Ensure long message trimming is unicode-aware

Co-authored-by: Scott Nonnenberg <scott@signal.org>

ts/jobs/helpers/sendNormalMessage.ts

@@ -31,7 +31,6 @@ import type {
   UploadedAttachmentType,
 } from '../../types/Attachment';
 import { copyCdnFields } from '../../util/attachments';
-import { LONG_ATTACHMENT_LIMIT } from '../../types/Message';
 import type { RawBodyRange } from '../../types/BodyRange';
 import type { EmbeddedContactWithUploadedAvatar } from '../../types/EmbeddedContact';
 import type { StoryContextType } from '../../types/Util';
@@ -55,6 +54,7 @@ import {
   getChangesForPropAtTimestamp,
 } from '../../util/editHelpers';
 import { getMessageSentTimestamp } from '../../util/getMessageSentTimestamp';
+import { isBodyTooLong, trimBody } from '../../util/longAttachment';
 
 const MAX_CONCURRENT_ATTACHMENT_UPLOADS = 5;
 
@@ -587,8 +587,8 @@ async function getMessageSendData({
     targetTimestamp,
   });
 
-  if (body && body.length > LONG_ATTACHMENT_LIMIT) {
-    body = body.slice(0, LONG_ATTACHMENT_LIMIT);
+  if (body && isBodyTooLong(body)) {
+    body = trimBody(body);
   }
 
   const uploadQueue = new PQueue({

ts/services/backups/export.ts

@@ -32,7 +32,6 @@ import {
   type ServiceIdString,
 } from '../../types/ServiceId';
 import type { RawBodyRange } from '../../types/BodyRange';
-import { LONG_ATTACHMENT_LIMIT } from '../../types/Message';
 import { PaymentEventKind } from '../../types/Payment';
 import { MessageRequestResponseEvent } from '../../types/MessageRequestResponseEvent';
 import type {
@@ -139,6 +138,7 @@ import { toAdminKeyBytes } from '../../util/callLinks';
 import { getRoomIdFromRootKey } from '../../util/callLinksRingrtc';
 import { SeenStatus } from '../../MessageSeenStatus';
 import { migrateAllMessages } from '../../messages/migrateMessageData';
+import { trimBody } from '../../util/longAttachment';
 
 const MAX_CONCURRENCY = 10;
 
@@ -2442,7 +2442,7 @@ export class BackupExportStream extends Readable {
       text:
         message.body != null
           ? {
-              body: message.body?.slice(0, LONG_ATTACHMENT_LIMIT),
+              body: message.body ? trimBody(message.body) : undefined,
               bodyRanges: message.bodyRanges?.map(range =>
                 this.toBodyRange(range)
               ),

ts/textsecure/MessageReceiver.ts

@@ -154,6 +154,7 @@ import {
 import { checkOurPniIdentityKey } from '../util/checkOurPniIdentityKey';
 import { CallLinkUpdateSyncType } from '../types/CallLink';
 import { bytesToUuid } from '../util/uuidToBytes';
+import { isBodyTooLong } from '../util/longAttachment';
 
 const GROUPV2_ID_LENGTH = 32;
 const RETRY_TIMEOUT = 2 * 60 * 1000;
@@ -2157,7 +2158,8 @@ export default class MessageReceiver
     envelope: ProcessedEnvelope,
     sentContainer: ProcessedSent
   ) {
-    log.info('MessageReceiver.handleSentMessage', getEnvelopeId(envelope));
+    const logId = `MessageReceiver.handleSentMessage/${getEnvelopeId(envelope)}`;
+    log.info(logId);
 
     logUnexpectedUrgentValue(envelope, 'sentSync');
 
@@ -2172,7 +2174,7 @@ export default class MessageReceiver
     } = sentContainer;
 
     if (!msg) {
-      throw new Error('MessageReceiver.handleSentMessage: message was falsey!');
+      throw new Error(`${logId}: message was falsey!`);
     }
 
     // TODO: DESKTOP-5804
@@ -2180,14 +2182,18 @@ export default class MessageReceiver
       if (destinationServiceId) {
         await this.handleEndSession(envelope, destinationServiceId);
       } else {
-        throw new Error(
-          'MessageReceiver.handleSentMessage: Cannot end session with falsey destination'
-        );
+        throw new Error(`${logId}: Cannot end session with falsey destination`);
       }
     }
 
     const message = this.processDecrypted(envelope, msg);
 
