public static function parse($doc, $options = array())
{
$terms = array();
libxml_use_internal_errors(true);
// Report XML errors
if (!$doc) {
foreach (libxml_get_errors() as $error) {
//TODO echo errors in template. Use custom validator?
var_dump($error);
}
}
$skos = new sfSkosPlugin();
$skos->xpath = new DOMXPath($doc);
// Create Xpath object, register namespaces
$skos->xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
$skos->xpath->registerNamespace('skos', 'http://www.w3.org/2004/02/skos/core#');
$skos->xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
// Set taxonomy
$skos->taxonomy = QubitTaxonomy::getById(QubitTaxonomy::SUBJECT_ID);
if (isset($options['taxonomy'])) {
$skos->taxonomy = $options['taxonomy'];
}
$skos->parent = QubitTerm::getById(QubitTerm::ROOT_ID);
if (isset($options['parent'])) {
$skos->parent = $options['parent'];
}
// XPath selector for expanded RDF syntax
$rdfsel = "rdf:Description[rdf:type[@rdf:resource='http://www.w3.org/2004/02/skos/core#Concept']]";
// Get all concepts
$concepts = $skos->xpath->query("skos:Concept | {$rdfsel}");
// Create terms from concepts
foreach ($concepts as $concept) {
if (!$concept instanceof domElement) {
continue;
}
$skos->addTerm($concept);
}
// Built term associations (including hierarchy)
foreach ($concepts as $concept) {
if (!$concept instanceof domElement) {
continue;
}
// Add parent
if (0 < $skos->xpath->query('./skos:broader', $concept)->length) {
$skos->setParent($concept);
}
// Add children
if (0 < $skos->xpath->query('./skos:narrower', $concept)->length) {
$skos->setChildren($concept);
}
// Add relations
if (0 < $skos->xpath->query('./skos:related', $concept)->length) {
$skos->addTermRelations($concept);
}
}
return $skos;
}
public function execute($request)
{
$this->form = new sfForm();
$this->timer = new QubitTimer();
$this->terms = array();
$this->termsPerPage = intval(sfConfig::get('app_hits_per_page'));
$this->taxonomy = null;
$this->parent = QubitTerm::getById(QubitTerm::ROOT_ID);
if (isset($this->getRoute()->resource)) {
$resource = $this->getRoute()->resource;
if ('QubitTaxonomy' == $resource->className) {
$this->taxonomy = QubitTaxonomy::getById($resource->id);
} else {
$this->parent = QubitTerm::getById($resource->id);
$this->taxonomy = $this->parent->taxonomy;
}
}
if (!isset($this->taxonomy)) {
$this->forward404();
}
// Check user authorization
if (!QubitAcl::check($this->parent, 'create')) {
QubitAcl::forwardUnauthorized();
}
$this->form->setWidget('file', new sfWidgetFormInputFile());
$this->form->setValidator('file', new sfValidatorFile());
if ($request->isMethod('post')) {
$this->form->bind($request->getPostParameters(), $request->getFiles());
if ($this->form->isValid()) {
if (null !== ($file = $this->form->getValue('file'))) {
$doc = new domDocument();
$doc->substituteEntities = true;
$doc->load($file->getTempName());
$this->skos = sfSkosPlugin::parse($doc, array('taxonomy' => $this->taxonomy, 'parent' => $this->parent));
}
}
} else {
$this->setTemplate('importSelect');
}
}
public function import($xmlFile, $options = array())
{
// load the XML document into a DOMXML object
$importDOM = $this->loadXML($xmlFile, $options);
// if we were unable to parse the XML file at all
if (empty($importDOM->documentElement)) {
$errorMsg = sfContext::getInstance()->i18n->__('Unable to parse XML file: malformed or unresolvable entities');
throw new Exception($errorMsg);
}
// if libxml threw errors, populate them to show in the template
if ($importDOM->libxmlerrors) {
// warning condition, XML file has errors (perhaps not well-formed or invalid?)
