public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$pageID = encodeLocalName($pageID);
// Remove Template as this is already extracted by the Infobox Extractor
// Find subtemplates and remove Subtemplates, which are listed as ignored!
preg_match_all('~\\{((?>[^{}]+)|(?R))*\\}~x', $pageSource, $subTemplates);
foreach ($subTemplates[0] as $key => $subTemplate) {
$subTemplate = preg_replace("/(^\\{\\{)|(\\}\\}\$)/", "", $subTemplate);
// Cut Brackets / {}
$pageSource = str_replace('{{' . $subTemplate . '}}', '', $pageSource);
}
// Extract internal Semantic Links
$findSemanticLinks = preg_match_all('/(\\[\\[)([a-zA-z0-9\\- _]+)(::)([^\\]]+)\\]\\]/', $pageSource, $matches, PREG_SET_ORDER);
foreach ($matches as $match) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::predicate(encodeLocalName($match[2])), RDFtriple::page($match[4]));
}
// Extract Literals
$findSemanticLinks = preg_match_all('/(\\[\\[)([a-zA-Z\\-_ ]+)(:=)([^\\]]+)\\]\\]/', $pageSource, $matches, PREG_SET_ORDER);
foreach ($matches as $match) {
$triple = array();
$triple = parseAttributeValue($match[4], $pageID, $match[2]);
// object, object_is, datatype(, language)
$lexicalForm = $triple[0];
$datatype = $triple[2];
$predicate = propertyToCamelCase(encodeLocalName($match[2]));
// Continue if empty String
if ($lexicalForm == null) {
continue;
}
$result->addTriple(RDFtriple::page($pageID), RDFTriple::predicate($predicate), RDFtriple::literal($lexicalForm, $datatype, 'en'));
}
return $result;
}
public function extractPage($pageID, $pageTitle, $pageSource)
{
$result = new ExtractionResult($pageID, $this->language, self::extractorID);
$pageID = encodeLocalName($pageID);
// Extract Wikipedia Link
if (preg_match('/\\{\\{wikipedia\\-c(\\-note)?\\}\\}/', $pageSource)) {
$result->addTriple(RDFtriple::page($pageID), RDFtriple::URI("http://www.w3.org/2002/07/owl#sameAs"), RDFtriple::URI("http://dbpedia.org/resource/" . $pageID));
}
return $result;
}
public function generate($subjectName, $propertyName, $value)
{
$result = array();
$links = $this->parser->parse($value);
foreach ($links as $link) {
$link = $this->mediaWikiUtil->toCanonicalWikiCase($link);
$link = encodeLocalName($link);
$resource = $this->basePath . $link;
$result[] = new RDFtriple(RDFtriple::page($subjectName), RDFtriple::URI(DB_ONTOLOGY_NS . $propertyName), RDFtriple::URI($resource));
}
return $result;
}
/**
* Writes the triple + additional information such as language, whether an object is a reference
* or a literal and the datatype into a global array ($parseResult)
*
* @param subject: String containing the triples subject
* @param predicate: String containing the triples predicate
* @param object: String containing the triples object
* @param file: Legacy, should be removed in the future
* @param object_is: 'r' if object is a reference, 'l' if object is a literal, 'b' if object is a blanknode
* @param dtype: String containing a literals XS D:datatype
* @param lang: String containing a literals language
*
* TODO: Should encodeLocalName be used for the whole URL? Should URI objects be used?
*
*/
function writeTripel($subject, $predicate, $object, $file = 'main', $object_is = 'r', $dtype = NULL, $lang = NULL)
{
global $parseResult;
if ($object_is == 'r' && !URI::validate(encodeLocalName($object))) {
return null;
}
// If $object_is == 'l', encodeLocalName shouldn't be used, the string will be encoded like e.g. \uBC18\uC57C
if ($object_is != 'l') {
$object = encodeLocalName($object);
}
$predicate = encodeLocalName($predicate);
if (USE_PERCENT_ENCODING) {
$predicate = str_replace("%", "_percent_", $predicate);
} else {
if (ereg("%([A-F0-9]{2})", substr($predicate, -3))) {
$predicate .= "_";
}
}
$parseResult[] = array(encodeLocalName($subject), $predicate, $object, $object_is, $dtype, $lang);
}
/**
* This function returns two results:
* [triples: the generated triples, metaTriples: meta triples 2: the used templates]
*
* Unfortunately I haven't renamed the variables yet - result
* is the array containing the result TRIPLES!! - its not this
* 2 element.
