Drink Recommender Skill

Alexa skill using decision template making drink recommendations.

Table of Contents

• Where do the questions come from?
• How do you deploy the skill?
• How can you automate testing?
• How to customize for Echo Show?

Data Source for Questions

The data folder contains the questions referenced in the skill. These are modeled in json.

Deploy from Command Line

To deploy the package, execute the build.sh script. All of the necessary files are zipped up in a package, staged to S3, then deployed as a lambda function. Here are the commands to setup.

# create temp zip file with build package contents
echo 'zipping up files'
zip -r drinkbot.zip index.js package.json data/questions.json node_modules/ > temp.log
echo 'build file created'

# stage the temp file in s3
aws s3 cp drinkbot.zip s3://drinkrecommender/binaries/

# remove the temporary file
rm drinkbot.zip

# update the lambda function with the binaries that have been staged
aws lambda update-function-code --function-name drinkRecommendationGreen --s3-bucket drinkrecommender --s3-key binaries/drinkbot.zip

Testing of Lambda from Command Line

Within the deployment script, there are CLI commands that read in a test object, then invoke the Lambda function directly.

# read in test data required to invoke the lambda function
echo 'test case 1 - basic request'
cd testing
request=$(<request.json)
cd ..

# invoke the new lambda function
aws lambda invoke --function-name drinkRecommendationGreen --payload "$request" testOutput.json

# read response file into local variable then print on the console
response=$(<testOutput.json)
echo $response
echo 'test case 1 complete'

Background Template for Echo Show

To render a background image on the Echo show, there are a few additional pieces to incorporate into the skill.

• Stage a 1024x600 image in an S3 bucket.
• Check the device type invoking the skill as you can only render an image response for an Echo.
• Render a template for the image.

Here is the relevant code within the skill that accomplishes just this.

var newSessionHandler = {
  'LaunchRequest': function () {
    this.handler.state = states.STARTMODE;
    // Display.RenderTemplate directives can be added to the response
    const builder = new Alexa.templateBuilders.BodyTemplate1Builder();
    const imageLoc = 'https://s3.amazonaws.com/drinkrecommender/media/drinkBackground.png';
    const template = builder.setTitle('Your Personal Bartender')
							.setBackgroundImage(makeImage(imageLoc))
							.setTextContent(makePlainText('Drink Recommender'))
							.build();

    if (this.event.context.System.device.supportedInterfaces.Display) {
	//this.response.speak(welcomeMessage).listen(repeatWelcomeMessage).renderTemplate(template);
	this.response.speak(welcomeMessage).listen(repeatWelcomeMessage);
        this.emit(':responseReady');
	console.log("this was requested by an Echo Show");
    } else {
    	this.emit(':ask', welcomeMessage, repeatWelcomeMessage);
    }
  }