21 Commits

Author SHA1 Message Date
b6950e11d2 v1.18.0 2026-04-10 17:25:34 +00:00
e4935fbf21 feat(readme): expand documentation for voicemail, IVR, audio engine, and API capabilities 2026-04-10 17:25:34 +00:00
f543ff1568 v1.17.2 2026-04-10 17:14:14 +00:00
c63a759689 fix(proxy-engine): use negotiated SDP payload types when wiring SIP legs and enable default nnnoiseless features for telephony denoising 2026-04-10 17:14:14 +00:00
a02146633b v1.17.1 2026-04-10 16:57:07 +00:00
f78639dd19 fix(proxy-engine,codec-lib,sip-proto,ts): preserve negotiated media details and improve RTP audio handling across call legs 2026-04-10 16:57:07 +00:00
2aca5f1510 v1.17.0 2026-04-10 15:58:41 +00:00
73b28f5f57 feat(proxy-engine): upgrade the internal audio bus to 48kHz f32 with per-leg denoising and improve SIP leg routing 2026-04-10 15:58:41 +00:00
10ad432a4c v1.16.0 2026-04-10 15:21:44 +00:00
66112091a2 feat(proxy-engine): integrate Kokoro TTS generation into proxy-engine and simplify TypeScript prompt handling to use cached WAV files 2026-04-10 15:21:44 +00:00
c9ae747c95 v1.15.0 2026-04-10 15:12:30 +00:00
45f9b9c15c feat(proxy-engine): add device leg, leg transfer, and leg replacement call controls 2026-04-10 15:12:30 +00:00
7d59361352 feat(mixer): enhance mixer functionality with interaction and tool legs
- Updated mixer to handle participant and isolated leg roles, allowing for IVR and consent interactions.
- Introduced commands for starting and canceling interactions, managing tool legs for recording and transcription.
- Implemented per-source audio handling for tool legs, enabling separate audio processing.
- Enhanced DTMF handling to forward events between participant legs only.
- Added support for PCM recording directly from tool legs, with WAV file generation.
- Updated TypeScript definitions and functions to support new interaction and tool leg features.
2026-04-10 14:54:21 +00:00
6a130db7c7 v1.14.0 2026-04-10 12:52:48 +00:00
93f671f1f9 feat(proxy-engine): add multiparty call mixing with dynamic SIP and WebRTC leg management 2026-04-10 12:52:48 +00:00
36eab44e28 v1.13.0 2026-04-10 12:19:20 +00:00
9e5aa35fee feat(proxy-engine,webrtc): add B2BUA SIP leg handling and WebRTC call bridging for outbound calls 2026-04-10 12:19:20 +00:00
82f2742db5 v1.12.0 2026-04-10 11:36:18 +00:00
239e2ac81d feat(proxy-engine): add Rust-based outbound calling, WebRTC bridging, and voicemail handling 2026-04-10 11:36:18 +00:00
ad253f823f v1.11.0 2026-04-10 09:57:27 +00:00
3132ba8cbb feat(rust-proxy-engine): add a Rust SIP proxy engine with shared SIP and codec libraries 2026-04-10 09:57:27 +00:00
75 changed files with 12519 additions and 8093 deletions

View File

@@ -1,5 +1,84 @@
# Changelog
## 2026-04-10 - 1.18.0 - feat(readme)
expand documentation for voicemail, IVR, audio engine, and API capabilities
- Updates the feature overview to document voicemail, IVR menus, call recording, enhanced TTS, and the 48kHz float audio engine
- Refreshes the architecture section to describe the TypeScript control plane, Rust proxy-engine data plane, and JSON-over-stdio IPC
- Clarifies REST API and WebSocket coverage with voicemail endpoints, incoming call events, and refined endpoint descriptions
## 2026-04-10 - 1.17.2 - fix(proxy-engine)
use negotiated SDP payload types when wiring SIP legs and enable default nnnoiseless features for telephony denoising
- Select the negotiated codec payload type from SDP answers instead of always using the first offered codec
- Preserve the device leg's preferred payload type from its own INVITE SDP when attaching it to the mixer
- Enable default nnnoiseless features in codec-lib and proxy-engine dependencies
## 2026-04-10 - 1.17.1 - fix(proxy-engine,codec-lib,sip-proto,ts)
preserve negotiated media details and improve RTP audio handling across call legs
- Use native Opus float encode/decode to avoid unnecessary i16 quantization in the f32 audio path.
- Parse full RTP headers including extensions and sequence numbers, then sort inbound packets before decoding to keep codec state stable for out-of-order audio.
- Capture negotiated codec payload types from SDP offers and answers and include codec, RTP port, remote media, and metadata in leg_added events.
- Emit leg_state_changed and leg_removed events more consistently so the dashboard reflects leg lifecycle updates accurately.
## 2026-04-10 - 1.17.0 - feat(proxy-engine)
upgrade the internal audio bus to 48kHz f32 with per-leg denoising and improve SIP leg routing
- switch mixer, prompt playback, and tool leg audio handling from 16kHz i16 to 48kHz f32 for higher-quality internal processing
- add f32 decode/encode and resampling support plus standalone RNNoise denoiser creation in codec-lib
- apply per-leg inbound noise suppression in the mixer before mix-minus generation
- fix passthrough call routing by matching the actual leg from the signaling source address when Call-IDs are shared
- correct dialed number extraction from bare SIP request URIs by parsing the user part directly
## 2026-04-10 - 1.16.0 - feat(proxy-engine)
integrate Kokoro TTS generation into proxy-engine and simplify TypeScript prompt handling to use cached WAV files
- adds a generate_tts command to proxy-engine with lazy-loaded Kokoro model support and WAV output generation
- removes standalone opus-codec and tts-engine workspace binaries by consolidating TTS generation into proxy-engine
- updates announcement and prompt cache flows to generate and cache WAV files on disk instead of pre-encoding RTP frames in TypeScript
## 2026-04-10 - 1.15.0 - feat(proxy-engine)
add device leg, leg transfer, and leg replacement call controls
- adds proxy-engine commands and call manager support for inviting a registered SIP device into an active call
- supports transferring an existing leg between calls while preserving the active connection and updating mixer routing
- supports replacing a call leg by removing the current leg and dialing a new outbound destination
- wires the frontend add-leg API and TypeScript bridge to the new device leg and leg control commands
## 2026-04-10 - 1.14.0 - feat(proxy-engine)
add multiparty call mixing with dynamic SIP and WebRTC leg management
- replace passthrough call handling with a mixer-backed call model that tracks multiple legs and exposes leg status in call state output
- add mixer and leg I/O infrastructure to bridge SIP RTP and WebRTC audio through channel-based mix-minus processing
- introduce add_leg and remove_leg proxy commands and wire frontend bridge APIs to manage external call legs
- emit leg lifecycle events for observability and mark unimplemented device-leg and transfer HTTP endpoints with 501 responses
## 2026-04-10 - 1.13.0 - feat(proxy-engine,webrtc)
add B2BUA SIP leg handling and WebRTC call bridging for outbound calls
- introduce a new SipLeg module to manage outbound provider dialogs, including INVITE lifecycle, digest auth retries, ACK handling, media endpoint tracking, and termination
- store outbound dashboard calls as B2BUA calls in the call manager and emit provider media details on call_answered for bridge setup
- separate SIP and WebRTC engine locking to avoid contention and deadlocks while linking sessions to call RTP sockets
- add bidirectional RTP bridging between provider SIP media and browser WebRTC audio using the allocated RTP socket
- wire browser webrtc-accept events in the frontend and sipproxy so session-to-call linking can occur when media and acceptance arrive in either order
## 2026-04-10 - 1.12.0 - feat(proxy-engine)
add Rust-based outbound calling, WebRTC bridging, and voicemail handling
- adds outbound call origination through the Rust proxy engine with dashboard make_call support
- routes unanswered inbound calls to voicemail, including greeting playback, beep generation, and WAV message recording
- introduces Rust WebRTC session handling and SIP audio bridging, replacing the previous TypeScript WebRTC path
- moves SIP registration and routing responsibilities further into the Rust proxy engine and removes legacy TypeScript call/SIP modules
## 2026-04-10 - 1.11.0 - feat(rust-proxy-engine)
add a Rust SIP proxy engine with shared SIP and codec libraries
- add new Rust workspace crates for proxy-engine, sip-proto, and codec-lib
- move transcoding logic out of opus-codec into reusable codec-lib and keep opus-codec as a thin CLI wrapper
- implement SIP message parsing, dialog handling, SDP/URI rewriting, provider registration, device registration, call management, RTP relay, and DTMF detection in Rust
- add a TypeScript proxy bridge and update the SIP proxy entrypoint to spawn and configure the Rust engine as the SIP data plane
## 2026-04-10 - 1.10.0 - feat(call, voicemail, ivr)
add voicemail and IVR call flows with DTMF handling, prompt playback, recording, and dashboard management

Binary file not shown.

Binary file not shown.

Binary file not shown.

Binary file not shown.

Binary file not shown.

Binary file not shown.

View File

@@ -1,6 +1,6 @@
{
"name": "siprouter",
"version": "1.10.0",
"version": "1.18.0",
"private": true,
"type": "module",
"scripts": {
@@ -14,7 +14,6 @@
"@design.estate/dees-element": "^2.2.4",
"@push.rocks/smartrust": "^1.3.2",
"@push.rocks/smartstate": "^2.3.0",
"werift": "^0.22.9",
"ws": "^8.20.0"
},
"devDependencies": {

283
readme.md
View File

@@ -1,6 +1,6 @@
# @serve.zone/siprouter
A production-grade **SIP B2BUA + WebRTC bridge** built with TypeScript and Rust. Routes calls between SIP providers, SIP hardware devices, and browser softphones — with real-time codec transcoding, ML noise suppression, neural TTS announcements, and a slick web dashboard.
A production-grade **SIP B2BUA + WebRTC bridge** built with TypeScript and Rust. Routes calls between SIP providers, SIP hardware devices, and browser softphones — with real-time codec transcoding, ML noise suppression, neural TTS, voicemail, IVR menus, and a slick web dashboard.
## Issue Reporting and Security
@@ -12,14 +12,16 @@ For reporting bugs, issues, or security vulnerabilities, please visit [community
siprouter sits between your SIP trunk providers and your endpoints — hardware phones, ATAs, browser softphones — and handles **everything** in between:
- 📞 **SIP B2BUA** — Terminates and re-originates calls with full RFC 3261 dialog state management
- 🌐 **WebRTC Bridge** — Browser-based softphone with bidirectional audio to the SIP network
- 🎛️ **Multi-Provider Trunking** — Register with multiple SIP providers simultaneously (sipgate, easybell, o2, etc.)
- 🔊 **Rust Codec Engine** — Real-time Opus ↔ G.722 ↔ PCMU ↔ PCMA transcoding in native Rust
- 🤖 **ML Noise Suppression** — RNNoise denoiser with per-direction state (to SIP / to browser)
- 🗣️ **Neural TTS** — Kokoro-powered "connecting your call" announcements, pre-encoded for instant playback
- 🔀 **Hub Model Calls** — N-leg calls with dynamic add/remove, transfer, and RTP fan-out
- 🖥 **Web Dashboard** — Real-time SPA with live call monitoring, browser phone, contact management, provider config
- 📞 **SIP B2BUA** — Terminates and re-originates calls with full RFC 3261 dialog state management, digest auth, and SDP negotiation
- 🌐 **WebRTC Bridge** — Browser-based softphone with bidirectional Opus audio to the SIP network
- 🎛️ **Multi-Provider Trunking** — Register with multiple SIP providers simultaneously (sipgate, easybell, etc.) with automatic failover
- 🎧 **48kHz f32 Audio Engine** — High-fidelity internal audio bus at 48kHz/32-bit float with native Opus float encode/decode, FFT-based resampling, and per-leg ML noise suppression
- 🔀 **N-Leg Mix-Minus Mixer** — Conference-grade mixing with dynamic leg add/remove, transfer, and per-source audio separation
- 📧 **Voicemail** — Configurable voicemail boxes with TTS greetings, recording, and web playback
- 🔢 **IVR Menus** — DTMF-navigable interactive voice response with nested menus, routing actions, and custom prompts
- 🗣 **Neural TTS** — Kokoro-powered announcements and greetings with 25+ voice presets, backed by espeak-ng fallback
- 🎙️ **Call Recording** — Per-source separated WAV recording at 48kHz via tool legs
- 🖥️ **Web Dashboard** — Real-time SPA with 9 views: live calls, browser phone, routing, voicemail, IVR, contacts, providers, and streaming logs
---
@@ -35,32 +37,38 @@ siprouter sits between your SIP trunk providers and your endpoints — hardware
┌──────────────────────────────────────┐
│ siprouter │
│ │
┌──────────┐ ┌──────────────────┐
│ Call Hub │ │ Rust Transcoder │
│ │ N legs │──│ Opus/G.722/PCM │ │
│ │ fan-out │ │ + RNNoise │ │
└────┬─────┘ └──────────────────┘
┌────┴─────┐ ┌──────────────────┐
│ SIP Stack│ │ Kokoro TTS │
│ │ Dialog SM│ │ (ONNX Runtime) │ │
└────┬─────┘ └──────────────────┘
┌────┴──────────────────────────┐
│ │ Local Registrar + Provider
│ │ Registration Engine
└───────────────────────────────┘
└──────────┬──────────────┬────────────┘
TypeScript Control Plane
┌────────────────────────────────┐
│ │ Config · WebRTC Signaling │ │
│ │ REST API · Web Dashboard │ │
│ Voicebox Manager · TTS Cache │
└────────────┬───────────────────┘
JSON-over-stdio IPC
┌────────────┴───────────────────┐
│ │ Rust proxy-engine (data plane) │ │
│ │
│ SIP Stack · Dialog SM · Auth
│ Call Manager · N-Leg Mixer │
│ │ 48kHz f32 Bus · RNNoise │ │
│ │ Codec Engine · RTP Port Pool │ │
│ WebRTC Engine · Kokoro TTS │
│ │ Voicemail · IVR · Recording │ │
│ └────┬──────────────────┬────────┘ │
└───────┤──────────────────┤───────────┘
│ │
┌──────┴──────┐ ─────┴──────┐
┌──────┴──────┐ ┌──────┴──────┐
│ SIP Devices │ │ SIP Trunk │
│ (HT801, etc)│ │ Providers │
└─────────────┘ ────────────┘
│ (HT801 etc) │ Providers
└─────────────┘ └─────────────┘
```
### The Hub Model
### 🧠 Key Design Decisions
Every call is a **hub** with N legs. Each leg is either a `SipLeg` (hardware device or provider) or a `WebRtcLeg` (browser). RTP flows through the hub — each leg's received audio is forwarded to all other legs, with codec transcoding handled transparently by the Rust engine.
- **Hub Model** — Every call is a hub with N legs. Each leg is a `SipLeg` (device/provider) or `WebRtcLeg` (browser). Legs can be dynamically added, removed, or transferred without tearing down the call.
- **Rust Data Plane** — All SIP protocol handling, codec transcoding, mixing, and RTP I/O runs in native Rust for real-time performance. TypeScript handles config, signaling, REST API, and dashboard.
- **48kHz f32 Internal Bus** — Audio is processed at maximum quality internally. Encoding/decoding to wire format (G.722, PCMU, Opus) happens solely at the leg boundary.
- **Per-Session Codec Isolation** — Each call leg gets its own encoder/decoder/resampler/denoiser state — no cross-call corruption.
- **SDP Codec Negotiation** — Outbound encoding uses the codec actually negotiated in SDP answers, not just the first offered codec.
---
@@ -70,15 +78,16 @@ Every call is a **hub** with N legs. Each leg is either a `SipLeg` (hardware dev
- **Node.js** ≥ 20 with `tsx` globally available
- **pnpm** for package management
- **Rust** toolchain (for building the codec engine and TTS)
- **Rust** toolchain (for building the proxy engine)
- **espeak-ng** (optional, for TTS fallback)
### Install & Build
```bash
# Clone and install
# Clone and install dependencies
pnpm install
# Build the Rust binaries (opus-codec + tts-engine)
# Build the Rust proxy-engine binary
pnpm run buildRust
# Bundle the web frontend
@@ -87,57 +96,92 @@ pnpm run bundle
### Configuration
Create `.nogit/config.json` with your setup:
Create `.nogit/config.json`:
```jsonc
{
"proxy": {
"lanIp": "192.168.1.100", // Your server's LAN IP
"lanPort": 5070, // SIP signaling port
"rtpPortRange": [20000, 20200],// RTP relay port pool (even ports)
"webUiPort": 3060 // Dashboard port
"publicIpSeed": "stun.example.com", // STUN server for public IP discovery
"rtpPortRange": { "min": 20000, "max": 20200 }, // RTP port pool (even ports)
"webUiPort": 3060 // Dashboard + REST API port
},
"providers": [
{
"id": "my-trunk",
"name": "My SIP Provider",
"host": "sip.provider.com",
"port": 5060,
"displayName": "My SIP Provider",
"domain": "sip.provider.com",
"outboundProxy": { "address": "sip.provider.com", "port": 5060 },
"username": "user",
"password": "pass",
"codecs": ["G.722", "PCMA", "PCMU"],
"registerExpiry": 3600
"codecs": [9, 0, 8, 101], // G.722, PCMU, PCMA, telephone-event
"registerIntervalSec": 300
}
],
"devices": [
{
"id": "desk-phone",
"name": "Desk Phone",
"type": "sip"
"displayName": "Desk Phone",
"expectedAddress": "192.168.1.50",
"extension": "100"
}
],
"routing": {
"inbound": {
"default": { "target": "all-devices", "ringBrowser": true }
"routes": [
{
"id": "inbound-default",
"name": "Ring all devices",
"priority": 100,
"direction": "inbound",
"match": {},
"action": {
"targets": ["desk-phone"],
"ringBrowsers": true,
"voicemailBox": "main",
"noAnswerTimeout": 25
}
},
{
"id": "outbound-default",
"name": "Route via trunk",
"priority": 100,
"direction": "outbound",
"match": {},
"action": { "provider": "my-trunk" }
}
]
},
"voiceboxes": [
{
"id": "main",
"enabled": true,
"greetingText": "Please leave a message after the beep.",
"greetingVoice": "af_bella",
"noAnswerTimeoutSec": 25,
"maxRecordingSec": 120,
"maxMessages": 50
}
],
"contacts": [
{ "id": "1", "name": "Alice", "number": "+491234567890", "starred": true }
]
}
```
### TTS Setup (Optional)
For neural "connecting your call" announcements, download the Kokoro TTS model:
For neural announcements and voicemail greetings, download the Kokoro TTS model:
```bash
mkdir -p .nogit/tts
# Download the full-quality model (310MB) + voices (27MB)
curl -L -o .nogit/tts/kokoro-v1.0.onnx \
https://github.com/mzdk100/kokoro/releases/download/V1.0/kokoro-v1.0.onnx
curl -L -o .nogit/tts/voices.bin \
https://github.com/mzdk100/kokoro/releases/download/V1.0/voices.bin
```
If the model files aren't present, the announcement feature is simply disabled — everything else works fine.
Without the model files, TTS falls back to `espeak-ng`. Without either, announcements are skipped — everything else works fine.
### Run
@@ -145,7 +189,7 @@ If the model files aren't present, the announcement feature is simply disabled
pnpm start
```
The SIP proxy starts on the configured port and the web dashboard is available at `http://<your-ip>:3060`.
The SIP proxy starts on the configured port and the web dashboard is available at `https://<your-ip>:3060`.
### HTTPS (Optional)
@@ -157,68 +201,91 @@ Place `cert.pem` and `key.pem` in `.nogit/` for TLS on the dashboard.
```
siprouter/
├── ts/ # TypeScript source
├── ts/ # TypeScript control plane
│ ├── sipproxy.ts # Main entry — bootstraps everything
│ ├── config.ts # Config loader & validation
│ ├── registrar.ts # Local SIP registrar for devices
│ ├── providerstate.ts # Per-provider upstream registration engine
│ ├── proxybridge.ts # Rust proxy-engine IPC bridge (smartrust)
│ ├── frontend.ts # Web dashboard HTTP/WS server + REST API
│ ├── webrtcbridge.ts # WebRTC signaling layer
│ ├── opusbridge.ts # Rust IPC bridge (smartrust)
│ ├── codec.ts # High-level RTP transcoding interface
│ ├── announcement.ts # Neural TTS announcement generator
── sip/ # Zero-dependency SIP protocol library
── message.ts # SIP message parser/builder/mutator
│ ├── dialog.ts # RFC 3261 dialog state machine
│ │ ├── helpers.ts # SDP builder, digest auth, codec registry
│ │ └── rewrite.ts # SIP URI + SDP body rewriting
│ └── call/ # Hub-model call management
│ ├── call-manager.ts # Central registry, factory, routing
│ ├── call.ts # Call hub — owns N legs, media fan-out
│ ├── sip-leg.ts # SIP device/provider connection
│ ├── webrtc-leg.ts # Browser WebRTC connection
│ └── rtp-port-pool.ts # UDP port allocation
│ ├── registrar.ts # Browser softphone registration
│ ├── announcement.ts # TTS announcement generator (espeak-ng / Kokoro)
│ ├── voicebox.ts # Voicemail box management
── call/
── prompt-cache.ts # Named audio prompt WAV management
├── ts_web/ # Web frontend (Lit-based SPA)
│ ├── elements/ # Web components (dashboard, phone, etc.)
│ ├── elements/ # Web components (9 dashboard views)
│ └── state/ # App state, WebRTC client, notifications
├── rust/ # Rust workspace
├── rust/ # Rust workspace (the data plane)
│ └── crates/
│ ├── opus-codec/ # Real-time audio transcoder (Opus/G.722/PCM)
── tts-engine/ # Kokoro neural TTS CLI
│ ├── codec-lib/ # Audio codec library (Opus/G.722/PCMU/PCMA)
── sip-proto/ # Zero-dependency SIP protocol library
│ └── proxy-engine/ # Main binary — SIP engine + mixer + RTP
├── html/ # Static HTML shell
├── .nogit/ # Secrets, config, models (gitignored)
└── dist_rust/ # Compiled Rust binaries (gitignored)
├── .nogit/ # Secrets, config, TTS models (gitignored)
└── dist_rust/ # Compiled Rust binary (gitignored)
```
---
## 🎧 Codec Engine (Rust)
## 🎧 Audio Engine (Rust)
The `opus-codec` binary handles all real-time audio processing via a JSON-over-stdio IPC protocol:
The `proxy-engine` binary handles all real-time audio processing with a **48kHz f32 internal bus** — encoding and decoding happens only at leg boundaries.
| Codec | Payload Type | Sample Rate | Use Case |
|-------|-------------|-------------|----------|
| **Opus** | 111 | 48 kHz | WebRTC browsers |
| **G.722** | 9 | 16 kHz | HD SIP devices |
### Supported Codecs
| Codec | PT | Native Rate | Use Case |
|-------|:--:|:-----------:|----------|
| **Opus** | 111 | 48 kHz | WebRTC browsers (native float encode/decode — zero i16 quantization) |
| **G.722** | 9 | 16 kHz | HD SIP devices & providers |
| **PCMU** (G.711 µ-law) | 0 | 8 kHz | Legacy SIP |
| **PCMA** (G.711 A-law) | 8 | 8 kHz | Legacy SIP |
**Features:**
- Per-call isolated codec sessions (no cross-call state corruption)
- FFT-based sample rate conversion via `rubato`
- **RNNoise ML noise suppression** with per-direction state — denoises audio flowing to SIP separately from audio flowing to the browser
- Raw PCM encoding for TTS frame processing
### Audio Pipeline
```
Inbound: Wire RTP → Decode → Resample to 48kHz → Denoise (RNNoise) → Mix Bus
Outbound: Mix Bus → Mix-Minus → Resample to codec rate → Encode → Wire RTP
```
- **FFT-based resampling** via `rubato` — high-quality sinc interpolation with cached resampler state for seamless inter-frame continuity
- **ML noise suppression** via `nnnoiseless` (RNNoise) — per-leg inbound denoising with SIMD acceleration (AVX/SSE). Skipped for WebRTC legs (browsers already denoise via getUserMedia)
- **Mix-minus mixing** — each participant hears everyone except themselves, accumulated in f64 precision
- **In-tick packet reorder** — inbound RTP packets are sorted by sequence number before decoding, protecting G.722 ADPCM state from out-of-order delivery
- **RFC 3550 compliant header parsing** — properly handles CSRC lists and header extensions
---
## 🗣️ Neural TTS (Rust)
## 🗣️ Neural TTS
The `tts-engine` binary uses [Kokoro TTS](https://github.com/mzdk100/kokoro) (82M parameter neural model) to synthesize announcements at startup:
Announcements and voicemail greetings are synthesized using [Kokoro TTS](https://github.com/mzdk100/kokoro) — an 82M parameter neural model running via ONNX Runtime directly in the Rust process:
- **24 kHz, 16-bit mono** output
- **25+ voice presets** — American/British, male/female (e.g., `af_bella`, `am_adam`, `bf_emma`, `bm_george`)
- **~800ms** synthesis time for a 3-second announcement
- Pre-encoded to G.722 + Opus for zero-latency RTP playback during call setup
- **~800ms** synthesis time for a 3-second phrase
- Lazy-loaded on first use — no startup cost if TTS is unused
- Falls back to `espeak-ng` if the ONNX model is not available
---
## 📧 Voicemail
- Configurable voicemail boxes with custom TTS greetings
- Automatic routing on no-answer timeout
- Recording with configurable max duration and message count
- Web dashboard playback and management
- WAV storage in `.nogit/voicemail/`
---
## 🔢 IVR (Interactive Voice Response)
- DTMF-navigable menus with configurable entries
- Actions: route to extension, route to voicemail, transfer, submenu, hangup, repeat prompt
- Custom TTS prompts per menu
- Nested menu support
---
@@ -228,25 +295,32 @@ The `tts-engine` binary uses [Kokoro TTS](https://github.com/mzdk100/kokoro) (82
| View | Description |
|------|-------------|
| **Overview** | Stats tiles — uptime, providers, devices, active calls |
| **Calls** | Active calls with leg details, codec info, packet counters. Add/remove legs, transfer, hangup |
| **Phone** | Browser softphone — mic/speaker selection, audio meters, dial pad, incoming call popup |
| **Contacts** | Contact management with click-to-call |
| **Providers** | SIP trunk config with registration status |
| **Log** | Live streaming log viewer |
| 📊 **Overview** | Stats tiles — uptime, providers, devices, active calls |
| 📞 **Calls** | Active calls with leg details, codec info, add/remove legs, transfer, hangup |
| ☎️ **Phone** | Browser softphone — mic/speaker selection, audio meters, dial pad, incoming call popup |
| 🔀 **Routes** | Routing rule management — match/action model with priority |
| 📧 **Voicemail** | Voicemail box management + message playback |
| 🔢 **IVR** | IVR menu builder — DTMF entries, TTS prompts, nested menus |
| 👤 **Contacts** | Contact management with click-to-call |
| 🔌 **Providers** | SIP trunk configuration and registration status |
| 📋 **Log** | Live streaming log viewer |
### REST API
| Endpoint | Method | Description |
|----------|--------|-------------|
| `/api/status` | GET | Full system status (providers, devices, calls) |
| `/api/status` | GET | Full system status (providers, devices, calls, history) |
| `/api/call` | POST | Originate a call |
| `/api/hangup` | POST | Hang up a call |
| `/api/call/:id/addleg` | POST | Add a leg to an active call |
| `/api/call/:id/addexternal` | POST | Add an external participant |
| `/api/call/:id/addleg` | POST | Add a device leg to an active call |
| `/api/call/:id/addexternal` | POST | Add an external participant via provider |
| `/api/call/:id/removeleg` | POST | Remove a leg from a call |
| `/api/transfer` | POST | Transfer a call |
| `/api/config` | GET/POST | Read or update configuration (hot-reload) |
| `/api/config` | GET | Read current configuration |
| `/api/config` | POST | Update configuration (hot-reload) |
| `/api/voicemail/:box` | GET | List voicemail messages |
| `/api/voicemail/:box/:id` | DELETE | Delete a voicemail message |
| `/api/voicemail/:box/:id/audio` | GET | Stream voicemail audio |
### WebSocket Events
@@ -255,6 +329,8 @@ Connect to `/ws` for real-time push:
```jsonc
{ "type": "status", "data": { ... } } // Full status snapshot (1s interval)
{ "type": "log", "data": { "message": "..." } } // Log lines in real-time
{ "type": "incoming_call", "data": { ... } } // Incoming call notification
{ "type": "call_ended", "data": { ... } } // Call ended notification
```
---
@@ -264,7 +340,7 @@ Connect to `/ws` for real-time push:
| Port | Protocol | Purpose |
|------|----------|---------|
| 5070 (configurable) | UDP | SIP signaling |
| 2000020200 (configurable) | UDP | RTP relay (even ports, per-call allocation) |
| 2000020200 (configurable) | UDP | RTP media (even ports, per-call allocation) |
| 3060 (configurable) | TCP | Web dashboard + WebSocket + REST API |
---
@@ -275,23 +351,16 @@ Connect to `/ws` for real-time push:
# Start in dev mode
pnpm start
# Build Rust crates
# Build Rust proxy-engine
pnpm run buildRust
# Bundle web frontend
pnpm run bundle
# Restart background server (build + bundle + restart)
# Build + bundle + restart background server
pnpm run restartBackground
```
### Key Design Decisions
- **Hub Model** — Calls are N-leg hubs, not point-to-point. This enables multi-party, dynamic leg manipulation, and transfer without tearing down the call.
- **Zero-dependency SIP library** — `ts/sip/` is a pure data-level SIP stack (parse/build/mutate/serialize). No transport or timer logic — those live in the application layer.
- **Rust for the hot path** — Codec transcoding and noise suppression run in native Rust for real-time performance. TypeScript handles signaling and orchestration.
- **Per-session codec isolation** — Each call gets its own Opus/G.722 encoder/decoder state in the Rust process, preventing stateful codec prediction from leaking between concurrent calls.
---
## License and Legal Information

2199
rust/Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@@ -1,5 +1,9 @@
[workspace]
members = ["crates/opus-codec", "crates/tts-engine"]
members = [
"crates/codec-lib",
"crates/sip-proto",
"crates/proxy-engine",
]
resolver = "2"
[profile.release]

View File

@@ -0,0 +1,10 @@
[package]
name = "codec-lib"
version = "0.1.0"
edition = "2021"
[dependencies]
audiopus = "0.3.0-rc.0"
ezk-g722 = "0.1"
rubato = "0.14"
nnnoiseless = "0.5"

View File

@@ -0,0 +1,472 @@
//! Audio codec library for the SIP router.
//!
//! Handles Opus ↔ G.722 ↔ PCMU/PCMA transcoding with ML noise suppression.
//! Used by the `proxy-engine` binary for all audio transcoding.
use audiopus::coder::{Decoder as OpusDecoder, Encoder as OpusEncoder};
use audiopus::packet::Packet as OpusPacket;
use audiopus::{Application, Bitrate as OpusBitrate, Channels, MutSignals, SampleRate};
use ezk_g722::libg722::{self, Bitrate};
use nnnoiseless::DenoiseState;
use rubato::{FftFixedIn, Resampler};
use std::collections::HashMap;
// ---- Payload type constants ------------------------------------------------
pub const PT_PCMU: u8 = 0;
pub const PT_PCMA: u8 = 8;
pub const PT_G722: u8 = 9;
pub const PT_OPUS: u8 = 111;
/// Return the native sample rate for a given payload type.
pub fn codec_sample_rate(pt: u8) -> u32 {
match pt {
PT_OPUS => 48000,
PT_G722 => 16000,
_ => 8000, // PCMU, PCMA
}
}
// ---- G.711 µ-law (PCMU) ---------------------------------------------------
pub fn mulaw_encode(sample: i16) -> u8 {
const BIAS: i16 = 0x84;
const CLIP: i16 = 32635;
let sign = if sample < 0 { 0x80u8 } else { 0 };
let mut s = (sample as i32).unsigned_abs().min(CLIP as u32) as i16;
s += BIAS;
let mut exp = 7u8;
let mut mask = 0x4000i16;
while exp > 0 && (s & mask) == 0 {
exp -= 1;
mask >>= 1;
}
let mantissa = ((s >> (exp + 3)) & 0x0f) as u8;
!(sign | (exp << 4) | mantissa)
}
pub fn mulaw_decode(mulaw: u8) -> i16 {
let v = !mulaw;
let sign = v & 0x80;
let exp = (v >> 4) & 0x07;
let mantissa = v & 0x0f;
// Use i32 to avoid overflow when exp=7, mantissa=15 (result > i16::MAX).
let mut sample = (((mantissa as i32) << 4) + 0x84) << exp;
sample -= 0x84;
let sample = if sign != 0 { -sample } else { sample };
sample.clamp(-32768, 32767) as i16
}
// ---- G.711 A-law (PCMA) ---------------------------------------------------
pub fn alaw_encode(sample: i16) -> u8 {
let sign = if sample >= 0 { 0x80u8 } else { 0 };
let s = (sample as i32).unsigned_abs().min(32767) as i16;
let mut exp = 7u8;
let mut mask = 0x4000i16;
while exp > 0 && (s & mask) == 0 {
exp -= 1;
mask >>= 1;
}
let mantissa = if exp > 0 {
((s >> (exp + 3)) & 0x0f) as u8
} else {
((s >> 4) & 0x0f) as u8
};
(sign | (exp << 4) | mantissa) ^ 0x55
}
pub fn alaw_decode(alaw: u8) -> i16 {
let v = alaw ^ 0x55;
let sign = v & 0x80;
let exp = (v >> 4) & 0x07;
let mantissa = v & 0x0f;
// Use i32 to avoid overflow for extreme values.
let sample = if exp == 0 {
((mantissa as i32) << 4) + 8
} else {
(((mantissa as i32) << 4) + 0x108) << (exp - 1)
};
let sample = if sign != 0 { sample } else { -sample };
sample.clamp(-32768, 32767) as i16
}
// ---- TranscodeState --------------------------------------------------------
/// Per-session codec state holding Opus, G.722, resampler, and denoiser instances.
///
/// Each concurrent call should get its own `TranscodeState` to prevent stateful
/// codecs (Opus, G.722 ADPCM) from corrupting each other.
pub struct TranscodeState {
opus_enc: OpusEncoder,
opus_dec: OpusDecoder,
g722_enc: libg722::encoder::Encoder,
g722_dec: libg722::decoder::Decoder,
/// Cached FFT resamplers keyed by (from_rate, to_rate, chunk_size).
resamplers: HashMap<(u32, u32, usize), FftFixedIn<f64>>,
/// Cached f32 FFT resamplers keyed by (from_rate, to_rate, chunk_size).
resamplers_f32: HashMap<(u32, u32, usize), FftFixedIn<f32>>,
/// ML noise suppression for the SIP-bound direction.
denoiser_to_sip: Box<DenoiseState<'static>>,
/// ML noise suppression for the browser-bound direction.
denoiser_to_browser: Box<DenoiseState<'static>>,
}
impl TranscodeState {
/// Create a new transcoding session with fresh codec state.
pub fn new() -> Result<Self, String> {
let mut opus_enc =
OpusEncoder::new(SampleRate::Hz48000, Channels::Mono, Application::Voip)
.map_err(|e| format!("opus encoder: {e}"))?;
opus_enc
.set_complexity(5)
.map_err(|e| format!("opus set_complexity: {e}"))?;
opus_enc
.set_bitrate(OpusBitrate::BitsPerSecond(24000))
.map_err(|e| format!("opus set_bitrate: {e}"))?;
let opus_dec = OpusDecoder::new(SampleRate::Hz48000, Channels::Mono)
.map_err(|e| format!("opus decoder: {e}"))?;
let g722_enc = libg722::encoder::Encoder::new(Bitrate::Mode1_64000, false, false);
let g722_dec = libg722::decoder::Decoder::new(Bitrate::Mode1_64000, false, false);
Ok(Self {
opus_enc,
opus_dec,
g722_enc,
g722_dec,
resamplers: HashMap::new(),
resamplers_f32: HashMap::new(),
denoiser_to_sip: DenoiseState::new(),
denoiser_to_browser: DenoiseState::new(),
})
}
/// High-quality sample rate conversion using rubato FFT resampler.
/// Resamplers are cached by (from_rate, to_rate, chunk_size) and reused,
/// maintaining proper inter-frame state for continuous audio streams.
pub fn resample(
&mut self,
pcm: &[i16],
from_rate: u32,
to_rate: u32,
) -> Result<Vec<i16>, String> {
if from_rate == to_rate || pcm.is_empty() {
return Ok(pcm.to_vec());
}
let chunk = pcm.len();
let key = (from_rate, to_rate, chunk);
if !self.resamplers.contains_key(&key) {
let r =
FftFixedIn::<f64>::new(from_rate as usize, to_rate as usize, chunk, 1, 1)
.map_err(|e| format!("resampler {from_rate}->{to_rate}: {e}"))?;
self.resamplers.insert(key, r);
}
let resampler = self.resamplers.get_mut(&key).unwrap();
let float_in: Vec<f64> = pcm.iter().map(|&s| s as f64 / 32768.0).collect();
let input = vec![float_in];
let result = resampler
.process(&input, None)
.map_err(|e| format!("resample {from_rate}->{to_rate}: {e}"))?;
Ok(result[0]
.iter()
.map(|&s| (s * 32767.0).round().clamp(-32768.0, 32767.0) as i16)
.collect())
}
/// Apply RNNoise ML noise suppression to 48kHz PCM audio.
/// Processes in 480-sample (10ms) frames. State persists across calls.
pub fn denoise(denoiser: &mut DenoiseState, pcm: &[i16]) -> Vec<i16> {
let frame_size = DenoiseState::FRAME_SIZE; // 480
let total = pcm.len();
let whole = (total / frame_size) * frame_size;
let mut output = Vec::with_capacity(total);
let mut out_buf = [0.0f32; 480];
for offset in (0..whole).step_by(frame_size) {
let input: Vec<f32> = pcm[offset..offset + frame_size]
.iter()
.map(|&s| s as f32)
.collect();
denoiser.process_frame(&mut out_buf, &input);
output.extend(
out_buf
.iter()
.map(|&s| s.round().clamp(-32768.0, 32767.0) as i16),
);
}
if whole < total {
output.extend_from_slice(&pcm[whole..]);
}
output
}
/// Transcode audio payload from one codec to another.
///
/// `direction`: `Some("to_sip")` or `Some("to_browser")` selects per-direction
/// denoiser. `None` skips denoising (backward compat).
pub fn transcode(
&mut self,
data: &[u8],
from_pt: u8,
to_pt: u8,
direction: Option<&str>,
) -> Result<Vec<u8>, String> {
if from_pt == to_pt {
return Ok(data.to_vec());
}
let (pcm, rate) = self.decode_to_pcm(data, from_pt)?;
let processed = if let Some(dir) = direction {
let pcm_48k = self.resample(&pcm, rate, 48000)?;
let denoiser = match dir {
"to_sip" => &mut self.denoiser_to_sip,
_ => &mut self.denoiser_to_browser,
};
let denoised = Self::denoise(denoiser, &pcm_48k);
let target_rate = codec_sample_rate(to_pt);
self.resample(&denoised, 48000, target_rate)?
} else {
let target_rate = codec_sample_rate(to_pt);
if rate == target_rate {
pcm
} else {
self.resample(&pcm, rate, target_rate)?
}
};
self.encode_from_pcm(&processed, to_pt)
}
/// Decode an encoded audio payload to raw 16-bit PCM samples.
/// Returns (samples, sample_rate).
pub fn decode_to_pcm(&mut self, data: &[u8], pt: u8) -> Result<(Vec<i16>, u32), String> {
match pt {
PT_OPUS => {
let mut pcm = vec![0i16; 5760]; // up to 120ms at 48kHz
let packet =
OpusPacket::try_from(data).map_err(|e| format!("opus packet: {e}"))?;
let out =
MutSignals::try_from(&mut pcm[..]).map_err(|e| format!("opus signals: {e}"))?;
let n: usize = self
.opus_dec
.decode(Some(packet), out, false)
.map_err(|e| format!("opus decode: {e}"))?
.into();
pcm.truncate(n);
Ok((pcm, 48000))
}
PT_G722 => {
let pcm = self.g722_dec.decode(data);
Ok((pcm, 16000))
}
PT_PCMU => {
let pcm: Vec<i16> = data.iter().map(|&b| mulaw_decode(b)).collect();
Ok((pcm, 8000))
}
PT_PCMA => {
let pcm: Vec<i16> = data.iter().map(|&b| alaw_decode(b)).collect();
Ok((pcm, 8000))
}
_ => Err(format!("unsupported source PT {pt}")),
}
}
/// Encode raw PCM samples to an audio codec.
pub fn encode_from_pcm(&mut self, pcm: &[i16], pt: u8) -> Result<Vec<u8>, String> {
match pt {
PT_OPUS => {
let mut buf = vec![0u8; 4000];
let n: usize = self
.opus_enc
.encode(pcm, &mut buf)
.map_err(|e| format!("opus encode: {e}"))?
.into();
buf.truncate(n);
Ok(buf)
}
PT_G722 => Ok(self.g722_enc.encode(pcm)),
PT_PCMU => Ok(pcm.iter().map(|&s| mulaw_encode(s)).collect()),
PT_PCMA => Ok(pcm.iter().map(|&s| alaw_encode(s)).collect()),
_ => Err(format!("unsupported target PT {pt}")),
}
}
// ---- f32 API for high-quality internal bus ----------------------------
/// Decode an encoded audio payload to f32 PCM samples in [-1.0, 1.0].
/// Returns (samples, sample_rate).
///
/// For Opus, uses native float decode (no i16 quantization).
/// For G.722/G.711, decodes to i16 then converts (codec is natively i16).
pub fn decode_to_f32(&mut self, data: &[u8], pt: u8) -> Result<(Vec<f32>, u32), String> {
match pt {
PT_OPUS => {
let mut pcm = vec![0.0f32; 5760]; // up to 120ms at 48kHz
let packet =
OpusPacket::try_from(data).map_err(|e| format!("opus packet: {e}"))?;
let out =
MutSignals::try_from(&mut pcm[..]).map_err(|e| format!("opus signals: {e}"))?;
let n: usize = self
.opus_dec
.decode_float(Some(packet), out, false)
.map_err(|e| format!("opus decode_float: {e}"))?
.into();
pcm.truncate(n);
Ok((pcm, 48000))
}
_ => {
// G.722, PCMU, PCMA: natively i16 codecs — decode then convert.
let (pcm_i16, rate) = self.decode_to_pcm(data, pt)?;
let pcm_f32 = pcm_i16.iter().map(|&s| s as f32 / 32768.0).collect();
Ok((pcm_f32, rate))
}
}
}
/// Encode f32 PCM samples ([-1.0, 1.0]) to an audio codec.
///
/// For Opus, uses native float encode (no i16 quantization).
/// For G.722/G.711, converts to i16 then encodes (codec is natively i16).
pub fn encode_from_f32(&mut self, pcm: &[f32], pt: u8) -> Result<Vec<u8>, String> {
match pt {
PT_OPUS => {
let mut buf = vec![0u8; 4000];
let n: usize = self
.opus_enc
.encode_float(pcm, &mut buf)
.map_err(|e| format!("opus encode_float: {e}"))?
.into();
buf.truncate(n);
Ok(buf)
}
_ => {
// G.722, PCMU, PCMA: natively i16 codecs.
let pcm_i16: Vec<i16> = pcm
.iter()
.map(|&s| (s * 32767.0).round().clamp(-32768.0, 32767.0) as i16)
.collect();
self.encode_from_pcm(&pcm_i16, pt)
}
}
}
/// High-quality sample rate conversion for f32 PCM using rubato FFT resampler.
/// Uses a separate cache from the i16 resampler.
pub fn resample_f32(
&mut self,
pcm: &[f32],
from_rate: u32,
to_rate: u32,
) -> Result<Vec<f32>, String> {
if from_rate == to_rate || pcm.is_empty() {
return Ok(pcm.to_vec());
}
let chunk = pcm.len();
let key = (from_rate, to_rate, chunk);
if !self.resamplers_f32.contains_key(&key) {
let r =
FftFixedIn::<f32>::new(from_rate as usize, to_rate as usize, chunk, 1, 1)
.map_err(|e| format!("resampler f32 {from_rate}->{to_rate}: {e}"))?;
self.resamplers_f32.insert(key, r);
}
let resampler = self.resamplers_f32.get_mut(&key).unwrap();
let input = vec![pcm.to_vec()];
let result = resampler
.process(&input, None)
.map_err(|e| format!("resample f32 {from_rate}->{to_rate}: {e}"))?;
Ok(result[0].clone())
}
/// Apply RNNoise ML noise suppression to 48kHz f32 PCM audio.
/// Processes in 480-sample (10ms) frames. State persists across calls.
/// Operates natively in f32 — no i16 conversion overhead.
pub fn denoise_f32(denoiser: &mut DenoiseState, pcm: &[f32]) -> Vec<f32> {
let frame_size = DenoiseState::FRAME_SIZE; // 480
let total = pcm.len();
let whole = (total / frame_size) * frame_size;
let mut output = Vec::with_capacity(total);
let mut out_buf = [0.0f32; 480];
// nnnoiseless expects f32 samples scaled as i16 range (-32768..32767).
for offset in (0..whole).step_by(frame_size) {
let input: Vec<f32> = pcm[offset..offset + frame_size]
.iter()
.map(|&s| s * 32768.0)
.collect();
denoiser.process_frame(&mut out_buf, &input);
output.extend(out_buf.iter().map(|&s| s / 32768.0));
}
if whole < total {
output.extend_from_slice(&pcm[whole..]);
}
output
}
}
/// Create a new standalone denoiser for per-leg inbound processing.
pub fn new_denoiser() -> Box<DenoiseState<'static>> {
DenoiseState::new()
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn mulaw_roundtrip() {
for sample in [-32768i16, -1000, -1, 0, 1, 1000, 32767] {
let encoded = mulaw_encode(sample);
let decoded = mulaw_decode(encoded);
// µ-law is lossy; verify the decoded value is close.
assert!((sample as i32 - decoded as i32).abs() < 1000,
"µ-law roundtrip failed for {sample}: got {decoded}");
}
}
#[test]
fn alaw_roundtrip() {
for sample in [-32768i16, -1000, -1, 0, 1, 1000, 32767] {
let encoded = alaw_encode(sample);
let decoded = alaw_decode(encoded);
assert!((sample as i32 - decoded as i32).abs() < 1000,
"A-law roundtrip failed for {sample}: got {decoded}");
}
}
#[test]
fn codec_sample_rates() {
assert_eq!(codec_sample_rate(PT_OPUS), 48000);
assert_eq!(codec_sample_rate(PT_G722), 16000);
assert_eq!(codec_sample_rate(PT_PCMU), 8000);
assert_eq!(codec_sample_rate(PT_PCMA), 8000);
}
#[test]
fn transcode_same_pt_is_passthrough() {
let mut st = TranscodeState::new().unwrap();
let data = vec![0u8; 160];
let result = st.transcode(&data, PT_PCMU, PT_PCMU, None).unwrap();
assert_eq!(result, data);
}
#[test]
fn pcmu_to_pcma_roundtrip() {
let mut st = TranscodeState::new().unwrap();
// 160 bytes = 20ms of PCMU at 8kHz
let pcmu_data: Vec<u8> = (0..160).map(|i| mulaw_encode((i as i16 * 200) - 16000)).collect();
let pcma = st.transcode(&pcmu_data, PT_PCMU, PT_PCMA, None).unwrap();
assert_eq!(pcma.len(), 160); // Same frame size
let back = st.transcode(&pcma, PT_PCMA, PT_PCMU, None).unwrap();
assert_eq!(back.len(), 160);
}
}

View File

@@ -1,17 +0,0 @@
[package]
name = "opus-codec"
version = "0.2.0"
edition = "2021"
[[bin]]
name = "opus-codec"
path = "src/main.rs"
[dependencies]
audiopus = "0.3.0-rc.0"
ezk-g722 = "0.1"
rubato = "0.14"
serde = { version = "1", features = ["derive"] }
serde_json = "1"
base64 = "0.22"
nnnoiseless = { version = "0.5", default-features = false }

View File

@@ -1,464 +0,0 @@
/// Audio transcoding bridge for smartrust.
///
/// Handles Opus ↔ G.722 ↔ PCMU transcoding for the SIP router.
/// Uses audiopus (libopus) for Opus and ezk-g722 (SpanDSP port) for G.722.
///
/// Supports per-session codec state so concurrent calls don't corrupt each
/// other's stateful codecs (Opus, G.722 ADPCM).
///
/// Protocol:
/// -> {"id":"1","method":"init","params":{}}
/// <- {"id":"1","success":true,"result":{}}
/// -> {"id":"2","method":"create_session","params":{"session_id":"call-abc"}}
/// <- {"id":"2","success":true,"result":{}}
/// -> {"id":"3","method":"transcode","params":{"session_id":"call-abc","data_b64":"...","from_pt":111,"to_pt":9}}
/// <- {"id":"3","success":true,"result":{"data_b64":"..."}}
/// -> {"id":"4","method":"destroy_session","params":{"session_id":"call-abc"}}
/// <- {"id":"4","success":true,"result":{}}
use audiopus::coder::{Decoder as OpusDecoder, Encoder as OpusEncoder};
use audiopus::packet::Packet as OpusPacket;
use audiopus::{Application, Bitrate as OpusBitrate, Channels, MutSignals, SampleRate};
use base64::engine::general_purpose::STANDARD as B64;
use base64::Engine as _;
use ezk_g722::libg722::{self, Bitrate};
use nnnoiseless::DenoiseState;
use rubato::{FftFixedIn, Resampler};
use serde::Deserialize;
use std::collections::HashMap;
use std::io::{self, BufRead, Write};
// Payload type constants.
const PT_PCMU: u8 = 0;
const PT_PCMA: u8 = 8;
const PT_G722: u8 = 9;
const PT_OPUS: u8 = 111;
#[derive(Deserialize)]
struct Request {
id: String,
method: String,
#[serde(default)]
params: serde_json::Value,
}
fn respond(out: &mut impl Write, id: &str, success: bool, result: Option<serde_json::Value>, error: Option<&str>) {
let mut resp = serde_json::json!({ "id": id, "success": success });
if let Some(r) = result { resp["result"] = r; }
if let Some(e) = error { resp["error"] = serde_json::Value::String(e.to_string()); }
let _ = writeln!(out, "{}", resp);
let _ = out.flush();
}
// ---------------------------------------------------------------------------
// Codec state
// ---------------------------------------------------------------------------
struct TranscodeState {
opus_enc: OpusEncoder,
opus_dec: OpusDecoder,
g722_enc: libg722::encoder::Encoder,
g722_dec: libg722::decoder::Decoder,
// Cached FFT resamplers keyed by (from_rate, to_rate, chunk_size).
resamplers: HashMap<(u32, u32, usize), FftFixedIn<f64>>,
// Per-direction ML noise suppression (RNNoise). Separate state per direction
// prevents the RNN hidden state from being corrupted by interleaved audio streams.
denoiser_to_sip: Box<DenoiseState<'static>>,
denoiser_to_browser: Box<DenoiseState<'static>>,
}
impl TranscodeState {
fn new() -> Result<Self, String> {
let mut opus_enc = OpusEncoder::new(SampleRate::Hz48000, Channels::Mono, Application::Voip)
.map_err(|e| format!("opus encoder: {e}"))?;
// Telephony-grade tuning: complexity 5 is sufficient for voice bridged to G.722.
opus_enc.set_complexity(5).map_err(|e| format!("opus set_complexity: {e}"))?;
opus_enc.set_bitrate(OpusBitrate::BitsPerSecond(24000)).map_err(|e| format!("opus set_bitrate: {e}"))?;
let opus_dec = OpusDecoder::new(SampleRate::Hz48000, Channels::Mono)
.map_err(|e| format!("opus decoder: {e}"))?;
let g722_enc = libg722::encoder::Encoder::new(Bitrate::Mode1_64000, false, false);
let g722_dec = libg722::decoder::Decoder::new(Bitrate::Mode1_64000, false, false);
Ok(Self {
opus_enc, opus_dec, g722_enc, g722_dec,
resamplers: HashMap::new(),
denoiser_to_sip: DenoiseState::new(),
denoiser_to_browser: DenoiseState::new(),
})
}
/// High-quality sample rate conversion using rubato FFT resampler.
/// Resamplers are cached by (from_rate, to_rate, chunk_size) and reused,
/// maintaining proper inter-frame state for continuous audio streams.
fn resample(&mut self, pcm: &[i16], from_rate: u32, to_rate: u32) -> Result<Vec<i16>, String> {
if from_rate == to_rate || pcm.is_empty() {
return Ok(pcm.to_vec());
}
let chunk = pcm.len();
let key = (from_rate, to_rate, chunk);
// Get or create cached resampler for this rate pair + chunk size.
if !self.resamplers.contains_key(&key) {
let r = FftFixedIn::<f64>::new(from_rate as usize, to_rate as usize, chunk, 1, 1)
.map_err(|e| format!("resampler {from_rate}->{to_rate}: {e}"))?;
self.resamplers.insert(key, r);
}
let resampler = self.resamplers.get_mut(&key).unwrap();
// i16 → f64 normalized to [-1.0, 1.0]
let float_in: Vec<f64> = pcm.iter().map(|&s| s as f64 / 32768.0).collect();
let input = vec![float_in];
let result = resampler.process(&input, None)
.map_err(|e| format!("resample {from_rate}->{to_rate}: {e}"))?;
// f64 → i16
Ok(result[0].iter()
.map(|&s| (s * 32767.0).round().clamp(-32768.0, 32767.0) as i16)
.collect())
}
/// Apply RNNoise ML noise suppression to 48kHz PCM audio.
/// Processes in 480-sample (10ms) frames. State persists across calls.
fn denoise(denoiser: &mut DenoiseState, pcm: &[i16]) -> Vec<i16> {
let frame_size = DenoiseState::FRAME_SIZE; // 480
let total = pcm.len();
// Round down to whole frames — don't process partial frames to avoid
// injecting artificial silence into the RNN state.
let whole = (total / frame_size) * frame_size;
let mut output = Vec::with_capacity(total);
let mut out_buf = [0.0f32; 480];
for offset in (0..whole).step_by(frame_size) {
let input: Vec<f32> = pcm[offset..offset + frame_size]
.iter().map(|&s| s as f32).collect();
denoiser.process_frame(&mut out_buf, &input);
output.extend(out_buf.iter()
.map(|&s| s.round().clamp(-32768.0, 32767.0) as i16));
}
// Pass through any trailing partial-frame samples unmodified.
if whole < total {
output.extend_from_slice(&pcm[whole..]);
}
output
}
/// Transcode audio payload from one codec to another.
/// `direction`: "to_sip" or "to_browser" — selects the per-direction denoiser.
/// If None, denoising is skipped (backward compat).
fn transcode(&mut self, data: &[u8], from_pt: u8, to_pt: u8, direction: Option<&str>) -> Result<Vec<u8>, String> {
if from_pt == to_pt {
return Ok(data.to_vec());
}
// Decode to PCM (at source sample rate).
let (pcm, rate) = self.decode_to_pcm(data, from_pt)?;
// Apply noise suppression if direction is specified.
let processed = if let Some(dir) = direction {
// Resample to 48kHz for denoising (no-op when already 48kHz).
let pcm_48k = self.resample(&pcm, rate, 48000)?;
let denoiser = match dir {
"to_sip" => &mut self.denoiser_to_sip,
_ => &mut self.denoiser_to_browser,
};
let denoised = Self::denoise(denoiser, &pcm_48k);
// Resample to target rate (no-op when target is 48kHz).
let target_rate = codec_sample_rate(to_pt);
self.resample(&denoised, 48000, target_rate)?
} else {
// No denoising — direct resample.
let target_rate = codec_sample_rate(to_pt);
if rate == target_rate { pcm } else { self.resample(&pcm, rate, target_rate)? }
};
// Encode from PCM.
self.encode_from_pcm(&processed, to_pt)
}
fn decode_to_pcm(&mut self, data: &[u8], pt: u8) -> Result<(Vec<i16>, u32), String> {
match pt {
PT_OPUS => {
let mut pcm = vec![0i16; 5760]; // up to 120ms at 48kHz (RFC 6716 max)
let packet = OpusPacket::try_from(data)
.map_err(|e| format!("opus packet: {e}"))?;
let out = MutSignals::try_from(&mut pcm[..])
.map_err(|e| format!("opus signals: {e}"))?;
let n: usize = self.opus_dec.decode(Some(packet), out, false)
.map_err(|e| format!("opus decode: {e}"))?.into();
pcm.truncate(n);
Ok((pcm, 48000))
}
PT_G722 => {
let pcm = self.g722_dec.decode(data);
Ok((pcm, 16000))
}
PT_PCMU => {
let pcm: Vec<i16> = data.iter().map(|&b| mulaw_decode(b)).collect();
Ok((pcm, 8000))
}
PT_PCMA => {
let pcm: Vec<i16> = data.iter().map(|&b| alaw_decode(b)).collect();
Ok((pcm, 8000))
}
_ => Err(format!("unsupported source PT {pt}")),
}
}
fn encode_from_pcm(&mut self, pcm: &[i16], pt: u8) -> Result<Vec<u8>, String> {
match pt {
PT_OPUS => {
let mut buf = vec![0u8; 4000];
let n: usize = self.opus_enc.encode(pcm, &mut buf)
.map_err(|e| format!("opus encode: {e}"))?.into();
buf.truncate(n);
Ok(buf)
}
PT_G722 => {
Ok(self.g722_enc.encode(pcm))
}
PT_PCMU => {
Ok(pcm.iter().map(|&s| mulaw_encode(s)).collect())
}
PT_PCMA => {
Ok(pcm.iter().map(|&s| alaw_encode(s)).collect())
}
_ => Err(format!("unsupported target PT {pt}")),
}
}
}
fn codec_sample_rate(pt: u8) -> u32 {
match pt {
PT_OPUS => 48000,
PT_G722 => 16000,
_ => 8000, // PCMU, PCMA
}
}
// ---------------------------------------------------------------------------
// G.711 µ-law (PCMU)
// ---------------------------------------------------------------------------
fn mulaw_encode(sample: i16) -> u8 {
const BIAS: i16 = 0x84;
const CLIP: i16 = 32635;
let sign = if sample < 0 { 0x80u8 } else { 0 };
// Use i32 to avoid overflow when sample == i16::MIN (-32768).
let mut s = (sample as i32).unsigned_abs().min(CLIP as u32) as i16;
s += BIAS;
let mut exp = 7u8;
let mut mask = 0x4000i16;
while exp > 0 && (s & mask) == 0 { exp -= 1; mask >>= 1; }
let mantissa = ((s >> (exp + 3)) & 0x0f) as u8;
!(sign | (exp << 4) | mantissa)
}
fn mulaw_decode(mulaw: u8) -> i16 {
let v = !mulaw;
let sign = v & 0x80;
let exp = (v >> 4) & 0x07;
let mantissa = v & 0x0f;
let mut sample = (((mantissa as i16) << 4) + 0x84) << exp;
sample -= 0x84;
if sign != 0 { -sample } else { sample }
}
// ---------------------------------------------------------------------------
// G.711 A-law (PCMA)
// ---------------------------------------------------------------------------
fn alaw_encode(sample: i16) -> u8 {
let sign = if sample >= 0 { 0x80u8 } else { 0 };
// Use i32 to avoid overflow when sample == i16::MIN (-32768).
let s = (sample as i32).unsigned_abs().min(32767) as i16;
let mut exp = 7u8;
let mut mask = 0x4000i16;
while exp > 0 && (s & mask) == 0 { exp -= 1; mask >>= 1; }
let mantissa = if exp > 0 { ((s >> (exp + 3)) & 0x0f) as u8 } else { ((s >> 4) & 0x0f) as u8 };
(sign | (exp << 4) | mantissa) ^ 0x55
}
fn alaw_decode(alaw: u8) -> i16 {
let v = alaw ^ 0x55;
let sign = v & 0x80;
let exp = (v >> 4) & 0x07;
let mantissa = v & 0x0f;
let sample = if exp == 0 {
((mantissa as i16) << 4) + 8
} else {
(((mantissa as i16) << 4) + 0x108) << (exp - 1)
};
if sign != 0 { sample } else { -sample }
}
// ---------------------------------------------------------------------------
// Main loop
// ---------------------------------------------------------------------------
/// Resolve a session: if session_id is provided, look it up in the sessions map;
/// otherwise fall back to the default state (backward compat with `init`).
fn get_session<'a>(
sessions: &'a mut HashMap<String, TranscodeState>,
default: &'a mut Option<TranscodeState>,
params: &serde_json::Value,
) -> Option<&'a mut TranscodeState> {
if let Some(sid) = params.get("session_id").and_then(|v| v.as_str()) {
sessions.get_mut(sid)
} else {
default.as_mut()
}
}
fn main() {
let stdin = io::stdin();
let stdout = io::stdout();
let mut out = io::BufWriter::new(stdout.lock());
let _ = writeln!(out, r#"{{"event":"ready","data":{{}}}}"#);
let _ = out.flush();
// Default state for backward-compat `init` (no session_id).
let mut default_state: Option<TranscodeState> = None;
// Per-session codec state for concurrent call isolation.
let mut sessions: HashMap<String, TranscodeState> = HashMap::new();
for line in stdin.lock().lines() {
let line = match line {
Ok(l) if !l.trim().is_empty() => l,
Ok(_) => continue,
Err(_) => break,
};
let req: Request = match serde_json::from_str(&line) {
Ok(r) => r,
Err(e) => {
respond(&mut out, "", false, None, Some(&format!("parse: {e}")));
continue;
}
};
match req.method.as_str() {
// Backward-compat: init the default (shared) session.
"init" => {
match TranscodeState::new() {
Ok(s) => {
default_state = Some(s);
respond(&mut out, &req.id, true, Some(serde_json::json!({})), None);
}
Err(e) => respond(&mut out, &req.id, false, None, Some(&e)),
}
}
// Create an isolated session with its own codec state.
"create_session" => {
let session_id = match req.params.get("session_id").and_then(|v| v.as_str()) {
Some(s) => s.to_string(),
None => { respond(&mut out, &req.id, false, None, Some("missing session_id")); continue; }
};
if sessions.contains_key(&session_id) {
respond(&mut out, &req.id, true, Some(serde_json::json!({})), None);
continue;
}
match TranscodeState::new() {
Ok(s) => {
sessions.insert(session_id, s);
respond(&mut out, &req.id, true, Some(serde_json::json!({})), None);
}
Err(e) => respond(&mut out, &req.id, false, None, Some(&e)),
}
}
// Destroy a session, freeing its codec state.
"destroy_session" => {
let session_id = match req.params.get("session_id").and_then(|v| v.as_str()) {
Some(s) => s,
None => { respond(&mut out, &req.id, false, None, Some("missing session_id")); continue; }
};
sessions.remove(session_id);
respond(&mut out, &req.id, true, Some(serde_json::json!({})), None);
}
// Transcode: uses session_id if provided, else default state.
"transcode" => {
let st = match get_session(&mut sessions, &mut default_state, &req.params) {
Some(s) => s,
None => { respond(&mut out, &req.id, false, None, Some("not initialized (no session or default state)")); continue; }
};
let data_b64 = match req.params.get("data_b64").and_then(|v| v.as_str()) {
Some(s) => s,
None => { respond(&mut out, &req.id, false, None, Some("missing data_b64")); continue; }
};
let from_pt = req.params.get("from_pt").and_then(|v| v.as_u64()).unwrap_or(0) as u8;
let to_pt = req.params.get("to_pt").and_then(|v| v.as_u64()).unwrap_or(0) as u8;
let direction = req.params.get("direction").and_then(|v| v.as_str());
let data = match B64.decode(data_b64) {
Ok(b) => b,
Err(e) => { respond(&mut out, &req.id, false, None, Some(&format!("b64: {e}"))); continue; }
};
match st.transcode(&data, from_pt, to_pt, direction) {
Ok(result) => {
respond(&mut out, &req.id, true, Some(serde_json::json!({ "data_b64": B64.encode(&result) })), None);
}
Err(e) => respond(&mut out, &req.id, false, None, Some(&e)),
}
}
// Encode raw 16-bit PCM to a target codec.
// Params: data_b64 (raw PCM bytes, 16-bit LE), sample_rate (input Hz), to_pt
// Optional: session_id for isolated codec state.
"encode_pcm" => {
let st = match get_session(&mut sessions, &mut default_state, &req.params) {
Some(s) => s,
None => { respond(&mut out, &req.id, false, None, Some("not initialized (no session or default state)")); continue; }
};
let data_b64 = match req.params.get("data_b64").and_then(|v| v.as_str()) {
Some(s) => s,
None => { respond(&mut out, &req.id, false, None, Some("missing data_b64")); continue; }
};
let sample_rate = req.params.get("sample_rate").and_then(|v| v.as_u64()).unwrap_or(22050) as u32;
let to_pt = req.params.get("to_pt").and_then(|v| v.as_u64()).unwrap_or(9) as u8;
let data = match B64.decode(data_b64) {
Ok(b) => b,
Err(e) => { respond(&mut out, &req.id, false, None, Some(&format!("b64: {e}"))); continue; }
};
if data.len() % 2 != 0 {
respond(&mut out, &req.id, false, None, Some("PCM data has odd byte count (expected 16-bit LE samples)"));
continue;
}
// Convert raw bytes to i16 samples.
let pcm: Vec<i16> = data.chunks_exact(2)
.map(|c| i16::from_le_bytes([c[0], c[1]]))
.collect();
// Resample to target codec's sample rate.
let target_rate = codec_sample_rate(to_pt);
let resampled = match st.resample(&pcm, sample_rate, target_rate) {
Ok(r) => r,
Err(e) => { respond(&mut out, &req.id, false, None, Some(&e)); continue; }
};
// Encode to target codec (reuse encode_from_pcm).
match st.encode_from_pcm(&resampled, to_pt) {
Ok(encoded) => {
respond(&mut out, &req.id, true, Some(serde_json::json!({ "data_b64": B64.encode(&encoded) })), None);
}
Err(e) => { respond(&mut out, &req.id, false, None, Some(&e)); continue; }
}
}
// Legacy commands (kept for backward compat).
"encode" | "decode" => {
respond(&mut out, &req.id, false, None, Some("use 'transcode' command instead"));
}
_ => respond(&mut out, &req.id, false, None, Some(&format!("unknown: {}", req.method))),
}
}
}

