now splitting buckets completely per section 4.2 of paper

kademlia/routing.py

@@ -59,6 +59,10 @@ class KBucket(object):
             return False
         return True
 
+    def depth(self):
+        sp = sharedPrefix([n.id for n in self.nodes.values()])
+        return len(sp)
+
     def head(self):
         return self.nodes.values()[0]
 
@@ -120,7 +124,6 @@ class RoutingTable(object):
         one, two = self.buckets[index].split()
         self.buckets[index] = one
         self.buckets.insert(index + 1, two)
-        # todo split one/two if needed based on section 4.2
 
     def getLonelyBuckets(self):
         """
@@ -141,7 +144,9 @@ class RoutingTable(object):
         if bucket.addNode(node):
             return
 
-        if bucket.hasInRange(self.node):
+        # Per section 4.2 of paper, split if the bucket has the node in it's range
+        # or if the depth is not congruent to 0 mod 5
+        if bucket.hasInRange(self.node) or bucket.depth() % 5 != 0:
             self.splitBucket(index)
             self.addContact(node)
         else:

kademlia/tests/utils.py

@@ -21,7 +21,7 @@ def mknode(id=None, ip=None, port=None, intid=None):
 
 class FakeProtocol(object):
     def __init__(self, sourceID, ksize=20):
-        self.router = RoutingTable(self, ksize)
+        self.router = RoutingTable(self, ksize, Node(sourceID))
         self.storage = {}
         self.sourceID = sourceID
 

kademlia/utils.py

@@ -39,3 +39,19 @@ class OrderedSet(list):
         if thing in self:
             self.remove(thing)
         self.append(thing)
+
+
+def sharedPrefix(args):
+    """
+    Find the shared prefix between the strings.
+
+    For instance, sharedPrefix(['blahblah', 'blahwhat']) is
+    'blah'.
+    """
+    i = 0
+    while i < min(map(len, args)):
+        if len(set(map(operator.itemgetter(i), args))) != 1:
+            break
+        i += 1
+    return args[0][:i]
+