Python 3: Convert some unicode/bytes uses (#3569)

changelog.d/3569.bugfix

@@ -0,0 +1 @@
+Unicode passwords are now normalised before hashing, preventing the instance where two different devices or browsers might send a different UTF-8 sequence for the password.

synapse/api/auth.py

@@ -252,10 +252,10 @@ class Auth(object):
             if ip_address not in app_service.ip_range_whitelist:
                 defer.returnValue((None, None))
 
-        if "user_id" not in request.args:
+        if b"user_id" not in request.args:
             defer.returnValue((app_service.sender, app_service))
 
-        user_id = request.args["user_id"][0]
+        user_id = request.args[b"user_id"][0].decode('utf8')
         if app_service.sender == user_id:
             defer.returnValue((app_service.sender, app_service))
 

synapse/federation/transport/server.py

@@ -165,7 +165,7 @@ def _parse_auth_header(header_bytes):
         param_dict = dict(kv.split("=") for kv in params)
 
         def strip_quotes(value):
-            if value.startswith(b"\""):
+            if value.startswith("\""):
                 return value[1:-1]
             else:
                 return value

synapse/handlers/auth.py

@@ -15,6 +15,7 @@
 # limitations under the License.
 
 import logging
+import unicodedata
 
 import attr
 import bcrypt
@@ -626,6 +627,7 @@ class AuthHandler(BaseHandler):
         # special case to check for "password" for the check_password interface
         # for the auth providers
         password = login_submission.get("password")
+
         if login_type == LoginType.PASSWORD:
             if not self._password_enabled:
                 raise SynapseError(400, "Password login has been disabled.")
@@ -707,9 +709,10 @@ class AuthHandler(BaseHandler):
         multiple inexact matches.
 
         Args:
-            user_id (str): complete @user:id
+            user_id (unicode): complete @user:id
+            password (unicode): the provided password
         Returns:
-            (str) the canonical_user_id, or None if unknown user / bad password
+            (unicode) the canonical_user_id, or None if unknown user / bad password
         """
         lookupres = yield self._find_user_id_and_pwd_hash(user_id)
         if not lookupres:
@@ -849,14 +852,19 @@ class AuthHandler(BaseHandler):
         """Computes a secure hash of password.
 
         Args:
-            password (str): Password to hash.
+            password (unicode): Password to hash.
 
         Returns:
-            Deferred(str): Hashed password.
+            Deferred(unicode): Hashed password.
         """
         def _do_hash():
-            return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper,
+            # Normalise the Unicode in the password
-                                 bcrypt.gensalt(self.bcrypt_rounds))
+            pw = unicodedata.normalize("NFKC", password)
+
+            return bcrypt.hashpw(
+                pw.encode('utf8') + self.hs.config.password_pepper.encode("utf8"),
+                bcrypt.gensalt(self.bcrypt_rounds),
+            ).decode('ascii')
 
         return make_deferred_yieldable(
             threads.deferToThreadPool(
@@ -868,16 +876,19 @@ class AuthHandler(BaseHandler):
         """Validates that self.hash(password) == stored_hash.
 
         Args:
-            password (str): Password to hash.
+            password (unicode): Password to hash.
-            stored_hash (str): Expected hash value.
+            stored_hash (unicode): Expected hash value.
 
         Returns:
             Deferred(bool): Whether self.hash(password) == stored_hash.
         """
 
         def _do_validate_hash():
+            # Normalise the Unicode in the password
+            pw = unicodedata.normalize("NFKC", password)
+
             return bcrypt.checkpw(
-                password.encode('utf8') + self.hs.config.password_pepper,
+                pw.encode('utf8') + self.hs.config.password_pepper.encode("utf8"),
                 stored_hash.encode('utf8')
             )
 

synapse/handlers/register.py

@@ -131,7 +131,7 @@ class RegistrationHandler(BaseHandler):
         Args:
             localpart : The local part of the user ID to register. If None,
               one will be generated.
-            password (str) : The password to assign to this user so they can
+            password (unicode) : The password to assign to this user so they can
               login again. This can be None which means they cannot login again
               via a password (e.g. the user is an application service user).
             generate_token (bool): Whether a new access token should be

synapse/http/server.py

@@ -13,12 +13,13 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import cgi
 import collections
 import logging
-import urllib
 
