test/html2txt.py

   1 #! /usr/bin/env python
   2 from __future__ import print_function
   3
   4 import sys
   5 from m_lib.net.www.html import HTMLFilter
   6
   7 PY2 = sys.version_info[0] < 3
   8 if PY2:
   9     with open(sys.argv[1], 'r') as f:
  10         html = f.read()
  11 else:
  12     with open(sys.argv[1], 'r', encoding='utf-8') as f:
  13         html = f.read()
  14
  15 filter = HTMLFilter()
  16 filter.feed(html)
  17 filter.close()
  18
  19 if PY2:
  20     print(filter.accumulator)
  21 else:
  22     if not isinstance(filter.accumulator, bytes):
  23         filter.accumulator = filter.accumulator.encode('utf-8')
  24     sys.stdout.buffer.write(filter.accumulator + b'\n')