skripte/python/lang_s/long_s_quasihyph.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf8 -*-
   3 # :Copyright: © 2011 Günter Milde.
   4 #             Released without warranty under the terms of the
   5 #             GNU General Public License (v. 2 or later)
   6 # :Id: $Id:  $
   7
   8 # long_s_quasihyph.py:
   9 # ============================================================
  10
  11 u"""Filter zum Wandeln von Wörtern mit langem S in
  12 Pseudo-Trennbeispiele (ausſagen -> auss-agen, eſſen -> es-s-en)."""
  13
  14 import sys, optparse, re
  15
  16
  17 usage = u'%prog [Optionen]\n' + __doc__
  18 parser = optparse.OptionParser(usage=usage)
  19 parser.add_option('-e', '--encoding', dest='encoding',
  20                     help=u'Kodierung der Ausgabe, '
  21                     u'Vorgabe "iso-8859-15"',
  22                     default='iso-8859-15')
  23
  24 (options, args) = parser.parse_args()
  25
  26 lines = (line.decode('utf-8') for line in sys.stdin)
  27
  28 words = (word.replace(u'ſ', u's-').replace(u'S', u's-')
  29          for word in lines)
  30
  31 sys.stdout.writelines(line.encode(options.encoding) for line in words)