feat: implement Meta AI msmsg decryption#2592
Conversation
Port msmsg (MessageSecretMessage) decryption: - Add src/Utils/meta-ai-msmsg.ts with HKDF+AES-GCM decryption logic - Track outgoing bot message secrets (setBotMessageSecret) in decode-wa-message.ts - Handle msmsg enc type in decryptMessageNode with multi-candidate secret fallback - Auto-decode richResponseMessage.text from submessages or unifiedResponse JSON - Remove the msmsg early-return bail-out in messages-recv.ts
|
Thanks for opening this pull request and contributing to the project! The next step is for the maintainers to review your changes. If everything looks good, it will be approved and merged into the main branch. In the meantime, anyone in the community is encouraged to test this pull request and provide feedback. ✅ How to confirm it worksIf you’ve tested this PR, please comment below with: This helps us speed up the review and merge process. 📦 To test this PR locally:If you encounter any issues or have feedback, feel free to comment as well. |
📝 WalkthroughWalkthroughThe PR adds support for decrypting Meta AI MSMSG bot messages by introducing a cryptographic module for bounded strategy-based AES-GCM decryption, integrating secret caching and MSMSG handling into the main message decoder, and removing special-case rejection so MSMSG flows through normal processing with retry. ChangesMSMSG bot message decryption support
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://github.com/_gh/avatars/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (1)
src/Utils/decode-wa-message.ts (1)
416-416: ⚡ Quick winReplace manual LID parsing with JID helper
meLid.split(':')[0]is a manual JID parse. Please use a JID helper (same reason as in the crypto helper file) to avoid malformed edge cases.As per coding guidelines, "Always use JID helper functions (
jidDecode,jidNormalizedUser,areJidsSameUser,isJidUser) instead of string operations like.split('@')or===comparisons."🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@src/Utils/decode-wa-message.ts` at line 416, Replace the manual parse meLid.split(':')[0] with the JID helper: import and call jidDecode(meLid) and use its user/local part (e.g., const { user } = jidDecode(meLid)) when building meId (so meId becomes metaTargetSenderJid || `${user}`@lid``), avoiding string-splitting and ensuring proper JID parsing; update any imports to pull jidDecode from the JID helpers module.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/Utils/decode-wa-message.ts`:
- Line 436: The catch blocks in decode-wa-message decryption paths use `any` and
unchecked casts which weaken type-safety: change `catch (e: any)` to `catch (e:
unknown)` and narrow `e` before accessing properties like `attemptedStrategies`
(use type guards or instanceof checks and rethrow if unknown), replace `as any`
enrichments for `richResponse` with a local structural type (e.g., `{ text?:
string }`) and cast/narrow the unknown value into that shape before mutating,
and likewise convert other `catch (err: any)` and the
`isSessionRecordError(error: any)` usage to accept `unknown` and perform
explicit narrowing inside `isSessionRecordError` (or add a type guard function)
so no `any` is introduced while preserving the same behavior in the decryption
and augmentation code paths.
- Around line 25-51: botMessageSecrets currently never evicts entries leading to
unbounded memory growth; modify setBotMessageSecret to enforce a fixed-cap
eviction policy (e.g., MAX_SECRETS_GLOBAL constant) and remove oldest entries
when the map exceeds that cap. Locate botMessageSecrets and the
setBotMessageSecret function and after inserting the new buf, check
botMessageSecrets.size and delete the oldest key(s) (use
botMessageSecrets.keys().next().value to find the oldest) until size <=
MAX_SECRETS_GLOBAL; keep pushRecentChatSecret and botRecentSecretsByChat logic
unchanged.
