sortemes: Simplify some logic.

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

diff --git a/collate/_abcollator.py b/collate/_abcollator.py
index 094a5de..0ae5d45 100644 (file)
--- a/collate/_abcollator.py
+++ b/collate/_abcollator.py
@@ -1,5 +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 along word boundries."""
+        if isinstance(string, str):
+            string = string.decode(self.encoding, 'replace')
+        return string.split()
+
+    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))