*
*/
private function myGenerate(BreadCrumb $breadCrumb, $value)
{
// result is the array containing: triples, meta triples, used templates
$result = array(array(), array(), array());
$rootSubjectUri = RDFTriple::page($breadCrumb->getRoot());
// this array is only relevant on depth 0
$relatedClasses = array();
//$metaTriples = array();
//$usedTemplateNames = array();
// 'parent' means the parent of the value - thus subject and predicate
$parentName = $this->breadCrumbTransformer->transform($breadCrumb);
$parentResource = RDFTriple::page($parentName);
$parentPropertyName = null;
$tmp = $breadCrumb->peekTop(0);
if (isset($tmp)) {
$parentPropertyName = $tmp->getPropertyName();
}
// Get all templates on this site, indexed by name
// (there may be multiple templates with the same name)
$nameToTemplates = SimpleWikiTemplateMatcher::match($value);
//print_r($value);
//print_r($nameToTemplates);
//echo "NOW COMES THE STORM\n";
foreach ($nameToTemplates as $templateName => $templates) {
if (strlen($templateName) < 1) {
continue;
}
//echo "GOT TEMPLATE NAME $templateName\n";
$templateName = $this->mediaWikiUtil->toCanonicalWikiCase($templateName);
if (!$this->templateNameFilter->doesAccept($templateName)) {
continue;
}
$templateUri = RDFTriple::URI(DB_TEMPLATE_NS . $templateName, false);
$result[2][$templateName] = 1;
// Get annotations for the template - if there are any
$lookupName = "Template:{$templateName}/doc";
if ($breadCrumb->getDepth() == 0) {
$ta = $this->templateDb->getTemplateAnnotation($lookupName);
// Create the triples for "relatesToClass"
// But only for the page itself (not for sub templates)
// if no related class exists, default to rdf:type owl:Thing
if (isset($ta)) {
foreach ($ta->getRelatedClasses() as $item) {
$relatedClasses[$item] = 1;
}
}
}
foreach ($templates as $templateIndex => $template) {
//echo "GOT TEMPLATE INDEX $templateIndex\n";
// Iterate over all arguments
$arguments = $template->getArguments();
foreach ($arguments as $argumentName => $values) {
//echo "GOT ARGUMENT NAME $argumentName\n";
// propertyNs defaults to DB_PROPERTY_NS unless there
// exists a mapping in the templatedb. In that case it will
// be set to DB_ONTOLOGY_NS
$propertyNs = DB_PROPERTY_NS;
$pa = null;
if (isset($ta)) {
$pas = $ta->getPropertyAnnotations();
if (array_key_exists($argumentName, $pas)) {
$pa = $pas[$argumentName];
$propertyNs = DB_ONTOLOGY_NS;
}
}
//print_r($ta);
//echo "PROPERTY NS : $lookupName - $argumentName = $propertyNs\n";
// Fake a property mapping if there was none in the db
// This maps argumentName back to iteself
if (!isset($pa)) {
// If there was no mapping we might ignore it
// depending on an option (We can prevent this extractor
// to generate triples with properties in the
// dbp:property namespace
// We allow such triples on subResources though.
if ($this->allowUnmappedProperties != true && $breadCrumb->getDepth() == 0) {
continue;
}
// If there was no mapping, also rename numeric
// argument names (e.g. 1 becomes property1)
// this is just cosmetic for the result
if (is_numeric($argumentName)) {
$argumentName = "property{$argumentName}";
}
$pa = new PropertyAnnotation($argumentName);
$pa->addMapping(new PropertyMapping($argumentName));
}
foreach ($pa->getMappings() as $pm) {
$parseHint = $pm->getParseHint();
//echo "Mapping $argumentName : {$pm->getRenamedValue()}\n\n";
// if the renamed value is not set, use the original
// name
// otherwise use the mapped value
if (!isEmptyString($pm->getRenamedValue())) {
$argumentName = $pm->getRenamedValue();
}
$argumentName = trim($argumentName);
//echo "Mapping $argumentName : {$pm->getRenamedValue()}\n\n";
// Skip empty properties
// FIXME does that even happen?
