#!/usr/bin/env python3

import sys, re
import urllib.request

if '-h' in sys.argv or len(sys.argv) < 2:
  print('./url_read.py op webpage_URL')
  print()
  print('reads the HTML of a webpage')
  print()
  print('options: -s = strip HTML tags')
  sys.exit()

if 'http' not in sys.argv[-1]:
  req='https://'+sys.argv[-1]
else:
  req=sys.argv[-1]

response=urllib.request.urlopen(req)
lines=response.read().decode('utf8')

for line in lines.split('\n'):
  if sys.argv[1] == '-s':
    print(re.sub('<.*?>','',line))
  else:
    print(line)