foreach ($importDOM->libxmlerrors as $libxmlerror) {
$xmlerrors[] = sfContext::getInstance()->i18n->__('libxml error %code% on line %line% in input file: %message%', array('%code%' => $libxmlerror->code, '%message%' => $libxmlerror->message, '%line%' => $libxmlerror->line));
}
$this->errors = array_merge((array) $this->errors, $xmlerrors);
}
if ('eac-cpf' == $importDOM->documentElement->tagName) {
$this->rootObject = new QubitActor();
$this->rootObject->parentId = QubitActor::ROOT_ID;
$eac = new sfEacPlugin($this->rootObject);
$eac->parse($importDOM);
$this->rootObject->save();
return $this;
}
// FIXME hardcoded until we decide how these will be developed
$validSchemas = array('+//ISBN 1-931666-00-8//DTD ead.dtd Encoded Archival Description (EAD) Version 2002//EN' => 'ead', '-//Society of American Archivists//DTD ead.dtd (Encoded Archival Description (EAD) Version 1.0)//EN' => 'ead1', 'http://www.loc.gov/METS/' => 'mets', 'http://www.loc.gov/mods/' => 'mods', 'http://www.loc.gov/MARC21/slim' => 'marc', 'record' => 'oai_dc_record', 'dc' => 'dc', 'oai_dc:dc' => 'dc', 'dublinCore' => 'dc', 'metadata' => 'dc', 'ead' => 'ead', 'add' => 'alouette', 'http://www.w3.org/2004/02/skos/core#' => 'skos');
// determine what kind of schema we're trying to import
$schemaDescriptors = array($importDOM->documentElement->tagName);
if (!empty($importDOM->namespaces)) {
krsort($importDOM->namespaces);
$schemaDescriptors = array_merge($schemaDescriptors, $importDOM->namespaces);
}
if (!empty($importDOM->doctype)) {
$schemaDescriptors = array_merge($schemaDescriptors, array($importDOM->doctype->name, $importDOM->doctype->systemId, $importDOM->doctype->publicId));
}
foreach ($schemaDescriptors as $descriptor) {
if (array_key_exists($descriptor, $validSchemas)) {
$importSchema = $validSchemas[$descriptor];
}
}
switch ($importSchema) {
case 'ead':
// just validate EAD import for now until we can get StrictXMLParsing working for all schemas in the self::LoadXML function. Having problems right now loading schemas.
$importDOM->validate();
// if libxml threw errors, populate them to show in the template
foreach (libxml_get_errors() as $libxmlerror) {
$this->errors[] = sfContext::getInstance()->i18n->__('libxml error %code% on line %line% in input file: %message%', array('%code%' => $libxmlerror->code, '%message%' => $libxmlerror->message, '%line%' => $libxmlerror->line));
}
break;
case 'skos':
$criteria = new Criteria();
$criteria->add(QubitSetting::NAME, 'plugins');
$setting = QubitSetting::getOne($criteria);
if (null === $setting || !in_array('sfSkosPlugin', unserialize($setting->getValue(array('sourceCulture' => true))))) {
throw new sfException(sfContext::getInstance()->i18n->__('The SKOS plugin is not enabled'));
}
$importTerms = sfSkosPlugin::parse($importDOM);
$this->rootObject = QubitTaxonomy::getById(QubitTaxonomy::SUBJECT_ID);
$this->count = count($importTerms);
return $this;
break;
}
$importMap = sfConfig::get('sf_app_module_dir') . DIRECTORY_SEPARATOR . 'object' . DIRECTORY_SEPARATOR . 'config' . DIRECTORY_SEPARATOR . 'import' . DIRECTORY_SEPARATOR . $importSchema . '.yml';
if (!file_exists($importMap)) {
// error condition, unknown schema or no import filter
$errorMsg = sfContext::getInstance()->i18n->__('Unknown schema or import format: "%format%"', array('%format%' => $importSchema));
throw new Exception($errorMsg);
}
$this->schemaMap = sfYaml::load($importMap);
// if XSLs are specified in the mapping, process them
if (!empty($this->schemaMap['processXSLT'])) {
// pre-filter through XSLs in order
foreach ((array) $this->schemaMap['processXSLT'] as $importXSL) {
$importXSL = sfConfig::get('sf_data_dir') . DIRECTORY_SEPARATOR . 'xslt' . DIRECTORY_SEPARATOR . $importXSL;
