def _parse_pages(self, book_path):
result = []
start = time.time()
with open(book_path, 'rb') as fh:
input_pdf = PdfFileReader(fh)
self._log.info('Start processing %s with %s pages...', book_path,
input_pdf.getNumPages())
if input_pdf.flattenedPages is None:
input_pdf._flatten()
for page_num, page in enumerate(input_pdf.flattenedPages, start=1):
result.append({
'page_num': str(page_num),
'text': page.extractText()
})
self._log.info('Finished processing %s in %s seconds.', book_path,
time.time() - start)
return result
