X-Git-Url: https://git.korewanetadesu.com/?p=python-collate.git;a=blobdiff_plain;f=collate%2Ficu%2F__init__.py;h=5f3ec05f98f19d051c4c8a6beda8b14fc0968081;hp=eeee418070483d0797b63b639d1aa7bca219442e;hb=9a7cf6459c40d53b58634f2df56386bf52c12f7c;hpb=dd764628164dd6f23c1ec98a13184ce7c69fa40f

diff --git a/collate/icu/__init__.py b/collate/icu/__init__.py
index eeee418..5f3ec05 100644
--- a/collate/icu/__init__.py
+++ b/collate/icu/__init__.py
@@ -11,22 +11,38 @@ Avoid this backend if...
 
 """
 
-import collate.icu._ucol
 import collate._abcollator
 import collate._locale
 import collate.errors
 
+from collate.icu import _icu
+
 class Collator(collate._abcollator.Collator):
     """ICU-based collation."""
 
     def __init__(self, locale, encoding=None):
         locale, encoding = collate._locale.getpair(locale, encoding)
-        self._collator = collate.icu._ucol.Collator(locale)
+        icu_locale = "root" if locale == "C" else locale
+        self._collator = _icu.Collator(icu_locale)
         self.locale = self._collator.locale
         self.encoding = collate._locale.encoding(encoding)
         if self._collator.used_default_information and locale != "C":
             raise collate.errors.InvalidLocaleError(locale)
 
+        try:
+            self._breaker = _icu.WordBreaker(icu_locale)
+        except ValueError:
+            # Thai is the only language with a special break locale,
+            # so this is a harmless error.
+            self._breaker = _icu.WordBreaker("root")
+
+    def words(self, string):
+        """Split the string along word boundries."""
+        if isinstance(string, str):
+            string = string.decode(self.encoding)
+        words = self._breaker.words(string)
+        return [w for w in words if not w.isspace()]
+
     def key(self, string):
         """Sort key for a string.
 
@@ -47,5 +63,5 @@ class Collator(collate._abcollator.Collator):
         if isinstance(a, str):
             a = a.decode(self.encoding, 'replace')
         if isinstance(b, str):
-            b = a.decode(self.encoding, 'replace')
+            b = b.decode(self.encoding, 'replace')
         return self._collator.cmp(a, b)