template.yaml

AWSTemplateFormatVersion: '2010-09-09'
Transform: AWS::Serverless-2016-10-31
Description: patterns-s3-docrepo

Parameters:
  DocumentsBucketName:
    Type: String
    Default: 'patterns-s3docrepo-documents'
  StagingBucketName:
    Type: String
    Default: 'patterns-s3docrepo-staging'
  QueuedBucketName:
    Type: String
    Default: 'patterns-s3docrepo-queued'
  ESdomain:
    Type: String
    Default: '<<YOUR-ENDPOINT-HERE>>'

Globals:
  Function:
    Timeout: 15
    Environment:
      Variables:
        language: 'en'

Resources:
  ## S3 buckets
  DocumentsBucket:
    Type: AWS::S3::Bucket  
    Properties:
      BucketName: !Ref DocumentsBucketName
  StagingBucket:
    Type: AWS::S3::Bucket
    Properties:
      BucketName: !Ref StagingBucketName
  QueuedBucket:
    Type: AWS::S3::Bucket
    Properties:
      BucketName: !Ref QueuedBucketName

  ## SQS queue
  MySqsQueue:
    Type: AWS::SQS::Queue

  # Lambda functions

  queryESFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: queryESindex/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 128
      Timeout: 10
      ReservedConcurrentExecutions: 1
      Environment:
        Variables:
          domain: !Ref ESdomain
      Events:
        HttpApiEvent:
          Type: HttpApi
          Properties:
            Path: /
            Method: GET

  AddToESFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: addToESindex/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 128
      Timeout: 10
      ReservedConcurrentExecutions: 1
      Environment:
        Variables:
          domain: !Ref ESdomain          
      Policies:
        - S3ReadPolicy:
            BucketName: !Ref QueuedBucketName        
        - Statement:
          - Effect: Allow
            Resource: '*'
            Action:
              - comprehend:*
      Events:
        MySQSEvent:
          Type: SQS
          Properties:
            Queue: !GetAtt MySqsQueue.Arn
            BatchSize: 1

  BatchingFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: batchingFunction/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 512
      Environment:
        Variables:
          OutputBucket: !Ref QueuedBucketName
      Policies:
        - S3CrudPolicy:
            BucketName: !Ref QueuedBucketName
        - S3ReadPolicy:
            BucketName: !Ref StagingBucketName
      Events:
        OriginalTextUpload:
          Type: S3
          Properties:
            Bucket: !Ref StagingBucket
            Events: s3:ObjectCreated:*
            Filter: 
              S3Key:
                Rules:
                  - Name: suffix
                    Value: '.txt'

  AddToQueueFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: addToQueueFunction/
      Handler: app.handler
      Runtime: nodejs12.x
      Timeout: 15
      MemorySize: 128
      Environment:
        Variables:
          SQSqueueName: !Ref MySqsQueue
      Policies:
        - SQSSendMessagePolicy:
            QueueName: !GetAtt MySqsQueue.QueueName
        - S3ReadPolicy:
            BucketName: !Ref QueuedBucketName
       
      Events:
        FileUpload:
          Type: S3
          Properties:
            Bucket: !Ref QueuedBucket
            Events: s3:ObjectCreated:*

  ProcessJPGFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: processJPG/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 128
      Environment:
        Variables:
          OutputBucket: !Ref QueuedBucketName
          MaxLabels: 5
          MinConfidence: 60
      Policies:
        - S3CrudPolicy:
            BucketName: !Ref QueuedBucketName
        - S3ReadPolicy:
            BucketName: !Ref DocumentsBucketName
        - Statement:
          - Effect: Allow
            Resource: '*'
            Action:
              - rekognition:*
      Events:
        Upload:
          Type: S3
          Properties:
            Bucket: !Ref DocumentsBucket
            Events: s3:ObjectCreated:*
            Filter: 
              S3Key:
                Rules:
                  - Name: suffix
                    Value: '.jpg'

  ProcessDOCXFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: processDOCX/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 512
      Environment:
        Variables:
          OutputBucket: !Ref StagingBucketName
      Policies:
        - S3CrudPolicy:
            BucketName: !Ref StagingBucketName
        - S3ReadPolicy:
            BucketName: !Ref DocumentsBucketName
      Events:
        Upload:
          Type: S3
          Properties:
            Bucket: !Ref DocumentsBucket
            Events: s3:ObjectCreated:*
            Filter: 
              S3Key:
                Rules:
                  - Name: suffix
                    Value: '.docx'

  ProcessPDFFunction:
    Type: AWS::Serverless::Function 
    Properties:
      CodeUri: processPDF/
      Handler: app.handler
      Runtime: nodejs12.x
      MemorySize: 512
      Environment:
        Variables:
          OutputBucket: !Ref StagingBucketName
      Policies:
        - S3CrudPolicy:
            BucketName: !Ref StagingBucketName
        - S3ReadPolicy:
            BucketName: !Ref DocumentsBucketName
      Events:
        Upload:
          Type: S3
          Properties:
            Bucket: !Ref DocumentsBucket
            Events: s3:ObjectCreated:*
            Filter: 
              S3Key:
                Rules:
                  - Name: suffix
                    Value: '.pdf'

## Outputs
Outputs: 
  DocumentsBucketName:
    Description: Documents Bucket Name
    Value: !Ref DocumentsBucket

  StagingBucketName:
    Description: Staging Bucket Name
    Value: !Ref StagingBucket

  QueuedBucketName:
    Description: Queued Bucket Name
    Value: !Ref QueuedBucket

  AddToESFunctionARN:
    Description: This AddToESFunction function ARN.
    Value: !GetAtt AddToESFunction.Arn

  AddToESFunctionRoleARN:
    Description: This ARN needs permission in the Elasticsearch configuration.
    Value: !GetAtt AddToESFunctionRole.Arn

  queryESFunctionARN:
    Description: This ARN needs permission in the Elasticsearch configuration.
    Value: !GetAtt queryESFunction.Arn