/**
* @param $task
* @param $args
*
* Arg[0] is one of "schema" (element set), "vocab" or "vocabulary"
* arg[1] is the vocabulary name.
* The file type is determined by the extension and must be one of "json", "rdf", "csv", "xml"
* arg[2] is the vocabulary id
* arg[3] is the batch id
* arg[4] [optional] is -d
*
* @throws Exception
*/
function run_import_vocabulary($task, $args)
{
//xdebug_break();
//check the argument counts
if (count($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];
$vocabId = $args[2];
$batchId = isset($args[3]) ? $args[3] : "";
$deleteMissing = (isset($args[4]) && ("-d" == $args[4]));
//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) {
$type = "vocab";
}
if (! is_numeric($vocabId)) {
throw new Exception('You must provide a valid ID');
}
//does the file exist?
if (! file_exists($filePath)) {
//default to the site upload path
$filePath = $GLOBALS['uploadPath'] . $filePath;
if (! file_exists($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) {
$vocabObj = VocabularyPeer::retrieveByPK($vocabId);
if (is_null($vocabObj)) {
throw new Exception('Invalid vocabulary ID');
}
//set some defaults
$baseDomain = $vocabObj->getBaseDomain();
$language = $vocabObj->getLanguage();
$statusId = $vocabObj->getStatusId();
$userId = $vocabObj->getCreatedUserId();
$agentId = $vocabObj->getAgentId();
//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 {
$import = new ImportVocab($type, $filePath, $vocabId);
}
/* From here on the process is the same regardless of UI */
// 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":
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(
$userId,
fixEncoding(rtrim($row[$map['prefLabel']])),
$language,
$statusId
);
} //don't update the concept if the preflabel matches
else if ($row[$map['prefLabel']] != $concept->getPrefLabel()) {
$concept->updateFromRequest($userId, fixEncoding(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()) {
$property->setObject(fixEncoding(rtrim($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
{
$import->setCsvReader($import->file);
$import->processProlog();
$import->getDataColumnIds();
$import->processData();
//todo: $results should be a class
$results[$vocabId] = $import->results;
$bacthId = $import->saveResults($batchId);
}
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 " Rows imported: " . count($results[$vocabId]['success']['rows']) . "\n From File:" . $filePath . "\nUse this ID for more in this batch: " . $bacthId;
// number of errors (link to error log)
}
/**
* @param \ConceptProperty | \ConceptProperty[] $dbElement
* @param $value
* @param $rowStatus
* @param \Concept $concept
* @param \ProfileProperty $profileProperty
* @param $language
*
* @param $key
*
* @return bool
* @throws \PropelException
*/
private function upsertConceptFromRow(
&$dbElement,
$value,
$rowStatus,
&$concept,
$profileProperty,
$language,
$key
) {
//set the keys to skip, because concepts don't maintain uri and status separately
$skipKeys = [ 59, 62 ];
if (in_array($key, $skipKeys)) {
return true;
}
if ( ! is_array($dbElement)) {
if (empty( $dbElement ) and $value) {
$dbElement = new \ConceptProperty();
$dbElement->setSkosPropertyId($profileProperty->getSkosId());
$dbElement->setConceptId($concept->getId());
$dbElement->setLanguage($language);
$dbElement->setCreatedUserId($this->userId);
$concept->addConceptPropertyRelatedByConceptId($dbElement);
}
if ($dbElement and $value !== $dbElement->getObject()) {
if (empty( $value )) {
$dbElement->delete();
} else {
$dbElement->setStatusId($rowStatus);
$dbElement->setObject($value);
$dbElement->setUpdatedUserId($this->userId);
$dbElement->importId = $this->importId;
$dbElement->setDeletedAt(null);
}
//$dbElement->save();
if ($profileProperty->getIsInForm()) {
if (( $profileProperty->getHasLanguage() and $concept->getLanguage() == $dbElement->getLanguage() ) or ! $profileProperty->getHasLanguage()) {
$this->setConceptValue($value,
$concept,
$profileProperty->getName(), ! $profileProperty->getIsObjectProp(),
$key);
$dbElement->setIsConceptProperty(true);
}
}
return true;
}
} else {
$foundOne = false;
foreach ($dbElement as &$oneElement) {
if ( ! $foundOne) {
$foundOne = $this->upsertConceptFromRow($oneElement,
$value,
$rowStatus,
$concept,
$profileProperty,
$language,
$key);
}
}
}
return true;
}
function run_import_marc_vocabulary($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)
}
public function executeEdit()
{
$this->setFlash('vocabID', $this->getVocabularyId());
if ($this->getRequest()->getMethod() == sfRequest::POST) {
//before the save...
