synapse-invite-checker/synapse_invite_checker/invite_checker.py

import treq
import json
import time
from twisted.internet.defer import inlineCallbacks, returnValue
from synapse.module_api import ModuleApi, errors
from synapse.types import UserID
import logging

logger = logging.getLogger(__name__)

class InviteCheckerConfig:
    def __init__(self, config):
        self.use_allowlist = config.get("use_allowlist", True)
        self.use_blocklist = config.get("use_blocklist", True)
        self.blocklist_allowlist_url = config.get("blocklist_allowlist_url", None)
        self.blocklist_rooms = config.get("blocklist_rooms", [])  # Blocklist for room names
        self.policy_room_ids = config.get("policy_room_ids", [])  # List of policy room IDs

    @staticmethod
    def parse_config(config):
        return InviteCheckerConfig(config)

class InviteChecker:
    def __init__(self, config, api: ModuleApi):
        self.api = api
        self.config = InviteCheckerConfig.parse_config(config)

        self.use_allowlist = self.config.use_allowlist
        self.use_blocklist = self.config.use_blocklist

        self.cache_expiry_time = 60
        self.cache_timestamp = 0
        self.blocklist = set()
        self.allowlist = set()
        self.allow_all_invites_on_error = False

        self.room_id_cache = {}
        self.blocklist_room_ids = set()

        self.api.register_spam_checker_callbacks(user_may_invite=self.user_may_invite)
        logger.info("InviteChecker initialized")

    @inlineCallbacks
    def fetch_json(self, url):
        logger.info(f"Fetching JSON data from: {url}")
        try:
            response = yield treq.get(url)
            if response.code == 200:
                try:
                    content = yield response.content()
                    data = json.loads(content.decode('utf-8'))
                    logger.debug(f"Received JSON data: {data}")
                    returnValue(data)
                except Exception as json_error:
                    logger.error(f"Failed to decode JSON data: {json_error}")
                    returnValue(None)
            else:
                logger.error(f"Failed to fetch JSON data. Status code: {response.code}")
                returnValue(None)
        except Exception as e:
            logger.error(f"Error while fetching JSON: {str(e)}")
            returnValue(None)

    @inlineCallbacks
    def fetch_policy_room_banlist(self):
        """Fetches the ban lists from multiple policy rooms using Synapse API."""
        if not self.config.policy_room_ids:
            return set()  # No policy rooms configured, return an empty set

        logger.info(f"Fetching ban lists from policy rooms: {self.config.policy_room_ids}")
        banned_entities = set()
        banned_entities_by_room = set()
        for room_id in self.config.policy_room_ids:
            logger.info(f"Fetching ban list from policy room: {room_id}")
            
            try:
                # Fetch all state events from the policy room
                state_events = yield self.api.get_room_state(room_id)
                
                if isinstance(state_events, dict):
                    logger.info(f"Received state events in dict format from room {room_id} with {len(state_events)} entries.")

                    # Loop over the dictionary of state events
                    for key, event in state_events.items():
                        event_type = event.get("type", "")
                        content = event.get("content", {})

                        # Check for ban events of type 'm.policy.rule.user' and 'm.policy.rule.server'
                        if event_type in ["m.policy.rule.user", "m.policy.rule.server"]:
                            entity = content.get('entity', '')
                            if entity:
                                banned_entities_by_room.add(entity)
                            #else:
                            #    logger.warning(f"Missing 'entity' in event: {event}")

                    logger.info(f"Fetched {len(banned_entities_by_room)} banned entities from policy room {room_id}.")
                    banned_entities = banned_entities_by_room.union(banned_entities)
                    banned_entities_by_room = set()
                    
                else:
                    logger.error(f"Unexpected response format from room {room_id}: {type(state_events)}")

            except Exception as e:
                logger.error(f"Failed to fetch policy room ban list from room {room_id}. Error: {str(e)}")

        logger.info(f"Total banned entities from all policy rooms: {len(banned_entities)}")
        return banned_entities

