Skip to content

README.md

Latest commit

 

History

History
49 lines (30 loc) · 1.75 KB

README.md

File metadata and controls

49 lines (30 loc) · 1.75 KB
Raw

Splintercom Python Client

Python device client that authenticates via OAuth2 device flow, streams camera video over WebRTC, and sends telemetry to the Splintercom API.

Tech Stack

  • Python 3.14+, uv
  • aiortc (WebRTC), OpenCV (camera capture), websockets (signaling), requests (HTTP)

Development

# From the repo root (recommended)
docker compose up -d clientpython

# Logs
docker compose logs -f clientpython

# Rebuild after code changes
docker compose up -d --build clientpython

Requires /dev/video0 on the host (mapped into the container). Tokens and OAuth config are shared with the utility service via Docker volumes.

Camera Architecture

A single SharedCameraSource opens cv2.VideoCapture once and fans frames to per-viewer asyncio.Queues. Each CameraVideoStreamTrack subscribes with a unique viewer key. This supports multiple simultaneous viewers without attempting to re-open the device. The camera is released when all viewers disconnect, and re-opened on the next connection.

CI/CD

Push to master triggers two workflows:

build-docker.yml — builds and pushes docker.kmanning.ie:5000/intercom-python-client:{latest,sha,version} (ARM runner)

build-deb.yml — runs pytest, then builds a native ARM64 binary with Nuitka and packages it as a .deb for Raspberry Pi installation. Uploads the artifact with 30-day retention.

Both workflows send a structured event to Honeycomb (github-actions dataset) at job completion — requires HONEYCOMB_API_KEY GitHub secret.

Installation on Raspberry Pi (from .deb)

sudo dpkg -i splintercom-client_<version>_arm64.deb

# Configure
sudo nano /etc/splintercom-client/env

# Enable and start
sudo systemctl enable --now splintercom-client