In `@src/Utils/meta-ai-msmsg.ts`:
- Around line 56-59: normalizeLidJid currently does manual string splitting;
replace that with JID helper usage: use jidDecode(jid) to parse the JID, check
the decoded server/domain equals 'lid' and that decoded user contains a ':'; if
so, take the portion before ':' from decoded.user and reassemble a normalized
JID using the jidNormalizedUser (or equivalent JID assembly helper) to return
e.g. "<userBeforeColon>`@lid`"; otherwise return the original jid or undefined as
before. Ensure you reference and import the helper(s) (jidDecode,
jidNormalizedUser) and update normalizeLidJid to use them instead of jid.split.
- Line 275: In decodeRichResponseMessage: remove the unnecessary any cast by
using richMsg.unifiedResponse?.data (typed as
IAIRichResponseUnifiedResponse.data | null) instead of (richMsg as
any).unifiedResponse?.data, and change the Buffer call to
Buffer.from(data).toString('utf8') (or, if data can be a base64 string, add a
small guard to detect string vs Uint8Array and decode accordingly) so that the
variable data and its decoding respect the declared types.
---
Nitpick comments:
In `@src/Utils/decode-wa-message.ts`:
- Line 416: Replace the manual parse meLid.split(':')[0] with the JID helper:
import and call jidDecode(meLid) and use its user/local part (e.g., const { user
} = jidDecode(meLid)) when building meId (so meId becomes metaTargetSenderJid ||
`${user}`@lid``), avoiding string-splitting and ensuring proper JID parsing;
update any imports to pull jidDecode from the JID helpers module.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: 2b99b022-7ca6-436f-a8cf-92cf191855b1
📒 Files selected for processing (3)
src/Socket/messages-recv.tssrc/Utils/decode-wa-message.tssrc/Utils/meta-ai-msmsg.ts
| const botMessageSecrets = new Map<string, Buffer>() | ||
| const botRecentSecretsByChat = new Map<string, { id: string; secret: Buffer }[]>() | ||
|
|
||
| const pushRecentChatSecret = (chatJid: string, id: string, secretBuf: Buffer): void => { | ||
| if (!chatJid || !secretBuf) return | ||
| const existing = botRecentSecretsByChat.get(chatJid) || [] | ||
| const filtered = existing.filter(item => item.id !== id && !item.secret.equals(secretBuf)) | ||
| filtered.unshift({ id, secret: secretBuf }) | ||
| if (filtered.length > MAX_SECRETS_PER_CHAT) filtered.length = MAX_SECRETS_PER_CHAT | ||
| botRecentSecretsByChat.set(chatJid, filtered) | ||
| } | ||
|
|
||
| export const setBotMessageSecret = (id: string, secret: Uint8Array | Buffer | string, chatJid?: string): void => { | ||
| if (!id || !secret) return | ||
| let buf: Buffer | ||
| if (Buffer.isBuffer(secret)) { | ||
| buf = secret | ||
| } else if (secret instanceof Uint8Array) { | ||
| buf = Buffer.from(secret.buffer, secret.byteOffset, secret.byteLength) | ||
| } else if (typeof secret === 'string') { | ||
| buf = Buffer.from(secret, 'base64') | ||
| } else { | ||
| return | ||
| } | ||
| botMessageSecrets.set(id, buf) | ||
| if (chatJid) pushRecentChatSecret(chatJid, id, buf) | ||
| } |
There was a problem hiding this comment.
Bound botMessageSecrets to prevent unbounded growth
botMessageSecrets never evicts entries. In long-lived sessions this can grow indefinitely and degrade memory usage over time.
