/**
* @param $key
* @param $value
* @param $propertyId
* @param $updateTime
* @param $rowStatusId
* @param $cellLanguage
* @param $cellType
* @param $schemaId
*
* @return array $results
* @throws \PropelException
*/
private function SetPropertyElement(
$key,
$value,
$propertyId,
$updateTime,
$rowStatusId,
$cellLanguage,
$cellType
) {
$profilePropertyId = $this->prolog['columns'][$key]['id'];
if (is_array($profilePropertyId)) {
$profilePropertyId = $profilePropertyId[0];
}
//check to see if the property already exists
//note that this also checks the object value as well, so there's no way to update or delete an existing triple
//the sheet would have to contain the identifier for the triple
//actually there is. We look for exact matches and skip the update of an exact match,
//but then we look at just the old properties that are left -- the ones where the property matches but the object doesn't
//and we update those. If we have any properties left after that, we add them as new.
//We need to add an instruction to the spreadsheet to treat an empty property cell as a skip or a delete
//and if an empty cell means delete, then we delete those
//get the language for this thing
//if there's a prolog set for the language for this column, use it
//use the default for the import (already set above)
//fall back to the default language of the vocabulary
//get the fqn if using curies
if ($cellType and $this->useCuries) {
$value = $this->getFqn($value);
}
$StatementCounter['status'] = 'skipped';
/** @var \SchemaPropertyElement[] $element */
$element = \SchemaPropertyElementPeer::lookupElement($propertyId, $profilePropertyId, $value, $cellLanguage);
if ($element) {
if (1 === count($element)) {
$element = $element[0];
//make sure we handle the special case of subproperty and subclass
if (empty( $value ) && $element->getIsSchemaProperty() && in_array($element->getProfilePropertyId(),
[ 6, 9 ])
) {
return $StatementCounter;
}
} else {
//it's ambiguous and we stop processing, logging an error to be dealt with later
$error['Message'] = "Ambiguous update";
$error['PropertyId'] = $propertyId;
$error['ProfilePropertyId'] = $profilePropertyId;
$error['UpdateValue'] = $value;
$this->results['errors'][] = $error;
}
}
//create a new propertyelement for each unmatched column
//we didn't find an existing element, make a new one
if ( ! $element && ! empty( $value )) {
$element = new \SchemaPropertyElement;
$element->setCreatedUserId($this->userId);
$element->setCreatedAt($updateTime);
$element->setProfilePropertyId($profilePropertyId);
$element->setSchemaPropertyId($propertyId);
$StatementCounter['status'] = 'created';
}
if ( ! $element) {
return $StatementCounter;
}
if (( $value != $element->getObject() ) || ( $cellLanguage != $element->getLanguage() )) {
/**
* @todo We need a check here for objectproperties and handle differently
* if it's a URI, and it uses namespaces, and we have the namespace, do the substitution
**/
if ( ! $cellType and $cellLanguage) {
if ($cellLanguage != $element->getLanguage()) {
$element->setLanguage($cellLanguage);
}
}
}
if ($value != $element->getObject()) {
if ( ! empty( $value )) {
$element->setObject($value);
} else {
$element->setDeletedAt($updateTime);
}
}
if ($element->isNew() or $element->isModified()) {
$element->setUpdatedUserId($this->userId);
$element->setUpdatedAt($updateTime);
$element->setStatusId($rowStatusId);
$element->importId = $this->importId;
$element->save();
if ($StatementCounter['status'] != 'created') {
$StatementCounter['status'] = 'modified';
}
if (is_array($element->getProfilePropertyId())) {
$profilePropertyId = $element->getProfilePropertyId()[0];
} else {
$profilePropertyId = $element->getProfilePropertyId();
}
/** @var \ProfileProperty $profileProperty */
$profileProperty = $this->profileProperties[$profilePropertyId];
$StatementCounter['column'] = $key;
$StatementCounter['id'] = $element->getId();
$StatementCounter['propertyId'] = $element->getProfilePropertyId();
$StatementCounter['object'] = $element->getObject();
$StatementCounter['type'] = $cellType;
$StatementCounter['language'] = $cellLanguage;
}
unset( $element );
return $StatementCounter;
}
function run_import_marc_vocabulary($task, $args)
{
//check the argument counts
if (count($args) < 1) {
throw new Exception('You must provide a vocabulary type.');
}
if (count($args) < 2) {
throw new Exception('You must provide a file name.');
}
if (count($args) < 3) {
throw new Exception('You must provide a vocabulary id.');
}
//set the arguments
$type = strtolower($args[0]);
$filePath = $args[1];
$id = $args[2];
$deleteMissing = (isset($args[3]) && ("-d" == $args[3]));
//do some basic validity checks
if (! in_array($type, array("schema", "vocab", "vocabulary"))) {
throw new Exception('You must import into a schema or a vocab');
}
if ("vocabulary" == $type) {
$type = "vocab";
}
if (! is_numeric($id)) {
throw new Exception('You must provide a valid ID');
}
//does the file exist?
