collate/_abcollator.py

   1 """Abstract base collator."""
   2
   3 import collate.strings
   4
   5 class Collator(object):
   6     """Abstract base class for Collators.
   7
   8     Attributes:
   9     locale - the collator follows rules for this locale
  10     encoding - assumed string encoding
  11     """
  12
  13     locale = "C"
  14     encoding = "ascii"
  15
  16     def __init__(self, locale=None, encoding=None):
  17         pass
  18
  19     def cmp(self, string1, string2):
  20         """Return negative if a < b, zero if a == b, positive if a > b."""
  21         return cmp(self.key(string1), self.key(string2))
  22
  23     def key(self, string):
  24         """Return a good sorting key for the string.
  25
  26         The sort key should be considered an opaque value which is
  27         only meaningful when compared to other sort keys from the same
  28         collator.
  29         """
  30         if isinstance(string, str):
  31             string = string.decode(self.encoding, 'replace')
  32         return string
  33
  34     def words(self, string):
  35         """Split the string along word boundries."""
  36         if isinstance(string, str):
  37             string = string.decode(self.encoding, 'replace')
  38         return string.split()
  39
  40     def sortemekey(self, string):
  41         """Return a key based on sortemes of a string.
  42
  43         A sorteme, by analogy with grapheme/morpheme/etc. is an atom
  44         of sort information. This is larger than a word boundry but
  45         smaller than a sentence boundry; roughly, a sorteme boundry
  46         occurs between letters and numbers, between numbers and
  47         numbers if 'too much' punctuation exists in between, between
  48         lines.
  49         """
  50         if isinstance(string, str):
  51             string = string.decode(self.encoding, 'replace')
  52
  53         # Shove the sortkeyed original string on the end to resolve
  54         # ties intelligently.
  55         return (collate.strings.sortemes(string, self.key),
  56                 self.key(string))
  57