$concept_property = $this->getRequestParameter('concept_property');
/**
* @todo the list of skos property types that require a related concept should be in a master configuration array
* this applies to the template too
**/
//check to see if the skosproperty requires a related concept
if (!in_array($concept_property['skos_property_id'], array('3', '16', '21', '32', '33', '34', '35', '36', '37'))) {
$concept_property['related_concept_id'] = null;
$concept_property['scheme_id'] = null;
}
$conceptPropertyId = $this->getRequestParameter('id');
$userId = $this->getUser()->getSubscriberId();
//does the user have editorial rights to the reciprocal...
$permission = false;
if (isset($concept_property['related_concept_id']) and $concept_property['related_concept_id']) {
//we want to lookup the URI of the related term
$related_concept = ConceptPeer::retrieveByPK($concept_property['related_concept_id']);
if ($related_concept) {
if ($this->getUser()->hasCredential(array(0 => 'administrator'))) {
$permission = true;
} else {
//get the maintainers of the reciprocal property
$maintainers = $related_concept->getVocabulary()->getVocabularyHasUsers();
/** @var VocabularyHasUser $maintainer */
foreach ($maintainers as $maintainer) {
if ($userId === $maintainer->getUserId() and $maintainer->getIsMaintainerFor()) {
$permission = true;
break;
}
}
}
}
}
if ($permission) {
if (isset($conceptPropertyId) && $conceptPropertyId) {
$this->deleteReciprocalProperty($conceptPropertyId, $concept_property['related_concept_id']);
}
if (isset($concept_property['related_concept_id']) and $concept_property['related_concept_id']) {
//we want to lookup the URI of the related term
$related_concept = ConceptPeer::retrieveByPK($concept_property['related_concept_id']);
if ($related_concept) {
//and overwrite whatever is in the current object TODO: move this into an javascript action in the user interface
$concept_property['object'] = $related_concept->getUri();
$this->getRequest()->getParameterHolder()->set('concept_property', $concept_property);
}
//lookup the inverse id
$InverseProfileId = ProfilePropertyPeer::retrieveBySkosID($concept_property['skos_property_id'])->getInverseProfilePropertyId();
$InverseSkosId = ProfilePropertyPeer::retrieveByPK($InverseProfileId)->getSkosId();
//then we create a new reciprocal property in the related term
$newProp = new ConceptProperty();
$newProp->setConceptId($concept_property['related_concept_id']);
$newProp->setSkosPropertyId($InverseSkosId);
$newProp->setSchemeId($this->concept->getVocabularyId());
$newProp->setRelatedConceptId($this->concept->GetId());
$newProp->setObject($this->concept->getUri());
$newProp->setStatusId($concept_property['status_id']);
$newProp->setIsGenerated(true);
$newProp->setCreatedUserId($this->getUser()->getSubscriberId());
$newProp->setUpdatedUserId($this->getUser()->getSubscriberId());
//TODO: make this the user's default language (actually the language is not relevant when defining relationships)
//$newProp->setLanguage($this->concept->getLanguage());
$newProp->setLanguage('');
$concept_property['language'] = '';
$newProp->save();
}
}
//save the array back to the request parameter
$this->requestParameterHolder->set('concept_property', $concept_property);
}
parent::executeEdit();
}