Uniformize spam-checker API, part 5: expand other spam-checker callbacks to return `Tuple[Codes, dict]` (#13044)

Signed-off-by: David Teller <davidt@element.io>
Co-authored-by: Brendan Abolivier <babolivier@matrix.org>

3 files changed, 249 insertions, 10 deletions

diff --git a/tests/rest/client/test_rooms.py b/tests/rest/client/test_rooms.py
index 1ccd96a207..e67844cfa1 100644
--- a/tests/rest/client/test_rooms.py
+++ b/tests/rest/client/test_rooms.py
@@ -22,7 +22,7 @@ from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
 from unittest.mock import Mock, call
 from urllib import parse as urlparse
 
-# `Literal` appears with Python 3.8.
+from parameterized import param, parameterized
 from typing_extensions import Literal
 
 from twisted.test.proto_helpers import MemoryReactor
@@ -815,14 +815,14 @@ class RoomsCreateTestCase(RoomBase):
         In this test, we use the more recent API in which callbacks return a `Union[Codes, Literal["NOT_SPAM"]]`.
         """
 
-        async def user_may_join_room(
+        async def user_may_join_room_codes(
             mxid: str,
             room_id: str,
             is_invite: bool,
         ) -> Codes:
             return Codes.CONSENT_NOT_GIVEN
 
-        join_mock = Mock(side_effect=user_may_join_room)
+        join_mock = Mock(side_effect=user_may_join_room_codes)
         self.hs.get_spam_checker()._user_may_join_room_callbacks.append(join_mock)
 
         channel = self.make_request(
@@ -834,6 +834,25 @@ class RoomsCreateTestCase(RoomBase):
 
         self.assertEqual(join_mock.call_count, 0)
 
+        # Now change the return value of the callback to deny any join. Since we're
+        # creating the room, despite the return value, we should be able to join.
+        async def user_may_join_room_tuple(
+            mxid: str,
+            room_id: str,
+            is_invite: bool,
+        ) -> Tuple[Codes, dict]:
+            return Codes.INCOMPATIBLE_ROOM_VERSION, {}
+
+        join_mock.side_effect = user_may_join_room_tuple
+
+        channel = self.make_request(
+            "POST",
+            "/createRoom",
+            {},
+        )
+        self.assertEqual(channel.code, 200, channel.json_body)
+        self.assertEqual(join_mock.call_count, 0)
+
 
 class RoomTopicTestCase(RoomBase):
     """Tests /rooms/$room_id/topic REST events."""
@@ -1113,13 +1132,15 @@ class RoomJoinTestCase(RoomBase):
         """
 
         # Register a dummy callback. Make it allow all room joins for now.
-        return_value: Union[Literal["NOT_SPAM"], Codes] = synapse.module_api.NOT_SPAM
+        return_value: Union[
+            Literal["NOT_SPAM"], Tuple[Codes, dict], Codes
+        ] = synapse.module_api.NOT_SPAM
 
         async def user_may_join_room(
             userid: str,
             room_id: str,
             is_invited: bool,
-        ) -> Union[Literal["NOT_SPAM"], Codes]:
+        ) -> Union[Literal["NOT_SPAM"], Tuple[Codes, dict], Codes]:
             return return_value
 
         # `spec` argument is needed for this function mock to have `__qualname__`, which
@@ -1163,8 +1184,28 @@ class RoomJoinTestCase(RoomBase):
         )
 
         # Now make the callback deny all room joins, and check that a join actually fails.
+        # We pick an arbitrary Codes rather than the default `Codes.FORBIDDEN`.
         return_value = Codes.CONSENT_NOT_GIVEN
-        self.helper.join(self.room3, self.user2, expect_code=403, tok=self.tok2)
+        self.helper.invite(self.room3, self.user1, self.user2, tok=self.tok1)
+        self.helper.join(
+            self.room3,
+            self.user2,
+            expect_code=403,
+            expect_errcode=return_value,
+            tok=self.tok2,
+        )
+
+        # Now make the callback deny all room joins, and check that a join actually fails.
+        # As above, with the experimental extension that lets us return dictionaries.
+        return_value = (Codes.BAD_ALIAS, {"another_field": "12345"})
+        self.helper.join(
+            self.room3,
+            self.user2,
+            expect_code=403,
+            expect_errcode=return_value[0],
+            tok=self.tok2,
+            expect_additional_fields=return_value[1],
+        )
 