View File

@@ -1,18 +1,26 @@
[package]
name = "tts-engine"
name = "proxy-engine"
version = "0.1.0"
edition = "2021"
[[bin]]
name = "tts-engine"
name = "proxy-engine"
path = "src/main.rs"
[dependencies]
codec-lib = { path = "../codec-lib" }
sip-proto = { path = "../sip-proto" }
nnnoiseless = "0.5"
tokio = { version = "1", features = ["full"] }
serde = { version = "1", features = ["derive"] }
serde_json = "1"
base64 = "0.22"
regex-lite = "0.1"
webrtc = "0.8"
rand = "0.8"
hound = "3.5"
kokoro-tts = { version = "0.3", default-features = false }
# Pin to rc.11 matching kokoro-tts's expectation; enable vendored TLS to avoid system libssl-dev.
ort = { version = "=2.0.0-rc.11", default-features = false, features = [
"std", "download-binaries", "copy-dylibs", "ndarray",
"tls-native-vendored"
] }
tokio = { version = "1", features = ["rt-multi-thread", "macros"] }
hound = "3.5"

View File

@@ -0,0 +1,240 @@
//! Audio player — reads a WAV file and streams it as RTP packets.
//! Also provides prompt preparation for the leg interaction system.
use crate::rtp::{build_rtp_header, rtp_clock_increment};
use codec_lib::{codec_sample_rate, TranscodeState};
use std::net::SocketAddr;
use std::path::Path;
use std::sync::Arc;
use tokio::net::UdpSocket;
use tokio::time::{self, Duration};
/// Mixing sample rate used by the mixer (must stay in sync with mixer::MIX_RATE).
const MIX_RATE: u32 = 48000;
/// Samples per 20ms frame at the mixing rate.
const MIX_FRAME_SIZE: usize = 960;
/// Play a WAV file as RTP to a destination.
/// Returns when playback is complete.
pub async fn play_wav_file(
file_path: &str,
socket: Arc<UdpSocket>,
dest: SocketAddr,
codec_pt: u8,
ssrc: u32,
) -> Result<u32, String> {
let path = Path::new(file_path);
if !path.exists() {
return Err(format!("WAV file not found: {file_path}"));
}
// Read WAV file.
let mut reader =
hound::WavReader::open(path).map_err(|e| format!("open WAV {file_path}: {e}"))?;
let spec = reader.spec();
let wav_rate = spec.sample_rate;
// Read all samples as i16.
let samples: Vec<i16> = if spec.bits_per_sample == 16 {
reader
.samples::<i16>()
.filter_map(|s| s.ok())
.collect()
} else if spec.bits_per_sample == 32 && spec.sample_format == hound::SampleFormat::Float {
reader
.samples::<f32>()
.filter_map(|s| s.ok())
.map(|s| (s * 32767.0).round().clamp(-32768.0, 32767.0) as i16)
.collect()
} else {
return Err(format!(
"unsupported WAV format: {}bit {:?}",
spec.bits_per_sample, spec.sample_format
));
};
if samples.is_empty() {
return Ok(0);
}
// Create codec state for encoding.
let mut transcoder = TranscodeState::new().map_err(|e| format!("codec init: {e}"))?;
// Resample to target codec rate.
let target_rate = codec_sample_rate(codec_pt);
let resampled = if wav_rate != target_rate {
transcoder
.resample(&samples, wav_rate, target_rate)
.map_err(|e| format!("resample: {e}"))?
} else {
samples
};
// Calculate frame size (20ms of audio at target rate).
let frame_samples = (target_rate as usize) / 50; // 20ms = 1/50 second
// Stream as RTP at 20ms intervals.
let mut seq: u16 = 0;
let mut ts: u32 = 0;
let mut offset = 0;
let mut interval = time::interval(Duration::from_millis(20));
let mut frames_sent = 0u32;
while offset < resampled.len() {
interval.tick().await;
let end = (offset + frame_samples).min(resampled.len());
let frame = &resampled[offset..end];
// Pad short final frame with silence.
let frame_data = if frame.len() < frame_samples {
let mut padded = frame.to_vec();
padded.resize(frame_samples, 0);
padded
} else {
frame.to_vec()
};
// Encode to target codec.
let encoded = match transcoder.encode_from_pcm(&frame_data, codec_pt) {
Ok(e) if !e.is_empty() => e,
_ => {
offset += frame_samples;
continue;
}
};
// Build RTP packet.
let header = build_rtp_header(codec_pt, seq, ts, ssrc);
let mut packet = header.to_vec();
packet.extend_from_slice(&encoded);
let _ = socket.send_to(&packet, dest).await;
seq = seq.wrapping_add(1);
ts = ts.wrapping_add(rtp_clock_increment(codec_pt));
offset += frame_samples;
frames_sent += 1;
}
Ok(frames_sent)
}
/// Generate and play a beep tone (sine wave) as RTP.
pub async fn play_beep(
socket: Arc<UdpSocket>,
dest: SocketAddr,
codec_pt: u8,
ssrc: u32,
start_seq: u16,
start_ts: u32,
freq_hz: u32,
duration_ms: u32,
) -> Result<(u16, u32), String> {
let mut transcoder = TranscodeState::new().map_err(|e| format!("codec init: {e}"))?;
let target_rate = codec_sample_rate(codec_pt);
let frame_samples = (target_rate as usize) / 50;
let total_samples = (target_rate as usize * duration_ms as usize) / 1000;
// Generate sine wave.
let amplitude = 16000i16;
let sine: Vec<i16> = (0..total_samples)
.map(|i| {
let t = i as f64 / target_rate as f64;
(amplitude as f64 * (2.0 * std::f64::consts::PI * freq_hz as f64 * t).sin()) as i16
})
.collect();
let mut seq = start_seq;
let mut ts = start_ts;
let mut offset = 0;
let mut interval = time::interval(Duration::from_millis(20));
while offset < sine.len() {
interval.tick().await;
let end = (offset + frame_samples).min(sine.len());
let mut frame = sine[offset..end].to_vec();
frame.resize(frame_samples, 0);
let encoded = match transcoder.encode_from_pcm(&frame, codec_pt) {
Ok(e) if !e.is_empty() => e,
_ => {
offset += frame_samples;
continue;
}
};
let header = build_rtp_header(codec_pt, seq, ts, ssrc);
let mut packet = header.to_vec();
packet.extend_from_slice(&encoded);
let _ = socket.send_to(&packet, dest).await;
seq = seq.wrapping_add(1);
ts = ts.wrapping_add(rtp_clock_increment(codec_pt));
offset += frame_samples;
}
Ok((seq, ts))
}
/// Load a WAV file and split it into 20ms f32 PCM frames at 48kHz.
/// Used by the leg interaction system to prepare prompt audio for the mixer.
pub fn load_prompt_pcm_frames(wav_path: &str) -> Result<Vec<Vec<f32>>, String> {
let path = Path::new(wav_path);
if !path.exists() {
return Err(format!("WAV file not found: {wav_path}"));
}
let mut reader =
hound::WavReader::open(path).map_err(|e| format!("open WAV {wav_path}: {e}"))?;
let spec = reader.spec();
let wav_rate = spec.sample_rate;
// Read all samples as f32 in [-1.0, 1.0].
let samples: Vec<f32> = if spec.bits_per_sample == 16 {
reader
.samples::<i16>()
.filter_map(|s| s.ok())
.map(|s| s as f32 / 32768.0)
.collect()
} else if spec.bits_per_sample == 32 && spec.sample_format == hound::SampleFormat::Float {
reader
.samples::<f32>()
.filter_map(|s| s.ok())
.collect()
} else {
return Err(format!(
"unsupported WAV format: {}bit {:?}",
spec.bits_per_sample, spec.sample_format
));
};
if samples.is_empty() {
return Ok(vec![]);
}
// Resample to MIX_RATE (48kHz) if needed.
let resampled = if wav_rate != MIX_RATE {
let mut transcoder = TranscodeState::new().map_err(|e| format!("codec init: {e}"))?;
transcoder
.resample_f32(&samples, wav_rate, MIX_RATE)
.map_err(|e| format!("resample: {e}"))?
} else {
samples
};
// Split into MIX_FRAME_SIZE (960) sample frames.
let mut frames = Vec::new();
let mut offset = 0;
while offset < resampled.len() {
let end = (offset + MIX_FRAME_SIZE).min(resampled.len());
let mut frame = resampled[offset..end].to_vec();
// Pad short final frame with silence.
frame.resize(MIX_FRAME_SIZE, 0.0);
frames.push(frame);
offset += MIX_FRAME_SIZE;
}
Ok(frames)
}

View File

@@ -0,0 +1,252 @@
//! Call hub — owns N legs and a mixer task.
//!
//! Every call has a central mixer that provides mix-minus audio to all
//! participants. Legs can be added and removed dynamically mid-call.
use crate::mixer::{MixerCommand, RtpPacket};
use crate::sip_leg::SipLeg;
use sip_proto::message::SipMessage;
use std::collections::HashMap;
use std::net::SocketAddr;
use std::sync::Arc;
use std::time::Instant;
use tokio::net::UdpSocket;
use tokio::sync::mpsc;
use tokio::task::JoinHandle;
pub type LegId = String;
/// Call state machine.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum CallState {
SettingUp,
Ringing,
Connected,
Voicemail,
Terminated,
}
impl CallState {
pub fn as_str(&self) -> &'static str {
match self {
Self::SettingUp => "setting-up",
Self::Ringing => "ringing",
Self::Connected => "connected",
Self::Voicemail => "voicemail",
Self::Terminated => "terminated",
}
}
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum CallDirection {
Inbound,
Outbound,
}
impl CallDirection {
pub fn as_str(&self) -> &'static str {
match self {
Self::Inbound => "inbound",
Self::Outbound => "outbound",
}
}
}
/// The type of a call leg.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum LegKind {
SipProvider,
SipDevice,
WebRtc,
Media, // voicemail playback, IVR, recording
Tool, // observer leg for recording, transcription, etc.
}
impl LegKind {
pub fn as_str(&self) -> &'static str {
match self {
Self::SipProvider => "sip-provider",
Self::SipDevice => "sip-device",
Self::WebRtc => "webrtc",
Self::Media => "media",
Self::Tool => "tool",
}
}
}
/// Per-leg state.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum LegState {
Inviting,
Ringing,
Connected,
Terminated,
}
impl LegState {
pub fn as_str(&self) -> &'static str {
match self {
Self::Inviting => "inviting",
Self::Ringing => "ringing",
Self::Connected => "connected",
Self::Terminated => "terminated",
}
}
}
/// Information about a single leg in a call.
pub struct LegInfo {
pub id: LegId,
pub kind: LegKind,
pub state: LegState,
pub codec_pt: u8,
/// For SIP legs: the SIP dialog manager (handles 407 auth, BYE, etc).
pub sip_leg: Option<SipLeg>,
/// For SIP legs: the SIP Call-ID for message routing.
pub sip_call_id: Option<String>,
/// For WebRTC legs: the session ID in WebRtcEngine.
pub webrtc_session_id: Option<String>,
/// The RTP socket allocated for this leg.
pub rtp_socket: Option<Arc<UdpSocket>>,
/// The RTP port number.
pub rtp_port: u16,
/// The remote media endpoint (learned from SDP or address learning).
pub remote_media: Option<SocketAddr>,
/// SIP signaling address (provider or device).
pub signaling_addr: Option<SocketAddr>,
/// Flexible key-value metadata (consent state, tool config, etc.).
/// Persisted into call history on call end.
pub metadata: HashMap<String, serde_json::Value>,
}
/// A multiparty call with N legs and a central mixer.
pub struct Call {
pub id: String,
pub state: CallState,
pub direction: CallDirection,
pub created_at: Instant,
// Metadata.
pub caller_number: Option<String>,
pub callee_number: Option<String>,
pub provider_id: String,
/// Original INVITE from the device (for device-originated outbound calls).
/// Used to construct proper 180/200/error responses back to the device.
pub device_invite: Option<SipMessage>,
/// All legs in this call, keyed by leg ID.
pub legs: HashMap<LegId, LegInfo>,
/// Channel to send commands to the mixer task.
pub mixer_cmd_tx: mpsc::Sender<MixerCommand>,
/// Handle to the mixer task (aborted on call teardown).
mixer_task: Option<JoinHandle<()>>,
}
impl Call {
pub fn new(
id: String,
direction: CallDirection,
provider_id: String,
mixer_cmd_tx: mpsc::Sender<MixerCommand>,
mixer_task: JoinHandle<()>,
) -> Self {
Self {
id,
state: CallState::SettingUp,
direction,
created_at: Instant::now(),
caller_number: None,
callee_number: None,
provider_id,
device_invite: None,
legs: HashMap::new(),
mixer_cmd_tx,
mixer_task: Some(mixer_task),
}
}
/// Add a leg to the mixer. Sends the AddLeg command with channel endpoints.
pub async fn add_leg_to_mixer(
&self,
leg_id: &str,
codec_pt: u8,
inbound_rx: mpsc::Receiver<RtpPacket>,
outbound_tx: mpsc::Sender<Vec<u8>>,
) {
let _ = self
.mixer_cmd_tx
.send(MixerCommand::AddLeg {
leg_id: leg_id.to_string(),
codec_pt,
inbound_rx,
outbound_tx,
})
.await;
}
/// Remove a leg from the mixer.
pub async fn remove_leg_from_mixer(&self, leg_id: &str) {
let _ = self
.mixer_cmd_tx
.send(MixerCommand::RemoveLeg {
leg_id: leg_id.to_string(),
})
.await;
}
pub fn duration_secs(&self) -> u64 {
self.created_at.elapsed().as_secs()
}
/// Shut down the mixer and abort its task.
pub async fn shutdown_mixer(&mut self) {
let _ = self.mixer_cmd_tx.send(MixerCommand::Shutdown).await;
if let Some(handle) = self.mixer_task.take() {
handle.abort();
}
}
/// Produce a JSON status snapshot for the dashboard.
pub fn to_status_json(&self) -> serde_json::Value {
let legs: Vec<serde_json::Value> = self
.legs
.values()
.filter(|l| l.state != LegState::Terminated)
.map(|l| {
let metadata: serde_json::Value = if l.metadata.is_empty() {
serde_json::json!({})
} else {
serde_json::Value::Object(
l.metadata.iter().map(|(k, v)| (k.clone(), v.clone())).collect(),
)
};
serde_json::json!({
"id": l.id,
"type": l.kind.as_str(),
"state": l.state.as_str(),
"codec": sip_proto::helpers::codec_name(l.codec_pt),
"rtpPort": l.rtp_port,
"remoteMedia": l.remote_media.map(|a| format!("{}:{}", a.ip(), a.port())),
"metadata": metadata,
})
})
.collect();
serde_json::json!({
"id": self.id,
"state": self.state.as_str(),
"direction": self.direction.as_str(),
"callerNumber": self.caller_number,
"calleeNumber": self.callee_number,
"providerUsed": self.provider_id,
"duration": self.duration_secs(),
"legs": legs,
})
}
}

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,325 @@
//! Configuration types received from the TypeScript control plane.
//!
//! TypeScript loads config from `.nogit/config.json` and sends it to the
//! proxy engine via the `configure` command. These types mirror the TS interfaces.
use serde::Deserialize;
use std::net::SocketAddr;
/// Network endpoint.
#[derive(Debug, Clone, Deserialize)]
pub struct Endpoint {
pub address: String,
pub port: u16,
}
impl Endpoint {
/// Resolve to a SocketAddr. Handles both IP addresses and hostnames.
pub fn to_socket_addr(&self) -> Option<SocketAddr> {
// Try direct parse first (IP address).
if let Ok(addr) = format!("{}:{}", self.address, self.port).parse() {
return Some(addr);
}
// DNS resolution for hostnames.
use std::net::ToSocketAddrs;
format!("{}:{}", self.address, self.port)
.to_socket_addrs()
.ok()
.and_then(|mut addrs| addrs.next())
}
}
/// Provider quirks for codec/protocol workarounds.
#[derive(Debug, Clone, Deserialize)]
pub struct Quirks {
#[serde(rename = "earlyMediaSilence")]
pub early_media_silence: bool,
#[serde(rename = "silencePayloadType")]
pub silence_payload_type: Option<u8>,
#[serde(rename = "silenceMaxPackets")]
pub silence_max_packets: Option<u32>,
}
/// A SIP trunk provider configuration.
#[derive(Debug, Clone, Deserialize)]
pub struct ProviderConfig {
pub id: String,
#[serde(rename = "displayName")]
pub display_name: String,
pub domain: String,
#[serde(rename = "outboundProxy")]
pub outbound_proxy: Endpoint,
pub username: String,
pub password: String,
#[serde(rename = "registerIntervalSec")]
pub register_interval_sec: u32,
pub codecs: Vec<u8>,
pub quirks: Quirks,
}
/// A SIP device (phone) configuration.
#[derive(Debug, Clone, Deserialize)]
pub struct DeviceConfig {
pub id: String,
#[serde(rename = "displayName")]
pub display_name: String,
#[serde(rename = "expectedAddress")]
pub expected_address: String,
pub extension: String,
}
/// Route match criteria.
#[derive(Debug, Clone, Deserialize)]
pub struct RouteMatch {
pub direction: String, // "inbound" | "outbound"
#[serde(rename = "numberPattern")]
pub number_pattern: Option<String>,
#[serde(rename = "callerPattern")]
pub caller_pattern: Option<String>,
#[serde(rename = "sourceProvider")]
pub source_provider: Option<String>,
#[serde(rename = "sourceDevice")]
pub source_device: Option<String>,
}
/// Route action.
#[derive(Debug, Clone, Deserialize)]
pub struct RouteAction {
pub targets: Option<Vec<String>>,
#[serde(rename = "ringBrowsers")]
pub ring_browsers: Option<bool>,
#[serde(rename = "voicemailBox")]
pub voicemail_box: Option<String>,
#[serde(rename = "ivrMenuId")]
pub ivr_menu_id: Option<String>,
#[serde(rename = "noAnswerTimeout")]
pub no_answer_timeout: Option<u32>,
pub provider: Option<String>,
#[serde(rename = "failoverProviders")]
pub failover_providers: Option<Vec<String>>,
#[serde(rename = "stripPrefix")]
pub strip_prefix: Option<String>,
#[serde(rename = "prependPrefix")]
pub prepend_prefix: Option<String>,
}
/// A routing rule.
#[derive(Debug, Clone, Deserialize)]
pub struct Route {
pub id: String,
pub name: String,
pub priority: i32,
pub enabled: bool,
#[serde(rename = "match")]
pub match_criteria: RouteMatch,
pub action: RouteAction,
}
/// Proxy network settings.
#[derive(Debug, Clone, Deserialize)]
pub struct ProxyConfig {
#[serde(rename = "lanIp")]
pub lan_ip: String,
#[serde(rename = "lanPort")]
pub lan_port: u16,
#[serde(rename = "publicIpSeed")]
pub public_ip_seed: Option<String>,
#[serde(rename = "rtpPortRange")]
pub rtp_port_range: RtpPortRange,
}
#[derive(Debug, Clone, Deserialize)]
pub struct RtpPortRange {
pub min: u16,
pub max: u16,
}
/// Full application config pushed from TypeScript.
#[derive(Debug, Clone, Deserialize)]
pub struct AppConfig {
pub proxy: ProxyConfig,
pub providers: Vec<ProviderConfig>,
pub devices: Vec<DeviceConfig>,
pub routing: RoutingConfig,
}
#[derive(Debug, Clone, Deserialize)]
pub struct RoutingConfig {
pub routes: Vec<Route>,
}
// ---------------------------------------------------------------------------
// Pattern matching (ported from ts/config.ts)
// ---------------------------------------------------------------------------
/// Test a value against a pattern string.
/// - None/empty: matches everything (wildcard)
/// - Trailing '*': prefix match
/// - Starts with '/': regex match
/// - Otherwise: exact match
pub fn matches_pattern(pattern: Option<&str>, value: &str) -> bool {
let pattern = match pattern {
None => return true,
Some(p) if p.is_empty() => return true,
Some(p) => p,
};
// Prefix match: "+49*"
if pattern.ends_with('*') {
return value.starts_with(&pattern[..pattern.len() - 1]);
}
// Regex match: "/^\\+49/" or "/pattern/i"
if pattern.starts_with('/') {
if let Some(last_slash) = pattern[1..].rfind('/') {
let re_str = &pattern[1..1 + last_slash];
let flags = &pattern[2 + last_slash..];
let case_insensitive = flags.contains('i');
if let Ok(re) = if case_insensitive {
regex_lite::Regex::new(&format!("(?i){re_str}"))
} else {
regex_lite::Regex::new(re_str)
} {
return re.is_match(value);
}
}
}
// Exact match.
value == pattern
}
/// Result of resolving an outbound route.
pub struct OutboundRouteResult {
pub provider: ProviderConfig,
pub transformed_number: String,
}
/// Result of resolving an inbound route.
pub struct InboundRouteResult {
pub device_ids: Vec<String>,
pub ring_browsers: bool,
pub voicemail_box: Option<String>,
pub ivr_menu_id: Option<String>,
pub no_answer_timeout: Option<u32>,
}
impl AppConfig {
/// Resolve which provider to use for an outbound call.
pub fn resolve_outbound_route(
&self,
dialed_number: &str,
source_device_id: Option<&str>,
is_provider_registered: &dyn Fn(&str) -> bool,
) -> Option<OutboundRouteResult> {
let mut routes: Vec<&Route> = self
.routing
.routes
.iter()
.filter(|r| r.enabled && r.match_criteria.direction == "outbound")
.collect();
routes.sort_by(|a, b| b.priority.cmp(&a.priority));
for route in &routes {
let m = &route.match_criteria;
if !matches_pattern(m.number_pattern.as_deref(), dialed_number) {
continue;
}
if let Some(sd) = &m.source_device {
if source_device_id != Some(sd.as_str()) {
continue;
}
}
// Find a registered provider.
let mut candidates: Vec<&str> = Vec::new();
if let Some(p) = &route.action.provider {
candidates.push(p);
}
if let Some(fps) = &route.action.failover_providers {
candidates.extend(fps.iter().map(|s| s.as_str()));
}
for pid in candidates {
let provider = match self.providers.iter().find(|p| p.id == pid) {
Some(p) => p,
None => continue,
};
if !is_provider_registered(pid) {
continue;
}
let mut num = dialed_number.to_string();
if let Some(strip) = &route.action.strip_prefix {
if num.starts_with(strip.as_str()) {
num = num[strip.len()..].to_string();
}
}
if let Some(prepend) = &route.action.prepend_prefix {
num = format!("{prepend}{num}");
}
return Some(OutboundRouteResult {
provider: provider.clone(),
transformed_number: num,
});
}
}
// Fallback: first provider.
self.providers.first().map(|p| OutboundRouteResult {
provider: p.clone(),
transformed_number: dialed_number.to_string(),
})
}
/// Resolve which devices to ring for an inbound call.
pub fn resolve_inbound_route(
&self,
provider_id: &str,
called_number: &str,
caller_number: &str,
) -> InboundRouteResult {
let mut routes: Vec<&Route> = self
.routing
.routes
.iter()
.filter(|r| r.enabled && r.match_criteria.direction == "inbound")
.collect();
routes.sort_by(|a, b| b.priority.cmp(&a.priority));
for route in &routes {
let m = &route.match_criteria;
if let Some(sp) = &m.source_provider {
if sp != provider_id {
continue;
}
}
if !matches_pattern(m.number_pattern.as_deref(), called_number) {
continue;
}
if !matches_pattern(m.caller_pattern.as_deref(), caller_number) {
continue;
}
return InboundRouteResult {
device_ids: route.action.targets.clone().unwrap_or_default(),
ring_browsers: route.action.ring_browsers.unwrap_or(false),
voicemail_box: route.action.voicemail_box.clone(),
ivr_menu_id: route.action.ivr_menu_id.clone(),
no_answer_timeout: route.action.no_answer_timeout,
};
}
// Fallback: ring all devices + browsers.
InboundRouteResult {
device_ids: vec![],
ring_browsers: true,
voicemail_box: None,
ivr_menu_id: None,
no_answer_timeout: None,
}
}
}

View File

@@ -0,0 +1,200 @@
//! DTMF detection — parses RFC 2833 telephone-event RTP packets.
//!
//! Deduplicates repeated packets (same digit sent multiple times with
//! increasing duration) and fires once per detected digit.
//!
//! Ported from ts/call/dtmf-detector.ts.
use crate::ipc::{emit_event, OutTx};
/// RFC 2833 event ID → character mapping.
const EVENT_CHARS: &[char] = &[
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '*', '#', 'A', 'B', 'C', 'D',
];
/// Safety timeout: report digit if no End packet arrives within this many ms.
const SAFETY_TIMEOUT_MS: u64 = 200;
/// DTMF detector for a single RTP stream.
pub struct DtmfDetector {
/// Negotiated telephone-event payload type (default 101).
telephone_event_pt: u8,
/// Clock rate for duration calculation (default 8000 Hz).
clock_rate: u32,
/// Call ID for event emission.
call_id: String,
// Deduplication state.
current_event_id: Option<u8>,
current_event_ts: Option<u32>,
current_event_reported: bool,
current_event_duration: u16,
out_tx: OutTx,
}
impl DtmfDetector {
pub fn new(call_id: String, out_tx: OutTx) -> Self {
Self {
telephone_event_pt: 101,
clock_rate: 8000,
call_id,
current_event_id: None,
current_event_ts: None,
current_event_reported: false,
current_event_duration: 0,
out_tx,
}
}
/// Feed an RTP packet. Checks PT; ignores non-DTMF packets.
/// Returns Some(digit_char) if a digit was detected.
pub fn process_rtp(&mut self, data: &[u8]) -> Option<char> {
if data.len() < 16 {
return None; // 12-byte header + 4-byte telephone-event minimum
}
let pt = data[1] & 0x7F;
if pt != self.telephone_event_pt {
return None;
}
let marker = (data[1] & 0x80) != 0;
let rtp_timestamp = u32::from_be_bytes([data[4], data[5], data[6], data[7]]);
// Parse telephone-event payload.
let event_id = data[12];
let end_bit = (data[13] & 0x80) != 0;
let duration = u16::from_be_bytes([data[14], data[15]]);
if event_id as usize >= EVENT_CHARS.len() {
return None;
}
// Detect new event.
let is_new = marker
|| self.current_event_id != Some(event_id)
|| self.current_event_ts != Some(rtp_timestamp);
if is_new {
// Report pending unreported event.
let pending = self.report_pending();
self.current_event_id = Some(event_id);
self.current_event_ts = Some(rtp_timestamp);
self.current_event_reported = false;
self.current_event_duration = duration;
if pending.is_some() {
return pending;
}
}
if duration > self.current_event_duration {
self.current_event_duration = duration;
}
// Report on End bit (first time only).
if end_bit && !self.current_event_reported {
self.current_event_reported = true;
let digit = EVENT_CHARS[event_id as usize];
let duration_ms = (self.current_event_duration as f64 / self.clock_rate as f64) * 1000.0;
emit_event(
&self.out_tx,
"dtmf_digit",
serde_json::json!({
"call_id": self.call_id,
"digit": digit.to_string(),
"duration_ms": duration_ms.round() as u32,
"source": "rfc2833",
}),
);
return Some(digit);
}
None
}
/// Report a pending unreported event.
fn report_pending(&mut self) -> Option<char> {
if let Some(event_id) = self.current_event_id {
if !self.current_event_reported && (event_id as usize) < EVENT_CHARS.len() {
self.current_event_reported = true;
let digit = EVENT_CHARS[event_id as usize];
let duration_ms =
(self.current_event_duration as f64 / self.clock_rate as f64) * 1000.0;
emit_event(
&self.out_tx,
"dtmf_digit",
serde_json::json!({
"call_id": self.call_id,
"digit": digit.to_string(),
"duration_ms": duration_ms.round() as u32,
"source": "rfc2833",
}),
);
return Some(digit);
}
}
None
}
/// Process a SIP INFO message body for DTMF.
pub fn process_sip_info(&mut self, content_type: &str, body: &str) -> Option<char> {
let ct = content_type.to_ascii_lowercase();
if ct.contains("application/dtmf-relay") {
// Format: "Signal= 5\r\nDuration= 160\r\n"
let signal = body
.lines()
.find(|l| l.to_ascii_lowercase().starts_with("signal"))
.and_then(|l| l.split('=').nth(1))
.map(|s| s.trim().to_string())?;
if signal.len() != 1 {
return None;
}
let digit = signal.chars().next()?.to_ascii_uppercase();
if !"0123456789*#ABCD".contains(digit) {
return None;
}
emit_event(
&self.out_tx,
"dtmf_digit",
serde_json::json!({
"call_id": self.call_id,
"digit": digit.to_string(),
"source": "sip-info",
}),
);
return Some(digit);
}
if ct.contains("application/dtmf") {
let digit = body.trim().chars().next()?.to_ascii_uppercase();
if !"0123456789*#ABCD".contains(digit) {
return None;
}
emit_event(
&self.out_tx,
"dtmf_digit",
serde_json::json!({
"call_id": self.call_id,
"digit": digit.to_string(),
"source": "sip-info",
}),
);
return Some(digit);
}
None
}
}

View File

@@ -0,0 +1,47 @@
//! IPC protocol — command dispatch and event emission.
//!
//! All communication with the TypeScript control plane goes through
//! JSON-line messages on stdin/stdout (smartrust protocol).
use serde::Deserialize;
use tokio::sync::mpsc;
/// Sender for serialized stdout output.
pub type OutTx = mpsc::UnboundedSender<String>;
/// A command received from the TypeScript control plane.
#[derive(Deserialize)]
pub struct Command {
pub id: String,
pub method: String,
#[serde(default)]
pub params: serde_json::Value,
}
/// Send a response to a command.
pub fn respond(tx: &OutTx, id: &str, success: bool, result: Option<serde_json::Value>, error: Option<&str>) {
let mut resp = serde_json::json!({ "id": id, "success": success });
if let Some(r) = result {
resp["result"] = r;
}
if let Some(e) = error {
resp["error"] = serde_json::Value::String(e.to_string());
}
let _ = tx.send(resp.to_string());
}
/// Send a success response.
pub fn respond_ok(tx: &OutTx, id: &str, result: serde_json::Value) {
respond(tx, id, true, Some(result), None);
}
/// Send an error response.
pub fn respond_err(tx: &OutTx, id: &str, error: &str) {
respond(tx, id, false, None, Some(error));
}
/// Emit an event to the TypeScript control plane.
pub fn emit_event(tx: &OutTx, event: &str, data: serde_json::Value) {
let msg = serde_json::json!({ "event": event, "data": data });
let _ = tx.send(msg.to_string());
}

View File

@@ -0,0 +1,100 @@
//! Leg I/O task spawners.
//!
//! Each SIP leg gets two tasks:
//! - Inbound: recv_from on RTP socket → strip header → send RtpPacket to mixer channel
//! - Outbound: recv encoded RTP from mixer channel → send_to remote media endpoint
//!
//! WebRTC leg I/O is handled inside webrtc_engine.rs (on_track + track.write).
use crate::mixer::RtpPacket;
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
use tokio::sync::mpsc;
/// Channel pair for connecting a leg to the mixer.
pub struct LegChannels {
/// Mixer receives decoded packets from this leg.
pub inbound_tx: mpsc::Sender<RtpPacket>,
pub inbound_rx: mpsc::Receiver<RtpPacket>,
/// Mixer sends encoded RTP to this leg.
pub outbound_tx: mpsc::Sender<Vec<u8>>,
pub outbound_rx: mpsc::Receiver<Vec<u8>>,
}
/// Create a channel pair for a leg.
pub fn create_leg_channels() -> LegChannels {
let (inbound_tx, inbound_rx) = mpsc::channel::<RtpPacket>(64);
let (outbound_tx, outbound_rx) = mpsc::channel::<Vec<u8>>(8);
LegChannels {
inbound_tx,
inbound_rx,
outbound_tx,
outbound_rx,
}
}
/// Spawn the inbound I/O task for a SIP leg.
/// Reads RTP from the socket, parses the variable-length header (RFC 3550),
/// and sends the payload to the mixer.
/// Returns the JoinHandle (exits when the inbound_tx channel is dropped).
pub fn spawn_sip_inbound(
rtp_socket: Arc<UdpSocket>,
inbound_tx: mpsc::Sender<RtpPacket>,
) -> tokio::task::JoinHandle<()> {
tokio::spawn(async move {
let mut buf = vec![0u8; 1500];
loop {
match rtp_socket.recv_from(&mut buf).await {
Ok((n, _from)) => {
if n < 12 {
continue; // Too small for RTP header.
}
let pt = buf[1] & 0x7F;
let marker = (buf[1] & 0x80) != 0;
let seq = u16::from_be_bytes([buf[2], buf[3]]);
let timestamp = u32::from_be_bytes([buf[4], buf[5], buf[6], buf[7]]);
// RFC 3550: header length = 12 + (CC * 4) + optional extension.
let cc = (buf[0] & 0x0F) as usize;
let has_extension = (buf[0] & 0x10) != 0;
let mut offset = 12 + cc * 4;
if has_extension {
if offset + 4 > n {
continue; // Malformed: extension header truncated.
}
let ext_len = u16::from_be_bytes([buf[offset + 2], buf[offset + 3]]) as usize;
offset += 4 + ext_len * 4;
}
if offset >= n {
continue; // No payload after header.
}
let payload = buf[offset..n].to_vec();
if payload.is_empty() {
continue;
}
if inbound_tx.send(RtpPacket { payload, payload_type: pt, marker, seq, timestamp }).await.is_err() {
break; // Channel closed — leg removed.
}
}
Err(_) => break, // Socket error.
}
}
})
}
/// Spawn the outbound I/O task for a SIP leg.
/// Reads encoded RTP packets from the mixer and sends them to the remote media endpoint.
/// Returns the JoinHandle (exits when the outbound_rx channel is closed).
pub fn spawn_sip_outbound(
rtp_socket: Arc<UdpSocket>,
remote_media: SocketAddr,
mut outbound_rx: mpsc::Receiver<Vec<u8>>,
) -> tokio::task::JoinHandle<()> {
tokio::spawn(async move {
while let Some(rtp_data) = outbound_rx.recv().await {
let _ = rtp_socket.send_to(&rtp_data, remote_media).await;
}
})
}

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,605 @@
//! Audio mixer — mix-minus engine for multiparty calls.
//!
//! Each Call spawns one mixer task. Legs communicate with the mixer via
//! tokio mpsc channels — no shared mutable state, no lock contention.
//!
//! Internal bus format: 48kHz f32 PCM (960 samples per 20ms frame).
//! All encoding/decoding happens at leg boundaries. Per-leg inbound denoising at 48kHz.
//!
//! The mixer runs a 20ms tick loop:
//! 1. Drain inbound channels, decode to f32, resample to 48kHz, denoise per-leg
//! 2. Compute total mix (sum of all **participant** legs' f32 PCM as f64)
//! 3. For each participant leg: mix-minus = total - own, resample to leg codec rate, encode, send
//! 4. For each isolated leg: play prompt frame or silence, check DTMF
//! 5. For each tool leg: send per-source unmerged audio batch
//! 6. Forward DTMF between participant legs only
use crate::ipc::{emit_event, OutTx};
use crate::rtp::{build_rtp_header, rtp_clock_increment};
use codec_lib::{codec_sample_rate, new_denoiser, TranscodeState};
use nnnoiseless::DenoiseState;
use std::collections::{HashMap, VecDeque};
use tokio::sync::{mpsc, oneshot};
use tokio::task::JoinHandle;
use tokio::time::{self, Duration, MissedTickBehavior};
/// Mixing sample rate — 48kHz. Opus is native, G.722 needs 3× upsample, G.711 needs 6× upsample.
/// All processing (denoising, mixing) happens at this rate in f32 for maximum quality.
const MIX_RATE: u32 = 48000;
/// Samples per 20ms frame at the mixing rate.
const MIX_FRAME_SIZE: usize = 960; // 48000 * 0.020
/// A raw RTP payload received from a leg (no RTP header).
pub struct RtpPacket {
pub payload: Vec<u8>,
pub payload_type: u8,
/// RTP marker bit (first packet of a DTMF event, etc.).
pub marker: bool,
/// RTP sequence number for reordering.
pub seq: u16,
/// RTP timestamp from the original packet header.
pub timestamp: u32,
}
// ---------------------------------------------------------------------------
// Leg roles
// ---------------------------------------------------------------------------
/// What role a leg currently plays in the mixer.
enum LegRole {
/// Normal participant: contributes to mix, receives mix-minus.
Participant,
/// Temporarily isolated for IVR/consent interaction.
Isolated(IsolationState),
}
struct IsolationState {
/// PCM frames at MIX_RATE (960 samples each, 48kHz f32) queued for playback.
prompt_frames: VecDeque<Vec<f32>>,
/// Digits that complete the interaction (e.g., ['1', '2']).
expected_digits: Vec<char>,
/// Ticks remaining before timeout (decremented each tick after prompt ends).
timeout_ticks_remaining: u32,
/// Whether we've finished playing the prompt.
prompt_done: bool,
/// Channel to send the result back to the command handler.
result_tx: Option<oneshot::Sender<InteractionResult>>,
}
/// Result of a leg interaction (consent prompt, IVR, etc.).
pub enum InteractionResult {
/// The participant pressed one of the expected digits.
Digit(char),
/// No digit was received within the timeout.
Timeout,
/// The leg was removed or the call tore down before completion.
Cancelled,
}
// ---------------------------------------------------------------------------
// Tool legs
// ---------------------------------------------------------------------------
/// Type of tool leg.
#[derive(Debug, Clone, Copy)]
pub enum ToolType {
Recording,
Transcription,
}
/// Per-source audio delivered to a tool leg each mixer tick.
pub struct ToolAudioBatch {
pub sources: Vec<ToolAudioSource>,
}
/// One participant's 20ms audio frame.
pub struct ToolAudioSource {
pub leg_id: String,
/// PCM at 48kHz f32, MIX_FRAME_SIZE (960) samples.
pub pcm_48k: Vec<f32>,
}
/// Internal storage for a tool leg inside the mixer.
struct ToolLegSlot {
#[allow(dead_code)]
tool_type: ToolType,
audio_tx: mpsc::Sender<ToolAudioBatch>,
}
// ---------------------------------------------------------------------------
// Commands
// ---------------------------------------------------------------------------
/// Commands sent to the mixer task via a control channel.
pub enum MixerCommand {
/// Add a new participant leg to the mix.
AddLeg {
leg_id: String,
codec_pt: u8,
inbound_rx: mpsc::Receiver<RtpPacket>,
outbound_tx: mpsc::Sender<Vec<u8>>,
},
/// Remove a leg from the mix (channels are dropped, I/O tasks exit).
RemoveLeg { leg_id: String },
/// Shut down the mixer.
Shutdown,
/// Isolate a leg and start an interaction (consent prompt, IVR).
/// The leg is removed from the mix and hears the prompt instead.
/// DTMF from the leg is checked against expected_digits.
StartInteraction {
leg_id: String,
/// PCM frames at MIX_RATE (48kHz f32), each 960 samples.
prompt_pcm_frames: Vec<Vec<f32>>,
expected_digits: Vec<char>,
timeout_ms: u32,
result_tx: oneshot::Sender<InteractionResult>,
},
/// Cancel an in-progress interaction (e.g., leg being removed).
CancelInteraction { leg_id: String },
/// Add a tool leg that receives per-source unmerged audio.
AddToolLeg {
leg_id: String,
tool_type: ToolType,
audio_tx: mpsc::Sender<ToolAudioBatch>,
},
/// Remove a tool leg (drops the channel, background task finalizes).
RemoveToolLeg { leg_id: String },
}
// ---------------------------------------------------------------------------
// Mixer internals
// ---------------------------------------------------------------------------
/// Internal per-leg state inside the mixer.
struct MixerLegSlot {
codec_pt: u8,
transcoder: TranscodeState,
/// Per-leg inbound denoiser (48kHz, 480-sample frames).
denoiser: Box<DenoiseState<'static>>,
inbound_rx: mpsc::Receiver<RtpPacket>,
outbound_tx: mpsc::Sender<Vec<u8>>,
/// Last decoded+denoised PCM frame at MIX_RATE (960 samples, 48kHz f32).
last_pcm_frame: Vec<f32>,
/// Number of consecutive ticks with no inbound packet.
silent_ticks: u32,
// RTP output state.
rtp_seq: u16,
rtp_ts: u32,
rtp_ssrc: u32,
/// Current role of this leg in the mixer.
role: LegRole,
}
/// Spawn the mixer task for a call. Returns the command sender and task handle.
pub fn spawn_mixer(
call_id: String,
out_tx: OutTx,
) -> (mpsc::Sender<MixerCommand>, JoinHandle<()>) {
let (cmd_tx, cmd_rx) = mpsc::channel::<MixerCommand>(32);
let handle = tokio::spawn(async move {
mixer_loop(call_id, cmd_rx, out_tx).await;
});
(cmd_tx, handle)
}
/// The 20ms mixing loop.
async fn mixer_loop(
call_id: String,
mut cmd_rx: mpsc::Receiver<MixerCommand>,
out_tx: OutTx,
) {
let mut legs: HashMap<String, MixerLegSlot> = HashMap::new();
let mut tool_legs: HashMap<String, ToolLegSlot> = HashMap::new();
let mut interval = time::interval(Duration::from_millis(20));
interval.set_missed_tick_behavior(MissedTickBehavior::Skip);
loop {
interval.tick().await;
// ── 1. Process control commands (non-blocking). ─────────────
loop {
match cmd_rx.try_recv() {
Ok(MixerCommand::AddLeg {
leg_id,
codec_pt,
inbound_rx,
outbound_tx,
}) => {
let transcoder = match TranscodeState::new() {
Ok(t) => t,
Err(e) => {
emit_event(
&out_tx,
"mixer_error",
serde_json::json!({
"call_id": call_id,
"leg_id": leg_id,
"error": format!("codec init: {e}"),
}),
);
continue;
}
};
legs.insert(
leg_id,
MixerLegSlot {
codec_pt,
transcoder,
denoiser: new_denoiser(),
inbound_rx,
outbound_tx,
last_pcm_frame: vec![0.0f32; MIX_FRAME_SIZE],
silent_ticks: 0,
rtp_seq: 0,
rtp_ts: 0,
rtp_ssrc: rand::random(),
role: LegRole::Participant,
},
);
}
Ok(MixerCommand::RemoveLeg { leg_id }) => {
// If the leg is isolated, send Cancelled before dropping.
if let Some(slot) = legs.get_mut(&leg_id) {
if let LegRole::Isolated(ref mut state) = slot.role {
if let Some(tx) = state.result_tx.take() {
let _ = tx.send(InteractionResult::Cancelled);
}
}
}
legs.remove(&leg_id);
// Channels drop → I/O tasks exit cleanly.
}
Ok(MixerCommand::Shutdown) => {
// Cancel all outstanding interactions before shutting down.
for slot in legs.values_mut() {
if let LegRole::Isolated(ref mut state) = slot.role {
if let Some(tx) = state.result_tx.take() {
let _ = tx.send(InteractionResult::Cancelled);
}
}
}
return;
}
Ok(MixerCommand::StartInteraction {
leg_id,
prompt_pcm_frames,
expected_digits,
timeout_ms,
result_tx,
}) => {
if let Some(slot) = legs.get_mut(&leg_id) {
// Cancel any existing interaction first.
if let LegRole::Isolated(ref mut old_state) = slot.role {
if let Some(tx) = old_state.result_tx.take() {
let _ = tx.send(InteractionResult::Cancelled);
}
}
let timeout_ticks = timeout_ms / 20;
slot.role = LegRole::Isolated(IsolationState {
prompt_frames: VecDeque::from(prompt_pcm_frames),
expected_digits,
timeout_ticks_remaining: timeout_ticks,
prompt_done: false,
result_tx: Some(result_tx),
});
} else {
// Leg not found — immediately cancel.
let _ = result_tx.send(InteractionResult::Cancelled);
}
}
Ok(MixerCommand::CancelInteraction { leg_id }) => {
if let Some(slot) = legs.get_mut(&leg_id) {
if let LegRole::Isolated(ref mut state) = slot.role {
if let Some(tx) = state.result_tx.take() {
let _ = tx.send(InteractionResult::Cancelled);
}
}
slot.role = LegRole::Participant;
}
}
Ok(MixerCommand::AddToolLeg {
leg_id,
tool_type,
audio_tx,
}) => {
tool_legs.insert(leg_id, ToolLegSlot { tool_type, audio_tx });
}
Ok(MixerCommand::RemoveToolLeg { leg_id }) => {
tool_legs.remove(&leg_id);
// Dropping the ToolLegSlot drops audio_tx → background task sees channel close.
}
Err(mpsc::error::TryRecvError::Empty) => break,
Err(mpsc::error::TryRecvError::Disconnected) => return,
}
}
if legs.is_empty() && tool_legs.is_empty() {
continue;
}
// ── 2. Drain inbound packets, decode to 48kHz f32 PCM. ────
// DTMF (PT 101) packets are collected separately.
// Audio packets are sorted by sequence number and decoded
// in order to maintain codec state (critical for G.722 ADPCM).
let leg_ids: Vec<String> = legs.keys().cloned().collect();
let mut dtmf_forward: Vec<(String, RtpPacket)> = Vec::new();
for lid in &leg_ids {
let slot = legs.get_mut(lid).unwrap();
// Drain channel — collect DTMF separately, collect ALL audio packets.
let mut audio_packets: Vec<RtpPacket> = Vec::new();
loop {
match slot.inbound_rx.try_recv() {
Ok(pkt) => {
if pkt.payload_type == 101 {
// DTMF telephone-event: collect for processing.
dtmf_forward.push((lid.clone(), pkt));
} else {
audio_packets.push(pkt);
}
}
Err(_) => break,
}
}
if !audio_packets.is_empty() {
slot.silent_ticks = 0;
// Sort by sequence number for correct codec state progression.
// This prevents G.722 ADPCM state corruption from out-of-order packets.
audio_packets.sort_by_key(|p| p.seq);
// Decode ALL packets in order (maintains codec state),
// but only keep the last decoded frame for mixing.
for pkt in &audio_packets {
match slot.transcoder.decode_to_f32(&pkt.payload, pkt.payload_type) {
Ok((pcm, rate)) => {
// Resample to 48kHz mixing rate if needed.
let pcm_48k = if rate == MIX_RATE {
pcm
} else {
slot.transcoder
.resample_f32(&pcm, rate, MIX_RATE)
.unwrap_or_else(|_| vec![0.0f32; MIX_FRAME_SIZE])
};
// Per-leg inbound denoising at 48kHz.
// Only for SIP telephony legs — WebRTC browsers
// already apply noise suppression via getUserMedia.
let processed = if slot.codec_pt != codec_lib::PT_OPUS {
TranscodeState::denoise_f32(&mut slot.denoiser, &pcm_48k)
} else {
pcm_48k
};
// Pad or truncate to exactly MIX_FRAME_SIZE.
let mut frame = processed;
frame.resize(MIX_FRAME_SIZE, 0.0);
slot.last_pcm_frame = frame;
}
Err(_) => {}
}
}
} else if dtmf_forward.iter().any(|(src, _)| src == lid) {
// Got DTMF but no audio — don't bump silent_ticks (DTMF counts as activity).
slot.silent_ticks = 0;
} else {
slot.silent_ticks += 1;
// After 150 ticks (3 seconds) of silence, zero out to avoid stale audio.
if slot.silent_ticks > 150 {
slot.last_pcm_frame = vec![0.0f32; MIX_FRAME_SIZE];
}
}
}
// ── 3. Compute total mix from PARTICIPANT legs only. ────────
// Accumulate as f64 to prevent precision loss when summing f32.
let mut total_mix = vec![0.0f64; MIX_FRAME_SIZE];
for slot in legs.values() {
if matches!(slot.role, LegRole::Participant) {
for (i, &s) in slot.last_pcm_frame.iter().enumerate().take(MIX_FRAME_SIZE) {
total_mix[i] += s as f64;
}
}
}
// ── 4. Per-leg output. ──────────────────────────────────────
// Collect interaction completions to apply after the loop
// (can't mutate role while iterating mutably for encode).
let mut completed_interactions: Vec<(String, InteractionResult)> = Vec::new();
for (lid, slot) in legs.iter_mut() {
match &mut slot.role {
LegRole::Participant => {
// Mix-minus: total minus this leg's own contribution, clamped to [-1.0, 1.0].
let mut mix_minus = Vec::with_capacity(MIX_FRAME_SIZE);
for i in 0..MIX_FRAME_SIZE {
let sample =
(total_mix[i] - slot.last_pcm_frame[i] as f64) as f32;
mix_minus.push(sample.clamp(-1.0, 1.0));
}
// Resample from 48kHz to the leg's codec native rate.
let target_rate = codec_sample_rate(slot.codec_pt);
let resampled = if target_rate == MIX_RATE {
mix_minus
} else {
slot.transcoder
.resample_f32(&mix_minus, MIX_RATE, target_rate)
.unwrap_or_default()
};
// Encode to the leg's codec (f32 → i16 → codec inside encode_from_f32).
let encoded =
match slot.transcoder.encode_from_f32(&resampled, slot.codec_pt) {
Ok(e) if !e.is_empty() => e,
_ => continue,
};
// Build RTP packet with header.
let header =
build_rtp_header(slot.codec_pt, slot.rtp_seq, slot.rtp_ts, slot.rtp_ssrc);
let mut rtp = header.to_vec();
rtp.extend_from_slice(&encoded);
slot.rtp_seq = slot.rtp_seq.wrapping_add(1);
slot.rtp_ts = slot.rtp_ts.wrapping_add(rtp_clock_increment(slot.codec_pt));
// Non-blocking send — drop frame if channel is full.
let _ = slot.outbound_tx.try_send(rtp);
}
LegRole::Isolated(state) => {
// Check for DTMF digit from this leg.
let mut matched_digit: Option<char> = None;
for (src_lid, dtmf_pkt) in &dtmf_forward {
if src_lid == lid && dtmf_pkt.payload.len() >= 4 {
let event_id = dtmf_pkt.payload[0];
let end_bit = (dtmf_pkt.payload[1] & 0x80) != 0;
if end_bit {
const EVENT_CHARS: &[char] = &[
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '*', '#',
'A', 'B', 'C', 'D',
];
if let Some(&ch) = EVENT_CHARS.get(event_id as usize) {
if state.expected_digits.contains(&ch) {
matched_digit = Some(ch);
break;
}
}
}
}
}
if let Some(digit) = matched_digit {
// Interaction complete — digit matched.
completed_interactions
.push((lid.clone(), InteractionResult::Digit(digit)));
} else {
// Play prompt frame or silence.
let pcm_frame = if let Some(frame) = state.prompt_frames.pop_front() {
frame
} else {
state.prompt_done = true;
vec![0.0f32; MIX_FRAME_SIZE]
};
// Encode prompt frame to the leg's codec.
let target_rate = codec_sample_rate(slot.codec_pt);
let resampled = if target_rate == MIX_RATE {
pcm_frame
} else {
slot.transcoder
.resample_f32(&pcm_frame, MIX_RATE, target_rate)
.unwrap_or_default()
};
if let Ok(encoded) =
slot.transcoder.encode_from_f32(&resampled, slot.codec_pt)
{
if !encoded.is_empty() {
let header = build_rtp_header(
slot.codec_pt,
slot.rtp_seq,
slot.rtp_ts,
slot.rtp_ssrc,
);
let mut rtp = header.to_vec();
rtp.extend_from_slice(&encoded);
slot.rtp_seq = slot.rtp_seq.wrapping_add(1);
slot.rtp_ts = slot
.rtp_ts
.wrapping_add(rtp_clock_increment(slot.codec_pt));
let _ = slot.outbound_tx.try_send(rtp);
}
}
// Check timeout (only after prompt finishes).
if state.prompt_done {
if state.timeout_ticks_remaining == 0 {
completed_interactions
.push((lid.clone(), InteractionResult::Timeout));
} else {
state.timeout_ticks_remaining -= 1;
}
}
}
}
}
}
// Apply completed interactions — revert legs to Participant.
for (lid, result) in completed_interactions {
if let Some(slot) = legs.get_mut(&lid) {
if let LegRole::Isolated(ref mut state) = slot.role {
if let Some(tx) = state.result_tx.take() {
let _ = tx.send(result);
}
}
slot.role = LegRole::Participant;
}
}
// ── 5. Distribute per-source audio to tool legs. ────────────
if !tool_legs.is_empty() {
// Collect participant PCM frames (computed in step 2).
let sources: Vec<ToolAudioSource> = legs
.iter()
.filter(|(_, s)| matches!(s.role, LegRole::Participant))
.map(|(lid, s)| ToolAudioSource {
leg_id: lid.clone(),
pcm_48k: s.last_pcm_frame.clone(),
})
.collect();
for tool in tool_legs.values() {
let batch = ToolAudioBatch {
sources: sources
.iter()
.map(|s| ToolAudioSource {
leg_id: s.leg_id.clone(),
pcm_48k: s.pcm_48k.clone(),
})
.collect(),
};
// Non-blocking send — drop batch if tool can't keep up.
let _ = tool.audio_tx.try_send(batch);
}
}
// ── 6. Forward DTMF packets between participant legs only. ──
for (source_lid, dtmf_pkt) in &dtmf_forward {
// Skip if the source is an isolated leg (its DTMF was handled in step 4).
if let Some(src_slot) = legs.get(source_lid) {
if matches!(src_slot.role, LegRole::Isolated(_)) {
continue;
}
}
for (target_lid, target_slot) in legs.iter_mut() {
if target_lid == source_lid {
continue; // Don't echo DTMF back to sender.
}
// Don't forward to isolated legs.
if matches!(target_slot.role, LegRole::Isolated(_)) {
continue;
}
let mut header = build_rtp_header(
101,
target_slot.rtp_seq,
target_slot.rtp_ts,
target_slot.rtp_ssrc,
);
if dtmf_pkt.marker {
header[1] |= 0x80; // Set marker bit.
}
let mut rtp_out = header.to_vec();
rtp_out.extend_from_slice(&dtmf_pkt.payload);
target_slot.rtp_seq = target_slot.rtp_seq.wrapping_add(1);
// Don't increment rtp_ts for DTMF — it shares timestamp context with audio.
let _ = target_slot.outbound_tx.try_send(rtp_out);
}
}
}
}