if (strlen($argumentName) < 1) {
continue;
}
//echo "TN = $templateName, AN = $argumentName\n";
$childBreadcrumb = $breadCrumb->createClone();
$childBreadcrumb->push(new BreadcrumbNode($templateName, $templateIndex, $argumentName));
//$templateChildName = $this->breadcrumbToSubject($childBreadcrumb);
$templateChildName = $this->breadCrumbTransformer->transform($childBreadcrumb);
// If there is no parse hint we might be able to derive it
if (!isset($parseHint)) {
$parseHint = $this->deriveParseHintFromName($argumentName);
}
// Attempt to obtain a triple generator
$tripleGenerator = $this->getTripleGenerator($parseHint);
// If we DONT have a triple generator
// we fall through to default handling
$localResult = array(array(), array(), array());
if (isset($tripleGenerator)) {
foreach ($values as $valueIndex => $value) {
//echo "GOT VALUE $value\n";
$value = trim($value);
// Skip empty values
if ($value == "") {
continue;
}
//echo "PROCESSING $templateChildName - $argumentName $value\n";
$tmp = $tripleGenerator->generate($templateChildName, $argumentName, $value);
$localResult[0] = array_merge($localResult[0], $tmp);
//echo "LOCALRESULT\n";
//print_r($localResult[0]);
//print_r($triples);
//echo "\nSigh\n";
//if(isset($triples))
// $result = array_merge($result, $triples);
}
// append the generated triples
//continue;
} else {
// No parse hint - default handling
// if property date and object an timespan
// we extract it with following special case
$argumentName = propertyToCamelCase($argumentName);
$argumentName = encodeLocalName($argumentName);
if (in_array($argumentName, $GLOBALS['W2RCFG']['ignoreProperties'])) {
continue;
}
// turn the argument name into a property name
$propertyName = $propertyNs . $argumentName;
foreach ($values as $valueIndex => $value) {
$value = trim($value);
// Skip empty values
if ($value == "") {
continue;
}
if ($argumentName == "date") {
$value = str_replace("[", "", $value);
$value = str_replace("]", "", $value);
$value = str_replace("–", "-", $value);
}
// Parse out sub templates
// if something was extracted:
// .) connect subject with subsubject
// .) indicate usage at wikipage
$subResources = $this->myGenerate($childBreadcrumb, $value);
for ($i = 0; $i < 3; ++$i) {
$localResult[$i] = array_merge($localResult[$i], $subResources[$i]);
}
//$result = array_merge($result, $triples);
//echo "GOT OBJECT $value\n";
$localResult[0] = array_merge($localResult[0], parseAttributeValueWrapper($value, $templateChildName, $propertyName, $this->language));
//$result = array_merge($result, $triples);
}
}
// For each triple add the ExtractedFromTemplate-Annotation
// Exclude triples with wikiPageUsesTemplate as predicate though
foreach ($localResult[0] as $triple) {
$triple->addExtractedFromTemplateAnnotation($templateUri);
}
// Add on delete cascade annotation
if ($breadCrumb->getDepth() > 1) {
foreach ($localResult[0] as $triple) {
$triple->addOnDeleteCascadeAnnotation($rootSubjectUri);
}
}
// merge the results
//for($i = 0; $i < 3; ++$i)
// $result[$i] = array_merge($result[$i], $localResult[$i]);
//}
for ($i = 0; $i < 3; ++$i) {
$result[$i] = array_merge($result[$i], $localResult[$i]);
}
}
}
/*
How to connect a sub-subject to the root subject?
