There are three requirements for getting these demos working:

1. Add the AWS SDK .jar and related files to your CF install.
2. Set up your own AWS credentials and add them to awsCredentials.cfc.
3. Set up your own copies of the required resources in SNS, Lambda, DynamoDB, S3, and Step Functions.

If you are running CF2018, the demos in this repo require that you have the following .jar file in your /cfusion/lib/ directory:

• aws-java-sdk-1.11.311 or later

If you are running CF2016 or earlier, you also need to add the following .jar files to your /cfusion/lib/ directory:

• jackson-annotations
• jackson-core
• jackson-databind
• joda-time

All of these files can be downloaded from https://aws.amazon.com/sdk-for-java/ Files other than the actual SDK .jar itself can be found in the /third-party directory within the SDK download.

You have to create your own AWS account and provide both the AccessKey and SecretKey in model/awsCredentials.cfc.

The account for which you are providing credentials must also have permissions for the following services:

• IAM
• S3
• SNS
• Lambda
• CloudWatch (for Lambda logging)
• DynamoDB
• Step Functions
• Rekognition
• Transcribe
• Translate
• Polly

For more infomration about IAM accounts, roles, and permissions, please review the IAM guide.

You need to set up the following resources within AWS for these demos to work:

1. SNS - create a topic to which messsages can be sent. The ARN (Amazon Resource Name, like a URL) of this topic must be added to application.cfc.
2. Lambda - create a Lambda function using the code in nodejs/lambda/lambda-returnDataToCaller.js. The Lambda runtime should be NodeJS 4.3 or later, and you do not need to configure a trigger for the function, as it will be invoked from this application. The ARN of the function must be added to application.cfc.
3. DynamoDB - create a DynamoDB table with a partition (primary) key of "userID" (String) and a sort key (range key) of "epochTime" (Number). The table name must be added to application.cfc.
4. Rekognition - add photos for Rekognition to analyze. There's a separate list of photos for matching faces, and a list for generating labels (image analysis). These photos and the name of the S3 bucket in which they can be found need to be added to the top of rekognition.cfm.
5. Step Functions: There are two workflows you can set up:

• Describe an Image

  a. Create the two Lambda functions used in this workflow using the code in nodejs/lambda/ -- generateRandomNumber.js and detectLabelsForImage.js.
  b. Add the ARNs of those functions to stateMachines/choiceDemoStateMachine.json.
  c. Add the name of the S3 bucket and the path to the photos that will be analyzed to stateMachines/choiceDemoStateMachine.json.
  d. Once you've added all the required information, use stateMachines/choiceDemoStateMachine.json to create a new Step Function state machine in the AWS Console.
  e. Add the ARN of the workflow to application.cfc as the application.awsResources.stepFunctionRandomImageARN value.

• Transcribe, Translate, and Speak a Video

a. Create the five Lambda functions used in this workflow using the code in nodejs/lambda/transcribeTranslateExample. You will need to add the name of your S3 bucket where you want the output from the workflow to go to getTranscriptionFile.js, translateText.js, and convertTextToSpeech.js.  
b. Add the ARNs of those functions to stateMachines/transcribeTranslateSpeakWorkflow.json.   
c. Once you've added all the required information, use stateMachines/transcribeTranslateSpeakWorkflow.json to create a new Step Function state machine in the AWS Console.    
d. Add the ARN of the workflow to application.cfc as the application.awsResources.stepFunctionTranscribeTranslateARN value.   
e. Modify stepFunctions.cfm (the inputStruct variable) to point to the URL of a MP4 file on S3.    

Remember, the AWS docs are pretty great. Use the Java API Reference often, as it'll tell you almost everything you need to know for working with a particular AWS service.

Enjoy!

There are a number of Python examples in this repo. You can get them running pretty easily by:

1. Installing boto via pip
2. Using your own ARNs as noted in the Python code in the /python directory of the repo
3. Setting python/pythonInvoke.sh to run as an executable on your machine

Boto makes it very easy to use AWS from Python and acts as the AWS-approved and supported SDK for Python.