$fwstk->execute();
}
- public static function extractTexts($pdf, $out, $ignoreSeparators = '', $force = true)
+ public static function extractTexts($pdf, $out, $textExtraction, $ignoreSeparators = '')
{
$out .= 'texts/';
- Files::mkdir($out);
-
- if (!$force && file_exists($out . 'fp1.txt')) {
- return;
+ if ($ignoreSeparators) {
+ $out .= 'sep_' . md5($ignoreSeparators);
}
+ $out = Files::mkdir($out) . '/';
$fwstk = new FWSTK();
$fwstk->setArg('--input ' . $pdf);
$fwstk->setArg('--extractTexts ' . $out . '%s%d.txt');
- $fwstk->setArg('--extractTextsMethod fluidbook');
+ $fwstk->setArg('--extractTextsMethod' . $textExtraction);
$fwstk->setArg('--threads 1');
if ($ignoreSeparators !== '') {
$fwstk->setArg('--ignoreSeparators ' . $ignoreSeparators);