if($breadCrumb->getDepth() == 0)
continue;
// Create the parent-child connection
$parentChildTriple = new RDFtriple(
$parentResource,
RDFtriple::URI(DB_PROPERTY_NS . encodeLocalName($parentPropertyName), false),
RDFtriple::page($templateChildName));
//$result[1][] = $parentChildTriple;
*/
}
}
if (count($relatedClasses) > 0) {
foreach ($relatedClasses as $relatedClass => $dummy) {
$result[1][] = new RDFtriple($parentResource, RDFtriple::URI(RDF_TYPE, false), RDFtriple::URI(DB_ONTOLOGY_NS . $relatedClass, false));
}
} else {
if ($breadCrumb->getDepth() == 0) {
$result[1][] = new RDFtriple($parentResource, RDFtriple::URI(RDF_TYPE, false), RDFtriple::URI(OWL_THING, false));
}
}
// Add the wiki page uses template triples - but only on depth 0
if ($breadCrumb->getDepth() == 0) {
foreach ($result[2] as $name => $dummy) {
$result[1][] = new RDFTriple($parentResource, self::$wikiPageUsesTemplateUri, RDFTriple::URI(DB_TEMPLATE_NS . $name, false));
}
}
$n = count($result[0]) + count($result[1]);
$this->log(TRACE, "Generated a total of {$n} triples at {$breadCrumb}");
foreach ($result[0] as $item) {
$this->log(TRACE, $item);
}
foreach ($result[1] as $item) {
$this->log(TRACE, $item);
}
return $result;
}
/**
* Parses internal Links:
* - If a Link is found: links to currencies are replaced with the respective symbol, external links are removed
* (these are usually references), links to dates are removed (if more than one link was found).
* - If only digits and currencies are at the beginning of the String, anything else is removed and the number
* is parsed for it's type (int, float, unit)
* - In any other cases, where internal links are mixed with text, the function compares the aggregated word-length
* of the links, with the length of text items. If the links are longer, the String is parsed as a link list, else
* the brackets are removed and the String is recognized as text.
*
*
*
*/
function catchLinkList(&$o, $s, $p, &$dtype)
{
// Match for any Link
$foundLink = preg_match_all("/(\\[{2})([^\\]]+)(\\]{2})/", $o, $matches);
if (!$foundLink) {
return false;
}
// Initialize object-type with literal
$object_is = 'l';
// echo "\n$o";
// Test whether property is included in known Linklists and parse Links
$knownLinkLists = $GLOBALS['linklistpredicates'];
// Remove DBpedia Base URI
$propertyName = substr($p, strlen($GLOBALS['W2RCFG']['propertyBase']), strlen($p));
// Compare property-name with known LinkList properties
foreach ($knownLinkLists as $linkList) {
if ($linkList == $propertyName) {
preg_match_all("/(\\[{2})([^\\]]+)(\\]{2})/", $o, $matches);
foreach ($matches[2] as $l) {
if (strlen($l) > 1) {
// Extract internal links of type [[abc|def]]
$pos = stripos($l, "|");
if ($pos) {
$l = substr($l, 0, $pos);
}
$object = $GLOBALS['W2RCFG']['wikipediaBase'] . ucwords(encodeLocalName($l));
$object_is = 'r';
writeTripel($s, $p, $object, 'main', $object_is);
unset($object);
}
}
return true;
}
}
// $weight: If text is mixed with length, this is the weight assigned to the links
// in order to decide whether the composite link/text String is parsed as link-list or text-litearal
// any value > 1, gives more weight to links, any value between 0 and 1, morr to the text part
$weight = 1.25;
// If an internal Link was found:
// Replace Links to currencies with the respective Symbol
$currencies = array("U.S. (D|d)ollar" => "\$", "United States (D|d)ollar" => "\$", "Dollar" => "\$", "Euro" => "€", "Yen" => "¥", "Pound" => "£");