Suggested change
const MAX_SECRETS_PER_CHAT = 20
+const MAX_SECRETS_GLOBAL = 2000
const botMessageSecrets = new Map<string, Buffer>()
+const botMessageSecretOrder: string[] = []
const botRecentSecretsByChat = new Map<string, { id: string; secret: Buffer }[]>()
@@
botMessageSecrets.set(id, buf)
+ botMessageSecretOrder.push(id)
+ if (botMessageSecretOrder.length > MAX_SECRETS_GLOBAL) {
+ const evictId = botMessageSecretOrder.shift()
+ if (evictId) botMessageSecrets.delete(evictId)
+ }
if (chatJid) pushRecentChatSecret(chatJid, id, buf)
}🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/Utils/decode-wa-message.ts` around lines 25 - 51, botMessageSecrets
currently never evicts entries leading to unbounded memory growth; modify
setBotMessageSecret to enforce a fixed-cap eviction policy (e.g.,
MAX_SECRETS_GLOBAL constant) and remove oldest entries when the map exceeds that
cap. Locate botMessageSecrets and the setBotMessageSecret function and after
inserting the new buf, check botMessageSecrets.size and delete the oldest key(s)
(use botMessageSecrets.keys().next().value to find the oldest) until size <=
MAX_SECRETS_GLOBAL; keep pushRecentChatSecret and botRecentSecretsByChat logic
unchanged.
| msgBuffer = await decryptMsmsgBotMessage(candidate.secret, helperKey, msMsg) | ||
| logger.debug({ source: candidate.source }, 'msmsg: decrypted with candidate secret') | ||
| break | ||
| } catch (e: any) { |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Verify explicit any usage in this file
rg -nP "\bas\s+any\b|:\s*any\b" src/Utils/decode-wa-message.tsRepository: WhiskeySockets/Baileys
Length of output: 365
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '400,560p' src/Utils/decode-wa-message.ts | nl -ba | sed -n '1,200p'Repository: WhiskeySockets/Baileys
Length of output: 109
🏁 Script executed:
#!/bin/bash
set -euo pipefail
awk 'NR>=420 && NR<=540 {printf "%d:%s\n", NR, $0}' src/Utils/decode-wa-message.ts
echo "----"
awk 'NR>=540 && NR<=575 {printf "%d:%s\n", NR, $0}' src/Utils/decode-wa-message.tsRepository: WhiskeySockets/Baileys
Length of output: 5196
Remove new any usage in msmsg/rich-response decryption paths
- Replace
catch (e: any)(line 436) withcatch (e: unknown)and narrow before readingattemptedStrategies/ rethrowing. - Replace
as anyrichResponse augmentation (lines 493 and 498) with a local structural type (e.g.,{ text?: string }) and cast/narrow fromunknowninstead of weakening types. - If this block was added in the same change, also avoid
catch (err: any)(line 512) /isSessionRecordError(error: any)(line 542) by switching tounknown+ narrowing.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/Utils/decode-wa-message.ts` at line 436, The catch blocks in
decode-wa-message decryption paths use `any` and unchecked casts which weaken
type-safety: change `catch (e: any)` to `catch (e: unknown)` and narrow `e`
before accessing properties like `attemptedStrategies` (use type guards or
instanceof checks and rethrow if unknown), replace `as any` enrichments for
`richResponse` with a local structural type (e.g., `{ text?: string }`) and
cast/narrow the unknown value into that shape before mutating, and likewise
convert other `catch (err: any)` and the `isSessionRecordError(error: any)`
usage to accept `unknown` and perform explicit narrowing inside
`isSessionRecordError` (or add a type guard function) so no `any` is introduced
while preserving the same behavior in the decryption and augmentation code
paths.
| const normalizeLidJid = (jid: string | undefined): string | undefined => { | ||
| if (!jid || !jid.endsWith('@lid') || !jid.includes(':')) return jid | ||
| return `${jid.split(':')[0]}@lid` | ||
| } |
There was a problem hiding this comment.
Use JID helpers instead of manual string splitting
normalizeLidJid currently uses jid.split(':')[0]. Please switch to JID helper parsing so non-standard forms don’t produce wrong normalization.
