private function generate($task_id, $filepath = false)
{
$task = $this->Task->getById($task_id);
if ($task === false || in_array($task->ttid, $this->task_types) === false) {
die('Task not providing datasplits.');
}
$values = $this->Task_inputs->getTaskValuesAssoc($task_id);
$estimation_procedure = null;
if (array_key_exists('estimation_procedure', $values)) {
$estimation_procedure = $this->Estimation_procedure->getById($values['estimation_procedure']);
}
if ($estimation_procedure == false) {
die('estimation procedure not found');
}
$dataset = $this->Dataset->getById($values['source_data']);
$epstr = $this->Estimation_procedure->toString($estimation_procedure);
$target_feature = $task->ttid == 7 ? $values['target_feature_event'] : $values['target_feature'];
// TODO: very important. sanity check input
$testset_str = array_key_exists('custom_testset', $values) && is_cs_natural_numbers($values['custom_testset']) ? '-test "' . $values['custom_testset'] . '"' : '';
$command = 'java -jar ' . $this->evaluation . ' -f "generate_folds" -d "' . $dataset->url . '" -e "' . $epstr . '" -c "' . $target_feature . '" -r "' . safe($dataset->row_id_attribute) . '" ' . $testset_str . $this->config;
if (array_key_exists('custom_testset', $values)) {
$command .= '-test "' . $values['custom_testset'] . '" ';
}
if ($filepath) {
$command .= ' -o ' . $filepath;
}
//if( $md5 ) $command .= ' -m';
$this->Log->cmd('API Splits::get(' . $task_id . ')', $command);
if (function_enabled('system')) {
header('Content-type: text/plain');
system(CMD_PREFIX . $command);
} else {
die('failed to generate arff file: php "system" function disabled. ');
}
}
/**
* given a file, return the mime type
*
* @param string $pExtension is the extension of the file or the complete file name
* @return mime type of entry and populates $this->mMimeTypes with existing mime types
* @access public
*/
function verifyMimeType($pFile)
{
$mime = NULL;
if (file_exists($pFile)) {
if (function_exists('finfo_open')) {
if (is_windows() && defined('PHP_MAGIC_PATH') && is_readable(PHP_MAGIC_PATH)) {
$finfo = finfo_open(FILEINFO_MIME, PHP_MAGIC_PATH);
} else {
$finfo = finfo_open(FILEINFO_MIME);
}
$mime = finfo_file($finfo, $pFile);
finfo_close($finfo);
} else {
if (function_enabled("escapeshellarg") && function_enabled("exec")) {
$mime = exec(trim('file -bi ' . escapeshellarg($pFile)));
}
}
if (empty($mime)) {
$mime = $this->lookupMimeType(substr($pFile, strrpos($pFile, '.') + 1));
}
if ($len = strpos($mime, ';')) {
$mime = substr($mime, 0, $len);
}
}
return $mime;
}
function validate_arff($to_folder, $filepath, $name, $did)
{
$ci =& get_instance();
$weka = PATH . APPPATH . 'third_party/OpenML/Java/weka.jar';
$res = array();
$code = 0;
$oldUrl = DATA_PATH . $filepath;
$newpath = strrev(implode(strrev('/openml_'), explode('/', strrev($filepath), 2)));
$newUrl = DATA_PATH . $newpath;
$heap = '-Xmx' . ($ci->input->is_cli_request() ? $ci->config->item('java_heap_space_cli') : $ci->config->item('java_heap_space_web'));
$command = "java {$heap} -cp {$weka} weka.core.converters.ArffLoader {$oldUrl} > {$newUrl}";
$ci->Log->cmd('ARFF Validation', $command);
if (function_enabled('exec') === false) {
return false;
}
exec(CMD_PREFIX . $command, $res, $code);
//Guess the id of the dataset and add it to the top of the file
$info = '% Data set "' . $name . '". For more information, see http:\\/\\/openml.org\\/d\\/' . $did;
$string = '1s/^/' . $info . '\\n/';
$command2 = "sed -i -e '{$string}' {$newUrl}";
exec(CMD_PREFIX . $command2, $res, $code);
if ($code == 0) {
return $newpath;
} else {
return false;
}
}
function owner($path)
{
$user = fileowner($path);
$group = filegroup($path);
$data = $user;
if (function_enabled('posix_getpwuid')) {
$u = posix_getpwuid($user);
$data .= " (" . $u["name"] . ")";
}
$data .= " <br> " . $group;
if (function_enabled('posix_getgrgid')) {
$g = posix_getgrgid($group);
$data .= " (" . $g["name"] . ") ";
}
return $data;
}
