/**
* 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 $documentContent;
}
public function convertToSdlxliff($file_path, $source_lang, $target_lang, $chosen_by_user_machine = false, $segm_rule = null)
{
if (!file_exists($file_path)) {
throw new Exception("Conversion Error : the file <{$file_path}> not exists");
}
$fileContent = file_get_contents($file_path);
$extension = pathinfo($file_path, PATHINFO_EXTENSION);
$filename = pathinfo($file_path, PATHINFO_FILENAME);
if (strtoupper($extension) == 'TXT' or strtoupper($extension) == 'STRINGS') {
$encoding = mb_detect_encoding($fileContent);
//in case of .strings, they may be in UTF-16
if (strtoupper($extension) == 'STRINGS') {
//use this function to convert stuff
$convertedFile = CatUtils::convertEncoding('UTF-8', $fileContent);
//retrieve new content
$fileContent = $convertedFile[1];
} else {
if ($encoding != 'UTF-8') {
$fileContent = iconv($encoding, "UTF-8//IGNORE", $fileContent);
}
}
if (!$this->hasBOM($fileContent)) {
$fileContent = $this->addBOM($fileContent);
}
}
//get random name for temporary location
$tmp_name = tempnam("/tmp", "MAT_FW");
//write encoded file to temporary location
$fileSize = file_put_contents($tmp_name, $fileContent);
//assign file pointer for POST
$data['documentContent'] = "@{$tmp_name}";
//flush memory
unset($fileContent);
//assign converter
if (!$chosen_by_user_machine) {
$this->ip = $this->pickRandConverter($segm_rule);
} else {
$this->ip = $chosen_by_user_machine;
}
$url = "{$this->ip}:{$this->port}/{$this->toXliffFunction}";
$data['fileExtension'] = $extension;
$data['fileName'] = "{$filename}.{$extension}";
$data['sourceLocale'] = $this->lang_handler->getLangRegionCode($source_lang);
$data['targetLocale'] = $this->lang_handler->getLangRegionCode($target_lang);
log::doLog($this->ip . " start conversion to xliff of {$file_path}");
$start_time = microtime(true);
$curl_result = $this->curl_post($url, $data, $this->opt);
$end_time = microtime(true);
$time_diff = $end_time - $start_time;
log::doLog($this->ip . " took {$time_diff} secs for {$file_path}");
$this->conversionObject->ip_machine = $this->ip;
$this->conversionObject->ip_client = Utils::getRealIpAddr();
$this->conversionObject->path_name = $file_path;
$this->conversionObject->file_name = $data['fileName'];
$this->conversionObject->direction = 'fw';
$this->conversionObject->src_lang = $data['sourceLocale'];
$this->conversionObject->trg_lang = $data['targetLocale'];
$this->conversionObject->file_size = $fileSize;
$this->conversionObject->conversion_time = $time_diff;
$decode = json_decode($curl_result, true);
$curl_result = null;
$res = $this->__parseOutput($decode);
//remove temporary file
unlink($tmp_name);
return $res;
}