Suggested change
+import { jidDecode } from '../WABinary'
+
const normalizeLidJid = (jid: string | undefined): string | undefined => {
- if (!jid || !jid.endsWith('`@lid`') || !jid.includes(':')) return jid
- return `${jid.split(':')[0]}`@lid``
+ if (!jid) return jid
+ const decoded = jidDecode(jid)
+ if (!decoded || decoded.server !== 'lid') return jid
+ return `${decoded.user}`@lid``
}As per coding guidelines, "Always use JID helper functions (jidDecode, jidNormalizedUser, areJidsSameUser, isJidUser) instead of string operations like .split('@') or === comparisons."
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| const normalizeLidJid = (jid: string | undefined): string | undefined => { | |
| if (!jid || !jid.endsWith('@lid') || !jid.includes(':')) return jid | |
| return `${jid.split(':')[0]}@lid` | |
| } | |
| import { jidDecode } from '../WABinary' | |
| const normalizeLidJid = (jid: string | undefined): string | undefined => { | |
| if (!jid) return jid | |
| const decoded = jidDecode(jid) | |
| if (!decoded || decoded.server !== 'lid') return jid | |
| return `${decoded.user}`@lid`` | |
| } |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/Utils/meta-ai-msmsg.ts` around lines 56 - 59, normalizeLidJid currently
does manual string splitting; replace that with JID helper usage: use
jidDecode(jid) to parse the JID, check the decoded server/domain equals 'lid'
and that decoded user contains a ':'; if so, take the portion before ':' from
decoded.user and reassemble a normalized JID using the jidNormalizedUser (or
equivalent JID assembly helper) to return e.g. "<userBeforeColon>`@lid`";
otherwise return the original jid or undefined as before. Ensure you reference
and import the helper(s) (jidDecode, jidNormalizedUser) and update
normalizeLidJid to use them instead of jid.split.
| .join('\n') | ||
| if (sub) return sub | ||
| } | ||
| const data = (richMsg as any).unifiedResponse?.data |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Verify remaining explicit `any` usage in this file
rg -nP "\bas\s+any\b|:\s*any\b" src/Utils/meta-ai-msmsg.tsRepository: WhiskeySockets/Baileys
Length of output: 124
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Show the exact context around the reported line
nl -ba src/Utils/meta-ai-msmsg.ts | sed -n '240,310p'
# Find where `richMsg` is declared/typed in this file
rg -n "richMsg" src/Utils/meta-ai-msmsg.tsRepository: WhiskeySockets/Baileys
Length of output: 109
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Context around the reported line (use cat -n since nl is unavailable)
cat -n src/Utils/meta-ai-msmsg.ts | sed -n '240,310p'
# Locate the type/definition of `richMsg` usages in this file
rg -n "richMsg" src/Utils/meta-ai-msmsg.tsRepository: WhiskeySockets/Baileys
Length of output: 2263
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Find how IAIRichResponseMessage is typed and whether unifiedResponse is part of it
rg -n "interface IAIRichResponseMessage|type IAIRichResponseMessage|IAIRichResponseMessage" src | head -n 50
rg -n "unifiedResponse" src | head -n 50Repository: WhiskeySockets/Baileys
Length of output: 295
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Locate the definition of IAIRichResponseMessage in the repo
rg -n "IAIRichResponseMessage" . -S --hidden
# Also locate where proto is imported from in this file
rg -n "from ['\"][^'\"]+['\"]" src/Utils/meta-ai-msmsg.tsRepository: WhiskeySockets/Baileys
Length of output: 1043
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect IAIRichResponseMessage definition around the reported lines
sed -n '520,580p' WAProto/index.d.tsRepository: WhiskeySockets/Baileys
Length of output: 3703
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect IAIRichResponseUnifiedResponse to confirm the `data` field's type
rg -n "interface IAIRichResponseUnifiedResponse" -n WAProto/index.d.ts
