/**
*/
public function download_and_parse()
{
$this->download_if_necessary();
$absolute_filename = $this->getLocalPath();
if (!file_exists($absolute_filename)) {
echo "Not Found: " . $this->id . "\n";
return;
}
$y = explode(".", $this->url);
$endung = mb_strtolower($y[count($y) - 1]);
$metadata = RISPDF2Text::document_pdf_metadata($absolute_filename);
$this->seiten_anzahl = $metadata["seiten"];
$this->datum_dokument = $metadata["datum"];
if ($this->datum_dokument == "") {
$this->datum_dokument = null;
}
if ($endung == "pdf") {
$this->text_pdf = RISPDF2Text::document_text_pdf($absolute_filename);
} else {
$this->text_pdf = "";
}
$this->text_ocr_raw = RISPDF2Text::document_text_ocr($absolute_filename, $this->seiten_anzahl);
$this->text_ocr_corrected = RISPDF2Text::ris_ocr_clean($this->text_ocr_raw);
$this->ocr_von = Dokument::$OCR_VON_TESSERACT;
copy($absolute_filename, OMNIPAGE_PDF_DIR . $this->id . "." . $endung);
}
