$ show WhiskeySockets/Baileys
Baileys is an unofficial, production-grade WhatsApp Web client library for Node.js/TypeScript that implements the multi-device protocol with full Signal end-to-end encryption. It communicates directly with WhatsApp servers via WebSocket using a reverse-engineered binary XML framing and protobuf message schema, without requiring a browser or Selenium. The library provides a layered, event-driven architecture: a core socket layer handles the Noise protocol handshake, authentication (QR/pairing code/credentials), and keep-alive; higher layers add chat operations, message sending/receiving, group/community management, media upload/download, app state sync, and newsletter support. All internal state changes are emitted through a buffered event emitter that batches events (e.g., `messages.upsert`, `chats.upsert`, `connection.update`) for efficient consumer processing. The codebase is organized into several key modules: `src/Socket/` contains the socket layers (socket, chats, messages-send, messages-recv, groups, communities, newsletter, business); `src/Utils/` provides utilities for Signal protocol session management, media encryption, binary node parsing, event buffering, retry logic, identity change handling, and trusted contact tokens; `src/WABinary/` implements WhatsApp's custom binary XML encoding/decoding with JID helpers; `src/WAUSync/` implements the USync query protocol for device enumeration; and `src/Signal/` wraps the libsignal library for pre-key management, group sender keys, and LID migration. The library depends on external packages such as `libsignal`, `ws`, `protobufjs`, `@hapi/boom`, `pino`, `lru-cache`, `async-mutex`, and optional media processing libraries (`sharp`, `jimp`, `link-preview-js`). Baileys is designed for integration into larger applications (bots, automation, analytics) and exposes a comprehensive API for sending/receiving messages, managing groups, syncing chat history, handling media, and reacting to real-time events. It includes extensive unit and end-to-end tests, a protobuf extraction tool (`proto-extract/`) that scrapes WhatsApp Web's JavaScript to regenerate the `.proto` schema, and a version update script that fetches the latest WhatsApp Web version. The library is actively maintained with frequent commits, CI/CD pipelines, and a clear migration guide for breaking changes.
The LLM identified 5 files worth highlighting for reuse:
Implements a generic event buffering system that batches multiple events (e.g., `chats.upsert`, `messages.upsert`, `contacts.upsert`) into a single emission after a configurable timeout or flush call. Uses a history cache to deduplicate and merge updates, and provides a `buffer()`/`flush()` API that can be extracted and reused in any event-driven system requiring coalesced updates.
Implements the Noise protocol handshake and encrypted framing for WhatsApp Web's transport layer. Manages handshake state (prologue, client hello, server hello, finish), encrypts/decrypts frames with counter-based nonces, and uses a mutex to serialize concurrent writes. The pattern is self-contained and could be adapted for any Noise-based protocol implementation.
Provides a retry manager for undecryptable messages with LRU caches for retry counters and session recreation. Detects MAC errors, schedules phone requests, and manages retry limits with exponential backoff. The component is isolated and demonstrates a robust pattern for handling transient decryption failures in E2EE systems.
A collection of pure functions for parsing, normalizing, and comparing WhatsApp JIDs (user identifiers). Handles device suffixes, LID/PN distinction, and domain extraction. Uses caching with WeakMap for performance. This module is fully self-contained and could be reused in any project that needs to manipulate WhatsApp-style identifiers.
Implements the companion registration client for QR and pairing code flows. Generates ephemeral key pairs, computes shared secrets, and constructs encrypted registration payloads. The module is self-contained with clear cryptographic primitives and could be extracted for any device pairing protocol requiring Noise-based key agreement.