if (file_exists($importXSL)) {
// instantiate an XSLT parser
$xslDOM = new DOMDocument();
$xslDOM->load($importXSL);
// Configure the transformer
$xsltProc = new XSLTProcessor();
$xsltProc->registerPHPFunctions();
$xsltProc->importStyleSheet($xslDOM);
$importDOM->loadXML($xsltProc->transformToXML($importDOM));
unset($xslDOM);
unset($xsltProc);
} else {
$this->errors[] = sfContext::getInstance()->i18n->__('Unable to load import XSL filter: "%importXSL%"', array('%importXSL%' => $importXSL));
}
}
// re-initialize xpath on the new XML
$importDOM->xpath = new DOMXPath($importDOM);
}
// switch source culture if langusage is set in an EAD document
if ($importSchema == 'ead') {
if (is_object($langusage = $importDOM->xpath->query('//eadheader/profiledesc/langusage/language/@langcode'))) {
$sf_user = sfContext::getInstance()->user;
$currentCulture = $sf_user->getCulture();
$langCodeConvertor = new fbISO639_Map();
foreach ($langusage as $language) {
$isocode = trim(preg_replace('/[\\n\\r\\s]+/', ' ', $language->nodeValue));
// convert to Symfony culture code
if (!($twoCharCode = strtolower($langCodeConvertor->getID2($isocode)))) {
$twoCharCode = $isocode;
}
// Check to make sure that the selected language is supported with a Symfony i18n data file.
// If not it will cause a fatal error in the Language List component on every response.
ProjectConfiguration::getActive()->loadHelpers('I18N');
try {
format_language($twoCharCode, $twoCharCode);
} catch (Exception $e) {
$this->errors[] = sfContext::getInstance()->i18n->__('EAD "langmaterial" is set to') . ': "' . $isocode . '". ' . sfContext::getInstance()->i18n->__('This language is currently not supported.');
continue;
}
if ($currentCulture !== $twoCharCode) {
$this->errors[] = sfContext::getInstance()->i18n->__('EAD "langmaterial" is set to') . ': "' . $isocode . '" (' . format_language($twoCharCode, 'en') . '). ' . sfContext::getInstance()->i18n->__('Your XML document has been saved in this language and your user interface has just been switched to this language.');
}
$sf_user->setCulture($twoCharCode);
// can only set to one language, so have to break once the first valid language is encountered
break;
}
}
}
unset($this->schemaMap['processXSLT']);
// go through schema map and populate objects/properties
foreach ($this->schemaMap as $name => $mapping) {
// if object is not defined or a valid class, we can't process this mapping
if (empty($mapping['Object']) || !class_exists('Qubit' . $mapping['Object'])) {
$this->errors[] = sfContext::getInstance()->i18n->__('Non-existent class defined in import mapping: "%class%"', array('%class%' => 'Qubit' . $mapping['Object']));
continue;
}
// get a list of XML nodes to process
$nodeList = $importDOM->xpath->query($mapping['XPath']);
foreach ($nodeList as $domNode) {
// create a new object
$class = 'Qubit' . $mapping['Object'];
$currentObject = new $class();
// set the rootObject to use for initial display in successful import
if (!$this->rootObject) {
$this->rootObject = $currentObject;
}
// if a parent path is specified, try to parent the node
if (empty($mapping['Parent'])) {
$parentNodes = new DOMNodeList();
} else {
$parentNodes = $importDOM->xpath->query('(' . $mapping['Parent'] . ')', $domNode);
}
if ($parentNodes->length > 0) {
// parent ID comes from last node in the list because XPath forces forward document order
$parentId = $parentNodes->item($parentNodes->length - 1)->getAttribute('xml:id');
unset($parentNodes);
if (!empty($parentId) && is_callable(array($currentObject, 'setParentId'))) {
$currentObject->parentId = $parentId;
}
} else {
// orphaned object, set root if possible
