fetchallcerts.py: calculate root hash

2 files changed, 61 insertions, 15 deletions

diff --git a/tools/certtools.py b/tools/certtools.py
index 16c2105..b132caa 100644
--- a/tools/certtools.py
+++ b/tools/certtools.py
@@ -226,3 +226,27 @@ def timing_point(timer_dict=None, name=None):
     else:
         timer_dict = {"deltatimes":[], "lasttime":t}
         return timer_dict
+
+def internal_hash(pair):
+    if len(pair) == 1:
+        return pair[0]
+    else:
+        hash = hashlib.sha256()
+        hash.update(struct.pack(">b", 1))
+        hash.update(pair[0])
+        hash.update(pair[1])
+        return hash.digest()
+
+def chunks(l, n):
+    return [l[i:i+n] for i in range(0, len(l), n)]
+
+def next_merkle_layer(layer):
+    return [internal_hash(pair) for pair in chunks(layer, 2)]
+
+def build_merkle_tree(layer0):
+    layers = []
+    current_layer = layer0
+    while len(current_layer) > 1:
+        current_layer = next_merkle_layer(current_layer)
+        layers.append(current_layer)
+    return layers
diff --git a/tools/fetchallcerts.py b/tools/fetchallcerts.py
index 801e296..dad5241 100644
--- a/tools/fetchallcerts.py
+++ b/tools/fetchallcerts.py
@@ -4,6 +4,7 @@
 # Copyright (c) 2014, NORDUnet A/S.
 # See LICENSE for licensing information.
 
+import argparse
 import urllib2
 import urllib
 import json
@@ -14,6 +15,11 @@ import hashlib
 import itertools
 from certtools import *
 
+parser = argparse.ArgumentParser(description='')
+parser.add_argument('baseurl', help="Base URL for CT server")
+parser.add_argument('--store', default=None, metavar="dir", help='Store certificates in directory dir')
+args = parser.parse_args()
+
 def extract_original_entry(entry):
     leaf_input =  base64.decodestring(entry["leaf_input"])
     (leaf_cert, timestamp) = unpack_mtl(leaf_input)
@@ -23,7 +29,7 @@ def extract_original_entry(entry):
 
 def get_entries_wrapper(baseurl, start, end):
     fetched_entries = []
-    while len(fetched_entries) < (end - start + 1):
+    while start + len(fetched_entries) < (end + 1):
         print "fetching from", start + len(fetched_entries)
         entries = get_entries(baseurl, start + len(fetched_entries), end)["entries"]
         if len(entries) == 0:
@@ -31,23 +37,39 @@ def get_entries_wrapper(baseurl, start, end):
         fetched_entries.extend(entries)
     return fetched_entries
 
-baseurl = sys.argv[1]
-destination_directory = sys.argv[2]
+def print_layer(layer):
+    for entry in layer:
+        print base64.b16encode(entry)
 
-sth = get_sth(baseurl)
+sth = get_sth(args.baseurl)
 tree_size = sth["tree_size"]
+root_hash = base64.decodestring(sth["sha256_root_hash"])
+
+print "tree size", tree_size
+print "root hash", base64.b16encode(root_hash)
+
+entries = get_entries_wrapper(args.baseurl, 0, tree_size - 1)
+
+print "fetched", len(entries), "entries"
+
+layer0 = [get_leaf_hash(base64.decodestring(entry["leaf_input"])) for entry in entries]
+
+tree = build_merkle_tree(layer0)
 
-print tree_size
+calculated_root_hash = tree[-1][0]
 
-entries = get_entries_wrapper(baseurl, 0, tree_size)
+print "calculated root hash", base64.b16encode(calculated_root_hash)
 
-print len(entries)
+if calculated_root_hash != root_hash:
+    print "fetched root hash and calculated root hash different, aborting"
+    sys.exit(1)
 
-for entry, i in zip(entries, range(0, len(entries))):
-    chain = extract_original_entry(entry)
-    f = open(destination_directory + "/" + ("%06d" % i), "w")
-    for cert in chain:
-        print >> f, "-----BEGIN CERTIFICATE-----"
-        print >> f, base64.encodestring(cert).rstrip()
-        print >> f, "-----END CERTIFICATE-----"
-        print >> f, ""
+if args.store:
+    for entry, i in zip(entries, range(0, len(entries))):
+        chain = extract_original_entry(entry)
+        f = open(args.store + "/" + ("%06d" % i), "w")
+        for cert in chain:
+            print >> f, "-----BEGIN CERTIFICATE-----"
+            print >> f, base64.encodestring(cert).rstrip()
+            print >> f, "-----END CERTIFICATE-----"
+            print >> f, ""