/**
* All of the following names will be encoded to 'Émile Zola':
* '%C3%89mile_Zola', '%C3%A9mile_Zola', ' %C3%A9mile Zola ', ' %C3%A9mile _ Zola ', ' Émile _ Zola '
*
* TODO: maybe we should expect (require) the name to be normalized, e.g. with uppercase
* first letter and without duplicate spaces or spaces at start or end?
* Would make this method much simpler.
*
* @param $name encoded MediaWiki page name, e.g. '%C3%89mile_Zola'.
* Must not include the namespace (e.g. 'Template:').
*/
public static function wikiDecode($name)
{
PhpUtil::assertString($name, 'name');
// make first character uppercase
$name = StringUtil::mb_ucfirst(self::cleanSpace(urldecode($name)));
return $name;
}
private function updateConfig($dir, $path, $source)
{
PhpUtil::assertString($path, 'path');
PhpUtil::assertString($source, 'source');
// TODO: add '#1' etc. if necessary on Windows for file names that only differ in case
// TODO: make suffix configurable
$file = $dir . $path . '.txt';
return $this->buildConfigFile($file, $source);
}
public function setDestination($destinationId, $destination)
{
PhpUtil::assertString($destinationId, 'destination id');
PhpUtil::assertType($destination, 'dbpedia\\destinations\\QuadDestination', 'destination');
if (isset($this->destinations[$destinationId])) {
throw new \InvalidArgumentException('destination for id [' . $destinationId . '] already set');
}
$this->destinations[$destinationId] = $destination;
}
public function addUnit($unit)
{
PhpUtil::assertType($unit, 'dbpedia\\ontology\\dataTypes\\UnitDataType', 'unit');
$unit->setDimension($this);
foreach ($unit->getLabels() as $label) {
if (isset($this->units[$label])) {
throw new \InvalidArgumentException($this . ' already has unit label ' . $label);
}
$this->units[$label] = $unit;
}
}
/**
* Loads all classes and properties but does not link them.
*
* @param $ontology The ontology instance
* @param $pageNode The page node of the configuration page
*/
private function createClasses($ontology, $pageNode)
{
// PHP keeps going if param type is wrong, but we want an exception
PhpUtil::assertType($ontology, 'dbpedia\\ontology\\Ontology', 'ontology');
PhpUtil::assertType($pageNode, 'dbpedia\\wikiparser\\PageNode', 'page node');
$name = self::getPageName($pageNode);
foreach ($pageNode->getChildren('TemplateNode') as $node) {
$templateName = $node->getTitle()->encoded();
if ($templateName == OntologyClass::TEMPLATE_NAME) {
$ontClass = new OntologyClass($name);
$labelProperty = $node->getProperty("rdfs:label");
if ($labelProperty) {
$ontClass->setLabel($labelProperty->getText());
} else {
$this->logger->warn("Class " . $ontClass->getUri() . " does not define any label.");
}
$ontology->addClass($ontClass);
} else {
if ($templateName == OntologyObjectProperty::TEMPLATE_NAME || $templateName == OntologyDataTypeProperty::TEMPLATE_NAME) {
if ($templateName == OntologyObjectProperty::TEMPLATE_NAME) {
$ontProperty = new OntologyObjectProperty($name);
} else {
$ontProperty = new OntologyDataTypeProperty($name);
}
$labelProperty = $node->getProperty("rdfs:label");
if ($labelProperty) {
$ontProperty->setLabel($labelProperty->getText());
} else {
$this->logger->warn("Property without any label found on page " . $pageName);
}
$typeProperty = $node->getProperty("rdf:type");
if ($typeProperty) {
if ($typeProperty->getText() == 'owl:FunctionalProperty') {
$ontProperty->setFunctional(true);
} else {
$this->logger->warn("Property with an invalid type found on page " . $pageName);
}
}
$ontology->addProperty($ontProperty);
}
}
}
}
/**
* @param $baseDir must end with a directory separator (slash or backslash)
* @param $skipNames names (not paths) of files and directories to skip, e.g. '.svn'.
* If not given, all files and directories will be included.
* @param $paths array of strings, paths of files to use, relative to base dir,
* using forward slashes. If not given, all files and directories will be included.
*/
public function __construct($baseDir, $skipNames = null, $paths = null)
{
PhpUtil::assertString($baseDir, 'base dir');
$baseDir = str_replace('\\', '/', realpath($baseDir));
if (!is_dir($baseDir)) {
throw new \InvalidArgumentException('base dir must be an existing directory, but is ' . $baseDir);
}
// make sure that $baseDir ends with /
if (!StringUtil::endsWith($baseDir, '/')) {
$baseDir .= '/';
}
if ($skipNames !== null) {
PhpUtil::assertArray($skipNames, 'skip names');
} else {
$skipNames = array();
}
if ($paths !== null) {
PhpUtil::assertArray($paths, 'paths');
}
$this->baseDir = $baseDir;
$this->skipNames = $skipNames;
$this->paths = $paths;
}
public function getDestination($destinationId)
{
PhpUtil::assertString($destinationId, 'destination id');
return $this->destination;
}
/**
* @param $nsCode namespace code
* @param $decoded decoded page name. URL-decoded, using normalized spaces (not underscores), first letter uppercase.
*/
public function __construct($nsCode, $decoded)
{
PhpUtil::assertInteger($nsCode, 'namespace code');
if (!isset(self::$nsNames[$nsCode])) {
throw new \InvalidArgumentException('unknown namespace code ' . $nsCode);
}
PhpUtil::assertString($decoded, 'page name');
if (strlen($decoded) === 0) {
throw new WikiParserException('page name must not be empty');
}
$this->nsCode = $nsCode;
$this->nsName = self::$nsNames[$nsCode][0];
$this->encoded = WikiUtil::wikiEncode($decoded);
// re-decode to make sure name is normalized
$this->decoded = WikiUtil::wikiDecode($this->encoded);
}
/**
* Sets the range of this property.
*
* @return string The range of this property.
*/
public function setRange($range)
{
if (!$range instanceof OntologyClass && !$range instanceof DataType) {
throw new \InvalidArgumentException('range must be an OntologyClass or a DataType, but has type ' . PhpUtil::typeNameOf($range));
}
$this->range = $range;
}
public function __construct($file)
{
PhpUtil::assertString($file, 'file');
$this->file = $file;
}
