mirror of
https://github.com/silverstripe/silverstripe-textextraction
synced 2024-06-27 23:19:39 +02:00
40 lines
1002 B
PHP
40 lines
1002 B
PHP
<?php
|
|
|
|
/**
|
|
* Text extractor that calls pdftotext to do the conversion.
|
|
* @author mstephens
|
|
*
|
|
*/
|
|
class PDFTextExtractor extends FileTextExtractor {
|
|
|
|
function isAvailable() {
|
|
$bin = $this->bin('pdftotext');
|
|
return (file_exists($bin) && is_executable($bin));
|
|
}
|
|
|
|
function supportedExtensions() {
|
|
return array("pdf");
|
|
}
|
|
|
|
/**
|
|
* Accessor to get the location of the binary
|
|
* @param $prog
|
|
* @return unknown_type
|
|
*/
|
|
function bin($prog='') {
|
|
if ($this->stat('binary_location')) $path = $this->stat('binary_location'); // By static from _config.php
|
|
elseif (file_exists('/usr/bin/pdftotext')) $path = '/usr/bin'; // By searching common directories
|
|
elseif (file_exists('/usr/local/bin/pdftotext')) $path = '/usr/local/bin';
|
|
else $path = '.'; // Hope it's in path
|
|
|
|
return ( $path ? $path . '/' : '' ) . $prog;
|
|
}
|
|
|
|
function getContent($path) {
|
|
if (!$path) return ""; // no file
|
|
$content = `{$this->bin('pdftotext')} "$path" -`;
|
|
return $content;
|
|
}
|
|
}
|
|
|
|
?>
|