Adding optional support for ONVIF Motion Events using Webhooks

[jeffreylanters]

In regards to the issue about motion detection not functioning on specific cameras (cough Tapo), I have done some further investigation. I have successfully replicated the problem using my own cameras using the same library which is being used in Scrypted (onvif), using Pullpoint communication. Upon inspecting the Scrypted source code, I observed (please correct me if I am mistaken) that it exclusively using Pullpoint communication for receiving motion detection events.

It is safe to say that the recent firmware update has, to the best of my ability to reproduce, has broken the camera's Pullpoint communication implementation. It is worth noting that TP-Link has not communicated any alterations in their implementation in the latest firmware updates. In response, I decided to experiment with Webhook communication, and I achieved positive results. Using a minimal server with Webhooks, I managed to receive both Motion and CellMotion detection events when motion initiated and ceased.

const http = require("http");
const { Cam } = require("onvif");
const { xml2js } = require("xml-js");

const serverConfig = {
  port: 8080,
  host: "", // Current machine's IP
};

const camConfig = {
  hostname: "",
  username: "",
  password: "",
  port: 2020,
  timeout: 10000,
};

const cam = new Cam(camConfig, handleCamConnect);
const server = http.createServer(handleRequest);
const uniqueSessionId = Date.now();

function handleServerListen() {
  console.log("Server is running", { serverConfig });
}

function handleCamConnect(error) {
  if (error) {
    throw new Error(error);
  }
  cam.getDeviceInformation(handleGetDeviceInformation);
  cam.getCapabilities(handleGetCapabilities);
}

function handleGetDeviceInformation(error, info, xml) {
  if (error) {
    throw new Error("Error getting device information", error);
  }
  console.log("Device Info", { info });
}

function handleGetCapabilities(error, capabilities, xml) {
  if (error) {
    throw new Error("Error getting capabilities", error);
  }
  if (!capabilities.events) {
    throw new Error("No events capabilities");
  }
  const authority = `http://${serverConfig.host}:${serverConfig.port}`;
  const path = `onvifnotify/${uniqueSessionId}`;
  const url = [authority, path].join("/");

  function renewSubscription() {
    cam.subscribe({ url }, handleSubscribe);
  }

  console.log("Event capabilities", { events: capabilities.events });
  setTimeout(renewSubscription, 1.5 * 60 * 1000);
  renewSubscription();
}

function handleSubscribe(error, subscription, xml) {
  if (error) {
    throw new Error("Error subscribing to events", error);
  }
  console.log("Subscribed", { subscription });
}

function handleRequest(request, response) {
  const data = [];

  function handleRequestData(chunk) {
    data.push(chunk);
  }

  function handleRequestEnd() {
    const text = Buffer.concat(data).toString();
    const body = xml2js(text, { compact: true });
    const notify = body["SOAP-ENV:Envelope"]["SOAP-ENV:Body"]["wsnt:Notify"];
    const notification = notify["wsnt:NotificationMessage"];
    const topicValue = notification["wsnt:Topic"]._text;
    const message = notification["wsnt:Message"]["tt:Message"];
    const messageValue = message["tt:Data"]["tt:SimpleItem"]._attributes.Value;

    console.log("Notification", { topicValue, messageValue });
  }

  request.on("data", handleRequestData);
  request.on("end", handleRequestEnd);
  response.writeHead(200);
  response.end("ok");

  console.log("Request", { url: request.url, method: request.method });
}

server.listen(serverConfig.port, serverConfig.host, handleServerListen);

Regrettably, I have not yet found the time to look into the process of integrating this into Scrypted. However, given that Scrypted does not currently utilize Webhooks at all while most cameras are working just fine using Pullpoint communication, it might not be a good idea to entirely replace the existing implementation. Instead, a more viable approach might involve enabling Webhooks via a toggle in the dashboard.

I am willing to contribute to the project if it is seems to be a valuable addition to the ONVIF plugin. I'm leaving this open for further discussion.

[koush]

Interesting, thanks I was thinking about writing my own onvif library as the one I'm currently using is a bit of a dated mess. The main issue with this approach is that the server needs to have a reachable address, which should generally be straightforward. However, if the server is behind a nat from the camera, it will not work without forwarding.

[jeffreylanters]

It indeed is -- taking a look at some other ONVIF libraries, when not taking certificates and encryption into account, it doesn't seem to be too complex to get a basic module up and running. I'm free to help building the implementation, feel free to let me know if you're planning something. Regarding the server, an insecure server seems te be required as well, since the protocol doesn't seem te be able to handle to call a webhook behind a secure protocol.

[koush]

no plans to implement this due to possible reachability issues and questionable additional value. I'll keep it in mind if/when I get to the onvif reimplementation.