umahüsla 2018 - Alexa Skill

Setup

Install the dependencies with

  yarn // or npm install

Start all Services

Once you have followed the setup guides below, you can start the app.

• Node: Start the app (both the Feed Server and the Alexa Server) with yarn start.
• Python (gql): start the gql server with docker-compose up and v/bin/server
• (Optional): Expose the services to the world with yarn start:public, you then do not need to run yarn start first.

Start all services with pm2

One can use pm2 to start the node services and keep them running (eg. with auto restart). Some important pm2 commands are listed below:

pm2 start pm2/pm2.config.js // starts all node services
pm2 stop pm2/pm2.config.js // stops all node services
pm2 kill // will kill all processes
pm2 log // see http://pm2.keymetrics.io/docs/usage/log-management/

You can then omit yarn start. For yarn start:public use pm2/pm2-public.config.js.

Setup

Localtunnel Setup

We use localtunnel to expose the local alexa and feed server to the world. One can configure the localtunnel Subdomain in the .env file. Otherwise the subdomain will change with every start of the localtunnel.

To setup localtunnel, you need to change the following in .env:

## Ports
ALEXA_SERVER_PORT=3000
FEED_SERVER_PORT=8000

## LOCALTUNNEL
LOCALTUNNEL_ENABLED=true
LOCALTUNNEL_FEED_DOMAIN=feed3000
LOCALTUNNEL_ALEXA_DOMAIN=alexa3000
LOCALTUNNEL_GQL_DOMAIN=gql3000

The localtunnel url needs to be added as the Alexa-Skill endpoint in the Alexa Skill settings.

GQL Server Setup (node)

Add the following to the .env:

## PYTHON SERVER
GQL_SERVER=http://localhost:9090/gql

The port must match with the gql python server port.

Feed & Twitter Setup

Setup the .env and add the following keys to setup twitter and the feed server:

## TWITTER
TWITTER_CONSUMER_KEY=
TWITTER_CONSUMER_SECRET=
TWITTER_ACCESS_TOKEN_KEY=
TWITTER_ACCESS_TOKEN_SECRET=

FEED_SERVER_PORT=8000

One can then start the feed server with yarn start:server. The feed is then available at http://localhost:8000/feed.

Available feeds:

• http://localhost:8000/feed
• http://localhost:8000/feed_with_audio
• http://localhost:8000/feed_with_video
• http://localhost:8000/feed_gql (only works when the python server runs as well)

Alexa-Skill Setup

Setup the .env and add the following keys:

## ALEXA SERVER
ALEXA_SERVER_PORT=3000
ALEXA_SERVER_DOMAIN=localhost

One can then start the alexa server with yarn start:alexa

Setup Video Skill

• https://developer.amazon.com/docs/custom-skills/videoapp-interface-reference.html#configure-your-skill-for-the-videoapp-directives

Intents and Utterances

Just start the Alexa server and open http://localhost:3000/alexa/umahuesla. The intents and utterances are at the bottom of the test-pages.

Python App Quick Guide

Local environment:

Run ./gradlew dev

Start Crate with docker-compose up -d.

Init DB Tables with v/bin/init_db.

Configure Env VARS (evertime or put it into .bash_profile):

export UH_TWITTER_API_CONSUMER_KEY=...
export UH_TWITTER_API_CONSUMER_SECRET=...
export UH_TWITTER_API_ACCESS_TOKEN_KEY=...
export UH_TWITTER_API_ACCESS_TOKEN_SECRET=...

Expose the GQL to the world by running yarn start:public (Note: this will also expose the other services). Optionally you can set the subdomain by adding yarn start:python --subdomain test123. This will result in (if subdomain is available): https://test123.localtunnel.me

Once the localtunnel is ready export another Env variable with the given localtunnel url. Set it with every start or put it into .bash_profile as well.

export UH_GQL_SERVER=...

Example: export UH_GQL_SERVER="https://gql-magic-3000.localtunnel.me/"

Now, start server with v/bin/server. Fetches new tweets every 2 minutes (latest 15 tweets as long as running).

Server exposes on http:/localhost:9090/gqlui and on the localtunnel url. The crate console is available at http://localhost:4200/#/.