{content.formatted_body}
" f"{prefix.formatted_body}
" f"{caption.formatted_body}
" ), format=Format.HTML, external_url=external_url, ) else: prefix.ensure_has_html() caption.ensure_has_html() combined_body = "\n".join((prefix.body, caption.body)) combined_formatted_body = ( f"{prefix.formatted_body}
{caption.formatted_body}
" ) combined = content combined["filename"] = content.body combined.body = combined_body combined["format"] = str(Format.HTML) combined["org.matrix.msc1767.caption"] = { "org.matrix.msc1767.text": combined_body, "org.matrix.msc1767.html": combined_formatted_body, } combined["formatted_body"] = combined_formatted_body if share_item.caption: combined["com.beeper.raw_caption_text"] = share_item.caption.text combined["com.beeper.instagram_item_username"] = share_item.caption.user.username if share_item.user: combined["com.beeper.instagram_item_username"] = share_item.user.username if item.direct_media_share and item.direct_media_share.media_share_type == "tag": combined["com.beeper.relation_preview_type"] = "post_mention" elif item_type_name == "clip": combined["com.beeper.relation_preview_type"] = "reel" return [(EventType.ROOM_MESSAGE, combined)] else: return [ (EventType.ROOM_MESSAGE, prefix), (EventType.ROOM_MESSAGE, content), (EventType.ROOM_MESSAGE, caption), ] async def _convert_instagram_xma_media_share( self, source: u.User, intent: IntentAPI, item: ThreadItem ) -> list[ConvertedMessage]: # N.B. _get_instagram_media_info also only supports downloading the first xma item xma_list = ( item.xma_media_share or item.xma_story_share or item.xma_reel_share or item.xma_reel_mention or item.xma_clip or item.generic_xma or item.avatar_sticker ) media = xma_list[0] if len(xma_list) != 1: self.log.warning(f"Item {item.item_id} has multiple xma media share parts") if media.xma_layout_type not in (0, 4): self.log.warning(f"Unrecognized xma layout type {media.xma_layout_type}") if media.preview_url or media.preview_url_info: _, content = await self._convert_instagram_media(source, intent, item) if item.xma_story_share: content["com.beeper.relation_preview_type"] = "story" content["com.beeper.instagram_item_username"] = media.header_title_text elif item.xma_reel_share: if item.message_item_type == "reaction": content["com.beeper.relation_preview_type"] = "story_reaction" content["com.beeper.raw_reaction"] = item.text elif item.message_item_type == "text": content["com.beeper.relation_preview_type"] = "story_reply" content["com.beeper.raw_reply_text"] = item.text elif item.xma_reel_mention: content["com.beeper.relation_preview_type"] = "story_mention" # You mentioned them if item.user_id == source.igpk: mention = await p.Puppet.get_by_pk(self.other_user_pk) if mention: content["com.beeper.instagram_mention"] = mention.username # They mentioned you else: owner = await p.Puppet.get_by_pk(item.user_id) if owner: content["com.beeper.instagram_item_username"] = owner.username else: content = None # Post shares (layout type 0): media title text # Reel shares/replies/reactions (layout type 4): item text caption_text = media.title_text or item.text or "" post_caption_text = None if media.subtitle_text: caption_text = ( f"{caption_text}\n{media.subtitle_text}" if caption_text else media.subtitle_text ) header_text = media.header_title_text or "" # Note replies have title_text for sender username, caption_body_text for the original note # and item.text for the reply itself. if not header_text and media.caption_body_text: header_text = caption_text caption_text = media.caption_body_text post_caption_text = item.text escaped_caption_text = html.escape(caption_text).replace("\n", "{escaped_caption_text}" if escaped_caption_text else "" ) if media.target_url and media.target_url.startswith("https://"): caption_body = ( f"> {caption_text}\n\n{media.target_url}" if caption_text else media.target_url ) target_url_pretty = str(URL(media.target_url).with_query(None)).replace( "https://www.", "" ) caption_formatted_body += ( f'' ) else: caption_body = f"> {caption_text}" if caption_text else "" if post_caption_text: caption_formatted_body += f"
{html.escape(post_caption_text)}
" caption_body += f"\n\n{post_caption_text}" # Add auxiliary text as prefix for caption if item.auxiliary_text: caption_formatted_body = ( f"{html.escape(item.auxiliary_text)}
{caption_formatted_body}" ) caption_body = f"{item.auxiliary_text}\n\n{caption_body}" elif len(xma_list) > 1: caption_formatted_body = f"Sent {len(xma_list)} items
{caption_formatted_body}" caption_body = f"Sent {len(xma_list)} items\n\n{caption_body}" caption = TextMessageEventContent( msgtype=MessageType.TEXT, body=caption_body, formatted_body=caption_formatted_body, format=Format.HTML, ) if content and media.target_url: content.external_url = media.target_url caption.external_url = media.target_url if not caption_body: if content is None: return [] return [(EventType.ROOM_MESSAGE, content)] elif content is None: return [(EventType.ROOM_MESSAGE, caption)] elif self.bridge.config["bridge.caption_in_message"]: if isinstance(content, TextMessageEventContent): content.ensure_has_html() caption.ensure_has_html() content.body += f"\n\n{caption.body}" content.formatted_body = ( f"{content.formatted_body}
