From 7e25e2c2da8dbcb1223ef812e2b95e3971c01fe0 Mon Sep 17 00:00:00 2001
From: Tulir Asokan <tulir@maunium.net>
Date: Mon, 16 Oct 2023 13:07:06 +0300
Subject: [PATCH] Fix legacy backfill not checking if conversations have
 messages

Fixes #631
---
 CHANGELOG.md            |  1 +
 database/historysync.go | 13 +++++++++++++
 historysync.go          | 19 ++++++++++---------
 3 files changed, 24 insertions(+), 9 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8bc2951..de5616f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -3,6 +3,7 @@
 * Added basic support for channels.
 * Added default mime type for outgoing attachments when the origin Matrix
   client forgets to specify the mime type.
+* Fixed legacy backfill creating portals for chats without messages.
 * Updated libwebp version used for encoding.
 
 # v0.10.2 (security update)
diff --git a/database/historysync.go b/database/historysync.go
index 2f88139..d896bbc 100644
--- a/database/historysync.go
+++ b/database/historysync.go
@@ -322,6 +322,19 @@ func (hsq *HistorySyncQuery) DeleteAllMessagesForPortal(userID id.UserID, portal
 	}
 }
 
+func (hsq *HistorySyncQuery) ConversationHasMessages(userID id.UserID, portalKey PortalKey) (exists bool) {
+	err := hsq.db.QueryRow(`
+		SELECT EXISTS(
+		    SELECT 1 FROM history_sync_message
+			WHERE user_mxid=$1 AND conversation_id=$2
+		)
+	`, userID, portalKey.JID).Scan(&exists)
+	if err != nil {
+		hsq.log.Warnfln("Failed to check if any messages are stored for %s/%s: %v", userID, portalKey.JID, err)
+	}
+	return
+}
+
 func (hsq *HistorySyncQuery) DeleteConversation(userID id.UserID, jid string) {
 	// This will also clear history_sync_message as there's a foreign key constraint
 	_, err := hsq.db.Exec(`
diff --git a/historysync.go b/historysync.go
index 578629d..bd16077 100644
--- a/historysync.go
+++ b/historysync.go
@@ -20,6 +20,7 @@ import (
 	"crypto/sha256"
 	"encoding/base64"
 	"fmt"
+	"strings"
 	"time"
 
 	"github.com/rs/zerolog"
@@ -138,8 +139,9 @@ func (user *User) backfillAll() {
 			Int("conversation_count", len(conversations)).
 			Msg("Probably received all history sync blobs, now backfilling conversations")
 		limit := user.bridge.Config.Bridge.HistorySync.MaxInitialConversations
+		bridgedCount := 0
 		// Find the portals for all the conversations.
-		for i, conv := range conversations {
+		for _, conv := range conversations {
 			jid, err := types.ParseJID(conv.ConversationID)
 			if err != nil {
 				user.zlog.Warn().Err(err).
@@ -153,7 +155,12 @@ func (user *User) backfillAll() {
 					Str("portal_jid", portal.Key.JID.String()).
 					Msg("Chat already has a room, deleting messages from database")
 				user.bridge.DB.HistorySync.DeleteConversation(user.MXID, portal.Key.JID.String())
-			} else if limit < 0 || i < limit {
+				bridgedCount++
+			} else if !user.bridge.DB.HistorySync.ConversationHasMessages(user.MXID, portal.Key) {
+				user.zlog.Debug().Str("portal_jid", portal.Key.JID.String()).Msg("Skipping chat with no messages in history sync")
+				user.bridge.DB.HistorySync.DeleteConversation(user.MXID, portal.Key.JID.String())
+			} else if limit < 0 || bridgedCount < limit {
+				bridgedCount++
 				err = portal.CreateMatrixRoom(user, nil, nil, true, true)
 				if err != nil {
 					user.zlog.Err(err).Msg("Failed to create Matrix room for backfill")
@@ -491,13 +498,7 @@ func (user *User) storeHistorySync(evt *waProto.HistorySync) {
 			}
 
 			msgType := getMessageType(msgEvt.Message)
-			if msgType == "unknown" || msgType == "ignore" || msgType == "unknown_protocol" {
-				unsupportedTypes++
-				continue
-			}
-
-			// Don't store unsupported messages.
-			if !containsSupportedMessage(msgEvt.Message) {
+			if msgType == "unknown" || msgType == "ignore" || strings.HasPrefix(msgType, "unknown_protocol_") || !containsSupportedMessage(msgEvt.Message) {
 				unsupportedTypes++
 				continue
 			}