Word-splitting.

[python-collate.git] / collate / syslocale.py

diff --git a/collate/syslocale.py b/collate/syslocale.py
index 1ee2924..e2aeed9 100644 (file)
--- a/collate/syslocale.py
+++ b/collate/syslocale.py
@@ -12,11 +12,10 @@ locale of all previous collators and anything else using the system
 locale information.
 
 Use this collation backend if...
- - You are on a system without ICU or UCA datafiles for the locale,
-   and DUCET results are not acceptable.
+ - You are on a system without ICU.
 
 Avoid this backend if...
- - ICU or UCA support is available for the current locale.
+ - ICU is available for the current locale.
  - You are sorting strings from alphabets outside the primary locale.
  - You need to support collating multiple locales at once.
  - You need the same results across multiple platforms.
@@ -24,6 +23,7 @@ Avoid this backend if...
 """
 
 import locale
+import re
 
 import collate.errors
 import collate._abcollator
@@ -66,3 +66,10 @@ class Collator(collate._abcollator.Collator):
             b = b.decode(self.encoding, "replace")
         return locale.strcoll(a, b)
                                   
+    def words(self, string, sep=re.compile(r"\W+", re.UNICODE)):
+        """Split the string into separate words.
+
+        This split is done using the locale's notion of a word boundry.
+        """
+        return re.split(sep, string)
+