{caption.formatted_body}" ) else: content["filename"] = content.body content.body = caption.body content["format"] = str(Format.HTML) content["formatted_body"] = caption.formatted_body content["org.matrix.msc1767.caption"] = { "org.matrix.msc1767.text": content.body, "org.matrix.msc1767.html": content["formatted_body"], } return [(EventType.ROOM_MESSAGE, content)] else: return [(EventType.ROOM_MESSAGE, content), (EventType.ROOM_MESSAGE, caption)] # TODO this is probably unused async def _convert_instagram_reel_share( self, source: u.User, intent: IntentAPI, item: ThreadItem ) -> list[ConvertedMessage]: assert item.reel_share media = item.reel_share.media prefix_html = None if item.reel_share.type == ReelShareType.REPLY: if item.reel_share.reel_owner_id == source.igpk: prefix = "Replied to your story" else: username = media.user.username prefix = f"Sent @{username}'s story" user_link = f'@{username}' prefix_html = f"Sent {user_link}'s story" elif item.reel_share.type == ReelShareType.REACTION: if item.reel_share.reel_owner_id == source.igpk: prefix = "Reacted to your story" elif item.user_id == source.igpk: prefix = "You reacted to their story" else: prefix = "Reacted to a story" elif item.reel_share.type == ReelShareType.MENTION: if item.reel_share.mentioned_user_id == source.igpk: prefix = "Mentioned you in their story" else: prefix = "You mentioned them in your story" else: self.log.debug(f"Unsupported reel share type {item.reel_share.type}") return [] prefix_content = TextMessageEventContent(msgtype=MessageType.NOTICE, body=prefix) if prefix_html: prefix_content.format = Format.HTML prefix_content.formatted_body = prefix_html caption_content = TextMessageEventContent( msgtype=MessageType.TEXT, body=item.reel_share.text ) if not caption_content.body and isinstance(media, MediaShareItem): caption_content.body = media.caption.text if media.caption else "" if not caption_content.body: caption_content.body = "{media_content.formatted_body}
" f"{prefix_content.formatted_body}
" f"{caption_content.formatted_body}
" ), format=Format.HTML, ) else: prefix_content.ensure_has_html() caption_content.ensure_has_html() combined_body = "\n".join((prefix_content.body, caption_content.body)) combined_formatted_body = ( f"{prefix_content.formatted_body}
" f"{caption_content.formatted_body}
" ) combined = media_content combined["filename"] = combined.body combined.body = combined_body combined["format"] = str(Format.HTML) combined["org.matrix.msc1767.caption"] = { "org.matrix.msc1767.text": combined_body, "org.matrix.msc1767.html": combined_formatted_body, } combined["formatted_body"] = combined_formatted_body else: combined = caption_content return [(EventType.ROOM_MESSAGE, combined)] else: await self._send_message(intent, prefix_content, timestamp=item.timestamp_ms) converted: list[ConvertedMessage] = [] if media_content: converted.append((EventType.ROOM_MESSAGE, media_content)) converted.append((EventType.ROOM_MESSAGE, caption_content)) return converted async def _convert_instagram_link( self, source: u.User, intent: IntentAPI, item: ThreadItem, ) -> ConvertedMessage: content = TextMessageEventContent(msgtype=MessageType.TEXT, body=item.link.text) link = item.link.link_context preview = { "og:url": link.link_url, "og:title": link.link_title, "og:description": link.link_summary, } if link.link_image_url: reuploaded = await self._reupload_instagram_file( source, link.link_image_url, msgtype=None, info=ImageInfo(), intent=intent ) preview["og:image"] = reuploaded.url preview["og:image:type"] = reuploaded.info.mimetype preview["og:image:width"] = reuploaded.info.width preview["og:image:height"] = reuploaded.info.height preview["matrix:image:size"] = reuploaded.info.size if reuploaded.file: preview["beeper:image:encryption"] = reuploaded.file.serialize() preview = {k: v for k, v in preview.items() if v} content["com.beeper.linkpreviews"] = [preview] if "og:title" in preview else [] await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_expired_placeholder( self, source: u.User, item: ThreadItem, action: str ) -> ConvertedMessage: if item.user_id == source.igpk: prefix = f"{action} your story" elif item.user_id == source.igpk: prefix = f"You {action.lower()} their story" else: prefix = f"{action} a story" body = f"{prefix}\n\nNo longer available" html = f"{prefix}
No longer available
" content = TextMessageEventContent( msgtype=MessageType.NOTICE, body=body, format=Format.HTML, formatted_body=html ) return EventType.ROOM_MESSAGE, content async def _convert_instagram_text(self, item: ThreadItem, text: str) -> ConvertedMessage: content = TextMessageEventContent(msgtype=MessageType.TEXT, body=text) content["com.beeper.linkpreviews"] = [] await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_instagram_placeholder(self, item: ThreadItem) -> ConvertedMessage: content = TextMessageEventContent( msgtype=MessageType.NOTICE, body=item.placeholder.message ) if content.body == "Update to the latest version of Instagram to view this message.": content.body = "This message type is not currently supported" content["com.beeper.linkpreviews"] = [] await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_instagram_unhandled(self, item: ThreadItem) -> ConvertedMessage: content = TextMessageEventContent( msgtype=MessageType.NOTICE, body=f"Unsupported message type {item.item_type.value}" ) await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_instagram_location(self, item: ThreadItem) -> ConvertedMessage | None: loc = item.location if not loc or not loc.lng or not loc.lat: # TODO handle somehow return None long_char = "E" if loc.lng > 0 else "W" lat_char = "N" if loc.lat > 0 else "S" body = ( f"{loc.name} - {round(abs(loc.lat), 4)}° {lat_char}, " f"{round(abs(loc.lng), 4)}° {long_char}" ) url = f"https://www.openstreetmap.org/#map=15/{loc.lat}/{loc.lng}" external_url = None if loc.external_source == "facebook_places": external_url = f"https://www.facebook.com/{loc.short_name}-{loc.facebook_places_id}" content = LocationMessageEventContent( msgtype=MessageType.LOCATION, geo_uri=f"geo:{loc.lat},{loc.lng}", body=f"Location: {body}\n{url}", external_url=external_url, ) content["format"] = str(Format.HTML) content["formatted_body"] = f"Location: {body}" await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_instagram_profile(self, item: ThreadItem) -> ConvertedMessage: username = item.profile.username user_link = f'@{username}' text = f"Shared @{username}'s profile" html = f"Shared {user_link}'s profile" content = TextMessageEventContent( msgtype=MessageType.TEXT, format=Format.HTML, body=text, formatted_body=html ) await self._add_instagram_reply(content, item.replied_to_message) return EventType.ROOM_MESSAGE, content async def _convert_instagram_xma_profile_share( self, item: ThreadItem ) -> list[ConvertedMessage]: assert item.xma_profile profile_messages = [] for profile in item.xma_profile: username = profile.header_title_text