sed -n '580,660p' WAProto/index.d.tsRepository: WhiskeySockets/Baileys
Length of output: 4924
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Find other references to unifiedResponse/data decoding
rg -n "unifiedResponse\\?\\.data|IAIRichResponseUnifiedResponse|unifiedResponse\\b" src WAProto -S
# Find how Buffer.from(...,'base64') is used with similar payloads
rg -n "Buffer\\.from\\([^\\)]*,'base64'\\)|Buffer\\.from\\([^\\)]*\"base64\"" src -SRepository: WhiskeySockets/Baileys
Length of output: 2369
🏁 Script executed:
node - <<'NODE'
const u = new Uint8Array([1,2,3,4]);
try {
const b = Buffer.from(u, 'base64');
console.log('buf len', b.length, 'contents', b);
} catch (e) {
console.error('error', e && e.message);
}
NODERepository: WhiskeySockets/Baileys
Length of output: 106
Remove any cast for unifiedResponse.data in decodeRichResponseMessage (src/Utils/meta-ai-msmsg.ts:275)
richMsg is already typed with unifiedResponse?: IAIRichResponseUnifiedResponse | null and IAIRichResponseUnifiedResponse.data?: Uint8Array | null, so replace const data = (richMsg as any).unifiedResponse?.data with const data = richMsg.unifiedResponse?.data, and update Buffer.from(data, 'base64') to Buffer.from(data).toString('utf8') (or add a guard only if data can be a base64 string).
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/Utils/meta-ai-msmsg.ts` at line 275, In decodeRichResponseMessage: remove
the unnecessary any cast by using richMsg.unifiedResponse?.data (typed as
IAIRichResponseUnifiedResponse.data | null) instead of (richMsg as
any).unifiedResponse?.data, and change the Buffer call to
Buffer.from(data).toString('utf8') (or, if data can be a base64 string, add a
small guard to detect string vs Uint8Array and decode accordingly) so that the
variable data and its decoding respect the declared types.
![cubic-dev-ai[bot]](https://github.com/_gh/avatars/in/1082092?s=60&v=4){kind=small}
There was a problem hiding this comment.
1 issue found across 3 files
Prompt for AI agents (unresolved issues)
Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.
<file name="src/Utils/decode-wa-message.ts">
<violation number="1" location="src/Utils/decode-wa-message.ts:25">
P1: Unbounded global message-secret cache risks memory growth and cross-session contamination</violation>
</file>
Reply with feedback, questions, or to request a fix.
Re-trigger cubic
|
|
||
| const MAX_SECRETS_PER_CHAT = 20 | ||
|
|
||
| const botMessageSecrets = new Map<string, Buffer>() |
There was a problem hiding this comment.
P1: Unbounded global message-secret cache risks memory growth and cross-session contamination
Prompt for AI agents
Check if this issue is valid — if so, understand the root cause and fix it. At src/Utils/decode-wa-message.ts, line 25:
<comment>Unbounded global message-secret cache risks memory growth and cross-session contamination</comment>
<file context>
@@ -18,6 +18,37 @@ import {
+
+const MAX_SECRETS_PER_CHAT = 20
+
+const botMessageSecrets = new Map<string, Buffer>()
+const botRecentSecretsByChat = new Map<string, { id: string; secret: Buffer }[]>()
+
</file context>
2 participants
Port msmsg (MessageSecretMessage) decryption:
Summary by cubic
Adds decryption support for Meta AI
msmsgso bot replies and edits are readable and processed end-to-end. Also auto-decodes rich response text and removes the old skip path.src/Utils/meta-ai-msmsg.tswith HKDF+AES-GCM decryption, bounded strategies, and rich response decoders.enctypemsmsgindecryptMessageNodewith multi-candidate secret fallback (by id and recent per-chat).messageSecretviasetBotMessageSecretto decrypt follow-ups and edits reliably.richResponseMessage.textfrom submessages orunifiedResponseJSON; removed themsmsgearly-return inmessages-recv.ts.Written for commit 6394514. Summary will update on new commits. Review in cubic
Summary by CodeRabbit
Bug Fixes
New Features