collate/codepoint.py

   1 """Codepoint-based collation.
   2
   3 This collation backend sorts using only the basic codepoint order. It
   4 is primarily intended to be used as a baseline and example for other
   5 collation backends.
   6
   7 Use this collation backend if...
   8  - You are writing tests for pycollate.
   9  - You are writing specialized Unicode software.
  10  - You are on a system with no locale module.
  11
  12 Avoid this backend if...
  13  - You are writing a normal program for a normal runtime environment.
  14  - You are sorting strings to show normal humans.
  15
  16 """
  17
  18 import collate._abcollator
  19 import collate._locale
  20
  21 __all__ = ["Collate"]
  22
  23 class Collator(collate._abcollator.Collator):
  24     """Codepoint-based collation.
  25
  26     Arguments
  27     locale - all parts but encoding ignored, always 'C'
  28     encoding - try to use this string encoding
  29     """
  30
  31     def __init__(self, locale=None, encoding=None):
  32         dummy, self.encoding = collate._locale.getpair(locale, encoding)
  33         self.locale = "C"
  34
  35     def key(self, string):
  36         """Sort key for a string.
  37
  38         If string is a str instance, it is first decoded according to
  39         the 'encoding' attribute of the Collator.
  40         """
  41         if isinstance(string, str):
  42             string = string.decode(self.encoding, 'replace')
  43         return string