//var_dump($DeathPlace);
//var_dump($Deathplacematch);
if (isset($PersonData['name']) && $PersonData['name'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(FOAF_NAME, false), RDFtriple::Literal($PersonData['name'], null, "de"));
}
if (isset($PersonData['givenname']) && $PersonData['givenname'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(FOAF_GIVENNAME, false), RDFtriple::Literal($PersonData['givenname'], null, "de"));
}
if (isset($PersonData['surname']) && $PersonData['surname'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(FOAF_SURNAME, false), RDFtriple::Literal($PersonData['surname'], null, "de"));
}
if (isset($BirthPlace) && $BirthPlace != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(DB_BIRTHPLACE, false), RDFtriple::page($BirthPlace));
// $result->addTriple(
// RDFtriple::page($pageID),
// RDFtriple::URI("http://purl.org/vocab/bio/0.1/event"),
// RDFtriple::URI("http://dbp edia.org/resource/" . URI::wikipediaEncode($pageID) . "/Birth"));
// $result->addTriple(
// RDFtriple::URI("http://dbp edia.org/resource/" . URI::wikipediaEncode($pageID) . "/Birth");
// RDFtriple::URI("http://purl.org/vocab/bio/0.1/place"),
// RDFtriple::page($BirthPlace));
}
if (isset($PersonData['birthdate']) && $PersonData['birthdate'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(DB_BIRTH, false), RDFtriple::Literal($PersonData['birthdate'], XS_DATE, null));
}
if (isset($DeathPlace) && $DeathPlace != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(DB_DEATHPLACE, false), RDFtriple::page($DeathPlace));
}
if (isset($PersonData['deathdate']) && $PersonData['deathdate'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(DB_DEATH, false), RDFtriple::Literal($PersonData['deathdate'], XS_DATE, null));
}
if (isset($PersonData['description']) && $PersonData['description'] != "") {
$result->addTriple($this->getPageURI(), RDFtriple::URI(DC_DESCRIPTION, false), RDFtriple::Literal($PersonData['description'], null, "de"));
}
$result->addTriple($this->getPageURI(), RDFtriple::URI(RDF_TYPE, false), RDFtriple::URI(FOAF_PERSON, false));
}
return $result;
}
private function addLiteral($result, $pageID, $class, $property, $value, $datatype = null, $lang = null)
{
$result->addTriple(RDFtriple::page($pageID), RDFtriple::property($class, $property, $this->flagNewSchema), RDFtriple::Literal($value, $datatype, $lang));
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2000/01/rdf-schema#label"), RDFtriple::Literal($this->decode_title($pageTitle), NULL, $this->language));
return $result;
}
public function addDCModifiedAnnotation()
{
$p = RDFtriple::URI(DC_MODIFIED, false);
$o = RDFtriple::Literal(date('c'), XS_DATETIME, "");
$this->addOWLAxiomAnnotation($p, $o);
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$foundCoordinates = array();
/* Main title */
if ($geoInfo = $this->extractGeoInfo($pageSource, true)) {
$pageG = new geo_param($geoInfo);
array_push($foundCoordinates, $pageG);
if (GeoExtractor::enableDebug) {
echo "<h3>Found title entry '" . implode("|", $geoInfo) . "'</h3>";
}
}
/* Coordinates provided in infobox formats */
$infoboxes = $this->getInfoboxes($pageSource);
foreach ($infoboxes[1] as $box) {
$boxProperties = $this->getBoxProperties($box, true);
foreach ($this->knownInfoboxFormats as $format) {
/* Initialize global defaults */
$pieces = array(null, 0, 0, 'N', null, 0, 0, 'E');
/* Apply template-specific NS/EW defaults */
if (isset($format[8])) {
/* NS default */
$pieces[3] = $format[8];
}
if (isset($format[9])) {
/* EW default */
$pieces[7] = $format[9];
}
/* Copy from template */
for ($i = 0; $i < count($format); $i++) {
$formatString = urldecode($format[$i]);
if ($formatString != "" && isset($boxProperties[strtolower($formatString)])) {
/* German coordinates: Treat 'O' (Ost) as 'E' (East) */
if ($i == 7 && $boxProperties[strtolower($formatString)] == 'O') {
$pieces[$i] = 'E';
} else {
$pieces[$i] = $boxProperties[strtolower($formatString)];
}
}
}
if (geo_param::is_lat($pieces[0]) && geo_param::is_long($pieces[4])) {
$g = new geo_param($pieces);
array_push($foundCoordinates, $g);
if (GeoExtractor::enableDebug) {
echo "<h3>Found format '" . $format[0] . "'</h3>";
}
}
}
/* $this->knownInfoboxFormats */
/*
* Look for coordinate tags inside the infobox
* Used widely, e.g. airports (=> Los_Angeles_International_Airport)
* These don't have to set the title, as they're in a first-level infobox
* We could simply take all references, but the idea here is to filter out
* cases where multiple locations are specified
*/
foreach ($this->knownTemplatePredicates as $coord) {
if (isset($boxProperties[strtolower(urldecode($coord))]) && ($geoInfo = $this->extractGeoInfo($boxProperties[strtolower(urldecode($coord))]))) {
$pageG = new geo_param($geoInfo);