if (! file_exists($filePath)) {
throw new Exception('You must supply a valid file to import: ' . $filePath);
}
//is the file a valid type?
if (preg_match('/^.+\.([[:alpha:]]{2,4})$/', $filePath, $matches)) {
if (! in_array(strtolower($matches[1]), array("json", "rdf", "csv", "xml"))) {
throw new Exception('You must provide a valid file type based on the extension');
}
} else {
throw new Exception("File type cannot be determined from the file extension");
}
$fileType = $matches[1];
//is the object a valid object?
if ('vocab' == $type) {
$vocabObj = VocabularyPeer::retrieveByPK($id);
if (is_null($vocabObj)) {
throw new Exception('Invalid vocabulary ID');
}
//set some defaults
$baseDomain = $vocabObj->getBaseDomain();
$language = $vocabObj->getLanguage();
$statusId = $vocabObj->getStatusId();
//get a skos property id map
$skosMap = SkosPropertyPeer::getPropertyNames();
//there has to be a hash or a slash
$tSlash = preg_match('@(/$)@i', $vocabObj->getUri()) ? '' : '/';
$tSlash = preg_match('/#$/', $vocabObj->getUri()) ? '' : $tSlash;
} else {
$schemaObj = SchemaPeer::retrieveByPK($id);
if (is_null($schemaObj)) {
throw new Exception('Invalid schema ID');
}
//set some defaults
$baseDomain = $schemaObj->getUri();
$language = $schemaObj->getLanguage();
$statusId = $schemaObj->getStatusId();
//get a element set property id map
$profileId = 1;
$profile = ProfilePeer::retrieveByPK($profileId);
$elementMap = $profile->getAllProperties();
//there has to be a hash or a slash
$tSlash = preg_match('@(/$)@i', $baseDomain) ? '' : '/';
$tSlash = preg_match('/#$/', $baseDomain) ? '' : $tSlash;
}
// insert jon's user id
$userId = 36;
/* From here on the process is the same regardless of UI */
//execute
// parse file to get the fields/columns and data
$file = fopen($filePath, "r");
if (! $file) {
throw new Exception("Can't read supplied file");
}
// check to see if file has been uploaded before
// check import history for file name
$importHistory = FileImportHistoryPeer::retrieveByLastFilePath($filePath);
// if reimport
// get last import history for filename
// unserialize column map
// match column names to AP based on map
// look for matches in unmatched field/column names to AP (ideal)
// csv table of data --
// row1: parsed field names/column headers
// row2: select dropdown with available fields from object AP (pre-select known matches)
// each select identified by column number
// row3: display datatype of selected field (updated dynamically when field selected)
// row4-13: first 10 rows of parsed data from file
// require a column that can match to 'URI' (maybe we'll allow an algorithm later)
// require columns that are required by AP
// on reimport there should be a flag to 'delete missing properties' from the current data
// note: at some point there will be a reimport process that allows URI changing
// this will require that there be an OMR identifier embedded in the incoming data
switch ($fileType) {
case "csv":
try {
$reader = new aCsvReader($filePath);
} catch(Exception $e) {
throw new Exception("Not a happy CSV file!");
}
if ('vocab' == $type) {
// Get array of heading names found
$headings = $reader->getHeadings();
$fields = ConceptPeer::getFieldNames();
//set the map
// $map[] = array("property" => "Uri", "column" => "URILocal");
// $map[] = array("property" => "prefLabel", "column" => "skos:prefLabel");
// $map[] = array("property" => "definition", "column" => "skos:definition");
// $map[] = array("property" => "notation", "column" => "skos:notation");
// $map[] = array("property" => "scopeNote", "column" => "skos:scopeNote");
$map = array(
"uri" => "URILocal",
"prefLabel" => "skos:prefLabel",
"definition" => "skos:definition",
"notation" => "skos:notation",
"scopeNote" => "skos:scopeNote"
);
$rows = 0;
//executeImport:
// serialize the column map
try {
while ($row = $reader->getRow()) {
$rows ++;
