pycollate

   1 #!/usr/bin/env python
   2
   3 import os
   4 import sys
   5 import locale
   6
   7 from optparse import OptionParser
   8
   9 import collate
  10
  11 def main(argv):
  12     lines = []
  13     encoding = locale.getpreferredencoding()
  14     try:
  15         locale.setlocale(locale.LC_ALL, '')
  16     except locale.Error:
  17         pass
  18
  19     parser = OptionParser()
  20     parser.add_option("-b", "--backend", dest="backend",
  21                       help="use backend BACKEND", metavar="BACKEND",
  22                       default="default")
  23     parser.add_option("-l", "--locale", dest="locale",
  24                       help="use locale LOCALE", metavar="LOCALE",
  25                       default=locale.getlocale()[0])
  26     parser.add_option("-e", "--encoding", dest="encoding",
  27                       help="use encoding ENCODING", metavar="ENCODING",
  28                       default=encoding)
  29     parser.add_option("-d", "--diagnostics", action="store_true",
  30                       dest="diagnostics", help="print diagnostic information")
  31
  32     options, args = parser.parse_args()
  33
  34     if options.backend or options.locale or options.encoding:
  35         backend = getattr(collate, options.backend)
  36         collate.set_default(backend, options.locale, options.encoding)
  37
  38     if options.diagnostics:
  39         print >>sys.stderr, "Backend: " + collate.default.__name__
  40         print >>sys.stderr, "Locale: " + collate.collator.locale
  41         print >>sys.stderr, "Encoding: " + collate.collator.encoding
  42
  43         if not args:
  44             raise SystemExit
  45
  46     if not args:
  47         args.append("-")
  48     for filename in args:
  49         if filename == "-":
  50             fileobj = sys.stdin
  51         else:
  52             fileobj = open(filename, "rU")
  53         for line in fileobj:
  54             line = line.strip()
  55             line = line.decode(encoding, "replace")
  56             lines.append(line)
  57     lines.sort(key=collate.collator.sortemekey)
  58
  59     for line in lines:
  60         print line.encode(encoding, "replace")
  61
  62 if __name__ == "__main__":
  63     main(sys.argv[1:])