// $z = str_replace('$','\$',$o);
foreach ($currencies as $key => $currency) {
// Do not match real Links to currencies e.g. United_States: currency = [[United States Dollar]] ($)
if (preg_match('/^\\s*\\[{2}' . $key . '\\s?\\|?[^\\]]*\\]{2}[\\(\\s ]*' . $currency . '[\\)\\s]*$/', $o)) {
break;
}
$o = trim(preg_replace('/(^[^' . $currency . ']*)(\\[{2}' . $key . '\\s?\\|?[^\\]]*\\]{2})/', '\\1' . $currency, $o));
// Old Version
// $o = trim(preg_replace("/\[{2}".$key."\s?\|?[^\]]*\]{2}(^$)/",$currency,$o));
}
// Remove External Links (these are usually references)
$o = trim(preg_replace("/\\[http:\\/\\/[^\\]]+\\]/", "", $o));
// Remove any Links between parentheses
// Remove links in parentheses. Bug: Destroys Links with "()" inside an internal Link. e.g. Boris_Becker: birthplace
$o = trim(preg_replace("/\\([^\\[\\]]*\\[{2}[^\\)\\]]*\\]{2}[^\\)]*\\)/", "", $o));
// If Link is a Date and more than one Link was found, remove Link
if ($foundLink > 1) {
$months = array("January", "February", "March", "April", "May", "June", "July", "August", "September", "November", "December");
foreach ($months as $month) {
// $o = trim(preg_replace("/\[{2}$month [0-9]{1,2}\]{2},?[\s]*,?/","",$o));
$o = trim(preg_replace("/\\[{2}{$month} [0-9]{1,2}\\]{2},?[\\s]*(,?[\\s]*(\\(?\\[{2}|\\()[0-9]{4}(\\]{2}\\)?|\\)))?/", "", $o));
}
// If Link is a year, remove Link
$o = trim(preg_replace("/\\(?\\[{1,2}[0-9]{4}\\]{1,2}\\)?/", "", $o));
}
// Initialize ResultString
$resultstring = "";
// String begins with Text and is followed by one or more Links
if (preg_match("/^([^\\[]+)(\\[{2})*/", $o, $stringStart)) {
// String ends with a Link -> this means String is like: "abc [[def | jjj ]] ghi [[ xyz ]]" (Problem "abc [[def]][[xyz]]")
if (preg_match("/\\]{2}\$/", $o)) {
// Match Text, followed by a Link
$found = preg_match_all("/([^\\[\\]]+)(\\[{2})([^\\]]+)(\\]{2})/", $o, $matches);
$linkPos = 3;
// Position of Links in $matches
$textPos = 1;
// Position of Text in $matches
// String ends with Text -> this means String is like: "abc [[def | jjj ]] ghi [[ xyz ]] klm"
} else {
// Initialize ResultString with "abc "
$resultstring = $stringStart[1];
// Match Link, followed by Text
$found = preg_match_all("/(\\[{2})([^\\]]+)(\\]{2})([^\\[]+)/", $o, $matches);
$linkPos = 2;
// Position of Links in $matches
$textPos = 4;
// Position of Text in $matches
}
// String starts with numbers and/or currency Symbols
if (preg_match("/(^[\\s]*([0-9\$€£¥]+[\\.,][0-9\$€£¥]+|[0-9\$€£¥]+)[\\s]*(((B|b)illion)?|((M|m)illion)?|((T|t)rillion)?|((Q|q)uadrillion)?))(.*)/", $o, $numberMatch) && strlen(trim($numberMatch[1])) > 2 || preg_match("/^([0-9\$€£¥]+[\\.,][0-9\$€£¥]+|[0-9\$€£¥]+)([\\s]*\$)/", $o)) {
// CodeBlock for parsing Numbers
// Remove any remaining Links
// $o = preg_replace("/\[{2}[^\]]+\]{2}/","",$o);
// echo "\n$o";
// Read Links to numbers, e.g., BMW: revenue => € 4.9 [[10000000 (number)| billion]]
if (preg_match("/(^[\\s]*([0-9\$€£¥]+[\\.,][0-9\$€£¥]+|[0-9\$€£¥]+)[\\s]*)(\\[{2}[^\\]\\|]*\\(number\\)[^\\]\\|]*\\|)([^\\]]+)(\\]{2})(.*\$)/", $o, $numberMatch)) {
$o = trim($numberMatch[1]) . " " . trim($numberMatch[4]);
} else {
if ($startPos = strpos($o, "[[")) {
// Remove anything after first Link
$o = substr($o, 0, $startPos);
}
}
// Remove year: e.g. p:revenue = "22 billion $ (2004) => "22 billion $";
if (preg_match("/^[\\s]*[0-9\$€£¥]+[^\\(]+\\([0-9]{4}\\).*/", $o)) {