+    if (message.body && isBodyTooLong(message.body)) {
+      const length = Buffer.byteLength(message.body);
+      this.removeFromCache(envelope);
+      log.warn(`${logId}: Dropping too-long message. Length: ${length}`);
+    }
+
     const ev = new SentEvent(
       {
         envelopeId: envelope.id,
@@ -2487,12 +2493,12 @@ export default class MessageReceiver
     envelope: UnsealedEnvelope,
     msg: Proto.IDataMessage
   ): Promise<void> {
-    const logId = getEnvelopeId(envelope);
-    log.info('MessageReceiver.handleDataMessage', logId);
+    const logId = `MessageReceiver.handleDataMessage/${getEnvelopeId(envelope)}`;
+    log.info(logId);
 
     if (getStoriesBlocked() && msg.storyContext) {
       log.info(
-        `MessageReceiver.handleDataMessage/${logId}: Dropping incoming dataMessage with storyContext field`
+        `${logId}: Dropping incoming dataMessage with storyContext field`
       );
       this.removeFromCache(envelope);
       return;
@@ -2501,15 +2507,10 @@ export default class MessageReceiver
     let p: Promise<void> = Promise.resolve();
     const { sourceServiceId: sourceAci } = envelope;
     if (!sourceAci) {
-      throw new Error(
-        'MessageReceiver.handleDataMessage: sourceAci was falsey'
-      );
+      throw new Error(`${logId}: sourceAci was falsey`);
     }
 
-    strictAssert(
-      isAciString(sourceAci),
-      'MessageReceiver.handleDataMessage: received message from PNI'
-    );
+    strictAssert(isAciString(sourceAci), `${logId}: received message from PNI`);
 
     if (this.isInvalidGroupData(msg, envelope)) {
       this.removeFromCache(envelope);
@@ -2542,10 +2543,10 @@ export default class MessageReceiver
       );
 
       if (isProfileKeyUpdate) {
-        return this.dispatchAndWait(logId, ev);
+        return this.dispatchAndWait(getEnvelopeId(envelope), ev);
       }
 
-      drop(this.dispatchAndWait(logId, ev));
+      drop(this.dispatchAndWait(getEnvelopeId(envelope), ev));
     }
     await p;
 
@@ -2573,13 +2574,17 @@ export default class MessageReceiver
     const isBlocked = groupId ? this.isGroupBlocked(groupId) : false;
 
     if (groupId && isBlocked) {
-      log.warn(
-        `Message ${getEnvelopeId(envelope)} ignored; destined for blocked group`
-      );
+      log.warn(`${logId}: message ignored; destined for blocked group`);
       this.removeFromCache(envelope);
       return undefined;
     }
 
+    if (message.body && isBodyTooLong(message.body)) {
+      const length = Buffer.byteLength(message.body);
+      this.removeFromCache(envelope);
+      log.warn(`${logId}: Dropping too-long message. Length: ${length}`);
+    }
+
     const ev = new MessageEvent(
       {
         envelopeId: envelope.id,

ts/types/Message.ts

@@ -6,8 +6,6 @@ import type { AttachmentType } from './Attachment';
 import type { EmbeddedContactType } from './EmbeddedContact';
 import type { IndexableBoolean, IndexablePresence } from './IndexedDB';
 
-export const LONG_ATTACHMENT_LIMIT = 2048;
-
 export function getMentionsRegex(): RegExp {
   return /\uFFFC/g;
 }

ts/types/Message2.ts

@@ -50,8 +50,8 @@ import {
 } from '../util/getLocalAttachmentUrl';
 import { encryptLegacyAttachment } from '../util/encryptLegacyAttachment';
 import { deepClone } from '../util/deepClone';
-import { LONG_ATTACHMENT_LIMIT } from './Message';
 import * as Bytes from '../Bytes';
+import { isBodyTooLong } from '../util/longAttachment';
 
 export const GROUP = 'group';
 export const PRIVATE = 'private';
@@ -1183,16 +1183,15 @@ export async function migrateBodyAttachmentToDisk(
     return message;
   }
 
-  if (!message.body || (message.body?.length ?? 0) <= LONG_ATTACHMENT_LIMIT) {
+  if (!message.body || !isBodyTooLong(message.body)) {
     return message;
   }
 
   logger.info(`${logId}: Writing bodyAttachment to disk`);
 
-  const data = Bytes.fromString(message.body);
   const bodyAttachment = {
     contentType: LONG_MESSAGE,
-    ...(await writeNewAttachmentData(data)),
+    ...(await writeNewAttachmentData(Bytes.fromString(message.body))),
   };
 
   return {

ts/util/longAttachment.ts

@@ -0,0 +1,21 @@
+// Copyright 2023 Signal Messenger, LLC
+// SPDX-License-Identifier: AGPL-3.0-only
+
+import { unicodeSlice } from './unicodeSlice';
+
+const LONG_ATTACHMENT_LIMIT = 2048;
+
+export function isBodyTooLong(body: string): boolean {
+  return Buffer.byteLength(body) > LONG_ATTACHMENT_LIMIT;
+}
+
+export function trimBody(body: string, length = LONG_ATTACHMENT_LIMIT): string {
+  const sliced = unicodeSlice(body, 0, length);
+
+  if (sliced.length > 0) {
+    return sliced;
+  }
+
+  // Failover, for degenerate cases
+  return '\uFFFE';
+}