collate/icu/__init__.py

   1 """ICU-based collation.
   2
   3 This collation backend uses the International Components for Unicode
   4 library to provide accurate and high-performance collation. It
   5 supports multiple locales and advanced sorting capabilities.
   6
   7 Use this collation backend if possible; it's by far the best.
   8
   9 Avoid this backend if...
  10  - ICU is not available for your system.
  11
  12 """
  13
  14 import collate.icu._ucol
  15 import collate._abcollator
  16 import collate._locale
  17 import collate.errors
  18
  19 class Collator(collate._abcollator.Collator):
  20     """ICU-based collation."""
  21
  22     def __init__(self, locale, encoding=None):
  23         locale, encoding = collate._locale.getpair(locale, encoding)
  24         self._collator = collate.icu._ucol.Collator(locale)
  25         self.locale = self._collator.locale
  26         self.encoding = collate._locale.encoding(encoding)
  27         if self._collator.used_default_information and locale != "C":
  28             raise collate.errors.InvalidLocaleError(locale)
  29
  30     def key(self, string):
  31         """Sort key for a string.
  32
  33         If the string is a str instance, it is decoded to a unicode
  34         instance according to the 'encoding' attribute of the
  35         Collator.
  36         """
  37         if isinstance(string, str):
  38             string = string.decode(self.encoding, 'replace')
  39         return self._collator.key(string)
  40
  41     def cmp(self, a, b):
  42         """Return negative if a < b, zero if a == b, positive if a > b.
  43
  44         If strs rather than unicodes are passed in, they are first
  45         decoded according to the 'encoding' attribute of the Collator.
  46         """
  47         if isinstance(a, str):
  48             a = a.decode(self.encoding, 'replace')
  49         if isinstance(b, str):
  50             b = a.decode(self.encoding, 'replace')
  51         return self._collator.cmp(a, b)