-from six.moves import http_client
+from six import PY3
+from six.moves import http_client, urllib
 
 from canonicaljson import encode_canonical_json, encode_pretty_printed_json, json
 
@@ -264,6 +265,7 @@ class JsonResource(HttpServer, resource.Resource):
         self.hs = hs
 
     def register_paths(self, method, path_patterns, callback):
+        method = method.encode("utf-8")  # method is bytes on py3
         for path_pattern in path_patterns:
             logger.debug("Registering for %s %s", method, path_pattern.pattern)
             self.path_regexs.setdefault(method, []).append(
@@ -296,8 +298,19 @@ class JsonResource(HttpServer, resource.Resource):
         # here. If it throws an exception, that is handled by the wrapper
         # installed by @request_handler.
 
+        def _unquote(s):
+            if PY3:
+                # On Python 3, unquote is unicode -> unicode
+                return urllib.parse.unquote(s)
+            else:
+                # On Python 2, unquote is bytes -> bytes We need to encode the
+                # URL again (as it was decoded by _get_handler_for request), as
+                # ASCII because it's a URL, and then decode it to get the UTF-8
+                # characters that were quoted.
+                return urllib.parse.unquote(s.encode('ascii')).decode('utf8')
+
         kwargs = intern_dict({
-            name: urllib.unquote(value).decode("UTF-8") if value else value
+            name: _unquote(value) if value else value
             for name, value in group_dict.items()
         })
 
@@ -313,9 +326,9 @@ class JsonResource(HttpServer, resource.Resource):
             request (twisted.web.http.Request):
 
         Returns:
-            Tuple[Callable, dict[str, str]]: callback method, and the dict
+            Tuple[Callable, dict[unicode, unicode]]: callback method, and the
-                mapping keys to path components as specified in the handler's
+                dict mapping keys to path components as specified in the
-                path match regexp.
+                handler's path match regexp.
 
                 The callback will normally be a method registered via
                 register_paths, so will return (possibly via Deferred) either
@@ -327,7 +340,7 @@ class JsonResource(HttpServer, resource.Resource):
         # Loop through all the registered callbacks to check if the method
         # and path regex match
         for path_entry in self.path_regexs.get(request.method, []):
-            m = path_entry.pattern.match(request.path)
+            m = path_entry.pattern.match(request.path.decode('ascii'))
             if m:
                 # We found a match!
                 return path_entry.callback, m.groupdict()
@@ -383,7 +396,7 @@ class RootRedirect(resource.Resource):
         self.url = path
 
     def render_GET(self, request):
-        return redirectTo(self.url, request)
+        return redirectTo(self.url.encode('ascii'), request)
 
     def getChild(self, name, request):
         if len(name) == 0:
@@ -404,12 +417,14 @@ def respond_with_json(request, code, json_object, send_cors=False,
         return
 
     if pretty_print:
-        json_bytes = encode_pretty_printed_json(json_object) + "\n"
+        json_bytes = (encode_pretty_printed_json(json_object) + "\n"
+                      ).encode("utf-8")
     else:
         if canonical_json or synapse.events.USE_FROZEN_DICTS:
+            # canonicaljson already encodes to bytes
             json_bytes = encode_canonical_json(json_object)
         else:
-            json_bytes = json.dumps(json_object)
+            json_bytes = json.dumps(json_object).encode("utf-8")
 
     return respond_with_json_bytes(
         request, code, json_bytes,

synapse/http/servlet.py

@@ -171,8 +171,16 @@ def parse_json_value_from_request(request, allow_empty_body=False):
     if not content_bytes and allow_empty_body:
         return None
 
+    # Decode to Unicode so that simplejson will return Unicode strings on
+    # Python 2
     try:
-        content = json.loads(content_bytes)
+        content_unicode = content_bytes.decode('utf8')
+    except UnicodeDecodeError:
+        logger.warn("Unable to decode UTF-8")
+        raise SynapseError(400, "Content not JSON.", errcode=Codes.NOT_JSON)
+
+    try:
+        content = json.loads(content_unicode)
     except Exception as e:
         logger.warn("Unable to parse JSON: %s", e)
         raise SynapseError(400, "Content not JSON.", errcode=Codes.NOT_JSON)

