misc corrections/optimizations in top() and similarity()

dlibrary/dlibrary.py

@@ -825,7 +825,7 @@ def copy_recursive(src, dest):
 memoized_similarities = {}
 
 def similarity(a, b):
-    if len(a) < len(b):
+    if len(a) < len(b) or (len(a) == len(b) and a < b):
         shorter = a
         longer = b
     else:
@@ -849,18 +849,17 @@ def similarity(a, b):
     return result
 
 def top(items, n, key):
-    if len(items) == 0:
-        return []
-    winners = items[:1]
-    for item in items[1:]:
-        if len(winners) < n or key(item) > key(winners[-1]):
-            for i in range(len(winners)):
-                if key(item) > key(winners[i]):
-                    winners.insert(i, item)
+    winners = []
+    for item in items:
+        score = key(item)
+        if len(winners) < n or score > winners[-1][1]:
+            for i in range(len(winners) + 1):
+                if i == len(winners) or score > winners[i][1]:
+                    winners.insert(i, (item, score))
                     break
-            if len(winners) > n:
+            while len(winners) > n:
                 winners.pop()
-    return winners
+    return [item for (item, score) in winners]
 
 def generate(args):
     jenv = Environment(