New approach - find split points based on Unicode categories.

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

diff --git a/collate/_abcollator.py b/collate/_abcollator.py
index a6ec268..fdd7783 100644 (file)
--- a/collate/_abcollator.py
+++ b/collate/_abcollator.py
@@ -1,14 +1,18 @@
+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.
-        """
-        return string.split()
-
-    def wordkeys(self, string):
-        return map(self.key, self.words)
+    def sortemekey(self, string, invalid=float('inf')):
+        keys = []
+        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))