synapse/rest/client/v1/admin.py

@@ -18,6 +18,7 @@ import hashlib
 import hmac
 import logging
 
+from six import text_type
 from six.moves import http_client
 
 from twisted.internet import defer
@@ -131,7 +132,10 @@ class UserRegisterServlet(ClientV1RestServlet):
                 400, "username must be specified", errcode=Codes.BAD_JSON,
             )
         else:
-            if (not isinstance(body['username'], str) or len(body['username']) > 512):
+            if (
+                not isinstance(body['username'], text_type)
+                or len(body['username']) > 512
+            ):
                 raise SynapseError(400, "Invalid username")
 
             username = body["username"].encode("utf-8")
@@ -143,7 +147,10 @@ class UserRegisterServlet(ClientV1RestServlet):
                 400, "password must be specified", errcode=Codes.BAD_JSON,
             )
         else:
-            if (not isinstance(body['password'], str) or len(body['password']) > 512):
+            if (
+                not isinstance(body['password'], text_type)
+                or len(body['password']) > 512
+            ):
                 raise SynapseError(400, "Invalid password")
 
             password = body["password"].encode("utf-8")
@@ -166,17 +173,18 @@ class UserRegisterServlet(ClientV1RestServlet):
         want_mac.update(b"admin" if admin else b"notadmin")
         want_mac = want_mac.hexdigest()
 
-        if not hmac.compare_digest(want_mac, got_mac):
+        if not hmac.compare_digest(want_mac, got_mac.encode('ascii')):
-            raise SynapseError(
+            raise SynapseError(403, "HMAC incorrect")
-                403, "HMAC incorrect",
-            )
 
         # Reuse the parts of RegisterRestServlet to reduce code duplication
         from synapse.rest.client.v2_alpha.register import RegisterRestServlet
+
         register = RegisterRestServlet(self.hs)
 
         (user_id, _) = yield register.registration_handler.register(
-            localpart=username.lower(), password=password, admin=bool(admin),
+            localpart=body['username'].lower(),
+            password=body["password"],
+            admin=bool(admin),
             generate_token=False,
         )
 

synapse/rest/client/v2_alpha/register.py

@@ -193,15 +193,15 @@ class RegisterRestServlet(RestServlet):
     def on_POST(self, request):
         body = parse_json_object_from_request(request)
 
-        kind = "user"
+        kind = b"user"
-        if "kind" in request.args:
+        if b"kind" in request.args:
-            kind = request.args["kind"][0]
+            kind = request.args[b"kind"][0]
 
-        if kind == "guest":
+        if kind == b"guest":
             ret = yield self._do_guest_registration(body)
             defer.returnValue(ret)
             return
-        elif kind != "user":
+        elif kind != b"user":
             raise UnrecognizedRequestError(
                 "Do not understand membership kind: %s" % (kind,)
             )
@@ -389,8 +389,8 @@ class RegisterRestServlet(RestServlet):
             assert_params_in_dict(params, ["password"])
 
             desired_username = params.get("username", None)
-            new_password = params.get("password", None)
             guest_access_token = params.get("guest_access_token", None)
+            new_password = params.get("password", None)
 
             if desired_username is not None:
                 desired_username = desired_username.lower()

synapse/rest/media/v1/media_storage.py

@@ -177,7 +177,7 @@ class MediaStorage(object):
             if res:
                 with res:
                     consumer = BackgroundFileConsumer(
-                        open(local_path, "w"), self.hs.get_reactor())
+                        open(local_path, "wb"), self.hs.get_reactor())
                     yield res.write_to_consumer(consumer)
                     yield consumer.wait()
                 defer.returnValue(local_path)

synapse/state.py

@@ -577,7 +577,7 @@ def _make_state_cache_entry(
 
 def _ordered_events(events):
     def key_func(e):
-        return -int(e.depth), hashlib.sha1(e.event_id.encode()).hexdigest()
+        return -int(e.depth), hashlib.sha1(e.event_id.encode('ascii')).hexdigest()
 
     return sorted(events, key=key_func)
 

synapse/storage/events.py

@@ -67,7 +67,13 @@ state_delta_reuse_delta_counter = Counter(
 
 
 def encode_json(json_object):
-    return frozendict_json_encoder.encode(json_object)
+    """
+    Encode a Python object as JSON and return it in a Unicode string.
+    """
+    out = frozendict_json_encoder.encode(json_object)
+    if isinstance(out, bytes):
+        out = out.decode('utf8')
+    return out
 