View File

@@ -0,0 +1,378 @@
//! Provider registration state machine.
//!
//! Handles the REGISTER cycle with upstream SIP providers:
//! - Sends periodic REGISTER messages
//! - Handles 401/407 Digest authentication challenges
//! - Detects public IP from Via received= parameter
//! - Emits registration state events to TypeScript
//!
//! Ported from ts/providerstate.ts.
use crate::config::ProviderConfig;
use crate::ipc::{emit_event, OutTx};
use sip_proto::helpers::{
compute_digest_auth, generate_branch, generate_call_id, generate_tag, parse_digest_challenge,
};
use sip_proto::message::{RequestOptions, SipMessage};
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
use tokio::sync::Mutex;
use tokio::time::{self, Duration};
/// Runtime state for a single SIP provider.
pub struct ProviderState {
pub config: ProviderConfig,
pub public_ip: Option<String>,
pub is_registered: bool,
pub registered_aor: String,
// Registration transaction state.
reg_call_id: String,
reg_cseq: u32,
reg_from_tag: String,
// Network.
lan_ip: String,
lan_port: u16,
}
impl ProviderState {
pub fn new(config: ProviderConfig, public_ip_seed: Option<&str>) -> Self {
let aor = format!("sip:{}@{}", config.username, config.domain);
Self {
public_ip: public_ip_seed.map(|s| s.to_string()),
is_registered: false,
registered_aor: aor,
reg_call_id: generate_call_id(None),
reg_cseq: 0,
reg_from_tag: generate_tag(),
lan_ip: String::new(),
lan_port: 0,
config,
}
}
/// Build and send a REGISTER request.
pub fn build_register(&mut self) -> Vec<u8> {
self.reg_cseq += 1;
let pub_ip = self.public_ip.as_deref().unwrap_or(&self.lan_ip);
let register = SipMessage::create_request(
"REGISTER",
&format!("sip:{}", self.config.domain),
RequestOptions {
via_host: pub_ip.to_string(),
via_port: self.lan_port,
via_transport: None,
via_branch: Some(generate_branch()),
from_uri: self.registered_aor.clone(),
from_display_name: None,
from_tag: Some(self.reg_from_tag.clone()),
to_uri: self.registered_aor.clone(),
to_display_name: None,
to_tag: None,
call_id: Some(self.reg_call_id.clone()),
cseq: Some(self.reg_cseq),
contact: Some(format!(
"<sip:{}@{}:{}>",
self.config.username, pub_ip, self.lan_port
)),
max_forwards: Some(70),
body: None,
content_type: None,
extra_headers: Some(vec![
(
"Expires".to_string(),
self.config.register_interval_sec.to_string(),
),
("User-Agent".to_string(), "SipRouter/1.0".to_string()),
(
"Allow".to_string(),
"INVITE, ACK, OPTIONS, CANCEL, BYE, SUBSCRIBE, NOTIFY, INFO, REFER, UPDATE"
.to_string(),
),
]),
},
);
register.serialize()
}
/// Handle a SIP response that might be for this provider's REGISTER.
/// Returns true if the message was consumed.
pub fn handle_registration_response(&mut self, msg: &SipMessage) -> Option<Vec<u8>> {
if !msg.is_response() {
return None;
}
if msg.call_id() != self.reg_call_id {
return None;
}
let cseq_method = msg.cseq_method().unwrap_or("");
if !cseq_method.eq_ignore_ascii_case("REGISTER") {
return None;
}
let code = msg.status_code().unwrap_or(0);
if code == 200 {
self.is_registered = true;
return Some(Vec::new()); // consumed, no reply needed
}
if code == 401 || code == 407 {
let challenge_header = if code == 401 {
msg.get_header("WWW-Authenticate")
} else {
msg.get_header("Proxy-Authenticate")
};
let challenge_header = match challenge_header {
Some(h) => h,
None => return Some(Vec::new()), // consumed but no challenge
};
let challenge = match parse_digest_challenge(challenge_header) {
Some(c) => c,
None => return Some(Vec::new()),
};
let auth_value = compute_digest_auth(
&self.config.username,
&self.config.password,
&challenge.realm,
&challenge.nonce,
"REGISTER",
&format!("sip:{}", self.config.domain),
challenge.algorithm.as_deref(),
challenge.opaque.as_deref(),
);
// Resend REGISTER with auth credentials.
self.reg_cseq += 1;
let pub_ip = self.public_ip.as_deref().unwrap_or(&self.lan_ip);
let auth_header_name = if code == 401 {
"Authorization"
} else {
"Proxy-Authorization"
};
let register = SipMessage::create_request(
"REGISTER",
&format!("sip:{}", self.config.domain),
RequestOptions {
via_host: pub_ip.to_string(),
via_port: self.lan_port,
via_transport: None,
via_branch: Some(generate_branch()),
from_uri: self.registered_aor.clone(),
from_display_name: None,
from_tag: Some(self.reg_from_tag.clone()),
to_uri: self.registered_aor.clone(),
to_display_name: None,
to_tag: None,
call_id: Some(self.reg_call_id.clone()),
cseq: Some(self.reg_cseq),
contact: Some(format!(
"<sip:{}@{}:{}>",
self.config.username, pub_ip, self.lan_port
)),
max_forwards: Some(70),
body: None,
content_type: None,
extra_headers: Some(vec![
(auth_header_name.to_string(), auth_value),
(
"Expires".to_string(),
self.config.register_interval_sec.to_string(),
),
("User-Agent".to_string(), "SipRouter/1.0".to_string()),
(
"Allow".to_string(),
"INVITE, ACK, OPTIONS, CANCEL, BYE, SUBSCRIBE, NOTIFY, INFO, REFER, UPDATE"
.to_string(),
),
]),
},
);
return Some(register.serialize());
}
if code >= 400 {
self.is_registered = false;
}
Some(Vec::new()) // consumed
}
/// Detect public IP from Via received= parameter.
pub fn detect_public_ip(&mut self, via: &str) {
if let Some(m) = via.find("received=") {
let rest = &via[m + 9..];
let end = rest
.find(|c: char| !c.is_ascii_digit() && c != '.')
.unwrap_or(rest.len());
let ip = &rest[..end];
if !ip.is_empty() && self.public_ip.as_deref() != Some(ip) {
self.public_ip = Some(ip.to_string());
}
}
}
pub fn set_network(&mut self, lan_ip: &str, lan_port: u16) {
self.lan_ip = lan_ip.to_string();
self.lan_port = lan_port;
}
}
/// Manages all provider states and their registration cycles.
pub struct ProviderManager {
providers: Vec<Arc<Mutex<ProviderState>>>,
out_tx: OutTx,
}
impl ProviderManager {
pub fn new(out_tx: OutTx) -> Self {
Self {
providers: Vec::new(),
out_tx,
}
}
/// Initialize providers from config and start registration cycles.
pub async fn configure(
&mut self,
configs: &[ProviderConfig],
public_ip_seed: Option<&str>,
lan_ip: &str,
lan_port: u16,
socket: Arc<UdpSocket>,
) {
self.providers.clear();
for cfg in configs {
let mut ps = ProviderState::new(cfg.clone(), public_ip_seed);
ps.set_network(lan_ip, lan_port);
let ps = Arc::new(Mutex::new(ps));
self.providers.push(ps.clone());
// Start the registration cycle.
let socket = socket.clone();
let out_tx = self.out_tx.clone();
tokio::spawn(async move {
provider_register_loop(ps, socket, out_tx).await;
});
}
}
/// Try to handle a SIP response as a provider registration response.
/// Returns true if consumed.
pub async fn handle_response(
&self,
msg: &SipMessage,
socket: &UdpSocket,
) -> bool {
for ps_arc in &self.providers {
let mut ps = ps_arc.lock().await;
let was_registered = ps.is_registered;
if let Some(reply) = ps.handle_registration_response(msg) {
// If there's a reply to send (e.g. auth retry).
if !reply.is_empty() {
if let Some(dest) = ps.config.outbound_proxy.to_socket_addr() {
let _ = socket.send_to(&reply, dest).await;
}
}
// Emit registration state change.
if ps.is_registered != was_registered {
emit_event(
&self.out_tx,
"provider_registered",
serde_json::json!({
"provider_id": ps.config.id,
"registered": ps.is_registered,
"public_ip": ps.public_ip,
}),
);
}
return true;
}
}
false
}
/// Find which provider sent a packet by matching source address.
pub async fn find_by_address(&self, addr: &SocketAddr) -> Option<Arc<Mutex<ProviderState>>> {
for ps_arc in &self.providers {
let ps = ps_arc.lock().await;
let proxy_addr = format!(
"{}:{}",
ps.config.outbound_proxy.address, ps.config.outbound_proxy.port
);
if let Ok(expected) = proxy_addr.parse::<SocketAddr>() {
if expected == *addr {
return Some(ps_arc.clone());
}
}
// Also match by IP only (port may differ).
if ps.config.outbound_proxy.address == addr.ip().to_string() {
return Some(ps_arc.clone());
}
}
None
}
/// Find a provider by its config ID (e.g. "easybell").
pub async fn find_by_provider_id(&self, provider_id: &str) -> Option<Arc<Mutex<ProviderState>>> {
for ps_arc in &self.providers {
let ps = ps_arc.lock().await;
if ps.config.id == provider_id {
return Some(ps_arc.clone());
}
}
None
}
/// Check if a provider is currently registered.
pub async fn is_registered(&self, provider_id: &str) -> bool {
for ps_arc in &self.providers {
let ps = ps_arc.lock().await;
if ps.config.id == provider_id {
return ps.is_registered;
}
}
false
}
}
/// Registration loop for a single provider.
async fn provider_register_loop(
ps: Arc<Mutex<ProviderState>>,
socket: Arc<UdpSocket>,
_out_tx: OutTx,
) {
// Initial registration.
{
let mut state = ps.lock().await;
let register_buf = state.build_register();
if let Some(dest) = state.config.outbound_proxy.to_socket_addr() {
let _ = socket.send_to(&register_buf, dest).await;
}
}
// Re-register periodically (85% of the interval).
let interval_sec = {
let state = ps.lock().await;
(state.config.register_interval_sec as f64 * 0.85) as u64
};
let mut interval = time::interval(Duration::from_secs(interval_sec.max(30)));
interval.tick().await; // skip first immediate tick
loop {
interval.tick().await;
let mut state = ps.lock().await;
let register_buf = state.build_register();
if let Some(dest) = state.config.outbound_proxy.to_socket_addr() {
let _ = socket.send_to(&register_buf, dest).await;
}
}
}

View File

@@ -0,0 +1,182 @@
//! Audio recorder — receives RTP packets and writes a WAV file.
use codec_lib::TranscodeState;
use std::path::Path;
/// Active recording session.
pub struct Recorder {
writer: hound::WavWriter<std::io::BufWriter<std::fs::File>>,
transcoder: TranscodeState,
source_pt: u8,
total_samples: u64,
sample_rate: u32,
max_samples: Option<u64>,
file_path: String,
}
impl Recorder {
/// Create a new recorder that writes to a WAV file.
/// `source_pt` is the RTP payload type of the incoming audio.
/// `max_duration_ms` optionally limits the recording length.
pub fn new(
file_path: &str,
source_pt: u8,
max_duration_ms: Option<u64>,
) -> Result<Self, String> {
// Ensure parent directory exists.
if let Some(parent) = Path::new(file_path).parent() {
std::fs::create_dir_all(parent)
.map_err(|e| format!("create dir: {e}"))?;
}
let sample_rate = 8000u32; // Record at 8kHz (standard telephony)
let spec = hound::WavSpec {
channels: 1,
sample_rate,
bits_per_sample: 16,
sample_format: hound::SampleFormat::Int,
};
let writer = hound::WavWriter::create(file_path, spec)
.map_err(|e| format!("create WAV {file_path}: {e}"))?;
let transcoder = TranscodeState::new().map_err(|e| format!("codec init: {e}"))?;
let max_samples = max_duration_ms.map(|ms| (sample_rate as u64 * ms) / 1000);
Ok(Self {
writer,
transcoder,
source_pt,
total_samples: 0,
sample_rate,
max_samples,
file_path: file_path.to_string(),
})
}
/// Create a recorder that writes raw PCM at a given sample rate.
/// Used by tool legs that already have decoded PCM (no RTP processing needed).
pub fn new_pcm(file_path: &str, sample_rate: u32, max_duration_ms: Option<u64>) -> Result<Self, String> {
if let Some(parent) = Path::new(file_path).parent() {
std::fs::create_dir_all(parent)
.map_err(|e| format!("create dir: {e}"))?;
}
let spec = hound::WavSpec {
channels: 1,
sample_rate,
bits_per_sample: 16,
sample_format: hound::SampleFormat::Int,
};
let writer = hound::WavWriter::create(file_path, spec)
.map_err(|e| format!("create WAV {file_path}: {e}"))?;
// source_pt is unused for PCM recording; set to 0.
let transcoder = TranscodeState::new().map_err(|e| format!("codec init: {e}"))?;
let max_samples = max_duration_ms.map(|ms| (sample_rate as u64 * ms) / 1000);
Ok(Self {
writer,
transcoder,
source_pt: 0,
total_samples: 0,
sample_rate,
max_samples,
file_path: file_path.to_string(),
})
}
/// Write raw PCM samples directly (no RTP decoding).
/// Returns true if recording should continue, false if max duration reached.
pub fn write_pcm(&mut self, samples: &[i16]) -> bool {
for &sample in samples {
if self.writer.write_sample(sample).is_err() {
return false;
}
self.total_samples += 1;
if let Some(max) = self.max_samples {
if self.total_samples >= max {
return false;
}
}
}
true
}
/// Process an incoming RTP packet (full packet with header).
/// Returns true if recording should continue, false if max duration reached.
pub fn process_rtp(&mut self, data: &[u8]) -> bool {
if data.len() <= 12 {
return true; // Too short, skip.
}
let pt = data[1] & 0x7F;
// Skip telephone-event (DTMF) packets.
if pt == 101 {
return true;
}
let payload = &data[12..];
if payload.is_empty() {
return true;
}
// Decode to PCM.
let (pcm, rate) = match self.transcoder.decode_to_pcm(payload, self.source_pt) {
Ok(result) => result,
Err(_) => return true, // Decode failed, skip packet.
};
// Resample to 8kHz if needed.
let pcm_8k = if rate != self.sample_rate {
match self.transcoder.resample(&pcm, rate, self.sample_rate) {
Ok(r) => r,
Err(_) => return true,
}
} else {
pcm
};
// Write samples.
for &sample in &pcm_8k {
if let Err(_) = self.writer.write_sample(sample) {
return false;
}
self.total_samples += 1;
if let Some(max) = self.max_samples {
if self.total_samples >= max {
return false; // Max duration reached.
}
}
}
true
}
/// Stop recording and finalize the WAV file.
pub fn stop(self) -> RecordingResult {
let duration_ms = if self.sample_rate > 0 {
(self.total_samples * 1000) / self.sample_rate as u64
} else {
0
};
// Writer is finalized on drop (writes RIFF header sizes).
drop(self.writer);
RecordingResult {
file_path: self.file_path,
duration_ms,
total_samples: self.total_samples,
}
}
}
pub struct RecordingResult {
pub file_path: String,
pub duration_ms: u64,
pub total_samples: u64,
}

View File

@@ -0,0 +1,171 @@
//! Device registrar — accepts REGISTER from SIP phones and tracks contacts.
//!
//! When a device sends REGISTER, the registrar responds with 200 OK
//! and stores the device's current contact (source IP:port).
//!
//! Ported from ts/registrar.ts.
use crate::config::DeviceConfig;
use crate::ipc::{emit_event, OutTx};
use sip_proto::helpers::generate_tag;
use sip_proto::message::{ResponseOptions, SipMessage};
use std::collections::HashMap;
use std::net::SocketAddr;
use std::time::{Duration, Instant};
const MAX_EXPIRES: u32 = 300;
/// A registered device entry.
#[derive(Debug, Clone)]
pub struct RegisteredDevice {
pub device_id: String,
pub display_name: String,
pub extension: String,
pub contact_addr: SocketAddr,
pub registered_at: Instant,
pub expires_at: Instant,
pub aor: String,
}
/// Manages device registrations.
pub struct Registrar {
/// Known device configs (from app config).
devices: Vec<DeviceConfig>,
/// Currently registered devices, keyed by device ID.
registered: HashMap<String, RegisteredDevice>,
out_tx: OutTx,
}
impl Registrar {
pub fn new(out_tx: OutTx) -> Self {
Self {
devices: Vec::new(),
registered: HashMap::new(),
out_tx,
}
}
/// Update the known device list from config.
pub fn configure(&mut self, devices: &[DeviceConfig]) {
self.devices = devices.to_vec();
}
/// Try to handle a SIP REGISTER from a device.
/// Returns Some(response_bytes) if handled, None if not a known device.
pub fn handle_register(
&mut self,
msg: &SipMessage,
from_addr: SocketAddr,
) -> Option<Vec<u8>> {
if msg.method() != Some("REGISTER") {
return None;
}
// Find the device by matching the source IP against expectedAddress.
let from_ip = from_addr.ip().to_string();
let device = self.devices.iter().find(|d| d.expected_address == from_ip)?;
let from_header = msg.get_header("From").unwrap_or("");
let aor = SipMessage::extract_uri(from_header)
.map(|s| s.to_string())
.unwrap_or_else(|| format!("sip:{}@{}", device.extension, from_ip));
let expires_header = msg.get_header("Expires");
let requested: u32 = expires_header
.and_then(|s| s.parse().ok())
.unwrap_or(3600);
let expires = requested.min(MAX_EXPIRES);
let entry = RegisteredDevice {
device_id: device.id.clone(),
display_name: device.display_name.clone(),
extension: device.extension.clone(),
contact_addr: from_addr,
registered_at: Instant::now(),
expires_at: Instant::now() + Duration::from_secs(expires as u64),
aor: aor.clone(),
};
self.registered.insert(device.id.clone(), entry);
// Emit event to TypeScript.
emit_event(
&self.out_tx,
"device_registered",
serde_json::json!({
"device_id": device.id,
"display_name": device.display_name,
"address": from_ip,
"port": from_addr.port(),
"aor": aor,
"expires": expires,
}),
);
// Build 200 OK response.
let contact = msg
.get_header("Contact")
.map(|s| s.to_string())
.unwrap_or_else(|| format!("<sip:{}:{}>", from_ip, from_addr.port()));
let response = SipMessage::create_response(
200,
"OK",
msg,
Some(ResponseOptions {
to_tag: Some(generate_tag()),
contact: Some(contact),
extra_headers: Some(vec![(
"Expires".to_string(),
expires.to_string(),
)]),
..Default::default()
}),
);
Some(response.serialize())
}
/// Get the contact address for a registered device.
pub fn get_device_contact(&self, device_id: &str) -> Option<SocketAddr> {
let entry = self.registered.get(device_id)?;
if Instant::now() > entry.expires_at {
return None;
}
Some(entry.contact_addr)
}
/// Check if a source address belongs to a known device.
pub fn is_known_device_address(&self, addr: &str) -> bool {
self.devices.iter().any(|d| d.expected_address == addr)
}
/// Find a registered device by its source IP address.
pub fn find_by_address(&self, addr: &SocketAddr) -> Option<&RegisteredDevice> {
let ip = addr.ip().to_string();
self.registered.values().find(|e| {
e.contact_addr.ip().to_string() == ip && Instant::now() <= e.expires_at
})
}
/// Get all device statuses for the dashboard.
pub fn get_all_statuses(&self) -> Vec<serde_json::Value> {
let now = Instant::now();
let mut result = Vec::new();
for dc in &self.devices {
let reg = self.registered.get(&dc.id);
let connected = reg.map(|r| now <= r.expires_at).unwrap_or(false);
result.push(serde_json::json!({
"id": dc.id,
"displayName": dc.display_name,
"address": reg.filter(|_| connected).map(|r| r.contact_addr.ip().to_string()),
"port": reg.filter(|_| connected).map(|r| r.contact_addr.port()),
"aor": reg.map(|r| r.aor.as_str()).unwrap_or(""),
"connected": connected,
"isBrowser": false,
}));
}
result
}
}

View File

@@ -0,0 +1,158 @@
//! RTP port pool and media forwarding.
//!
//! Manages a pool of even-numbered UDP ports for RTP media.
//! Each port gets a bound tokio UdpSocket. Supports:
//! - Direct forwarding (SIP-to-SIP, no transcoding)
//! - Transcoding forwarding (via codec-lib, e.g. G.722 ↔ Opus)
//! - Silence generation
//! - NAT priming
//!
//! Ported from ts/call/rtp-port-pool.ts + sip-leg.ts RTP handling.
use std::collections::HashMap;
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
/// A single RTP port allocation.
pub struct RtpAllocation {
pub port: u16,
pub socket: Arc<UdpSocket>,
}
/// RTP port pool — allocates even-numbered UDP ports.
pub struct RtpPortPool {
min: u16,
max: u16,
allocated: HashMap<u16, Arc<UdpSocket>>,
}
impl RtpPortPool {
pub fn new(min: u16, max: u16) -> Self {
let min = if min % 2 == 0 { min } else { min + 1 };
Self {
min,
max,
allocated: HashMap::new(),
}
}
/// Allocate an even-numbered port and bind a UDP socket.
pub async fn allocate(&mut self) -> Option<RtpAllocation> {
let mut port = self.min;
while port < self.max {
if !self.allocated.contains_key(&port) {
match UdpSocket::bind(format!("0.0.0.0:{port}")).await {
Ok(sock) => {
let sock = Arc::new(sock);
self.allocated.insert(port, sock.clone());
return Some(RtpAllocation { port, socket: sock });
}
Err(_) => {
// Port in use, try next.
}
}
}
port += 2;
}
None // Pool exhausted.
}
/// Release a port back to the pool.
pub fn release(&mut self, port: u16) {
self.allocated.remove(&port);
// Socket is dropped when the last Arc reference goes away.
}
pub fn size(&self) -> usize {
self.allocated.len()
}
pub fn capacity(&self) -> usize {
((self.max - self.min) / 2) as usize
}
}
/// An active RTP relay between two endpoints.
/// Receives on `local_socket` and forwards to `remote_addr`.
pub struct RtpRelay {
pub local_port: u16,
pub local_socket: Arc<UdpSocket>,
pub remote_addr: Option<SocketAddr>,
/// If set, transcode packets using this codec session before forwarding.
pub transcode: Option<TranscodeConfig>,
/// Packets received counter.
pub pkt_received: u64,
/// Packets sent counter.
pub pkt_sent: u64,
}
pub struct TranscodeConfig {
pub from_pt: u8,
pub to_pt: u8,
pub session_id: String,
}
impl RtpRelay {
pub fn new(port: u16, socket: Arc<UdpSocket>) -> Self {
Self {
local_port: port,
local_socket: socket,
remote_addr: None,
transcode: None,
pkt_received: 0,
pkt_sent: 0,
}
}
pub fn set_remote(&mut self, addr: SocketAddr) {
self.remote_addr = Some(addr);
}
}
/// Send a 1-byte NAT priming packet to open a pinhole.
pub async fn prime_nat(socket: &UdpSocket, remote: SocketAddr) {
let _ = socket.send_to(&[0u8], remote).await;
}
/// Build an RTP silence frame for PCMU (payload type 0).
pub fn silence_frame_pcmu() -> Vec<u8> {
// 12-byte RTP header + 160 bytes of µ-law silence (0xFF)
let mut frame = vec![0u8; 172];
frame[0] = 0x80; // V=2
frame[1] = 0; // PT=0 (PCMU)
// seq, timestamp, ssrc left as 0 — caller should set these
frame[12..].fill(0xFF); // µ-law silence
frame
}
/// Build an RTP silence frame for G.722 (payload type 9).
pub fn silence_frame_g722() -> Vec<u8> {
// 12-byte RTP header + 160 bytes of G.722 silence
let mut frame = vec![0u8; 172];
frame[0] = 0x80; // V=2
frame[1] = 9; // PT=9 (G.722)
// G.722 silence: all zeros is valid silence
frame
}
/// Build an RTP header with the given parameters.
pub fn build_rtp_header(pt: u8, seq: u16, timestamp: u32, ssrc: u32) -> [u8; 12] {
let mut header = [0u8; 12];
header[0] = 0x80; // V=2
header[1] = pt & 0x7F;
header[2..4].copy_from_slice(&seq.to_be_bytes());
header[4..8].copy_from_slice(&timestamp.to_be_bytes());
header[8..12].copy_from_slice(&ssrc.to_be_bytes());
header
}
/// Get the RTP clock increment per 20ms frame for a payload type.
pub fn rtp_clock_increment(pt: u8) -> u32 {
match pt {
9 => 160, // G.722: 8000 Hz clock rate (despite 16kHz audio) × 0.02s
0 | 8 => 160, // PCMU/PCMA: 8000 × 0.02
111 => 960, // Opus: 48000 × 0.02
_ => 160,
}
}

View File

@@ -0,0 +1,475 @@
//! SipLeg — manages one side of a B2BUA call.
//!
//! Handles the full INVITE lifecycle:
//! - Send INVITE with SDP
//! - Handle 407 Proxy Authentication (digest auth retry)
//! - Handle 200 OK (ACK, learn media endpoint)
//! - Handle BYE/CANCEL (teardown)
//! - Track SIP dialog state (early → confirmed → terminated)
//!
//! Ported from ts/call/sip-leg.ts.
use sip_proto::dialog::{DialogState, SipDialog};
use sip_proto::helpers::{
build_sdp, compute_digest_auth, generate_branch, generate_tag, parse_digest_challenge,
parse_sdp_endpoint, SdpOptions,
};
use sip_proto::message::{RequestOptions, SipMessage};
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
/// State of a SIP leg.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum LegState {
Inviting,
Ringing,
Connected,
Terminating,
Terminated,
}
/// Configuration for creating a SIP leg.
pub struct SipLegConfig {
/// Proxy LAN IP (for Via, Contact, SDP).
pub lan_ip: String,
/// Proxy LAN port.
pub lan_port: u16,
/// Public IP (for provider-facing legs).
pub public_ip: Option<String>,
/// SIP target endpoint (provider outbound proxy or device address).
pub sip_target: SocketAddr,
/// Provider credentials (for 407 auth).
pub username: Option<String>,
pub password: Option<String>,
pub registered_aor: Option<String>,
/// Codec payload types to offer.
pub codecs: Vec<u8>,
/// Our RTP port for SDP.
pub rtp_port: u16,
}
/// A SIP leg with full dialog management.
pub struct SipLeg {
pub id: String,
pub state: LegState,
pub config: SipLegConfig,
pub dialog: Option<SipDialog>,
/// The INVITE we sent (needed for CANCEL and 407 ACK).
invite: Option<SipMessage>,
/// Original unauthenticated INVITE (for re-ACKing retransmitted 407s).
orig_invite: Option<SipMessage>,
/// Whether we've attempted digest auth.
auth_attempted: bool,
/// Remote media endpoint (learned from SDP in 200 OK).
pub remote_media: Option<SocketAddr>,
}
impl SipLeg {
pub fn new(id: String, config: SipLegConfig) -> Self {
Self {
id,
state: LegState::Inviting,
config,
dialog: None,
invite: None,
orig_invite: None,
auth_attempted: false,
remote_media: None,
}
}
/// Build and send an INVITE to establish this leg.
pub async fn send_invite(
&mut self,
from_uri: &str,
to_uri: &str,
sip_call_id: &str,
socket: &UdpSocket,
) {
let ip = self
.config
.public_ip
.as_deref()
.unwrap_or(&self.config.lan_ip);
let sdp = build_sdp(&SdpOptions {
ip,
port: self.config.rtp_port,
payload_types: &self.config.codecs,
..Default::default()
});
let invite = SipMessage::create_request(
"INVITE",
to_uri,
RequestOptions {
via_host: ip.to_string(),
via_port: self.config.lan_port,
via_transport: None,
via_branch: Some(generate_branch()),
from_uri: from_uri.to_string(),
from_display_name: None,
from_tag: Some(generate_tag()),
to_uri: to_uri.to_string(),
to_display_name: None,
to_tag: None,
call_id: Some(sip_call_id.to_string()),
cseq: Some(1),
contact: Some(format!("<sip:{ip}:{}>", self.config.lan_port)),
max_forwards: Some(70),
body: Some(sdp),
content_type: Some("application/sdp".to_string()),
extra_headers: Some(vec![
("User-Agent".to_string(), "SipRouter/1.0".to_string()),
]),
},
);
self.dialog = Some(SipDialog::from_uac_invite(&invite, ip, self.config.lan_port));
self.invite = Some(invite.clone());
self.state = LegState::Inviting;
let _ = socket.send_to(&invite.serialize(), self.config.sip_target).await;
}
/// Handle an incoming SIP message routed to this leg.
/// Returns an optional reply to send (e.g. ACK, auth retry INVITE).
pub fn handle_message(&mut self, msg: &SipMessage) -> SipLegAction {
if msg.is_response() {
self.handle_response(msg)
} else {
self.handle_request(msg)
}
}
fn handle_response(&mut self, msg: &SipMessage) -> SipLegAction {
let code = msg.status_code().unwrap_or(0);
let cseq_method = msg.cseq_method().unwrap_or("").to_uppercase();
if cseq_method != "INVITE" {
return SipLegAction::None;
}
// Handle retransmitted 407 for the original unauthenticated INVITE.
if self.auth_attempted {
if let Some(dialog) = &self.dialog {
let response_cseq: u32 = msg
.get_header("CSeq")
.and_then(|s| s.split_whitespace().next())
.and_then(|s| s.parse().ok())
.unwrap_or(0);
if response_cseq < dialog.local_cseq && code >= 400 {
// ACK the retransmitted error response.
if let Some(orig) = &self.orig_invite {
let ack = build_non_2xx_ack(orig, msg);
return SipLegAction::Send(ack.serialize());
}
return SipLegAction::None;
}
}
}
// Handle 407 Proxy Authentication Required.
if code == 407 {
return self.handle_auth_challenge(msg);
}
// Update dialog state.
if let Some(dialog) = &mut self.dialog {
dialog.process_response(msg);
}
if code == 180 || code == 183 {
self.state = LegState::Ringing;
SipLegAction::StateChange(LegState::Ringing)
} else if code >= 200 && code < 300 {
// ACK the 200 OK.
let ack_buf = if let Some(dialog) = &self.dialog {
let ack = dialog.create_ack();
Some(ack.serialize())
} else {
None
};
// If already connected (200 retransmit), just re-ACK.
if self.state == LegState::Connected {
return match ack_buf {
Some(buf) => SipLegAction::Send(buf),
None => SipLegAction::None,
};
}
// Learn media endpoint from SDP.
if msg.has_sdp_body() {
if let Some(ep) = parse_sdp_endpoint(&msg.body) {
if let Ok(addr) = format!("{}:{}", ep.address, ep.port).parse() {
self.remote_media = Some(addr);
}
}
}
self.state = LegState::Connected;
match ack_buf {
Some(buf) => SipLegAction::ConnectedWithAck(buf),
None => SipLegAction::StateChange(LegState::Connected),
}
} else if code >= 300 {
self.state = LegState::Terminated;
if let Some(dialog) = &mut self.dialog {
dialog.terminate();
}
SipLegAction::Terminated(format!("rejected_{code}"))
} else {
SipLegAction::None // 1xx provisional
}
}
fn handle_auth_challenge(&mut self, msg: &SipMessage) -> SipLegAction {
if self.auth_attempted {
self.state = LegState::Terminated;
if let Some(dialog) = &mut self.dialog {
dialog.terminate();
}
return SipLegAction::Terminated("auth_rejected".to_string());
}
self.auth_attempted = true;
let challenge_header = match msg.get_header("Proxy-Authenticate") {
Some(h) => h,
None => {
self.state = LegState::Terminated;
return SipLegAction::Terminated("407_no_challenge".to_string());
}
};
let challenge = match parse_digest_challenge(challenge_header) {
Some(c) => c,
None => {
self.state = LegState::Terminated;
return SipLegAction::Terminated("407_bad_challenge".to_string());
}
};
let password = match &self.config.password {
Some(p) => p.clone(),
None => {
self.state = LegState::Terminated;
return SipLegAction::Terminated("407_no_password".to_string());
}
};
let aor = match &self.config.registered_aor {
Some(a) => a.clone(),
None => {
self.state = LegState::Terminated;
return SipLegAction::Terminated("407_no_aor".to_string());
}
};
let username = aor
.trim_start_matches("sip:")
.trim_start_matches("sips:")
.split('@')
.next()
.unwrap_or("")
.to_string();
let dest_uri = self
.invite
.as_ref()
.and_then(|i| i.request_uri())
.unwrap_or("")
.to_string();
let auth_value = compute_digest_auth(
&username,
&password,
&challenge.realm,
&challenge.nonce,
"INVITE",
&dest_uri,
challenge.algorithm.as_deref(),
challenge.opaque.as_deref(),
);
// ACK the 407.
let mut ack_buf = None;
if let Some(invite) = &self.invite {
let ack = build_non_2xx_ack(invite, msg);
ack_buf = Some(ack.serialize());
}
// Save original INVITE for retransmission handling.
self.orig_invite = self.invite.clone();
// Build authenticated INVITE with same From tag, CSeq=2.
let ip = self
.config
.public_ip
.as_deref()
.unwrap_or(&self.config.lan_ip);
let from_tag = self
.dialog
.as_ref()
.map(|d| d.local_tag.clone())
.unwrap_or_else(generate_tag);
let sdp = build_sdp(&SdpOptions {
ip,
port: self.config.rtp_port,
payload_types: &self.config.codecs,
..Default::default()
});
let call_id = self
.dialog
.as_ref()
.map(|d| d.call_id.clone())
.unwrap_or_default();
let invite_auth = SipMessage::create_request(
"INVITE",
&dest_uri,
RequestOptions {
via_host: ip.to_string(),
via_port: self.config.lan_port,
via_transport: None,
via_branch: Some(generate_branch()),
from_uri: aor,
from_display_name: None,
from_tag: Some(from_tag),
to_uri: dest_uri.clone(),
to_display_name: None,
to_tag: None,
call_id: Some(call_id),
cseq: Some(2),
contact: Some(format!("<sip:{ip}:{}>", self.config.lan_port)),
max_forwards: Some(70),
body: Some(sdp),
content_type: Some("application/sdp".to_string()),
extra_headers: Some(vec![
("Proxy-Authorization".to_string(), auth_value),
("User-Agent".to_string(), "SipRouter/1.0".to_string()),
]),
},
);
self.invite = Some(invite_auth.clone());
if let Some(dialog) = &mut self.dialog {
dialog.local_cseq = 2;
}
// Return both the ACK for the 407 and the new authenticated INVITE.
let invite_buf = invite_auth.serialize();
SipLegAction::AuthRetry {
ack_407: ack_buf,
invite_with_auth: invite_buf,
}
}
fn handle_request(&mut self, msg: &SipMessage) -> SipLegAction {
let method = msg.method().unwrap_or("");
if method == "BYE" {
let ok = SipMessage::create_response(200, "OK", msg, None);
self.state = LegState::Terminated;
if let Some(dialog) = &mut self.dialog {
dialog.terminate();
}
return SipLegAction::SendAndTerminate(ok.serialize(), "bye".to_string());
}
if method == "INFO" {
let ok = SipMessage::create_response(200, "OK", msg, None);
return SipLegAction::Send(ok.serialize());
}
SipLegAction::None
}
/// Build a BYE or CANCEL to tear down this leg.
pub fn build_hangup(&mut self) -> Option<Vec<u8>> {
let dialog = self.dialog.as_mut()?;
let msg = if dialog.state == DialogState::Confirmed {
dialog.create_request("BYE", None, None, None)
} else if dialog.state == DialogState::Early {
if let Some(invite) = &self.invite {
dialog.create_cancel(invite)
} else {
return None;
}
} else {
return None;
};
self.state = LegState::Terminating;
dialog.terminate();
Some(msg.serialize())
}
/// Get the SIP Call-ID for routing.
pub fn sip_call_id(&self) -> Option<&str> {
self.dialog.as_ref().map(|d| d.call_id.as_str())
}
}
/// Actions produced by the SipLeg message handler.
pub enum SipLegAction {
/// No action needed.
None,
/// Send a SIP message (ACK, 200 OK to INFO, etc.).
Send(Vec<u8>),
/// Leg state changed.
StateChange(LegState),
/// Connected — send this ACK.
ConnectedWithAck(Vec<u8>),
/// Terminated with a reason.
Terminated(String),
/// Send 200 OK and terminate.
SendAndTerminate(Vec<u8>, String),
/// 407 auth retry — send ACK for 407, then send new INVITE with auth.
AuthRetry {
ack_407: Option<Vec<u8>>,
invite_with_auth: Vec<u8>,
},
}
/// Build an ACK for a non-2xx response (same transaction as the INVITE).
fn build_non_2xx_ack(original_invite: &SipMessage, response: &SipMessage) -> SipMessage {
let via = original_invite.get_header("Via").unwrap_or("").to_string();
let from = original_invite
.get_header("From")
.unwrap_or("")
.to_string();
let to = response.get_header("To").unwrap_or("").to_string();
let call_id = original_invite.call_id().to_string();
let cseq_num: u32 = original_invite
.get_header("CSeq")
.and_then(|s| s.split_whitespace().next())
.and_then(|s| s.parse().ok())
.unwrap_or(1);
let ruri = original_invite
.request_uri()
.unwrap_or("sip:unknown")
.to_string();
SipMessage::new(
format!("ACK {ruri} SIP/2.0"),
vec![
("Via".to_string(), via),
("From".to_string(), from),
("To".to_string(), to),
("Call-ID".to_string(), call_id),
("CSeq".to_string(), format!("{cseq_num} ACK")),
("Max-Forwards".to_string(), "70".to_string()),
("Content-Length".to_string(), "0".to_string()),
],
String::new(),
)
}

View File

@@ -0,0 +1,67 @@
//! SIP UDP transport — owns the main SIP socket.
//!
//! Binds a UDP socket, receives SIP messages, and provides a send method.
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
/// The SIP UDP transport layer.
pub struct SipTransport {
socket: Arc<UdpSocket>,
}
impl SipTransport {
/// Bind a UDP socket on the given address (e.g. "0.0.0.0:5070").
pub async fn bind(bind_addr: &str) -> Result<Self, String> {
let socket = UdpSocket::bind(bind_addr)
.await
.map_err(|e| format!("bind {bind_addr}: {e}"))?;
Ok(Self {
socket: Arc::new(socket),
})
}
/// Get a clone of the socket Arc for the receiver task.
pub fn socket(&self) -> Arc<UdpSocket> {
self.socket.clone()
}
/// Send a raw SIP message to a destination.
pub async fn send_to(&self, data: &[u8], dest: SocketAddr) -> Result<usize, String> {
self.socket
.send_to(data, dest)
.await
.map_err(|e| format!("send to {dest}: {e}"))
}
/// Send a raw SIP message to an address:port pair.
pub async fn send_to_addr(&self, data: &[u8], addr: &str, port: u16) -> Result<usize, String> {
let dest: SocketAddr = format!("{addr}:{port}")
.parse()
.map_err(|e| format!("bad address {addr}:{port}: {e}"))?;
self.send_to(data, dest).await
}
/// Spawn the UDP receive loop. Calls the handler for every received packet.
pub fn spawn_receiver<F>(
&self,
handler: F,
) where
F: Fn(&[u8], SocketAddr) + Send + 'static,
{
let socket = self.socket.clone();
tokio::spawn(async move {
let mut buf = vec![0u8; 65535];
loop {
match socket.recv_from(&mut buf).await {
Ok((n, addr)) => handler(&buf[..n], addr),
Err(e) => {
eprintln!("[sip_transport] recv error: {e}");
tokio::time::sleep(std::time::Duration::from_millis(100)).await;
}
}
}
});
}
}

View File

@@ -0,0 +1,143 @@
//! Tool leg consumers — background tasks that process per-source unmerged audio.
//!
//! Tool legs are observer legs that receive individual audio streams from each
//! participant in a call. The mixer pipes `ToolAudioBatch` every 20ms containing
//! each participant's decoded PCM@48kHz f32 tagged with source leg ID.
//!
//! Consumers:
//! - **Recording**: writes per-source WAV files for speaker-separated recording.
//! - **Transcription**: stub for future Whisper integration (accumulates audio in Rust).
use crate::ipc::{emit_event, OutTx};
use crate::mixer::ToolAudioBatch;
use crate::recorder::Recorder;
use std::collections::HashMap;
use tokio::sync::mpsc;
use tokio::task::JoinHandle;
// ---------------------------------------------------------------------------
// Recording consumer
// ---------------------------------------------------------------------------
/// Spawn a recording tool leg that writes per-source WAV files.
///
/// Returns the channel sender (for the mixer to send batches) and the task handle.
/// When the channel is closed (tool leg removed), all WAV files are finalized
/// and a `tool_recording_done` event is emitted.
pub fn spawn_recording_tool(
tool_leg_id: String,
call_id: String,
base_dir: String,
out_tx: OutTx,
) -> (mpsc::Sender<ToolAudioBatch>, JoinHandle<()>) {
let (tx, mut rx) = mpsc::channel::<ToolAudioBatch>(64);
let handle = tokio::spawn(async move {
let mut recorders: HashMap<String, Recorder> = HashMap::new();
while let Some(batch) = rx.recv().await {
for source in &batch.sources {
// Skip silence-only frames (near-zero = no audio activity).
let has_audio = source.pcm_48k.iter().any(|&s| s.abs() > 1e-6);
if !has_audio && !recorders.contains_key(&source.leg_id) {
continue; // Don't create a file for silence-only sources.
}
let recorder = recorders.entry(source.leg_id.clone()).or_insert_with(|| {
let path = format!("{}/{}-{}.wav", base_dir, call_id, source.leg_id);
Recorder::new_pcm(&path, 48000, None).unwrap_or_else(|e| {
panic!("failed to create recorder for {}: {e}", source.leg_id);
})
});
// Convert f32 [-1.0, 1.0] to i16 for WAV writing.
let pcm_i16: Vec<i16> = source.pcm_48k
.iter()
.map(|&s| (s * 32767.0).round().clamp(-32768.0, 32767.0) as i16)
.collect();
if !recorder.write_pcm(&pcm_i16) {
// Max duration reached — stop recording this source.
break;
}
}
}
// Channel closed — finalize all recordings.
let mut files = Vec::new();
for (leg_id, rec) in recorders {
let result = rec.stop();
files.push(serde_json::json!({
"source_leg_id": leg_id,
"file_path": result.file_path,
"duration_ms": result.duration_ms,
}));
}
emit_event(
&out_tx,
"tool_recording_done",
serde_json::json!({
"call_id": call_id,
"tool_leg_id": tool_leg_id,
"files": files,
}),
);
});
(tx, handle)
}
// ---------------------------------------------------------------------------
// Transcription consumer (stub — real plumbing, stub consumer)
// ---------------------------------------------------------------------------
/// Spawn a transcription tool leg.
///
/// The plumbing is fully real: it receives per-source unmerged PCM@48kHz f32 from
/// the mixer every 20ms. The consumer is a stub that accumulates audio and
/// reports metadata on close. Future: will stream to a Whisper HTTP endpoint.
pub fn spawn_transcription_tool(
tool_leg_id: String,
call_id: String,
out_tx: OutTx,
) -> (mpsc::Sender<ToolAudioBatch>, JoinHandle<()>) {
let (tx, mut rx) = mpsc::channel::<ToolAudioBatch>(64);
let handle = tokio::spawn(async move {
// Track per-source sample counts for duration reporting.
let mut source_samples: HashMap<String, u64> = HashMap::new();
while let Some(batch) = rx.recv().await {
for source in &batch.sources {
*source_samples.entry(source.leg_id.clone()).or_insert(0) +=
source.pcm_48k.len() as u64;
// TODO: Future — accumulate chunks and stream to Whisper endpoint.
// For now, the audio is received and counted but not processed.
}
}
// Channel closed — report metadata.
let sources: Vec<serde_json::Value> = source_samples
.iter()
.map(|(leg_id, samples)| {
serde_json::json!({
"source_leg_id": leg_id,
"duration_ms": (samples * 1000) / 48000,
})
})
.collect();
emit_event(
&out_tx,
"tool_transcription_done",
serde_json::json!({
"call_id": call_id,
"tool_leg_id": tool_leg_id,
"sources": sources,
}),
);
});
(tx, handle)
}

View File

@@ -0,0 +1,138 @@
//! Text-to-speech engine — synthesizes text to WAV files using Kokoro neural TTS.
//!
//! The model is loaded lazily on first use. If the model/voices files are not
//! present, the generate command returns an error and the TS side falls back
//! to espeak-ng.
use kokoro_tts::{KokoroTts, Voice};
use std::path::Path;
/// Wraps the Kokoro TTS engine with lazy model loading.
pub struct TtsEngine {
tts: Option<KokoroTts>,
/// Path that was used to load the current model (for cache invalidation).
loaded_model_path: String,
loaded_voices_path: String,
}
impl TtsEngine {
pub fn new() -> Self {
Self {
tts: None,
loaded_model_path: String::new(),
loaded_voices_path: String::new(),
}
}
/// Generate a WAV file from text.
///
/// Params (from IPC JSON):
/// - `model`: path to the ONNX model file
/// - `voices`: path to the voices.bin file
/// - `voice`: voice name (e.g. "af_bella")
/// - `text`: text to synthesize
/// - `output`: output WAV file path
pub async fn generate(&mut self, params: &serde_json::Value) -> Result<serde_json::Value, String> {
let model_path = params.get("model").and_then(|v| v.as_str())
.ok_or("missing 'model' param")?;
let voices_path = params.get("voices").and_then(|v| v.as_str())
.ok_or("missing 'voices' param")?;
let voice_name = params.get("voice").and_then(|v| v.as_str())
.unwrap_or("af_bella");
let text = params.get("text").and_then(|v| v.as_str())
.ok_or("missing 'text' param")?;
let output_path = params.get("output").and_then(|v| v.as_str())
.ok_or("missing 'output' param")?;
if text.is_empty() {
return Err("empty text".into());
}
// Check that model/voices files exist.
if !Path::new(model_path).exists() {
return Err(format!("model not found: {model_path}"));
}
if !Path::new(voices_path).exists() {
return Err(format!("voices not found: {voices_path}"));
}
// Lazy-load or reload if paths changed.
if self.tts.is_none()
|| self.loaded_model_path != model_path
|| self.loaded_voices_path != voices_path
{
eprintln!("[tts] loading model: {model_path}");
let tts = KokoroTts::new(model_path, voices_path)
.await
.map_err(|e| format!("model load failed: {e:?}"))?;
self.tts = Some(tts);
self.loaded_model_path = model_path.to_string();
self.loaded_voices_path = voices_path.to_string();
}
let tts = self.tts.as_ref().unwrap();
let voice = select_voice(voice_name);
eprintln!("[tts] synthesizing voice '{voice_name}': \"{text}\"");
let (samples, duration) = tts.synth(text, voice)
.await
.map_err(|e| format!("synthesis failed: {e:?}"))?;
eprintln!("[tts] synthesized {} samples in {duration:?}", samples.len());
// Write 24kHz 16-bit mono WAV.
let spec = hound::WavSpec {
channels: 1,
sample_rate: 24000,
bits_per_sample: 16,
sample_format: hound::SampleFormat::Int,
};
let mut writer = hound::WavWriter::create(output_path, spec)
.map_err(|e| format!("WAV create failed: {e}"))?;
for &sample in &samples {
let s16 = (sample * 32767.0).round().clamp(-32768.0, 32767.0) as i16;
writer.write_sample(s16).map_err(|e| format!("WAV write: {e}"))?;
}
writer.finalize().map_err(|e| format!("WAV finalize: {e}"))?;
eprintln!("[tts] wrote {output_path}");
Ok(serde_json::json!({ "output": output_path }))
}
}
/// Map voice name string to Kokoro Voice enum variant.
fn select_voice(name: &str) -> Voice {
match name {
"af_bella" => Voice::AfBella(1.0),
"af_heart" => Voice::AfHeart(1.0),
"af_jessica" => Voice::AfJessica(1.0),
"af_nicole" => Voice::AfNicole(1.0),
"af_nova" => Voice::AfNova(1.0),
"af_sarah" => Voice::AfSarah(1.0),
"af_sky" => Voice::AfSky(1.0),
"af_river" => Voice::AfRiver(1.0),
"af_alloy" => Voice::AfAlloy(1.0),
"af_aoede" => Voice::AfAoede(1.0),
"af_kore" => Voice::AfKore(1.0),
"am_adam" => Voice::AmAdam(1.0),
"am_echo" => Voice::AmEcho(1.0),
"am_eric" => Voice::AmEric(1.0),
"am_fenrir" => Voice::AmFenrir(1.0),
"am_liam" => Voice::AmLiam(1.0),
"am_michael" => Voice::AmMichael(1.0),
"am_onyx" => Voice::AmOnyx(1.0),
"am_puck" => Voice::AmPuck(1.0),
"bf_alice" => Voice::BfAlice(1.0),
"bf_emma" => Voice::BfEmma(1.0),
"bf_isabella" => Voice::BfIsabella(1.0),
"bf_lily" => Voice::BfLily(1.0),
"bm_daniel" => Voice::BmDaniel(1.0),
"bm_fable" => Voice::BmFable(1.0),
"bm_george" => Voice::BmGeorge(1.0),
"bm_lewis" => Voice::BmLewis(1.0),
_ => {
eprintln!("[tts] unknown voice '{name}', falling back to af_bella");
Voice::AfBella(1.0)
}
}
}

View File

@@ -0,0 +1,137 @@
//! Voicemail session — answer → play greeting → beep → record → done.
use crate::audio_player::{play_beep, play_wav_file};
use crate::ipc::{emit_event, OutTx};
use crate::recorder::Recorder;
use std::net::SocketAddr;
use std::sync::Arc;
use tokio::net::UdpSocket;
/// Run a voicemail session on an RTP port.
///
/// 1. Plays the greeting WAV file to the caller
/// 2. Plays a beep tone
/// 3. Records the caller's message until BYE or max duration
///
/// The RTP receive loop is separate — it feeds packets to the recorder
/// via the returned channel.
pub async fn run_voicemail_session(
rtp_socket: Arc<UdpSocket>,
provider_media: SocketAddr,
codec_pt: u8,
greeting_wav: Option<String>,
recording_path: String,
max_recording_ms: u64,
call_id: String,
caller_number: String,
out_tx: OutTx,
) {
let ssrc: u32 = rand::random();
emit_event(
&out_tx,
"voicemail_started",
serde_json::json!({
"call_id": call_id,
"caller_number": caller_number,
}),
);
// Step 1: Play greeting.
let mut next_seq: u16 = 0;
let mut next_ts: u32 = 0;
if let Some(wav_path) = &greeting_wav {
match play_wav_file(wav_path, rtp_socket.clone(), provider_media, codec_pt, ssrc).await {
Ok(frames) => {
next_seq = frames as u16;
next_ts = frames * crate::rtp::rtp_clock_increment(codec_pt);
}
Err(e) => {
emit_event(
&out_tx,
"voicemail_error",
serde_json::json!({ "call_id": call_id, "error": format!("greeting: {e}") }),
);
}
}
}
// Step 2: Play beep (1kHz, 500ms).
match play_beep(
rtp_socket.clone(),
provider_media,
codec_pt,
ssrc,
next_seq,
next_ts,
1000,
500,
)
.await
{
Ok((_seq, _ts)) => {}
Err(e) => {
emit_event(
&out_tx,
"voicemail_error",
serde_json::json!({ "call_id": call_id, "error": format!("beep: {e}") }),
);
}
}
// Step 3: Record incoming audio.
let recorder = match Recorder::new(&recording_path, codec_pt, Some(max_recording_ms)) {
Ok(r) => r,
Err(e) => {
emit_event(
&out_tx,
"voicemail_error",
serde_json::json!({ "call_id": call_id, "error": format!("recorder: {e}") }),
);
return;
}
};
// Receive RTP and feed to recorder.
let result = record_from_socket(rtp_socket, recorder, max_recording_ms).await;
// Step 4: Done — emit recording result.
emit_event(
&out_tx,
"recording_done",
serde_json::json!({
"call_id": call_id,
"file_path": result.file_path,
"duration_ms": result.duration_ms,
"caller_number": caller_number,
}),
);
}
/// Read RTP packets from the socket and feed them to the recorder.
/// Returns when the socket errors out (BYE closes the call/socket)
/// or max duration is reached.
async fn record_from_socket(
socket: Arc<UdpSocket>,
mut recorder: Recorder,
max_ms: u64,
) -> crate::recorder::RecordingResult {
let mut buf = vec![0u8; 65535];
let deadline = tokio::time::Instant::now() + tokio::time::Duration::from_millis(max_ms + 2000);
loop {
let timeout = tokio::time::timeout_at(deadline, socket.recv_from(&mut buf));
match timeout.await {
Ok(Ok((n, _addr))) => {
if !recorder.process_rtp(&buf[..n]) {
break; // Max duration reached.
}
}
Ok(Err(_)) => break, // Socket error (closed).
Err(_) => break, // Timeout (max duration + grace).
}
}
recorder.stop()
}