$o = trim(preg_replace("/\\([0-9]{4}\\)/", "", $o));
}
$o = trim($o);
// echo "\n parsing for values $o";
list($o, $o_is, $dtype, $lang) = parseAttributeValue($o, $s, $p);
if ($o !== NULL) {
writeTripel($s, $p, $o, 'main', $o_is, $dtype, $lang);
}
return true;
} else {
// Calculate aggregate length of text and Links
$lengthLink = 0;
$lengthText = strlen($resultstring);
foreach ($matches[$linkPos] as $match) {
if ($pos = strpos($match, "|")) {
$lengthLink += strlen(preg_replace("/\\s/", "", substr($match, $pos, strlen($match) - $pos)));
} else {
$lengthLink += strlen(preg_replace("/\\s/", "", $match));
}
}
foreach ($matches[$textPos] as $match) {
$lengthText += strlen(preg_replace("/\\s/", "", $match));
}
// compare aggregated length of links and literals ($weight is defined at the top of this function)
if ($weight * $lengthLink >= $lengthText) {
// CodeBlock for Start with Text and more Links than Text
// echo "Start with Text: more Links ($lengthLink,$lengthText)";
foreach ($matches[$linkPos] as $l) {
if (strlen($l) > 1) {
// Extract internal links of type [[abc|def]]
$pos = stripos($l, "|");
if ($pos) {
$l = substr($l, 0, $pos);
}
$object = $GLOBALS['W2RCFG']['wikipediaBase'] . ucwords(encodeLocalName($l));
$object_is = 'r';
writeTripel($s, $p, $object, 'main', $object_is);
unset($object);
}
}
return true;
} else {
// CodeBlock for Start with Text and more Text than Links
// echo "Start with Text: more Text ($lengthLink,$lengthText) ($o)";
// Replace Links with their Labels
$o = preg_replace_callback("/(\\[{2}[^\\|^\\]]+)(\\|)([^\\]]+)(\\]{2})/", 'getLabelForLink', $o);
// Replace simple links with their link-text
$o = preg_replace("/\\[{2}|\\]{2}/", "", $o);
writeTripel($s, $p, $o, 'main', $object_is);
return true;
}
}
// String begins with Links and is followed by Text or Links)
} else {
if (preg_match("/^(\\[{2})([^\\]]+)(\\]{2})/", $o, $stringStart)) {
// String ends with a Link -> this means String is like: "[[def | jjj ]] ghi [[ xyz ]]"
if (preg_match("/\\]{2}\$/", $o)) {
// Initialize ResultString with "[[def | jjj ]]"
$resultstring = $stringStart[2];
// Match Text, followed by a Link
$found = preg_match_all("/([^\\[\\]]+)(\\[{2})([^\\]]+)(\\]{2})/", $o, $matches);
$linkPos = 3;
// Position of Links in $matches
$textPos = 1;
// Position of Text in $matches
// String ends with Text -> this means String is like: "[[def | jjj ]] ghi [[ xyz ]] klm"
} else {
// Match Link, followed by Text
$found = preg_match_all("/(\\[{2})([^\\]]+)(\\]{2})([^\\[]+)/", $o, $matches);
$linkPos = 2;
// Position of Links in $matches
$textPos = 4;
// Position of Text in $matches
}
// String is composed only of Links -> this means String is like: "[[abc]][[def]]"
if (!$found) {
// CodeBlock for returning only Links
$found = preg_match_all("/(\\[{2})([^\\]]+)(\\]{2})/", $o, $matches);
foreach ($matches[2] as $l) {
if (strlen($l) > 1) {
// Extract internal links of type [[abc|def]]
$pos = stripos($l, "|");
if ($pos) {
$l = substr($l, 0, $pos);
}
$object = $GLOBALS['W2RCFG']['wikipediaBase'] . ucwords(encodeLocalName($l));
$object_is = 'r';
writeTripel($s, $p, $object, 'main', $object_is);
unset($object);
}
}
return true;
} else {
// Calculate aggregate length of text and Links
// If String starts and ends with Link, add length of first Link
$lengthLink = strlen($resultstring);
if ($lengthLink > 0) {
// If first Links of type [[abc | def]] only count "def"
if ($pos = strpos($resultstring, "|")) {