user_link = f'@{username}' text = f"Shared @{username}'s profile" html = f"Shared {user_link}'s profile" content = TextMessageEventContent( msgtype=MessageType.TEXT, format=Format.HTML, body=text, formatted_body=html ) await self._add_instagram_reply(content, item.replied_to_message) profile_messages.append((EventType.ROOM_MESSAGE, content)) return profile_messages async def _add_instagram_reply( self, content: MessageEventContent, reply_to: ThreadItem | None ) -> None: if not reply_to: return message = await DBMessage.get_by_item_id(reply_to.item_id, self.receiver) if not message: return content.set_reply(message.mxid) if not isinstance(content, TextMessageEventContent): return try: evt = await self.main_intent.get_event(message.mx_room, message.mxid) except (MNotFound, MForbidden): evt = None if not evt: return if evt.type == EventType.ROOM_ENCRYPTED: try: evt = await self.matrix.e2ee.decrypt(evt, wait_session_timeout=0) except DecryptionError: return if isinstance(evt.content, TextMessageEventContent): evt.content.trim_reply_fallback() content.set_reply(evt) async def handle_instagram_item( self, source: u.User, sender: p.Puppet, item: MessageSyncMessage ): client_context = item.client_context link_client_context = item.link.client_context if item.link else None cc = client_context if link_client_context: if not client_context: cc = f"link:{link_client_context}" elif client_context != link_client_context: cc = f"{client_context}/link:{link_client_context}" if client_context and client_context in self._reqid_dedup: self.log.debug( f"Ignoring message {item.item_id} ({cc}) by {item.user_id}" " as it was sent by us (client_context in dedup queue)" ) return [] elif link_client_context and link_client_context in self._reqid_dedup: self.log.debug( f"Ignoring message {item.item_id} ({cc}) by {item.user_id}" " as it was sent by us (link.client_context in dedup queue)" ) return [] # Check in-memory queues for duplicates if item.item_id in self._msgid_dedup: self.log.debug( f"Ignoring message {item.item_id} ({item.client_context}) by {item.user_id}" " as it was already handled (message.id in dedup queue)" ) return self._msgid_dedup.appendleft(item.item_id) # Check database for duplicates if await DBMessage.get_by_item_id(item.item_id, self.receiver) is not None: self.log.debug( f"Ignoring message {item.item_id} ({item.client_context}) by {item.user_id}" " as it was already handled (message.id in database)" ) return self.log.debug( f"Handling Instagram message {item.item_id} ({item.client_context}) by {item.user_id}" ) intent = sender.intent_for(self) background_task.create(intent.set_typing(self.mxid, timeout=0)) event_ids = [] for event_type, content in await self.convert_instagram_item(source, sender, item): event_ids.append( await self._send_message( intent, content, event_type=event_type, timestamp=item.timestamp_ms ) ) event_ids = [event_id for event_id in event_ids if event_id] if not event_ids: self.log.warning(f"Unhandled Instagram message {item.item_id}") return self.log.debug(f"Handled Instagram message {item.item_id} -> {event_ids}") await DBMessage( mxid=event_ids[-1], mx_room=self.mxid, item_id=item.item_id, client_context=item.client_context, receiver=self.receiver, sender=sender.igpk, ig_timestamp=item.timestamp, ).insert() await self._send_delivery_receipt(event_ids[-1]) async def convert_instagram_item( self, source: u.User, sender: p.Puppet, item: ThreadItem ) -> list[ConvertedMessage]: if not isinstance(item, ThreadItem): # Parsing these items failed, they should have been logged already return [] try: return await self._convert_instagram_item(source, sender, item) except Exception: self.log.exception("Fatal error converting Instagram item") self.log.trace("Item content: %s", item.serialize()) return [] async def _convert_instagram_item( self, source: u.User, sender: p.Puppet, item: ThreadItem ) -> list[ConvertedMessage]: intent = sender.intent_for(self) if ( item.xma_media_share or item.xma_reel_share or item.xma_reel_mention or item.xma_story_share or item.xma_clip or item.generic_xma or item.avatar_sticker ): return await self._convert_instagram_xma_media_share(source, intent, item) converted: list[ConvertedMessage] = [] handle_text = True if item.media or item.animated_media or item.voice_media or item.visual_media: converted.append(await self._convert_instagram_media(source, intent, item)) elif item.location: if loc_content := await self._convert_instagram_location(item): converted.append(loc_content) elif item.profile: converted.append(await self._convert_instagram_profile(item)) elif item.xma_profile: converted.extend(await self._convert_instagram_xma_profile_share(item)) elif item.reel_share: converted.extend(await self._convert_instagram_reel_share(source, intent, item)) elif ( item.media_share or item.direct_media_share or item.story_share or item.clip or item.felix_share ): converted.extend(await self._convert_instagram_media_share(source, intent, item)) elif item.item_type == ThreadItemType.EXPIRED_PLACEHOLDER: if item.message_item_type == "reaction": action = "Reacted to" else: action = "Shared" msg_type, expired = await self._convert_expired_placeholder(source, item, action) if self.bridge.config["bridge.caption_in_message"] and item.text: _, text = await self._convert_instagram_text(item, item.text) expired.ensure_has_html() text.ensure_has_html() combined = TextMessageEventContent( msgtype=MessageType.TEXT, body="\n".join((expired.body, text.body)), formatted_body=f"{expired.formatted_body}{text.formatted_body}