View File

@@ -0,0 +1,327 @@
//! WebRTC engine — manages browser PeerConnections.
//!
//! Audio bridging is now channel-based:
//! - Browser Opus audio → on_track → mixer inbound channel
//! - Mixer outbound channel → Opus RTP → TrackLocalStaticRTP → browser
//!
//! The mixer handles all transcoding. The WebRTC engine just shuttles raw Opus.
use crate::ipc::{emit_event, OutTx};
use crate::mixer::RtpPacket;
use codec_lib::PT_OPUS;
use std::collections::HashMap;
use std::sync::Arc;
use tokio::sync::{mpsc, Mutex};
use webrtc::api::media_engine::MediaEngine;
use webrtc::api::APIBuilder;
use webrtc::ice_transport::ice_candidate::RTCIceCandidateInit;
use webrtc::peer_connection::configuration::RTCConfiguration;
use webrtc::peer_connection::peer_connection_state::RTCPeerConnectionState;
use webrtc::peer_connection::sdp::session_description::RTCSessionDescription;
use webrtc::peer_connection::RTCPeerConnection;
use webrtc::rtp_transceiver::rtp_codec::RTCRtpCodecCapability;
use webrtc::track::track_local::track_local_static_rtp::TrackLocalStaticRTP;
use webrtc::track::track_local::{TrackLocal, TrackLocalWriter};
/// A managed WebRTC session.
struct WebRtcSession {
pc: Arc<RTCPeerConnection>,
local_track: Arc<TrackLocalStaticRTP>,
call_id: Option<String>,
/// Channel sender for forwarding browser Opus audio to the mixer.
/// Set when the session is linked to a call via link_to_mixer().
mixer_tx: Arc<Mutex<Option<mpsc::Sender<RtpPacket>>>>,
}
/// Manages all WebRTC sessions.
pub struct WebRtcEngine {
sessions: HashMap<String, WebRtcSession>,
out_tx: OutTx,
}
impl WebRtcEngine {
pub fn new(out_tx: OutTx) -> Self {
Self {
sessions: HashMap::new(),
out_tx,
}
}
/// Handle a WebRTC offer from a browser — create PeerConnection, return SDP answer.
pub async fn handle_offer(
&mut self,
session_id: &str,
offer_sdp: &str,
) -> Result<String, String> {
let mut media_engine = MediaEngine::default();
media_engine
.register_default_codecs()
.map_err(|e| format!("register codecs: {e}"))?;
let api = APIBuilder::new()
.with_media_engine(media_engine)
.build();
let config = RTCConfiguration {
ice_servers: vec![],
..Default::default()
};
let pc = api
.new_peer_connection(config)
.await
.map_err(|e| format!("create peer connection: {e}"))?;
let pc = Arc::new(pc);
// Local audio track for sending audio to browser (Opus).
let local_track = Arc::new(TrackLocalStaticRTP::new(
RTCRtpCodecCapability {
mime_type: "audio/opus".to_string(),
clock_rate: 48000,
channels: 1,
..Default::default()
},
"audio".to_string(),
"siprouter".to_string(),
));
let _sender = pc
.add_track(local_track.clone() as Arc<dyn TrackLocal + Send + Sync>)
.await
.map_err(|e| format!("add track: {e}"))?;
// Shared mixer channel sender (populated when linked to a call).
let mixer_tx: Arc<Mutex<Option<mpsc::Sender<RtpPacket>>>> =
Arc::new(Mutex::new(None));
// ICE candidate handler.
let out_tx_ice = self.out_tx.clone();
let sid_ice = session_id.to_string();
pc.on_ice_candidate(Box::new(move |candidate| {
let out_tx = out_tx_ice.clone();
let sid = sid_ice.clone();
Box::pin(async move {
if let Some(c) = candidate {
if let Ok(json) = c.to_json() {
emit_event(
&out_tx,
"webrtc_ice_candidate",
serde_json::json!({
"session_id": sid,
"candidate": json.candidate,
"sdp_mid": json.sdp_mid,
"sdp_mline_index": json.sdp_mline_index,
}),
);
}
}
})
}));
// Connection state handler.
let out_tx_state = self.out_tx.clone();
let sid_state = session_id.to_string();
pc.on_peer_connection_state_change(Box::new(move |state| {
let out_tx = out_tx_state.clone();
let sid = sid_state.clone();
Box::pin(async move {
let state_str = match state {
RTCPeerConnectionState::Connected => "connected",
RTCPeerConnectionState::Disconnected => "disconnected",
RTCPeerConnectionState::Failed => "failed",
RTCPeerConnectionState::Closed => "closed",
RTCPeerConnectionState::New => "new",
RTCPeerConnectionState::Connecting => "connecting",
_ => "unknown",
};
emit_event(
&out_tx,
"webrtc_state",
serde_json::json!({ "session_id": sid, "state": state_str }),
);
})
}));
// Track handler — receives Opus audio from the browser.
// Forwards raw Opus payload to the mixer channel (when linked).
let out_tx_track = self.out_tx.clone();
let sid_track = session_id.to_string();
let mixer_tx_for_track = mixer_tx.clone();
pc.on_track(Box::new(move |track, _receiver, _transceiver| {
let out_tx = out_tx_track.clone();
let sid = sid_track.clone();
let mixer_tx = mixer_tx_for_track.clone();
Box::pin(async move {
let codec_info = track.codec();
emit_event(
&out_tx,
"webrtc_track",
serde_json::json!({
"session_id": sid,
"kind": track.kind().to_string(),
"codec": codec_info.capability.mime_type,
}),
);
// Spawn browser→mixer forwarding task.
tokio::spawn(browser_to_mixer_loop(track, mixer_tx, out_tx, sid));
})
}));
// Set remote offer.
let offer = RTCSessionDescription::offer(offer_sdp.to_string())
.map_err(|e| format!("parse offer: {e}"))?;
pc.set_remote_description(offer)
.await
.map_err(|e| format!("set remote description: {e}"))?;
// Create answer.
let answer = pc
.create_answer(None)
.await
.map_err(|e| format!("create answer: {e}"))?;
let answer_sdp = answer.sdp.clone();
pc.set_local_description(answer)
.await
.map_err(|e| format!("set local description: {e}"))?;
self.sessions.insert(
session_id.to_string(),
WebRtcSession {
pc,
local_track,
call_id: None,
mixer_tx,
},
);
Ok(answer_sdp)
}
/// Link a WebRTC session to a call's mixer via channels.
/// - `inbound_tx`: browser audio goes TO the mixer through this channel
/// - `outbound_rx`: mixed audio comes FROM the mixer through this channel
pub async fn link_to_mixer(
&mut self,
session_id: &str,
call_id: &str,
inbound_tx: mpsc::Sender<RtpPacket>,
outbound_rx: mpsc::Receiver<Vec<u8>>,
) -> bool {
let session = match self.sessions.get_mut(session_id) {
Some(s) => s,
None => return false,
};
session.call_id = Some(call_id.to_string());
// Set the mixer sender so the on_track loop starts forwarding.
{
let mut tx = session.mixer_tx.lock().await;
*tx = Some(inbound_tx);
}
// Spawn mixer→browser outbound task.
let local_track = session.local_track.clone();
tokio::spawn(mixer_to_browser_loop(outbound_rx, local_track));
true
}
pub async fn add_ice_candidate(
&self,
session_id: &str,
candidate: &str,
sdp_mid: Option<&str>,
sdp_mline_index: Option<u16>,
) -> Result<(), String> {
let session = self
.sessions
.get(session_id)
.ok_or_else(|| format!("session {session_id} not found"))?;
let init = RTCIceCandidateInit {
candidate: candidate.to_string(),
sdp_mid: sdp_mid.map(|s| s.to_string()),
sdp_mline_index,
..Default::default()
};
session
.pc
.add_ice_candidate(init)
.await
.map_err(|e| format!("add ICE: {e}"))
}
pub async fn close_session(&mut self, session_id: &str) -> Result<(), String> {
if let Some(session) = self.sessions.remove(session_id) {
session.pc.close().await.map_err(|e| format!("close: {e}"))?;
}
Ok(())
}
}
/// Browser → Mixer audio forwarding loop.
/// Reads Opus RTP from the browser track, sends raw Opus payload to the mixer channel.
async fn browser_to_mixer_loop(
track: Arc<webrtc::track::track_remote::TrackRemote>,
mixer_tx: Arc<Mutex<Option<mpsc::Sender<RtpPacket>>>>,
out_tx: OutTx,
session_id: String,
) {
let mut buf = vec![0u8; 1500];
let mut count = 0u64;
loop {
match track.read(&mut buf).await {
Ok((rtp_packet, _attributes)) => {
count += 1;
let payload = &rtp_packet.payload;
if payload.is_empty() {
continue;
}
// Send raw Opus payload to mixer (if linked).
let tx = mixer_tx.lock().await;
if let Some(ref tx) = *tx {
let _ = tx
.send(RtpPacket {
payload: payload.to_vec(),
payload_type: PT_OPUS,
marker: rtp_packet.header.marker,
seq: rtp_packet.header.sequence_number,
timestamp: rtp_packet.header.timestamp,
})
.await;
}
drop(tx);
if count == 1 || count == 50 || count % 500 == 0 {
emit_event(
&out_tx,
"webrtc_audio_rx",
serde_json::json!({
"session_id": session_id,
"direction": "browser_to_mixer",
"packet_count": count,
}),
);
}
}
Err(_) => break, // Track ended.
}
}
}
/// Mixer → Browser audio forwarding loop.
/// Reads Opus-encoded RTP packets from the mixer and writes to the WebRTC track.
async fn mixer_to_browser_loop(
mut outbound_rx: mpsc::Receiver<Vec<u8>>,
local_track: Arc<TrackLocalStaticRTP>,
) {
while let Some(rtp_data) = outbound_rx.recv().await {
let _ = local_track.write(&rtp_data).await;
}
}

View File

@@ -0,0 +1,8 @@
[package]
name = "sip-proto"
version = "0.1.0"
edition = "2021"
[dependencies]
md-5 = "0.10"
rand = "0.8"

View File

@@ -0,0 +1,408 @@
//! SIP dialog state machine (RFC 3261 §12).
//!
//! Tracks local/remote tags, CSeq counters, route set, and remote target.
//! Provides methods to build in-dialog requests (BYE, re-INVITE, ACK, CANCEL).
//!
//! Ported from ts/sip/dialog.ts.
use crate::helpers::{generate_branch, generate_tag};
use crate::message::SipMessage;
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum DialogState {
Early,
Confirmed,
Terminated,
}
/// SIP dialog state per RFC 3261 §12.
#[derive(Debug, Clone)]
pub struct SipDialog {
pub call_id: String,
pub local_tag: String,
pub remote_tag: Option<String>,
pub local_uri: String,
pub remote_uri: String,
pub local_cseq: u32,
pub remote_cseq: u32,
pub route_set: Vec<String>,
pub remote_target: String,
pub state: DialogState,
pub local_host: String,
pub local_port: u16,
}
impl SipDialog {
/// Create a dialog from an INVITE we are sending (UAC side).
/// The dialog enters Early state; call `process_response()` when responses arrive.
pub fn from_uac_invite(invite: &SipMessage, local_host: &str, local_port: u16) -> Self {
let from = invite.get_header("From").unwrap_or("");
let to = invite.get_header("To").unwrap_or("");
let local_cseq = invite
.get_header("CSeq")
.and_then(|c| c.split_whitespace().next())
.and_then(|s| s.parse().ok())
.unwrap_or(1);
Self {
call_id: invite.call_id().to_string(),
local_tag: SipMessage::extract_tag(from)
.map(|s| s.to_string())
.unwrap_or_else(generate_tag),
remote_tag: None,
local_uri: SipMessage::extract_uri(from)
.unwrap_or("")
.to_string(),
remote_uri: SipMessage::extract_uri(to).unwrap_or("").to_string(),
local_cseq,
remote_cseq: 0,
route_set: Vec::new(),
remote_target: invite
.request_uri()
.or_else(|| SipMessage::extract_uri(to))
.unwrap_or("")
.to_string(),
state: DialogState::Early,
local_host: local_host.to_string(),
local_port,
}
}
/// Create a dialog from an INVITE we received (UAS side).
pub fn from_uas_invite(
invite: &SipMessage,
local_tag: &str,
local_host: &str,
local_port: u16,
) -> Self {
let from = invite.get_header("From").unwrap_or("");
let to = invite.get_header("To").unwrap_or("");
let contact = invite.get_header("Contact");
let remote_target = contact
.and_then(SipMessage::extract_uri)
.or_else(|| SipMessage::extract_uri(from))
.unwrap_or("")
.to_string();
Self {
call_id: invite.call_id().to_string(),
local_tag: local_tag.to_string(),
remote_tag: SipMessage::extract_tag(from).map(|s| s.to_string()),
local_uri: SipMessage::extract_uri(to).unwrap_or("").to_string(),
remote_uri: SipMessage::extract_uri(from).unwrap_or("").to_string(),
local_cseq: 0,
remote_cseq: 0,
route_set: Vec::new(),
remote_target,
state: DialogState::Early,
local_host: local_host.to_string(),
local_port,
}
}
/// Update dialog state from a received response.
pub fn process_response(&mut self, response: &SipMessage) {
let to = response.get_header("To").unwrap_or("");
let tag = SipMessage::extract_tag(to).map(|s| s.to_string());
let code = response.status_code().unwrap_or(0);
// Always update remoteTag from 2xx (RFC 3261 §12.1.2).
if let Some(ref t) = tag {
if code >= 200 && code < 300 {
self.remote_tag = Some(t.clone());
} else if self.remote_tag.is_none() {
self.remote_tag = Some(t.clone());
}
}
// Update remote target from Contact.
if let Some(contact) = response.get_header("Contact") {
if let Some(uri) = SipMessage::extract_uri(contact) {
self.remote_target = uri.to_string();
}
}
// Record-Route → route set (in reverse for UAC).
if self.state == DialogState::Early {
let rr: Vec<String> = response
.headers
.iter()
.filter(|(n, _)| n.to_ascii_lowercase() == "record-route")
.map(|(_, v)| v.clone())
.collect();
if !rr.is_empty() {
let mut reversed = rr;
reversed.reverse();
self.route_set = reversed;
}
}
if code >= 200 && code < 300 {
self.state = DialogState::Confirmed;
} else if code >= 300 {
self.state = DialogState::Terminated;
}
}
/// Build an in-dialog request (BYE, re-INVITE, INFO, ...).
/// Automatically increments the local CSeq.
pub fn create_request(
&mut self,
method: &str,
body: Option<&str>,
content_type: Option<&str>,
extra_headers: Option<Vec<(String, String)>>,
) -> SipMessage {
self.local_cseq += 1;
let branch = generate_branch();
let remote_tag_str = self
.remote_tag
.as_ref()
.map(|t| format!(";tag={t}"))
.unwrap_or_default();
let mut headers = vec![
(
"Via".to_string(),
format!(
"SIP/2.0/UDP {}:{};branch={branch};rport",
self.local_host, self.local_port
),
),
(
"From".to_string(),
format!("<{}>;tag={}", self.local_uri, self.local_tag),
),
(
"To".to_string(),
format!("<{}>{remote_tag_str}", self.remote_uri),
),
("Call-ID".to_string(), self.call_id.clone()),
(
"CSeq".to_string(),
format!("{} {method}", self.local_cseq),
),
("Max-Forwards".to_string(), "70".to_string()),
];
for route in &self.route_set {
headers.push(("Route".to_string(), route.clone()));
}
headers.push((
"Contact".to_string(),
format!("<sip:{}:{}>", self.local_host, self.local_port),
));
if let Some(extra) = extra_headers {
headers.extend(extra);
}
let body_str = body.unwrap_or("");
if !body_str.is_empty() {
if let Some(ct) = content_type {
headers.push(("Content-Type".to_string(), ct.to_string()));
}
}
headers.push(("Content-Length".to_string(), body_str.len().to_string()));
let ruri = self.resolve_ruri();
SipMessage::new(
format!("{method} {ruri} SIP/2.0"),
headers,
body_str.to_string(),
)
}
/// Build an ACK for a 2xx response to INVITE (RFC 3261 §13.2.2.4).
pub fn create_ack(&self) -> SipMessage {
let branch = generate_branch();
let remote_tag_str = self
.remote_tag
.as_ref()
.map(|t| format!(";tag={t}"))
.unwrap_or_default();
let mut headers = vec![
(
"Via".to_string(),
format!(
"SIP/2.0/UDP {}:{};branch={branch};rport",
self.local_host, self.local_port
),
),
(
"From".to_string(),
format!("<{}>;tag={}", self.local_uri, self.local_tag),
),
(
"To".to_string(),
format!("<{}>{remote_tag_str}", self.remote_uri),
),
("Call-ID".to_string(), self.call_id.clone()),
(
"CSeq".to_string(),
format!("{} ACK", self.local_cseq),
),
("Max-Forwards".to_string(), "70".to_string()),
];
for route in &self.route_set {
headers.push(("Route".to_string(), route.clone()));
}
headers.push(("Content-Length".to_string(), "0".to_string()));
let ruri = self.resolve_ruri();
SipMessage::new(format!("ACK {ruri} SIP/2.0"), headers, String::new())
}
/// Build a CANCEL for the original INVITE (same branch, CSeq).
pub fn create_cancel(&self, original_invite: &SipMessage) -> SipMessage {
let via = original_invite.get_header("Via").unwrap_or("").to_string();
let from = original_invite.get_header("From").unwrap_or("").to_string();
let to = original_invite.get_header("To").unwrap_or("").to_string();
let headers = vec![
("Via".to_string(), via),
("From".to_string(), from),
("To".to_string(), to),
("Call-ID".to_string(), self.call_id.clone()),
(
"CSeq".to_string(),
format!("{} CANCEL", self.local_cseq),
),
("Max-Forwards".to_string(), "70".to_string()),
("Content-Length".to_string(), "0".to_string()),
];
let ruri = original_invite
.request_uri()
.unwrap_or(&self.remote_target)
.to_string();
SipMessage::new(
format!("CANCEL {ruri} SIP/2.0"),
headers,
String::new(),
)
}
/// Transition the dialog to terminated state.
pub fn terminate(&mut self) {
self.state = DialogState::Terminated;
}
/// Resolve Request-URI from route set or remote target.
fn resolve_ruri(&self) -> &str {
if !self.route_set.is_empty() {
if let Some(top_route) = SipMessage::extract_uri(&self.route_set[0]) {
if top_route.contains(";lr") {
return &self.remote_target; // loose routing
}
return top_route; // strict routing
}
}
&self.remote_target
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::message::RequestOptions;
fn make_invite() -> SipMessage {
SipMessage::create_request(
"INVITE",
"sip:callee@host",
RequestOptions {
via_host: "192.168.1.1".to_string(),
via_port: 5070,
via_transport: None,
via_branch: Some("z9hG4bK-test".to_string()),
from_uri: "sip:caller@proxy".to_string(),
from_display_name: None,
from_tag: Some("from-tag".to_string()),
to_uri: "sip:callee@host".to_string(),
to_display_name: None,
to_tag: None,
call_id: Some("test-dialog-call".to_string()),
cseq: Some(1),
contact: Some("<sip:caller@192.168.1.1:5070>".to_string()),
max_forwards: None,
body: None,
content_type: None,
extra_headers: None,
},
)
}
#[test]
fn uac_dialog_lifecycle() {
let invite = make_invite();
let mut dialog = SipDialog::from_uac_invite(&invite, "192.168.1.1", 5070);
assert_eq!(dialog.state, DialogState::Early);
assert_eq!(dialog.call_id, "test-dialog-call");
assert_eq!(dialog.local_tag, "from-tag");
assert!(dialog.remote_tag.is_none());
// Simulate 200 OK
let response = SipMessage::create_response(
200,
"OK",
&invite,
Some(crate::message::ResponseOptions {
to_tag: Some("remote-tag".to_string()),
contact: Some("<sip:callee@10.0.0.1:5060>".to_string()),
..Default::default()
}),
);
dialog.process_response(&response);
assert_eq!(dialog.state, DialogState::Confirmed);
assert_eq!(dialog.remote_tag.as_deref(), Some("remote-tag"));
assert_eq!(dialog.remote_target, "sip:callee@10.0.0.1:5060");
}
#[test]
fn create_bye() {
let invite = make_invite();
let mut dialog = SipDialog::from_uac_invite(&invite, "192.168.1.1", 5070);
dialog.remote_tag = Some("remote-tag".to_string());
dialog.state = DialogState::Confirmed;
let bye = dialog.create_request("BYE", None, None, None);
assert_eq!(bye.method(), Some("BYE"));
assert_eq!(bye.call_id(), "test-dialog-call");
assert_eq!(dialog.local_cseq, 2);
let to = bye.get_header("To").unwrap();
assert!(to.contains("tag=remote-tag"));
}
#[test]
fn create_ack() {
let invite = make_invite();
let mut dialog = SipDialog::from_uac_invite(&invite, "192.168.1.1", 5070);
dialog.remote_tag = Some("remote-tag".to_string());
let ack = dialog.create_ack();
assert_eq!(ack.method(), Some("ACK"));
assert!(ack.get_header("CSeq").unwrap().contains("ACK"));
}
#[test]
fn create_cancel() {
let invite = make_invite();
let dialog = SipDialog::from_uac_invite(&invite, "192.168.1.1", 5070);
let cancel = dialog.create_cancel(&invite);
assert_eq!(cancel.method(), Some("CANCEL"));
assert!(cancel.get_header("CSeq").unwrap().contains("CANCEL"));
assert!(cancel.start_line.contains("sip:callee@host"));
}
}

View File

@@ -0,0 +1,339 @@
//! SIP helper utilities — ID generation, codec registry, SDP builder,
//! Digest authentication, SDP parser, and MWI body builder.
use md5::{Digest, Md5};
use rand::Rng;
// ---- ID generators ---------------------------------------------------------
/// Generate a random SIP Call-ID (32 hex chars).
pub fn generate_call_id(domain: Option<&str>) -> String {
let id = random_hex(16);
match domain {
Some(d) => format!("{id}@{d}"),
None => id,
}
}
/// Generate a random SIP From/To tag (16 hex chars).
pub fn generate_tag() -> String {
random_hex(8)
}
/// Generate an RFC 3261 compliant Via branch (starts with `z9hG4bK` magic cookie).
pub fn generate_branch() -> String {
format!("z9hG4bK-{}", random_hex(8))
}
fn random_hex(bytes: usize) -> String {
let mut rng = rand::thread_rng();
(0..bytes).map(|_| format!("{:02x}", rng.gen::<u8>())).collect()
}
// ---- Codec registry --------------------------------------------------------
/// Look up the rtpmap name for a static payload type.
pub fn codec_name(pt: u8) -> &'static str {
match pt {
0 => "PCMU/8000",
3 => "GSM/8000",
4 => "G723/8000",
8 => "PCMA/8000",
9 => "G722/8000",
18 => "G729/8000",
101 => "telephone-event/8000",
_ => "unknown",
}
}
// ---- SDP builder -----------------------------------------------------------
/// Options for building an SDP body.
pub struct SdpOptions<'a> {
pub ip: &'a str,
pub port: u16,
pub payload_types: &'a [u8],
pub session_id: Option<&'a str>,
pub session_name: Option<&'a str>,
pub direction: Option<&'a str>,
pub attributes: &'a [&'a str],
}
impl<'a> Default for SdpOptions<'a> {
fn default() -> Self {
Self {
ip: "0.0.0.0",
port: 0,
payload_types: &[9, 0, 8, 101],
session_id: None,
session_name: None,
direction: None,
attributes: &[],
}
}
}
/// Build a minimal SDP body suitable for SIP INVITE offers/answers.
pub fn build_sdp(opts: &SdpOptions) -> String {
let session_id = opts
.session_id
.map(|s| s.to_string())
.unwrap_or_else(|| format!("{}", rand::thread_rng().gen_range(0..1_000_000_000u64)));
let session_name = opts.session_name.unwrap_or("-");
let direction = opts.direction.unwrap_or("sendrecv");
let pts: Vec<String> = opts.payload_types.iter().map(|pt| pt.to_string()).collect();
let mut lines = vec![
"v=0".to_string(),
format!("o=- {session_id} {session_id} IN IP4 {}", opts.ip),
format!("s={session_name}"),
format!("c=IN IP4 {}", opts.ip),
"t=0 0".to_string(),
format!("m=audio {} RTP/AVP {}", opts.port, pts.join(" ")),
];
for &pt in opts.payload_types {
let name = codec_name(pt);
if name != "unknown" {
lines.push(format!("a=rtpmap:{pt} {name}"));
}
if pt == 101 {
lines.push("a=fmtp:101 0-16".to_string());
}
}
lines.push(format!("a={direction}"));
for attr in opts.attributes {
lines.push(format!("a={attr}"));
}
lines.push(String::new()); // trailing CRLF
lines.join("\r\n")
}
// ---- SIP Digest authentication (RFC 2617) ----------------------------------
/// Parsed fields from a Proxy-Authenticate or WWW-Authenticate header.
#[derive(Debug, Clone)]
pub struct DigestChallenge {
pub realm: String,
pub nonce: String,
pub algorithm: Option<String>,
pub opaque: Option<String>,
pub qop: Option<String>,
}
/// Parse a `Proxy-Authenticate` or `WWW-Authenticate` header value.
pub fn parse_digest_challenge(header: &str) -> Option<DigestChallenge> {
let lower = header.to_ascii_lowercase();
if !lower.starts_with("digest ") {
return None;
}
let params = &header[7..];
let get = |key: &str| -> Option<String> {
// Try quoted value first.
let pat = format!("{}=", key);
if let Some(pos) = params.to_ascii_lowercase().find(&pat) {
let after = &params[pos + pat.len()..];
let after = after.trim_start();
if after.starts_with('"') {
let end = after[1..].find('"')?;
return Some(after[1..1 + end].to_string());
}
// Unquoted value.
let end = after.find(|c: char| c == ',' || c.is_whitespace()).unwrap_or(after.len());
return Some(after[..end].to_string());
}
None
};
let realm = get("realm")?;
let nonce = get("nonce")?;
Some(DigestChallenge {
realm,
nonce,
algorithm: get("algorithm"),
opaque: get("opaque"),
qop: get("qop"),
})
}
fn md5_hex(s: &str) -> String {
let mut hasher = Md5::new();
hasher.update(s.as_bytes());
format!("{:x}", hasher.finalize())
}
/// Compute a SIP Digest Authorization header value.
pub fn compute_digest_auth(
username: &str,
password: &str,
realm: &str,
nonce: &str,
method: &str,
uri: &str,
algorithm: Option<&str>,
opaque: Option<&str>,
) -> String {
let ha1 = md5_hex(&format!("{username}:{realm}:{password}"));
let ha2 = md5_hex(&format!("{method}:{uri}"));
let response = md5_hex(&format!("{ha1}:{nonce}:{ha2}"));
let alg = algorithm.unwrap_or("MD5");
let mut header = format!(
"Digest username=\"{username}\", realm=\"{realm}\", \
nonce=\"{nonce}\", uri=\"{uri}\", response=\"{response}\", \
algorithm={alg}"
);
if let Some(op) = opaque {
header.push_str(&format!(", opaque=\"{op}\""));
}
header
}
// ---- SDP parser ------------------------------------------------------------
use crate::Endpoint;
/// Parse the audio media port, connection address, and preferred codec from an SDP body.
pub fn parse_sdp_endpoint(sdp: &str) -> Option<Endpoint> {
let mut addr: Option<&str> = None;
let mut port: Option<u16> = None;
let mut codec_pt: Option<u8> = None;
let normalized = sdp.replace("\r\n", "\n");
for raw in normalized.split('\n') {
let line = raw.trim();
if let Some(rest) = line.strip_prefix("c=IN IP4 ") {
addr = Some(rest.trim());
} else if let Some(rest) = line.strip_prefix("m=audio ") {
// m=audio <port> RTP/AVP <pt1> [<pt2> ...]
let parts: Vec<&str> = rest.split_whitespace().collect();
if !parts.is_empty() {
port = parts[0].parse().ok();
}
// parts[1] is "RTP/AVP" or similar, parts[2..] are payload types.
// The first PT is the preferred codec.
if parts.len() > 2 {
codec_pt = parts[2].parse::<u8>().ok();
}
}
}
match (addr, port) {
(Some(a), Some(p)) => Some(Endpoint {
address: a.to_string(),
port: p,
codec_pt,
}),
_ => None,
}
}
// ---- MWI (RFC 3842) --------------------------------------------------------
/// Build the body and extra headers for an MWI NOTIFY (RFC 3842 message-summary).
pub struct MwiResult {
pub body: String,
pub content_type: &'static str,
pub extra_headers: Vec<(String, String)>,
}
pub fn build_mwi_body(
new_messages: u32,
old_messages: u32,
account_uri: &str,
) -> MwiResult {
let waiting = if new_messages > 0 { "yes" } else { "no" };
let body = format!(
"Messages-Waiting: {waiting}\r\n\
Message-Account: {account_uri}\r\n\
Voice-Message: {new_messages}/{old_messages}\r\n"
);
MwiResult {
body,
content_type: "application/simple-message-summary",
extra_headers: vec![
("Event".to_string(), "message-summary".to_string()),
(
"Subscription-State".to_string(),
"terminated;reason=noresource".to_string(),
),
],
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_generate_branch_has_magic_cookie() {
let branch = generate_branch();
assert!(branch.starts_with("z9hG4bK-"));
assert!(branch.len() > 8);
}
#[test]
fn test_codec_name() {
assert_eq!(codec_name(0), "PCMU/8000");
assert_eq!(codec_name(9), "G722/8000");
assert_eq!(codec_name(101), "telephone-event/8000");
assert_eq!(codec_name(255), "unknown");
}
#[test]
fn test_build_sdp() {
let sdp = build_sdp(&SdpOptions {
ip: "192.168.1.1",
port: 20000,
payload_types: &[9, 0, 101],
..Default::default()
});
assert!(sdp.contains("m=audio 20000 RTP/AVP 9 0 101"));
assert!(sdp.contains("c=IN IP4 192.168.1.1"));
assert!(sdp.contains("a=rtpmap:9 G722/8000"));
assert!(sdp.contains("a=fmtp:101 0-16"));
assert!(sdp.contains("a=sendrecv"));
}
#[test]
fn test_parse_digest_challenge() {
let header = r#"Digest realm="asterisk", nonce="abc123", algorithm=MD5, opaque="xyz""#;
let ch = parse_digest_challenge(header).unwrap();
assert_eq!(ch.realm, "asterisk");
assert_eq!(ch.nonce, "abc123");
assert_eq!(ch.algorithm.as_deref(), Some("MD5"));
assert_eq!(ch.opaque.as_deref(), Some("xyz"));
}
#[test]
fn test_compute_digest_auth() {
let auth = compute_digest_auth(
"user", "pass", "realm", "nonce", "REGISTER", "sip:host", None, None,
);
assert!(auth.starts_with("Digest "));
assert!(auth.contains("username=\"user\""));
assert!(auth.contains("realm=\"realm\""));
assert!(auth.contains("response=\""));
}
#[test]
fn test_parse_sdp_endpoint() {
let sdp = "v=0\r\nc=IN IP4 10.0.0.1\r\nm=audio 5060 RTP/AVP 0\r\n";
let ep = parse_sdp_endpoint(sdp).unwrap();
assert_eq!(ep.address, "10.0.0.1");
assert_eq!(ep.port, 5060);
}
#[test]
fn test_build_mwi_body() {
let mwi = build_mwi_body(3, 5, "sip:user@host");
assert!(mwi.body.contains("Messages-Waiting: yes"));
assert!(mwi.body.contains("Voice-Message: 3/5"));
assert_eq!(mwi.content_type, "application/simple-message-summary");
}
}

View File

@@ -0,0 +1,19 @@
//! SIP protocol library for the proxy engine.
//!
//! Provides SIP message parsing/serialization, dialog state management,
//! SDP handling, Digest authentication, and URI rewriting.
//! Ported from the TypeScript `ts/sip/` library.
pub mod message;
pub mod dialog;
pub mod helpers;
pub mod rewrite;
/// Network endpoint (address + port + optional negotiated codec).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Endpoint {
pub address: String,
pub port: u16,
/// First payload type from the SDP `m=audio` line (the preferred codec).
pub codec_pt: Option<u8>,
}

View File

@@ -0,0 +1,563 @@
//! SIP message parsing, serialization, inspection, mutation, and factory methods.
//!
//! Ported from ts/sip/message.ts.
use crate::helpers::{generate_branch, generate_call_id, generate_tag};
/// A parsed SIP message (request or response).
#[derive(Debug, Clone)]
pub struct SipMessage {
pub start_line: String,
pub headers: Vec<(String, String)>,
pub body: String,
}
impl SipMessage {
pub fn new(start_line: String, headers: Vec<(String, String)>, body: String) -> Self {
Self { start_line, headers, body }
}
// ---- Parsing -----------------------------------------------------------
/// Parse a raw buffer into a SipMessage. Returns None for invalid data.
pub fn parse(buf: &[u8]) -> Option<Self> {
if buf.is_empty() {
return None;
}
// First byte must be ASCII A-z.
if buf[0] < 0x41 || buf[0] > 0x7a {
return None;
}
let text = std::str::from_utf8(buf).ok()?;
let (head, body) = if let Some(sep) = text.find("\r\n\r\n") {
(&text[..sep], &text[sep + 4..])
} else if let Some(sep) = text.find("\n\n") {
(&text[..sep], &text[sep + 2..])
} else {
(text, "")
};
let normalized = head.replace("\r\n", "\n");
let lines: Vec<&str> = normalized.split('\n').collect();
if lines.is_empty() || lines[0].is_empty() {
return None;
}
let start_line = lines[0];
// Validate: must be a SIP request or response start line.
if !is_sip_first_line(start_line) {
return None;
}
let mut headers = Vec::new();
for &line in &lines[1..] {
let trimmed = line.trim();
if trimmed.is_empty() {
continue;
}
if let Some(colon) = line.find(':') {
let name = line[..colon].trim().to_string();
let value = line[colon + 1..].trim().to_string();
headers.push((name, value));
}
}
Some(SipMessage {
start_line: start_line.to_string(),
headers,
body: body.to_string(),
})
}
// ---- Serialization -----------------------------------------------------
/// Serialize the message to a byte buffer suitable for UDP transmission.
pub fn serialize(&self) -> Vec<u8> {
let mut head = self.start_line.clone();
for (name, value) in &self.headers {
head.push_str("\r\n");
head.push_str(name);
head.push_str(": ");
head.push_str(value);
}
head.push_str("\r\n\r\n");
let mut buf = head.into_bytes();
if !self.body.is_empty() {
buf.extend_from_slice(self.body.as_bytes());
}
buf
}
// ---- Inspectors --------------------------------------------------------
pub fn is_request(&self) -> bool {
!self.start_line.starts_with("SIP/")
}
pub fn is_response(&self) -> bool {
self.start_line.starts_with("SIP/")
}
/// Request method (INVITE, REGISTER, ...) or None for responses.
pub fn method(&self) -> Option<&str> {
if !self.is_request() {
return None;
}
self.start_line.split_whitespace().next()
}
/// Response status code or None for requests.
pub fn status_code(&self) -> Option<u16> {
if !self.is_response() {
return None;
}
self.start_line
.split_whitespace()
.nth(1)
.and_then(|s| s.parse().ok())
}
pub fn call_id(&self) -> &str {
self.get_header("Call-ID").unwrap_or("noid")
}
/// Method from the CSeq header (e.g. "INVITE").
pub fn cseq_method(&self) -> Option<&str> {
let cseq = self.get_header("CSeq")?;
cseq.split_whitespace().nth(1)
}
/// True for INVITE, SUBSCRIBE, REFER, NOTIFY, UPDATE.
pub fn is_dialog_establishing(&self) -> bool {
matches!(
self.method(),
Some("INVITE" | "SUBSCRIBE" | "REFER" | "NOTIFY" | "UPDATE")
)
}
/// True when the body carries an SDP payload.
pub fn has_sdp_body(&self) -> bool {
if self.body.is_empty() {
return false;
}
let ct = self.get_header("Content-Type").unwrap_or("");
ct.to_ascii_lowercase().starts_with("application/sdp")
}
// ---- Header accessors --------------------------------------------------
/// Get the first header value matching `name` (case-insensitive).
pub fn get_header(&self, name: &str) -> Option<&str> {
let nl = name.to_ascii_lowercase();
for (n, v) in &self.headers {
if n.to_ascii_lowercase() == nl {
return Some(v.as_str());
}
}
None
}
/// Overwrites the first header with the given name, or appends it.
pub fn set_header(&mut self, name: &str, value: &str) -> &mut Self {
let nl = name.to_ascii_lowercase();
for h in &mut self.headers {
if h.0.to_ascii_lowercase() == nl {
h.1 = value.to_string();
return self;
}
}
self.headers.push((name.to_string(), value.to_string()));
self
}
/// Inserts a header at the top of the header list.
pub fn prepend_header(&mut self, name: &str, value: &str) -> &mut Self {
self.headers.insert(0, (name.to_string(), value.to_string()));
self
}
/// Removes all headers with the given name.
pub fn remove_header(&mut self, name: &str) -> &mut Self {
let nl = name.to_ascii_lowercase();
self.headers.retain(|(n, _)| n.to_ascii_lowercase() != nl);
self
}
/// Recalculates Content-Length to match the current body.
pub fn update_content_length(&mut self) -> &mut Self {
let len = self.body.len();
self.set_header("Content-Length", &len.to_string())
}
// ---- Start-line mutation -----------------------------------------------
/// Replace the Request-URI (second token) of a request start line.
pub fn set_request_uri(&mut self, uri: &str) -> &mut Self {
if !self.is_request() {
return self;
}
let parts: Vec<&str> = self.start_line.splitn(3, ' ').collect();
if parts.len() >= 3 {
self.start_line = format!("{} {} {}", parts[0], uri, parts[2]);
}
self
}
/// Returns the Request-URI (second token) of a request start line.
pub fn request_uri(&self) -> Option<&str> {
if !self.is_request() {
return None;
}
self.start_line.split_whitespace().nth(1)
}
// ---- Factory methods ---------------------------------------------------
/// Build a new SIP request.
pub fn create_request(method: &str, request_uri: &str, opts: RequestOptions) -> Self {
let branch = opts.via_branch.unwrap_or_else(|| generate_branch());
let transport = opts.via_transport.unwrap_or_else(|| "UDP".to_string());
let from_tag = opts.from_tag.unwrap_or_else(|| generate_tag());
let call_id = opts.call_id.unwrap_or_else(|| generate_call_id(None));
let cseq = opts.cseq.unwrap_or(1);
let max_forwards = opts.max_forwards.unwrap_or(70);
let from_display = opts
.from_display_name
.map(|d| format!("\"{d}\" "))
.unwrap_or_default();
let to_display = opts
.to_display_name
.map(|d| format!("\"{d}\" "))
.unwrap_or_default();
let to_tag_str = opts
.to_tag
.map(|t| format!(";tag={t}"))
.unwrap_or_default();
let mut headers = vec![
(
"Via".to_string(),
format!(
"SIP/2.0/{transport} {}:{};branch={branch};rport",
opts.via_host, opts.via_port
),
),
(
"From".to_string(),
format!("{from_display}<{}>;tag={from_tag}", opts.from_uri),
),
(
"To".to_string(),
format!("{to_display}<{}>{to_tag_str}", opts.to_uri),
),
("Call-ID".to_string(), call_id),
("CSeq".to_string(), format!("{cseq} {method}")),
("Max-Forwards".to_string(), max_forwards.to_string()),
];
if let Some(contact) = &opts.contact {
headers.push(("Contact".to_string(), contact.clone()));
}
if let Some(extra) = opts.extra_headers {
headers.extend(extra);
}
let body = opts.body.unwrap_or_default();
if !body.is_empty() {
if let Some(ct) = &opts.content_type {
headers.push(("Content-Type".to_string(), ct.clone()));
}
}
headers.push(("Content-Length".to_string(), body.len().to_string()));
SipMessage {
start_line: format!("{method} {request_uri} SIP/2.0"),
headers,
body,
}
}
/// Build a SIP response to an incoming request.
/// Copies Via, From, To, Call-ID, and CSeq from the original request.
pub fn create_response(
status_code: u16,
reason_phrase: &str,
request: &SipMessage,
opts: Option<ResponseOptions>,
) -> Self {
let opts = opts.unwrap_or_default();
let mut headers: Vec<(String, String)> = Vec::new();
// Copy all Via headers (order matters).
for (n, v) in &request.headers {
if n.to_ascii_lowercase() == "via" {
headers.push(("Via".to_string(), v.clone()));
}
}
// From — copied verbatim.
if let Some(from) = request.get_header("From") {
headers.push(("From".to_string(), from.to_string()));
}
// To — add tag if provided and not already present.
let mut to = request.get_header("To").unwrap_or("").to_string();
if let Some(tag) = &opts.to_tag {
if !to.contains("tag=") {
to.push_str(&format!(";tag={tag}"));
}
}
headers.push(("To".to_string(), to));
headers.push(("Call-ID".to_string(), request.call_id().to_string()));
if let Some(cseq) = request.get_header("CSeq") {
headers.push(("CSeq".to_string(), cseq.to_string()));
}
if let Some(contact) = &opts.contact {
headers.push(("Contact".to_string(), contact.clone()));
}
if let Some(extra) = opts.extra_headers {
headers.extend(extra);
}
let body = opts.body.unwrap_or_default();
if !body.is_empty() {
if let Some(ct) = &opts.content_type {
headers.push(("Content-Type".to_string(), ct.clone()));
}
}
headers.push(("Content-Length".to_string(), body.len().to_string()));
SipMessage {
start_line: format!("SIP/2.0 {status_code} {reason_phrase}"),
headers,
body,
}
}
/// Extract the tag from a From or To header value.
pub fn extract_tag(header_value: &str) -> Option<&str> {
let idx = header_value.find(";tag=")?;
let rest = &header_value[idx + 5..];
let end = rest
.find(|c: char| c.is_whitespace() || c == ';' || c == '>')
.unwrap_or(rest.len());
Some(&rest[..end])
}
/// Extract the URI from an addr-spec or name-addr (From/To/Contact).
pub fn extract_uri(header_value: &str) -> Option<&str> {
if let Some(start) = header_value.find('<') {
let end = header_value[start..].find('>')?;
Some(&header_value[start + 1..start + end])
} else {
let trimmed = header_value.trim();
let end = trimmed
.find(|c: char| c == ';' || c == '>')
.unwrap_or(trimmed.len());
let result = &trimmed[..end];
if result.is_empty() { None } else { Some(result) }
}
}
}
/// Options for `SipMessage::create_request`.
pub struct RequestOptions {
pub via_host: String,
pub via_port: u16,
pub via_transport: Option<String>,
pub via_branch: Option<String>,
pub from_uri: String,
pub from_display_name: Option<String>,
pub from_tag: Option<String>,
pub to_uri: String,
pub to_display_name: Option<String>,
pub to_tag: Option<String>,
pub call_id: Option<String>,
pub cseq: Option<u32>,
pub contact: Option<String>,
pub max_forwards: Option<u16>,
pub body: Option<String>,
pub content_type: Option<String>,
pub extra_headers: Option<Vec<(String, String)>>,
}
/// Options for `SipMessage::create_response`.
#[derive(Default)]
pub struct ResponseOptions {
pub to_tag: Option<String>,
pub contact: Option<String>,
pub body: Option<String>,
pub content_type: Option<String>,
pub extra_headers: Option<Vec<(String, String)>>,
}
/// Check if a string matches the SIP first-line pattern.
fn is_sip_first_line(line: &str) -> bool {
// Request: METHOD SP URI SP SIP/X.Y
// Response: SIP/X.Y SP STATUS SP REASON
if line.starts_with("SIP/") {
// Response: SIP/2.0 200 OK
let parts: Vec<&str> = line.splitn(3, ' ').collect();
if parts.len() >= 2 {
return parts[1].chars().all(|c| c.is_ascii_digit());
}
} else {
// Request: INVITE sip:user@host SIP/2.0
let parts: Vec<&str> = line.splitn(3, ' ').collect();
if parts.len() >= 3 {
return parts[0].chars().all(|c| c.is_ascii_uppercase())
&& parts[2].starts_with("SIP/");
}
}
false
}
#[cfg(test)]
mod tests {
use super::*;
const INVITE_RAW: &str = "INVITE sip:user@host SIP/2.0\r\n\
Via: SIP/2.0/UDP 192.168.1.1:5060;branch=z9hG4bK-test\r\n\
From: <sip:caller@host>;tag=abc\r\n\
To: <sip:user@host>\r\n\
Call-ID: test-call-id\r\n\
CSeq: 1 INVITE\r\n\
Content-Length: 0\r\n\r\n";
#[test]
fn parse_invite() {
let msg = SipMessage::parse(INVITE_RAW.as_bytes()).unwrap();
assert!(msg.is_request());
assert!(!msg.is_response());
assert_eq!(msg.method(), Some("INVITE"));
assert_eq!(msg.call_id(), "test-call-id");
assert_eq!(msg.cseq_method(), Some("INVITE"));
assert!(msg.is_dialog_establishing());
assert_eq!(msg.request_uri(), Some("sip:user@host"));
}
#[test]
fn parse_response() {
let raw = "SIP/2.0 200 OK\r\n\
Via: SIP/2.0/UDP 192.168.1.1:5060;branch=z9hG4bK-test\r\n\
From: <sip:caller@host>;tag=abc\r\n\
To: <sip:user@host>;tag=def\r\n\
Call-ID: test-call-id\r\n\
CSeq: 1 INVITE\r\n\
Content-Length: 0\r\n\r\n";
let msg = SipMessage::parse(raw.as_bytes()).unwrap();
assert!(msg.is_response());
assert_eq!(msg.status_code(), Some(200));
assert_eq!(msg.cseq_method(), Some("INVITE"));
}
#[test]
fn serialize_roundtrip() {
let msg = SipMessage::parse(INVITE_RAW.as_bytes()).unwrap();
let serialized = msg.serialize();
let reparsed = SipMessage::parse(&serialized).unwrap();
assert_eq!(reparsed.call_id(), "test-call-id");
assert_eq!(reparsed.method(), Some("INVITE"));
assert_eq!(reparsed.headers.len(), msg.headers.len());
}
#[test]
fn header_mutation() {
let mut msg = SipMessage::parse(INVITE_RAW.as_bytes()).unwrap();
msg.set_header("X-Custom", "value1");
assert_eq!(msg.get_header("X-Custom"), Some("value1"));
msg.set_header("X-Custom", "value2");
assert_eq!(msg.get_header("X-Custom"), Some("value2"));
msg.prepend_header("X-First", "first");
assert_eq!(msg.headers[0].0, "X-First");
msg.remove_header("X-Custom");
assert_eq!(msg.get_header("X-Custom"), None);
}
#[test]
fn set_request_uri() {
let mut msg = SipMessage::parse(INVITE_RAW.as_bytes()).unwrap();
msg.set_request_uri("sip:new@host");
assert_eq!(msg.request_uri(), Some("sip:new@host"));
assert!(msg.start_line.starts_with("INVITE sip:new@host SIP/2.0"));
}
#[test]
fn extract_tag_and_uri() {
assert_eq!(
SipMessage::extract_tag("<sip:user@host>;tag=abc123"),
Some("abc123")
);
assert_eq!(SipMessage::extract_tag("<sip:user@host>"), None);
assert_eq!(
SipMessage::extract_uri("<sip:user@host>"),
Some("sip:user@host")
);
assert_eq!(
SipMessage::extract_uri("\"Name\" <sip:user@host>;tag=abc"),
Some("sip:user@host")
);
}
#[test]
fn create_request_and_response() {
let invite = SipMessage::create_request(
"INVITE",
"sip:user@host",
RequestOptions {
via_host: "192.168.1.1".to_string(),
via_port: 5070,
via_transport: None,
via_branch: None,
from_uri: "sip:caller@proxy".to_string(),
from_display_name: None,
from_tag: Some("mytag".to_string()),
to_uri: "sip:user@host".to_string(),
to_display_name: None,
to_tag: None,
call_id: Some("test-123".to_string()),
cseq: Some(1),
contact: Some("<sip:caller@192.168.1.1:5070>".to_string()),
max_forwards: None,
body: None,
content_type: None,
extra_headers: None,
},
);
assert_eq!(invite.method(), Some("INVITE"));
assert_eq!(invite.call_id(), "test-123");
assert!(invite.get_header("Via").unwrap().contains("192.168.1.1:5070"));
let response = SipMessage::create_response(
200,
"OK",
&invite,
Some(ResponseOptions {
to_tag: Some("remotetag".to_string()),
..Default::default()
}),
);
assert!(response.is_response());
assert_eq!(response.status_code(), Some(200));
let to = response.get_header("To").unwrap();
assert!(to.contains("tag=remotetag"));
}
#[test]
fn has_sdp_body() {
let mut msg = SipMessage::parse(INVITE_RAW.as_bytes()).unwrap();
assert!(!msg.has_sdp_body());
msg.body = "v=0\r\no=- 1 1 IN IP4 0.0.0.0\r\n".to_string();
msg.set_header("Content-Type", "application/sdp");
assert!(msg.has_sdp_body());
}
}

View File

@@ -0,0 +1,130 @@
//! SIP URI and SDP body rewriting helpers.
//!
//! Ported from ts/sip/rewrite.ts.
use crate::Endpoint;
/// Replaces the host:port in every `sip:` / `sips:` URI found in `value`.
pub fn rewrite_sip_uri(value: &str, host: &str, port: u16) -> String {
let mut result = String::with_capacity(value.len());
let mut i = 0;
let bytes = value.as_bytes();
while i < bytes.len() {
// Look for "sip:" or "sips:"
let scheme_len = if i + 4 <= bytes.len()
&& (bytes[i..].starts_with(b"sip:") || bytes[i..].starts_with(b"SIP:"))
{
4
} else if i + 5 <= bytes.len()
&& (bytes[i..].starts_with(b"sips:") || bytes[i..].starts_with(b"SIPS:"))
{
5
} else {
result.push(value[i..].chars().next().unwrap());
i += value[i..].chars().next().unwrap().len_utf8();
continue;
};
let scheme = &value[i..i + scheme_len];
let rest = &value[i + scheme_len..];
// Check for userpart (contains '@')
let (userpart, host_start) = if let Some(at) = rest.find('@') {
// Make sure @ comes before any delimiters
let delim = rest.find(|c: char| c == '>' || c == ';' || c == ',' || c.is_whitespace());
if delim.is_none() || at < delim.unwrap() {
(&rest[..=at], at + 1)
} else {
("", 0)
}
} else {
("", 0)
};
// Find the end of the host:port portion
let host_rest = &rest[host_start..];
let end = host_rest
.find(|c: char| c == '>' || c == ';' || c == ',' || c.is_whitespace())
.unwrap_or(host_rest.len());
result.push_str(scheme);
result.push_str(userpart);
result.push_str(&format!("{host}:{port}"));
i += scheme_len + host_start + end;
}
result
}
/// Rewrites the connection address (`c=`) and audio media port (`m=audio`)
/// in an SDP body. Returns the rewritten body together with the original
/// endpoint that was replaced (if any).
pub fn rewrite_sdp(body: &str, ip: &str, port: u16) -> (String, Option<Endpoint>) {
let mut orig_addr: Option<String> = None;
let mut orig_port: Option<u16> = None;
let lines: Vec<String> = body
.replace("\r\n", "\n")
.split('\n')
.map(|line| {
if let Some(rest) = line.strip_prefix("c=IN IP4 ") {
orig_addr = Some(rest.trim().to_string());
format!("c=IN IP4 {ip}")
} else if line.starts_with("m=audio ") {
let parts: Vec<&str> = line.split(' ').collect();
if parts.len() >= 2 {
orig_port = parts[1].parse().ok();
let mut rebuilt = parts[0].to_string();
rebuilt.push(' ');
rebuilt.push_str(&port.to_string());
for part in &parts[2..] {
rebuilt.push(' ');
rebuilt.push_str(part);
}
return rebuilt;
}
line.to_string()
} else {
line.to_string()
}
})
.collect();
let original = match (orig_addr, orig_port) {
(Some(a), Some(p)) => Some(Endpoint { address: a, port: p, codec_pt: None }),
_ => None,
};
(lines.join("\r\n"), original)
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_rewrite_sip_uri() {
let input = "<sip:user@10.0.0.1:5060>";
let result = rewrite_sip_uri(input, "192.168.1.1", 5070);
assert_eq!(result, "<sip:user@192.168.1.1:5070>");
}
#[test]
fn test_rewrite_sip_uri_no_port() {
let input = "sip:user@10.0.0.1";
let result = rewrite_sip_uri(input, "192.168.1.1", 5070);
assert_eq!(result, "sip:user@192.168.1.1:5070");
}
#[test]
fn test_rewrite_sdp() {
let sdp = "v=0\r\nc=IN IP4 10.0.0.1\r\nm=audio 5060 RTP/AVP 0 9\r\na=sendrecv\r\n";
let (rewritten, orig) = rewrite_sdp(sdp, "192.168.1.1", 20000);
assert!(rewritten.contains("c=IN IP4 192.168.1.1"));
assert!(rewritten.contains("m=audio 20000 RTP/AVP 0 9"));
let ep = orig.unwrap();
assert_eq!(ep.address, "10.0.0.1");
assert_eq!(ep.port, 5060);
}
}

View File

@@ -1,149 +0,0 @@
/// TTS engine CLI — synthesizes text to a WAV file using Kokoro neural TTS.
///
/// Usage:
/// echo "Hello world" | tts-engine --model kokoro-v1.0.onnx --voices voices.bin --output out.wav
/// tts-engine --model kokoro-v1.0.onnx --voices voices.bin --output out.wav --text "Hello world"
///
/// Outputs 24kHz 16-bit mono WAV.
use kokoro_tts::{KokoroTts, Voice};
use std::io::{self, Read};
fn parse_args() -> Result<(String, String, String, String, Option<String>), String> {
let args: Vec<String> = std::env::args().collect();
let mut model = String::new();
let mut voices = String::new();
let mut output = String::new();
let mut text: Option<String> = None;
let mut voice_name: Option<String> = None;
let mut i = 1;
while i < args.len() {
match args[i].as_str() {
"--model" => { i += 1; model = args.get(i).cloned().unwrap_or_default(); }
"--voices" => { i += 1; voices = args.get(i).cloned().unwrap_or_default(); }
"--output" | "--output_file" => { i += 1; output = args.get(i).cloned().unwrap_or_default(); }
"--text" => { i += 1; text = args.get(i).cloned(); }
"--voice" => { i += 1; voice_name = args.get(i).cloned(); }
_ => {}
}
i += 1;
}
if model.is_empty() { return Err("--model required".into()); }
if voices.is_empty() { return Err("--voices required".into()); }
if output.is_empty() { return Err("--output required".into()); }
let voice_str = voice_name.unwrap_or_else(|| "af_bella".into());
Ok((model, voices, output, voice_str, text))
}
fn select_voice(name: &str) -> Voice {
match name {
"af_bella" => Voice::AfBella(1.0),
"af_heart" => Voice::AfHeart(1.0),
"af_jessica" => Voice::AfJessica(1.0),
"af_nicole" => Voice::AfNicole(1.0),
"af_nova" => Voice::AfNova(1.0),
"af_sarah" => Voice::AfSarah(1.0),
"af_sky" => Voice::AfSky(1.0),
"af_river" => Voice::AfRiver(1.0),
"af_alloy" => Voice::AfAlloy(1.0),
"af_aoede" => Voice::AfAoede(1.0),
"af_kore" => Voice::AfKore(1.0),
"am_adam" => Voice::AmAdam(1.0),
"am_echo" => Voice::AmEcho(1.0),
"am_eric" => Voice::AmEric(1.0),
"am_fenrir" => Voice::AmFenrir(1.0),
"am_liam" => Voice::AmLiam(1.0),
"am_michael" => Voice::AmMichael(1.0),
"am_onyx" => Voice::AmOnyx(1.0),
"am_puck" => Voice::AmPuck(1.0),
"bf_alice" => Voice::BfAlice(1.0),
"bf_emma" => Voice::BfEmma(1.0),
"bf_isabella" => Voice::BfIsabella(1.0),
"bf_lily" => Voice::BfLily(1.0),
"bm_daniel" => Voice::BmDaniel(1.0),
"bm_fable" => Voice::BmFable(1.0),
"bm_george" => Voice::BmGeorge(1.0),
"bm_lewis" => Voice::BmLewis(1.0),
_ => {
eprintln!("[tts-engine] unknown voice '{}', falling back to af_bella", name);
Voice::AfBella(1.0)
}
}
}
#[tokio::main]
async fn main() {
let (model_path, voices_path, output_path, voice_name, text_arg) = match parse_args() {
Ok(v) => v,
Err(e) => {
eprintln!("Error: {}", e);
eprintln!("Usage: tts-engine --model <model.onnx> --voices <voices.bin> --output <output.wav> [--text <text>] [--voice <voice_name>]");
std::process::exit(1);
}
};
// Get text from --text arg or stdin.
let text = match text_arg {
Some(t) => t,
None => {
let mut buf = String::new();
io::stdin().read_to_string(&mut buf).expect("failed to read stdin");
buf.trim().to_string()
}
};
if text.is_empty() {
eprintln!("[tts-engine] no text provided");
std::process::exit(1);
}
eprintln!("[tts-engine] loading model: {}", model_path);
let tts = match KokoroTts::new(&model_path, &voices_path).await {
Ok(t) => t,
Err(e) => {
eprintln!("[tts-engine] failed to load model: {:?}", e);
std::process::exit(1);
}
};
let voice = select_voice(&voice_name);
eprintln!("[tts-engine] synthesizing with voice '{}': \"{}\"", voice_name, text);
let (samples, duration) = match tts.synth(&text, voice).await {
Ok(r) => r,
Err(e) => {
eprintln!("[tts-engine] synthesis failed: {:?}", e);
std::process::exit(1);
}
};
eprintln!("[tts-engine] synthesized {} samples in {:?}", samples.len(), duration);
// Write WAV: 24kHz, 16-bit, mono (same format announcement.ts expects).
let spec = hound::WavSpec {
channels: 1,
sample_rate: 24000,
bits_per_sample: 16,
sample_format: hound::SampleFormat::Int,
};
let mut writer = match hound::WavWriter::create(&output_path, spec) {
Ok(w) => w,
Err(e) => {
eprintln!("[tts-engine] failed to create WAV: {}", e);
std::process::exit(1);
}
};
for &sample in &samples {
let s16 = (sample * 32767.0).round().clamp(-32768.0, 32767.0) as i16;
writer.write_sample(s16).unwrap();
}
writer.finalize().unwrap();
eprintln!("[tts-engine] wrote {}", output_path);
}

