Merge pull request #120 from majestrate/pysogs-better-filter-2022-08-23

apply profanity filter to username (in addition to post body)

Author: majestrate

sogs/model/post.py

@@ -0,0 +1,34 @@
+from .. import utils
+from .. import session_pb2 as protobuf
+
+
+class Post:
+    """Class representing a post made in an open group"""
+
+    _proto = None
+
+    def __init__(self, raw=None, *, user=None, text=None):
+        if isinstance(raw, bytes) or isinstance(raw, memoryview):
+            msg = protobuf.Content()
+            msg.ParseFromString(utils.remove_session_message_padding(raw))
+            self._proto = msg.dataMessage
+        if self._proto is None:
+            # TODO: implement other kinds of construction methods for Posts
+            raise ValueError('must provide raw message bytes')
+
+    @property
+    def text(self):
+        """ accessor for the post body """
+        return self._proto.body
+
+    @property
+    def username(self):
+        """ accessor for the username of the post's author """
+        if self.profile is None:
+            return
+        return self.profile.displayName
+
+    @property
+    def profile(self):
+        """ accessor for the user profile data containing things like username etc """
+        return self._proto.profile

sogs/model/room.py

@@ -4,6 +4,7 @@
 from ..web import app
 from .user import User
 from .file import File
+from .post import Post
 from .exc import (
     NoSuchRoom,
     NoSuchFile,
@@ -712,12 +713,14 @@ def should_filter(self, user: User, data: bytes):
         if config.PROFANITY_FILTER and not self.check_admin(user):
             import better_profanity
 
-            if better_profanity.profanity.contains_profanity(utils.message_body(data)):
-                if config.PROFANITY_SILENT:
-                    return True
-                else:
-                    # FIXME: can we send back some error code that makes Session not retry?
-                    raise PostRejected("filtration rejected message")
+            msg = Post(raw=data)
+            for part in (msg.text, msg.username):
+                if better_profanity.profanity.contains_profanity(part):
+                    if config.PROFANITY_SILENT:
+                        return True
+                    else:
+                        # FIXME: can we send back some error code that makes Session not retry?
+                        raise PostRejected("filtration rejected message")
         return False
 
     def _own_files(self, msg_id: int, files: List[int], user):

sogs/utils.py

@@ -1,21 +1,13 @@
 from . import crypto
 from . import config
 from . import http
-from . import session_pb2 as protobuf
 
 import base64
 from flask import request, abort, Response
 import json
 from typing import Union, Tuple
 
 
-def message_body(data: bytes):
-    """given a bunch of bytes for a protobuf message return the message's body"""
-    msg = protobuf.Content()
-    msg.ParseFromString(remove_session_message_padding(data))
-    return msg.dataMessage.body
-
-
 def encode_base64(data: bytes):
     return base64.b64encode(data).decode()