From 748366ad6a9dc4b2a269d5499ae1d5d7e8526762 Mon Sep 17 00:00:00 2001
From: Bohdan Horbeshko <bodqhrohro@gmail.com>
Date: Sat, 22 Jul 2023 10:46:35 -0400
Subject: [PATCH] Avoid webpage preview updates being sent as message edits (by
 hash matching)

---
 telegram/client.go   |  7 ++++++
 telegram/handlers.go |  6 ++++-
 telegram/utils.go    | 57 ++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/telegram/client.go b/telegram/client.go
index 61d46aa..5b0217a 100644
--- a/telegram/client.go
+++ b/telegram/client.go
@@ -2,6 +2,7 @@ package telegram
 
 import (
 	"github.com/pkg/errors"
+	"hash/maphash"
 	"path/filepath"
 	"strconv"
 	"sync"
@@ -60,6 +61,9 @@ type Client struct {
 	DelayedStatuses     map[int64]*DelayedStatus
 	DelayedStatusesLock sync.Mutex
 
+	lastMsgHashes map[int64]uint64
+	msgHashSeed   maphash.Seed
+
 	locks           clientLocks
 	SendMessageLock sync.Mutex
 }
@@ -69,6 +73,7 @@ type clientLocks struct {
 	chatMessageLocks   map[int64]*sync.Mutex
 	resourcesLock      sync.Mutex
 	outboxLock         sync.Mutex
+	lastMsgHashesLock  sync.Mutex
 }
 
 // NewClient instantiates a Telegram App
@@ -129,6 +134,8 @@ func NewClient(conf config.TelegramConfig, jid string, component *xmpp.Component
 		cache:           cache.NewCache(),
 		options:         options,
 		DelayedStatuses: make(map[int64]*DelayedStatus),
+		lastMsgHashes:   make(map[int64]uint64),
+		msgHashSeed:     maphash.MakeSeed(),
 		locks: clientLocks{
 			chatMessageLocks: make(map[int64]*sync.Mutex),
 		},
diff --git a/telegram/handlers.go b/telegram/handlers.go
index abc1f5d..cc6e635 100644
--- a/telegram/handlers.go
+++ b/telegram/handlers.go
@@ -226,6 +226,8 @@ func (c *Client) updateNewMessage(update *client.UpdateNewMessage) {
 		}).Warn("New message from chat")
 
 		c.ProcessIncomingMessage(chatId, update.Message, ignoredResource)
+
+		c.updateLastMessageHash(update.Message.ChatId, update.Message.Id, update.Message.Content)
 	}()
 }
 
@@ -233,6 +235,8 @@ func (c *Client) updateNewMessage(update *client.UpdateNewMessage) {
 func (c *Client) updateMessageContent(update *client.UpdateMessageContent) {
 	markupFunction := formatter.EntityToXEP0393
 
+	defer c.updateLastMessageHash(update.ChatId, update.MessageId, update.NewContent)
+
 	c.SendMessageLock.Lock()
 	c.SendMessageLock.Unlock()
 	xmppId, err := gateway.IdsDB.GetByTgIds(c.Session.Login, c.jid, update.ChatId, update.MessageId)
@@ -250,7 +254,7 @@ func (c *Client) updateMessageContent(update *client.UpdateMessageContent) {
 		return
 	}
 
-	if update.NewContent.MessageContentType() == client.TypeMessageText {
+	if update.NewContent.MessageContentType() == client.TypeMessageText && c.hasLastMessageHashChanged(update.ChatId, update.MessageId, update.NewContent) {
 		textContent := update.NewContent.(*client.MessageText)
 		var editChar string
 		if c.Session.AsciiArrows {
diff --git a/telegram/utils.go b/telegram/utils.go
index 9486349..4835696 100644
--- a/telegram/utils.go
+++ b/telegram/utils.go
@@ -2,8 +2,10 @@ package telegram
 
 import (
 	"crypto/sha1"
+	"encoding/binary"
 	"fmt"
 	"github.com/pkg/errors"
+	"hash/maphash"
 	"io"
 	"io/ioutil"
 	"net/http"
@@ -1384,3 +1386,58 @@ func (c *Client) getCarbonFullJids(isOutgoing bool, ignoredResource string) []st
 	}
 	return jids
 }
+
+func (c *Client) calculateMessageHash(messageId int64, content client.MessageContent) uint64 {
+	var h maphash.Hash
+	h.SetSeed(c.msgHashSeed)
+
+	buf8 := make([]byte, 8)
+	binary.BigEndian.PutUint64(buf8, uint64(messageId))
+	h.Write(buf8)
+
+	if content != nil && content.MessageContentType() == client.TypeMessageText {
+		textContent, ok := content.(*client.MessageText)
+		if !ok {
+			uhOh()
+		}
+
+		if textContent.Text != nil {
+			h.WriteString(textContent.Text.Text)
+			for _, entity := range textContent.Text.Entities {
+				buf4 := make([]byte, 4)
+				binary.BigEndian.PutUint32(buf4, uint32(entity.Offset))
+				h.Write(buf4)
+				binary.BigEndian.PutUint32(buf4, uint32(entity.Length))
+				h.Write(buf4)
+				h.WriteString(entity.Type.TextEntityTypeType())
+			}
+		}
+	}
+
+	return h.Sum64()
+}
+
+func (c *Client) updateLastMessageHash(chatId, messageId int64, content client.MessageContent) {
+	c.locks.lastMsgHashesLock.Lock()
+	defer c.locks.lastMsgHashesLock.Unlock()
+
+	c.lastMsgHashes[chatId] = c.calculateMessageHash(messageId, content)
+}
+
+func (c *Client) hasLastMessageHashChanged(chatId, messageId int64, content client.MessageContent) bool {
+	c.locks.lastMsgHashesLock.Lock()
+	defer c.locks.lastMsgHashesLock.Unlock()
+
+	oldHash, ok := c.lastMsgHashes[chatId]
+	newHash := c.calculateMessageHash(messageId, content)
+
+	if !ok {
+		log.Warnf("Last message hash for chat %v does not exist", chatId)
+	}
+	log.WithFields(log.Fields{
+		"old hash": oldHash,
+		"new hash": newHash,
+	}).Info("Message hashes")
+
+	return !ok || oldHash != newHash
+}