/**
* @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)
}