View File

@@ -3,6 +3,6 @@
*/
export const commitinfo = {
name: 'siprouter',
version: '1.10.0',
version: '1.18.0',
description: 'undefined'
}

View File

@@ -1,42 +1,22 @@
/**
* TTS announcement module — pre-generates audio announcements using espeak-ng
* and caches them as encoded RTP packets for playback during call setup.
* TTS announcement module — generates announcement WAV files at startup.
*
* On startup, generates the announcement WAV via espeak-ng (formant-based TTS
* with highly accurate pronunciation), encodes each 20ms frame to G.722 (for
* SIP) and Opus (for WebRTC) via the Rust transcoder, and caches the packets.
* Engine priority: espeak-ng (formant TTS, fast) → Kokoro neural TTS via
* proxy-engine → disabled.
*
* Falls back to the Rust tts-engine (Kokoro neural TTS) if espeak-ng is not
* installed, and disables announcements if neither is available.
* The generated WAV is left on disk for Rust's audio_player / start_interaction
* to play during calls. No encoding or RTP playback happens in TypeScript.
*/
import { execSync } from 'node:child_process';
import fs from 'node:fs';
import path from 'node:path';
import { Buffer } from 'node:buffer';
import { buildRtpHeader, rtpClockIncrement } from './call/leg.ts';
import { encodePcm, isCodecReady } from './opusbridge.ts';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
/** A pre-encoded announcement ready for RTP playback. */
export interface IAnnouncementCache {
/** G.722 encoded frames (each is a 20ms frame payload, no RTP header). */
g722Frames: Buffer[];
/** Opus encoded frames for WebRTC playback. */
opusFrames: Buffer[];
/** Total duration in milliseconds. */
durationMs: number;
}
import { sendProxyCommand, isProxyReady } from './proxybridge.ts';
// ---------------------------------------------------------------------------
// State
// ---------------------------------------------------------------------------
let cachedAnnouncement: IAnnouncementCache | null = null;
const TTS_DIR = path.join(process.cwd(), '.nogit', 'tts');
const ANNOUNCEMENT_TEXT = "Hello. I'm connecting your call now.";
const CACHE_WAV = path.join(TTS_DIR, 'announcement.wav');
@@ -47,12 +27,10 @@ const KOKORO_VOICES = 'voices.bin';
const KOKORO_VOICE = 'af_bella';
// ---------------------------------------------------------------------------
// Initialization
// TTS generators
// ---------------------------------------------------------------------------
/**
* Check if espeak-ng is available on the system.
*/
/** Check if espeak-ng is available on the system. */
function isEspeakAvailable(): boolean {
try {
execSync('which espeak-ng', { stdio: 'pipe' });
@@ -62,10 +40,7 @@ function isEspeakAvailable(): boolean {
}
}
/**
* Generate announcement WAV via espeak-ng (primary engine).
* Returns true on success.
*/
/** Generate announcement WAV via espeak-ng (primary engine). */
function generateViaEspeak(wavPath: string, text: string, log: (msg: string) => void): boolean {
log('[tts] generating announcement audio via espeak-ng...');
try {
@@ -81,11 +56,8 @@ function generateViaEspeak(wavPath: string, text: string, log: (msg: string) =>
}
}
/**
* Generate announcement WAV via Kokoro TTS (fallback engine).
* Returns true on success.
*/
function generateViaKokoro(wavPath: string, text: string, log: (msg: string) => void): boolean {
/** Generate announcement WAV via Kokoro TTS (fallback, runs inside proxy-engine). */
async function generateViaKokoro(wavPath: string, text: string, log: (msg: string) => void): Promise<boolean> {
const modelPath = path.join(TTS_DIR, KOKORO_MODEL);
const voicesPath = path.join(TTS_DIR, KOKORO_VOICES);
@@ -94,25 +66,21 @@ function generateViaKokoro(wavPath: string, text: string, log: (msg: string) =>
return false;
}
const root = process.cwd();
const ttsBinPaths = [
path.join(root, 'dist_rust', 'tts-engine'),
path.join(root, 'rust', 'target', 'release', 'tts-engine'),
path.join(root, 'rust', 'target', 'debug', 'tts-engine'),
];
const ttsBin = ttsBinPaths.find((p) => fs.existsSync(p));
if (!ttsBin) {
log('[tts] tts-engine binary not found — Kokoro fallback unavailable');
if (!isProxyReady()) {
log('[tts] proxy-engine not ready — Kokoro fallback unavailable');
return false;
}
log('[tts] generating announcement audio via Kokoro TTS (fallback)...');
try {
execSync(
`"${ttsBin}" --model "${modelPath}" --voices "${voicesPath}" --voice "${KOKORO_VOICE}" --output "${wavPath}" --text "${text}"`,
{ timeout: 120000, stdio: 'pipe' },
);
log('[tts] Kokoro WAV generated');
await sendProxyCommand('generate_tts', {
model: modelPath,
voices: voicesPath,
voice: KOKORO_VOICE,
text,
output: wavPath,
});
log('[tts] Kokoro WAV generated (via proxy-engine)');
return true;
} catch (e: any) {
log(`[tts] Kokoro failed: ${e.message}`);
@@ -120,40 +88,13 @@ function generateViaKokoro(wavPath: string, text: string, log: (msg: string) =>
}
}
/**
* Read a WAV file and detect its sample rate from the fmt chunk.
* Returns { pcm, sampleRate } or null on failure.
*/
function readWavWithRate(wavPath: string): { pcm: Buffer; sampleRate: number } | null {
const wav = fs.readFileSync(wavPath);
if (wav.length < 44) return null;
if (wav.toString('ascii', 0, 4) !== 'RIFF') return null;
if (wav.toString('ascii', 8, 12) !== 'WAVE') return null;
let sampleRate = 22050; // default
let offset = 12;
let pcm: Buffer | null = null;
while (offset < wav.length - 8) {
const chunkId = wav.toString('ascii', offset, offset + 4);
const chunkSize = wav.readUInt32LE(offset + 4);
if (chunkId === 'fmt ') {
sampleRate = wav.readUInt32LE(offset + 12);
}
if (chunkId === 'data') {
pcm = wav.subarray(offset + 8, offset + 8 + chunkSize);
}
offset += 8 + chunkSize;
if (offset % 2 !== 0) offset++;
}
if (!pcm) return null;
return { pcm, sampleRate };
}
// ---------------------------------------------------------------------------
// Initialization
// ---------------------------------------------------------------------------
/**
* Pre-generate the announcement audio and encode to G.722 + Opus frames.
* Must be called after the codec bridge is initialized.
* Pre-generate the announcement WAV file.
* Must be called after the proxy engine is initialized.
*
* Engine priority: espeak-ng → Kokoro → disabled.
*/
@@ -161,7 +102,6 @@ export async function initAnnouncement(log: (msg: string) => void): Promise<bool
fs.mkdirSync(TTS_DIR, { recursive: true });
try {
// Generate WAV if not cached.
if (!fs.existsSync(CACHE_WAV)) {
let generated = false;
@@ -172,9 +112,9 @@ export async function initAnnouncement(log: (msg: string) => void): Promise<bool
log('[tts] espeak-ng not installed — trying Kokoro fallback');
}
// Fall back to Kokoro.
// Fall back to Kokoro (via proxy-engine).
if (!generated) {
generated = generateViaKokoro(CACHE_WAV, ANNOUNCEMENT_TEXT, log);
generated = await generateViaKokoro(CACHE_WAV, ANNOUNCEMENT_TEXT, log);
}
if (!generated) {
@@ -183,49 +123,7 @@ export async function initAnnouncement(log: (msg: string) => void): Promise<bool
}
}
// Read WAV and extract raw PCM + sample rate.
const result = readWavWithRate(CACHE_WAV);
if (!result) {
log('[tts] failed to parse WAV file');
return false;
}
const { pcm, sampleRate } = result;
// Wait for codec bridge to be ready.
if (!isCodecReady()) {
log('[tts] codec bridge not ready — will retry');
return false;
}
// Encode in 20ms chunks. The Rust encoder resamples to each codec's native rate.
const FRAME_SAMPLES = Math.floor(sampleRate * 0.02);
const FRAME_BYTES = FRAME_SAMPLES * 2; // 16-bit = 2 bytes per sample
const totalFrames = Math.floor(pcm.length / FRAME_BYTES);
const g722Frames: Buffer[] = [];
const opusFrames: Buffer[] = [];
log(`[tts] encoding ${totalFrames} frames (${FRAME_SAMPLES} samples/frame @ ${sampleRate}Hz)...`);
for (let i = 0; i < totalFrames; i++) {
const framePcm = pcm.subarray(i * FRAME_BYTES, (i + 1) * FRAME_BYTES);
const pcmBuf = Buffer.from(framePcm);
const [g722, opus] = await Promise.all([
encodePcm(pcmBuf, sampleRate, 9), // G.722 for SIP devices
encodePcm(pcmBuf, sampleRate, 111), // Opus for WebRTC browsers
]);
if (g722) g722Frames.push(g722);
if (opus) opusFrames.push(opus);
if (!g722 && !opus && i < 3) log(`[tts] frame ${i} encode failed`);
}
cachedAnnouncement = {
g722Frames,
opusFrames,
durationMs: totalFrames * 20,
};
log(`[tts] announcement cached: ${g722Frames.length} frames (${(totalFrames * 20 / 1000).toFixed(1)}s)`);
log('[tts] announcement WAV ready');
return true;
} catch (e: any) {
log(`[tts] init error: ${e.message}`);
@@ -233,100 +131,7 @@ export async function initAnnouncement(log: (msg: string) => void): Promise<bool
}
}
// ---------------------------------------------------------------------------
// Playback
// ---------------------------------------------------------------------------
/**
* Play the pre-cached announcement to an RTP endpoint.
*
* @param sendPacket - function to send a raw RTP packet
* @param ssrc - SSRC to use in RTP headers
* @param onDone - called when the announcement finishes
* @returns a cancel function, or null if no announcement is cached
*/
export function playAnnouncement(
sendPacket: (pkt: Buffer) => void,
ssrc: number,
onDone?: () => void,
): (() => void) | null {
if (!cachedAnnouncement || cachedAnnouncement.g722Frames.length === 0) {
onDone?.();
return null;
/** Get the path to the cached announcement WAV, or null if not generated. */
export function getAnnouncementWavPath(): string | null {
return fs.existsSync(CACHE_WAV) ? CACHE_WAV : null;
}
const frames = cachedAnnouncement.g722Frames;
const PT = 9; // G.722
let frameIdx = 0;
let seq = Math.floor(Math.random() * 0xffff);
let rtpTs = Math.floor(Math.random() * 0xffffffff);
const timer = setInterval(() => {
if (frameIdx >= frames.length) {
clearInterval(timer);
onDone?.();
return;
}
const payload = frames[frameIdx];
const hdr = buildRtpHeader(PT, seq & 0xffff, rtpTs >>> 0, ssrc >>> 0, frameIdx === 0);
const pkt = Buffer.concat([hdr, payload]);
sendPacket(pkt);
seq++;
rtpTs += rtpClockIncrement(PT);
frameIdx++;
}, 20);
// Return cancel function.
return () => clearInterval(timer);
}
/**
* Play pre-cached Opus announcement to a WebRTC PeerConnection sender.
*
* @param sendRtpPacket - function to send a raw RTP packet via sender.sendRtp()
* @param ssrc - SSRC to use in RTP headers
* @param onDone - called when announcement finishes
* @returns cancel function, or null if no announcement cached
*/
export function playAnnouncementToWebRtc(
sendRtpPacket: (pkt: Buffer) => void,
ssrc: number,
counters: { seq: number; ts: number },
onDone?: () => void,
): (() => void) | null {
if (!cachedAnnouncement || cachedAnnouncement.opusFrames.length === 0) {
onDone?.();
return null;
}
const frames = cachedAnnouncement.opusFrames;
const PT = 111; // Opus
let frameIdx = 0;
const timer = setInterval(() => {
if (frameIdx >= frames.length) {
clearInterval(timer);
onDone?.();
return;
}
const payload = frames[frameIdx];
const hdr = buildRtpHeader(PT, counters.seq & 0xffff, counters.ts >>> 0, ssrc >>> 0, frameIdx === 0);
const pkt = Buffer.concat([hdr, payload]);
sendRtpPacket(pkt);
counters.seq++;
counters.ts += 960; // Opus at 48kHz: 960 samples per 20ms
frameIdx++;
}, 20);
return () => clearInterval(timer);
}
/** Check if an announcement is cached and ready. */
export function isAnnouncementReady(): boolean {
return cachedAnnouncement !== null && cachedAnnouncement.g722Frames.length > 0;
}

View File

@@ -1,323 +0,0 @@
/**
* Audio recorder — captures RTP packets from a single direction,
* decodes them to PCM, and writes a WAV file.
*
* Uses the Rust codec bridge to transcode incoming audio (G.722, Opus,
* PCMU, PCMA) to PCMU, then decodes mu-law to 16-bit PCM in TypeScript.
* Output: 8kHz 16-bit mono WAV (standard telephony quality).
*
* Supports:
* - Max recording duration limit
* - Silence detection (stop after N seconds of silence)
* - Manual stop
* - DTMF packets (PT 101) are automatically skipped
*/
import { Buffer } from 'node:buffer';
import fs from 'node:fs';
import path from 'node:path';
import { WavWriter } from './wav-writer.ts';
import type { IWavWriterResult } from './wav-writer.ts';
import { transcode, createSession, destroySession } from '../opusbridge.ts';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
export interface IRecordingOptions {
/** Output directory for WAV files. */
outputDir: string;
/** Target sample rate for the WAV output (default 8000). */
sampleRate?: number;
/** Maximum recording duration in seconds. 0 = unlimited. Default 120. */
maxDurationSec?: number;
/** Stop after this many consecutive seconds of silence. 0 = disabled. Default 5. */
silenceTimeoutSec?: number;
/** Silence threshold: max PCM amplitude below this is "silent". Default 200. */
silenceThreshold?: number;
/** Logging function. */
log: (msg: string) => void;
}
export interface IRecordingResult {
/** Full path to the WAV file. */
filePath: string;
/** Duration in milliseconds. */
durationMs: number;
/** Sample rate of the WAV. */
sampleRate: number;
/** Size of the WAV file in bytes. */
fileSize: number;
/** Why the recording was stopped. */
stopReason: TRecordingStopReason;
}
export type TRecordingStopReason = 'manual' | 'max-duration' | 'silence' | 'cancelled';
// ---------------------------------------------------------------------------
// Mu-law decode table (ITU-T G.711)
// ---------------------------------------------------------------------------
/** Pre-computed mu-law → 16-bit linear PCM lookup table (256 entries). */
const MULAW_DECODE: Int16Array = buildMulawDecodeTable();
function buildMulawDecodeTable(): Int16Array {
const table = new Int16Array(256);
for (let i = 0; i < 256; i++) {
// Invert all bits per mu-law standard.
let mu = ~i & 0xff;
const sign = mu & 0x80;
const exponent = (mu >> 4) & 0x07;
const mantissa = mu & 0x0f;
let magnitude = ((mantissa << 1) + 33) << (exponent + 2);
magnitude -= 0x84; // Bias adjustment
table[i] = sign ? -magnitude : magnitude;
}
return table;
}
/** Decode a PCMU payload to 16-bit LE PCM. */
function decodeMulaw(mulaw: Buffer): Buffer {
const pcm = Buffer.alloc(mulaw.length * 2);
for (let i = 0; i < mulaw.length; i++) {
pcm.writeInt16LE(MULAW_DECODE[mulaw[i]], i * 2);
}
return pcm;
}
// ---------------------------------------------------------------------------
// AudioRecorder
// ---------------------------------------------------------------------------
export class AudioRecorder {
/** Current state. */
state: 'idle' | 'recording' | 'stopped' = 'idle';
/** Called when recording stops automatically (silence or max duration). */
onStopped: ((result: IRecordingResult) => void) | null = null;
private outputDir: string;
private sampleRate: number;
private maxDurationSec: number;
private silenceTimeoutSec: number;
private silenceThreshold: number;
private log: (msg: string) => void;
private wavWriter: WavWriter | null = null;
private filePath: string = '';
private codecSessionId: string | null = null;
private stopReason: TRecordingStopReason = 'manual';
// Silence detection.
private consecutiveSilentFrames = 0;
/** Number of 20ms frames that constitute silence timeout. */
private silenceFrameThreshold = 0;
// Max duration timer.
private maxDurationTimer: ReturnType<typeof setTimeout> | null = null;
// Processing queue to avoid concurrent transcodes.
private processQueue: Promise<void> = Promise.resolve();
constructor(options: IRecordingOptions) {
this.outputDir = options.outputDir;
this.sampleRate = options.sampleRate ?? 8000;
this.maxDurationSec = options.maxDurationSec ?? 120;
this.silenceTimeoutSec = options.silenceTimeoutSec ?? 5;
this.silenceThreshold = options.silenceThreshold ?? 200;
this.log = options.log;
}
/**
* Start recording. Creates the output directory, WAV file, and codec session.
* @param fileId - unique ID for the recording file name
*/
async start(fileId?: string): Promise<void> {
if (this.state !== 'idle') return;
// Ensure output directory exists.
if (!fs.existsSync(this.outputDir)) {
fs.mkdirSync(this.outputDir, { recursive: true });
}
// Generate file path.
const id = fileId ?? `rec-${Date.now()}-${Math.random().toString(36).slice(2, 8)}`;
this.filePath = path.join(this.outputDir, `${id}.wav`);
// Create a codec session for isolated decoding.
this.codecSessionId = `recorder-${id}`;
await createSession(this.codecSessionId);
// Open WAV writer.
this.wavWriter = new WavWriter({
filePath: this.filePath,
sampleRate: this.sampleRate,
});
this.wavWriter.open();
// Silence detection threshold: frames in timeout period.
this.silenceFrameThreshold = this.silenceTimeoutSec > 0
? Math.ceil((this.silenceTimeoutSec * 1000) / 20)
: 0;
this.consecutiveSilentFrames = 0;
// Max duration timer.
if (this.maxDurationSec > 0) {
this.maxDurationTimer = setTimeout(() => {
if (this.state === 'recording') {
this.stopReason = 'max-duration';
this.log(`[recorder] max duration reached (${this.maxDurationSec}s)`);
this.stop().then((result) => this.onStopped?.(result));
}
}, this.maxDurationSec * 1000);
}
this.state = 'recording';
this.stopReason = 'manual';
this.log(`[recorder] started → ${this.filePath}`);
}
/**
* Feed an RTP packet. Strips the 12-byte header, transcodes the payload
* to PCMU via the Rust bridge, decodes to PCM, and writes to WAV.
* Skips telephone-event (DTMF) and comfort noise packets.
*/
processRtp(data: Buffer): void {
if (this.state !== 'recording') return;
if (data.length < 13) return; // too short
const pt = data[1] & 0x7f;
// Skip DTMF (telephone-event) and comfort noise.
if (pt === 101 || pt === 13) return;
const payload = data.subarray(12);
if (payload.length === 0) return;
// Queue processing to avoid concurrent transcodes corrupting codec state.
this.processQueue = this.processQueue.then(() => this.decodeAndWrite(payload, pt));
}
/** Decode a single RTP payload to PCM and write to WAV. */
private async decodeAndWrite(payload: Buffer, pt: number): Promise<void> {
if (this.state !== 'recording' || !this.wavWriter) return;
let pcm: Buffer;
if (pt === 0) {
// PCMU: decode directly in TypeScript (no Rust round-trip needed).
pcm = decodeMulaw(payload);
} else {
// All other codecs: transcode to PCMU via Rust, then decode mu-law.
const mulaw = await transcode(payload, pt, 0, this.codecSessionId ?? undefined);
if (!mulaw) return;
pcm = decodeMulaw(mulaw);
}
// Silence detection.
if (this.silenceFrameThreshold > 0) {
if (isSilent(pcm, this.silenceThreshold)) {
this.consecutiveSilentFrames++;
if (this.consecutiveSilentFrames >= this.silenceFrameThreshold) {
this.stopReason = 'silence';
this.log(`[recorder] silence detected (${this.silenceTimeoutSec}s)`);
this.stop().then((result) => this.onStopped?.(result));
return;
}
} else {
this.consecutiveSilentFrames = 0;
}
}
this.wavWriter.write(pcm);
}
/**
* Stop recording and finalize the WAV file.
*/
async stop(): Promise<IRecordingResult> {
if (this.state === 'stopped' || this.state === 'idle') {
return {
filePath: this.filePath,
durationMs: 0,
sampleRate: this.sampleRate,
fileSize: 0,
stopReason: this.stopReason,
};
}
this.state = 'stopped';
// Wait for pending decode operations to finish.
await this.processQueue;
// Clear timers.
if (this.maxDurationTimer) {
clearTimeout(this.maxDurationTimer);
this.maxDurationTimer = null;
}
// Finalize WAV.
let wavResult: IWavWriterResult | null = null;
if (this.wavWriter) {
wavResult = this.wavWriter.close();
this.wavWriter = null;
}
// Destroy codec session.
if (this.codecSessionId) {
await destroySession(this.codecSessionId);
this.codecSessionId = null;
}
const result: IRecordingResult = {
filePath: this.filePath,
durationMs: wavResult?.durationMs ?? 0,
sampleRate: this.sampleRate,
fileSize: wavResult?.fileSize ?? 0,
stopReason: this.stopReason,
};
this.log(`[recorder] stopped (${result.stopReason}): ${result.durationMs}ms → ${this.filePath}`);
return result;
}
/** Cancel recording — stops and deletes the WAV file. */
async cancel(): Promise<void> {
this.stopReason = 'cancelled';
await this.stop();
// Delete the incomplete file.
try {
if (fs.existsSync(this.filePath)) {
fs.unlinkSync(this.filePath);
this.log(`[recorder] cancelled — deleted ${this.filePath}`);
}
} catch { /* best effort */ }
}
/** Clean up all resources. */
destroy(): void {
if (this.state === 'recording') {
this.cancel();
}
this.onStopped = null;
}
}
// ---------------------------------------------------------------------------
// Helpers
// ---------------------------------------------------------------------------
/** Check if a PCM buffer is "silent" (max amplitude below threshold). */
function isSilent(pcm: Buffer, threshold: number): boolean {
let maxAmp = 0;
for (let i = 0; i < pcm.length - 1; i += 2) {
const sample = pcm.readInt16LE(i);
const abs = sample < 0 ? -sample : sample;
if (abs > maxAmp) maxAmp = abs;
// Early exit: already above threshold.
if (maxAmp >= threshold) return false;
}
return true;
}

File diff suppressed because it is too large Load Diff

View File

@@ -1,265 +0,0 @@
/**
* Call — the hub entity in the hub model.
*
* A Call owns N legs and bridges their media. For 2-party calls, RTP packets
* from leg A are forwarded to leg B and vice versa. For N>2 party calls,
* packets from each leg are forwarded to all other legs (fan-out).
*
* Transcoding is applied per-leg when codecs differ.
*/
import { Buffer } from 'node:buffer';
import type { ILeg } from './leg.ts';
import type { TCallState, TCallDirection, ICallStatus } from './types.ts';
import { RtpPortPool } from './rtp-port-pool.ts';
import type { SipLeg } from './sip-leg.ts';
export class Call {
readonly id: string;
state: TCallState = 'setting-up';
direction: TCallDirection;
readonly createdAt: number;
callerNumber: string | null = null;
calleeNumber: string | null = null;
providerUsed: string | null = null;
/** All legs in this call. */
private legs = new Map<string, ILeg>();
/** Codec payload type for the "native" audio in the call (usually the first SIP leg's codec). */
private nativeCodec: number | null = null;
/** Port pool reference for cleanup. */
private portPool: RtpPortPool;
private log: (msg: string) => void;
private onChange: ((call: Call) => void) | null = null;
constructor(options: {
id: string;
direction: TCallDirection;
portPool: RtpPortPool;
log: (msg: string) => void;
onChange?: (call: Call) => void;
}) {
this.id = options.id;
this.direction = options.direction;
this.createdAt = Date.now();
this.portPool = options.portPool;
this.log = options.log;
this.onChange = options.onChange ?? null;
}
// -------------------------------------------------------------------------
// Leg management
// -------------------------------------------------------------------------
/** Add a leg to this call and wire up media forwarding. */
addLeg(leg: ILeg): void {
this.legs.set(leg.id, leg);
// Wire up RTP forwarding: when this leg receives a packet, forward to all other legs.
leg.onRtpReceived = (data: Buffer) => {
this.forwardRtp(leg.id, data);
};
this.log(`[call:${this.id}] added leg ${leg.id} (${leg.type}), total=${this.legs.size}`);
this.updateState();
}
/** Remove a leg from this call, tear it down, and release its port. */
removeLeg(legId: string): void {
const leg = this.legs.get(legId);
if (!leg) return;
leg.onRtpReceived = null;
leg.teardown();
if (leg.rtpPort) {
this.portPool.release(leg.rtpPort);
}
this.legs.delete(legId);
this.log(`[call:${this.id}] removed leg ${legId}, total=${this.legs.size}`);
this.updateState();
}
getLeg(legId: string): ILeg | null {
return this.legs.get(legId) ?? null;
}
getLegs(): ILeg[] {
return [...this.legs.values()];
}
getLegByType(type: string): ILeg | null {
for (const leg of this.legs.values()) {
if (leg.type === type) return leg;
}
return null;
}
getLegBySipCallId(sipCallId: string): ILeg | null {
for (const leg of this.legs.values()) {
if (leg.sipCallId === sipCallId) return leg;
}
return null;
}
get legCount(): number {
return this.legs.size;
}
// -------------------------------------------------------------------------
// Media forwarding (the hub)
// -------------------------------------------------------------------------
private forwardRtp(fromLegId: string, data: Buffer): void {
for (const [id, leg] of this.legs) {
if (id === fromLegId) continue;
if (leg.state !== 'connected') continue;
// For WebRTC legs, sendRtp calls forwardToBrowser which handles transcoding internally.
// For SIP legs, forward the raw packet (same codec path) or let the leg handle it.
// The Call hub does NOT transcode — that's the leg's responsibility.
leg.sendRtp(data);
}
}
// -------------------------------------------------------------------------
// State management
// -------------------------------------------------------------------------
private updateState(): void {
if (this.state === 'terminated' || this.state === 'terminating') return;
const legs = [...this.legs.values()];
if (legs.length === 0) {
this.state = 'terminated';
} else if (legs.every((l) => l.state === 'terminated')) {
this.state = 'terminated';
} else if (legs.some((l) => l.state === 'connected') && legs.filter((l) => l.state !== 'terminated').length >= 2) {
// If a system leg is connected, report voicemail/ivr state for the dashboard.
const systemLeg = legs.find((l) => l.type === 'system');
if (systemLeg) {
// Keep voicemail/ivr state if already set; otherwise set connected.
if (this.state !== 'voicemail' && this.state !== 'ivr') {
this.state = 'connected';
}
} else {
this.state = 'connected';
}
} else if (legs.some((l) => l.state === 'ringing')) {
this.state = 'ringing';
} else {
this.state = 'setting-up';
}
this.onChange?.(this);
}
/** Notify the call that a leg's state has changed. */
notifyLegStateChange(_leg: ILeg): void {
this.updateState();
}
// -------------------------------------------------------------------------
// Hangup
// -------------------------------------------------------------------------
/** Tear down all legs and terminate the call. */
hangup(): void {
if (this.state === 'terminated' || this.state === 'terminating') return;
this.state = 'terminating';
this.log(`[call:${this.id}] hanging up (${this.legs.size} legs)`);
for (const [id, leg] of this.legs) {
// Send BYE/CANCEL for SIP legs (system legs have no SIP signaling).
if (leg.type === 'sip-device' || leg.type === 'sip-provider') {
(leg as SipLeg).sendHangup();
}
leg.teardown();
if (leg.rtpPort) {
this.portPool.release(leg.rtpPort);
}
}
this.legs.clear();
this.state = 'terminated';
this.onChange?.(this);
}
/**
* Handle a BYE from one leg — tear down the other legs.
* Called by CallManager when a SipLeg receives a BYE.
*/
handleLegTerminated(terminatedLegId: string): void {
const terminatedLeg = this.legs.get(terminatedLegId);
if (!terminatedLeg) return;
// Remove the terminated leg.
terminatedLeg.onRtpReceived = null;
if (terminatedLeg.rtpPort) {
this.portPool.release(terminatedLeg.rtpPort);
}
this.legs.delete(terminatedLegId);
// If this is a 2-party call, hang up the other leg too.
if (this.legs.size <= 1) {
for (const [id, leg] of this.legs) {
// Send BYE/CANCEL for SIP legs (system legs just get torn down).
if (leg.type === 'sip-device' || leg.type === 'sip-provider') {
(leg as SipLeg).sendHangup();
}
leg.teardown();
if (leg.rtpPort) {
this.portPool.release(leg.rtpPort);
}
}
this.legs.clear();
this.state = 'terminated';
this.log(`[call:${this.id}] terminated`);
this.onChange?.(this);
} else {
this.log(`[call:${this.id}] leg ${terminatedLegId} removed, ${this.legs.size} remaining`);
this.updateState();
}
}
// -------------------------------------------------------------------------
// Transfer
// -------------------------------------------------------------------------
/**
* Detach a leg from this call (without tearing it down).
* The leg can then be added to another call.
*/
detachLeg(legId: string): ILeg | null {
const leg = this.legs.get(legId);
if (!leg) return null;
leg.onRtpReceived = null;
this.legs.delete(legId);
this.log(`[call:${this.id}] detached leg ${legId}`);
this.updateState();
return leg;
}
// -------------------------------------------------------------------------
// Status
// -------------------------------------------------------------------------
getStatus(): ICallStatus {
return {
id: this.id,
state: this.state,
direction: this.direction,
callerNumber: this.callerNumber,
calleeNumber: this.calleeNumber,
providerUsed: this.providerUsed,
createdAt: this.createdAt,
duration: Math.floor((Date.now() - this.createdAt) / 1000),
legs: [...this.legs.values()].map((l) => l.getStatus()),
};
}
}

View File

@@ -1,272 +0,0 @@
/**
* DTMF detection — parses RFC 2833 telephone-event RTP packets
* and SIP INFO (application/dtmf-relay) messages.
*
* Designed to be attached to any leg or RTP stream. The detector
* deduplicates repeated telephone-event packets (same digit is sent
* multiple times with increasing duration) and fires a callback
* once per detected digit.
*/
import { Buffer } from 'node:buffer';
import type { SipMessage } from '../sip/index.ts';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
/** A single detected DTMF digit. */
export interface IDtmfDigit {
/** The digit character: '0'-'9', '*', '#', 'A'-'D'. */
digit: string;
/** Duration in milliseconds. */
durationMs: number;
/** Detection source. */
source: 'rfc2833' | 'sip-info';
/** Wall-clock timestamp when the digit was detected. */
timestamp: number;
}
/** Callback fired once per detected DTMF digit. */
export type TDtmfCallback = (digit: IDtmfDigit) => void;
// ---------------------------------------------------------------------------
// Constants
// ---------------------------------------------------------------------------
/** RFC 2833 event ID → character mapping. */
const EVENT_CHARS: string[] = [
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9',
'*', '#', 'A', 'B', 'C', 'D',
];
/** Safety timeout: report digit if no End packet arrives within this many ms. */
const SAFETY_TIMEOUT_MS = 200;
// ---------------------------------------------------------------------------
// DtmfDetector
// ---------------------------------------------------------------------------
/**
* Detects DTMF digits from RFC 2833 RTP packets and SIP INFO messages.
*
* Usage:
* ```
* const detector = new DtmfDetector(log);
* detector.onDigit = (d) => console.log('DTMF:', d.digit);
* // Feed every RTP packet (detector checks PT internally):
* detector.processRtp(rtpPacket);
* // Or feed a SIP INFO message:
* detector.processSipInfo(sipMsg);
* ```
*/
export class DtmfDetector {
/** Callback fired once per detected digit. */
onDigit: TDtmfCallback | null = null;
/** Negotiated telephone-event payload type (default 101). */
private telephoneEventPt: number;
/** Clock rate for duration calculation (default 8000 Hz). */
private clockRate: number;
// -- Deduplication state for RFC 2833 --
/** Event ID of the digit currently being received. */
private currentEventId: number | null = null;
/** RTP timestamp of the first packet for the current event. */
private currentEventTs: number | null = null;
/** Whether the current event has already been reported. */
private currentEventReported = false;
/** Latest duration value seen (in clock ticks). */
private currentEventDuration = 0;
/** Latest volume value seen (dBm0, 0 = loudest). */
private currentEventVolume = 0;
/** Safety timer: fires if no End packet arrives. */
private safetyTimer: ReturnType<typeof setTimeout> | null = null;
private log: (msg: string) => void;
constructor(
log: (msg: string) => void,
telephoneEventPt = 101,
clockRate = 8000,
) {
this.log = log;
this.telephoneEventPt = telephoneEventPt;
this.clockRate = clockRate;
}
// -------------------------------------------------------------------------
// RFC 2833 RTP processing
// -------------------------------------------------------------------------
/**
* Feed an RTP packet. Checks PT; ignores non-DTMF packets.
* Expects the full RTP packet (12-byte header + payload).
*/
processRtp(data: Buffer): void {
if (data.length < 16) return; // 12-byte header + 4-byte telephone-event payload minimum
const pt = data[1] & 0x7f;
if (pt !== this.telephoneEventPt) return;
// Parse RTP header fields we need.
const marker = (data[1] & 0x80) !== 0;
const rtpTimestamp = data.readUInt32BE(4);
// Parse telephone-event payload (4 bytes starting at offset 12).
const eventId = data[12];
const endBit = (data[13] & 0x80) !== 0;
const volume = data[13] & 0x3f;
const duration = data.readUInt16BE(14);
// Validate event ID.
if (eventId >= EVENT_CHARS.length) return;
// Detect new event: marker bit, different event ID, or different RTP timestamp.
const isNewEvent =
marker ||
eventId !== this.currentEventId ||
rtpTimestamp !== this.currentEventTs;
if (isNewEvent) {
// If there was an unreported previous event, report it now (fallback).
this.reportPendingEvent();
// Start tracking the new event.
this.currentEventId = eventId;
this.currentEventTs = rtpTimestamp;
this.currentEventReported = false;
this.currentEventDuration = duration;
this.currentEventVolume = volume;
// Start safety timer.
this.clearSafetyTimer();
this.safetyTimer = setTimeout(() => {
this.reportPendingEvent();
}, SAFETY_TIMEOUT_MS);
}
// Update duration (it increases with each retransmission).
if (duration > this.currentEventDuration) {
this.currentEventDuration = duration;
}
// Report on End bit (first time only).
if (endBit && !this.currentEventReported) {
this.currentEventReported = true;
this.clearSafetyTimer();
const digit = EVENT_CHARS[eventId];
const durationMs = (this.currentEventDuration / this.clockRate) * 1000;
this.log(`[dtmf] RFC 2833 digit '${digit}' (${Math.round(durationMs)}ms)`);
this.onDigit?.({
digit,
durationMs,
source: 'rfc2833',
timestamp: Date.now(),
});
}
}
/** Report a pending (unreported) event — called by safety timer or on new event start. */
private reportPendingEvent(): void {
if (
this.currentEventId !== null &&
!this.currentEventReported &&
this.currentEventId < EVENT_CHARS.length
) {
this.currentEventReported = true;
this.clearSafetyTimer();
const digit = EVENT_CHARS[this.currentEventId];
const durationMs = (this.currentEventDuration / this.clockRate) * 1000;
this.log(`[dtmf] RFC 2833 digit '${digit}' (${Math.round(durationMs)}ms, safety timeout)`);
this.onDigit?.({
digit,
durationMs,
source: 'rfc2833',
timestamp: Date.now(),
});
}
}
private clearSafetyTimer(): void {
if (this.safetyTimer) {
clearTimeout(this.safetyTimer);
this.safetyTimer = null;
}
}
// -------------------------------------------------------------------------
// SIP INFO processing
// -------------------------------------------------------------------------
/**
* Parse a SIP INFO message carrying DTMF.
* Supports Content-Type: application/dtmf-relay (Signal=X / Duration=Y).
*/
processSipInfo(msg: SipMessage): void {
const ct = (msg.getHeader('Content-Type') || '').toLowerCase();
if (!ct.includes('application/dtmf-relay') && !ct.includes('application/dtmf')) return;
const body = msg.body || '';
if (ct.includes('application/dtmf-relay')) {
// Format: "Signal= 5\r\nDuration= 160\r\n"
const signalMatch = body.match(/Signal\s*=\s*(\S+)/i);
const durationMatch = body.match(/Duration\s*=\s*(\d+)/i);
if (!signalMatch) return;
const signal = signalMatch[1];
const durationTicks = durationMatch ? parseInt(durationMatch[1], 10) : 160;
// Validate digit.
if (signal.length !== 1 || !/[0-9*#A-Da-d]/.test(signal)) return;
const digit = signal.toUpperCase();
const durationMs = (durationTicks / this.clockRate) * 1000;
this.log(`[dtmf] SIP INFO digit '${digit}' (${Math.round(durationMs)}ms)`);
this.onDigit?.({
digit,
durationMs,
source: 'sip-info',
timestamp: Date.now(),
});
} else if (ct.includes('application/dtmf')) {
// Simple format: just the digit character in the body.
const digit = body.trim().toUpperCase();
if (digit.length !== 1 || !/[0-9*#A-D]/.test(digit)) return;
this.log(`[dtmf] SIP INFO digit '${digit}' (application/dtmf)`);
this.onDigit?.({
digit,
durationMs: 250, // default duration
source: 'sip-info',
timestamp: Date.now(),
});
}
}
// -------------------------------------------------------------------------
// Lifecycle
// -------------------------------------------------------------------------
/** Reset detection state (e.g., between calls). */
reset(): void {
this.currentEventId = null;
this.currentEventTs = null;
this.currentEventReported = false;
this.currentEventDuration = 0;
this.currentEventVolume = 0;
this.clearSafetyTimer();
}
/** Clean up timers and references. */
destroy(): void {
this.clearSafetyTimer();
this.onDigit = null;
}
}

View File

@@ -1,12 +0,0 @@
export type { TCallState, TLegState, TLegType, TCallDirection, ICallStatus, ILegStatus, ICallHistoryEntry } from './types.ts';
export type { ILeg } from './leg.ts';
export { rtpClockIncrement, buildRtpHeader, codecDisplayName } from './leg.ts';
export { RtpPortPool } from './rtp-port-pool.ts';
export type { IRtpAllocation } from './rtp-port-pool.ts';
export { SipLeg } from './sip-leg.ts';
export type { ISipLegConfig } from './sip-leg.ts';
export { WebRtcLeg } from './webrtc-leg.ts';
export type { IWebRtcLegConfig } from './webrtc-leg.ts';
export { Call } from './call.ts';
export { CallManager } from './call-manager.ts';
export type { ICallManagerConfig } from './call-manager.ts';

View File

@@ -1,104 +0,0 @@
/**
* ILeg interface — abstract connection from a Call hub to an endpoint.
*
* Concrete implementations: SipLeg (SIP devices + providers) and WebRtcLeg (browsers).
* Shared RTP utilities (header building, clock rates) are also defined here.
*/
import { Buffer } from 'node:buffer';
import type dgram from 'node:dgram';
import type { IEndpoint } from '../sip/index.ts';
import type { TLegState, TLegType, ILegStatus } from './types.ts';
import type { IRtpTranscoder } from '../codec.ts';
import type { SipDialog } from '../sip/index.ts';
import type { SipMessage } from '../sip/index.ts';
// ---------------------------------------------------------------------------
// ILeg interface
// ---------------------------------------------------------------------------
export interface ILeg {
readonly id: string;
readonly type: TLegType;
state: TLegState;
/** The SIP Call-ID used by this leg (for CallManager routing). */
readonly sipCallId: string;
/** Where this leg sends/receives RTP. */
readonly rtpPort: number | null;
readonly rtpSock: dgram.Socket | null;
remoteMedia: IEndpoint | null;
/** Negotiated codec payload type (e.g. 9 = G.722, 111 = Opus). */
codec: number | null;
/** Transcoder for converting to this leg's codec (set by Call when codecs differ). */
transcoder: IRtpTranscoder | null;
/** Packet counters. */
pktSent: number;
pktReceived: number;
/** SIP dialog (SipLegs only, null for WebRtcLegs). */
readonly dialog: SipDialog | null;
/**
* Send an RTP packet toward this leg's remote endpoint.
* If a transcoder is set, the Call should transcode before calling this.
*/
sendRtp(data: Buffer): void;
/**
* Callback set by the owning Call — invoked when this leg receives an RTP packet.
* The Call uses this to forward to other legs.
*/
onRtpReceived: ((data: Buffer) => void) | null;
/**
* Handle an incoming SIP message routed to this leg (SipLegs only).
* Returns a SipMessage response if one needs to be sent, or null.
*/
handleSipMessage(msg: SipMessage, rinfo: IEndpoint): void;
/** Release all resources (sockets, peer connections, etc.). */
teardown(): void;
/** Status snapshot for the dashboard. */
getStatus(): ILegStatus;
}
// ---------------------------------------------------------------------------
// Shared RTP utilities
// ---------------------------------------------------------------------------
/** RTP clock increment per 20ms frame for each codec. */
export function rtpClockIncrement(pt: number): number {
if (pt === 111) return 960; // Opus: 48000 Hz x 0.02s
if (pt === 9) return 160; // G.722: 8000 Hz x 0.02s (SDP clock rate quirk)
return 160; // PCMU/PCMA: 8000 Hz x 0.02s
}
/** Build a fresh RTP header with correct PT, timestamp, seq, SSRC. */
export function buildRtpHeader(pt: number, seq: number, ts: number, ssrc: number, marker: boolean): Buffer {
const hdr = Buffer.alloc(12);
hdr[0] = 0x80; // V=2
hdr[1] = (marker ? 0x80 : 0) | (pt & 0x7f);
hdr.writeUInt16BE(seq & 0xffff, 2);
hdr.writeUInt32BE(ts >>> 0, 4);
hdr.writeUInt32BE(ssrc >>> 0, 8);
return hdr;
}
/** Codec name for status display. */
export function codecDisplayName(pt: number | null): string | null {
if (pt === null) return null;
switch (pt) {
case 0: return 'PCMU';
case 8: return 'PCMA';
case 9: return 'G.722';
case 111: return 'Opus';
case 101: return 'telephone-event';
default: return `PT${pt}`;
}
}

View File

@@ -1,38 +1,31 @@
/**
* PromptCache — manages multiple named audio prompts for IVR and voicemail.
* PromptCache — manages named audio prompt WAV files for IVR and voicemail.
*
* Each prompt is pre-encoded as both G.722 frames (for SIP legs) and Opus
* frames (for WebRTC legs), ready for 20ms RTP playback.
* Generates WAV files via espeak-ng (primary) or Kokoro TTS through the
* proxy-engine (fallback). Also supports loading pre-existing WAV files
* and programmatic tone generation.
*
* Supports three sources:
* 1. TTS generation via espeak-ng (primary) or Kokoro (fallback)
* 2. Loading from a pre-existing WAV file
* 3. Programmatic tone generation (beep, etc.)
*
* The existing announcement.ts system continues to work independently;
* this module provides generalized prompt management for IVR/voicemail.
* All audio playback happens in Rust (audio_player / start_interaction).
* This module only manages WAV files on disk.
*/
import { execSync } from 'node:child_process';
import fs from 'node:fs';
import path from 'node:path';
import { Buffer } from 'node:buffer';
import { buildRtpHeader, rtpClockIncrement } from './leg.ts';
import { encodePcm, isCodecReady } from '../opusbridge.ts';
import { sendProxyCommand, isProxyReady } from '../proxybridge.ts';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
/** A pre-encoded prompt ready for RTP playback. */
/** A cached prompt — just a WAV file path and metadata. */
export interface ICachedPrompt {
/** Unique prompt identifier. */
id: string;
/** G.722 encoded frames (20ms each, no RTP header). */
g722Frames: Buffer[];
/** Opus encoded frames (20ms each, no RTP header). */
opusFrames: Buffer[];
/** Total duration in milliseconds. */
/** Path to the WAV file on disk. */
wavPath: string;
/** Total duration in milliseconds (approximate, from WAV header). */
durationMs: number;
}
@@ -65,84 +58,61 @@ function generateViaEspeak(wavPath: string, text: string): boolean {
}
}
/** Generate WAV via Kokoro TTS. */
function generateViaKokoro(wavPath: string, text: string, voice: string): boolean {
/** Generate WAV via Kokoro TTS (runs inside proxy-engine). */
async function generateViaKokoro(wavPath: string, text: string, voice: string): Promise<boolean> {
const modelPath = path.join(TTS_DIR, 'kokoro-v1.0.onnx');
const voicesPath = path.join(TTS_DIR, 'voices.bin');
if (!fs.existsSync(modelPath) || !fs.existsSync(voicesPath)) return false;
const root = process.cwd();
const ttsBin = [
path.join(root, 'dist_rust', 'tts-engine'),
path.join(root, 'rust', 'target', 'release', 'tts-engine'),
path.join(root, 'rust', 'target', 'debug', 'tts-engine'),
].find((p) => fs.existsSync(p));
if (!ttsBin) return false;
if (!isProxyReady()) return false;
try {
execSync(
`"${ttsBin}" --model "${modelPath}" --voices "${voicesPath}" --voice "${voice}" --output "${wavPath}" --text "${text}"`,
{ timeout: 120000, stdio: 'pipe' },
);
await sendProxyCommand('generate_tts', {
model: modelPath,
voices: voicesPath,
voice,
text,
output: wavPath,
});
return true;
} catch {
return false;
}
}
/** Read a WAV file and return raw PCM + sample rate. */
function readWavWithRate(wavPath: string): { pcm: Buffer; sampleRate: number } | null {
/** Read a WAV file's duration from its header. */
function getWavDurationMs(wavPath: string): number {
try {
const wav = fs.readFileSync(wavPath);
if (wav.length < 44) return null;
if (wav.toString('ascii', 0, 4) !== 'RIFF') return null;
if (wav.toString('ascii', 8, 12) !== 'WAVE') return null;
if (wav.length < 44) return 0;
if (wav.toString('ascii', 0, 4) !== 'RIFF') return 0;
let sampleRate = 22050;
let pcm: Buffer | null = null;
let sampleRate = 16000;
let dataSize = 0;
let bitsPerSample = 16;
let channels = 1;
let offset = 12;
while (offset < wav.length - 8) {
const chunkId = wav.toString('ascii', offset, offset + 4);
const chunkSize = wav.readUInt32LE(offset + 4);
if (chunkId === 'fmt ') {
channels = wav.readUInt16LE(offset + 10);
sampleRate = wav.readUInt32LE(offset + 12);
bitsPerSample = wav.readUInt16LE(offset + 22);
}
if (chunkId === 'data') {
pcm = wav.subarray(offset + 8, offset + 8 + chunkSize);
dataSize = chunkSize;
}
offset += 8 + chunkSize;
if (offset % 2 !== 0) offset++;
}
return pcm ? { pcm, sampleRate } : null;
const bytesPerSample = (bitsPerSample / 8) * channels;
const totalSamples = bytesPerSample > 0 ? dataSize / bytesPerSample : 0;
return sampleRate > 0 ? Math.round((totalSamples / sampleRate) * 1000) : 0;
} catch {
return 0;
}
/** Encode raw PCM frames to G.722 + Opus. */
async function encodePcmFrames(
pcm: Buffer,
sampleRate: number,
log: (msg: string) => void,
): Promise<{ g722Frames: Buffer[]; opusFrames: Buffer[] } | null> {
if (!isCodecReady()) return null;
const frameSamples = Math.floor(sampleRate * 0.02); // 20ms
const frameBytes = frameSamples * 2; // 16-bit
const totalFrames = Math.floor(pcm.length / frameBytes);
const g722Frames: Buffer[] = [];
const opusFrames: Buffer[] = [];
for (let i = 0; i < totalFrames; i++) {
const framePcm = Buffer.from(pcm.subarray(i * frameBytes, (i + 1) * frameBytes));
const [g722, opus] = await Promise.all([
encodePcm(framePcm, sampleRate, 9), // G.722
encodePcm(framePcm, sampleRate, 111), // Opus
]);
if (g722) g722Frames.push(g722);
if (opus) opusFrames.push(opus);
}
return { g722Frames, opusFrames };
}
// ---------------------------------------------------------------------------
@@ -178,7 +148,7 @@ export class PromptCache {
}
/**
* Generate a TTS prompt and cache it.
* Generate a TTS prompt WAV and cache its path.
* Uses espeak-ng (primary) or Kokoro (fallback).
*/
async generatePrompt(id: string, text: string, voice = 'af_bella'): Promise<ICachedPrompt | null> {
@@ -190,14 +160,14 @@ export class PromptCache {
this.espeakAvailable = isEspeakAvailable();
}
// Generate WAV.
let generated = false;
// Generate WAV if not already on disk.
if (!fs.existsSync(wavPath)) {
let generated = false;
if (this.espeakAvailable) {
generated = generateViaEspeak(wavPath, text);
}
if (!generated) {
generated = generateViaKokoro(wavPath, text, voice);
generated = await generateViaKokoro(wavPath, text, voice);
}
if (!generated) {
this.log(`[prompt-cache] failed to generate TTS for "${id}"`);
@@ -206,49 +176,22 @@ export class PromptCache {
this.log(`[prompt-cache] generated WAV for "${id}"`);
}
return this.loadWavPrompt(id, wavPath);
return this.registerWav(id, wavPath);
}
/**
* Load a WAV file as a prompt and cache it.
* Load a pre-existing WAV file as a prompt.
*/
async loadWavPrompt(id: string, wavPath: string): Promise<ICachedPrompt | null> {
if (!fs.existsSync(wavPath)) {
this.log(`[prompt-cache] WAV not found: ${wavPath}`);
return null;
}
const result = readWavWithRate(wavPath);
if (!result) {
this.log(`[prompt-cache] failed to parse WAV: ${wavPath}`);
return null;
}
const encoded = await encodePcmFrames(result.pcm, result.sampleRate, this.log);
if (!encoded) {
this.log(`[prompt-cache] encoding failed for "${id}" (codec bridge not ready?)`);
return null;
}
const durationMs = encoded.g722Frames.length * 20;
const prompt: ICachedPrompt = {
id,
g722Frames: encoded.g722Frames,
opusFrames: encoded.opusFrames,
durationMs,
};
this.prompts.set(id, prompt);
this.log(`[prompt-cache] cached "${id}": ${encoded.g722Frames.length} frames (${(durationMs / 1000).toFixed(1)}s)`);
return prompt;
return this.registerWav(id, wavPath);
}
/**
* Generate a beep tone prompt (sine wave).
* @param id - prompt ID
* @param freqHz - tone frequency (default 1000 Hz)
* @param durationMs - tone duration (default 500ms)
* @param amplitude - 16-bit amplitude (default 8000)
* Generate a beep tone WAV and cache it.
*/
async generateBeep(
id: string,
@@ -256,14 +199,17 @@ export class PromptCache {
durationMs = 500,
amplitude = 8000,
): Promise<ICachedPrompt | null> {
// Generate at 16kHz for decent quality.
fs.mkdirSync(TTS_DIR, { recursive: true });
const wavPath = path.join(TTS_DIR, `prompt-${id}.wav`);
if (!fs.existsSync(wavPath)) {
// Generate 16kHz 16-bit mono sine wave WAV.
const sampleRate = 16000;
const totalSamples = Math.floor((sampleRate * durationMs) / 1000);
const pcm = Buffer.alloc(totalSamples * 2);
for (let i = 0; i < totalSamples; i++) {
const t = i / sampleRate;
// Apply a short fade-in/fade-out to avoid click artifacts.
const fadeLen = Math.floor(sampleRate * 0.01); // 10ms fade
let envelope = 1.0;
if (i < fadeLen) envelope = i / fadeLen;
@@ -273,132 +219,57 @@ export class PromptCache {
pcm.writeInt16LE(Math.max(-32768, Math.min(32767, sample)), i * 2);
}
const encoded = await encodePcmFrames(pcm, sampleRate, this.log);
if (!encoded) {
this.log(`[prompt-cache] beep encoding failed for "${id}"`);
return null;
// Write WAV file.
const headerSize = 44;
const dataSize = pcm.length;
const wav = Buffer.alloc(headerSize + dataSize);
// RIFF header
wav.write('RIFF', 0);
wav.writeUInt32LE(36 + dataSize, 4);
wav.write('WAVE', 8);
// fmt chunk
wav.write('fmt ', 12);
wav.writeUInt32LE(16, 16); // chunk size
wav.writeUInt16LE(1, 20); // PCM format
wav.writeUInt16LE(1, 22); // mono
wav.writeUInt32LE(sampleRate, 24);
wav.writeUInt32LE(sampleRate * 2, 28); // byte rate
wav.writeUInt16LE(2, 32); // block align
wav.writeUInt16LE(16, 34); // bits per sample
// data chunk
wav.write('data', 36);
wav.writeUInt32LE(dataSize, 40);
pcm.copy(wav, 44);
fs.writeFileSync(wavPath, wav);
this.log(`[prompt-cache] beep WAV generated for "${id}"`);
}
const actualDuration = encoded.g722Frames.length * 20;
const prompt: ICachedPrompt = {
id,
g722Frames: encoded.g722Frames,
opusFrames: encoded.opusFrames,
durationMs: actualDuration,
};
this.prompts.set(id, prompt);
this.log(`[prompt-cache] beep "${id}" cached: ${actualDuration}ms @ ${freqHz}Hz`);
return prompt;
return this.registerWav(id, wavPath);
}
/**
* Remove a prompt from the cache.
*/
/** Remove a prompt from the cache. */
remove(id: string): void {
this.prompts.delete(id);
}
/**
* Clear all cached prompts.
*/
/** Clear all cached prompts. */
clear(): void {
this.prompts.clear();
}
// -------------------------------------------------------------------------
// Internal
// -------------------------------------------------------------------------
private registerWav(id: string, wavPath: string): ICachedPrompt {
const durationMs = getWavDurationMs(wavPath);
const prompt: ICachedPrompt = { id, wavPath, durationMs };
this.prompts.set(id, prompt);
this.log(`[prompt-cache] cached "${id}": ${wavPath} (${(durationMs / 1000).toFixed(1)}s)`);
return prompt;
}
// ---------------------------------------------------------------------------
// Standalone playback helpers (for use by SystemLeg)
// ---------------------------------------------------------------------------
/**
* Play a cached prompt's G.722 frames as RTP packets at 20ms intervals.
*
* @param prompt - the cached prompt to play
* @param sendPacket - function to send a raw RTP packet (12-byte header + payload)
* @param ssrc - SSRC for RTP headers
* @param onDone - called when playback finishes
* @returns cancel function, or null if prompt has no G.722 frames
*/
export function playPromptG722(
prompt: ICachedPrompt,
sendPacket: (pkt: Buffer) => void,
ssrc: number,
onDone?: () => void,
): (() => void) | null {
if (prompt.g722Frames.length === 0) {
onDone?.();
return null;
}
const frames = prompt.g722Frames;
const PT = 9;
let frameIdx = 0;
let seq = Math.floor(Math.random() * 0xffff);
let rtpTs = Math.floor(Math.random() * 0xffffffff);
const timer = setInterval(() => {
if (frameIdx >= frames.length) {
clearInterval(timer);
onDone?.();
return;
}
const payload = frames[frameIdx];
const hdr = buildRtpHeader(PT, seq & 0xffff, rtpTs >>> 0, ssrc >>> 0, frameIdx === 0);
const pkt = Buffer.concat([hdr, payload]);
sendPacket(pkt);
seq++;
rtpTs += rtpClockIncrement(PT);
frameIdx++;
}, 20);
return () => clearInterval(timer);
}
/**
* Play a cached prompt's Opus frames as RTP packets at 20ms intervals.
*
* @param prompt - the cached prompt to play
* @param sendPacket - function to send a raw RTP packet
* @param ssrc - SSRC for RTP headers
* @param counters - shared seq/ts counters (mutated in place for seamless transitions)
* @param onDone - called when playback finishes
* @returns cancel function, or null if prompt has no Opus frames
*/
export function playPromptOpus(
prompt: ICachedPrompt,
sendPacket: (pkt: Buffer) => void,
ssrc: number,
counters: { seq: number; ts: number },
onDone?: () => void,
): (() => void) | null {
if (prompt.opusFrames.length === 0) {
onDone?.();
return null;
}
const frames = prompt.opusFrames;
const PT = 111;
let frameIdx = 0;
const timer = setInterval(() => {
if (frameIdx >= frames.length) {
clearInterval(timer);
onDone?.();
return;
}
const payload = frames[frameIdx];
const hdr = buildRtpHeader(PT, counters.seq & 0xffff, counters.ts >>> 0, ssrc >>> 0, frameIdx === 0);
const pkt = Buffer.concat([hdr, payload]);
sendPacket(pkt);
counters.seq++;
counters.ts += 960; // Opus 48kHz: 960 samples per 20ms
frameIdx++;
}, 20);
return () => clearInterval(timer);
}