// lookup the URI (or the OMR ID if available) for a match
$uri = $baseDomain . $row[$map["uri"]];
$concept = ConceptPeer::getConceptByUri($uri);
$updateTime = time();
$language = (isset($map['language'])) ? $row[$map['language']] : $vocabObj->getLanguage();
if (! $concept) {
// create a new concept or element
$concept = new Concept();
$concept->setVocabulary($vocabObj);
$concept->setUri($uri);
/**
* @todo Need to handle updates for topconcept here, just like language
**/
$concept->setIsTopConcept(false);
$concept->updateFromRequest(
$userId,
fixMarcEncoding(rtrim($row[$map['prefLabel']])),
$language,
$statusId
);
} //don't update the concept if the preflabel matches
else if ($row[$map['prefLabel']] != $concept->getPrefLabel()) {
$concept->updateFromRequest($userId, fixMarcEncoding(rtrim($row[$map['prefLabel']])));
}
//there needs to be a language to lookup the properties unless it's an objectProperty
$rowLanguage = (isset($map['language'])) ? $row[$map['language']] : $concept->getLanguage();
foreach ($map as $key => $value) {
//we skip because we already did them
if (! in_array($key, array('uri', 'prefLabel', 'language'))) {
$skosId = $skosMap[$key];
//check to see if the property already exists
$property =
ConceptPropertyPeer::lookupProperty($concept->getId(), $skosId, $rowLanguage);
//create a new property for each unmatched column
if (! empty($row[$value])) {
if (! $property) {
$property = new ConceptProperty();
$property->setCreatedUserId($userId);
$property->setConceptId($concept->getId());
$property->setCreatedAt($updateTime);
$property->setSkosPropertyId($skosId);
}
if (($row[$value] != $property->getObject()) ||
($rowLanguage != $property->getLanguage())
) {
/**
* @todo We need a check here for skos objectproperties and handle differently
**/
if ($rowLanguage != $property->getLanguage()) {
$property->setLanguage($rowLanguage);
}
if ($row[$value] != $property->getObject()) {
$property->setObject(fixMarcEncoding(rtrim($row[$value])));
}
$property->setUpdatedUserId($userId);
$property->setUpdatedAt($updateTime);
$property->save();
}
} //the row value is empty
else if ($deleteMissing && $property) {
$property->delete();
}
}
}
// else
// lookup and update concept or element
// lookup and update each property
// update the history for each property, action is 'import', should be a single timestamp for all (this should be automatic)
// if 'delete missing properties' is true
// delete each existing, non-required property that wasn't updated by the import
}
} catch(Exception $e) {
// catch
// if there's an error of any kind, write to error log and continue
echo "Error on row: " . $rows . ", " . $uri . "\n" . $e . "\n";
continue;
}
$objects = $vocabObj->countConcepts();
} else //it's an element set
{
$map = array(
"uri" => "uriLocalPart",
"name" => "reg:name",
"definition" => "skos:definition",
"label" => "rdfs:label",
"note" => array("tag" => "tagCap", "ind1" => "ind1Cap", "ind2" => "ind2Cap", "sub" => "subCap")
);
$rows = 0;
//executeImport:
// serialize the column map
try {
while ($row = $reader->getRow()) {
// lookup the URI (or the OMR ID if available) for a match
//There always has to be a URI on either update or create
if (! isset($row[$map["uri"]])) {
throw new Exception('Missing URI for row: ' . $reader->getRowCount());
continue;
}
$rows ++;
$uri = $baseDomain . $tSlash . $row[$map["uri"]];
$property = SchemaPropertyPeer::retrieveByUri($uri);
$updateTime = time();
$rowLanguage = (isset($map['language'])) ? $row[$map['language']] : $language;
$rowStatusId = (isset($map['status'])) ? $row[$map['status']] : $statusId;
if (! $property) {
// create a new property
/** @var SchemaProperty * */
$property = new SchemaProperty();
$property->setSchema($schemaObj);
$property->setUri($uri);
$property->setCreatedUserId($userId);
$property->setCreatedAt($updateTime);
}
$property->setLanguage($rowLanguage);