", format=Format.HTML, ) handle_text = False converted.append((msg_type, combined)) else: converted.append((msg_type, expired)) elif item.video_call_event: msg_type = ( MessageType.NOTICE if item.video_call_event.action == "video_call_ended" else MessageType.TEXT ) video_call_notification = TextMessageEventContent( msgtype=msg_type, body=item.video_call_event.description, formatted_body=f"{item.video_call_event.description}", format=Format.HTML, ) converted.append((EventType.ROOM_MESSAGE, video_call_notification)) elif item.action_log: # These probably don't need to be bridged self.log.debug(f"Ignoring action log message {item.item_id}") return [] # TODO handle item.clip? # TODO should these be put into a caption? if handle_text and item.text: converted.append(await self._convert_instagram_text(item, item.text)) elif item.like: # We handle likes as text because Matrix clients do big emoji on their own. converted.append(await self._convert_instagram_text(item, item.like)) elif item.link: converted.append(await self._convert_instagram_link(source, intent, item)) elif item.placeholder and len(converted) == 0: self.log.warning( f"Got placeholder item in {item.item_id}: {item.placeholder.serialize()}" ) converted.append(await self._convert_instagram_placeholder(item)) if len(converted) == 0: self.log.debug(f"Unhandled Instagram message {item.item_id}") converted.append(await self._convert_instagram_unhandled(item)) return converted def _deterministic_event_id( self, sender: p.Puppet, item_id: str, part_name: int | None = None ) -> EventID: hash_content = f"{self.mxid}/instagram/{sender.igpk}/{item_id}" if part_name: hash_content += f"/{part_name}" hashed = hashlib.sha256(hash_content.encode("utf-8")).digest() b64hash = base64.urlsafe_b64encode(hashed).decode("utf-8").rstrip("=") return EventID(f"${b64hash}:instagram.com") async def handle_instagram_remove(self, item_id: str) -> None: message = await DBMessage.get_by_item_id(item_id, self.receiver) if message is None: return await message.delete() if message.mxid: sender = await p.Puppet.get_by_pk(message.sender) try: await sender.intent_for(self).redact(self.mxid, message.mxid) except MForbidden: await self.main_intent.redact(self.mxid, message.mxid) self.log.debug(f"Redacted {message.mxid} after Instagram unsend") async def handle_instagram_reaction(self, item: ThreadItem, remove: bool) -> None: sender = await p.Puppet.get_by_pk(item.new_reaction.sender_id) message = await DBMessage.get_by_item_id(item.item_id, self.receiver) if not message: self.log.debug(f"Dropping reaction by {sender.pk} to unknown message {item.item_id}") return emoji = item.new_reaction.emoji async with self._reaction_lock: existing = await DBReaction.get_by_item_id(item.item_id, self.receiver, sender.pk) if not existing and remove: self.log.debug( f"Ignoring duplicate reaction removal by {sender.pk} to {item.item_id}" ) return elif not remove and existing and existing.reaction == emoji: self.log.debug(f"Ignoring duplicate reaction by {sender.pk} to {item.item_id}") return intent = sender.intent_for(self) if remove: await existing.delete() await intent.redact(self.mxid, existing.mxid) self.log.debug( f"Removed {sender.pk}'s reaction to {item.item_id} (redacted {existing.mxid})" ) else: timestamp = item.new_reaction.timestamp_ms reaction_event_id = await intent.react( self.mxid, message.mxid, key=emoji, timestamp=timestamp ) await self._upsert_reaction( existing, intent, reaction_event_id, message, sender, emoji, timestamp ) self.log.debug( f"Handled {sender.pk}'s reaction to {item.item_id} -> {reaction_event_id}" ) async def _handle_instagram_reactions( self, message: DBMessage, reactions: list[Reaction] ) -> None: old_reactions: dict[int, DBReaction] old_reactions = { reaction.ig_sender: reaction for reaction in await DBReaction.get_all_by_item_id(message.item_id, self.receiver) } for new_reaction in reactions: old_reaction = old_reactions.pop(new_reaction.sender_id, None) if old_reaction and old_reaction.reaction == new_reaction.emoji: continue puppet = await p.Puppet.get_by_pk(new_reaction.sender_id) intent = puppet.intent_for(self) timestamp = int(time.time() * 1000) reaction_event_id = await intent.react( self.mxid, message.mxid, new_reaction.emoji, timestamp=timestamp ) await self._upsert_reaction( old_reaction, intent, reaction_event_id, message, puppet, new_reaction.emoji, timestamp, ) for old_reaction in old_reactions.values(): await old_reaction.delete() puppet = await p.Puppet.get_by_pk(old_reaction.ig_sender) await puppet.intent_for(self).redact(self.mxid, old_reaction.mxid) async def handle_instagram_update(self, item: MessageSyncMessage) -> None: message = await DBMessage.get_by_item_id(item.item_id, self.receiver) if not message: return if item.has_seen: puppet = await p.Puppet.get_by_pk(item.has_seen, create=False) if puppet: await puppet.intent_for(self).mark_read(self.mxid, message.mxid) else: async with self._reaction_lock: await self._handle_instagram_reactions( message, (item.reactions.emojis if item.reactions else []) ) # endregion # region Updating portal info def _get_thread_name(self, thread: Thread) -> str: if self.is_direct: if self.other_user_pk == thread.viewer_id and len(thread.users) == 0: return "Instagram chat with yourself" elif len(thread.users) == 1: tpl = self.config["bridge.private_chat_name_template"] ui = thread.users[0] return tpl.format( displayname=ui.full_name or ui.username, id=ui.pk, username=ui.username ) elif thread.thread_title: return self.config["bridge.group_chat_name_template"].format(name=thread.thread_title) return "" async def update_thread_image( self, source: u.User, thread_image: ThreadImage, sender: p.Puppet | None = None ) -> bool: if ( self.is_direct or not thread_image or (self.thread_image_id == thread_image.id and self.avatar_set) ): return False best = thread_image.best_image if not best: return False data, mimetype = await self._download_instagram_file(source, best.url) if not data: return False self.thread_image_id = thread_image.id self.avatar_set = False mxc = await self.main_intent.upload_media( data=data, mime_type=mimetype, filename=str(thread_image.id), async_upload=self.config["homeserver.async_media"], ) return await self._update_photo(mxc, sender=sender) async def update_info(self, thread: Thread, source: u.User) -> None: changed = await self._update_name(self._get_thread_name(thread)) changed = await self.update_thread_image(source, thread.thread_image) or changed changed = await self._update_participants(thread.users, source) or changed if changed: await self.update_bridge_info() await