Add MAC authentication to encrypted docs.

* Fix review comments:
    * Use of literal string instead of self.STORAGE_SECRETS_KEY
    * Add mac_method param to mac_doc()
    * Verify mac_method in mac_doc() and raise in there if unknown method
* Use different parts of storage_secret for generating doc passphrase and mac key.
* Add changes file.

1 files changed, 79 insertions, 4 deletions

diff --git a/src/leap/soledad/backends/leap_backend.py b/src/leap/soledad/backends/leap_backend.py
index 6f1f9546..87f63432 100644
--- a/src/leap/soledad/backends/leap_backend.py
+++ b/src/leap/soledad/backends/leap_backend.py
@@ -25,6 +25,8 @@ try:
     import simplejson as json
 except ImportError:
     import json  # noqa
+import hashlib
+import hmac
 
 
 from u1db import Document
@@ -37,6 +39,7 @@ from leap.common.keymanager import KeyManager
 from leap.common.check import leap_assert
 from leap.soledad.auth import TokenBasedAuth
 
+
 #
 # Exceptions
 #
@@ -52,6 +55,21 @@ class UnknownEncryptionScheme(Exception):
     """
     Raised when trying to decrypt from unknown encryption schemes.
     """
+    pass
+
+
+class UnknownMacMethod(Exception):
+    """
+    Raised when trying to authenticate document's content with unknown MAC
+    mehtod.
+    """
+    pass
+
+
+class WrongMac(Exception):
+    """
+    Raised when failing to authenticate document's contents based on MAC.
+    """
 
 
 #
@@ -68,12 +86,55 @@ class EncryptionSchemes(object):
     PUBKEY = 'pubkey'
 
 
+class MacMethods(object):
+    """
+    Representation of MAC methods used to authenticate document's contents.
+    """
+
+    HMAC = 'hmac'
+
+
 #
 # Crypto utilities for a LeapDocument.
 #
 
 ENC_JSON_KEY = '_enc_json'
 ENC_SCHEME_KEY = '_enc_scheme'
+MAC_KEY = '_mac'
+MAC_METHOD_KEY = '_mac_method'
+
+
+def mac_doc(crypto, doc_id, doc_rev, ciphertext, mac_method):
+    """
+    Calculate a MAC for C{doc} using C{ciphertext}.
+
+    Current MAC method used is HMAC, with the following parameters:
+
+        * key: sha256(storage_secret, doc_id)
+        * msg: doc_id + doc_rev + ciphertext
+        * digestmod: sha256
+
+    @param crypto: A SoledadCryto instance used to perform the encryption.
+    @type crypto: leap.soledad.crypto.SoledadCrypto
+    @param doc_id: The id of the document.
+    @type doc_id: str
+    @param doc_rev: The revision of the document.
+    @type doc_rev: str
+    @param ciphertext: The content of the document.
+    @type ciphertext: str
+    @param mac_method: The MAC method to use.
+    @type mac_method: str
+
+    @return: The calculated MAC.
+    @rtype: str
+    """
+    if mac_method == MacMethods.HMAC:
+        return hmac.new(
+            crypto.doc_mac_key(doc_id),
+            str(doc_id) + str(doc_rev) + ciphertext,
+            hashlib.sha256).hexdigest()
+    # raise if we do not know how to handle this MAC method
+    raise UnknownMacMethod('Unknown MAC method: %s.' % mac_method)
 
 
 def encrypt_doc(crypto, doc):
@@ -85,6 +146,8 @@ def encrypt_doc(crypto, doc):
         {
             ENC_JSON_KEY: '<encrypted doc JSON string>',
             ENC_SCHEME_KEY: 'symkey',
+            MAC_KEY: '<mac>'
+            MAC_METHOD_KEY: 'hmac'
         }
 
     @param crypto: A SoledadCryto instance used to perform the encryption.
@@ -100,16 +163,17 @@ def encrypt_doc(crypto, doc):
     # encrypt content
     ciphertext = crypto.encrypt_sym(
         doc.get_json(),
-        crypto.passphrase_hash(doc.doc_id))
+        crypto.doc_passphrase(doc.doc_id))
     # verify it is indeed encrypted
     if not crypto.is_encrypted_sym(ciphertext):
         raise DocumentNotEncrypted('Failed encrypting document.')
-    # calculate hmac
-    #mac = hmac.new(doc_id,, doc_json, 
     # update doc's content with encrypted version
     return json.dumps({
         ENC_JSON_KEY: ciphertext,
         ENC_SCHEME_KEY: EncryptionSchemes.SYMKEY,
+        MAC_KEY: mac_doc(
+            crypto, doc.doc_id, doc.rev, ciphertext, MacMethods.HMAC),
+        MAC_METHOD_KEY: MacMethods.HMAC,
     })
 
 
@@ -124,6 +188,8 @@ def decrypt_doc(crypto, doc):
         {
             ENC_JSON_KEY: '<enc_blob>',
             ENC_SCHEME_KEY: '<enc_scheme>',
+            MAC_KEY: '<mac>'
+            MAC_METHOD_KEY: 'hmac'
         }
 
     C{enc_blob} is the encryption of the JSON serialization of the document's
@@ -141,6 +207,15 @@ def decrypt_doc(crypto, doc):
     leap_assert(doc.is_tombstone() is False)
     leap_assert(ENC_JSON_KEY in doc.content)
     leap_assert(ENC_SCHEME_KEY in doc.content)
+    leap_assert(MAC_KEY in doc.content)
+    leap_assert(MAC_METHOD_KEY in doc.content)
+    # verify MAC
+    mac = mac_doc(
+        crypto, doc.doc_id, doc.rev,
+        doc.content[ENC_JSON_KEY],
+        doc.content[MAC_METHOD_KEY])
+    if doc.content[MAC_KEY] != mac:
+        raise WrongMac('Could not authenticate document\'s contents.')
     # decrypt doc's content
     ciphertext = doc.content[ENC_JSON_KEY]
     enc_scheme = doc.content[ENC_SCHEME_KEY]
@@ -153,7 +228,7 @@ def decrypt_doc(crypto, doc):
                 'symmetric key.')
         plainjson = crypto.decrypt_sym(
             ciphertext,
-            crypto.passphrase_hash(doc.doc_id))
+            crypto.doc_passphrase(doc.doc_id))
     else:
         raise UnknownEncryptionScheme(enc_scheme)
     return plainjson