/**
* Backwards compatibility method and forward
*
* Works by Reference variable
*
* @param $fileMetaData [
* "id_file",
* "filename",
* "source",
* "mime_type",
* "sha1_original_file"
* ]
*/
private function migrateFileDB2FS(&$fileMetaData)
{
//create temporary storage to place stuff
$tempDir = "/tmp" . DIRECTORY_SEPARATOR . uniqid("", true);
mkdir($tempDir, 0755);
//fetch xliff from the files database
$xliffContent = $this->getXliffFromDB($fileMetaData['id_file']);
//try pulling the original content too (if it's empty it means that it was an unconverted xliff)
$fileContent = $this->getOriginalFromDB($fileMetaData['id_file']);
if (!empty($fileContent)) {
//it's a converted file
//i'd like to know it's real extension....
//create temporary file with appropriately modified name
$result = DetectProprietaryXliff::getInfoByStringData($xliffContent);
if ($result['proprietary_short_name'] == 'trados') {
$tempXliff = $tempDir . DIRECTORY_SEPARATOR . $fileMetaData['filename'] . ".sdlxliff";
} else {
$tempXliff = $tempDir . DIRECTORY_SEPARATOR . $fileMetaData['filename'] . ".xlf";
}
//create file
$tempOriginal = $tempDir . DIRECTORY_SEPARATOR . $fileMetaData['filename'];
//flush content
file_put_contents($tempOriginal, $fileContent);
//get hash, based on original
$sha1 = sha1($fileContent);
//free memory
unset($fileContent);
} else {
//if it's a unconverted xliff
//create temporary file with original name
$tempXliff = $tempDir . DIRECTORY_SEPARATOR . $fileMetaData['filename'];
// set original to empty
$tempOriginal = false;
//get hash
$sha1 = sha1($xliffContent);
}
//flush xliff file content
file_put_contents($tempXliff, $xliffContent);
//free memory
unset($xliffContent);
if (stripos($fileMetaData['sha1_original_file'], DIRECTORY_SEPARATOR) === false) {
$query = "select create_date from projects where id = {$fileMetaData['id_project']}";
$db = Database::obtain();
$results = $db->fetch_array($query);
$dateHashPath = date_create($results[0]['create_date'])->format('Ymd') . DIRECTORY_SEPARATOR . $sha1;
$db->update('files', array("sha1_original_file" => $dateHashPath), 'id = ' . $fileMetaData['id_file']);
//update Reference
$fileMetaData['sha1_original_file'] = $dateHashPath;
}
//build a cache package
$this->makeCachePackage($sha1, $fileMetaData['source'], $tempOriginal, $tempXliff);
//build a file package
$this->moveFromCacheToFileDir($fileMetaData['sha1_original_file'], $fileMetaData['source'], $fileMetaData['id_file']);
//clean temporary stuff
Utils::deleteDir($tempDir);
}
/**
* Remove the tag mrk if the file is an xlif and if the file is a globalsight file
*
* Also, check for encoding and transform utf16 to utf8 and back
*
* @param $documentContent
* @param $path
*
* @return string
*/
public function ifGlobalSightXliffRemoveTargetMarks($documentContent, $path)
{
$extension = FilesStorage::pathinfo_fix($path);
if (!DetectProprietaryXliff::isXliffExtension($extension)) {
return $documentContent;
}
$is_utf8 = true;
$original_charset = 'utf-8';
//not used, useful only to avoid IDE warning for not used variable
//The file is UTF-16 Encoded
if (stripos(substr($documentContent, 0, 100), "<?xml ") === false) {
$is_utf8 = false;
list($original_charset, $documentContent) = CatUtils::convertEncoding('UTF-8', $documentContent);
}
//avoid in memory copy of very large files if possible
$detect_result = DetectProprietaryXliff::getInfoByStringData(substr($documentContent, 0, 1024));
//clean mrk tags for GlobalSight application compatibility
//this should be a sax parser instead of in memory copy for every trans-unit
if ($detect_result['proprietary_short_name'] == 'globalsight') {
// Getting Trans-units
$trans_units = explode('<trans-unit', $documentContent);
foreach ($trans_units as $pos => $trans_unit) {
// First element in the XLIFF split is the header, not the first file
if ($pos > 0) {
//remove seg-source tags
$trans_unit = preg_replace('|<seg-source.*?</seg-source>|si', '', $trans_unit);
//take the target content
$trans_unit = preg_replace('#<mrk[^>]+>|</mrk>#si', '', $trans_unit);
$trans_units[$pos] = $trans_unit;
}
}
// End of trans-units
$documentContent = implode('<trans-unit', $trans_units);
}
if (!$is_utf8) {
list($__utf8, $documentContent) = CatUtils::convertEncoding($original_charset, $documentContent);
}
return $documentContent;
}