if (is_callable(array($currentObject, 'setRoot'))) {
$currentObject->setRoot();
}
}
// go through methods and populate properties
foreach ($mapping['Methods'] as $name => $methodMap) {
// if method is not defined, we can't process this mapping
if (empty($methodMap['Method']) || !is_callable(array($currentObject, $methodMap['Method']))) {
$this->errors[] = sfContext::getInstance()->i18n->__('Non-existent method defined in import mapping: "%method%"', array('%method%' => $methodMap['Method']));
continue;
}
// get a list of XML nodes to process
$nodeList2 = $importDOM->xpath->query($methodMap['XPath'], $domNode);
if (is_object($nodeList2)) {
switch ($name) {
// hack: some multi-value elements (e.g. 'languages') need to get passed as one array instead of individual nodes values
case 'languages':
$langCodeConvertor = new fbISO639_Map();
$value = array();
foreach ($nodeList2 as $nodeee) {
if ($twoCharCode = $langCodeConvertor->getID2($nodeee->nodeValue)) {
$value[] = strtolower($twoCharCode);
} else {
$value[] = $nodeee->nodeValue;
}
}
$currentObject->language = $value;
break;
case 'flocat':
$resources = array();
foreach ($nodeList2 as $nodeee) {
$resources[] = $nodeee->nodeValue;
}
if (0 < count($resources)) {
$currentObject->importDigitalObjectFromUri($resources);
}
break;
default:
foreach ($nodeList2 as $domNode2) {
// normalize the node text (trim whitespace manually); NB: this will strip any child elements, eg. HTML tags
$nodeValue = trim(preg_replace('/[\\n\\r\\s]+/', ' ', $domNode2->nodeValue));
// if you want the full XML from the node, use this
$nodeXML = $domNode2->ownerDocument->saveXML($domNode2);
// set the parameters for the method call
if (empty($methodMap['Parameters'])) {
$parameters = array($nodeValue);
} else {
$parameters = array();
foreach ((array) $methodMap['Parameters'] as $parameter) {
// if the parameter begins with %, evaluate it as an XPath expression relative to the current node
if ('%' == substr($parameter, 0, 1)) {
// evaluate the XPath expression
$xPath = substr($parameter, 1);
$result = $importDOM->xpath->query($xPath, $domNode2);
if ($result->length > 1) {
// convert nodelist into an array
foreach ($result as $element) {
$resultArray[] = $element->nodeValue;
}
$parameters[] = $resultArray;
} else {
// pass the node value unaltered; this provides an alternative to $nodeValue above
$parameters[] = $result->item(0)->nodeValue;
}
} else {
// Confirm DOMXML node exists to avoid warnings at run-time
if (false !== preg_match_all('/\\$importDOM->xpath->query\\(\'@\\w+\', \\$domNode2\\)->item\\(0\\)->nodeValue/', $parameter, $matches)) {
foreach ($matches[0] as $match) {
$str = str_replace('->nodeValue', '', $match);
if (null !== ($node = eval('return ' . $str . ';'))) {
// Substitute node value for search string
$parameter = str_replace($match, '\'' . $node->nodeValue . '\'', $parameter);
} else {
// Replace empty nodes with null in parameter string
$parameter = str_replace($match, 'null', $parameter);
}
}
}
eval('$parameters[] = ' . $parameter . ';');
}
}
}
// invoke the object and method defined in the schema map
call_user_func_array(array(&$currentObject, $methodMap['Method']), $parameters);
}
}
unset($nodeList2);
}
}
// make sure we have a publication status set before indexing
if ($currentObject instanceof QubitInformationObject && count($currentObject->statuss) == 0) {
$currentObject->setPublicationStatus(sfConfig::get('app_defaultPubStatus', QubitTerm::PUBLICATION_STATUS_DRAFT_ID));
}
// save the object after it's fully-populated
$currentObject->save();
// write the ID onto the current XML node for tracking
$domNode->setAttribute('xml:id', $currentObject->id);
}
}
return $this;
}