View File

@@ -1,71 +0,0 @@
/**
* Unified RTP port pool — replaces the three separate allocators
* in sipproxy.ts, calloriginator.ts, and webrtcbridge.ts.
*
* Allocates even-numbered UDP ports from a configured range.
* Each allocation binds a dgram socket and returns it ready to use.
*/
import dgram from 'node:dgram';
export interface IRtpAllocation {
port: number;
sock: dgram.Socket;
}
export class RtpPortPool {
private min: number;
private max: number;
private allocated = new Map<number, dgram.Socket>();
private log: (msg: string) => void;
constructor(min: number, max: number, log: (msg: string) => void) {
this.min = min % 2 === 0 ? min : min + 1; // ensure even start
this.max = max;
this.log = log;
}
/**
* Allocate an even-numbered port and bind a UDP socket to it.
* Returns null if the pool is exhausted.
*/
allocate(): IRtpAllocation | null {
for (let port = this.min; port < this.max; port += 2) {
if (this.allocated.has(port)) continue;
const sock = dgram.createSocket('udp4');
try {
sock.bind(port, '0.0.0.0');
} catch {
try { sock.close(); } catch { /* ignore */ }
continue;
}
this.allocated.set(port, sock);
this.log(`[rtp-pool] allocated port ${port} (${this.allocated.size} in use)`);
return { port, sock };
}
this.log('[rtp-pool] WARN: port pool exhausted');
return null;
}
/**
* Release a port back to the pool and close its socket.
*/
release(port: number): void {
const sock = this.allocated.get(port);
if (!sock) return;
try { sock.close(); } catch { /* ignore */ }
this.allocated.delete(port);
this.log(`[rtp-pool] released port ${port} (${this.allocated.size} in use)`);
}
/** Number of currently allocated ports. */
get size(): number {
return this.allocated.size;
}
/** Total capacity (number of even ports in range). */
get capacity(): number {
return Math.floor((this.max - this.min) / 2);
}
}

View File

@@ -1,633 +0,0 @@
/**
* SipLeg — a SIP connection from the Call hub to a device or provider.
*
* Wraps a SipDialog and an RTP socket. Handles:
* - INVITE/ACK/BYE/CANCEL lifecycle
* - SDP rewriting (LAN IP for devices, public IP for providers)
* - Digest auth for provider legs (407/401)
* - Early-media silence for providers with quirks
* - Record-Route insertion for dialog-establishing requests
*/
import dgram from 'node:dgram';
import { Buffer } from 'node:buffer';
import {
SipMessage,
SipDialog,
buildSdp,
parseSdpEndpoint,
rewriteSdp,
rewriteSipUri,
parseDigestChallenge,
computeDigestAuth,
generateTag,
} from '../sip/index.ts';
import type { IEndpoint } from '../sip/index.ts';
import type { IProviderConfig, IQuirks } from '../config.ts';
import type { TLegState, TLegType, ILegStatus } from './types.ts';
import type { ILeg } from './leg.ts';
import { codecDisplayName } from './leg.ts';
import type { IRtpTranscoder } from '../codec.ts';
// ---------------------------------------------------------------------------
// SipLeg config
// ---------------------------------------------------------------------------
export interface ISipLegConfig {
/** Whether this leg faces a device (LAN) or a provider (WAN). */
role: 'device' | 'provider';
/** Proxy LAN IP (for SDP rewriting toward devices). */
lanIp: string;
/** Proxy LAN port (for Via, Contact, Record-Route). */
lanPort: number;
/** Public IP (for SDP rewriting toward providers). */
getPublicIp: () => string | null;
/** Send a SIP message via the main UDP socket. */
sendSip: (buf: Buffer, dest: IEndpoint) => void;
/** Logging function. */
log: (msg: string) => void;
/** Provider config (for provider legs: auth, codecs, quirks, outbound proxy). */
provider?: IProviderConfig;
/** The endpoint to send SIP messages to (device address or provider outbound proxy). */
sipTarget: IEndpoint;
/** RTP port and socket (pre-allocated from the pool). */
rtpPort: number;
rtpSock: dgram.Socket;
/** Payload types to offer in SDP. */
payloadTypes?: number[];
/** Registered AOR (for From header in provider leg). */
getRegisteredAor?: () => string | null;
/** SIP password (for digest auth). */
getSipPassword?: () => string | null;
}
// ---------------------------------------------------------------------------
// SipLeg
// ---------------------------------------------------------------------------
export class SipLeg implements ILeg {
readonly id: string;
readonly type: TLegType;
state: TLegState = 'inviting';
readonly config: ISipLegConfig;
/** The SIP dialog for this leg. */
dialog: SipDialog | null = null;
/** Original INVITE (needed for CANCEL). */
invite: SipMessage | null = null;
/** Original unauthenticated INVITE (for re-ACKing retransmitted 407s). */
private origInvite: SipMessage | null = null;
/** Whether we've attempted digest auth on this leg. */
private authAttempted = false;
/** RTP socket and port. */
readonly rtpPort: number;
readonly rtpSock: dgram.Socket;
/** Remote media endpoint (learned from SDP). */
remoteMedia: IEndpoint | null = null;
/** Negotiated codec. */
codec: number | null = null;
/** Transcoder (set by Call when codecs differ between legs). */
transcoder: IRtpTranscoder | null = null;
/** Stable SSRC for this leg (used for silence + forwarded audio). */
readonly ssrc: number = (Math.random() * 0xffffffff) >>> 0;
/** Packet counters. */
pktSent = 0;
pktReceived = 0;
/** Callback set by Call to receive RTP. */
onRtpReceived: ((data: Buffer) => void) | null = null;
/** Silence stream timer (for provider quirks). */
private silenceTimer: ReturnType<typeof setInterval> | null = null;
/** Callbacks for lifecycle events. */
onStateChange: ((leg: SipLeg) => void) | null = null;
onConnected: ((leg: SipLeg) => void) | null = null;
onTerminated: ((leg: SipLeg) => void) | null = null;
/** Callback for SIP INFO messages (used for DTMF relay). */
onInfoReceived: ((msg: SipMessage) => void) | null = null;
constructor(id: string, config: ISipLegConfig) {
this.id = id;
this.type = config.role === 'device' ? 'sip-device' : 'sip-provider';
this.config = config;
this.rtpPort = config.rtpPort;
this.rtpSock = config.rtpSock;
// Set up RTP receive handler.
this.rtpSock.on('message', (data: Buffer, rinfo: dgram.RemoteInfo) => {
this.pktReceived++;
// Learn remote media endpoint from first packet if not yet known.
if (!this.remoteMedia) {
this.remoteMedia = { address: rinfo.address, port: rinfo.port };
this.config.log(`[sip-leg:${this.id}] learned remote media: ${rinfo.address}:${rinfo.port}`);
}
// Forward to the Call hub.
if (this.onRtpReceived) {
this.onRtpReceived(data);
}
});
this.rtpSock.on('error', (e: Error) => {
this.config.log(`[sip-leg:${this.id}] rtp error: ${e.message}`);
});
}
get sipCallId(): string {
return this.dialog?.callId || 'no-dialog';
}
// -------------------------------------------------------------------------
// Outbound INVITE (B2BUA mode — create a new dialog)
// -------------------------------------------------------------------------
/**
* Send an INVITE to establish this leg.
* Creates a new SipDialog (UAC side).
*/
sendInvite(options: {
fromUri: string;
toUri: string;
callId: string;
fromTag?: string;
fromDisplayName?: string;
cseq?: number;
extraHeaders?: [string, string][];
}): void {
const ip = this.type === 'sip-provider'
? (this.config.getPublicIp() || this.config.lanIp)
: this.config.lanIp;
const pts = this.config.payloadTypes || [9, 0, 8, 101];
const sdp = buildSdp({ ip, port: this.rtpPort, payloadTypes: pts });
const invite = SipMessage.createRequest('INVITE', options.toUri, {
via: { host: ip, port: this.config.lanPort },
from: { uri: options.fromUri, displayName: options.fromDisplayName, tag: options.fromTag },
to: { uri: options.toUri },
callId: options.callId,
cseq: options.cseq,
contact: `<sip:${ip}:${this.config.lanPort}>`,
body: sdp,
contentType: 'application/sdp',
extraHeaders: options.extraHeaders,
});
this.invite = invite;
this.dialog = SipDialog.fromUacInvite(invite, ip, this.config.lanPort);
this.state = 'inviting';
this.config.log(`[sip-leg:${this.id}] INVITE -> ${this.config.sipTarget.address}:${this.config.sipTarget.port}`);
this.config.sendSip(invite.serialize(), this.config.sipTarget);
}
// -------------------------------------------------------------------------
// Passthrough mode — forward a SIP message with rewriting
// -------------------------------------------------------------------------
/**
* Accept an incoming INVITE as a UAS (for passthrough inbound calls).
* Creates a SipDialog on the UAS side.
*/
acceptIncoming(invite: SipMessage): void {
const localTag = generateTag();
this.dialog = SipDialog.fromUasInvite(invite, localTag, this.config.lanIp, this.config.lanPort);
this.invite = invite;
this.state = 'inviting';
// Learn remote media from SDP.
if (invite.hasSdpBody) {
const ep = parseSdpEndpoint(invite.body);
if (ep) {
this.remoteMedia = ep;
this.config.log(`[sip-leg:${this.id}] media from SDP: ${ep.address}:${ep.port}`);
}
}
}
/**
* Forward a SIP message through this leg with SDP rewriting.
* Used for passthrough calls where the proxy relays messages.
*/
forwardMessage(msg: SipMessage, dest: IEndpoint): void {
const rewriteIp = this.type === 'sip-provider'
? (this.config.getPublicIp() || this.config.lanIp)
: this.config.lanIp;
// Rewrite SDP if present.
if (msg.hasSdpBody) {
const { body, original } = rewriteSdp(msg.body, rewriteIp, this.rtpPort);
msg.body = body;
msg.updateContentLength();
if (original) {
this.remoteMedia = original;
this.config.log(`[sip-leg:${this.id}] media from SDP rewrite: ${original.address}:${original.port}`);
}
}
// Record-Route for dialog-establishing requests.
if (msg.isRequest && msg.isDialogEstablishing) {
msg.prependHeader('Record-Route', `<sip:${this.config.lanIp}:${this.config.lanPort};lr>`);
}
// Rewrite Contact.
if (this.type === 'sip-provider') {
const contact = msg.getHeader('Contact');
if (contact) {
const nc = rewriteSipUri(contact, rewriteIp, this.config.lanPort);
if (nc !== contact) msg.setHeader('Contact', nc);
}
}
// Rewrite Request-URI for inbound messages going to device.
if (this.type === 'sip-device' && msg.isRequest) {
msg.setRequestUri(rewriteSipUri(msg.requestUri!, dest.address, dest.port));
}
this.config.sendSip(msg.serialize(), dest);
}
// -------------------------------------------------------------------------
// SIP message handling (routed by CallManager)
// -------------------------------------------------------------------------
handleSipMessage(msg: SipMessage, rinfo: IEndpoint): void {
if (msg.isResponse) {
this.handleResponse(msg, rinfo);
} else {
this.handleRequest(msg, rinfo);
}
}
private handleResponse(msg: SipMessage, _rinfo: IEndpoint): void {
const code = msg.statusCode ?? 0;
const method = msg.cseqMethod?.toUpperCase();
this.config.log(`[sip-leg:${this.id}] <- ${code} (${method})`);
if (method === 'INVITE') {
this.handleInviteResponse(msg, code);
}
// BYE/CANCEL responses don't need action beyond logging.
}
private handleInviteResponse(msg: SipMessage, code: number): void {
// Handle retransmitted 407 for the original unauthenticated INVITE.
if (this.authAttempted && this.dialog) {
const responseCSeqNum = parseInt((msg.getHeader('CSeq') || '').split(/\s+/)[0], 10);
if (responseCSeqNum < this.dialog.localCSeq && code >= 400) {
if (this.origInvite) {
const ack = buildNon2xxAck(this.origInvite, msg);
this.config.sendSip(ack.serialize(), this.config.sipTarget);
}
return;
}
}
// Handle 407 Proxy Authentication Required.
if (code === 407 && this.type === 'sip-provider') {
this.handleAuthChallenge(msg);
return;
}
// Update dialog state.
if (this.dialog) {
this.dialog.processResponse(msg);
}
if (code === 180 || code === 183) {
this.state = 'ringing';
this.onStateChange?.(this);
} else if (code >= 200 && code < 300) {
// ACK the 200 OK.
if (this.dialog) {
const ack = this.dialog.createAck();
this.config.sendSip(ack.serialize(), this.config.sipTarget);
this.config.log(`[sip-leg:${this.id}] ACK sent`);
}
// If already connected (200 retransmit), just re-ACK.
if (this.state === 'connected') {
this.config.log(`[sip-leg:${this.id}] re-ACK (200 retransmit)`);
return;
}
// Learn media endpoint from SDP.
if (msg.hasSdpBody) {
const ep = parseSdpEndpoint(msg.body);
if (ep) {
this.remoteMedia = ep;
this.config.log(`[sip-leg:${this.id}] media = ${ep.address}:${ep.port}`);
}
}
this.state = 'connected';
this.config.log(`[sip-leg:${this.id}] CONNECTED`);
// Start silence for provider legs with early media quirks.
if (this.type === 'sip-provider') {
this.startSilence();
}
// Prime the RTP path.
if (this.remoteMedia) {
this.primeRtp(this.remoteMedia);
}
this.onConnected?.(this);
this.onStateChange?.(this);
} else if (code >= 300) {
this.config.log(`[sip-leg:${this.id}] rejected ${code}`);
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
this.onStateChange?.(this);
}
}
private handleAuthChallenge(msg: SipMessage): void {
if (this.authAttempted) {
this.config.log(`[sip-leg:${this.id}] 407 after auth attempt — credentials rejected`);
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
return;
}
this.authAttempted = true;
const challenge = msg.getHeader('Proxy-Authenticate');
if (!challenge) {
this.config.log(`[sip-leg:${this.id}] 407 but no Proxy-Authenticate`);
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
return;
}
const parsed = parseDigestChallenge(challenge);
if (!parsed) {
this.config.log(`[sip-leg:${this.id}] could not parse digest challenge`);
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
return;
}
const password = this.config.getSipPassword?.();
const aor = this.config.getRegisteredAor?.();
if (!password || !aor) {
this.config.log(`[sip-leg:${this.id}] 407 but no password or AOR`);
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
return;
}
const username = aor.replace(/^sips?:/, '').split('@')[0];
const destUri = this.invite?.requestUri || '';
const authValue = computeDigestAuth({
username,
password,
realm: parsed.realm,
nonce: parsed.nonce,
method: 'INVITE',
uri: destUri,
algorithm: parsed.algorithm,
opaque: parsed.opaque,
});
// ACK the 407.
if (this.invite) {
const ack407 = buildNon2xxAck(this.invite, msg);
this.config.sendSip(ack407.serialize(), this.config.sipTarget);
this.config.log(`[sip-leg:${this.id}] ACK-407 sent`);
}
// Keep original INVITE for re-ACKing retransmitted 407s.
this.origInvite = this.invite;
// Resend INVITE with auth, same From tag, incremented CSeq.
const ip = this.config.getPublicIp() || this.config.lanIp;
const fromTag = this.dialog!.localTag;
const pts = this.config.payloadTypes || [9, 0, 8, 101];
const sdp = buildSdp({ ip, port: this.rtpPort, payloadTypes: pts });
const inviteAuth = SipMessage.createRequest('INVITE', destUri, {
via: { host: ip, port: this.config.lanPort },
from: { uri: aor, tag: fromTag },
to: { uri: destUri },
callId: this.dialog!.callId,
cseq: 2,
contact: `<sip:${ip}:${this.config.lanPort}>`,
body: sdp,
contentType: 'application/sdp',
extraHeaders: [['Proxy-Authorization', authValue]],
});
this.invite = inviteAuth;
this.dialog!.localCSeq = 2;
this.config.log(`[sip-leg:${this.id}] resending INVITE with auth`);
this.config.sendSip(inviteAuth.serialize(), this.config.sipTarget);
}
private handleRequest(msg: SipMessage, rinfo: IEndpoint): void {
const method = msg.method;
this.config.log(`[sip-leg:${this.id}] <- ${method} from ${rinfo.address}:${rinfo.port}`);
if (method === 'BYE') {
// Send 200 OK to the BYE.
const ok = SipMessage.createResponse(200, 'OK', msg);
this.config.sendSip(ok.serialize(), { address: rinfo.address, port: rinfo.port });
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
this.onTerminated?.(this);
this.onStateChange?.(this);
}
if (method === 'INFO') {
// Respond 200 OK to the INFO request.
const ok = SipMessage.createResponse(200, 'OK', msg);
this.config.sendSip(ok.serialize(), { address: rinfo.address, port: rinfo.port });
// Forward to DTMF handler (if attached).
this.onInfoReceived?.(msg);
}
// Other in-dialog requests (re-INVITE, etc.) can be handled here in the future.
}
// -------------------------------------------------------------------------
// Send BYE / CANCEL
// -------------------------------------------------------------------------
/** Send BYE (if confirmed) or CANCEL (if early) to tear down this leg. */
sendHangup(): void {
if (!this.dialog) return;
if (this.dialog.state === 'confirmed') {
const bye = this.dialog.createRequest('BYE');
this.config.sendSip(bye.serialize(), this.config.sipTarget);
this.config.log(`[sip-leg:${this.id}] BYE sent`);
} else if (this.dialog.state === 'early' && this.invite) {
const cancel = this.dialog.createCancel(this.invite);
this.config.sendSip(cancel.serialize(), this.config.sipTarget);
this.config.log(`[sip-leg:${this.id}] CANCEL sent`);
}
this.state = 'terminating';
this.dialog.terminate();
}
// -------------------------------------------------------------------------
// RTP
// -------------------------------------------------------------------------
sendRtp(data: Buffer): void {
if (!this.remoteMedia) return;
this.rtpSock.send(data, this.remoteMedia.port, this.remoteMedia.address);
this.pktSent++;
}
/** Send a 1-byte UDP packet to punch NAT hole. */
private primeRtp(peer: IEndpoint): void {
try {
this.rtpSock.send(Buffer.alloc(1), peer.port, peer.address);
this.config.log(`[sip-leg:${this.id}] RTP primed -> ${peer.address}:${peer.port}`);
} catch (e: any) {
this.config.log(`[sip-leg:${this.id}] prime error: ${e.message}`);
}
}
// -------------------------------------------------------------------------
// Silence stream (provider quirks)
// -------------------------------------------------------------------------
private startSilence(): void {
if (this.silenceTimer) return;
const quirks = this.config.provider?.quirks;
if (!quirks?.earlyMediaSilence) return;
if (!this.remoteMedia) return;
const PT = quirks.silencePayloadType ?? 9;
const MAX = quirks.silenceMaxPackets ?? 250;
const PAYLOAD = 160;
let seq = Math.floor(Math.random() * 0xffff);
let rtpTs = Math.floor(Math.random() * 0xffffffff);
let count = 0;
// Use proper silence byte for the codec (0x00 is NOT silence for most codecs).
const silenceByte = silenceByteForPT(PT);
this.silenceTimer = setInterval(() => {
if (this.pktReceived > 0 || count >= MAX) {
clearInterval(this.silenceTimer!);
this.silenceTimer = null;
this.config.log(`[sip-leg:${this.id}] silence stop after ${count} pkts`);
return;
}
const pkt = Buffer.alloc(12 + PAYLOAD, silenceByte);
// RTP header (first 12 bytes).
pkt[0] = 0x80;
pkt[1] = PT;
pkt.writeUInt16BE(seq & 0xffff, 2);
pkt.writeUInt32BE(rtpTs >>> 0, 4);
pkt.writeUInt32BE(this.ssrc >>> 0, 8); // stable SSRC
this.rtpSock.send(pkt, this.remoteMedia!.port, this.remoteMedia!.address);
seq++;
rtpTs += PAYLOAD;
count++;
}, 20);
this.config.log(`[sip-leg:${this.id}] silence start -> ${this.remoteMedia.address}:${this.remoteMedia.port} (ssrc=${this.ssrc})`);
}
// -------------------------------------------------------------------------
// Lifecycle
// -------------------------------------------------------------------------
teardown(): void {
if (this.silenceTimer) {
clearInterval(this.silenceTimer);
this.silenceTimer = null;
}
this.state = 'terminated';
if (this.dialog) this.dialog.terminate();
// Note: RTP socket is NOT closed here — the RtpPortPool manages that.
}
getStatus(): ILegStatus {
return {
id: this.id,
type: this.type,
state: this.state,
remoteMedia: this.remoteMedia,
rtpPort: this.rtpPort,
pktSent: this.pktSent,
pktReceived: this.pktReceived,
codec: codecDisplayName(this.codec),
transcoding: this.transcoder !== null,
};
}
}
// ---------------------------------------------------------------------------
// Helper: proper silence byte per codec
// ---------------------------------------------------------------------------
/** Return the byte value representing digital silence for a given RTP payload type. */
function silenceByteForPT(pt: number): number {
switch (pt) {
case 0: return 0xFF; // PCMU: μ-law silence (zero amplitude)
case 8: return 0xD5; // PCMA: A-law silence (zero amplitude)
case 9: return 0xD5; // G.722: sub-band silence (zero amplitude)
default: return 0xFF; // safe default
}
}
// ---------------------------------------------------------------------------
// Helper: ACK for non-2xx (same transaction)
// ---------------------------------------------------------------------------
function buildNon2xxAck(originalInvite: SipMessage, response: SipMessage): SipMessage {
const via = originalInvite.getHeader('Via') || '';
const from = originalInvite.getHeader('From') || '';
const toFromResponse = response.getHeader('To') || '';
const callId = originalInvite.callId;
const cseqNum = parseInt((originalInvite.getHeader('CSeq') || '1').split(/\s+/)[0], 10);
return new SipMessage(
`ACK ${originalInvite.requestUri} SIP/2.0`,
[
['Via', via],
['From', from],
['To', toFromResponse],
['Call-ID', callId],
['CSeq', `${cseqNum} ACK`],
['Max-Forwards', '70'],
['Content-Length', '0'],
],
'',
);
}

View File

@@ -1,336 +0,0 @@
/**
* SystemLeg — virtual ILeg for IVR menus and voicemail.
*
* Plugs into the Call hub exactly like SipLeg or WebRtcLeg:
* - Receives caller audio via sendRtp() (called by Call.forwardRtp)
* - Plays prompts by firing onRtpReceived (picked up by Call.forwardRtp → caller's leg)
* - Detects DTMF from caller's audio (RFC 2833 telephone-event)
* - Records caller's audio to WAV files (for voicemail)
*
* No UDP socket or SIP dialog needed — purely virtual.
*/
import { Buffer } from 'node:buffer';
import type dgram from 'node:dgram';
import type { IEndpoint } from '../sip/index.ts';
import type { SipMessage } from '../sip/index.ts';
import type { SipDialog } from '../sip/index.ts';
import type { IRtpTranscoder } from '../codec.ts';
import type { ILeg } from './leg.ts';
import type { TLegState, TLegType, ILegStatus } from './types.ts';
import { DtmfDetector } from './dtmf-detector.ts';
import type { IDtmfDigit } from './dtmf-detector.ts';
import { AudioRecorder } from './audio-recorder.ts';
import type { IRecordingResult } from './audio-recorder.ts';
import { PromptCache, playPromptG722, playPromptOpus } from './prompt-cache.ts';
import type { ICachedPrompt } from './prompt-cache.ts';
import { buildRtpHeader, rtpClockIncrement } from './leg.ts';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
export type TSystemLegMode = 'ivr' | 'voicemail-greeting' | 'voicemail-recording' | 'idle';
export interface ISystemLegConfig {
/** Logging function. */
log: (msg: string) => void;
/** The prompt cache for TTS playback. */
promptCache: PromptCache;
/**
* Codec payload type used by the caller's leg.
* Determines whether G.722 (9) or Opus (111) frames are played.
* Default: 9 (G.722, typical for SIP callers).
*/
callerCodecPt?: number;
/** Called when a DTMF digit is detected. */
onDtmfDigit?: (digit: IDtmfDigit) => void;
/** Called when a voicemail recording is complete. */
onRecordingComplete?: (result: IRecordingResult) => void;
/** Called when the SystemLeg wants to signal an IVR action. */
onAction?: (action: string, data?: any) => void;
}
// ---------------------------------------------------------------------------
// SystemLeg
// ---------------------------------------------------------------------------
export class SystemLeg implements ILeg {
readonly id: string;
readonly type: TLegType = 'system';
state: TLegState = 'connected'; // Immediately "connected" — no setup phase.
/** Current operating mode. */
mode: TSystemLegMode = 'idle';
// --- ILeg required fields (virtual — no real network resources) ---
readonly sipCallId: string;
readonly rtpPort: number | null = null;
readonly rtpSock: dgram.Socket | null = null;
remoteMedia: IEndpoint | null = null;
codec: number | null = null;
transcoder: IRtpTranscoder | null = null;
pktSent = 0;
pktReceived = 0;
readonly dialog: SipDialog | null = null;
/**
* Set by Call.addLeg() — firing this injects audio into the Call hub,
* which forwards it to the caller's leg.
*/
onRtpReceived: ((data: Buffer) => void) | null = null;
// --- Internal components ---
private dtmfDetector: DtmfDetector;
private recorder: AudioRecorder | null = null;
private promptCache: PromptCache;
private promptCancel: (() => void) | null = null;
private callerCodecPt: number;
private log: (msg: string) => void;
readonly config: ISystemLegConfig;
/** Stable SSRC for all prompt playback (random, stays constant for the leg's lifetime). */
private ssrc: number;
/** Sequence/timestamp counters for Opus prompt playback (shared for seamless transitions). */
private opusCounters = { seq: 0, ts: 0 };
constructor(id: string, config: ISystemLegConfig) {
this.id = id;
this.sipCallId = `system-${id}`; // Virtual Call-ID — not a real SIP dialog.
this.config = config;
this.log = config.log;
this.promptCache = config.promptCache;
this.callerCodecPt = config.callerCodecPt ?? 9; // Default G.722
this.ssrc = (Math.random() * 0xffffffff) >>> 0;
this.opusCounters.seq = Math.floor(Math.random() * 0xffff);
this.opusCounters.ts = Math.floor(Math.random() * 0xffffffff);
// Initialize DTMF detector.
this.dtmfDetector = new DtmfDetector(this.log);
this.dtmfDetector.onDigit = (digit) => {
this.log(`[system-leg:${this.id}] DTMF '${digit.digit}' (${digit.source})`);
this.config.onDtmfDigit?.(digit);
};
}
// -------------------------------------------------------------------------
// ILeg: sendRtp — receives caller's audio from the Call hub
// -------------------------------------------------------------------------
/**
* Called by the Call hub (via forwardRtp) to deliver the caller's audio
* to this leg. We use this for DTMF detection and recording.
*/
sendRtp(data: Buffer): void {
this.pktReceived++;
// Feed DTMF detector (it checks PT internally, ignores non-101 packets).
this.dtmfDetector.processRtp(data);
// Feed recorder if active.
if (this.mode === 'voicemail-recording' && this.recorder) {
this.recorder.processRtp(data);
}
}
// -------------------------------------------------------------------------
// ILeg: handleSipMessage — handles SIP INFO for DTMF
// -------------------------------------------------------------------------
/**
* Handle a SIP message routed to this leg. Only SIP INFO (DTMF) is relevant.
*/
handleSipMessage(msg: SipMessage, _rinfo: IEndpoint): void {
if (msg.method === 'INFO') {
this.dtmfDetector.processSipInfo(msg);
}
}
// -------------------------------------------------------------------------
// Prompt playback
// -------------------------------------------------------------------------
/**
* Play a cached prompt by ID.
* The audio is injected into the Call hub via onRtpReceived.
*
* @param promptId - ID of the prompt in the PromptCache
* @param onDone - called when playback completes (not on cancel)
* @returns true if playback started, false if prompt not found
*/
playPrompt(promptId: string, onDone?: () => void): boolean {
const prompt = this.promptCache.get(promptId);
if (!prompt) {
this.log(`[system-leg:${this.id}] prompt "${promptId}" not found`);
onDone?.();
return false;
}
// Cancel any in-progress playback.
this.cancelPrompt();
this.log(`[system-leg:${this.id}] playing prompt "${promptId}" (${prompt.durationMs}ms)`);
// Select G.722 or Opus frames based on caller codec.
if (this.callerCodecPt === 111) {
// WebRTC caller: play Opus frames.
this.promptCancel = playPromptOpus(
prompt,
(pkt) => this.injectPacket(pkt),
this.ssrc,
this.opusCounters,
() => {
this.promptCancel = null;
onDone?.();
},
);
} else {
// SIP caller: play G.722 frames (works for all SIP codecs since the
// SipLeg's RTP socket sends whatever we give it — the provider's
// media endpoint accepts the codec negotiated in the SDP).
this.promptCancel = playPromptG722(
prompt,
(pkt) => this.injectPacket(pkt),
this.ssrc,
() => {
this.promptCancel = null;
onDone?.();
},
);
}
return this.promptCancel !== null;
}
/**
* Play a sequence of prompts, one after another.
*/
playPromptSequence(promptIds: string[], onDone?: () => void): void {
let index = 0;
const playNext = () => {
if (index >= promptIds.length) {
onDone?.();
return;
}
const id = promptIds[index++];
if (!this.playPrompt(id, playNext)) {
// Prompt not found — skip and play next.
playNext();
}
};
playNext();
}
/** Cancel any in-progress prompt playback. */
cancelPrompt(): void {
if (this.promptCancel) {
this.promptCancel();
this.promptCancel = null;
}
}
/** Whether a prompt is currently playing. */
get isPlaying(): boolean {
return this.promptCancel !== null;
}
/**
* Inject an RTP packet into the Call hub.
* This simulates "receiving" audio on this leg — the hub
* will forward it to the caller's leg.
*/
private injectPacket(pkt: Buffer): void {
this.pktSent++;
this.onRtpReceived?.(pkt);
}
// -------------------------------------------------------------------------
// Recording
// -------------------------------------------------------------------------
/**
* Start recording the caller's audio.
* @param outputDir - directory to write the WAV file
* @param fileId - unique ID for the file name
*/
async startRecording(outputDir: string, fileId?: string): Promise<void> {
if (this.recorder) {
await this.recorder.stop();
}
this.recorder = new AudioRecorder({
outputDir,
log: this.log,
maxDurationSec: 120,
silenceTimeoutSec: 5,
});
this.recorder.onStopped = (result) => {
this.log(`[system-leg:${this.id}] recording auto-stopped (${result.stopReason})`);
this.config.onRecordingComplete?.(result);
};
this.mode = 'voicemail-recording';
await this.recorder.start(fileId);
}
/**
* Stop recording and finalize the WAV file.
*/
async stopRecording(): Promise<IRecordingResult | null> {
if (!this.recorder) return null;
const result = await this.recorder.stop();
this.recorder = null;
return result;
}
/** Cancel recording — stops and deletes the file. */
async cancelRecording(): Promise<void> {
if (this.recorder) {
await this.recorder.cancel();
this.recorder = null;
}
}
// -------------------------------------------------------------------------
// Lifecycle
// -------------------------------------------------------------------------
/** Release all resources. */
teardown(): void {
this.cancelPrompt();
// Stop recording gracefully.
if (this.recorder && this.recorder.state === 'recording') {
this.recorder.stop().then((result) => {
this.config.onRecordingComplete?.(result);
});
this.recorder = null;
}
this.dtmfDetector.destroy();
this.state = 'terminated';
this.mode = 'idle';
this.onRtpReceived = null;
this.log(`[system-leg:${this.id}] torn down`);
}
/** Status snapshot for the dashboard. */
getStatus(): ILegStatus {
return {
id: this.id,
type: this.type,
state: this.state,
remoteMedia: null,
rtpPort: null,
pktSent: this.pktSent,
pktReceived: this.pktReceived,
codec: this.callerCodecPt === 111 ? 'Opus' : 'G.722',
transcoding: false,
};
}
}

View File

@@ -1,70 +0,0 @@
/**
* Hub model type definitions — Call, Leg, and status types.
*/
import type { IEndpoint } from '../sip/index.ts';
// ---------------------------------------------------------------------------
// State types
// ---------------------------------------------------------------------------
export type TCallState =
| 'setting-up'
| 'ringing'
| 'connected'
| 'on-hold'
| 'voicemail'
| 'ivr'
| 'transferring'
| 'terminating'
| 'terminated';
export type TLegState =
| 'inviting'
| 'ringing'
| 'connected'
| 'on-hold'
| 'terminating'
| 'terminated';
export type TLegType = 'sip-device' | 'sip-provider' | 'webrtc' | 'system';
export type TCallDirection = 'inbound' | 'outbound' | 'internal';
// ---------------------------------------------------------------------------
// Status interfaces (for frontend dashboard)
// ---------------------------------------------------------------------------
export interface ILegStatus {
id: string;
type: TLegType;
state: TLegState;
remoteMedia: IEndpoint | null;
rtpPort: number | null;
pktSent: number;
pktReceived: number;
codec: string | null;
transcoding: boolean;
}
export interface ICallStatus {
id: string;
state: TCallState;
direction: TCallDirection;
callerNumber: string | null;
calleeNumber: string | null;
providerUsed: string | null;
createdAt: number;
duration: number;
legs: ILegStatus[];
}
export interface ICallHistoryEntry {
id: string;
direction: TCallDirection;
callerNumber: string | null;
calleeNumber: string | null;
providerUsed: string | null;
startedAt: number;
duration: number;
}

View File

@@ -1,163 +0,0 @@
/**
* Streaming WAV file writer — opens a file, writes a placeholder header,
* appends raw PCM data in chunks, and finalizes (patches sizes) on close.
*
* Produces standard RIFF/WAVE format compatible with the WAV parser
* in announcement.ts (extractPcmFromWav).
*/
import fs from 'node:fs';
import { Buffer } from 'node:buffer';
// ---------------------------------------------------------------------------
// Types
// ---------------------------------------------------------------------------
export interface IWavWriterOptions {
/** Full path to the output WAV file. */
filePath: string;
/** Sample rate in Hz (e.g. 16000). */
sampleRate: number;
/** Number of channels (default 1 = mono). */
channels?: number;
/** Bits per sample (default 16). */
bitsPerSample?: number;
}
export interface IWavWriterResult {
/** Full path to the WAV file. */
filePath: string;
/** Total duration in milliseconds. */
durationMs: number;
/** Sample rate of the output file. */
sampleRate: number;
/** Total number of audio samples written. */
totalSamples: number;
/** File size in bytes. */
fileSize: number;
}
// ---------------------------------------------------------------------------
// WAV header constants
// ---------------------------------------------------------------------------
/** Standard WAV header size: RIFF(12) + fmt(24) + data-header(8) = 44 bytes. */
const HEADER_SIZE = 44;
// ---------------------------------------------------------------------------
// WavWriter
// ---------------------------------------------------------------------------
export class WavWriter {
private fd: number | null = null;
private totalDataBytes = 0;
private closed = false;
private filePath: string;
private sampleRate: number;
private channels: number;
private bitsPerSample: number;
constructor(options: IWavWriterOptions) {
this.filePath = options.filePath;
this.sampleRate = options.sampleRate;
this.channels = options.channels ?? 1;
this.bitsPerSample = options.bitsPerSample ?? 16;
}
/** Open the file and write a placeholder 44-byte WAV header. */
open(): void {
if (this.fd !== null) throw new Error('WavWriter already open');
this.fd = fs.openSync(this.filePath, 'w');
this.totalDataBytes = 0;
this.closed = false;
// Write 44 bytes of zeros as placeholder — patched in close().
const placeholder = Buffer.alloc(HEADER_SIZE);
fs.writeSync(this.fd, placeholder, 0, HEADER_SIZE, 0);
}
/** Append raw 16-bit little-endian PCM samples. */
write(pcm: Buffer): void {
if (this.fd === null || this.closed) return;
if (pcm.length === 0) return;
fs.writeSync(this.fd, pcm, 0, pcm.length);
this.totalDataBytes += pcm.length;
}
/**
* Finalize: rewrite the RIFF and data chunk sizes in the header, close the file.
* Returns metadata about the written WAV.
*/
close(): IWavWriterResult {
if (this.fd === null || this.closed) {
return {
filePath: this.filePath,
durationMs: 0,
sampleRate: this.sampleRate,
totalSamples: 0,
fileSize: HEADER_SIZE,
};
}
this.closed = true;
const blockAlign = this.channels * (this.bitsPerSample / 8);
const byteRate = this.sampleRate * blockAlign;
const fileSize = HEADER_SIZE + this.totalDataBytes;
// Build the complete 44-byte header.
const hdr = Buffer.alloc(HEADER_SIZE);
let offset = 0;
// RIFF chunk descriptor.
hdr.write('RIFF', offset); offset += 4;
hdr.writeUInt32LE(fileSize - 8, offset); offset += 4; // ChunkSize = fileSize - 8
hdr.write('WAVE', offset); offset += 4;
// fmt sub-chunk.
hdr.write('fmt ', offset); offset += 4;
hdr.writeUInt32LE(16, offset); offset += 4; // Subchunk1Size (PCM = 16)
hdr.writeUInt16LE(1, offset); offset += 2; // AudioFormat (1 = PCM)
hdr.writeUInt16LE(this.channels, offset); offset += 2;
hdr.writeUInt32LE(this.sampleRate, offset); offset += 4;
hdr.writeUInt32LE(byteRate, offset); offset += 4;
hdr.writeUInt16LE(blockAlign, offset); offset += 2;
hdr.writeUInt16LE(this.bitsPerSample, offset); offset += 2;
// data sub-chunk.
hdr.write('data', offset); offset += 4;
hdr.writeUInt32LE(this.totalDataBytes, offset); offset += 4;
// Patch the header at the beginning of the file.
fs.writeSync(this.fd, hdr, 0, HEADER_SIZE, 0);
fs.closeSync(this.fd);
this.fd = null;
const bytesPerSample = this.bitsPerSample / 8;
const totalSamples = Math.floor(this.totalDataBytes / (bytesPerSample * this.channels));
const durationMs = (totalSamples / this.sampleRate) * 1000;
return {
filePath: this.filePath,
durationMs: Math.round(durationMs),
sampleRate: this.sampleRate,
totalSamples,
fileSize,
};
}
/** Current recording duration in milliseconds. */
get durationMs(): number {
const bytesPerSample = this.bitsPerSample / 8;
const totalSamples = Math.floor(this.totalDataBytes / (bytesPerSample * this.channels));
return (totalSamples / this.sampleRate) * 1000;
}
/** Whether the writer is still open and accepting data. */
get isOpen(): boolean {
return this.fd !== null && !this.closed;
}
}

View File

@@ -1,417 +0,0 @@
/**
* WebRtcLeg — a WebRTC connection from the Call hub to a browser client.
*
* Wraps a werift RTCPeerConnection and handles:
* - WebRTC offer/answer/ICE negotiation
* - Opus <-> G.722/PCMU/PCMA transcoding via Rust IPC
* - RTP header rebuilding with correct PT, timestamp, SSRC
*/
import dgram from 'node:dgram';
import { Buffer } from 'node:buffer';
import { WebSocket } from 'ws';
import type { IEndpoint } from '../sip/index.ts';
import type { TLegState, ILegStatus } from './types.ts';
import type { ILeg } from './leg.ts';
import { rtpClockIncrement, buildRtpHeader, codecDisplayName } from './leg.ts';
import { createTranscoder, OPUS_PT } from '../codec.ts';
import type { IRtpTranscoder } from '../codec.ts';
import { createSession, destroySession } from '../opusbridge.ts';
import type { SipDialog } from '../sip/index.ts';
import type { SipMessage } from '../sip/index.ts';
// ---------------------------------------------------------------------------
// WebRtcLeg config
// ---------------------------------------------------------------------------
export interface IWebRtcLegConfig {
/** The browser's WebSocket connection. */
ws: WebSocket;
/** The browser's session ID. */
sessionId: string;
/** RTP port and socket (pre-allocated from the pool). */
rtpPort: number;
rtpSock: dgram.Socket;
/** Logging function. */
log: (msg: string) => void;
}
// ---------------------------------------------------------------------------
// WebRtcLeg
// ---------------------------------------------------------------------------
export class WebRtcLeg implements ILeg {
readonly id: string;
readonly type = 'webrtc' as const;
state: TLegState = 'inviting';
readonly sessionId: string;
/** The werift RTCPeerConnection instance. */
private pc: any = null;
/** RTP socket for bridging to SIP. */
readonly rtpSock: dgram.Socket;
readonly rtpPort: number;
/** Remote media endpoint (the other side of the bridge, set by Call). */
remoteMedia: IEndpoint | null = null;
/** Negotiated WebRTC codec payload type. */
codec: number | null = null;
/** Transcoders for WebRTC <-> SIP conversion. */
transcoder: IRtpTranscoder | null = null; // used by Call for fan-out
private toSipTranscoder: IRtpTranscoder | null = null;
private fromSipTranscoder: IRtpTranscoder | null = null;
/** RTP counters for outgoing (to SIP) direction. */
private toSipSeq = 0;
private toSipTs = 0;
private toSipSsrc = (Math.random() * 0xffffffff) >>> 0;
/** RTP counters for incoming (from SIP) direction.
* Initialized to random values so announcements and provider audio share
* a continuous sequence — prevents the browser jitter buffer from discarding
* packets after the announcement→provider transition. */
readonly fromSipCounters = {
seq: Math.floor(Math.random() * 0xffff),
ts: Math.floor(Math.random() * 0xffffffff),
};
fromSipSsrc = (Math.random() * 0xffffffff) >>> 0;
/** Packet counters. */
pktSent = 0;
pktReceived = 0;
/** Callback set by Call. */
onRtpReceived: ((data: Buffer) => void) | null = null;
/** Callback to send transcoded RTP to the provider via the SipLeg's socket.
* Set by CallManager when the bridge is established. If null, falls back to own rtpSock. */
onSendToProvider: ((data: Buffer, dest: IEndpoint) => void) | null = null;
/** Lifecycle callbacks. */
onConnected: ((leg: WebRtcLeg) => void) | null = null;
onTerminated: ((leg: WebRtcLeg) => void) | null = null;
/** Cancel handle for an in-progress announcement. */
announcementCancel: (() => void) | null = null;
private ws: WebSocket;
private config: IWebRtcLegConfig;
private pendingIceCandidates: any[] = [];
// SipDialog is not applicable for WebRTC legs.
readonly dialog: SipDialog | null = null;
readonly sipCallId: string;
constructor(id: string, config: IWebRtcLegConfig) {
this.id = id;
this.sessionId = config.sessionId;
this.ws = config.ws;
this.rtpSock = config.rtpSock;
this.rtpPort = config.rtpPort;
this.config = config;
this.sipCallId = `webrtc-${id}`;
// Log RTP arriving on this socket (symmetric RTP from provider).
// Audio forwarding is handled by the Call hub: SipLeg → forwardRtp → WebRtcLeg.sendRtp.
// We do NOT transcode here to avoid double-processing (the SipLeg also receives these packets).
let sipRxCount = 0;
this.rtpSock.on('message', (data: Buffer) => {
sipRxCount++;
if (sipRxCount === 1 || sipRxCount === 50 || sipRxCount % 500 === 0) {
this.config.log(`[webrtc-leg:${this.id}] SIP->browser rtp #${sipRxCount} (${data.length}b) [symmetric, ignored]`);
}
});
}
// -------------------------------------------------------------------------
// WebRTC offer/answer
// -------------------------------------------------------------------------
/**
* Handle a WebRTC offer from the browser. Creates the PeerConnection,
* sets remote offer, creates answer, and sends it back.
*/
async handleOffer(offerSdp: string): Promise<void> {
this.config.log(`[webrtc-leg:${this.id}] received offer`);
try {
const werift = await import('werift');
this.pc = new werift.RTCPeerConnection({ iceServers: [] });
// Add sendrecv transceiver before setRemoteDescription.
this.pc.addTransceiver('audio', { direction: 'sendrecv' });
// Handle incoming audio from browser.
this.pc.ontrack = (event: any) => {
const track = event.track;
this.config.log(`[webrtc-leg:${this.id}] got track: ${track.kind}`);
let rxCount = 0;
track.onReceiveRtp.subscribe((rtp: any) => {
if (!this.remoteMedia) return;
rxCount++;
if (rxCount === 1 || rxCount === 50 || rxCount % 500 === 0) {
this.config.log(`[webrtc-leg:${this.id}] browser->SIP rtp #${rxCount}`);
}
this.forwardToSip(rtp, rxCount);
});
};
// ICE candidate handling.
this.pc.onicecandidate = (candidate: any) => {
if (candidate) {
const json = candidate.toJSON?.() || candidate;
this.wsSend({ type: 'webrtc-ice', sessionId: this.sessionId, candidate: json });
}
};
this.pc.onconnectionstatechange = () => {
this.config.log(`[webrtc-leg:${this.id}] connection state: ${this.pc.connectionState}`);
if (this.pc.connectionState === 'connected') {
this.state = 'connected';
this.onConnected?.(this);
} else if (this.pc.connectionState === 'failed' || this.pc.connectionState === 'closed') {
this.state = 'terminated';
this.onTerminated?.(this);
}
};
if (this.pc.oniceconnectionstatechange !== undefined) {
this.pc.oniceconnectionstatechange = () => {
this.config.log(`[webrtc-leg:${this.id}] ICE state: ${this.pc.iceConnectionState}`);
};
}
// Set remote offer and create answer.
await this.pc.setRemoteDescription({ type: 'offer', sdp: offerSdp });
const answer = await this.pc.createAnswer();
await this.pc.setLocalDescription(answer);
const sdp: string = this.pc.localDescription!.sdp;
// Detect negotiated codec.
const mAudio = sdp.match(/m=audio\s+\d+\s+\S+\s+(\d+)/);
if (mAudio) {
this.codec = parseInt(mAudio[1], 10);
this.config.log(`[webrtc-leg:${this.id}] negotiated audio PT=${this.codec}`);
}
// Extract sender SSRC from SDP.
const ssrcMatch = sdp.match(/a=ssrc:(\d+)\s/);
if (ssrcMatch) {
this.fromSipSsrc = parseInt(ssrcMatch[1], 10);
}
// Also try from sender object.
const senders = this.pc.getSenders();
if (senders[0]) {
const senderSsrc = (senders[0] as any).ssrc ?? (senders[0] as any)._ssrc;
if (senderSsrc) this.fromSipSsrc = senderSsrc;
}
// Send answer to browser.
this.wsSend({ type: 'webrtc-answer', sessionId: this.sessionId, sdp });
this.config.log(`[webrtc-leg:${this.id}] sent answer, rtp port=${this.rtpPort}`);
// Process buffered ICE candidates.
for (const c of this.pendingIceCandidates) {
try { await this.pc.addIceCandidate(c); } catch { /* ignore */ }
}
this.pendingIceCandidates = [];
} catch (err: any) {
this.config.log(`[webrtc-leg:${this.id}] offer error: ${err.message}`);
this.wsSend({ type: 'webrtc-error', sessionId: this.sessionId, error: err.message });
this.state = 'terminated';
this.onTerminated?.(this);
}
}
/** Add an ICE candidate from the browser. */
async addIceCandidate(candidate: any): Promise<void> {
if (!this.pc) {
this.pendingIceCandidates.push(candidate);
return;
}
try {
if (candidate) await this.pc.addIceCandidate(candidate);
} catch (err: any) {
this.config.log(`[webrtc-leg:${this.id}] ICE error: ${err.message}`);
}
}
// -------------------------------------------------------------------------
// Transcoding setup
// -------------------------------------------------------------------------
/** Codec session ID for isolated Rust codec state (unique per leg). */
private codecSessionId = `webrtc-${Date.now()}-${Math.random().toString(36).slice(2, 8)}`;
/**
* Set up transcoders for bridging between WebRTC and SIP codecs.
* Called by the Call when the remote media endpoint is known.
* Creates an isolated Rust codec session so concurrent calls don't
* corrupt each other's stateful codec state (Opus/G.722 ADPCM).
*/
async setupTranscoders(sipPT: number): Promise<void> {
const webrtcPT = this.codec ?? OPUS_PT;
// Create isolated codec session for this leg.
await createSession(this.codecSessionId);
this.toSipTranscoder = createTranscoder(webrtcPT, sipPT, this.codecSessionId, 'to_sip');
this.fromSipTranscoder = createTranscoder(sipPT, webrtcPT, this.codecSessionId, 'to_browser');
const mode = this.toSipTranscoder ? `transcoding PT ${webrtcPT}<->${sipPT}` : `pass-through PT ${webrtcPT}`;
this.config.log(`[webrtc-leg:${this.id}] ${mode} (session: ${this.codecSessionId})`);
}
// -------------------------------------------------------------------------
// RTP forwarding
// -------------------------------------------------------------------------
/** Forward RTP from SIP side to browser via WebRTC. */
private forwardToBrowser(data: Buffer, count: number): void {
const sender = this.pc?.getSenders()[0];
if (!sender) return;
if (this.fromSipTranscoder && data.length > 12) {
const payload = Buffer.from(data.subarray(12));
// Stop announcement if still playing — provider audio takes over.
if (this.announcementCancel) {
this.announcementCancel();
this.announcementCancel = null;
}
// Capture seq/ts BEFORE async transcode to preserve ordering.
const toPT = this.fromSipTranscoder.toPT;
const seq = this.fromSipCounters.seq++;
const ts = this.fromSipCounters.ts;
this.fromSipCounters.ts += rtpClockIncrement(toPT);
const result = this.fromSipTranscoder.payload(payload);
const sendTranscoded = (transcoded: Buffer) => {
if (transcoded.length === 0) return; // transcoding failed
try {
const hdr = buildRtpHeader(toPT, seq, ts, this.fromSipSsrc, false);
const out = Buffer.concat([hdr, transcoded]);
const r = sender.sendRtp(out);
if (r instanceof Promise) r.catch(() => {});
} catch { /* ignore */ }
};
if (result instanceof Promise) result.then(sendTranscoded).catch(() => {});
else sendTranscoded(result);
} else if (!this.fromSipTranscoder) {
// No transcoder — either same codec or not set up yet.
// Only forward if we don't expect transcoding.
if (this.codec === null) {
try { sender.sendRtp(data); } catch { /* ignore */ }
}
}
}
/** Forward RTP from browser to SIP side. */
private forwardToSip(rtp: any, count: number): void {
if (!this.remoteMedia) return;
if (this.toSipTranscoder) {
const payload: Buffer = rtp.payload;
if (!payload || payload.length === 0) return;
// Capture seq/ts BEFORE async transcode to preserve ordering.
const toPT = this.toSipTranscoder.toPT;
const seq = this.toSipSeq++;
const ts = this.toSipTs;
this.toSipTs += rtpClockIncrement(toPT);
const result = this.toSipTranscoder.payload(payload);
const sendTranscoded = (transcoded: Buffer) => {
if (transcoded.length === 0) return; // transcoding failed
const hdr = buildRtpHeader(toPT, seq, ts, this.toSipSsrc, false);
const out = Buffer.concat([hdr, transcoded]);
if (this.onSendToProvider) {
this.onSendToProvider(out, this.remoteMedia!);
} else {
this.rtpSock.send(out, this.remoteMedia!.port, this.remoteMedia!.address);
}
this.pktSent++;
};
if (result instanceof Promise) result.then(sendTranscoded).catch(() => {});
else sendTranscoded(result);
} else if (this.codec === null) {
// Same codec (no transcoding needed) — pass through.
const raw = rtp.serialize();
if (this.onSendToProvider) {
this.onSendToProvider(raw, this.remoteMedia);
} else {
this.rtpSock.send(raw, this.remoteMedia.port, this.remoteMedia.address);
}
this.pktSent++;
}
// If codec is set but transcoder is null, drop the packet — transcoder not ready yet.
// This prevents raw Opus from being sent to a G.722 endpoint.
}
/**
* Send RTP to the browser via WebRTC (used by Call hub for fan-out).
* This transcodes and sends through the PeerConnection, NOT to a UDP address.
*/
sendRtp(data: Buffer): void {
this.forwardToBrowser(data, this.pktSent);
this.pktSent++;
}
/**
* Send a pre-encoded RTP packet directly to the browser via PeerConnection.
* Used for announcements — the packet must already be in the correct codec (Opus).
*/
sendDirectToBrowser(pkt: Buffer): void {
const sender = this.pc?.getSenders()[0];
if (!sender) return;
try {
const r = sender.sendRtp(pkt);
if (r instanceof Promise) r.catch(() => {});
} catch { /* ignore */ }
}
/** No-op: WebRTC legs don't process SIP messages. */
handleSipMessage(_msg: SipMessage, _rinfo: IEndpoint): void {
// WebRTC legs don't handle SIP messages.
}
// -------------------------------------------------------------------------
// Lifecycle
// -------------------------------------------------------------------------
teardown(): void {
this.state = 'terminated';
try { this.pc?.close(); } catch { /* ignore */ }
this.pc = null;
// Destroy the isolated Rust codec session for this leg.
destroySession(this.codecSessionId).catch(() => {});
// Note: RTP socket is NOT closed here — the RtpPortPool manages that.
}
getStatus(): ILegStatus {
return {
id: this.id,
type: this.type,
state: this.state,
remoteMedia: this.remoteMedia,
rtpPort: this.rtpPort,
pktSent: this.pktSent,
pktReceived: this.pktReceived,
codec: codecDisplayName(this.codec),
transcoding: this.toSipTranscoder !== null || this.fromSipTranscoder !== null,
};
}
// -------------------------------------------------------------------------
// Helpers
// -------------------------------------------------------------------------
private wsSend(data: unknown): void {
try {
if (this.ws.readyState === WebSocket.OPEN) {
this.ws.send(JSON.stringify(data));
}
} catch { /* ignore */ }
}
}

View File

@@ -1,40 +0,0 @@
/**
* Audio codec translation layer for bridging between WebRTC and SIP.
*
* All actual codec work (Opus, G.722, PCMU, PCMA) is done in Rust via
* the smartrust bridge. This module provides the RTP-level transcoding
* interface used by the webrtcbridge.
*/
import { Buffer } from 'node:buffer';
import { transcode, isCodecReady } from './opusbridge.ts';
/** Opus dynamic payload type (standard WebRTC assignment). */
export const OPUS_PT = 111;
export interface IRtpTranscoder {
/** Transcode an RTP payload. Always async (Rust IPC). */
payload(data: Buffer): Promise<Buffer>;
fromPT: number;
toPT: number;
}
/**
* Create a transcoder that converts RTP payloads between two codecs.
* Returns null if the codecs are the same or the Rust bridge isn't ready.
*
* @param sessionId - optional Rust codec session for isolated state per call
*/
export function createTranscoder(fromPT: number, toPT: number, sessionId?: string, direction?: string): IRtpTranscoder | null {
if (fromPT === toPT) return null;
if (!isCodecReady()) return null;
return {
fromPT,
toPT,
async payload(data: Buffer): Promise<Buffer> {
const result = await transcode(data, fromPT, toPT, sessionId, direction);
return result || Buffer.alloc(0); // return empty on failure — never pass raw codec bytes
},
};
}

View File

@@ -8,7 +8,15 @@
import fs from 'node:fs';
import path from 'node:path';
import type { IEndpoint } from './sip/index.ts';
// ---------------------------------------------------------------------------
// Shared types (previously in ts/sip/types.ts, now inlined)
// ---------------------------------------------------------------------------
export interface IEndpoint {
address: string;
port: number;
}
// ---------------------------------------------------------------------------
// Config interfaces
@@ -319,175 +327,5 @@ export function loadConfig(): IAppConfig {
return cfg;
}
// ---------------------------------------------------------------------------
// Pattern matching
// ---------------------------------------------------------------------------
/**
* Test a value against a pattern string.
* - undefined/empty pattern: matches everything (wildcard)
* - Prefix: "pattern*" matches values starting with "pattern"
* - Regex: "/pattern/" or "/pattern/i" compiles as RegExp
* - Otherwise: exact match
*/
export function matchesPattern(pattern: string | undefined, value: string): boolean {
if (!pattern) return true;
// Prefix match: "+49*"
if (pattern.endsWith('*')) {
return value.startsWith(pattern.slice(0, -1));
}
// Regex match: "/^\\+49/" or "/pattern/i"
if (pattern.startsWith('/')) {
const lastSlash = pattern.lastIndexOf('/');
if (lastSlash > 0) {
const re = new RegExp(pattern.slice(1, lastSlash), pattern.slice(lastSlash + 1));
return re.test(value);
}
}
// Exact match.
return value === pattern;
}
// ---------------------------------------------------------------------------
// Route resolution
// ---------------------------------------------------------------------------
/** Result of resolving an outbound route. */
export interface IOutboundRouteResult {
provider: IProviderConfig;
transformedNumber: string;
}
/** Result of resolving an inbound route. */
export interface IInboundRouteResult {
/** Device IDs to ring (empty = all devices). */
deviceIds: string[];
ringBrowsers: boolean;
/** If set, route directly to this voicemail box (skip ringing). */
voicemailBox?: string;
/** If set, route to this IVR menu (skip ringing). */
ivrMenuId?: string;
/** Override for no-answer timeout in seconds. */
noAnswerTimeout?: number;
}
/**
* Resolve which provider to use for an outbound call, and transform the number.
*
* @param cfg - app config
* @param dialedNumber - the number being dialed
* @param sourceDeviceId - optional device originating the call
* @param isProviderRegistered - callback to check if a provider is currently registered
*/
export function resolveOutboundRoute(
cfg: IAppConfig,
dialedNumber: string,
sourceDeviceId?: string,
isProviderRegistered?: (providerId: string) => boolean,
): IOutboundRouteResult | null {
const routes = cfg.routing.routes
.filter((r) => r.enabled && r.match.direction === 'outbound')
.sort((a, b) => b.priority - a.priority);
for (const route of routes) {
const m = route.match;
if (!matchesPattern(m.numberPattern, dialedNumber)) continue;
if (m.sourceDevice && m.sourceDevice !== sourceDeviceId) continue;
// Find a registered provider (primary + failovers).
const candidates = [route.action.provider, ...(route.action.failoverProviders || [])].filter(Boolean) as string[];
for (const pid of candidates) {
const provider = getProvider(cfg, pid);
if (!provider) continue;
if (isProviderRegistered && !isProviderRegistered(pid)) continue;
// Apply number transformation.
let num = dialedNumber;
if (route.action.stripPrefix && num.startsWith(route.action.stripPrefix)) {
num = num.slice(route.action.stripPrefix.length);
}
if (route.action.prependPrefix) {
num = route.action.prependPrefix + num;
}
return { provider, transformedNumber: num };
}
// Route matched but no provider is available — continue to next route.
}
// Fallback: first available provider.
const fallback = cfg.providers[0];
return fallback ? { provider: fallback, transformedNumber: dialedNumber } : null;
}
/**
* Resolve which devices/browsers to ring for an inbound call.
*
* @param cfg - app config
* @param providerId - the provider the call is coming from
* @param calledNumber - the DID / called number (from Request-URI)
* @param callerNumber - the caller ID (from From header)
*/
export function resolveInboundRoute(
cfg: IAppConfig,
providerId: string,
calledNumber: string,
callerNumber: string,
): IInboundRouteResult {
const routes = cfg.routing.routes
.filter((r) => r.enabled && r.match.direction === 'inbound')
.sort((a, b) => b.priority - a.priority);
for (const route of routes) {
const m = route.match;
if (m.sourceProvider && m.sourceProvider !== providerId) continue;
if (!matchesPattern(m.numberPattern, calledNumber)) continue;
if (!matchesPattern(m.callerPattern, callerNumber)) continue;
return {
deviceIds: route.action.targets || [],
ringBrowsers: route.action.ringBrowsers ?? false,
voicemailBox: route.action.voicemailBox,
ivrMenuId: route.action.ivrMenuId,
noAnswerTimeout: route.action.noAnswerTimeout,
};
}
// Fallback: ring all devices + browsers.
return { deviceIds: [], ringBrowsers: true };
}
// ---------------------------------------------------------------------------
// Lookup helpers
// ---------------------------------------------------------------------------
export function getProvider(cfg: IAppConfig, id: string): IProviderConfig | null {
return cfg.providers.find((p) => p.id === id) ?? null;
}
export function getDevice(cfg: IAppConfig, id: string): IDeviceConfig | null {
return cfg.devices.find((d) => d.id === id) ?? null;
}
/**
* @deprecated Use resolveOutboundRoute() instead. Kept for backward compat.
*/
export function getProviderForOutbound(cfg: IAppConfig): IProviderConfig | null {
const result = resolveOutboundRoute(cfg, '');
return result?.provider ?? null;
}
/**
* @deprecated Use resolveInboundRoute() instead. Kept for backward compat.
*/
export function getDevicesForInbound(cfg: IAppConfig, providerId: string): IDeviceConfig[] {
const result = resolveInboundRoute(cfg, providerId, '', '');
if (!result.deviceIds.length) return cfg.devices;
return result.deviceIds.map((id) => getDevice(cfg, id)).filter(Boolean) as IDeviceConfig[];
}
// Route resolution, pattern matching, and provider/device lookup
// are now handled entirely by the Rust proxy-engine.