self.update() # TODO update power levels with thread.admin_user_ids async def update_info_from_puppet(self, puppet: p.Puppet | None = None) -> None: if not self.is_direct: return if not puppet: puppet = await self.get_dm_puppet() await self._update_photo(puppet.photo_mxc) if self.name and not self.name_set: await self._update_name(self.name) async def _update_name(self, name: str) -> bool: if name and (self.name != name or (not self.name_set and self.set_dm_room_metadata)): self.name = name if self.mxid: try: await self.main_intent.set_room_name(self.mxid, name) self.name_set = True except Exception: self.log.exception("Failed to update name") self.name_set = False return True return False async def _update_photo(self, photo_mxc: ContentURI, sender: p.Puppet | None = None) -> bool: if self.avatar_url == photo_mxc and (self.avatar_set or not self.set_dm_room_metadata): return False self.avatar_url = photo_mxc self.avatar_set = False if self.mxid and self.set_dm_room_metadata: try: # TODO use sender intent await self.main_intent.set_room_avatar(self.mxid, photo_mxc) self.avatar_set = True except Exception: self.log.exception("Failed to set room avatar") return True async def _update_participants(self, users: list[ThreadUser], source: u.User) -> bool: meta_changed = False # Make sure puppets who should be here are here for user in users: puppet = await p.Puppet.get_by_pk(user.pk) await puppet.update_info(user, source) if self.mxid: await puppet.intent_for(self).ensure_joined(self.mxid) if puppet.pk == self.other_user_pk: meta_changed = await self._update_photo(puppet.photo_mxc) if self.mxid: # Kick puppets who shouldn't be here current_members = {int(user.pk) for user in users} for user_id in await self.main_intent.get_room_members(self.mxid): pk = p.Puppet.get_id_from_mxid(user_id) if pk and pk not in current_members and pk != self.other_user_pk: await self.main_intent.kick_user( self.mxid, p.Puppet.get_mxid_from_id(pk), reason="User had left this Instagram DM", ) return meta_changed async def _update_read_receipts(self, receipts: dict[int | str, ThreadUserLastSeenAt]) -> None: for user_id, receipt in receipts.items(): message: DBMessage | DBReaction message = await DBMessage.get_by_item_id(receipt.item_id, self.receiver) if not message: reaction: DBReaction message, reaction = await asyncio.gather( DBMessage.get_closest(self.mxid, int(receipt.timestamp)), DBReaction.get_closest(self.mxid, receipt.timestamp_ms), ) if (not message or not message.mxid) and not reaction: self.log.debug( "Couldn't find message %s to mark as read by %s", receipt, user_id ) continue elif not message or (reaction and reaction.mx_timestamp > message.ig_timestamp_ms): message = reaction puppet = await p.Puppet.get_by_pk(int(user_id), create=False) if not puppet: continue try: await puppet.intent_for(self).mark_read(message.mx_room, message.mxid) except Exception: self.log.warning( f"Failed to mark {message.mxid} in {message.mx_room} " f"as read by {puppet.intent.mxid}", exc_info=True, ) async def get_dm_puppet(self) -> p.Puppet | None: if not self.is_direct: return None return await p.Puppet.get_by_pk(self.other_user_pk) # endregion # region Backfill async def enqueue_immediate_backfill(self, source: u.User, priority: int) -> None: assert self.config["bridge.backfill.msc2716"] max_pages = self.config["bridge.backfill.incremental.max_pages"] max_total_pages = self.config["bridge.backfill.incremental.max_total_pages"] if max_pages <= 0 or max_total_pages == 0: return if not await Backfill.get(source.mxid, self.thread_id, self.receiver): await Backfill.new( source.mxid, priority, self.thread_id, self.receiver, max_pages, self.config["bridge.backfill.incremental.page_delay"], self.config["bridge.backfill.incremental.post_batch_delay"], max_total_pages, ).insert() async def backfill(self, source: u.User, backfill_request: Backfill) -> None: try: last_message_ig_timestamp = await self._backfill(source, backfill_request) if ( last_message_ig_timestamp is not None and not self.bridge.homeserver_software.is_hungry and self.config["bridge.backfill.msc2716"] ): await self.send_post_backfill_dummy(last_message_ig_timestamp) finally: # Always sleep after the backfill request is finished processing, even if it errors. await asyncio.sleep(backfill_request.post_batch_delay) async def _backfill(self, source: u.User, backfill_request: Backfill) -> int | None: assert source.client self.log.debug("Backfill request: %s", backfill_request) num_pages = backfill_request.num_pages self.log.debug( "Backfilling up to %d pages of history in %s through %s", num_pages, self.mxid, source.mxid, ) try: if self.cursor: self.log.debug( f"There is a cursor for the chat, fetching messages before {self.cursor}" ) resp = await source.client.get_thread( self.thread_id, seq_id=source.seq_id, cursor=self.cursor ) else: self.log.debug( "There is no first message in the chat, starting with the most recent messages" ) resp = await source.client.get_thread(self.thread_id, seq_id=source.seq_id) except IGRateLimitError as e: backoff = self.config.get("bridge.backfill.backoff.message_history", 300) self.log.warning( f"Backfilling failed due to rate limit. Waiting for {backoff} seconds before " f"resuming. Error: {e}" ) await asyncio.sleep(backoff) raise async def dedup_messages(messages: list[ThreadItem]) -> list[ThreadItem]: deduped = [] # Sometimes (seems like on Facebook chats) it fetches the first message in the chat over # and over again. for item in messages: # Check in-memory queues for duplicates if item.item_id in self._msgid_dedup: self.log.debug( f"Ignoring message {item.item_id} ({item.client_context}) by {item.user_id}" " as it was already handled (message.id in dedup queue)" ) continue self._msgid_dedup.appendleft(item.item_id) # Check database for duplicates if await DBMessage.get_by_item_id(item.item_id, self.receiver) is not None: self.log.debug( f"Ignoring message {item.item_id} ({item.client_context}) by {item.user_id}" " as it was already handled (message.id in database)" ) continue deduped.append(item) return