$lengthLink += strlen(preg_replace("/\\s/", "", substr($resultstring, $pos, strlen($resultstring) - $pos)));
}
}
$lengthText = 0;
// add length of current link (in $matches) to aggregate length (if link is like [[abc|def]], only def counts)
foreach ($matches[$linkPos] as $match) {
if ($pos = strpos($match, "|")) {
$lengthLink += strlen(preg_replace("/\\s/", "", substr($match, $pos, strlen($match) - $pos)));
} else {
$lengthLink += strlen(preg_replace("/\\s/", "", $match));
}
}
// add length of literals to aggregate text-length
foreach ($matches[$textPos] as $match) {
$lengthText += strlen(preg_replace("/\\s/", "", $match));
}
// compare aggregated length of links and literals ($weight is defined at the top of this function)
if ($weight * $lengthLink >= $lengthText) {
// CodeBlock for Start with Link and more Links than Text
// echo "Start with Link: more Links ($lengthLink,$lengthText) ($o)";
if (strlen($resultstring) > 1) {
array_unshift($matches[$linkPos], $resultstring);
}
foreach ($matches[$linkPos] as $l) {
if (strlen($l) > 1) {
// Extract internal links of type [[abc|def]]
$pos = stripos($l, "|");
if ($pos) {
$l = substr($l, 0, $pos);
}
$object = $GLOBALS['W2RCFG']['wikipediaBase'] . ucwords(encodeLocalName($l));
$object_is = 'r';
writeTripel($s, $p, $object, 'main', $object_is);
unset($object);
}
}
return true;
} else {
// CodeBlock for Start with Link and more Text than Links
// echo "Start with Link: more Text ($lengthLink,$lengthText) ($o)";
// Replace Links with their Labels
$o = preg_replace_callback("/(\\[{2}[^\\|^\\]]+)(\\|)([^\\]]+)(\\]{2})/", 'getLabelForLink', $o);
// Replace simple links with their link-text
$o = preg_replace("/\\[{2}|\\]{2}/", "", $o);
writeTripel($s, $p, $o, 'main', $object_is);
return true;
}
}
}
}
}
function printexplicitTyping($name, $filename, $name_is, $object_is = 'n')
{
static $namearray = array();
static $predicatetypearray = array();
if ($name_is == 'c') {
$save = $name . ':Cat';
}
if ($name_is == 't') {
$save = $name . ':Temp';
}
if ($name_is == 'p') {
$save = $name . ':Pred';
}
if (!arrayMultiSearch($save, $namearray[$filename])) {
$namearray[$filename][] = $save;
if ($object_is != 'n') {
$predicatetypearray[$save][$filename]['is'] = $object_is;
}
$filedecisionTemplate = $GLOBALS['rdftypeProperty'] != $GLOBALS['W2RCFG']['templateProperty'] ? 'main' : 'type';
$filedecisionCategory = $GLOBALS['rdftypeProperty'] != $GLOBALS['W2RCFG']['categoryProperty'] ? 'main' : 'type';
if ($name_is == 'c' && $filedecisionCategory == 'type') {
writeTripel($name, $GLOBALS['W2RCFG']['categoryProperty'], $GLOBALS['W2RCFG']['classBase'], 'type');
}
if ($name_is == 'c' && $filedecisionCategory == 'main') {
printexplicitTyping($GLOBALS['W2RCFG']['categoryProperty'], $filename, 'p', 'r');
}
if ($name_is == 't' && $filedecisionTemplate == 'type') {
writeTripel($GLOBALS['W2RCFG']['wikipediaBase'] . $GLOBALS['templateLabel'] . ':' . encodeLocalName($name), $GLOBALS['rdftypeProperty'], $GLOBALS['W2RCFG']['classBase'], 'type');
}
if ($name_is == 't' && $filedecisionTemplate == 'main') {
printexplicitTyping($GLOBALS['W2RCFG']['templateProperty'], $filename, 'p', 'r');
}
if ($name_is == 'p') {
writeTripel($name, $GLOBALS['rdftypeProperty'], $object_is == 'l' ? $GLOBALS['W2RCFG']['datatypePropertyBase'] : $GLOBALS['W2RCFG']['objectPropertyBase'], 'type');
}
return;
} else {
return $predicatetypearray[$save][$filename]['is'];
}
}
