/**
* Parses one table entry and returns the table name.
*
* @param \SimpleXml $result The original XML body loaded in XML.
*
* @return string
*/
private function _parseOneTable($result)
{
$query = ".//{$this->_dataServicesMetadataPrefix}:properties/";
$query .= "{$this->_dataServicesPrefix}:TableName";
$tableName = $result->xpath($query);
$table = (string) $tableName[0];
return $table;
}
/**
* Get all data for a record (item or file).
*
* @see OaiPmhStaticRepository_Harvest_Document::_getDataForRecord()
*
* @param SimpleXml $record
* @return array The document array.
*/
protected function _getDataForRecord($record)
{
$current = array();
// Set default values to avoid notices.
$current['process'] = array();
$current['specific'] = array();
$current['metadata'] = array();
$current['extra'] = array();
// Process flat Dublin Core.
$record->registerXPathNamespace('', self::XML_NAMESPACE);
$record->registerXPathNamespace(self::XML_PREFIX, self::XML_NAMESPACE);
$record->registerXPathNamespace(self::DC_PREFIX, self::DC_NAMESPACE);
$record->registerXPathNamespace(self::DCTERMS_PREFIX, self::DCTERMS_NAMESPACE);
$xpath = 'dc:*|dcterms:*';
$dcs = $record->xpath($xpath);
foreach ($dcs as $dc) {
$name = strtolower($dc->getName());
if (isset($this->_dcTerms[$name])) {
$text = $this->_innerXML($dc);
$current['metadata']['Dublin Core'][$this->_dcTerms[$name]][] = $text;
}
}
// The xml needs the Dublin Core namespaces in some cases.
if (!empty($dcs)) {
$recordDom = dom_import_simplexml($record);
if ($recordDom) {
$recordDom->setAttribute('xmlns:' . self::DCTERMS_PREFIX, self::DCTERMS_NAMESPACE);
$recordDom->setAttribute('xmlns:' . self::DC_PREFIX, self::DC_NAMESPACE);
}
}
// Process hierarchical elements.
$elementSets = $record->elementSet;
foreach ($elementSets as $elementSet) {
$elementSetName = trim($this->_getXmlAttribute($elementSet, 'name'));
// Unmanageable.
if (strlen($elementSetName) == 0) {
continue;
}
$elements = $elementSet->element;
foreach ($elements as $element) {
$elementName = trim($this->_getXmlAttribute($element, 'name'));
// Unmanageable.
if (strlen($elementName) == 0) {
continue;
}
$data = $element->data;
foreach ($data as $value) {
$text = $this->_innerXML($value);
$current['metadata'][$elementSetName][$elementName][] = $text;
}
}
}
// Save all attributes as extra data.
foreach ($record->attributes() as $name => $data) {
$current['extra'][$name][] = (string) $data;
}
// Process extra data.
$extra = $record->extra;
if (!empty($extra)) {
$extraData = $extra->data;
foreach ($extraData as $data) {
$name = trim($this->_getXmlAttribute($data, 'name'));
if (strlen($name) > 0) {
$text = $this->_innerXML($data);
$current['extra'][$name][] = $text;
}
}
}
// Normalize special data, keeping original order.
// Filling data during loop is unpredictable.
$extraLower = array();
foreach ($current['extra'] as $field => $data) {
$lowerField = $this->_spaceFromUppercase($field);
if (isset($this->_specialData[$lowerField])) {
// Multiple values are allowed (for example tags). Keep order.
if ($this->_specialData[$lowerField]) {
// Manage the tags exception (may be "tags" or "tag").
if ($lowerField == 'tag') {
$lowerField = 'tags';
}
$extraLower[$lowerField] = empty($extraLower[$lowerField]) ? $data : array_merge($extraLower[$lowerField], $data);
} else {
$extraLower[$lowerField] = is_array($data) ? array_pop($data) : $data;
}
unset($current['extra'][$field]);
}
}
if ($extraLower) {
$current['extra'] = array_merge($current['extra'], $extraLower);
}
// Exceptions.
// Normalize "path" (exception: can be "file" or "path").
if (isset($current['extra']['file'])) {
$current['specific']['path'] = $current['extra']['file'];
unset($current['extra']['file']);
}
// Normalize true extra data.
if (!empty($current['extra'])) {
$extraData = array_diff_key($current['extra'], $this->_specialData);
if ($extraData) {
// Step 1: set single value as string, else let it as array.
$value = null;
foreach ($extraData as $name => &$value) {
if (is_array($value)) {
// Normalize empty value.
if (count($value) == 0) {
$value = '';
} elseif (count($value) == 1) {
$value = reset($value);
}
}
}
// Required, because $value is a generic reference used just before.
unset($value);
// Step 2: Normalize extra data names like geolocation[latitude]
// (array notation). They will be imported via a pseudo post.
$extra = array();
foreach ($extraData as $key => $value) {
$array = $this->_convertArrayNotation($key);
$array = $this->_nestArray($array, $value);
$value = reset($array);
$name = key($array);
$extra[] = array($name => $value);
}
$finalExtraData = array();
foreach ($extra as $data) {
$finalExtraData = array_merge_recursive($finalExtraData, $data);
}
$specialData = array_intersect_key($current['extra'], $this->_specialData);
$current['extra'] = array_merge($finalExtraData, $specialData);
}
}
// Avoid useless metadata.
unset($current['extra']['xmlns:' . self::DC_PREFIX]);
unset($current['extra']['xmlns:' . self::DCTERMS_PREFIX]);
return $current;
}