deduped messages = await dedup_messages(resp.thread.items) cursor = resp.thread.oldest_cursor backfill_more = resp.thread.has_older if len(messages) == 0: self.log.debug("No messages to backfill.") return None last_message_timestamp = messages[-1].timestamp_ms pages_to_backfill = backfill_request.num_pages if backfill_request.max_total_pages > -1: pages_to_backfill = min(pages_to_backfill, backfill_request.max_total_pages) pages_backfilled = 0 for i in range(pages_to_backfill): base_insertion_event_id = await self.backfill_message_page( source, list(reversed(messages)) ) self.cursor = cursor await self.save() pages_backfilled += 1 if base_insertion_event_id: self.historical_base_insertion_event_id = base_insertion_event_id await self.save() if backfill_more and i < pages_to_backfill - 1: # Sleep before fetching another page of messages. await asyncio.sleep(backfill_request.page_delay) # Fetch more messages try: resp = await source.client.get_thread( self.thread_id, seq_id=source.seq_id, cursor=self.cursor ) messages = await dedup_messages(resp.thread.items) cursor = resp.thread.oldest_cursor backfill_more &= resp.thread.has_older except IGRateLimitError as e: backoff = self.config.get("bridge.backfill.backoff.message_history", 300) self.log.warning( f"Backfilling failed due to rate limit. Waiting for {backoff} seconds " "before resuming." ) await asyncio.sleep(backoff) # If we hit the rate limit, then we will want to give up for now, but enqueue # additional backfill to do later. break if backfill_request.max_total_pages == -1: new_max_total_pages = -1 else: new_max_total_pages = backfill_request.max_total_pages - pages_backfilled if new_max_total_pages <= 0: backfill_more = False if backfill_more: self.log.debug("Enqueueing more backfill") await Backfill.new( source.mxid, # Always enqueue subsequent backfills at the lowest priority 2, self.thread_id, self.receiver, backfill_request.num_pages, backfill_request.page_delay, backfill_request.post_batch_delay, new_max_total_pages, ).insert() else: self.log.debug("No more messages to backfill") await self._update_read_receipts(resp.thread.last_seen_at) return last_message_timestamp async def backfill_message_page( self, source: u.User, message_page: list[ThreadItem], forward: bool = False, last_message: DBMessage | None = None, mark_read: bool = False, ) -> EventID | None: """ Backfills a page of messages to Matrix. The messages should be in order from oldest to newest. Returns: a tuple containing the number of messages that were actually bridged, the timestamp of the oldest bridged message and the base insertion event ID if it exists. """ assert source.client if len(message_page) == 0: return None if forward: assert (last_message and last_message.mxid) or self.first_event_id prev_event_id = last_message.mxid if last_message else self.first_event_id else: assert self.config["bridge.backfill.msc2716"] assert self.first_event_id prev_event_id = self.first_event_id assert self.mxid oldest_message_in_page = message_page[0] oldest_msg_timestamp = oldest_message_in_page.timestamp_ms batch_messages: list[BatchSendEvent] = [] state_events_at_start: list[BatchSendStateEvent] = [] added_members = set() current_members = await self.main_intent.state_store.get_members( self.mxid, memberships=(Membership.JOIN,) ) def add_member(puppet: p.Puppet, mxid: UserID): assert self.mxid if mxid in added_members: return if ( self.bridge.homeserver_software.is_hungry or not self.config["bridge.backfill.msc2716"] ): # Hungryserv doesn't expect or check state events at start. added_members.add(mxid) return content_args = {"avatar_url": puppet.photo_mxc, "displayname": puppet.name} state_events_at_start.extend( [ BatchSendStateEvent( content=MemberStateEventContent(Membership.INVITE, **content_args), type=EventType.ROOM_MEMBER, sender=self.main_intent.mxid, state_key=mxid, timestamp=oldest_msg_timestamp, ), BatchSendStateEvent( content=MemberStateEventContent(Membership.JOIN, **content_args), type=EventType.ROOM_MEMBER, sender=mxid, state_key=mxid, timestamp=oldest_msg_timestamp, ), ] ) added_members.add(mxid) async def intent_for(user_id: int) -> tuple[p.Puppet, IntentAPI]: puppet: p.Puppet = await p.Puppet.get_by_pk(user_id) if puppet: intent = puppet.intent_for(self) else: intent = self.main_intent if puppet.is_real_user and not self._can_double_puppet_backfill(intent.mxid): intent = puppet.default_mxid_intent return puppet, intent message_infos: list[tuple[ThreadItem | Reaction, int]] = [] intents: list[IntentAPI] = [] for message in message_page: puppet, intent = await intent_for(message.user_id) # Convert the message converted = await self.convert_instagram_item(source, puppet, message) if not converted: self.log.debug(f"Skipping unsupported message in backfill {message.item_id}") continue if intent.mxid not in current_members: add_member(puppet, intent.mxid) d_event_id = None for index, (event_type, content) in enumerate(converted): if self.encrypted and self.matrix.e2ee: event_type, content = await self.matrix.e2ee.encrypt( self.mxid, event_type, content ) if intent.api.is_real_user and intent.api.bridge_name is not None: content[DOUBLE_PUPPET_SOURCE_KEY] = intent.api.bridge_name if self.bridge.homeserver_software.is_hungry: d_event_id = self._deterministic_event_id(puppet, message.item_id, index) message_infos.append((message, index)) batch_messages.append( BatchSendEvent( content=content, type=event_type, sender=intent.mxid, timestamp=message.timestamp_ms, event_id=d_event_id, ) ) intents.append(intent) if self.bridge.homeserver_software.is_hungry and message.reactions: for reaction in message.reactions.emojis: puppet, intent = await intent_for(reaction.sender_id) reaction_event = ReactionEventContent() reaction_event.relates_to = RelatesTo( rel_type=RelationType.ANNOTATION, event_id=d_event_id, key=reaction.emoji ) if intent.api.is_real_user and intent.api.bridge_name is not None: reaction_event[DOUBLE_PUPPET_SOURCE_KEY] = intent.api.bridge_name message_infos.append((reaction, 