array_push($foundCoordinates, $pageG);
if (GeoExtractor::enableDebug) {
echo "<h3>Found infobox entry '" . implode("|", $geoInfo) . "'</h3>";
}
}
}
}
/* infoboxes */
$numResults = 0;
foreach ($foundCoordinates as $g) {
if ($g->is_valid()) {
if (GeoExtractor::enablePreview) {
?>
<iframe src="http://christianhbecker.com/da/workspace/gmap.php?lat=<?php
echo $g->latdeg;
?>
&long=<?php
echo $g->londeg;
?>
&scale=5" style="width: 500px; height: 300px; border: none;" scrolling="no">
</iframe>
<?php
}
/* Only process the first result */
if (++$numResults == 1) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2003/01/geo/wgs84_pos#lat"), RDFtriple::Literal((string) $g->latdeg, "http://www.w3.org/2001/XMLSchema#float", NULL));
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2003/01/geo/wgs84_pos#long"), RDFtriple::Literal((string) $g->londeg, "http://www.w3.org/2001/XMLSchema#float", NULL));
/*
* Process additional attributes
* See http://en.wikipedia.org/wiki/Wikipedia:WikiProject_Geographical_coordinates#Parameters
*/
if (isset($g->attr['type'])) {
if (isset($this->typeToGeoNames[$g->attr['type']])) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.geonames.org/ontology#featureClass"), RDFtriple::URI("http://www.geonames.org/ontology#" . $this->typeToGeoNames[$g->attr['type']][0]));
if (isset($this->typeToGeoNames[$g->attr['type']][1])) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.geonames.org/ontology#featureCode"), RDFtriple::URI("http://www.geonames.org/ontology#" . $this->typeToGeoNames[$g->attr['type']][0] . '.' . $this->typeToGeoNames[$g->attr['type']][1]));
}
}
/* city(pop): City, town or village with specified population */
if (strtolower($g->attr['type'] == 'city') && isset($g->attr['arg:type'])) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.geonames.org/ontology#population"), RDFtriple::Literal($g->attr['arg:type'], "http://www.w3.org/2001/XMLSchema#integer", NULL));
}
/*
* landmark: Cultural landmark, building of special interest, tourist attration and other points of interest
* Mapped to YAGO; could in theory use http://www.eionet.europa.eu/gemet/concept/8525 ("tourist facility"),
* but it seems as if any POI not matching the above categories is tagged as a landmark
* (=> "Google" etc.)
*/
if (strtolower($g->attr['type'] == 'landmark')) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), RDFtriple::URI("http://dbpedia.org/class/yago/Landmark108624891"));
}
}
/* type */
/*
* region: ISO 3166-1 alpha-2 country code or ISO 3166-2 region code to GeoNames inCountry
* Very unreliable, as it just sets a preferred map view,
* i.e. Germany's and The Czech Republic's regions are set to "EN"....
*/
/*if (isset($g->attr['region']))
{
$result->addTriple(
RDFtriple::page($pageID),
RDFtriple::URI("http://www.geonames.org/ontology#inCountry"),
RDFtriple::URI("http://www.geonames.org/countries/#".
substr($g->attr['region'], 0, 2)));
} */
}
}
/* is_valid */
}
/* $foundCoordinates */
return $result;
}
/**
* A wrapper for parseAttributeValue
*
*
* @global <type> $parseResult
* @param <type> $value
* @param <type> $templateChildName
* @param <type> $propertyName
* @param <type> $language
* @return <type>
*/
function parseAttributeValueWrapper($value, $templateChildName, $propertyName, $language)
{
$result = array();
global $parseResult;
$parseResult = null;
$localResult = parseAttributeValue($value, $templateChildName, $propertyName, $language);
$items = array();
// remap local and global results into a uniform schema
if (isset($parseResult)) {
foreach ($parseResult as $item) {
list(, , $o, $ot, $dt, $ol) = $item;
$items[] = array($o, $ot, $dt, $ol);
}
}
$parseResult = null;
if (isset($localResult)) {
list($o, $ot, $dt, $ol) = $localResult;
$items[] = array($o, $ot, $dt, $ol);
}
foreach ($items as $item) {
$object = $item[0];
$objectType = $item[1];
$dataType = $item[2];
$objectLanguage = $item[3];
// And another hack... we pass the language to the
// parse function, and we don't get it back...
if (!isset($objectLanguage)) {
$objectLanguage = $language;
}
// special newline handling
$br = array('<br>', '<br/>', '<br />');
if ($objectType == 'l') {
$object = str_replace($br, "\n", $object);
} else {
if ($objectType == 'r') {
$object = str_replace($br, '', $object);
}
}
//echo "Got object type '$objectType'\n";
if ($objectType == "r") {
$object = RDFtriple::URI($object);
} else {
if ($objectType == "l") {
$object = RDFtriple::Literal($object, $dataType, $objectLanguage);
} else {
Logger::warn("Shouldn't happen - found a blank node where none expected - objectType = {$objectType}");
continue;
}
}
$result[] = new RDFtriple(RDFtriple::page($templateChildName), RDFtriple::URI($propertyName), $object);
}
return $result;
}