 
 class RoomJoinRatelimitTestCase(RoomBase):
@@ -1314,6 +1355,97 @@ class RoomMessagesTestCase(RoomBase):
         channel = self.make_request("PUT", path, content)
         self.assertEqual(200, channel.code, msg=channel.result["body"])
 
+    @parameterized.expand(
+        [
+            # Allow
+            param(
+                name="NOT_SPAM", value="NOT_SPAM", expected_code=200, expected_fields={}
+            ),
+            param(name="False", value=False, expected_code=200, expected_fields={}),
+            # Block
+            param(
+                name="scalene string",
+                value="ANY OTHER STRING",
+                expected_code=403,
+                expected_fields={"errcode": "M_FORBIDDEN"},
+            ),
+            param(
+                name="True",
+                value=True,
+                expected_code=403,
+                expected_fields={"errcode": "M_FORBIDDEN"},
+            ),
+            param(
+                name="Code",
+                value=Codes.LIMIT_EXCEEDED,
+                expected_code=403,
+                expected_fields={"errcode": "M_LIMIT_EXCEEDED"},
+            ),
+            param(
+                name="Tuple",
+                value=(Codes.SERVER_NOT_TRUSTED, {"additional_field": "12345"}),
+                expected_code=403,
+                expected_fields={
+                    "errcode": "M_SERVER_NOT_TRUSTED",
+                    "additional_field": "12345",
+                },
+            ),
+        ]
+    )
+    def test_spam_checker_check_event_for_spam(
+        self,
+        name: str,
+        value: Union[str, bool, Codes, Tuple[Codes, JsonDict]],
+        expected_code: int,
+        expected_fields: dict,
+    ) -> None:
+        class SpamCheck:
+            mock_return_value: Union[
+                str, bool, Codes, Tuple[Codes, JsonDict], bool
+            ] = "NOT_SPAM"
+            mock_content: Optional[JsonDict] = None
+
+            async def check_event_for_spam(
+                self,
+                event: synapse.events.EventBase,
+            ) -> Union[str, Codes, Tuple[Codes, JsonDict], bool]:
+                self.mock_content = event.content
+                return self.mock_return_value
+
+        spam_checker = SpamCheck()
+
+        self.hs.get_spam_checker()._check_event_for_spam_callbacks.append(
+            spam_checker.check_event_for_spam
+        )
+
+        # Inject `value` as mock_return_value
+        spam_checker.mock_return_value = value
+        path = "/rooms/%s/send/m.room.message/check_event_for_spam_%s" % (
+            urlparse.quote(self.room_id),
+            urlparse.quote(name),
+        )
+        body = "test-%s" % name
+        content = '{"body":"%s","msgtype":"m.text"}' % body
+        channel = self.make_request("PUT", path, content)
+
+        # Check that the callback has witnessed the correct event.
+        self.assertIsNotNone(spam_checker.mock_content)
+        if (
+            spam_checker.mock_content is not None
+        ):  # Checked just above, but mypy doesn't know about that.
+            self.assertEqual(
+                spam_checker.mock_content["body"], body, spam_checker.mock_content
+            )
+
+        # Check that we have the correct result.
+        self.assertEqual(expected_code, channel.code, msg=channel.result["body"])
+        for expected_key, expected_value in expected_fields.items():
+            self.assertEqual(
+                channel.json_body.get(expected_key, None),
+                expected_value,
+                "Field %s absent or invalid " % expected_key,
+            )
+
 
 class RoomPowerLevelOverridesTestCase(RoomBase):
     """Tests that the power levels can be overridden with server config."""
@@ -3235,7 +3367,8 @@ class ThreepidInviteTestCase(unittest.HomeserverTestCase):
         make_invite_mock.assert_called_once()
 
