/**
* get Content of DOC file
*
* @param string $file
* @return string The extracted content of the file
*/
public function getContent($file)
{
// create the tempfile which will contain the content
if (TYPO3_VERSION_INTEGER >= 7000000) {
$tempFileName = TYPO3\CMS\Core\Utility\GeneralUtility::tempnam('doc_files-Indexer');
} else {
$tempFileName = t3lib_div::tempnam('doc_files-Indexer');
}
// Delete if exists, just to be safe.
@unlink($tempFileName);
// generate and execute the pdftotext commandline tool
$cmd = $this->app['catdoc'] . ' -s8859-1 -dutf-8 ' . escapeshellarg($file) . ' > ' . escapeshellarg($tempFileName);
if (TYPO3_VERSION_INTEGER >= 7000000) {
TYPO3\CMS\Core\Utility\CommandUtility::exec($cmd);
} else {
t3lib_utility_Command::exec($cmd);
}
// check if the tempFile was successfully created
if (@is_file($tempFileName)) {
if (TYPO3_VERSION_INTEGER >= 7000000) {
$content = TYPO3\CMS\Core\Utility\GeneralUtility::getUrl($tempFileName);
} else {
$content = t3lib_div::getUrl($tempFileName);
}
unlink($tempFileName);
} else {
return false;
}
// check if content was found
if (strlen($content)) {
return $content;
} else {
return false;
}
}
/**
* get Content of DOC file
*
* @param string $file
* @return string The extracted content of the file
*/
public function getContent($file)
{
// create the tempfile which will contain the content
$tempFileName = TYPO3\CMS\Core\Utility\GeneralUtility::tempnam('xls_files-Indexer');
// Delete if exists, just to be safe.
@unlink($tempFileName);
// generate and execute the pdftotext commandline tool
$cmd = $this->app['xls2csv'] . ' -c \' \' -q 0 -s8859-1 -dutf-8 ' . escapeshellarg($file) . ' > ' . escapeshellarg($tempFileName);
TYPO3\CMS\Core\Utility\CommandUtility::exec($cmd);
// check if the tempFile was successfully created
if (@is_file($tempFileName)) {
$content = TYPO3\CMS\Core\Utility\GeneralUtility::getUrl($tempFileName);
unlink($tempFileName);
} else {
return false;
}
// check if content was found
if (strlen($content)) {
return $content;
} else {
return false;
}
}
/**
* get Content of PDF file
*
* @param string $file
* @return string The extracted content of the file
*/
public function getContent($file)
{
$this->fileInfo = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance('tx_kesearch_lib_fileinfo');
$this->fileInfo->setFile($file);
// get PDF informations
if (!($pdfInfo = $this->getPdfInfo($file))) {
return false;
}
// proceed only of there are any pages found
if (intval($pdfInfo['pages']) && $this->isAppArraySet) {
// create the tempfile which will contain the content
$tempFileName = TYPO3\CMS\Core\Utility\GeneralUtility::tempnam('pdf_files-Indexer');
// Delete if exists, just to be safe.
@unlink($tempFileName);
// generate and execute the pdftotext commandline tool
$cmd = $this->app['pdftotext'] . ' -enc UTF-8 -q ' . escapeshellarg($file) . ' ' . escapeshellarg($tempFileName);
TYPO3\CMS\Core\Utility\CommandUtility::exec($cmd);
// check if the tempFile was successfully created
if (@is_file($tempFileName)) {
$content = TYPO3\CMS\Core\Utility\GeneralUtility::getUrl($tempFileName);
unlink($tempFileName);
} else {
$this->addError('Content for file ' . $file . ' could not be extracted. Maybe it is encrypted?');
// return empty string if no content was found
$content = '';
}
return $this->removeEndJunk($content);
} else {
return false;
}
}