public static function processObjectFiles($assetsID, $options)
{
// Disable PHP's max execution time
set_time_limit(0);
$saveBase = mfcs::config('convertedPath');
$originalsFilepath = self::getSaveDir($assetsID, 'archive');
$originalFiles = scandir($originalsFilepath);
// Setup return array
$return = array('processed' => array(), 'combine' => array(), 'thumbs' => array(), 'ocr' => array());
// Remove dot files from array
foreach ($originalFiles as $I => $filename) {
if ($filename[0] == '.') {
unset($originalFiles[$I]);
}
}
// Needed to put the files in the right order for processing
if (natcasesort($originalFiles) === FALSE) {
return FALSE;
}
try {
// If combine files is checked, read this image and add it to the combined object
if (isset($options['combine']) && str2bool($options['combine'])) {
try {
$errors = array();
$createThumb = TRUE;
// Create us some temp working space
$tmpDir = mfcs::config('mfcstmp') . DIRECTORY_SEPARATOR . uniqid();
mkdir($tmpDir, 0777, TRUE);
// Ensure that the HOCR file is created
if (!self::createHOCR("{$saveBase}/hocr.cfg")) {
return FALSE;
}
$gsTemp = $tmpDir . DIRECTORY_SEPARATOR . uniqid();
touch($gsTemp);
foreach ($originalFiles as $filename) {
// Figure some stuff out about the file
$originalFile = $originalsFilepath . DIRECTORY_SEPARATOR . $filename;
$_filename = pathinfo($originalFile);
$filename = $_filename['filename'];
$baseFilename = $tmpDir . DIRECTORY_SEPARATOR . $filename;
// Create a thumbnail of the first image
if ($createThumb === TRUE) {
if (($return['combine'][] = self::createThumbnail($originalFile, $filename, $options, $assetsID, TRUE)) === FALSE) {
throw new Exception("Failed to create thumbnail: " . $filename);
}
// Prevent making multiple thumbnails
$createThumb = FALSE;
}
// perform hOCR on the original uploaded file which gets stored in combined as an HTML file
$_exec = shell_exec(sprintf('tesseract %s %s -l eng %s 2>&1', escapeshellarg($originalFile), escapeshellarg($baseFilename), escapeshellarg("{$saveBase}/hocr.cfg")));
// If a new-line char is in the output, assume it's an error
// Tesseract failed, let's normalize the image and try again
if (strpos(trim($_exec), "\n") !== FALSE) {
$errors[] = "Unable to process OCR for " . basename($originalFile) . ". Continuing…";
errorHandle::warningMsg("Unable to process OCR for " . basename($originalFile) . ". Continuing…");
// Ensure HTML file exists
touch($baseFilename . ".html");
}
// Create an OCR'd pdf of the file
$_exec = shell_exec(sprintf('hocr2pdf -i %s -s -o %s < %s 2>&1', escapeshellarg($originalFile), escapeshellarg($baseFilename . ".pdf"), escapeshellarg($baseFilename . ".html")));
if (trim($_exec) !== 'Writing unmodified DCT buffer.') {
if (strpos($_exec, 'Warning:') !== FALSE) {
errorHandle::newError("hocr2pdf Warning: " . $_exec, errorHandle::DEBUG);
} else {
errorHandle::errorMsg("Failed to Create PDF: " . basename($filename, "jpg") . ".pdf");
throw new Exception("hocr2pdf Error: " . $_exec);
}
}
// Add this pdf to a temp file that will be read in by gs
file_put_contents($gsTemp, $baseFilename . ".pdf" . PHP_EOL, FILE_APPEND);
// We're done with this file, delete it
unlink($baseFilename . ".html");
}
// Combine all PDF files in directory
$_exec = shell_exec(sprintf('gs -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook -dNOPAUSE -dQUIET -dBATCH -sOutputFile=%s @%s 2>&1', self::getSaveDir($assetsID, 'combine') . "combined.pdf", $gsTemp));
if (!is_empty($_exec)) {
