/**
* Save file attributes from File
* @param File $file
* @param boolean $update
*/
private function setAutoFileAttributes($file, $update = false)
{
// We download the file first in /tmp/ folder
if (!file_exists(ReadFile::TEMP_FOLDER . $file->name)) {
ReadFile::downloadRemoteFile($file->location, $file->name);
}
// from update page, we delete all auto attributes and make them again
if ($update) {
$criteria = new CDbCriteria();
$criteria->addCondition('file_id = ' . $file->id);
$fileAttributes = FileAttributes::model()->findAll($criteria);
foreach ($fileAttributes as $fileAttribute) {
if (strpos(Attribute::AUTO_ATTRIBUTE, $fileAttribute->attribute->structured_comment_name)) {
$fileAttribute->delete();
}
}
}
$fileAttribute = new FileAttributes();
$fileAttribute->file_id = $file->id;
/*$name = strpos($file->name, '.') ? explode('.', $file->name) : $file->name;
$name = is_array($name) ? $name[0] : $name;*/
$extension = str_replace('.', '', $file->extension);
switch ($extension) {
case 'fasta':
case 'fa':
case 'fastq':
case 'seq':
case 'bam':
case 'sam':
case 'cram':
// Launche the python script to get the result
list($aminoAcids, $nucleotides) = ReadFile::readPythonFile($file->name);
// Number of Amino Acids
$numberAminoAcids = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_amino_acids'));
$numberAminoAcids->attribute_id = $attribute->id;
$numberAminoAcids->value = $aminoAcids;
$numberAminoAcids->save();
// Number of nucleotides
$numberNucleotides = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_nucleotides'));
$numberNucleotides->value = $nucleotides;
$numberNucleotides->attribute_id = $attribute->id;
$numberNucleotides->save();
break;
case 'txt':
case 'text':
case 'doc':
case 'docx':
case 'rtf':
case 'odt':
case 'wpd':
case 'lwd':
case 'readme':
case 'README':
case 'pdf':
case '':
$result = null;
if (in_array($extension, array('txt', 'text', 'readme', 'README', '', 'wpd', 'lwd'))) {
$result = ReadFile::readTextFile($file->name);
}
if (in_array($extension, array('docx', 'odt'))) {
$result = ReadFile::readDocxOdtFile($file->name);
}
if ($extension == 'doc') {
$result = ReadFile::readDocFile($file->name);
}
if ($extension == 'rtf') {
$result = ReadFile::readRtfFile($file->name);
}
if ($extension == 'pdf') {
$result = ReadFile::readPdfFile($file->name);
}
$numberOfLines = $numberOfWords = 0;
$numberOfWords = str_word_count($result);
$numberOfLines = substr_count($result, "\n");
// Number of words
$numberWords = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_words'));
$numberWords->attribute_id = $attribute->id;
$numberWords->value = $numberOfWords;
$numberWords->save();
// Number of lines
$numberLines = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_lines'));
$numberLines->attribute_id = $attribute->id;
$numberLines->value = $numberOfLines;
$numberLines->save();
break;
case 'csv':
case 'tsv':
case 'xls':
case 'xlsx':
case 'tab':
case 'sdc':
case 'ods':
case 'gff':
case 'gff3':
case 'snp':
case 'vcf':
case 'ipr':
$rows = $columns = 0;
if ($extension == 'xls') {
list($rows, $columns) = ReadFile::readXlsFile($file->name);
}
if ($extension == 'ods') {
list($rows, $columns) = ReadFile::readOdsFile($file->name);
}
if ($extension == 'xlsx') {
list($rows, $columns) = ReadFile::readXlsxFile($file->name);
}
if (in_array($extension, array('csv', 'tsv', 'vcf', 'gff'))) {
list($rows, $columns) = ReadFile::readTableFile($file->name, $extension);
}
// Number of rows
$numberRows = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_rows'));
$numberRows->attribute_id = $attribute->id;
$numberRows->value = $rows;
$numberRows->save();
// Number of columns
$numberColumns = clone $fileAttribute;
$attribute = Attribute::model()->findByAttributes(array('structured_comment_name' => 'num_columns'));
$numberColumns->value = $columns;
$numberColumns->attribute_id = $attribute->id;
$numberColumns->save();
break;
}
}