         # Now change the return value of the callback to deny any invite and test that
-        # we can't send the invite.
+        # we can't send the invite. We pick an arbitrary error code to be able to check
+        # that the same code has been returned
         mock.return_value = make_awaitable(Codes.CONSENT_NOT_GIVEN)
         channel = self.make_request(
             method="POST",
@@ -3249,6 +3382,27 @@ class ThreepidInviteTestCase(unittest.HomeserverTestCase):
             access_token=self.tok,
         )
         self.assertEqual(channel.code, 403)
+        self.assertEqual(channel.json_body["errcode"], Codes.CONSENT_NOT_GIVEN)
+
+        # Also check that it stopped before calling _make_and_store_3pid_invite.
+        make_invite_mock.assert_called_once()
+
+        # Run variant with `Tuple[Codes, dict]`.
+        mock.return_value = make_awaitable((Codes.EXPIRED_ACCOUNT, {"field": "value"}))
+        channel = self.make_request(
+            method="POST",
+            path="/rooms/" + self.room_id + "/invite",
+            content={
+                "id_server": "example.com",
+                "id_access_token": "sometoken",
+                "medium": "email",
+                "address": email_to_invite,
+            },
+            access_token=self.tok,
+        )
+        self.assertEqual(channel.code, 403)
+        self.assertEqual(channel.json_body["errcode"], Codes.EXPIRED_ACCOUNT)
+        self.assertEqual(channel.json_body["field"], "value")
 
         # Also check that it stopped before calling _make_and_store_3pid_invite.
         make_invite_mock.assert_called_once()
diff --git a/tests/rest/client/utils.py b/tests/rest/client/utils.py
index a0788b1bb0..93f749744d 100644
--- a/tests/rest/client/utils.py
+++ b/tests/rest/client/utils.py
@@ -41,6 +41,7 @@ from twisted.web.resource import Resource
 from twisted.web.server import Site
 
 from synapse.api.constants import Membership
+from synapse.api.errors import Codes
 from synapse.server import HomeServer
 from synapse.types import JsonDict
 
@@ -171,6 +172,8 @@ class RestHelper:
         expect_code: int = HTTPStatus.OK,
         tok: Optional[str] = None,
         appservice_user_id: Optional[str] = None,
+        expect_errcode: Optional[Codes] = None,
+        expect_additional_fields: Optional[dict] = None,
     ) -> None:
         self.change_membership(
             room=room,
@@ -180,6 +183,8 @@ class RestHelper:
             appservice_user_id=appservice_user_id,
             membership=Membership.JOIN,
             expect_code=expect_code,
+            expect_errcode=expect_errcode,
+            expect_additional_fields=expect_additional_fields,
         )
 
     def knock(
@@ -263,6 +268,7 @@ class RestHelper:
         appservice_user_id: Optional[str] = None,
         expect_code: int = HTTPStatus.OK,
         expect_errcode: Optional[str] = None,
+        expect_additional_fields: Optional[dict] = None,
     ) -> None:
         """
         Send a membership state event into a room.
@@ -323,6 +329,21 @@ class RestHelper:
                 channel.result["body"],
             )
 
+        if expect_additional_fields is not None:
+            for expect_key, expect_value in expect_additional_fields.items():
+                assert expect_key in channel.json_body, "Expected field %s, got %s" % (
+                    expect_key,
+                    channel.json_body,
+                )
+                assert (
+                    channel.json_body[expect_key] == expect_value
+                ), "Expected: %s at %s, got: %s, resp: %s" % (
+                    expect_value,
+                    expect_key,
+                    channel.json_body[expect_key],
+                    channel.json_body,
+                )
+
         self.auth_user_id = temp_id
 
     def send(
diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py
index 1c67e1ca91..79727c430f 100644
--- a/tests/rest/media/v1/test_media_storage.py
+++ b/tests/rest/media/v1/test_media_storage.py
@@ -23,11 +23,13 @@ from urllib import parse
 import attr
 from parameterized import parameterized, parameterized_class
 from PIL import Image as Image
+from typing_extensions import Literal
 
 from twisted.internet import defer
 from twisted.internet.defer import Deferred
 from twisted.test.proto_helpers import MemoryReactor
 
