collate/syslocale.py

   1 """C library locale-based collation.
   2
   3 This collation backend uses the system's C library to sort strings. It
   4 is fast and almost always available, but may sort strings outside of
   5 the user's native locale incorrectly or confusingly (for example,
   6 en_US tends to ignore hiragana characters; ja_JP does not case-fold
   7 Latin characters).
   8
   9 Since the C library only supports one locale active at a time per
  10 process, instantiating a Collator from this module will affect the
  11 locale of all previous collators and anything else using the system
  12 locale information.
  13
  14 Use this collation backend if...
  15  - You are on a system without ICU.
  16
  17 Avoid this backend if...
  18  - ICU is available for the current locale.
  19  - You are sorting strings from alphabets outside the primary locale.
  20  - You need to support collating multiple locales at once.
  21  - You need the same results across multiple platforms.
  22
  23 """
  24
  25 import locale
  26 import re
  27
  28 import collate.errors
  29 import collate._abcollator
  30 import collate._locale
  31
  32 class Collator(collate._abcollator.Collator):
  33     """C library locale-based collation."""
  34
  35     def __init__(self, locale_code, encoding=None):
  36         locale_code, encoding = collate._locale.getpair(locale_code, encoding)
  37         try:
  38             setlocale = locale_code + "." + encoding
  39             locale.setlocale(locale.LC_COLLATE, setlocale)
  40         except locale.Error:
  41             raise collate.errors.InvalidLocaleError(setlocale)
  42         self.locale = locale.getlocale(locale.LC_COLLATE)[0]
  43         self.encoding = collate._locale.encoding(encoding)
  44
  45     def key(self, string):
  46         """Sort key for a string.
  47
  48         If string is a unicode instance that cannot be processed by
  49         the system locale library, it is first encoded according to
  50         the 'encoding' attribute of the Collator.
  51         """
  52         try:
  53             return locale.strxfrm(string)
  54         except UnicodeEncodeError:
  55             return locale.strxfrm(string.encode(self.encoding, "replace"))
  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 locale.strcoll(a, b)
  68
  69     def words(self, string, sep=re.compile(r"\W+", re.UNICODE)):
  70         """Split the string into separate words.
  71
  72         This split is done using the locale's notion of a word boundry.
  73         """
  74         return re.split(sep, string)
  75