View File

@@ -11,10 +11,13 @@ import path from 'node:path';
import http from 'node:http';
import https from 'node:https';
import { WebSocketServer, WebSocket } from 'ws';
import type { CallManager } from './call/index.ts';
import { handleWebRtcSignaling } from './webrtcbridge.ts';
import type { VoiceboxManager } from './voicebox.ts';
// CallManager was previously used for WebRTC call handling. Now replaced by Rust proxy-engine.
// Kept as `any` type for backward compat with the function signature until full WebRTC port.
type CallManager = any;
const CONFIG_PATH = path.join(process.cwd(), '.nogit', 'config.json');
// ---------------------------------------------------------------------------
@@ -125,14 +128,19 @@ async function handleRequest(
}
}
// API: add leg to call.
// API: add a SIP device to a call (mid-call INVITE to desk phone).
if (url.pathname.startsWith('/api/call/') && url.pathname.endsWith('/addleg') && method === 'POST') {
try {
const callId = url.pathname.split('/')[3];
const body = await readJsonBody(req);
if (!body?.deviceId) return sendJson(res, { ok: false, error: 'missing deviceId' }, 400);
const ok = callManager?.addDeviceToCall(callId, body.deviceId) ?? false;
return sendJson(res, { ok });
const { addDeviceLeg } = await import('./proxybridge.ts');
const legId = await addDeviceLeg(callId, body.deviceId);
if (legId) {
return sendJson(res, { ok: true, legId });
} else {
return sendJson(res, { ok: false, error: 'device not registered or call not found' }, 404);
}
} catch (e: any) {
return sendJson(res, { ok: false, error: e.message }, 400);
}
@@ -144,8 +152,9 @@ async function handleRequest(
const callId = url.pathname.split('/')[3];
const body = await readJsonBody(req);
if (!body?.number) return sendJson(res, { ok: false, error: 'missing number' }, 400);
const ok = callManager?.addExternalToCall(callId, body.number, body.providerId) ?? false;
return sendJson(res, { ok });
const { addLeg: addLegFn } = await import('./proxybridge.ts');
const legId = await addLegFn(callId, body.number, body.providerId);
return sendJson(res, { ok: !!legId, legId });
} catch (e: any) {
return sendJson(res, { ok: false, error: e.message }, 400);
}
@@ -157,22 +166,22 @@ async function handleRequest(
const callId = url.pathname.split('/')[3];
const body = await readJsonBody(req);
if (!body?.legId) return sendJson(res, { ok: false, error: 'missing legId' }, 400);
const ok = callManager?.removeLegFromCall(callId, body.legId) ?? false;
const { removeLeg: removeLegFn } = await import('./proxybridge.ts');
const ok = await removeLegFn(callId, body.legId);
return sendJson(res, { ok });
} catch (e: any) {
return sendJson(res, { ok: false, error: e.message }, 400);
}
}
// API: transfer leg.
// API: transfer leg (not yet implemented).
if (url.pathname === '/api/transfer' && method === 'POST') {
try {
const body = await readJsonBody(req);
if (!body?.sourceCallId || !body?.legId || !body?.targetCallId) {
return sendJson(res, { ok: false, error: 'missing sourceCallId, legId, or targetCallId' }, 400);
}
const ok = callManager?.transferLeg(body.sourceCallId, body.legId, body.targetCallId) ?? false;
return sendJson(res, { ok });
return sendJson(res, { ok: false, error: 'not yet implemented' }, 501);
} catch (e: any) {
return sendJson(res, { ok: false, error: e.message }, 400);
}
@@ -337,6 +346,12 @@ export function initWebUi(
onConfigSaved?: () => void,
callManager?: CallManager,
voiceboxManager?: VoiceboxManager,
/** WebRTC signaling handlers — forwarded to Rust proxy-engine. */
onWebRtcOffer?: (sessionId: string, sdp: string, ws: WebSocket) => Promise<void>,
onWebRtcIce?: (sessionId: string, candidate: any) => Promise<void>,
onWebRtcClose?: (sessionId: string) => Promise<void>,
/** Called when browser sends webrtc-accept (callId + sessionId linking). */
onWebRtcAccept?: (callId: string, sessionId: string) => void,
): void {
const WEB_PORT = 3060;
@@ -372,17 +387,26 @@ export function initWebUi(
socket.on('message', (raw) => {
try {
const msg = JSON.parse(raw.toString());
if (msg.type === 'webrtc-accept' && msg.callId) {
log(`[webrtc] browser accepted call ${msg.callId} session=${msg.sessionId || 'none'}`);
const ok = callManager?.acceptBrowserCall(msg.callId, msg.sessionId) ?? false;
log(`[webrtc] acceptBrowserCall result: ${ok}`);
} else if (msg.type === 'webrtc-offer' && msg.sessionId) {
callManager?.handleWebRtcOffer(msg.sessionId, msg.sdp, socket as any).catch((e: any) =>
if (msg.type === 'webrtc-offer' && msg.sessionId) {
// Forward to Rust proxy-engine for WebRTC handling.
if (onWebRtcOffer) {
log(`[webrtc-ws] offer msg keys: ${Object.keys(msg).join(',')}, sdp type: ${typeof msg.sdp}, sdp len: ${msg.sdp?.length || 0}`);
onWebRtcOffer(msg.sessionId, msg.sdp, socket as any).catch((e: any) =>
log(`[webrtc] offer error: ${e.message}`));
}
} else if (msg.type === 'webrtc-ice' && msg.sessionId) {
callManager?.handleWebRtcIce(msg.sessionId, msg.candidate).catch(() => {});
if (onWebRtcIce) {
onWebRtcIce(msg.sessionId, msg.candidate).catch(() => {});
}
} else if (msg.type === 'webrtc-hangup' && msg.sessionId) {
callManager?.handleWebRtcHangup(msg.sessionId);
if (onWebRtcClose) {
onWebRtcClose(msg.sessionId).catch(() => {});
}
} else if (msg.type === 'webrtc-accept' && msg.callId) {
log(`[webrtc] accept: call=${msg.callId} session=${msg.sessionId || 'none'}`);
if (onWebRtcAccept && msg.sessionId) {
onWebRtcAccept(msg.callId, msg.sessionId);
}
} else if (msg.type?.startsWith('webrtc-')) {
msg._remoteIp = remoteIp;
handleWebRtcSignaling(socket as any, msg);

209
ts/ivr.ts
View File

@@ -1,209 +0,0 @@
/**
* IVR engine — state machine that navigates callers through menus
* based on DTMF digit input.
*
* The IvrEngine is instantiated per-call and drives a SystemLeg:
* - Plays menu prompts via the SystemLeg's prompt playback
* - Receives DTMF digits and resolves them to actions
* - Fires an onAction callback for the CallManager to execute
* (route to extension, voicemail, transfer, etc.)
*/
import type { IIvrConfig, IIvrMenu, TIvrAction } from './config.ts';
import type { SystemLeg } from './call/system-leg.ts';
// ---------------------------------------------------------------------------
// IVR Engine
// ---------------------------------------------------------------------------
export class IvrEngine {
private config: IIvrConfig;
private systemLeg: SystemLeg;
private onAction: (action: TIvrAction) => void;
private log: (msg: string) => void;
/** The currently active menu. */
private currentMenu: IIvrMenu | null = null;
/** How many times the current menu has been replayed (for retry limit). */
private retryCount = 0;
/** Timer for digit input timeout. */
private digitTimeout: ReturnType<typeof setTimeout> | null = null;
/** Whether the engine is waiting for a digit (prompt finished playing). */
private waitingForDigit = false;
/** Whether the engine has been destroyed. */
private destroyed = false;
constructor(
config: IIvrConfig,
systemLeg: SystemLeg,
onAction: (action: TIvrAction) => void,
log: (msg: string) => void,
) {
this.config = config;
this.systemLeg = systemLeg;
this.onAction = onAction;
this.log = log;
}
// -------------------------------------------------------------------------
// Public API
// -------------------------------------------------------------------------
/**
* Start the IVR — navigates to the entry menu and plays its prompt.
*/
start(): void {
const entryMenu = this.getMenu(this.config.entryMenuId);
if (!entryMenu) {
this.log(`[ivr] entry menu "${this.config.entryMenuId}" not found — hanging up`);
this.onAction({ type: 'hangup' });
return;
}
this.navigateToMenu(entryMenu);
}
/**
* Handle a DTMF digit from the caller.
*/
handleDigit(digit: string): void {
if (this.destroyed || !this.currentMenu) return;
// Clear the timeout — caller pressed something.
this.clearDigitTimeout();
// Cancel any playing prompt (caller interrupted it).
this.systemLeg.cancelPrompt();
this.waitingForDigit = false;
this.log(`[ivr] digit '${digit}' in menu "${this.currentMenu.id}"`);
// Look up the digit in the current menu.
const entry = this.currentMenu.entries.find((e) => e.digit === digit);
if (entry) {
this.executeAction(entry.action);
} else {
this.log(`[ivr] invalid digit '${digit}' in menu "${this.currentMenu.id}"`);
this.executeAction(this.currentMenu.invalidAction);
}
}
/**
* Clean up timers and state.
*/
destroy(): void {
this.destroyed = true;
this.clearDigitTimeout();
this.currentMenu = null;
}
// -------------------------------------------------------------------------
// Internal
// -------------------------------------------------------------------------
/** Navigate to a menu: play its prompt, then wait for digit. */
private navigateToMenu(menu: IIvrMenu): void {
if (this.destroyed) return;
this.currentMenu = menu;
this.waitingForDigit = false;
this.clearDigitTimeout();
const promptId = `ivr-menu-${menu.id}`;
this.log(`[ivr] playing menu "${menu.id}" prompt`);
this.systemLeg.playPrompt(promptId, () => {
if (this.destroyed) return;
// Prompt finished — start digit timeout.
this.waitingForDigit = true;
this.startDigitTimeout();
});
}
/** Start the timeout timer for digit input. */
private startDigitTimeout(): void {
const timeoutSec = this.currentMenu?.timeoutSec ?? 5;
this.digitTimeout = setTimeout(() => {
if (this.destroyed || !this.currentMenu) return;
this.log(`[ivr] digit timeout in menu "${this.currentMenu.id}"`);
this.handleTimeout();
}, timeoutSec * 1000);
}
/** Handle timeout (no digit pressed). */
private handleTimeout(): void {
if (!this.currentMenu) return;
this.retryCount++;
const maxRetries = this.currentMenu.maxRetries ?? 3;
if (this.retryCount >= maxRetries) {
this.log(`[ivr] max retries (${maxRetries}) reached in menu "${this.currentMenu.id}"`);
this.executeAction(this.currentMenu.timeoutAction);
} else {
this.log(`[ivr] retry ${this.retryCount}/${maxRetries} in menu "${this.currentMenu.id}"`);
// Replay the current menu.
this.navigateToMenu(this.currentMenu);
}
}
/** Execute an IVR action. */
private executeAction(action: TIvrAction): void {
if (this.destroyed) return;
switch (action.type) {
case 'submenu': {
const submenu = this.getMenu(action.menuId);
if (submenu) {
this.retryCount = 0;
this.navigateToMenu(submenu);
} else {
this.log(`[ivr] submenu "${action.menuId}" not found — hanging up`);
this.onAction({ type: 'hangup' });
}
break;
}
case 'repeat': {
if (this.currentMenu) {
this.navigateToMenu(this.currentMenu);
}
break;
}
case 'play-message': {
// Play a message prompt, then return to the current menu.
this.systemLeg.playPrompt(action.promptId, () => {
if (this.destroyed || !this.currentMenu) return;
this.navigateToMenu(this.currentMenu);
});
break;
}
default:
// All other actions (route-extension, route-voicemail, transfer, hangup)
// are handled by the CallManager via the onAction callback.
this.log(`[ivr] action: ${action.type}`);
this.onAction(action);
break;
}
}
/** Look up a menu by ID. */
private getMenu(menuId: string): IIvrMenu | null {
return this.config.menus.find((m) => m.id === menuId) ?? null;
}
/** Clear the digit timeout timer. */
private clearDigitTimeout(): void {
if (this.digitTimeout) {
clearTimeout(this.digitTimeout);
this.digitTimeout = null;
}
}
}

View File

@@ -1,199 +0,0 @@
/**
* Audio transcoding bridge — uses smartrust to communicate with the Rust
* opus-codec binary, which handles Opus ↔ G.722 ↔ PCMU/PCMA transcoding.
*
* All codec conversion happens in Rust (libopus + SpanDSP G.722 port).
* The TypeScript side just passes raw payloads back and forth.
*/
import path from 'node:path';
import { RustBridge } from '@push.rocks/smartrust';
// ---------------------------------------------------------------------------
// Command type map for smartrust
// ---------------------------------------------------------------------------
type TCodecCommands = {
init: {
params: Record<string, never>;
result: Record<string, never>;
};
create_session: {
params: { session_id: string };
result: Record<string, never>;
};
destroy_session: {
params: { session_id: string };
result: Record<string, never>;
};
transcode: {
params: { data_b64: string; from_pt: number; to_pt: number; session_id?: string; direction?: string };
result: { data_b64: string };
};
encode_pcm: {
params: { data_b64: string; sample_rate: number; to_pt: number; session_id?: string };
result: { data_b64: string };
};
};
// ---------------------------------------------------------------------------
// Bridge singleton
// ---------------------------------------------------------------------------
let bridge: RustBridge<TCodecCommands> | null = null;
let initialized = false;
function buildLocalPaths(): string[] {
const root = process.cwd();
return [
path.join(root, 'dist_rust', 'opus-codec'),
path.join(root, 'rust', 'target', 'release', 'opus-codec'),
path.join(root, 'rust', 'target', 'debug', 'opus-codec'),
];
}
let logFn: ((msg: string) => void) | undefined;
/**
* Initialize the audio transcoding bridge. Spawns the Rust binary.
*/
export async function initCodecBridge(log?: (msg: string) => void): Promise<boolean> {
if (initialized && bridge) return true;
logFn = log;
try {
bridge = new RustBridge<TCodecCommands>({
binaryName: 'opus-codec',
localPaths: buildLocalPaths(),
});
const spawned = await bridge.spawn();
if (!spawned) {
log?.('[codec] failed to spawn opus-codec binary');
bridge = null;
return false;
}
// Auto-restart: reset state when the Rust process exits so the next
// transcode attempt triggers re-initialization instead of silent failure.
bridge.on('exit', () => {
logFn?.('[codec] Rust audio transcoder process exited — will re-init on next use');
bridge = null;
initialized = false;
});
await bridge.sendCommand('init', {} as any);
initialized = true;
log?.('[codec] Rust audio transcoder initialized (Opus + G.722 + PCMU/PCMA)');
return true;
} catch (e: any) {
log?.(`[codec] init error: ${e.message}`);
bridge = null;
return false;
}
}
// ---------------------------------------------------------------------------
// Session management — per-call codec isolation
// ---------------------------------------------------------------------------
/**
* Create an isolated codec session. Each session gets its own Opus/G.722
* encoder/decoder state, preventing concurrent calls from corrupting each
* other's stateful codec predictions.
*/
export async function createSession(sessionId: string): Promise<boolean> {
if (!bridge || !initialized) {
// Attempt auto-reinit if bridge died.
const ok = await initCodecBridge(logFn);
if (!ok) return false;
}
try {
await bridge!.sendCommand('create_session', { session_id: sessionId });
return true;
} catch (e: any) {
logFn?.(`[codec] create_session error: ${e?.message || e}`);
return false;
}
}
/**
* Destroy a codec session, freeing its encoder/decoder state.
*/
export async function destroySession(sessionId: string): Promise<void> {
if (!bridge || !initialized) return;
try {
await bridge.sendCommand('destroy_session', { session_id: sessionId });
} catch {
// Best-effort cleanup.
}
}
// ---------------------------------------------------------------------------
// Transcoding
// ---------------------------------------------------------------------------
/**
* Transcode an RTP payload between two codecs.
* All codec work (Opus, G.722, PCMU, PCMA) + resampling happens in Rust.
*
* @param data - raw RTP payload (no header)
* @param fromPT - source payload type (0=PCMU, 8=PCMA, 9=G.722, 111=Opus)
* @param toPT - target payload type
* @param sessionId - optional session for isolated codec state
* @returns transcoded payload, or null on failure
*/
export async function transcode(data: Buffer, fromPT: number, toPT: number, sessionId?: string, direction?: string): Promise<Buffer | null> {
if (!bridge || !initialized) return null;
try {
const params: any = {
data_b64: data.toString('base64'),
from_pt: fromPT,
to_pt: toPT,
};
if (sessionId) params.session_id = sessionId;
if (direction) params.direction = direction;
const result = await bridge.sendCommand('transcode', params);
return Buffer.from(result.data_b64, 'base64');
} catch {
return null;
}
}
/**
* Encode raw 16-bit PCM to a target codec.
* @param pcmData - raw 16-bit LE PCM bytes
* @param sampleRate - input sample rate (e.g. 22050 for Piper TTS)
* @param toPT - target payload type (9=G.722, 111=Opus, 0=PCMU, 8=PCMA)
* @param sessionId - optional session for isolated codec state
*/
export async function encodePcm(pcmData: Buffer, sampleRate: number, toPT: number, sessionId?: string): Promise<Buffer | null> {
if (!bridge || !initialized) return null;
try {
const params: any = {
data_b64: pcmData.toString('base64'),
sample_rate: sampleRate,
to_pt: toPT,
};
if (sessionId) params.session_id = sessionId;
const result = await bridge.sendCommand('encode_pcm', params);
return Buffer.from(result.data_b64, 'base64');
} catch (e: any) {
console.error('[encodePcm] error:', e?.message || e);
return null;
}
}
/** Check if the codec bridge is ready. */
export function isCodecReady(): boolean {
return initialized && bridge !== null;
}
/** Shut down the codec bridge. */
export function shutdownCodecBridge(): void {
if (bridge) {
try { bridge.kill(); } catch { /* ignore */ }
bridge = null;
initialized = false;
}
}

View File

@@ -1,306 +0,0 @@
/**
* Per-provider runtime state and upstream registration.
*
* Each configured provider gets its own ProviderState instance tracking
* registration status, public IP, and the periodic REGISTER cycle.
*/
import { Buffer } from 'node:buffer';
import {
SipMessage,
generateCallId,
generateTag,
generateBranch,
parseDigestChallenge,
computeDigestAuth,
} from './sip/index.ts';
import type { IEndpoint } from './sip/index.ts';
import type { IProviderConfig } from './config.ts';
// ---------------------------------------------------------------------------
// Provider state
// ---------------------------------------------------------------------------
export class ProviderState {
readonly config: IProviderConfig;
publicIp: string | null;
isRegistered = false;
registeredAor: string;
// Registration transaction state.
private regCallId: string;
private regCSeq = 0;
private regFromTag: string;
private regTimer: ReturnType<typeof setInterval> | null = null;
private sendSip: ((buf: Buffer, dest: IEndpoint) => void) | null = null;
private logFn: ((msg: string) => void) | null = null;
private onRegistrationChange: ((provider: ProviderState) => void) | null = null;
constructor(config: IProviderConfig, publicIpSeed: string | null) {
this.config = config;
this.publicIp = publicIpSeed;
this.registeredAor = `sip:${config.username}@${config.domain}`;
this.regCallId = generateCallId();
this.regFromTag = generateTag();
}
private log(msg: string): void {
this.logFn?.(`[provider:${this.config.id}] ${msg}`);
}
// -------------------------------------------------------------------------
// Upstream registration
// -------------------------------------------------------------------------
/**
* Start the periodic REGISTER cycle with this provider.
*/
startRegistration(
lanIp: string,
lanPort: number,
sendSip: (buf: Buffer, dest: IEndpoint) => void,
log: (msg: string) => void,
onRegistrationChange: (provider: ProviderState) => void,
): void {
this.sendSip = sendSip;
this.logFn = log;
this.onRegistrationChange = onRegistrationChange;
// Initial registration.
this.sendRegister(lanIp, lanPort);
// Re-register periodically.
const intervalMs = (this.config.registerIntervalSec * 0.85) * 1000;
this.regTimer = setInterval(() => this.sendRegister(lanIp, lanPort), intervalMs);
}
stopRegistration(): void {
if (this.regTimer) {
clearInterval(this.regTimer);
this.regTimer = null;
}
}
private sendRegister(lanIp: string, lanPort: number): void {
this.regCSeq++;
const pub = this.publicIp || lanIp;
const { config } = this;
const register = SipMessage.createRequest('REGISTER', `sip:${config.domain}`, {
via: { host: pub, port: lanPort },
from: { uri: this.registeredAor, tag: this.regFromTag },
to: { uri: this.registeredAor },
callId: this.regCallId,
cseq: this.regCSeq,
contact: `<sip:${config.username}@${pub}:${lanPort}>`,
maxForwards: 70,
extraHeaders: [
['Expires', String(config.registerIntervalSec)],
['User-Agent', 'SipRouter/1.0'],
['Allow', 'INVITE, ACK, OPTIONS, CANCEL, BYE, SUBSCRIBE, NOTIFY, INFO, REFER, UPDATE'],
],
});
this.log(`REGISTER -> ${config.outboundProxy.address}:${config.outboundProxy.port} (CSeq ${this.regCSeq})`);
this.sendSip!(register.serialize(), config.outboundProxy);
}
/**
* Handle an incoming SIP response that belongs to this provider's registration.
* Returns true if the message was consumed.
*/
handleRegistrationResponse(msg: SipMessage): boolean {
if (!msg.isResponse) return false;
if (msg.callId !== this.regCallId) return false;
if (msg.cseqMethod?.toUpperCase() !== 'REGISTER') return false;
const code = msg.statusCode ?? 0;
this.log(`REGISTER <- ${code}`);
if (code === 200) {
const wasRegistered = this.isRegistered;
this.isRegistered = true;
if (!wasRegistered) {
this.log('registered');
this.onRegistrationChange?.(this);
}
return true;
}
if (code === 401 || code === 407) {
const challengeHeader = code === 401
? msg.getHeader('WWW-Authenticate')
: msg.getHeader('Proxy-Authenticate');
if (!challengeHeader) {
this.log(`${code} but no challenge header`);
return true;
}
const challenge = parseDigestChallenge(challengeHeader);
if (!challenge) {
this.log(`${code} could not parse digest challenge`);
return true;
}
const authValue = computeDigestAuth({
username: this.config.username,
password: this.config.password,
realm: challenge.realm,
nonce: challenge.nonce,
method: 'REGISTER',
uri: `sip:${this.config.domain}`,
algorithm: challenge.algorithm,
opaque: challenge.opaque,
});
// Resend REGISTER with auth.
this.regCSeq++;
const pub = this.publicIp || 'unknown';
// We need lanIp/lanPort but don't have them here — reconstruct from Via.
const via = msg.getHeader('Via') || '';
const viaHost = via.match(/SIP\/2\.0\/UDP\s+([^;:]+)/)?.[1] || pub;
const viaPort = parseInt(via.match(/:(\d+)/)?.[1] || '5070', 10);
const register = SipMessage.createRequest('REGISTER', `sip:${this.config.domain}`, {
via: { host: viaHost, port: viaPort },
from: { uri: this.registeredAor, tag: this.regFromTag },
to: { uri: this.registeredAor },
callId: this.regCallId,
cseq: this.regCSeq,
contact: `<sip:${this.config.username}@${viaHost}:${viaPort}>`,
maxForwards: 70,
extraHeaders: [
[code === 401 ? 'Authorization' : 'Proxy-Authorization', authValue],
['Expires', String(this.config.registerIntervalSec)],
['User-Agent', 'SipRouter/1.0'],
['Allow', 'INVITE, ACK, OPTIONS, CANCEL, BYE, SUBSCRIBE, NOTIFY, INFO, REFER, UPDATE'],
],
});
this.log(`REGISTER -> (with auth, CSeq ${this.regCSeq})`);
this.sendSip!(register.serialize(), this.config.outboundProxy);
return true;
}
if (code >= 400) {
const wasRegistered = this.isRegistered;
this.isRegistered = false;
if (wasRegistered) {
this.log(`registration lost (${code})`);
this.onRegistrationChange?.(this);
}
return true;
}
return true; // consume 1xx etc.
}
/**
* Update public IP from Via received= parameter.
*/
detectPublicIp(via: string): void {
const m = via.match(/received=([\d.]+)/);
if (m && m[1] !== this.publicIp) {
this.log(`publicIp = ${m[1]}`);
this.publicIp = m[1];
}
}
}
// ---------------------------------------------------------------------------
// Provider state management
// ---------------------------------------------------------------------------
let providerStates: Map<string, ProviderState>;
export function initProviderStates(
providers: IProviderConfig[],
publicIpSeed: string | null,
): Map<string, ProviderState> {
providerStates = new Map();
for (const p of providers) {
providerStates.set(p.id, new ProviderState(p, publicIpSeed));
}
return providerStates;
}
export function getProviderState(id: string): ProviderState | null {
return providerStates?.get(id) ?? null;
}
/**
* Sync running provider states with updated config.
* - New providers: create state + start registration.
* - Removed providers: stop registration + delete state.
* - Changed providers: stop old, create new, start registration (preserves detected publicIp).
*/
export function syncProviderStates(
newProviders: IProviderConfig[],
publicIpSeed: string | null,
lanIp: string,
lanPort: number,
sendSip: (buf: Buffer, dest: IEndpoint) => void,
log: (msg: string) => void,
onRegistrationChange: (provider: ProviderState) => void,
): void {
if (!providerStates) return;
const newIds = new Set(newProviders.map(p => p.id));
const oldIds = new Set(providerStates.keys());
// Remove providers no longer in config.
for (const id of oldIds) {
if (!newIds.has(id)) {
const ps = providerStates.get(id)!;
ps.stopRegistration();
providerStates.delete(id);
log(`[provider:${id}] removed`);
}
}
for (const p of newProviders) {
if (!oldIds.has(p.id)) {
// New provider.
const ps = new ProviderState(p, publicIpSeed);
providerStates.set(p.id, ps);
ps.startRegistration(lanIp, lanPort, sendSip, log, onRegistrationChange);
log(`[provider:${p.id}] added — registration started`);
} else {
// Existing provider — check if config changed.
const existing = providerStates.get(p.id)!;
if (JSON.stringify(existing.config) !== JSON.stringify(p)) {
existing.stopRegistration();
const ps = new ProviderState(p, existing.publicIp || publicIpSeed);
providerStates.set(p.id, ps);
ps.startRegistration(lanIp, lanPort, sendSip, log, onRegistrationChange);
log(`[provider:${p.id}] config changed — re-registering`);
}
}
}
}
/**
* Find which provider sent a packet, by matching the source address
* against all providers' outbound proxy addresses.
*/
export function getProviderByUpstreamAddress(address: string, port: number): ProviderState | null {
if (!providerStates) return null;
for (const ps of providerStates.values()) {
if (ps.config.outboundProxy.address === address && ps.config.outboundProxy.port === port) {
return ps;
}
}
return null;
}
/**
* Check whether a response belongs to any provider's registration transaction.
*/
export function handleProviderRegistrationResponse(msg: SipMessage): boolean {
if (!providerStates || !msg.isResponse) return false;
for (const ps of providerStates.values()) {
if (ps.handleRegistrationResponse(msg)) return true;
}
return false;
}

516
ts/proxybridge.ts Normal file
View File

@@ -0,0 +1,516 @@
/**
* Proxy engine bridge — manages the Rust proxy-engine subprocess.
*
* The proxy-engine handles ALL SIP protocol mechanics. TypeScript only:
* - Sends configuration
* - Receives high-level events (incoming_call, call_ended, etc.)
* - Sends high-level commands (hangup, make_call, play_audio)
*
* No raw SIP ever touches TypeScript.
*/
import path from 'node:path';
import { RustBridge } from '@push.rocks/smartrust';
// ---------------------------------------------------------------------------
// Command type map for smartrust
// ---------------------------------------------------------------------------
type TProxyCommands = {
configure: {
params: Record<string, unknown>;
result: { bound: string };
};
hangup: {
params: { call_id: string };
result: Record<string, never>;
};
make_call: {
params: { number: string; device_id?: string; provider_id?: string };
result: { call_id: string };
};
play_audio: {
params: { call_id: string; leg_id?: string; file_path: string; codec?: number };
result: Record<string, never>;
};
start_recording: {
params: { call_id: string; file_path: string; max_duration_ms?: number };
result: Record<string, never>;
};
stop_recording: {
params: { call_id: string };
result: { file_path: string; duration_ms: number };
};
add_device_leg: {
params: { call_id: string; device_id: string };
result: { leg_id: string };
};
transfer_leg: {
params: { source_call_id: string; leg_id: string; target_call_id: string };
result: Record<string, never>;
};
replace_leg: {
params: { call_id: string; old_leg_id: string; number: string; provider_id?: string };
result: { new_leg_id: string };
};
start_interaction: {
params: {
call_id: string;
leg_id: string;
prompt_wav: string;
expected_digits: string;
timeout_ms: number;
};
result: { result: 'digit' | 'timeout' | 'cancelled'; digit?: string };
};
add_tool_leg: {
params: {
call_id: string;
tool_type: 'recording' | 'transcription';
config?: Record<string, unknown>;
};
result: { tool_leg_id: string };
};
remove_tool_leg: {
params: { call_id: string; tool_leg_id: string };
result: Record<string, never>;
};
set_leg_metadata: {
params: { call_id: string; leg_id: string; key: string; value: unknown };
result: Record<string, never>;
};
generate_tts: {
params: { model: string; voices: string; voice: string; text: string; output: string };
result: { output: string };
};
};
// ---------------------------------------------------------------------------
// Event types from Rust
// ---------------------------------------------------------------------------
export interface IIncomingCallEvent {
call_id: string;
from_uri: string;
to_number: string;
provider_id: string;
}
export interface IOutboundCallEvent {
call_id: string;
from_device: string | null;
to_number: string;
}
export interface ICallEndedEvent {
call_id: string;
reason: string;
duration: number;
from_side?: string;
}
export interface IProviderRegisteredEvent {
provider_id: string;
registered: boolean;
public_ip: string | null;
}
export interface IDeviceRegisteredEvent {
device_id: string;
display_name: string;
address: string;
port: number;
aor: string;
expires: number;
}
// ---------------------------------------------------------------------------
// Bridge singleton
// ---------------------------------------------------------------------------
let bridge: RustBridge<TProxyCommands> | null = null;
let initialized = false;
let logFn: ((msg: string) => void) | undefined;
function buildLocalPaths(): string[] {
const root = process.cwd();
return [
path.join(root, 'dist_rust', 'proxy-engine'),
path.join(root, 'rust', 'target', 'release', 'proxy-engine'),
path.join(root, 'rust', 'target', 'debug', 'proxy-engine'),
];
}
/**
* Initialize the proxy engine — spawn the Rust binary.
* Call configure() separately to push config and start SIP.
*/
export async function initProxyEngine(log?: (msg: string) => void): Promise<boolean> {
if (initialized && bridge) return true;
logFn = log;
try {
bridge = new RustBridge<TProxyCommands>({
binaryName: 'proxy-engine',
localPaths: buildLocalPaths(),
});
const spawned = await bridge.spawn();
if (!spawned) {
log?.('[proxy-engine] failed to spawn binary');
bridge = null;
return false;
}
bridge.on('exit', () => {
logFn?.('[proxy-engine] process exited — will need re-init');
bridge = null;
initialized = false;
});
// Forward stderr for debugging.
bridge.on('stderr', (line: string) => {
logFn?.(`[proxy-engine:stderr] ${line}`);
});
initialized = true;
log?.('[proxy-engine] spawned and ready');
return true;
} catch (e: any) {
log?.(`[proxy-engine] init error: ${e.message}`);
bridge = null;
return false;
}
}
/**
* Send the full app config to the proxy engine.
* This binds the SIP socket, starts provider registrations, etc.
*/
export async function configureProxyEngine(config: Record<string, unknown>): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
const result = await bridge.sendCommand('configure', config as any);
logFn?.(`[proxy-engine] configured, SIP bound on ${(result as any)?.bound || '?'}`);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] configure error: ${e.message}`);
return false;
}
}
/**
* Initiate an outbound call via Rust. Returns the call ID or null on failure.
*/
export async function makeCall(number: string, deviceId?: string, providerId?: string): Promise<string | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('make_call', {
number,
device_id: deviceId,
provider_id: providerId,
} as any);
return (result as any)?.call_id || null;
} catch (e: any) {
logFn?.(`[proxy-engine] make_call error: ${e?.message || e}`);
return null;
}
}
/**
* Send a hangup command.
*/
export async function hangupCall(callId: string): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('hangup', { call_id: callId } as any);
return true;
} catch {
return false;
}
}
/**
* Send a WebRTC offer to the proxy engine. Returns the SDP answer.
*/
export async function webrtcOffer(sessionId: string, sdp: string): Promise<{ sdp: string } | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('webrtc_offer', { session_id: sessionId, sdp } as any);
return result as any;
} catch (e: any) {
logFn?.(`[proxy-engine] webrtc_offer error: ${e?.message || e}`);
return null;
}
}
/**
* Forward an ICE candidate to the proxy engine.
*/
export async function webrtcIce(sessionId: string, candidate: any): Promise<void> {
if (!bridge || !initialized) return;
try {
await bridge.sendCommand('webrtc_ice', {
session_id: sessionId,
candidate: candidate?.candidate || candidate,
sdp_mid: candidate?.sdpMid,
sdp_mline_index: candidate?.sdpMLineIndex,
} as any);
} catch { /* ignore */ }
}
/**
* Link a WebRTC session to a SIP call — enables audio bridging.
* The browser's Opus audio will be transcoded and sent to the provider.
*/
export async function webrtcLink(sessionId: string, callId: string, providerMediaAddr: string, providerMediaPort: number, sipPt: number = 9): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('webrtc_link', {
session_id: sessionId,
call_id: callId,
provider_media_addr: providerMediaAddr,
provider_media_port: providerMediaPort,
sip_pt: sipPt,
} as any);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] webrtc_link error: ${e?.message || e}`);
return false;
}
}
/**
* Add an external SIP leg to an existing call (multiparty).
*/
export async function addLeg(callId: string, number: string, providerId?: string): Promise<string | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('add_leg', {
call_id: callId,
number,
provider_id: providerId,
} as any);
return (result as any)?.leg_id || null;
} catch (e: any) {
logFn?.(`[proxy-engine] add_leg error: ${e?.message || e}`);
return null;
}
}
/**
* Remove a leg from a call.
*/
export async function removeLeg(callId: string, legId: string): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('remove_leg', { call_id: callId, leg_id: legId } as any);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] remove_leg error: ${e?.message || e}`);
return false;
}
}
/**
* Close a WebRTC session.
*/
export async function webrtcClose(sessionId: string): Promise<void> {
if (!bridge || !initialized) return;
try {
await bridge.sendCommand('webrtc_close', { session_id: sessionId } as any);
} catch { /* ignore */ }
}
// ---------------------------------------------------------------------------
// Device leg & interaction commands
// ---------------------------------------------------------------------------
/**
* Add a local SIP device to an existing call (mid-call INVITE to desk phone).
*/
export async function addDeviceLeg(callId: string, deviceId: string): Promise<string | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('add_device_leg', {
call_id: callId,
device_id: deviceId,
} as any);
return (result as any)?.leg_id || null;
} catch (e: any) {
logFn?.(`[proxy-engine] add_device_leg error: ${e?.message || e}`);
return null;
}
}
/**
* Transfer a leg from one call to another (leg stays connected, switches mixer).
*/
export async function transferLeg(
sourceCallId: string,
legId: string,
targetCallId: string,
): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('transfer_leg', {
source_call_id: sourceCallId,
leg_id: legId,
target_call_id: targetCallId,
} as any);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] transfer_leg error: ${e?.message || e}`);
return false;
}
}
/**
* Replace a leg: terminate the old leg and dial a new number into the same call.
*/
export async function replaceLeg(
callId: string,
oldLegId: string,
number: string,
providerId?: string,
): Promise<string | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('replace_leg', {
call_id: callId,
old_leg_id: oldLegId,
number,
provider_id: providerId,
} as any);
return (result as any)?.new_leg_id || null;
} catch (e: any) {
logFn?.(`[proxy-engine] replace_leg error: ${e?.message || e}`);
return null;
}
}
/**
* Start an interaction on a specific leg — isolate it, play a prompt, collect DTMF.
* Blocks until the interaction completes (digit pressed, timeout, or cancelled).
*/
export async function startInteraction(
callId: string,
legId: string,
promptWav: string,
expectedDigits: string,
timeoutMs: number,
): Promise<{ result: 'digit' | 'timeout' | 'cancelled'; digit?: string } | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('start_interaction', {
call_id: callId,
leg_id: legId,
prompt_wav: promptWav,
expected_digits: expectedDigits,
timeout_ms: timeoutMs,
} as any);
return result as any;
} catch (e: any) {
logFn?.(`[proxy-engine] start_interaction error: ${e?.message || e}`);
return null;
}
}
/**
* Add a tool leg (recording or transcription) to a call.
* Tool legs receive per-source unmerged audio from all participants.
*/
export async function addToolLeg(
callId: string,
toolType: 'recording' | 'transcription',
config?: Record<string, unknown>,
): Promise<string | null> {
if (!bridge || !initialized) return null;
try {
const result = await bridge.sendCommand('add_tool_leg', {
call_id: callId,
tool_type: toolType,
config,
} as any);
return (result as any)?.tool_leg_id || null;
} catch (e: any) {
logFn?.(`[proxy-engine] add_tool_leg error: ${e?.message || e}`);
return null;
}
}
/**
* Remove a tool leg from a call. Triggers finalization (WAV files, metadata).
*/
export async function removeToolLeg(callId: string, toolLegId: string): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('remove_tool_leg', {
call_id: callId,
tool_leg_id: toolLegId,
} as any);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] remove_tool_leg error: ${e?.message || e}`);
return false;
}
}
/**
* Set a metadata key-value pair on a leg.
*/
export async function setLegMetadata(
callId: string,
legId: string,
key: string,
value: unknown,
): Promise<boolean> {
if (!bridge || !initialized) return false;
try {
await bridge.sendCommand('set_leg_metadata', {
call_id: callId,
leg_id: legId,
key,
value,
} as any);
return true;
} catch (e: any) {
logFn?.(`[proxy-engine] set_leg_metadata error: ${e?.message || e}`);
return false;
}
}
/**
* Subscribe to an event from the proxy engine.
* Event names: incoming_call, outbound_device_call, call_ringing,
* call_answered, call_ended, provider_registered, device_registered,
* dtmf_digit, recording_done, tool_recording_done, tool_transcription_done,
* leg_added, leg_removed, sip_unhandled
*/
export function onProxyEvent(event: string, handler: (data: any) => void): void {
if (!bridge) throw new Error('proxy engine not initialized');
bridge.on(`management:${event}`, handler);
}
/** Check if the proxy engine is ready. */
export function isProxyReady(): boolean {
return initialized && bridge !== null;
}
/** Send an arbitrary command to the proxy engine bridge. */
export async function sendProxyCommand<K extends keyof TProxyCommands>(
method: K,
params: TProxyCommands[K]['params'],
): Promise<TProxyCommands[K]['result']> {
if (!bridge || !initialized) throw new Error('proxy engine not initialized');
return bridge.sendCommand(method as string, params as any) as any;
}
/** Shut down the proxy engine. */
export function shutdownProxyEngine(): void {
if (bridge) {
try { bridge.kill(); } catch { /* ignore */ }
bridge = null;
initialized = false;
}
}

View File

@@ -1,118 +1,27 @@
/**
* Local SIP registrar — accepts REGISTER from devices and browser clients.
* Browser device registration.
*
* Devices point their SIP registration at the proxy instead of the upstream
* provider. The registrar responds with 200 OK and stores the device's
* current contact (source IP:port). Browser softphones register via
* WebSocket signaling.
* SIP device registration is now handled entirely by the Rust proxy-engine.
* This module only handles browser softphone registration via WebSocket.
*/
import { createHash } from 'node:crypto';
import {
SipMessage,
generateTag,
} from './sip/index.ts';
/** Hash a string to a 6-char hex ID. */
export function shortHash(input: string): string {
return createHash('sha256').update(input).digest('hex').slice(0, 6);
}
import type { IEndpoint } from './sip/index.ts';
import type { IDeviceConfig } from './config.ts';
// ---------------------------------------------------------------------------
// Types
// Browser device registration
// ---------------------------------------------------------------------------
export interface IRegisteredDevice {
deviceConfig: IDeviceConfig;
contact: IEndpoint | null;
registeredAt: number;
expiresAt: number;
aor: string;
connected: boolean;
isBrowser: boolean;
}
export interface IDeviceStatusEntry {
id: string;
displayName: string;
contact: IEndpoint | null;
aor: string;
connected: boolean;
isBrowser: boolean;
}
// ---------------------------------------------------------------------------
// State
// ---------------------------------------------------------------------------
const registeredDevices = new Map<string, IRegisteredDevice>();
const browserDevices = new Map<string, IRegisteredDevice>();
let knownDevices: IDeviceConfig[] = [];
let logFn: (msg: string) => void = () => {};
// ---------------------------------------------------------------------------
// Public API
// ---------------------------------------------------------------------------
export function initRegistrar(
devices: IDeviceConfig[],
log: (msg: string) => void,
): void {
knownDevices = devices;
logFn = log;
}
/**
* Process a REGISTER from a SIP device. Returns a 200 OK response to send back,
* or null if this REGISTER should not be handled by the local registrar.
*/
export function handleDeviceRegister(
msg: SipMessage,
rinfo: IEndpoint,
): SipMessage | null {
if (msg.method !== 'REGISTER') return null;
const device = knownDevices.find((d) => d.expectedAddress === rinfo.address);
if (!device) return null;
const from = msg.getHeader('From');
const aor = from ? SipMessage.extractUri(from) || `sip:${device.extension}@${rinfo.address}` : `sip:${device.extension}@${rinfo.address}`;
const MAX_EXPIRES = 300;
const expiresHeader = msg.getHeader('Expires');
const requested = expiresHeader ? parseInt(expiresHeader, 10) : 3600;
const expires = Math.min(requested, MAX_EXPIRES);
const entry: IRegisteredDevice = {
deviceConfig: device,
contact: { address: rinfo.address, port: rinfo.port },
registeredAt: Date.now(),
expiresAt: Date.now() + expires * 1000,
aor,
connected: true,
isBrowser: false,
};
registeredDevices.set(device.id, entry);
logFn(`[registrar] ${device.displayName} (${device.id}) registered from ${rinfo.address}:${rinfo.port} expires=${expires}`);
const contact = msg.getHeader('Contact') || `<sip:${rinfo.address}:${rinfo.port}>`;
const response = SipMessage.createResponse(200, 'OK', msg, {
toTag: generateTag(),
contact,
extraHeaders: [['Expires', String(expires)]],
});
return response;
}
const browserDevices = new Map<string, { deviceId: string; displayName: string; remoteIp: string }>();
/**
* Register a browser softphone as a device.
*/
export function registerBrowserDevice(sessionId: string, userAgent?: string, remoteIp?: string): void {
// Extract a short browser name from the UA string.
let browserName = 'Browser';
if (userAgent) {
if (userAgent.includes('Firefox/')) browserName = 'Firefox';
@@ -121,21 +30,11 @@ export function registerBrowserDevice(sessionId: string, userAgent?: string, rem
else if (userAgent.includes('Safari/') && !userAgent.includes('Chrome/')) browserName = 'Safari';
}
const entry: IRegisteredDevice = {
deviceConfig: {
id: `browser-${shortHash(sessionId)}`,
browserDevices.set(sessionId, {
deviceId: `browser-${shortHash(sessionId)}`,
displayName: browserName,
expectedAddress: remoteIp || '127.0.0.1',
extension: 'webrtc',
},
contact: null,
registeredAt: Date.now(),
expiresAt: Date.now() + 60 * 1000, // 60s — browser must re-register to stay alive
aor: `sip:webrtc@browser`,
connected: true,
isBrowser: true,
};
browserDevices.set(sessionId, entry);
remoteIp: remoteIp || '127.0.0.1',
});
}
/**
@@ -144,96 +43,3 @@ export function registerBrowserDevice(sessionId: string, userAgent?: string, rem
export function unregisterBrowserDevice(sessionId: string): void {
browserDevices.delete(sessionId);
}
/**
* Get a registered device by its config ID.
*/
export function getRegisteredDevice(deviceId: string): IRegisteredDevice | null {
const entry = registeredDevices.get(deviceId);
if (!entry) return null;
if (Date.now() > entry.expiresAt) {
registeredDevices.delete(deviceId);
return null;
}
return entry;
}
/**
* Get a registered device by source IP address.
*/
export function getRegisteredDeviceByAddress(address: string): IRegisteredDevice | null {
for (const entry of registeredDevices.values()) {
if (entry.contact?.address === address && Date.now() <= entry.expiresAt) {
return entry;
}
}
return null;
}
/**
* Check whether an address belongs to a known device (by config expectedAddress).
*/
export function isKnownDeviceAddress(address: string): boolean {
return knownDevices.some((d) => d.expectedAddress === address);
}
/**
* Get all devices for the dashboard.
* - Configured devices always show (connected or not).
* - Browser devices only show while connected.
*/
export function getAllDeviceStatuses(): IDeviceStatusEntry[] {
const now = Date.now();
const result: IDeviceStatusEntry[] = [];
// Configured devices — always show.
for (const dc of knownDevices) {
const reg = registeredDevices.get(dc.id);
const connected = reg ? now <= reg.expiresAt : false;
if (reg && now > reg.expiresAt) {
registeredDevices.delete(dc.id);
}
result.push({
id: dc.id,
displayName: dc.displayName,
contact: connected && reg ? reg.contact : null,
aor: reg?.aor || `sip:${dc.extension}@${dc.expectedAddress}`,
connected,
isBrowser: false,
});
}
// Browser devices — only while connected.
for (const [, entry] of browserDevices) {
const ip = entry.deviceConfig.expectedAddress;
result.push({
id: entry.deviceConfig.id,
displayName: entry.deviceConfig.displayName,
contact: ip && ip !== '127.0.0.1' ? { address: ip, port: 0 } : null,
aor: entry.aor,
connected: true,
isBrowser: true,
});
}
return result;
}
/**
* Get all currently registered (connected) SIP devices.
*/
export function getAllRegisteredDevices(): IRegisteredDevice[] {
const now = Date.now();
const result: IRegisteredDevice[] = [];
for (const [id, entry] of registeredDevices) {
if (now > entry.expiresAt) {
registeredDevices.delete(id);
} else {
result.push(entry);
}
}
for (const [, entry] of browserDevices) {
result.push(entry);
}
return result;
}

View File

@@ -1,280 +0,0 @@
/**
* SipDialog — tracks the state of a SIP dialog (RFC 3261 §12).
*
* A dialog is created by a dialog-establishing request (INVITE, SUBSCRIBE, …)
* and its 1xx/2xx response. It manages local/remote tags, CSeq counters,
* the route set, and provides helpers to build in-dialog requests (ACK, BYE,
* re-INVITE, …).
*
* Usage:
* ```ts
* // Caller (UAC) side — create from the outgoing INVITE we just sent:
* const dialog = SipDialog.fromUacInvite(invite);
*
* // When a 200 OK arrives:
* dialog.processResponse(response200);
*
* // Build ACK for the 2xx:
* const ack = dialog.createAck();
*
* // Later — hang up:
* const bye = dialog.createRequest('BYE');
* ```
*/
import { SipMessage } from './message.ts';
import { generateTag, generateBranch } from './helpers.ts';
import { Buffer } from 'node:buffer';
export type TDialogState = 'early' | 'confirmed' | 'terminated';
export class SipDialog {
callId: string;
localTag: string;
remoteTag: string | null = null;
localUri: string;
remoteUri: string;
localCSeq: number;
remoteCSeq: number = 0;
routeSet: string[] = [];
remoteTarget: string; // Contact URI of the remote party
state: TDialogState = 'early';
// Transport info for sending in-dialog messages.
localHost: string;
localPort: number;
constructor(options: {
callId: string;
localTag: string;
remoteTag?: string;
localUri: string;
remoteUri: string;
localCSeq: number;
remoteTarget: string;
localHost: string;
localPort: number;
routeSet?: string[];
}) {
this.callId = options.callId;
this.localTag = options.localTag;
this.remoteTag = options.remoteTag ?? null;
this.localUri = options.localUri;
this.remoteUri = options.remoteUri;
this.localCSeq = options.localCSeq;
this.remoteTarget = options.remoteTarget;
this.localHost = options.localHost;
this.localPort = options.localPort;
this.routeSet = options.routeSet ?? [];
}
// -------------------------------------------------------------------------
// Factory: create dialog from an outgoing INVITE (UAC side)
// -------------------------------------------------------------------------
/**
* Create a dialog from an INVITE we are sending.
* The dialog enters "early" state; call `processResponse()` when
* provisional or final responses arrive.
*/
static fromUacInvite(invite: SipMessage, localHost: string, localPort: number): SipDialog {
const from = invite.getHeader('From') || '';
const to = invite.getHeader('To') || '';
return new SipDialog({
callId: invite.callId,
localTag: SipMessage.extractTag(from) || generateTag(),
localUri: SipMessage.extractUri(from) || '',
remoteUri: SipMessage.extractUri(to) || '',
localCSeq: parseInt((invite.getHeader('CSeq') || '1').split(/\s+/)[0], 10),
remoteTarget: invite.requestUri || SipMessage.extractUri(to) || '',
localHost: localHost,
localPort: localPort,
});
}
// -------------------------------------------------------------------------
// Factory: create dialog from an incoming INVITE (UAS side)
// -------------------------------------------------------------------------
/**
* Create a dialog from an INVITE we received.
* Typically used when acting as a UAS (e.g. for call-back scenarios).
*/
static fromUasInvite(invite: SipMessage, localTag: string, localHost: string, localPort: number): SipDialog {
const from = invite.getHeader('From') || '';
const to = invite.getHeader('To') || '';
const contact = invite.getHeader('Contact');
return new SipDialog({
callId: invite.callId,
localTag,
remoteTag: SipMessage.extractTag(from) || undefined,
localUri: SipMessage.extractUri(to) || '',
remoteUri: SipMessage.extractUri(from) || '',
localCSeq: 0,
remoteTarget: (contact ? SipMessage.extractUri(contact) : null) || SipMessage.extractUri(from) || '',
localHost,
localPort,
});
}
// -------------------------------------------------------------------------
// Response processing
// -------------------------------------------------------------------------
/**
* Update dialog state from a received response.
* - 1xx with To-tag → early dialog
* - 2xx → confirmed dialog
* - 3xx6xx → terminated
*/
processResponse(response: SipMessage): void {
const to = response.getHeader('To') || '';
const tag = SipMessage.extractTag(to);
const code = response.statusCode ?? 0;
// Always update remoteTag from 2xx (RFC 3261 §12.1.2: tag in 2xx is definitive).
if (tag && (code >= 200 && code < 300)) {
this.remoteTag = tag;
} else if (tag && !this.remoteTag) {
this.remoteTag = tag;
}
// Update remote target from Contact.
const contact = response.getHeader('Contact');
if (contact) {
const uri = SipMessage.extractUri(contact);
if (uri) this.remoteTarget = uri;
}
// Record-Route → route set (in reverse for UAC).
if (this.state === 'early') {
const rr: string[] = [];
for (const [n, v] of response.headers) {
if (n.toLowerCase() === 'record-route') rr.push(v);
}
if (rr.length) this.routeSet = rr.reverse();
}
if (code >= 200 && code < 300) {
this.state = 'confirmed';
} else if (code >= 300) {
this.state = 'terminated';
}
}
// -------------------------------------------------------------------------
// Request building
// -------------------------------------------------------------------------
/**
* Build an in-dialog request (BYE, re-INVITE, INFO, …).
* Automatically increments the local CSeq.
*/
createRequest(method: string, options?: {
body?: string;
contentType?: string;
extraHeaders?: [string, string][];
}): SipMessage {
this.localCSeq++;
const branch = generateBranch();
const headers: [string, string][] = [
['Via', `SIP/2.0/UDP ${this.localHost}:${this.localPort};branch=${branch};rport`],
['From', `<${this.localUri}>;tag=${this.localTag}`],
['To', `<${this.remoteUri}>${this.remoteTag ? `;tag=${this.remoteTag}` : ''}`],
['Call-ID', this.callId],
['CSeq', `${this.localCSeq} ${method}`],
['Max-Forwards', '70'],
];
// Route set → Route headers.
for (const route of this.routeSet) {
headers.push(['Route', route]);
}
headers.push(['Contact', `<sip:${this.localHost}:${this.localPort}>`]);
if (options?.extraHeaders) headers.push(...options.extraHeaders);
const body = options?.body || '';
if (body && options?.contentType) {
headers.push(['Content-Type', options.contentType]);
}
headers.push(['Content-Length', String(Buffer.byteLength(body, 'utf8'))]);
// Determine Request-URI from route set or remote target.
let ruri = this.remoteTarget;
if (this.routeSet.length) {
const topRoute = SipMessage.extractUri(this.routeSet[0]);
if (topRoute && topRoute.includes(';lr')) {
ruri = this.remoteTarget; // loose routing — RURI stays as remote target
} else if (topRoute) {
ruri = topRoute; // strict routing — top route becomes RURI
}
}
return new SipMessage(`${method} ${ruri} SIP/2.0`, headers, body);
}
/**
* Build an ACK for a 2xx response to INVITE (RFC 3261 §13.2.2.4).
* ACK for 2xx is a new transaction, so it gets its own Via/branch.
*/
createAck(): SipMessage {
const branch = generateBranch();
const headers: [string, string][] = [
['Via', `SIP/2.0/UDP ${this.localHost}:${this.localPort};branch=${branch};rport`],
['From', `<${this.localUri}>;tag=${this.localTag}`],
['To', `<${this.remoteUri}>${this.remoteTag ? `;tag=${this.remoteTag}` : ''}`],
['Call-ID', this.callId],
['CSeq', `${this.localCSeq} ACK`],
['Max-Forwards', '70'],
];
for (const route of this.routeSet) {
headers.push(['Route', route]);
}
headers.push(['Content-Length', '0']);
let ruri = this.remoteTarget;
if (this.routeSet.length) {
const topRoute = SipMessage.extractUri(this.routeSet[0]);
if (topRoute && topRoute.includes(';lr')) {
ruri = this.remoteTarget;
} else if (topRoute) {
ruri = topRoute;
}
}
return new SipMessage(`ACK ${ruri} SIP/2.0`, headers, '');
}
/**
* Build a CANCEL for the original INVITE (same branch, CSeq).
* Used before the dialog is confirmed.
*/
createCancel(originalInvite: SipMessage): SipMessage {
const via = originalInvite.getHeader('Via') || '';
const from = originalInvite.getHeader('From') || '';
const to = originalInvite.getHeader('To') || '';
const headers: [string, string][] = [
['Via', via],
['From', from],
['To', to],
['Call-ID', this.callId],
['CSeq', `${this.localCSeq} CANCEL`],
['Max-Forwards', '70'],
['Content-Length', '0'],
];
const ruri = originalInvite.requestUri || this.remoteTarget;
return new SipMessage(`CANCEL ${ruri} SIP/2.0`, headers, '');
}
/** Transition the dialog to terminated state. */
terminate(): void {
this.state = 'terminated';
}
}