 
 class _EventPeristenceQueue(object):
@@ -1058,7 +1064,7 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
 
                 metadata_json = encode_json(
                     event.internal_metadata.get_dict()
-                ).decode("UTF-8")
+                )
 
                 sql = (
                     "UPDATE event_json SET internal_metadata = ?"
@@ -1172,8 +1178,8 @@ class EventsStore(EventFederationStore, EventsWorkerStore, BackgroundUpdateStore
                     "room_id": event.room_id,
                     "internal_metadata": encode_json(
                         event.internal_metadata.get_dict()
-                    ).decode("UTF-8"),
+                    ),
-                    "json": encode_json(event_dict(event)).decode("UTF-8"),
+                    "json": encode_json(event_dict(event)),
                 }
                 for event, _ in events_and_contexts
             ],

synapse/storage/signatures.py

@@ -74,7 +74,7 @@ class SignatureWorkerStore(SQLBaseStore):
             txn (cursor):
             event_id (str): Id for the Event.
         Returns:
-            A dict of algorithm -> hash.
+            A dict[unicode, bytes] of algorithm -> hash.
         """
         query = (
             "SELECT algorithm, hash"

synapse/types.py

@@ -137,7 +137,7 @@ class DomainSpecificString(
     @classmethod
     def from_string(cls, s):
         """Parse the string given by 's' into a structure object."""
-        if len(s) < 1 or s[0] != cls.SIGIL:
+        if len(s) < 1 or s[0:1] != cls.SIGIL:
             raise SynapseError(400, "Expected %s string to start with '%s'" % (
                 cls.__name__, cls.SIGIL,
             ))

synapse/util/frozenutils.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from six import string_types
+from six import binary_type, text_type
 
 from canonicaljson import json
 from frozendict import frozendict
@@ -26,7 +26,7 @@ def freeze(o):
     if isinstance(o, frozendict):
         return o
 
-    if isinstance(o, string_types):
+    if isinstance(o, (binary_type, text_type)):
         return o
 
     try:
@@ -41,7 +41,7 @@ def unfreeze(o):
     if isinstance(o, (dict, frozendict)):
         return dict({k: unfreeze(v) for k, v in o.items()})
 
-    if isinstance(o, string_types):
+    if isinstance(o, (binary_type, text_type)):
         return o
 
     try:

tests/api/test_auth.py

@@ -46,7 +46,7 @@ class AuthTestCase(unittest.TestCase):
         self.auth = Auth(self.hs)
 
         self.test_user = "@foo:bar"
-        self.test_token = "_test_token_"
+        self.test_token = b"_test_token_"
 
         # this is overridden for the appservice tests
         self.store.get_app_service_by_token = Mock(return_value=None)
@@ -61,7 +61,7 @@ class AuthTestCase(unittest.TestCase):
         self.store.get_user_by_access_token = Mock(return_value=user_info)
 
         request = Mock(args={})
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         requester = yield self.auth.get_user_by_req(request)
         self.assertEquals(requester.user.to_string(), self.test_user)
@@ -70,7 +70,7 @@ class AuthTestCase(unittest.TestCase):
         self.store.get_user_by_access_token = Mock(return_value=None)
 
         request = Mock(args={})
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         d = self.auth.get_user_by_req(request)
         self.failureResultOf(d, AuthError)
@@ -98,7 +98,7 @@ class AuthTestCase(unittest.TestCase):
 
         request = Mock(args={})
         request.getClientIP.return_value = "127.0.0.1"
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         requester = yield self.auth.get_user_by_req(request)
         self.assertEquals(requester.user.to_string(), self.test_user)
@@ -115,7 +115,7 @@ class AuthTestCase(unittest.TestCase):
 
         request = Mock(args={})
         request.getClientIP.return_value = "192.168.10.10"
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         requester = yield self.auth.get_user_by_req(request)
         self.assertEquals(requester.user.to_string(), self.test_user)
@@ -131,7 +131,7 @@ class AuthTestCase(unittest.TestCase):
 
         request = Mock(args={})
         request.getClientIP.return_value = "131.111.8.42"
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         d = self.auth.get_user_by_req(request)
         self.failureResultOf(d, AuthError)
@@ -141,7 +141,7 @@ class AuthTestCase(unittest.TestCase):
         self.store.get_user_by_access_token = Mock(return_value=None)
 
         request = Mock(args={})
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         d = self.auth.get_user_by_req(request)
         self.failureResultOf(d, AuthError)
@@ -158,7 +158,7 @@ class AuthTestCase(unittest.TestCase):
 
