From 7e8083eb48fc22e1dde8b3a3d171f16acdfd6a25 Mon Sep 17 00:00:00 2001
From: Erik Johnston <erik@matrix.org>
Date: Wed, 3 Feb 2021 16:44:16 +0000
Subject: Add check_media_file_for_spam spam checker hook

---
 tests/rest/media/v1/test_media_storage.py | 94 +++++++++++++++++++++++++++++++
 1 file changed, 94 insertions(+)

(limited to 'tests')

diff --git a/tests/rest/media/v1/test_media_storage.py b/tests/rest/media/v1/test_media_storage.py
index a6c6985173..c279eb49e3 100644
--- a/tests/rest/media/v1/test_media_storage.py
+++ b/tests/rest/media/v1/test_media_storage.py
@@ -30,6 +30,8 @@ from twisted.internet import defer
 from twisted.internet.defer import Deferred
 
 from synapse.logging.context import make_deferred_yieldable
+from synapse.rest import admin
+from synapse.rest.client.v1 import login
 from synapse.rest.media.v1._base import FileInfo
 from synapse.rest.media.v1.filepath import MediaFilePaths
 from synapse.rest.media.v1.media_storage import MediaStorage
@@ -37,6 +39,7 @@ from synapse.rest.media.v1.storage_provider import FileStorageProviderBackend
 
 from tests import unittest
 from tests.server import FakeSite, make_request
+from tests.utils import default_config
 
 
 class MediaStorageTests(unittest.HomeserverTestCase):
@@ -398,3 +401,94 @@ class MediaRepoTests(unittest.HomeserverTestCase):
             headers.getRawHeaders(b"X-Robots-Tag"),
             [b"noindex, nofollow, noarchive, noimageindex"],
         )
+
+
+class TestSpamChecker:
+    """A spam checker module that rejects all media that includes the bytes
+    `evil`.
+    """
+
+    def __init__(self, config, api):
+        self.config = config
+        self.api = api
+
+    def parse_config(config):
+        return config
+
+    async def check_event_for_spam(self, foo):
+        return False  # allow all events
+
+    async def user_may_invite(self, inviter_userid, invitee_userid, room_id):
+        return True  # allow all invites
+
+    async def user_may_create_room(self, userid):
+        return True  # allow all room creations
+
+    async def user_may_create_room_alias(self, userid, room_alias):
+        return True  # allow all room aliases
+
+    async def user_may_publish_room(self, userid, room_id):
+        return True  # allow publishing of all rooms
+
+    async def check_media_file_for_spam(self, file_wrapper, file_info) -> bool:
+        buf = BytesIO()
+        await file_wrapper.write_chunks_to(buf.write)
+
+        return b"evil" in buf.getvalue()
+
+
+class SpamCheckerTestCase(unittest.HomeserverTestCase):
+    servlets = [
+        login.register_servlets,
+        admin.register_servlets,
+    ]
+
+    def prepare(self, reactor, clock, hs):
+        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"]
+
+    def default_config(self):
+        config = default_config("test")
+
+        config.update(
+            {
+                "spam_checker": [
+                    {
+                        "module": TestSpamChecker.__module__ + ".TestSpamChecker",
+                        "config": {},
+                    }
+                ]
+            }
+        )
+
+        return config
+
+    def test_upload_innocent(self):
+        """Attempt to upload some innocent data that should be allowed.
+        """
+
+        image_data = unhexlify(
+            b"89504e470d0a1a0a0000000d4948445200000001000000010806"
+            b"0000001f15c4890000000a49444154789c63000100000500010d"
+            b"0a2db40000000049454e44ae426082"
+        )
+
+        self.helper.upload_media(
+            self.upload_resource, image_data, tok=self.tok, expect_code=200
+        )
+
+    def test_upload_ban(self):
+        """Attempt to upload some data that includes bytes "evil", which should
+        get rejected by the spam checker.
+        """
+
+        data = b"Some evil data"
+
+        self.helper.upload_media(
+            self.upload_resource, data, tok=self.tok, expect_code=400
+        )
-- 
cgit 1.4.1