/apprtc

The video chat demo app based on WebRTC

Primary LanguageJavaScriptBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

Build Status

AppRTC Demo Code

Development

Detailed information on developing in the webrtc github repo can be found in the WebRTC GitHub repo developer's guide.

The development AppRTC server can be accessed by visiting http://localhost:8080.

Running AppRTC locally requires the Google App Engine SDK for Python and Grunt.

Detailed instructions for running on Ubuntu Linux are provided below.

Running on Ubuntu Linux

Install grunt by first installing npm. npm is distributed as part of nodejs.

sudo apt-get install nodejs
sudo npm install -g npm

On Ubuntu 14.04 the default packages installs /usr/bin/nodejs but the /usr/bin/node executable is required for grunt. This is installed on some Ubuntu package sets; if it is missing, you can add this by installing the nodejs-legacy package,

sudo apt-get install nodejs-legacy

It is easiest to install a shared version of grunt-cli from npm using the -g flag. This will allow you access the grunt command from /usr/local/bin. More information can be found on gruntjs Getting Started.

sudo npm -g install grunt-cli

Omitting the -g flag will install grunt-cli to the current directory under the node_modules directory.

Finally, you will want to install grunt and required grunt dependencies. This can be done from any directory under your checkout of the webrtc/apprtc repository.

npm install

On Ubuntu, you will also need to install the webtest package:

sudo apt-get install python-webtest

Before you start the AppRTC dev server and *everytime you update the source code you need to recompile the App Engine package by running,

grunt build

Start the AppRTC dev server from the out/app_engine directory by running the Google App Engine SDK dev server,

<path to sdk>/dev_appserver.py ./out/app_engine

Then navigate to http://localhost:8080 in your browser (given it's on the same machine).

Testing

All tests by running grunt.

To run only the Python tests you can call,

grunt runPythonTests

Enabling Local Logging

Note that logging is automatically enabled when running on Google App Engine using an implicit service account.

By default, logging to a BigQuery from the development server is disabled. Log information is presented on the console. Unless you are modifying the analytics API you will not need to enable remote logging.

Logging to BigQuery when running LOCALLY requires a secrets.json containing Service Account credentials to a Google Developer project where BigQuery is enabled. DO NOT COMMIT secrets.json TO THE REPOSITORY.

To generate a secrets.json file in the Google Developers Console for your project:

  1. Go to the project page.
  2. Under APIs & auth select Credentials.
  3. Confirm a Service Account already exists or create it by selecting Create new Client ID.
  4. Select Generate new JSON key from the Service Account area to create and download JSON credentials.
  5. Rename the downloaded file to secrets.json and place in the directory containing analytics.py.

When the Analytics class detects that AppRTC is running locally, all data is logged to analytics table in the dev dataset. You can bootstrap the dev dataset by following the instructions in the Bootstrapping/Updating BigQuery.

BigQuery

When running on App Engine the Analytics class will log to analytics table in the prod dataset for whatever project is defined in app.yaml.

Schema

bigquery/analytics_schema.json contains the fields used in the BigQuery table. New fields can be added to the schema and the table updated. However, fields cannot be renamed or removed. Caution should be taken when updating the production table as reverting schema updates is difficult.

Update the BigQuery table from the schema by running,

bq update -t prod.analytics bigquery/analytics_schema.json

Bootstrapping

Initialize the required BigQuery datasets and tables with the following,

bq mk prod
bq mk -t prod.analytics bigquery/analytics_schema.json

Deployment

Note this assumes you are setting up AppRTC (Web client and backend), Collider (signalling server) and the Coturn TURN server on the same machine and that it is running Linux, if you are not just perform the steps of each application on the machine you want to run them on. Instructions were performed on Ubuntu 14.04 using Python 2.7.9 and Go 1.6.3.

  1. Clone the AppRTC repository on the machine you want to host it on (git clone )
  2. Do all the steps in the Collider instructions.
  3. In the AppRTC git checkout navigate to src/app_engine/constants.py and change the WSS_INSTANCE_HOST_KEY: to the address and port Collider is listening too, in this case localhost:8089.
  4. If you want just want to setup AppRTC with a TURN server directly then continue to step 5 otherwise jump to 7.
  5. Install and start the Coturn TURN server according to their instructions
  6. Now navigate to src/web_app/js/utils.js and change the requestIceServers function to the following:
function requestIceServers(iceServerRequestUrl, iceTransports) {
  return new Promise(function(resolve, reject) {
    var servers = [{
        credential: "turnPassword",
        username: "turnUser",
        urls: [
          "turn:localhost:3478?transport=udp",
          "turn:localhost:3478?transport=tcp"
        ]
    }];
    resolve(servers);
  });
}

7. (Only do this if you skipped step 5 and 6) AppRTC by default uses an ICE server provider to get TURN servers, it's basically just a web server with authentication that returns a JSON response containing TURN servers with credentials, note that before it provides a response, it checks where the user is connecting from, checks if there are any TURN servers in that area, if not it spins up an instance and gets it's reachable address and credentials. If you have such a service then change the ICE server constants in constants.py to point to that.

8. Now build AppRTC using grunt build and then start it using dev appserver provided by GAE pathToGAESDK/dev_appserver.py out/app_engine/.

9. Open a WebRTC enabled browser and navigate to http://localhost:8080 (http://localhost:8080?wstls=false if you have TLS disabled on Collider)