View File

@@ -1,241 +0,0 @@
/**
* SIP helper utilities — ID generation and SDP construction.
*/
import { randomBytes, createHash } from 'node:crypto';
// ---------------------------------------------------------------------------
// ID generators
// ---------------------------------------------------------------------------
/** Generate a random SIP Call-ID. */
export function generateCallId(domain?: string): string {
const id = randomBytes(16).toString('hex');
return domain ? `${id}@${domain}` : id;
}
/** Generate a random SIP From/To tag. */
export function generateTag(): string {
return randomBytes(8).toString('hex');
}
/** Generate a RFC 3261 compliant Via branch (starts with z9hG4bK magic cookie). */
export function generateBranch(): string {
return `z9hG4bK-${randomBytes(8).toString('hex')}`;
}
// ---------------------------------------------------------------------------
// Codec registry
// ---------------------------------------------------------------------------
const CODEC_NAMES: Record<number, string> = {
0: 'PCMU/8000',
3: 'GSM/8000',
4: 'G723/8000',
8: 'PCMA/8000',
9: 'G722/8000',
18: 'G729/8000',
101: 'telephone-event/8000',
};
/** Look up the rtpmap name for a static payload type. */
export function codecName(pt: number): string {
return CODEC_NAMES[pt] || `unknown/${pt}`;
}
// ---------------------------------------------------------------------------
// SDP builder
// ---------------------------------------------------------------------------
export interface ISdpOptions {
/** IP address for the c= and o= lines. */
ip: string;
/** Audio port for the m=audio line. */
port: number;
/** RTP payload type numbers (e.g. [9, 0, 8, 101]). */
payloadTypes?: number[];
/** SDP session ID (random if omitted). */
sessionId?: string;
/** Session name for the s= line (defaults to '-'). */
sessionName?: string;
/** Direction attribute (defaults to 'sendrecv'). */
direction?: 'sendrecv' | 'recvonly' | 'sendonly' | 'inactive';
/** Extra a= lines to append (without "a=" prefix). */
attributes?: string[];
}
/**
* Build a minimal SDP body suitable for SIP INVITE offers/answers.
*
* ```ts
* const sdp = buildSdp({
* ip: '192.168.5.66',
* port: 20000,
* payloadTypes: [9, 0, 101],
* });
* ```
*/
export function buildSdp(options: ISdpOptions): string {
const {
ip,
port,
payloadTypes = [9, 0, 8, 101],
sessionId = String(Math.floor(Math.random() * 1e9)),
sessionName = '-',
direction = 'sendrecv',
attributes = [],
} = options;
const lines: string[] = [
'v=0',
`o=- ${sessionId} ${sessionId} IN IP4 ${ip}`,
`s=${sessionName}`,
`c=IN IP4 ${ip}`,
't=0 0',
`m=audio ${port} RTP/AVP ${payloadTypes.join(' ')}`,
];
for (const pt of payloadTypes) {
const name = CODEC_NAMES[pt];
if (name) lines.push(`a=rtpmap:${pt} ${name}`);
if (pt === 101) lines.push(`a=fmtp:101 0-16`);
}
lines.push(`a=${direction}`);
for (const attr of attributes) lines.push(`a=${attr}`);
lines.push(''); // trailing CRLF
return lines.join('\r\n');
}
// ---------------------------------------------------------------------------
// SIP Digest authentication (RFC 2617)
// ---------------------------------------------------------------------------
export interface IDigestChallenge {
realm: string;
nonce: string;
algorithm?: string;
opaque?: string;
qop?: string;
}
/**
* Parse a `Proxy-Authenticate` or `WWW-Authenticate` header value
* into its constituent fields.
*/
export function parseDigestChallenge(header: string): IDigestChallenge | null {
if (!header.toLowerCase().startsWith('digest ')) return null;
const params = header.slice(7);
const get = (key: string): string | undefined => {
const re = new RegExp(`${key}\\s*=\\s*"([^"]*)"`, 'i');
const m = params.match(re);
if (m) return m[1];
// unquoted value
const re2 = new RegExp(`${key}\\s*=\\s*([^,\\s]+)`, 'i');
const m2 = params.match(re2);
return m2 ? m2[1] : undefined;
};
const realm = get('realm');
const nonce = get('nonce');
if (!realm || !nonce) return null;
return { realm, nonce, algorithm: get('algorithm'), opaque: get('opaque'), qop: get('qop') };
}
function md5(s: string): string {
return createHash('md5').update(s).digest('hex');
}
/**
* Compute a SIP Digest `Proxy-Authorization` or `Authorization` header value.
*/
export function computeDigestAuth(options: {
username: string;
password: string;
realm: string;
nonce: string;
method: string;
uri: string;
algorithm?: string;
opaque?: string;
}): string {
const ha1 = md5(`${options.username}:${options.realm}:${options.password}`);
const ha2 = md5(`${options.method}:${options.uri}`);
const response = md5(`${ha1}:${options.nonce}:${ha2}`);
let header = `Digest username="${options.username}", realm="${options.realm}", ` +
`nonce="${options.nonce}", uri="${options.uri}", response="${response}", ` +
`algorithm=${options.algorithm || 'MD5'}`;
if (options.opaque) header += `, opaque="${options.opaque}"`;
return header;
}
/**
* Parse the audio media port and connection address from an SDP body.
* Returns null when no c= + m=audio pair is found.
*/
export function parseSdpEndpoint(sdp: string): { address: string; port: number } | null {
let addr: string | null = null;
let port: number | null = null;
for (const raw of sdp.replace(/\r\n/g, '\n').split('\n')) {
const line = raw.trim();
if (line.startsWith('c=IN IP4 ')) {
addr = line.slice('c=IN IP4 '.length).trim();
} else if (line.startsWith('m=audio ')) {
const parts = line.split(' ');
if (parts.length >= 2) port = parseInt(parts[1], 10);
}
}
return addr && port ? { address: addr, port } : null;
}
// ---------------------------------------------------------------------------
// MWI (Message Waiting Indicator) — RFC 3842
// ---------------------------------------------------------------------------
/**
* Build a SIP NOTIFY request for Message Waiting Indicator.
*
* Sent out-of-dialog to notify a device about voicemail message counts.
* Uses the message-summary event package per RFC 3842.
*/
export interface IMwiOptions {
/** Proxy LAN IP and port (Via / From / Contact). */
proxyHost: string;
proxyPort: number;
/** Target device URI (e.g. "sip:user@192.168.5.100:5060"). */
targetUri: string;
/** Account URI for the voicebox (used in the From header). */
accountUri: string;
/** Number of new (unheard) voice messages. */
newMessages: number;
/** Number of old (heard) voice messages. */
oldMessages: number;
}
/**
* Build the body and headers for an MWI NOTIFY (RFC 3842 message-summary).
*
* Returns the body string and extra headers needed. The caller builds
* the SipMessage via SipMessage.createRequest('NOTIFY', ...).
*/
export function buildMwiBody(newMessages: number, oldMessages: number, accountUri: string): {
body: string;
contentType: string;
extraHeaders: [string, string][];
} {
const hasNew = newMessages > 0;
const body =
`Messages-Waiting: ${hasNew ? 'yes' : 'no'}\r\n` +
`Message-Account: ${accountUri}\r\n` +
`Voice-Message: ${newMessages}/${oldMessages}\r\n`;
return {
body,
contentType: 'application/simple-message-summary',
extraHeaders: [
['Event', 'message-summary'],
['Subscription-State', 'terminated;reason=noresource'],
],
};
}

View File

@@ -1,17 +0,0 @@
export { SipMessage } from './message.ts';
export { SipDialog } from './dialog.ts';
export type { TDialogState } from './dialog.ts';
export { rewriteSipUri, rewriteSdp } from './rewrite.ts';
export {
generateCallId,
generateTag,
generateBranch,
codecName,
buildSdp,
parseSdpEndpoint,
parseDigestChallenge,
computeDigestAuth,
buildMwiBody,
} from './helpers.ts';
export type { ISdpOptions, IDigestChallenge, IMwiOptions } from './helpers.ts';
export type { IEndpoint } from './types.ts';

View File

@@ -1,316 +0,0 @@
/**
* SipMessage — parse, inspect, mutate, and serialize SIP messages.
*
* Provides a fluent (builder-style) API so callers can chain header
* manipulations before serializing:
*
* const buf = SipMessage.parse(raw)!
* .setHeader('Contact', newContact)
* .prependHeader('Record-Route', rr)
* .updateContentLength()
* .serialize();
*/
import { Buffer } from 'node:buffer';
import { generateCallId, generateTag, generateBranch } from './helpers.ts';
const SIP_FIRST_LINE_RE = /^(?:[A-Z]+\s+\S+\s+SIP\/\d\.\d|SIP\/\d\.\d\s+\d+\s+)/;
export class SipMessage {
startLine: string;
headers: [string, string][];
body: string;
constructor(startLine: string, headers: [string, string][], body: string) {
this.startLine = startLine;
this.headers = headers;
this.body = body;
}
// -------------------------------------------------------------------------
// Parsing
// -------------------------------------------------------------------------
static parse(buf: Buffer): SipMessage | null {
if (!buf.length) return null;
if (buf[0] < 0x41 || buf[0] > 0x7a) return null;
let text: string;
try { text = buf.toString('utf8'); } catch { return null; }
let head: string;
let body: string;
let sep = text.indexOf('\r\n\r\n');
if (sep !== -1) {
head = text.slice(0, sep);
body = text.slice(sep + 4);
} else {
sep = text.indexOf('\n\n');
if (sep !== -1) {
head = text.slice(0, sep);
body = text.slice(sep + 2);
} else {
head = text;
body = '';
}
}
const lines = head.replace(/\r\n/g, '\n').split('\n');
if (!lines.length || !lines[0]) return null;
const startLine = lines[0];
if (!SIP_FIRST_LINE_RE.test(startLine)) return null;
const headers: [string, string][] = [];
for (const line of lines.slice(1)) {
if (!line.trim()) continue;
const colon = line.indexOf(':');
if (colon === -1) continue;
headers.push([line.slice(0, colon).trim(), line.slice(colon + 1).trim()]);
}
return new SipMessage(startLine, headers, body);
}
// -------------------------------------------------------------------------
// Serialization
// -------------------------------------------------------------------------
serialize(): Buffer {
const head = [this.startLine, ...this.headers.map(([n, v]) => `${n}: ${v}`)].join('\r\n') + '\r\n\r\n';
return Buffer.concat([Buffer.from(head, 'utf8'), Buffer.from(this.body || '', 'utf8')]);
}
// -------------------------------------------------------------------------
// Inspectors
// -------------------------------------------------------------------------
get isRequest(): boolean {
return !this.startLine.startsWith('SIP/');
}
get isResponse(): boolean {
return this.startLine.startsWith('SIP/');
}
/** Request method (INVITE, REGISTER, ...) or null for responses. */
get method(): string | null {
if (!this.isRequest) return null;
return this.startLine.split(' ')[0];
}
/** Response status code or null for requests. */
get statusCode(): number | null {
if (!this.isResponse) return null;
return parseInt(this.startLine.split(' ')[1], 10);
}
get callId(): string {
return this.getHeader('Call-ID') || 'noid';
}
/** Method from the CSeq header (e.g. "INVITE"). */
get cseqMethod(): string | null {
const cseq = this.getHeader('CSeq');
if (!cseq) return null;
const parts = cseq.trim().split(/\s+/);
return parts.length >= 2 ? parts[1] : null;
}
/** True for INVITE, SUBSCRIBE, REFER, NOTIFY, UPDATE. */
get isDialogEstablishing(): boolean {
return /^(INVITE|SUBSCRIBE|REFER|NOTIFY|UPDATE)\s/.test(this.startLine);
}
/** True when the body carries an SDP payload. */
get hasSdpBody(): boolean {
const ct = (this.getHeader('Content-Type') || '').toLowerCase();
return !!this.body && ct.startsWith('application/sdp');
}
// -------------------------------------------------------------------------
// Header accessors (fluent)
// -------------------------------------------------------------------------
getHeader(name: string): string | null {
const nl = name.toLowerCase();
for (const [n, v] of this.headers) if (n.toLowerCase() === nl) return v;
return null;
}
/** Overwrites the first header with the given name, or appends it. */
setHeader(name: string, value: string): this {
const nl = name.toLowerCase();
for (const h of this.headers) {
if (h[0].toLowerCase() === nl) { h[1] = value; return this; }
}
this.headers.push([name, value]);
return this;
}
/** Inserts a header at the top of the header list. */
prependHeader(name: string, value: string): this {
this.headers.unshift([name, value]);
return this;
}
/** Removes all headers with the given name. */
removeHeader(name: string): this {
const nl = name.toLowerCase();
this.headers = this.headers.filter(([n]) => n.toLowerCase() !== nl);
return this;
}
/** Recalculates Content-Length to match the current body. */
updateContentLength(): this {
const len = Buffer.byteLength(this.body || '', 'utf8');
return this.setHeader('Content-Length', String(len));
}
// -------------------------------------------------------------------------
// Start-line mutation
// -------------------------------------------------------------------------
/** Replaces the Request-URI (second token) of a request start line. */
setRequestUri(uri: string): this {
if (!this.isRequest) return this;
const parts = this.startLine.split(' ');
if (parts.length >= 2) {
parts[1] = uri;
this.startLine = parts.join(' ');
}
return this;
}
/** Returns the Request-URI (second token) of a request start line. */
get requestUri(): string | null {
if (!this.isRequest) return null;
return this.startLine.split(' ')[1] || null;
}
// -------------------------------------------------------------------------
// Factory methods — build new SIP messages from scratch
// -------------------------------------------------------------------------
/**
* Build a new SIP request.
*
* ```ts
* const invite = SipMessage.createRequest('INVITE', 'sip:user@host', {
* from: { uri: 'sip:me@proxy', tag: 'abc' },
* to: { uri: 'sip:user@host' },
* via: { host: '192.168.5.66', port: 5070 },
* contact: '<sip:me@192.168.5.66:5070>',
* });
* ```
*/
static createRequest(method: string, requestUri: string, options: {
via: { host: string; port: number; transport?: string; branch?: string };
from: { uri: string; displayName?: string; tag?: string };
to: { uri: string; displayName?: string; tag?: string };
callId?: string;
cseq?: number;
contact?: string;
maxForwards?: number;
body?: string;
contentType?: string;
extraHeaders?: [string, string][];
}): SipMessage {
const branch = options.via.branch || generateBranch();
const transport = options.via.transport || 'UDP';
const fromTag = options.from.tag || generateTag();
const callId = options.callId || generateCallId();
const cseq = options.cseq ?? 1;
const fromDisplay = options.from.displayName ? `"${options.from.displayName}" ` : '';
const toDisplay = options.to.displayName ? `"${options.to.displayName}" ` : '';
const toTag = options.to.tag ? `;tag=${options.to.tag}` : '';
const headers: [string, string][] = [
['Via', `SIP/2.0/${transport} ${options.via.host}:${options.via.port};branch=${branch};rport`],
['From', `${fromDisplay}<${options.from.uri}>;tag=${fromTag}`],
['To', `${toDisplay}<${options.to.uri}>${toTag}`],
['Call-ID', callId],
['CSeq', `${cseq} ${method}`],
['Max-Forwards', String(options.maxForwards ?? 70)],
];
if (options.contact) {
headers.push(['Contact', options.contact]);
}
if (options.extraHeaders) {
headers.push(...options.extraHeaders);
}
const body = options.body || '';
if (body && options.contentType) {
headers.push(['Content-Type', options.contentType]);
}
headers.push(['Content-Length', String(Buffer.byteLength(body, 'utf8'))]);
return new SipMessage(`${method} ${requestUri} SIP/2.0`, headers, body);
}
/**
* Build a SIP response to an incoming request.
*
* Copies Via, From, To, Call-ID, and CSeq from the original request.
*/
static createResponse(
statusCode: number,
reasonPhrase: string,
request: SipMessage,
options?: {
toTag?: string;
contact?: string;
body?: string;
contentType?: string;
extraHeaders?: [string, string][];
},
): SipMessage {
const headers: [string, string][] = [];
// Copy all Via headers (order matters).
for (const [n, v] of request.headers) {
if (n.toLowerCase() === 'via') headers.push(['Via', v]);
}
// From — copied verbatim.
const from = request.getHeader('From');
if (from) headers.push(['From', from]);
// To — add tag if provided and not already present.
let to = request.getHeader('To') || '';
if (options?.toTag && !to.includes('tag=')) {
to += `;tag=${options.toTag}`;
}
headers.push(['To', to]);
headers.push(['Call-ID', request.callId]);
const cseq = request.getHeader('CSeq');
if (cseq) headers.push(['CSeq', cseq]);
if (options?.contact) headers.push(['Contact', options.contact]);
if (options?.extraHeaders) headers.push(...options.extraHeaders);
const body = options?.body || '';
if (body && options?.contentType) {
headers.push(['Content-Type', options.contentType]);
}
headers.push(['Content-Length', String(Buffer.byteLength(body, 'utf8'))]);
return new SipMessage(`SIP/2.0 ${statusCode} ${reasonPhrase}`, headers, body);
}
/** Extract the tag from a From or To header value. */
static extractTag(headerValue: string): string | null {
const m = headerValue.match(/;tag=([^\s;>]+)/);
return m ? m[1] : null;
}
/** Extract the URI from an addr-spec or name-addr (From/To/Contact). */
static extractUri(headerValue: string): string | null {
const m = headerValue.match(/<([^>]+)>/);
return m ? m[1] : headerValue.trim().split(/[;>]/)[0] || null;
}
}

View File

@@ -1,228 +0,0 @@
# ts/sip — SIP Protocol Library
A zero-dependency SIP (Session Initiation Protocol) library for Deno / Node.
Provides parsing, construction, mutation, and dialog management for SIP
messages, plus helpers for SDP bodies and URI rewriting.
## Modules
| File | Purpose |
|------|---------|
| `message.ts` | `SipMessage` class — parse, inspect, mutate, serialize |
| `dialog.ts` | `SipDialog` class — track dialog state, build in-dialog requests |
| `helpers.ts` | ID generators, codec registry, SDP builder/parser |
| `rewrite.ts` | SIP URI and SDP body rewriting |
| `types.ts` | Shared types (`IEndpoint`) |
| `index.ts` | Barrel re-export |
## Quick Start
```ts
import {
SipMessage,
SipDialog,
buildSdp,
parseSdpEndpoint,
rewriteSipUri,
rewriteSdp,
generateCallId,
generateTag,
generateBranch,
} from './sip/index.ts';
```
## SipMessage
### Parsing
```ts
import { Buffer } from 'node:buffer';
const raw = Buffer.from(
'INVITE sip:user@example.com SIP/2.0\r\n' +
'Via: SIP/2.0/UDP 10.0.0.1:5060;branch=z9hG4bK776\r\n' +
'From: <sip:alice@example.com>;tag=abc\r\n' +
'To: <sip:bob@example.com>\r\n' +
'Call-ID: a84b4c76e66710@10.0.0.1\r\n' +
'CSeq: 1 INVITE\r\n' +
'Content-Length: 0\r\n\r\n'
);
const msg = SipMessage.parse(raw);
// msg.method → "INVITE"
// msg.isRequest → true
// msg.callId → "a84b4c76e66710@10.0.0.1"
// msg.cseqMethod → "INVITE"
// msg.isDialogEstablishing → true
```
### Fluent mutation
All setter methods return `this` for chaining:
```ts
const buf = SipMessage.parse(raw)!
.setHeader('Contact', '<sip:proxy@192.168.1.1:5070>')
.prependHeader('Record-Route', '<sip:192.168.1.1:5070;lr>')
.updateContentLength()
.serialize();
```
### Building requests from scratch
```ts
const invite = SipMessage.createRequest('INVITE', 'sip:+4930123@voip.example.com', {
via: { host: '192.168.5.66', port: 5070 },
from: { uri: 'sip:alice@example.com', displayName: 'Alice' },
to: { uri: 'sip:+4930123@voip.example.com' },
contact: '<sip:192.168.5.66:5070>',
body: sdpBody,
contentType: 'application/sdp',
});
// Call-ID, From tag, Via branch are auto-generated if not provided.
```
### Building responses
```ts
const ok = SipMessage.createResponse(200, 'OK', incomingInvite, {
toTag: generateTag(),
contact: '<sip:192.168.5.66:5070>',
body: answerSdp,
contentType: 'application/sdp',
});
```
### Inspectors
| Property | Type | Description |
|----------|------|-------------|
| `isRequest` | `boolean` | True for requests (INVITE, BYE, ...) |
| `isResponse` | `boolean` | True for responses (SIP/2.0 200 OK, ...) |
| `method` | `string \| null` | Request method or null |
| `statusCode` | `number \| null` | Response status code or null |
| `callId` | `string` | Call-ID header value |
| `cseqMethod` | `string \| null` | Method from CSeq header |
| `requestUri` | `string \| null` | Request-URI (second token of start line) |
| `isDialogEstablishing` | `boolean` | INVITE, SUBSCRIBE, REFER, NOTIFY, UPDATE |
| `hasSdpBody` | `boolean` | Body present with Content-Type: application/sdp |
### Static helpers
```ts
SipMessage.extractTag('<sip:alice@x.com>;tag=abc') // → "abc"
SipMessage.extractUri('"Alice" <sip:alice@x.com>') // → "sip:alice@x.com"
```
## SipDialog
Tracks dialog state per RFC 3261 §12. A dialog is created from a
dialog-establishing request and updated as responses arrive.
### UAC (caller) side
```ts
// 1. Build and send INVITE
const invite = SipMessage.createRequest('INVITE', destUri, { ... });
const dialog = SipDialog.fromUacInvite(invite, '192.168.5.66', 5070);
// 2. Process responses as they arrive
dialog.processResponse(trying100); // state stays 'early'
dialog.processResponse(ringing180); // state stays 'early', remoteTag learned
dialog.processResponse(ok200); // state → 'confirmed'
// 3. ACK the 200
const ack = dialog.createAck();
// 4. In-dialog requests
const bye = dialog.createRequest('BYE');
dialog.terminate();
```
### UAS (callee) side
```ts
const dialog = SipDialog.fromUasInvite(incomingInvite, generateTag(), localHost, localPort);
```
### CANCEL (before answer)
```ts
const cancel = dialog.createCancel(originalInvite);
```
### Dialog states
`'early'``'confirmed'``'terminated'`
## Helpers
### ID generation
```ts
generateCallId() // → "a3f8b2c1d4e5f6a7b8c9d0e1f2a3b4c5"
generateCallId('example.com') // → "a3f8b2c1...@example.com"
generateTag() // → "1a2b3c4d5e6f7a8b"
generateBranch() // → "z9hG4bK-1a2b3c4d5e6f7a8b"
```
### SDP builder
```ts
const sdp = buildSdp({
ip: '192.168.5.66',
port: 20000,
payloadTypes: [9, 0, 8, 101], // G.722, PCMU, PCMA, telephone-event
direction: 'sendrecv',
});
```
### SDP parser
```ts
const ep = parseSdpEndpoint(sdpBody);
// → { address: '10.0.0.1', port: 20000 } or null
```
### Codec names
```ts
codecName(9) // → "G722/8000"
codecName(0) // → "PCMU/8000"
codecName(101) // → "telephone-event/8000"
```
## Rewriting
### SIP URI
Replaces the host:port in all `sip:` / `sips:` URIs found in a header value:
```ts
rewriteSipUri('<sip:user@10.0.0.1:5060>', '203.0.113.1', 5070)
// → '<sip:user@203.0.113.1:5070>'
```
### SDP body
Rewrites the connection address and audio media port, returning the original
endpoint that was replaced:
```ts
const { body, original } = rewriteSdp(sdpBody, '203.0.113.1', 20000);
// original → { address: '10.0.0.1', port: 8000 }
```
## Architecture Notes
This library is intentionally low-level — it operates on individual messages
and dialogs rather than providing a full SIP stack with transport and
transaction layers. This makes it suitable for building:
- **SIP proxies** — parse, rewrite headers/SDP, serialize, forward
- **B2BUA (back-to-back user agents)** — manage two dialogs, bridge media
- **SIP testing tools** — craft and send arbitrary messages
- **Protocol analyzers** — parse and inspect SIP traffic
The library does not manage sockets, timers, or retransmissions — those
concerns belong to the application layer.

View File

@@ -1,54 +0,0 @@
/**
* SIP URI and SDP body rewriting helpers.
*/
import type { IEndpoint } from './types.ts';
const SIP_URI_RE = /(sips?:)([^@>;,\s]+@)?([^>;,\s:]+)(:\d+)?/g;
/**
* Replaces the host:port in every `sip:` / `sips:` URI found in `value`.
*/
export function rewriteSipUri(value: string, host: string, port: number): string {
return value.replace(SIP_URI_RE, (_m, scheme: string, userpart?: string) =>
`${scheme}${userpart || ''}${host}:${port}`);
}
/**
* Rewrites the connection address (`c=`) and audio media port (`m=audio`)
* in an SDP body. Returns the rewritten body together with the original
* endpoint that was replaced (if any).
*/
export function rewriteSdp(
body: string,
ip: string,
port: number,
): { body: string; original: IEndpoint | null } {
let origAddr: string | null = null;
let origPort: number | null = null;
const out = body
.replace(/\r\n/g, '\n')
.split('\n')
.map((line) => {
if (line.startsWith('c=IN IP4 ')) {
origAddr = line.slice('c=IN IP4 '.length).trim();
return `c=IN IP4 ${ip}`;
}
if (line.startsWith('m=audio ')) {
const parts = line.split(' ');
if (parts.length >= 2) {
origPort = parseInt(parts[1], 10);
parts[1] = String(port);
}
return parts.join(' ');
}
return line;
})
.join('\r\n');
return {
body: out,
original: origAddr && origPort ? { address: origAddr, port: origPort } : null,
};
}

View File

@@ -1,8 +0,0 @@
/**
* Shared SIP types.
*/
export interface IEndpoint {
address: string;
port: number;
}

View File

@@ -1,39 +1,22 @@
/**
* SIP proxy — hub model entry point.
* SIP proxy — entry point.
*
* Thin bootstrap that wires together:
* - UDP socket for all SIP signaling
* - CallManager (the hub model core)
* - Provider registration
* - Local device registrar
* - WebRTC signaling
* - Web dashboard
* - Rust codec bridge
* Spawns the Rust proxy-engine which handles ALL SIP protocol mechanics.
* TypeScript is the control plane:
* - Loads config and pushes it to Rust
* - Receives high-level events (incoming calls, registration, etc.)
* - Drives the web dashboard
* - Manages IVR, voicemail, announcements
* - Handles WebRTC browser signaling (forwarded to Rust in Phase 2)
*
* All call/media logic lives in ts/call/.
* No raw SIP ever touches TypeScript.
*/
import dgram from 'node:dgram';
import fs from 'node:fs';
import path from 'node:path';
import { Buffer } from 'node:buffer';
import { SipMessage } from './sip/index.ts';
import type { IEndpoint } from './sip/index.ts';
import { loadConfig, resolveOutboundRoute } from './config.ts';
import type { IAppConfig, IProviderConfig } from './config.ts';
import {
initProviderStates,
syncProviderStates,
getProviderByUpstreamAddress,
handleProviderRegistrationResponse,
} from './providerstate.ts';
import {
initRegistrar,
handleDeviceRegister,
isKnownDeviceAddress,
getAllDeviceStatuses,
} from './registrar.ts';
import { loadConfig } from './config.ts';
import type { IAppConfig } from './config.ts';
import { broadcastWs, initWebUi } from './frontend.ts';
import {
initWebRtcSignaling,
@@ -41,21 +24,36 @@ import {
getAllBrowserDeviceIds,
getBrowserDeviceWs,
} from './webrtcbridge.ts';
import { initCodecBridge } from './opusbridge.ts';
import { initAnnouncement } from './announcement.ts';
import { CallManager } from './call/index.ts';
import { PromptCache } from './call/prompt-cache.ts';
import { VoiceboxManager } from './voicebox.ts';
import {
initProxyEngine,
configureProxyEngine,
onProxyEvent,
hangupCall,
makeCall,
shutdownProxyEngine,
webrtcOffer,
webrtcIce,
webrtcLink,
webrtcClose,
addLeg,
removeLeg,
} from './proxybridge.ts';
import type {
IIncomingCallEvent,
IOutboundCallEvent,
ICallEndedEvent,
IProviderRegisteredEvent,
IDeviceRegisteredEvent,
} from './proxybridge.ts';
// ---------------------------------------------------------------------------
// Config
// ---------------------------------------------------------------------------
let appConfig: IAppConfig = loadConfig();
const { proxy } = appConfig;
const LAN_IP = proxy.lanIp;
const LAN_PORT = proxy.lanPort;
const LOG_PATH = path.join(process.cwd(), 'sip_trace.log');
@@ -77,42 +75,106 @@ function log(msg: string): void {
broadcastWs('log', { message: msg });
}
function logPacket(label: string, data: Buffer): void {
const head = `\n========== ${now()} ${label} (${data.length}b) ==========\n`;
const looksText = data.length > 0 && data[0] >= 0x41 && data[0] <= 0x7a;
const body = looksText
? data.toString('utf8')
: `[${data.length} bytes binary] ${data.toString('hex').slice(0, 80)}`;
fs.appendFileSync(LOG_PATH, head + body + '\n');
// ---------------------------------------------------------------------------
// Shadow state — maintained from Rust events for the dashboard
// ---------------------------------------------------------------------------
interface IProviderStatus {
id: string;
displayName: string;
registered: boolean;
publicIp: string | null;
}
interface IDeviceStatus {
id: string;
displayName: string;
address: string | null;
port: number;
connected: boolean;
isBrowser: boolean;
}
interface IActiveLeg {
id: string;
type: 'sip-device' | 'sip-provider' | 'webrtc' | 'tool';
state: string;
codec: string | null;
rtpPort: number | null;
remoteMedia: string | null;
metadata: Record<string, unknown>;
}
interface IActiveCall {
id: string;
direction: string;
callerNumber: string | null;
calleeNumber: string | null;
providerUsed: string | null;
state: string;
startedAt: number;
legs: Map<string, IActiveLeg>;
}
interface IHistoryLeg {
id: string;
type: string;
metadata: Record<string, unknown>;
}
interface ICallHistoryEntry {
id: string;
direction: string;
callerNumber: string | null;
calleeNumber: string | null;
startedAt: number;
duration: number;
legs: IHistoryLeg[];
}
const providerStatuses = new Map<string, IProviderStatus>();
const deviceStatuses = new Map<string, IDeviceStatus>();
const activeCalls = new Map<string, IActiveCall>();
const callHistory: ICallHistoryEntry[] = [];
const MAX_HISTORY = 100;
// WebRTC session ↔ call linking state.
// Both pieces (session accept + call media info) can arrive in any order.
const webrtcSessionToCall = new Map<string, string>(); // sessionId → callId
const webrtcCallToSession = new Map<string, string>(); // callId → sessionId
const pendingCallMedia = new Map<string, { addr: string; port: number; sipPt: number }>(); // callId → provider media info
// Initialize provider statuses from config (all start as unregistered).
for (const p of appConfig.providers) {
providerStatuses.set(p.id, {
id: p.id,
displayName: p.displayName,
registered: false,
publicIp: null,
});
}
// Initialize device statuses from config.
for (const d of appConfig.devices) {
deviceStatuses.set(d.id, {
id: d.id,
displayName: d.displayName,
address: null,
port: 0,
connected: false,
isBrowser: false,
});
}
// ---------------------------------------------------------------------------
// Initialize subsystems
// ---------------------------------------------------------------------------
const providerStates = initProviderStates(appConfig.providers, proxy.publicIpSeed);
initRegistrar(appConfig.devices, log);
// Initialize voicemail and IVR subsystems.
const promptCache = new PromptCache(log);
const voiceboxManager = new VoiceboxManager(log);
voiceboxManager.init(appConfig.voiceboxes ?? []);
const callManager = new CallManager({
appConfig,
sendSip: (buf, dest) => sock.send(buf, dest.port, dest.address),
log,
broadcastWs,
getProviderState: (id) => providerStates.get(id),
getAllBrowserDeviceIds,
sendToBrowserDevice,
getBrowserDeviceWs,
promptCache,
voiceboxManager,
});
// Initialize WebRTC signaling (browser device registration only).
// WebRTC signaling (browser device registration).
initWebRtcSignaling({ log });
// ---------------------------------------------------------------------------
@@ -120,177 +182,372 @@ initWebRtcSignaling({ log });
// ---------------------------------------------------------------------------
function getStatus() {
const providers: unknown[] = [];
for (const ps of providerStates.values()) {
providers.push({
id: ps.config.id,
displayName: ps.config.displayName,
registered: ps.isRegistered,
publicIp: ps.publicIp,
// Merge SIP devices (from Rust) + browser devices (from TS WebSocket).
const devices = [...deviceStatuses.values()];
for (const bid of getAllBrowserDeviceIds()) {
devices.push({
id: bid,
displayName: 'Browser',
address: null,
port: 0,
connected: true,
isBrowser: true,
});
}
return {
instanceId,
uptime: Math.floor((Date.now() - startTime) / 1000),
lanIp: LAN_IP,
providers,
devices: getAllDeviceStatuses(),
calls: callManager.getStatus(),
callHistory: callManager.getHistory(),
lanIp: appConfig.proxy.lanIp,
providers: [...providerStatuses.values()],
devices,
calls: [...activeCalls.values()].map((c) => ({
...c,
duration: Math.floor((Date.now() - c.startedAt) / 1000),
legs: [...c.legs.values()].map((l) => ({
id: l.id,
type: l.type,
state: l.state,
codec: l.codec,
rtpPort: l.rtpPort,
remoteMedia: l.remoteMedia,
metadata: l.metadata || {},
pktSent: 0,
pktReceived: 0,
transcoding: false,
})),
})),
callHistory,
contacts: appConfig.contacts || [],
voicemailCounts: voiceboxManager.getAllUnheardCounts(),
};
}
// ---------------------------------------------------------------------------
// Main UDP socket
// Start Rust proxy engine
// ---------------------------------------------------------------------------
const sock = dgram.createSocket('udp4');
async function startProxyEngine(): Promise<void> {
const ok = await initProxyEngine(log);
if (!ok) {
log('[FATAL] failed to start proxy engine');
process.exit(1);
}
sock.on('message', (data: Buffer, rinfo: dgram.RemoteInfo) => {
try {
const ps = getProviderByUpstreamAddress(rinfo.address, rinfo.port);
const msg = SipMessage.parse(data);
if (!msg) {
// Non-SIP data — forward raw based on direction.
// Subscribe to events from Rust BEFORE sending configure.
onProxyEvent('provider_registered', (data: IProviderRegisteredEvent) => {
const ps = providerStatuses.get(data.provider_id);
if (ps) {
// From provider, forward to... nowhere useful without a call context.
logPacket(`UP->DN RAW (unparsed) from ${rinfo.address}:${rinfo.port}`, data);
} else {
// From device, forward to default provider.
const dp = resolveOutboundRoute(appConfig, '');
if (dp) sock.send(data, dp.provider.outboundProxy.port, dp.provider.outboundProxy.address);
const wasRegistered = ps.registered;
ps.registered = data.registered;
ps.publicIp = data.public_ip;
if (data.registered && !wasRegistered) {
log(`[provider:${data.provider_id}] registered (publicIp=${data.public_ip})`);
} else if (!data.registered && wasRegistered) {
log(`[provider:${data.provider_id}] registration lost`);
}
return;
}
// 1. Provider registration responses — consumed by providerstate.
if (handleProviderRegistrationResponse(msg)) return;
// 2. Device REGISTER — handled by local registrar.
if (!ps && msg.method === 'REGISTER') {
const response = handleDeviceRegister(msg, { address: rinfo.address, port: rinfo.port });
if (response) {
sock.send(response.serialize(), rinfo.port, rinfo.address);
return;
}
}
// 3. Route to existing call by SIP Call-ID.
if (callManager.routeSipMessage(msg, rinfo)) {
return;
}
// 4. New inbound call from provider.
if (ps && msg.isRequest && msg.method === 'INVITE') {
logPacket(`[new inbound] INVITE from ${rinfo.address}:${rinfo.port}`, data);
// Detect public IP from Via.
const via = msg.getHeader('Via');
if (via) ps.detectPublicIp(via);
callManager.createInboundCall(ps, msg, { address: rinfo.address, port: rinfo.port });
return;
}
// 5. New outbound call from device (passthrough).
if (!ps && msg.isRequest && msg.method === 'INVITE') {
logPacket(`[new outbound] INVITE from ${rinfo.address}:${rinfo.port}`, data);
const dialedNumber = SipMessage.extractUri(msg.requestUri || '') || '';
const routeResult = resolveOutboundRoute(
appConfig,
dialedNumber,
undefined,
(pid) => !!providerStates.get(pid)?.registeredAor,
);
if (routeResult) {
const provState = providerStates.get(routeResult.provider.id);
if (provState) {
// Apply number transformation to the INVITE if needed.
if (routeResult.transformedNumber !== dialedNumber) {
const newUri = msg.requestUri?.replace(dialedNumber, routeResult.transformedNumber);
if (newUri) msg.setRequestUri(newUri);
}
callManager.handlePassthroughOutbound(msg, { address: rinfo.address, port: rinfo.port }, routeResult.provider, provState);
}
}
return;
}
// 6. Fallback: forward based on direction (for mid-dialog messages
// that don't match any tracked call, e.g. OPTIONS, NOTIFY).
if (ps) {
// From provider -> forward to device.
logPacket(`[fallback inbound] from ${rinfo.address}:${rinfo.port}`, data);
const via = msg.getHeader('Via');
if (via) ps.detectPublicIp(via);
// Try to figure out where to send it...
// For now, just log. These should become rare once all calls are tracked.
log(`[fallback] unrouted inbound ${msg.isRequest ? msg.method : msg.statusCode} Call-ID=${msg.callId.slice(0, 30)}`);
} else {
// From device -> forward to provider.
logPacket(`[fallback outbound] from ${rinfo.address}:${rinfo.port}`, data);
const fallback = resolveOutboundRoute(appConfig, '');
if (fallback) sock.send(msg.serialize(), fallback.provider.outboundProxy.port, fallback.provider.outboundProxy.address);
}
} catch (e: any) {
log(`[err] ${e?.stack || e}`);
broadcastWs('registration', { providerId: data.provider_id, registered: data.registered });
}
});
sock.on('error', (err: Error) => log(`[main] sock err: ${err.message}`));
onProxyEvent('device_registered', (data: IDeviceRegisteredEvent) => {
const ds = deviceStatuses.get(data.device_id);
if (ds) {
ds.address = data.address;
ds.port = data.port;
ds.connected = true;
log(`[registrar] ${data.display_name} registered from ${data.address}:${data.port}`);
}
});
sock.bind(LAN_PORT, '0.0.0.0', () => {
const providerList = appConfig.providers.map((p) => p.displayName).join(', ');
const deviceList = appConfig.devices.map((d) => d.displayName).join(', ');
log(`sip proxy bound 0.0.0.0:${LAN_PORT} | providers: ${providerList} | devices: ${deviceList}`);
onProxyEvent('incoming_call', (data: IIncomingCallEvent) => {
log(`[call] incoming: ${data.from_uri}${data.to_number} via ${data.provider_id} (${data.call_id})`);
activeCalls.set(data.call_id, {
id: data.call_id,
direction: 'inbound',
callerNumber: data.from_uri,
calleeNumber: data.to_number,
providerUsed: data.provider_id,
state: 'ringing',
startedAt: Date.now(),
legs: new Map(),
});
// Start upstream provider registrations.
for (const ps of providerStates.values()) {
ps.startRegistration(
LAN_IP,
LAN_PORT,
(buf, dest) => sock.send(buf, dest.port, dest.address),
log,
(provider) => broadcastWs('registration', { providerId: provider.config.id, registered: provider.isRegistered }),
);
// Notify browsers of incoming call.
const browserIds = getAllBrowserDeviceIds();
for (const bid of browserIds) {
sendToBrowserDevice(bid, {
type: 'webrtc-incoming',
callId: data.call_id,
from: data.from_uri,
deviceId: bid,
});
}
});
onProxyEvent('outbound_device_call', (data: IOutboundCallEvent) => {
log(`[call] outbound: device ${data.from_device}${data.to_number} (${data.call_id})`);
activeCalls.set(data.call_id, {
id: data.call_id,
direction: 'outbound',
callerNumber: data.from_device,
calleeNumber: data.to_number,
providerUsed: null,
state: 'setting-up',
startedAt: Date.now(),
legs: new Map(),
});
});
onProxyEvent('outbound_call_started', (data: any) => {
log(`[call] outbound started: ${data.call_id}${data.number} via ${data.provider_id}`);
activeCalls.set(data.call_id, {
id: data.call_id,
direction: 'outbound',
callerNumber: null,
calleeNumber: data.number,
providerUsed: data.provider_id,
state: 'setting-up',
startedAt: Date.now(),
legs: new Map(),
});
// Notify all browser devices — they can connect via WebRTC to listen/talk.
const browserIds = getAllBrowserDeviceIds();
for (const bid of browserIds) {
sendToBrowserDevice(bid, {
type: 'webrtc-incoming',
callId: data.call_id,
from: data.number,
deviceId: bid,
});
}
});
onProxyEvent('call_ringing', (data: { call_id: string }) => {
const call = activeCalls.get(data.call_id);
if (call) call.state = 'ringing';
});
onProxyEvent('call_answered', (data: { call_id: string; provider_media_addr?: string; provider_media_port?: number; sip_pt?: number }) => {
const call = activeCalls.get(data.call_id);
if (call) {
call.state = 'connected';
log(`[call] ${data.call_id} connected`);
// Enrich provider leg with media info from the answered event.
if (data.provider_media_addr && data.provider_media_port) {
for (const leg of call.legs.values()) {
if (leg.type === 'sip-provider') {
leg.remoteMedia = `${data.provider_media_addr}:${data.provider_media_port}`;
if (data.sip_pt !== undefined) {
const codecNames: Record<number, string> = { 0: 'PCMU', 8: 'PCMA', 9: 'G.722', 111: 'Opus' };
leg.codec = codecNames[data.sip_pt] || `PT${data.sip_pt}`;
}
break;
}
}
}
}
// Initialize audio codec bridge (Rust binary via smartrust).
initCodecBridge(log)
.then(() => initAnnouncement(log))
.then(async () => {
// Pre-generate voicemail beep tone.
await promptCache.generateBeep('voicemail-beep', 1000, 500, 8000);
// Try to link WebRTC session to this call for audio bridging.
if (data.provider_media_addr && data.provider_media_port) {
const sessionId = webrtcCallToSession.get(data.call_id);
if (sessionId) {
// Both session and media info available — link now.
const sipPt = data.sip_pt ?? 9;
log(`[webrtc] linking session=${sessionId.slice(0, 8)} to call=${data.call_id} media=${data.provider_media_addr}:${data.provider_media_port} pt=${sipPt}`);
webrtcLink(sessionId, data.call_id, data.provider_media_addr, data.provider_media_port, sipPt).then((ok) => {
log(`[webrtc] link result: ${ok}`);
});
} else {
// Session not yet accepted — store media info for when it arrives.
pendingCallMedia.set(data.call_id, {
addr: data.provider_media_addr,
port: data.provider_media_port,
sipPt: data.sip_pt ?? 9,
});
log(`[webrtc] media info cached for call=${data.call_id}, waiting for session accept`);
}
}
});
// Pre-generate voicemail greetings for all configured voiceboxes.
onProxyEvent('call_ended', (data: ICallEndedEvent) => {
const call = activeCalls.get(data.call_id);
if (call) {
log(`[call] ${data.call_id} ended: ${data.reason} (${data.duration}s)`);
// Snapshot legs with metadata for history.
const historyLegs: IHistoryLeg[] = [];
for (const [, leg] of call.legs) {
historyLegs.push({
id: leg.id,
type: leg.type,
metadata: leg.metadata || {},
});
}
// Move to history.
callHistory.unshift({
id: call.id,
direction: call.direction,
callerNumber: call.callerNumber,
calleeNumber: call.calleeNumber,
startedAt: call.startedAt,
duration: data.duration,
legs: historyLegs,
});
if (callHistory.length > MAX_HISTORY) callHistory.pop();
activeCalls.delete(data.call_id);
// Notify browser(s) that the call ended.
broadcastWs('webrtc-call-ended', { callId: data.call_id });
// Clean up WebRTC session mappings.
const sessionId = webrtcCallToSession.get(data.call_id);
if (sessionId) {
webrtcCallToSession.delete(data.call_id);
webrtcSessionToCall.delete(sessionId);
webrtcClose(sessionId).catch(() => {});
}
pendingCallMedia.delete(data.call_id);
}
});
onProxyEvent('sip_unhandled', (data: any) => {
log(`[sip] unhandled ${data.method_or_status} Call-ID=${data.call_id?.slice(0, 20)} from=${data.from_addr}:${data.from_port}`);
});
// Leg events (multiparty) — update shadow state so the dashboard shows legs.
onProxyEvent('leg_added', (data: any) => {
log(`[leg] added: call=${data.call_id} leg=${data.leg_id} kind=${data.kind} state=${data.state}`);
const call = activeCalls.get(data.call_id);
if (call) {
call.legs.set(data.leg_id, {
id: data.leg_id,
type: data.kind,
state: data.state,
codec: data.codec ?? null,
rtpPort: data.rtpPort ?? null,
remoteMedia: data.remoteMedia ?? null,
metadata: data.metadata || {},
});
}
});
onProxyEvent('leg_removed', (data: any) => {
log(`[leg] removed: call=${data.call_id} leg=${data.leg_id}`);
activeCalls.get(data.call_id)?.legs.delete(data.leg_id);
});
onProxyEvent('leg_state_changed', (data: any) => {
log(`[leg] state: call=${data.call_id} leg=${data.leg_id}${data.state}`);
const call = activeCalls.get(data.call_id);
if (!call) return;
const leg = call.legs.get(data.leg_id);
if (leg) {
leg.state = data.state;
if (data.metadata) leg.metadata = data.metadata;
} else {
// Initial legs (provider/device) don't emit leg_added — create on first state change.
const legId: string = data.leg_id;
const type = legId.includes('-prov') ? 'sip-provider' : legId.includes('-dev') ? 'sip-device' : 'webrtc';
call.legs.set(data.leg_id, {
id: data.leg_id,
type,
state: data.state,
codec: null,
rtpPort: null,
remoteMedia: null,
metadata: data.metadata || {},
});
}
});
// WebRTC events from Rust — forward ICE candidates to browser via WebSocket.
onProxyEvent('webrtc_ice_candidate', (data: any) => {
// Find the browser's WebSocket by session ID and send the ICE candidate.
broadcastWs('webrtc-ice', {
sessionId: data.session_id,
candidate: { candidate: data.candidate, sdpMid: data.sdp_mid, sdpMLineIndex: data.sdp_mline_index },
});
});
onProxyEvent('webrtc_state', (data: any) => {
log(`[webrtc] session=${data.session_id?.slice(0, 8)} state=${data.state}`);
});
onProxyEvent('webrtc_track', (data: any) => {
log(`[webrtc] session=${data.session_id?.slice(0, 8)} track=${data.kind} codec=${data.codec}`);
});
onProxyEvent('webrtc_audio_rx', (data: any) => {
if (data.packet_count === 1 || data.packet_count === 50) {
log(`[webrtc] session=${data.session_id?.slice(0, 8)} browser audio rx #${data.packet_count}`);
}
});
// Voicemail events.
onProxyEvent('voicemail_started', (data: any) => {
log(`[voicemail] started for call ${data.call_id} caller=${data.caller_number}`);
});
onProxyEvent('recording_done', (data: any) => {
log(`[voicemail] recording done: ${data.file_path} (${data.duration_ms}ms) caller=${data.caller_number}`);
// Save voicemail metadata via VoiceboxManager.
voiceboxManager.addMessage?.('default', {
callerNumber: data.caller_number || 'Unknown',
callerName: null,
fileName: data.file_path,
durationMs: data.duration_ms,
});
});
onProxyEvent('voicemail_error', (data: any) => {
log(`[voicemail] error: ${data.error} call=${data.call_id}`);
});
// Send full config to Rust — this binds the SIP socket and starts registrations.
const configured = await configureProxyEngine({
proxy: appConfig.proxy,
providers: appConfig.providers,
devices: appConfig.devices,
routing: appConfig.routing,
});
if (!configured) {
log('[FATAL] failed to configure proxy engine');
process.exit(1);
}
const providerList = appConfig.providers.map((p) => p.displayName).join(', ');
const deviceList = appConfig.devices.map((d) => d.displayName).join(', ');
log(`proxy engine started | LAN ${appConfig.proxy.lanIp}:${appConfig.proxy.lanPort} | providers: ${providerList} | devices: ${deviceList}`);
// Generate TTS audio (WAV files on disk, played by Rust audio_player).
try {
await initAnnouncement(log);
// Pre-generate prompts.
await promptCache.generateBeep('voicemail-beep', 1000, 500, 8000);
for (const vb of appConfig.voiceboxes ?? []) {
if (!vb.enabled) continue;
const promptId = `voicemail-greeting-${vb.id}`;
const wavPath = vb.greetingWavPath;
if (wavPath) {
await promptCache.loadWavPrompt(promptId, wavPath);
if (vb.greetingWavPath) {
await promptCache.loadWavPrompt(promptId, vb.greetingWavPath);
} else {
const text = vb.greetingText || 'The person you are trying to reach is not available. Please leave a message after the tone.';
await promptCache.generatePrompt(promptId, text, vb.greetingVoice || 'af_bella');
}
}
// Pre-generate IVR menu prompts.
if (appConfig.ivr?.enabled) {
for (const menu of appConfig.ivr.menus) {
const promptId = `ivr-menu-${menu.id}`;
await promptCache.generatePrompt(promptId, menu.promptText, menu.promptVoice || 'af_bella');
await promptCache.generatePrompt(`ivr-menu-${menu.id}`, menu.promptText, menu.promptVoice || 'af_bella');
}
}
log(`[startup] prompts cached: ${promptCache.listIds().join(', ') || 'none'}`);
})
.catch((e) => log(`[codec] init failed: ${e}`));
});
} catch (e) {
log(`[tts] init failed: ${e}`);
}
}
// ---------------------------------------------------------------------------
// Web UI
@@ -300,33 +557,121 @@ initWebUi(
getStatus,
log,
(number, deviceId, providerId) => {
const call = callManager.createOutboundCall(number, deviceId, providerId);
return call ? { id: call.id } : null;
// Outbound calls from dashboard — send make_call command to Rust.
log(`[dashboard] start call: ${number} device=${deviceId || 'any'} provider=${providerId || 'auto'}`);
// Fire-and-forget — the async result comes via events.
makeCall(number, deviceId, providerId).then((callId) => {
if (callId) {
log(`[dashboard] call started: ${callId}`);
activeCalls.set(callId, {
id: callId,
direction: 'outbound',
callerNumber: null,
calleeNumber: number,
providerUsed: providerId || null,
state: 'setting-up',
startedAt: Date.now(),
legs: new Map(),
});
} else {
log(`[dashboard] call failed for ${number}`);
}
});
// Return a temporary ID so the frontend doesn't show "failed" immediately.
return { id: `pending-${Date.now()}` };
},
(callId) => {
hangupCall(callId);
return true;
},
(callId) => callManager.hangup(callId),
() => {
// Reload config after UI save.
// Config saved — reconfigure Rust engine.
try {
const fresh = loadConfig();
Object.assign(appConfig, fresh);
// Sync provider registrations: add new, remove deleted, re-register changed.
syncProviderStates(
fresh.providers,
proxy.publicIpSeed,
LAN_IP,
LAN_PORT,
(buf, dest) => sock.send(buf, dest.port, dest.address),
log,
(provider) => broadcastWs('registration', { providerId: provider.config.id, registered: provider.isRegistered }),
);
log('[config] reloaded config after save');
// Update shadow state.
for (const p of fresh.providers) {
if (!providerStatuses.has(p.id)) {
providerStatuses.set(p.id, {
id: p.id, displayName: p.displayName, registered: false, publicIp: null,
});
}
}
for (const d of fresh.devices) {
if (!deviceStatuses.has(d.id)) {
deviceStatuses.set(d.id, {
id: d.id, displayName: d.displayName, address: null, port: 0, connected: false, isBrowser: false,
});
}
}
// Re-send config to Rust.
configureProxyEngine({
proxy: fresh.proxy,
providers: fresh.providers,
devices: fresh.devices,
routing: fresh.routing,
}).then((ok) => {
if (ok) log('[config] reloaded — proxy engine reconfigured');
else log('[config] reload failed — proxy engine rejected config');
});
} catch (e: any) {
log(`[config] reload failed: ${e.message}`);
}
},
callManager,
voiceboxManager,
undefined, // callManager — legacy, replaced by Rust proxy-engine
voiceboxManager, // voiceboxManager
// WebRTC signaling → forwarded to Rust proxy-engine.
async (sessionId, sdp, ws) => {
log(`[webrtc] offer from browser session=${sessionId.slice(0, 8)} sdp_type=${typeof sdp} sdp_len=${sdp?.length || 0}`);
if (!sdp || typeof sdp !== 'string' || sdp.length < 10) {
log(`[webrtc] WARNING: invalid SDP (type=${typeof sdp}), skipping offer`);
return;
}
log(`[webrtc] sending offer to Rust (${sdp.length}b)...`);
const result = await webrtcOffer(sessionId, sdp);
log(`[webrtc] Rust result: ${JSON.stringify(result)?.slice(0, 200)}`);
if (result?.sdp) {
ws.send(JSON.stringify({ type: 'webrtc-answer', sessionId, sdp: result.sdp }));
log(`[webrtc] answer sent to browser session=${sessionId.slice(0, 8)}`);
} else {
log(`[webrtc] ERROR: no answer SDP from Rust`);
}
},
async (sessionId, candidate) => {
await webrtcIce(sessionId, candidate);
},
async (sessionId) => {
await webrtcClose(sessionId);
},
// onWebRtcAccept — browser has accepted a call, linking session to call.
(callId: string, sessionId: string) => {
log(`[webrtc] accept: callId=${callId} sessionId=${sessionId.slice(0, 8)}`);
// Store bidirectional mapping.
webrtcSessionToCall.set(sessionId, callId);
webrtcCallToSession.set(callId, sessionId);
// Check if we already have media info for this call (provider answered first).
const media = pendingCallMedia.get(callId);
if (media) {
pendingCallMedia.delete(callId);
log(`[webrtc] linking session=${sessionId.slice(0, 8)} to call=${callId} media=${media.addr}:${media.port} pt=${media.sipPt}`);
webrtcLink(sessionId, callId, media.addr, media.port, media.sipPt).then((ok) => {
log(`[webrtc] link result: ${ok}`);
});
} else {
log(`[webrtc] session ${sessionId.slice(0, 8)} accepted, waiting for call_answered media info`);
}
},
);
process.on('SIGINT', () => { log('SIGINT, exiting'); process.exit(0); });
process.on('SIGTERM', () => { log('SIGTERM, exiting'); process.exit(0); });
// ---------------------------------------------------------------------------
// Start
// ---------------------------------------------------------------------------
startProxyEngine();
process.on('SIGINT', () => { log('SIGINT, exiting'); shutdownProxyEngine(); process.exit(0); });
process.on('SIGTERM', () => { log('SIGTERM, exiting'); shutdownProxyEngine(); process.exit(0); });

View File

@@ -3,6 +3,6 @@
*/
export const commitinfo = {
name: 'siprouter',
version: '1.10.0',
version: '1.18.0',
description: 'undefined'
}

View File

@@ -20,7 +20,7 @@ export interface IDeviceStatus {
export interface ILegStatus {
id: string;
type: 'sip-device' | 'sip-provider' | 'webrtc';
type: 'sip-device' | 'sip-provider' | 'webrtc' | 'tool';
state: string;
remoteMedia: { address: string; port: number } | null;
rtpPort: number | null;
@@ -28,6 +28,7 @@ export interface ILegStatus {
pktReceived: number;
codec: string | null;
transcoding: boolean;
metadata?: Record<string, unknown>;
}
export interface ICallStatus {
@@ -42,6 +43,12 @@ export interface ICallStatus {
legs: ILegStatus[];
}
export interface IHistoryLeg {
id: string;
type: string;
metadata: Record<string, unknown>;
}
export interface ICallHistoryEntry {
id: string;
direction: 'inbound' | 'outbound' | 'internal';
@@ -50,6 +57,7 @@ export interface ICallHistoryEntry {
providerUsed: string | null;
startedAt: number;
duration: number;
legs?: IHistoryLeg[];
}
export interface IContact {