0)) batch_messages.append( BatchSendEvent( content=reaction_event, type=EventType.REACTION, sender=intent.mxid, timestamp=message.timestamp_ms, ) ) if not batch_messages: return None if ( not self.bridge.homeserver_software.is_hungry and self.config["bridge.backfill.msc2716"] and (forward or self.next_batch_id is None) ): self.log.debug("Sending dummy event to avoid forward extremity errors") await self.main_intent.send_message_event( self.mxid, EventType("fi.mau.dummy.pre_backfill", EventType.Class.MESSAGE), {} ) self.log.info( "Sending %d %s messages to %s with batch ID %s and previous event ID %s", len(batch_messages), "new" if forward else "historical", self.mxid, self.next_batch_id, prev_event_id, ) if self.bridge.homeserver_software.is_hungry: self.log.debug("Batch message event IDs %s", [m.event_id for m in batch_messages]) base_insertion_event_id = None if self.config["bridge.backfill.msc2716"]: batch_send_resp = await self.main_intent.batch_send( self.mxid, prev_event_id, batch_id=self.next_batch_id, events=batch_messages, state_events_at_start=state_events_at_start, beeper_new_messages=forward, beeper_mark_read_by=source.mxid if mark_read else None, ) base_insertion_event_id = batch_send_resp.base_insertion_event_id event_ids = batch_send_resp.event_ids else: batch_send_resp = None event_ids = [ await intent.send_message_event( self.mxid, evt.type, evt.content, timestamp=evt.timestamp ) for evt, intent in zip(batch_messages, intents) ] await self._finish_batch(event_ids, message_infos) if not forward: assert batch_send_resp self.log.debug("Got next batch ID %s for %s", batch_send_resp.next_batch_id, self.mxid) self.next_batch_id = batch_send_resp.next_batch_id await self.save() return base_insertion_event_id def _can_double_puppet_backfill(self, custom_mxid: UserID) -> bool: return self.config["bridge.backfill.double_puppet_backfill"] and ( # Hungryserv can batch send any users self.bridge.homeserver_software.is_hungry # Non-MSC2716 backfill can use any double puppet or not self.config["bridge.backfill.msc2716"] # Local users can be double puppeted even with MSC2716 or (custom_mxid[custom_mxid.index(":") + 1 :] == self.config["homeserver.domain"]) ) async def _finish_batch( self, event_ids: list[EventID], message_infos: list[tuple[ThreadItem | Reaction, int]] ): # We have to do this slightly annoying processing of the event IDs and message infos so # that we only map the last event ID to the message. # When inline captions are enabled, this will have no effect since index will always be 0 # since there's only ever one event per message. current_message = None messages = [] reactions = [] message_id = None for event_id, (message_or_reaction, index) in zip(event_ids, message_infos): if isinstance(message_or_reaction, ThreadItem): message = message_or_reaction if index == 0 and current_message: # This means that all of the events for the previous message have been processed, # and the current_message is the most recent event for that message. messages.append(current_message) current_message = DBMessage( mxid=event_id, mx_room=self.mxid, item_id=message.item_id, client_context=message.client_context, receiver=self.receiver, sender=message.user_id, ig_timestamp=message.timestamp, ) message_id = message.item_id else: assert message_id reaction = message_or_reaction reactions.append( DBReaction( mxid=event_id, mx_room=self.mxid, ig_item_id=message_id, ig_receiver=self.receiver, ig_sender=reaction.sender_id, reaction=reaction.emoji, mx_timestamp=reaction.timestamp_ms, ) ) if current_message: messages.append(current_message) try: await DBMessage.bulk_insert(messages) except Exception: self.log.exception("Failed to store batch message IDs") try: for reaction in reactions: await reaction.insert() except Exception: self.log.exception("Failed to store backfilled reactions") async def send_post_backfill_dummy( self, last_message_ig_timestamp: int, base_insertion_event_id: EventID | None = None, ): if not self.config["bridge.backfill.msc2716"]: return assert self.mxid if not base_insertion_event_id: base_insertion_event_id = self.historical_base_insertion_event_id if not base_insertion_event_id: self.log.debug( "No base insertion event ID in database or from batch send response. Not sending" " dummy event." ) return event_id = await self.main_intent.send_message_event( self.mxid, event_type=HistorySyncMarkerMessage, content={ "org.matrix.msc2716.marker.insertion": base_insertion_event_id, "m.marker.insertion": base_insertion_event_id, }, ) await DBMessage( mxid=event_id, mx_room=self.mxid, item_id=f"fi.mau.instagram.post_backfill_dummy.{last_message_ig_timestamp}", client_context=None, receiver=self.receiver, sender=0, ig_timestamp=last_message_ig_timestamp, ).insert() # endregion # region Bridge info state event @property def bridge_info_state_key(self) -> str: return f"net.maunium.instagram://instagram/{self.thread_id}" @property def bridge_info(self) -> dict[str, Any]: return { "bridgebot": self.az.bot_mxid, "creator": self.main_intent.mxid, "protocol": { "id": "instagram", "displayname": "Instagram DM", "avatar_url": self.config["appservice.bot_avatar"], }, "channel": { "id": self.thread_id, "displayname": self.name, "avatar_url": self.avatar_url, }, } async def update_bridge_info(self) -> None: if not self.mxid: self.log.debug("Not updating bridge info: no Matrix room created") return try: self.log.debug("Updating bridge info...") await self.main_intent.send_state_event( self.mxid, StateBridge, self.bridge_info, self.bridge_info_state_key ) # TODO remove this once https://github.com/matrix-org/matrix-doc/pull/2346 is in spec await self.main_intent.send_state_event( self.mxid, StateHalfShotBridge, self.bridge_info, self.bridge_info_state_key ) except Exception: self.log.warning("Failed to update bridge info", exc_info=True) # endregion # region Creating Matrix rooms async def create_matrix_room(self, source: u.User, info: Thread) -> RoomID | None: if self.mxid: try: await self.update_matrix_room(source, info) except Exception: self.log.exception("Failed to update portal") return self.mxid async with self._create_room_lock: try: