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._abcollator
  15 import collate._locale
  16 import collate.errors
  17
  18 from collate.icu import _icu
  19
  20 class Collator(collate._abcollator.Collator):
  21     """ICU-based collation."""
  22
  23     def __init__(self, locale, encoding=None):
  24         locale, encoding = collate._locale.getpair(locale, encoding)
  25         icu_locale = "root" if locale == "C" else locale
  26         self._collator = _icu.Collator(icu_locale)
  27         self.locale = self._collator.locale
  28         self.encoding = collate._locale.encoding(encoding)
  29         if self._collator.used_default_information and locale != "C":
  30             raise collate.errors.InvalidLocaleError(locale)
  31
  32         try:
  33             self._breaker = _icu.WordBreaker(icu_locale)
  34         except ValueError:
  35             # Thai is the only language with a special break locale,
  36             # so this is a harmless error.
  37             self._breaker = _icu.WordBreaker("root")
  38
  39     def words(self, string):
  40         """Split the string along word boundries."""
  41         if isinstance(string, str):
  42             string = string.decode(self.encoding)
  43         words = self._breaker.words(string)
  44         return [w for w in words if not w.isspace()]
  45
  46     def key(self, string):
  47         """Sort key for a string.
  48
  49         If the string is a str instance, it is decoded to a unicode
  50         instance according to the 'encoding' attribute of the
  51         Collator.
  52         """
  53         if isinstance(string, str):
  54             string = string.decode(self.encoding, 'replace')
  55         return self._collator.key(string)
  56
  57     def cmp(self, a, b):
  58         """Return negative if a < b, zero if a == b, positive if a > b.
  59
  60         If strs rather than unicodes are passed in, they are first
  61         decoded according to the 'encoding' attribute of the Collator.
  62         """
  63         if isinstance(a, str):
  64             a = a.decode(self.encoding, 'replace')
  65         if isinstance(b, str):
  66             b = b.decode(self.encoding, 'replace')
  67         return self._collator.cmp(a, b)