|
|
|
@ -40,8 +40,8 @@ function islandora_text_extraction_media_presave(MediaInterface $media) {
|
|
|
|
|
$file = File::load($file_id); |
|
|
|
|
if ($file) { |
|
|
|
|
$data = file_get_contents($file->getFileUri()); |
|
|
|
|
// Check if it's already markup like hOCR |
|
|
|
|
if (substr($data, 0, 4) == '<xml') { |
|
|
|
|
// Check if it's already markup like hOCR. |
|
|
|
|
if (substr($data, 0, 5) == '<?xml') { |
|
|
|
|
return; |
|
|
|
|
} |
|
|
|
|
$data = nl2br($data); |
|
|
|
|