+from synapse.api.errors import Codes
 from synapse.events import EventBase
 from synapse.events.spamcheck import load_legacy_spam_checkers
 from synapse.logging.context import make_deferred_yieldable
@@ -570,9 +572,11 @@ class MediaRepoTests(unittest.HomeserverTestCase):
         )
 
 
-class TestSpamChecker:
+class TestSpamCheckerLegacy:
     """A spam checker module that rejects all media that includes the bytes
     `evil`.
+
+    Uses the legacy Spam-Checker API.
     """
 
     def __init__(self, config: Dict[str, Any], api: ModuleApi) -> None:
@@ -613,7 +617,7 @@ class TestSpamChecker:
         return b"evil" in buf.getvalue()
 
 
-class SpamCheckerTestCase(unittest.HomeserverTestCase):
+class SpamCheckerTestCaseLegacy(unittest.HomeserverTestCase):
     servlets = [
         login.register_servlets,
         admin.register_servlets,
@@ -637,7 +641,8 @@ class SpamCheckerTestCase(unittest.HomeserverTestCase):
             {
                 "spam_checker": [
                     {
-                        "module": TestSpamChecker.__module__ + ".TestSpamChecker",
+                        "module": TestSpamCheckerLegacy.__module__
+                        + ".TestSpamCheckerLegacy",
                         "config": {},
                     }
                 ]
@@ -662,3 +667,62 @@ class SpamCheckerTestCase(unittest.HomeserverTestCase):
         self.helper.upload_media(
             self.upload_resource, data, tok=self.tok, expect_code=400
         )
+
+
+EVIL_DATA = b"Some evil data"
+EVIL_DATA_EXPERIMENT = b"Some evil data to trigger the experimental tuple API"
+
+
+class SpamCheckerTestCase(unittest.HomeserverTestCase):
+    servlets = [
+        login.register_servlets,
+        admin.register_servlets,
+    ]
+
+    def prepare(self, reactor: MemoryReactor, clock: Clock, hs: HomeServer) -> None:
+        self.user = self.register_user("user", "pass")
+        self.tok = self.login("user", "pass")
+
+        # Allow for uploading and downloading to/from the media repo
+        self.media_repo = hs.get_media_repository_resource()
+        self.download_resource = self.media_repo.children[b"download"]
+        self.upload_resource = self.media_repo.children[b"upload"]
+
+        hs.get_module_api().register_spam_checker_callbacks(
+            check_media_file_for_spam=self.check_media_file_for_spam
+        )
+
+    async def check_media_file_for_spam(
+        self, file_wrapper: ReadableFileWrapper, file_info: FileInfo
+    ) -> Union[Codes, Literal["NOT_SPAM"]]:
+        buf = BytesIO()
+        await file_wrapper.write_chunks_to(buf.write)
+
+        if buf.getvalue() == EVIL_DATA:
+            return Codes.FORBIDDEN
+        elif buf.getvalue() == EVIL_DATA_EXPERIMENT:
+            return (Codes.FORBIDDEN, {})
+        else:
+            return "NOT_SPAM"
+
+    def test_upload_innocent(self) -> None:
+        """Attempt to upload some innocent data that should be allowed."""
+        self.helper.upload_media(
+            self.upload_resource, SMALL_PNG, tok=self.tok, expect_code=200
+        )
+
+    def test_upload_ban(self) -> None:
+        """Attempt to upload some data that includes bytes "evil", which should
+        get rejected by the spam checker.
+        """
+
+        self.helper.upload_media(
+            self.upload_resource, EVIL_DATA, tok=self.tok, expect_code=400
+        )
+
+        self.helper.upload_media(
+            self.upload_resource,
+            EVIL_DATA_EXPERIMENT,
+            tok=self.tok,
+            expect_code=400,
+        )