return await self._create_matrix_room(source, info) except Exception: self.log.exception("Failed to create portal") return None def _get_invite_content(self, double_puppet: p.Puppet | None) -> dict[str, bool]: invite_content = {} if double_puppet: invite_content["fi.mau.will_auto_accept"] = True if self.is_direct: invite_content["is_direct"] = True return invite_content async def update_matrix_room(self, source: u.User, info: Thread) -> None: puppet = await p.Puppet.get_by_custom_mxid(source.mxid) await self.main_intent.invite_user( self.mxid, source.mxid, check_cache=True, extra_content=self._get_invite_content(puppet), ) if puppet: did_join = await puppet.intent.ensure_joined(self.mxid) if did_join and self.is_direct: await source.update_direct_chats({self.main_intent.mxid: [self.mxid]}) await self.update_info(info, source) await self._update_read_receipts(info.last_seen_at) async def _create_matrix_room(self, source: u.User, info: Thread) -> RoomID | None: if self.mxid: await self.update_matrix_room(source, info) return self.mxid await self.update_info(info, source) self.log.debug("Creating Matrix room") initial_state = [ { "type": str(StateBridge), "state_key": self.bridge_info_state_key, "content": self.bridge_info, }, # TODO remove this once https://github.com/matrix-org/matrix-doc/pull/2346 is in spec { "type": str(StateHalfShotBridge), "state_key": self.bridge_info_state_key, "content": self.bridge_info, }, ] invites = [] if self.config["bridge.encryption.default"] and self.matrix.e2ee: self.encrypted = True initial_state.append( { "type": "m.room.encryption", "content": self.get_encryption_state_event_json(), } ) if self.is_direct: invites.append(self.az.bot_mxid) if self.set_dm_room_metadata: self.name_set = bool(self.name) if self.avatar_url is not None: initial_state.append( { "type": str(EventType.ROOM_AVATAR), "content": {"url": self.avatar_url}, } ) self.avatar_set = True creation_content = {} if not self.config["bridge.federate_rooms"]: creation_content["m.federate"] = False self.mxid = await self.main_intent.create_room( name=self.name if self.set_dm_room_metadata else None, is_direct=self.is_direct, initial_state=initial_state, invitees=invites, creation_content=creation_content, ) if not self.mxid: raise Exception("Failed to create room: no mxid returned") if self.encrypted and self.matrix.e2ee and self.is_direct: try: await self.az.intent.ensure_joined(self.mxid) except Exception: self.log.warning(f"Failed to add bridge bot to new private chat {self.mxid}") await self.update() self.log.debug(f"Matrix room created: {self.mxid}") self.by_mxid[self.mxid] = self puppet = await p.Puppet.get_by_custom_mxid(source.mxid) await self.main_intent.invite_user( self.mxid, source.mxid, extra_content=self._get_invite_content(puppet) ) if puppet: try: if self.is_direct: await source.update_direct_chats({self.main_intent.mxid: [self.mxid]}) await puppet.intent.join_room_by_id(self.mxid) except MatrixError: self.log.debug( "Failed to join custom puppet into newly created portal", exc_info=True ) await self._update_participants(info.users, source) self.log.trace("Sending portal post-create dummy event") self.first_event_id = await self.main_intent.send_message_event( self.mxid, PortalCreateDummy, {} ) await self.update() return self.mxid # endregion # region Database getters async def postinit(self) -> None: self.by_thread_id[(self.thread_id, self.receiver)] = self if self.mxid: self.by_mxid[self.mxid] = self self._main_intent = ( (await p.Puppet.get_by_pk(self.other_user_pk)).default_mxid_intent if self.other_user_pk else self.az.intent ) async def delete(self) -> None: await DBMessage.delete_all(self.mxid) self.by_mxid.pop(self.mxid, None) self.mxid = None self.encrypted = False await self.update() async def save(self) -> None: await self.update() @classmethod def all_with_room(cls) -> AsyncGenerator[Portal, None]: return cls._db_to_portals(super().all_with_room()) @classmethod def find_private_chats_with(cls, other_user: int) -> AsyncGenerator[Portal, None]: return cls._db_to_portals(super().find_private_chats_with(other_user)) @classmethod async def find_private_chat(cls, receiver: int, other_user: int) -> Portal | None: thread_id = await super().find_private_chat_id(receiver, other_user) if not thread_id: return None return await cls.get_by_thread_id(thread_id, receiver=receiver, is_group=False) @classmethod async def _db_to_portals(cls, query: Awaitable[list[Portal]]) -> AsyncGenerator[Portal, None]: portals = await query for index, portal in enumerate(portals): try: yield cls.by_thread_id[(portal.thread_id, portal.receiver)] except KeyError: await portal.postinit() yield portal @classmethod @async_getter_lock async def get_by_mxid(cls, mxid: RoomID) -> Portal | None: try: return cls.by_mxid[mxid] except KeyError: pass portal = cast(cls, await super().get_by_mxid(mxid)) if portal is not None: await portal.postinit() return portal return None @classmethod @async_getter_lock async def get_by_thread_id( cls, thread_id: str, *, receiver: int, is_group: bool | None = None, other_user_pk: int | None = None, ) -> Portal | None: if is_group and receiver != 0: receiver = 0 try: return cls.by_thread_id[(thread_id, receiver)] except KeyError: pass if is_group is None and receiver != 0: try: return cls.by_thread_id[(thread_id, 0)] except KeyError: pass portal = cast( cls, await super().get_by_thread_id( thread_id, receiver=receiver, rec_must_match=is_group is not None ), ) if portal is not None: await portal.postinit() return portal if is_group is not None: portal = cls(thread_id, receiver, other_user_pk=other_user_pk) await portal.insert() await portal.postinit() return portal return None @classmethod async def get_by_thread(cls, thread: Thread, receiver: int) -> Portal | None: if thread.is_group: receiver = 0 other_user_pk = None else: if len(thread.users) == 0: other_user_pk = receiver else: other_user_pk = thread.users[0].pk return await cls.get_by_thread_id( thread.thread_id, receiver=receiver, is_group=thread.is_group, other_user_pk=other_user_pk, ) # endregion