    @inlineCallbacks
    def update_blocklist_allowlist(self):
        """Fetch and update the blocklist, allowlist, and blocklisted room IDs."""
        logger.info("Updating blocklist, allowlist, and room blocklist")
        
        json_data = yield self.fetch_json(self.config.blocklist_allowlist_url)

        if json_data:
            self.allow_all_invites_on_error = False
            self.use_allowlist = json_data.get('use_allowlist', True)
            self.use_blocklist = json_data.get('use_blocklist', True)
            self.blocklist = set(json_data.get('blocklist', []))
            self.allowlist = set(json_data.get('allowlist', []))

            # Fetch and merge the policy room ban lists from multiple rooms
            policy_banlist = yield self.fetch_policy_room_banlist()
            self.blocklist.update(policy_banlist)

            self.blocklist_room_ids = set()
            for room_entry in json_data.get('blocklist_rooms', []):
                if room_entry.startswith('!'):
                    logger.info(f"Blocklisting room ID directly: {room_entry}")
                    self.blocklist_room_ids.add(room_entry)
                else:
                    room_id = yield self.resolve_room_id(room_entry)
                    if room_id:
                        logger.info(f"Blocklisting room: {room_entry} -> {room_id}")
                        self.blocklist_room_ids.add(room_id)
                    else:
                        logger.error(f"Failed to blocklist room: {room_entry}")

            logger.info(f"Updated blocklist with {len(self.blocklist)} entries and {len(self.blocklist_room_ids)} room IDs.")
        else:
            logger.error("Failed to update allowlist/blocklist due to missing JSON data.")
            self.allow_all_invites_on_error = True

    @inlineCallbacks
    def resolve_room_id(self, room_alias):
        """Resolve a room alias to a room_id and cache the result."""
        if room_alias in self.room_id_cache:
            returnValue(self.room_id_cache[room_alias])

        logger.info(f"Resolving room alias to room_id: {room_alias}")
        try:
            room_id = yield self.api.resolve_room_alias(room_alias)
            self.room_id_cache[room_alias] = room_id
            returnValue(room_id)
        except Exception as e:
            logger.error(f"Failed to resolve room alias {room_alias}: {e}")
            returnValue(None)

    @inlineCallbacks
    def get_blocklist_allowlist(self):
        current_time = time.time()

        if current_time - self.cache_timestamp > self.cache_expiry_time:
            yield self.update_blocklist_allowlist()

        if self.allow_all_invites_on_error:
            logger.info("Skipping allowlist/blocklist checks because of previous JSON fetch failure.")
            returnValue((set(), set(), set()))

        returnValue((self.blocklist, self.allowlist, self.blocklist_room_ids))

    @inlineCallbacks
    def user_may_invite(self, inviter: str, invitee: str, room_id: str):
        logger.info(f"Checking invite from {inviter} to {invitee} for room {room_id}")

        if self.allow_all_invites_on_error:
            logger.info(f"Allowing invite from {inviter} to {invitee} due to previous JSON fetch failure.")
            returnValue("NOT_SPAM")

        blocklist, allowlist, blocklist_room_ids = yield self.get_blocklist_allowlist()
        inviter_domain = UserID.from_string(inviter).domain

        logger.debug(f"Blocklist: {blocklist}, Allowlist: {allowlist}, Blocklist Room IDs: {blocklist_room_ids}")

        if room_id in blocklist_room_ids:
            logger.info(f"Invite blocked: room {room_id} is blocklisted")
            returnValue(errors.Codes.FORBIDDEN)

        if self.use_allowlist:
            logger.info(f"Allowlist enabled. Checking domain: {inviter_domain}")
            if inviter_domain in allowlist:
                logger.info(f"Invite allowed: {inviter_domain} is in the allowlist")
                returnValue("NOT_SPAM")

        if self.use_blocklist:
            logger.info(f"Blocklist enabled. Checking domain: {inviter_domain}")
            if inviter_domain in blocklist:
                logger.info(f"Invite blocked: {inviter_domain} is in the blocklist")
                returnValue(errors.Codes.FORBIDDEN)

        if self.use_allowlist and inviter_domain not in allowlist:
            logger.info(f"Invite blocked: {inviter_domain} is not in the allowlist")
            returnValue(errors.Codes.FORBIDDEN)

        logger.info(f"Invite allowed by default: {inviter_domain}")
        returnValue("NOT_SPAM")