X-Git-Url: https://git.korewanetadesu.com/?p=python-collate.git;a=blobdiff_plain;f=collate%2F_abcollator.py;h=0ae5d451f59defb65cb34fb4a2ca6b2f36152212;hp=a6ec268d9bb6512bffd6b1d752088798fde6ea09;hb=b0490f43bebb2aca90496fcdb5b65fa89642b409;hpb=96cd5d3ad9dd1390c7739a6c3b9fa03ac3a2b4ff

diff --git a/collate/_abcollator.py b/collate/_abcollator.py
index a6ec268..0ae5d45 100644
--- a/collate/_abcollator.py
+++ b/collate/_abcollator.py
@@ -1,14 +1,32 @@
+import collate.strings
+
 class Collator(object):
     def cmp(self, string1, string2):
         """Return negative if a < b, zero if a == b, positive if a > b."""
         return cmp(self.key(string1), self.key(string2))
 
     def words(self, string):
-        """Split the string into separate words.
-
-        This split is done using Unicode's definition of whitespace.
-        """
+        """Split the string along word boundries."""
+        if isinstance(string, str):
+            string = string.decode(self.encoding, 'replace')
         return string.split()
 
-    def wordkeys(self, string):
-        return map(self.key, self.words)
+    def sortemekey(self, string, invalid=float('inf')):
+        """Return a key based on sortemes of a string.
+
+        If the string is a str instance, it is decoded to a unicode
+        instance according to the 'encoding' attribute of the
+        Collator.
+        """
+        keys = []
+        if isinstance(string, str):
+            string = string.decode(self.encoding, 'replace')
+        for sorteme in collate.strings.sortemes(string):
+            num, alpha = collate.strings.numeric(sorteme, invalid)
+            if num == invalid:
+                keys.append(self.key(alpha))
+            else:
+                keys.append(num)
+        # Shove the sortkeyed original string on the end to resolve
+        # ties intelligently.
+        return (keys, self.key(string))