def extract_from_file(sender, instance, **kwargs):
if Tika.ping():
tika_handle = Tika.from_file(instance.original_file.path)
doc_text = UnicodeDammit(tika_handle.text()).unicode_markup
instance.extracted_text = doc_text
instance.extracted_html = tika_handle.html()
instance.metadata = tika_handle.meta()
def tika_handle(self):
return (Tika.ping() and self.path) \
and Tika.from_file(self.path) \
or None