     @defer.inlineCallbacks
     def test_get_user_by_req_appservice_valid_token_valid_user_id(self):
-        masquerading_user_id = "@doppelganger:matrix.org"
+        masquerading_user_id = b"@doppelganger:matrix.org"
         app_service = Mock(
             token="foobar", url="a_url", sender=self.test_user,
             ip_range_whitelist=None,
@@ -169,14 +169,17 @@ class AuthTestCase(unittest.TestCase):
 
         request = Mock(args={})
         request.getClientIP.return_value = "127.0.0.1"
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
-        request.args["user_id"] = [masquerading_user_id]
+        request.args[b"user_id"] = [masquerading_user_id]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         requester = yield self.auth.get_user_by_req(request)
-        self.assertEquals(requester.user.to_string(), masquerading_user_id)
+        self.assertEquals(
+            requester.user.to_string(),
+            masquerading_user_id.decode('utf8')
+        )
 
     def test_get_user_by_req_appservice_valid_token_bad_user_id(self):
-        masquerading_user_id = "@doppelganger:matrix.org"
+        masquerading_user_id = b"@doppelganger:matrix.org"
         app_service = Mock(
             token="foobar", url="a_url", sender=self.test_user,
             ip_range_whitelist=None,
@@ -187,8 +190,8 @@ class AuthTestCase(unittest.TestCase):
 
         request = Mock(args={})
         request.getClientIP.return_value = "127.0.0.1"
-        request.args["access_token"] = [self.test_token]
+        request.args[b"access_token"] = [self.test_token]
-        request.args["user_id"] = [masquerading_user_id]
+        request.args[b"user_id"] = [masquerading_user_id]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         d = self.auth.get_user_by_req(request)
         self.failureResultOf(d, AuthError)
@@ -418,7 +421,7 @@ class AuthTestCase(unittest.TestCase):
 
         # check the token works
         request = Mock(args={})
-        request.args["access_token"] = [token]
+        request.args[b"access_token"] = [token.encode('ascii')]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
         requester = yield self.auth.get_user_by_req(request, allow_guest=True)
         self.assertEqual(UserID.from_string(USER_ID), requester.user)
@@ -431,7 +434,7 @@ class AuthTestCase(unittest.TestCase):
 
         # the token should *not* work now
         request = Mock(args={})
-        request.args["access_token"] = [guest_tok]
+        request.args[b"access_token"] = [guest_tok.encode('ascii')]
         request.requestHeaders.getRawHeaders = mock_getRawHeaders()
 
         with self.assertRaises(AuthError) as cm:

tests/utils.py

@@ -193,7 +193,7 @@ class MockHttpResource(HttpServer):
         self.prefix = prefix
 
     def trigger_get(self, path):
-        return self.trigger("GET", path, None)
+        return self.trigger(b"GET", path, None)
 
     @patch('twisted.web.http.Request')
     @defer.inlineCallbacks
@@ -227,7 +227,7 @@ class MockHttpResource(HttpServer):
 
         headers = {}
         if federation_auth:
-            headers[b"Authorization"] = ["X-Matrix origin=test,key=,sig="]
+            headers[b"Authorization"] = [b"X-Matrix origin=test,key=,sig="]
         mock_request.requestHeaders.getRawHeaders = mock_getRawHeaders(headers)
 
         # return the right path if the event requires it
@@ -241,6 +241,9 @@ class MockHttpResource(HttpServer):
         except Exception:
             pass
 
+        if isinstance(path, bytes):
+            path = path.decode('utf8')
+
         for (method, pattern, func) in self.callbacks:
             if http_method != method:
                 continue
@@ -249,7 +252,7 @@ class MockHttpResource(HttpServer):
             if matcher:
                 try:
                     args = [
-                        urlparse.unquote(u).decode("UTF-8")
+                        urlparse.unquote(u)
                         for u in matcher.groups()
                     ]