$property->setStatusId($rowStatusId);
$property->setUpdatedUserId($userId);
$property->setUpdatedAt($updateTime);
if (isset($row[$map["label"]])) {
$property->setLabel($row[$map["label"]]);
}
if (isset($row[$map["name"]])) {
$property->setName($row[$map["name"]]);
}
if (isset($row[$map["definition"]])) {
$property->setDefinition($row[$map["definition"]]);
}
if (is_array($map["note"])) {
$note = '';
foreach ($map["note"] as $key => $value) {
$caption = ! empty($row[$value]) ? " (" . $row[$value] . ")" : ' (no caption)';
$note .= ! empty($row[$key]) ? $key . ": " . $row[$key] . $caption . "<br />" : "";
}
$property->setNote($note);
} else {
if (isset($row[$map["note"]])) {
$property->setNote($row[$map["note"]]);
}
}
$property->saveSchemaProperty($userId);
/**
* @todo Need to handle domain and range
**/
foreach ($map as $key => $value) {
//we skip because we already did them
if (! in_array(
$key,
array('uri', 'status', 'language', 'label', 'name', 'definition', 'comment', 'note')
)
) {
$elementId = $elementMap[$key];
//check to see if the property already exists
//note that this also checks the object value as well, so there's no way to update or delete an existing triple
//the sheet would have to conatin the identifier for the triple
$element = SchemaPropertyElementPeer::lookupElement(
$schemaObj->getId(),
$elementId,
$map[$value]
);
//create a new property for each unmatched column
if (! empty($row[$value])) {
if (! $element) {
$element = new SchemaPropertyElement();
$element->setCreatedUserId($userId);
$element->setCreatedAt($updateTime);
$element->setProfilePropertyId($elementId);
}
if (($row[$value] != $element->getObject()) ||
($rowLanguage != $element->getLanguage())
) {
/**
* @todo We need a check here for objectproperties and handle differently
**/
if ($rowLanguage != $element->getLanguage()) {
$element->setLanguage($rowLanguage);
}
if ($row[$value] != $element->getObject()) {
$element->setObject($row[$value]);
}
$element->setUpdatedUserId($userId);
$element->setUpdatedAt($updateTime);
$element->save();
}
} //the row value is empty
else if ($deleteMissing && $element) {
$element->delete();
}
}
}
// else
// lookup and update concept or element
// lookup and update each property
// update the history for each property, action is 'import', should be a single timestamp for all (this should be automatic)
// if 'delete missing properties' is true
// delete each existing, non-required property that wasn't updated by the import
}
} catch(Exception $e) {
// catch
// if there's an error of any kind, write to error log and continue
echo "Error on row: " . $rows . ", " . $uri . "\n" . $e . "\n";
continue;
}
$objects = $schemaObj->countSchemaPropertys();
}
// save the import history file (match timestamp to history entries)
break;
case "json":
break;
case "rdf":
break;
case "xml":
break;
default:
}
/* output to stdout*/
// number of objects imported (link to history, filtered on timestamp of import)
echo "File:" . $filePath . ";\n Objects imported: " . $objects . "; Rows read: " . $rows . "\n";
// number of errors (link to error log)
}
/**
* create a new element
*
* @param SchemaPropertyElement $element
* @param integer $userId
* @param $field
* @param string $object
* @param \Connection $con
*
* @return bool|\SchemaPropertyElement
*/
public function updateElement(SchemaPropertyElement $element, $userId, $field, $object, $con)
{
if ($element) {
$element->setIsSchemaProperty(true);
$element->setUpdatedUserId($userId);
//SchemaPropertyElementPeer::updateElement($schema_property, $element, $userId, $field, $con);
if ('is_subproperty_of' == $field || 'is_subclass_of' == $field) {
if ($this->getIsSubpropertyOf() && empty($object)) {
$element->setRelatedSchemaPropertyId($this->getIsSubpropertyOf());
$object = $this->getParentUri();
}
}
$element->setObject($object ? $object : '');
$element->save($con);
}
return $element;
}