Setting up the Chatbot Server

To avoid unnecessary complications, perform the steps in the following order:

1. Clone the repository
2. Set up an Amazon DynamoDB instance
3. Set up the AWS Lambda function
4. Link the Lambda function to DynamoDB
5. Set up the Amazon Lex Model

Clone the Repository

The following example shows how to clone the github.com/elasticpath/lex-chatbot repository and install the project dependencies.

# Clone the Git repository
git clone https://github.com/elasticpath/lex-chatbot.git

# Go into the cloned directory
cd lex-chatbot/ep-lambda-function

# Install all the dependencies for the project
npm install

Set up an Amazon DynamoDB Instance

The Reference Chatbot uses a simple Amazon-hosted noSQL database instance to track purchase state across multiple active devices.

1. In the AWS console, go to the DynamoDB service.
2. Create a new table.
3. Set the name to DynamoCache.
4. Set the primary key to responseId.
5. In the Table Overview, locate the Time to Live Attribute and click Enable TTL.
6. Set the TTL attribute to ttl and click Continue.

Set up the AWS Lambda Function

You must have a valid Elastic Path Commerce platform. The platform is used for the backend of the lambda.

1. Zip up the contents of the current directory, including the node_modules folder.
2. In the Amazon Lambda Console, create a new Function called EPConversationalLambda.
3. Select Author From Scratch template.
4. Choose Node.js 10.x as the runtime language.
5. Under the Function Code pane, change Code Entry Type to Upload a .zip folder.
6. Locate and upload the newly zipped folder.
7. Under Environment Variables, add and set the following values then save the function:

Key	Value
CACHE_TABLE	DynamoCache
CACHE_REGION	<Your_AWS_Region>
SCOPE	<Your_Cortex_Scope>
CORTEX_URL	<Your_Cortex_URL>
HOST_URL	<Your_Host_Service_URL>
SKU_IMAGES_URL	<Your_SKU_Images_URL>
ROLE	PUBLIC
GRANT_TYPE	password
USERNAME	[empty]
PASSWORD	[empty]

Link the Lambda Function to DynamoDB

Link the lambda function to DynamoDB by using IAM (Identity and Access Management) to read and write to the database table. This action preserves the state of the transaction.

1. In the AWS Lambda Designer, click the Key icon to view Permissions.
2. In the Execution Role pane, find the roleName value. Copy the value to the clipboard.
3. Browse Amazon Services, and search for IAM (Identity and Access Management).
4. In the IAM Console, go to the AWS Account field, select Roles.
5. In the search bar, paste the roleName and select the EPConversationalLambda role.
6. In the Permissions tab, select the AWSLambdaBasicExecutionRole.
7. Click Edit Policy.
8. Click Add additional permissions and select DynamoDB as the service.
9. Under the Actions drop-down list, enable Access level for List, Read, and Write.
10. Under the Resources drop-down list, enable All resources.
11. Click Review Policy.
12. Ensure the DynamoDB Access level is accurate.
13. Click to save your changes.

Set up the Amazon Lex Model

Import the Lex Model into Amazon Lex and then set the fulfillment logic.

1. Go to the lex-chatbot/ep-lex-models directory.
2. Add the lex-model.json file to a zip folder.
3. Go to the Amazon Lex console to view your bots. Note: You must have at least one bot available to import an existing model. Create one of the pre-existing bots as a sample to access this feature.
4. Under the Actions drop-down list, select Import.
5. Upload the lex-model.zip file.
6. Set the Fulfillment logic to AWS Lambda function, and select your function from the drop-down list. Save the Intent.
7. Repeat step 6 for each Intent.