def job(x):
# document conversion
global pdf2txt
global output_dir
global count
global upload
y = x.replace('.pdf', '.txt')
if output_dir:
y = output_dir + y.split('/')[-1]
if not os.path.isfile(y):
if output_dir:
os.system('python3 {} {} > {}'.format(pdf2txt, x, y))
else:
os.system('python3 {} {}'.format(pdf2txt, x))
if os.stat(y).st_size <= MIN_BYTES:
logging.info('{}: File Size unsatisfactory. Performing OCR'.format(x))
ocr.pdfocr2txt(x, y, resolution=resolution, tmp=tmp)
logging.info('{} Done'.format(x))
else:
logging.info('{} already a converted file'.format(x))
count.value += 1
logging.info('Complete {} out of {}'.format(int(count.value), total))
return y
def pdf_to_txt(input_pdf, output_txt):
# document conversion
try:
global pdf2txt
if not os.path.isfile(output_txt):
if os.path.isfile(input_pdf):
os.system('python3 {} {} > {}'.format(
pdf2txt, input_pdf, output_txt))
else:
logging.info('{}: File not found!'.format(input_pdf))
return False
if os.stat(output_txt).st_size <= MIN_BYTES:
logging.info(
'{}: File Size unsatisfactory. Performing OCR'.format(input_pdf))
ocr.pdfocr2txt(input_pdf, output_txt,
resolution=resolution, tmp=tmp)
logging.info('{} Done'.format(input_pdf))
return True
else:
logging.info('{} already a converted file'.format(input_pdf))
return False
except Exception as e:
logging.error(
"Exception occurred while converting file %s to txt", input_pdf, exc_info=True)
return False