errorHandle::errorMsg("Failed to combine PDFs into single PDF.");
throw new Exception("GhostScript Error: " . $_exec);
}
$return['combine'][] = array('name' => 'combined.pdf', 'path' => self::getSaveDir($assetsID, 'combine', FALSE), 'size' => filesize(self::getSaveDir($assetsID, 'combine') . 'combined.pdf'), 'type' => 'application/pdf', 'errors' => $errors);
// Lastly, we delete our temp working dir (always nice to cleanup after yourself)
if (self::cleanupTempDirectory($tmpDir) === FALSE) {
errorHandle::errorMsg("Unable to clean up temporary directory: " . $tmpDir);
throw new Exception("Unable to clean up temporary directory: " . $tmpDir);
}
} catch (Exception $e) {
// We need to delete our working dir
if (isset($tmpDir) && is_dir($tmpDir)) {
if (self::cleanupTempDirectory($tmpDir) === FALSE) {
errorHandle::errorMsg("Unable to clean up temporary directory (in Exception): " . $tmpDir);
}
}
throw new Exception($e->getMessage(), $e->getCode(), $e);
}
}
// If Combine
// This conditional needs updated when different conversion options are added or removed.
// If the file has no processing to do, don't do any ...
if (!isset($options['convert']) && !isset($options['thumbnail']) && !isset($options['ocr']) && !isset($options['mp3'])) {
return $return;
}
foreach ($originalFiles as $filename) {
$originalFile = $originalsFilepath . DIRECTORY_SEPARATOR . $filename;
$_filename = pathinfo($originalFile);
$filename = $_filename['filename'];
// Convert uploaded files into some ofhter size/format/etc
if (isset($options['convert']) && str2bool($options['convert'])) {
// we create the Imagick object here so that we can pass it to thumbnail creation
$image = new Imagick();
$image->readImage($originalFile);
// Convert it
if (($image = self::convertImage($image, $options, $assetsID, $filename)) === FALSE) {
throw new Exception("Failed to create processed image: " . $originalFile);
}
$filename = $filename . '.' . strtolower($image->getImageFormat());
// Create a thumbnail that includes converted options
if (isset($options['thumbnail']) && str2bool($options['thumbnail'])) {
if (($return['thumbs'][] = self::createThumbnail($image, $filename, $options, $assetsID)) === FALSE) {
throw new Exception("Failed to create thumbnail: " . $filename);
}
}
// Set the return array
$return['processed'][] = array('name' => $filename, 'path' => self::getSaveDir($assetsID, 'processed', FALSE), 'size' => filesize(self::getSaveDir($assetsID, 'processed') . $filename), 'type' => self::getMimeType(self::getSaveDir($assetsID, 'processed') . $filename), 'errors' => '');
} else {
if (isset($options['thumbnail']) && str2bool($options['thumbnail'])) {
if (($return['thumbs'][] = self::createThumbnail($originalFile, $filename, $options, $assetsID)) === FALSE) {
throw new Exception("Failed to create thumbnail: " . $filename);
}
}
}
// Create an OCR text file
if (isset($options['ocr']) && str2bool($options['ocr'])) {
if (($return['ocr'][] = self::createOCRTextFile($originalFile, $assetsID, $filename)) === FALSE) {
errorHandle::errorMsg("Failed to create OCR text file: " . $filename);
throw new Exception("Failed to create OCR file for {$filename}");
}
}
// Create MP3
if (isset($options['mp3']) && str2bool($options['mp3'])) {
$return['mp3'][] = self::createMP3($originalFile);
}
}
// Foreach File
} catch (Exception $e) {
errorHandle::newError(__METHOD__ . "() - {$e->getMessage()} {$e->getLine()}:{$e->getFile()}", errorHandle::HIGH);
}
return $return;
}
