---
AWSTemplateFormatVersion: '2010-09-09'
Description: >-
  (WWPS-GLS-WF-CROMWELL-AIO) Creates the complete set of resources needed to run
  genomics workflows using Cromwell on AWS Batch


Mappings:
  TagMap:
    default:
      architecture: "genomics-workflows"
      solution: "cromwell"
      tags:
        - Key: "architecture"
          Value: "genomics-workflows"
        - Key: "solution"
          Value: "cromwell"


Metadata:
  AWS::CloudFormation::Interface:
    ParameterGroups:
      - Label:
          default: "VPC"
        Parameters:
          - Namespace
          - VpcId
          - SubnetIds
          - ServerSubnetID
          - DBSubnetIDs
      - Label:
          default: "Data Storage"
        Parameters:
          - S3DataBucketName
          - ExistingDataBucket
      - Label:
          default: "AWS Batch"
        Parameters:
          - DefaultCEMinvCpus
          - DefaultCEMaxvCpus
          - PriorityCEMinvCpus
          - PriorityCEMaxvCpus
      - Label:
          default: "Cromwell Resources"
        Parameters:
          - CromwellVersion
          - CromwellVersionSpecified
          - CromwellJarUrl
          - S3DataBucketARNs
          - DBUsername
          - DBPassword

    ParameterLabels:
      Namespace:
        default: Namespace
      VpcId:
        default: VPC ID
      SubnetIds:
        default: VPC Subnet IDs
      ServerSubnetID:
        default: "Server Subnet ID"
      S3DataBucketName:
        default: S3 Data Bucket Name
      ExistingDataBucket:
        default: Existing Data Bucket?
      DefaultCEMinvCpus:
        default: Default Min vCPU
      DefaultCEMaxvCpus:
        default: Default Max vCPU
      PriorityCEMinvCpus:
        default: High Priority Min vCPU
      PriorityCEMaxvCpus:
        default: High Priority Max vCPU
      S3DataBucketARNs:
        default: "S3 Open Data Bucket ARNs"
      CromwellVersion:
        default: "Cromwell Version"
      CromwellVersionSpecified:
        default: "Cromwell Version Specified"
      DBSubnetIDs:
        default: "Database Subnet IDs"
      DBUsername:
        default: "Cromwell Database Username"
      DBPassword:
        default: "Cromwell Database Password"
      TemplateRootUrl:
        default: Template Root URL


Parameters:
  Namespace:
    Type: String
    Description: Namespace (e.g. project name) to use to label resources. Between 4-20 letters, digits, and '-', starting with a letter.
    Default: ""

  VpcId:
    Type: AWS::EC2::VPC::Id
    Description: 'The VPC to create security groups and deploy AWS Batch to. NOTE: Must be the same VPC as the provided subnet IDs.'
  SubnetIds:
    Type: List<AWS::EC2::Subnet::Id>
    Description: 'Subnets you want your batch compute environment to launch in. We recommend private subnets. NOTE: Must be from the VPC provided.'
  ServerSubnetID:
    Type: AWS::EC2::Subnet::Id
    Description: Subnet for the Cromwell server. For public access, use a public subnet.

  S3DataBucketName:
    Description: >-
      A S3 bucket name for storing analysis results
      The bucket name must respect the S3 bucket naming conventions 
      (can contain lowercase letters, numbers, periods and hyphens).
      If left blank a unique bucket name will be generated.
    Type: String
    Default: ""
    AllowedPattern: "((?=^.{3,63}$)(?!^(\\d+\\.)+\\d+$)(^(([a-z0-9]|[a-z0-9][a-z0-9\\-]*[a-z0-9])\\.)*([a-z0-9]|[a-z0-9][a-z0-9\\-]*[a-z0-9])$)|(^.{0}$))"
    ConstraintDescription: "Must respect S3 bucket naming conventions"

  ExistingDataBucket:
    Description: Does this data bucket already exist?
    Type: String
    AllowedValues:
      - "Yes"
      - "No"
    Default: "No"
  DefaultCEMinvCpus:
    Description: The minimum number of CPUs for the default Batch Compute Environment
    Type: Number
    Default: 0
  DefaultCEMaxvCpus:
    Description: The maximum number of CPUs for the default Batch Compute Environment
    Type: Number
    Default: 100
  PriorityCEMinvCpus:
    Description: The minimum number of CPUs for the high-priority Batch Compute Environment
    Type: Number
    Default: 0
  PriorityCEMaxvCpus:
    Description: The maximum number of CPUs for the high-priority Batch Compute Environment
    Type: Number
    Default: 100
  
  CromwellVersion:
    Type: String
    Description: >-
      Version of Cromwell to install. "latest" will retrieve the currently released
      version of Cromwell from Github.
    AllowedValues:
      - latest
      - specify ...
    Default: latest
  CromwellVersionSpecified:
    Type: String
    Description: >-
      Specific version of Cromwell to install.  Must match a released version number.
      For example, 52, 52.1, etc. The minimum supported version is 52. Ignored if "Cromwell Version" is set to "latest".
    Default: ""
  CromwellJarUrl:
    Type: String
    Description: >-
      URL to a pre-built cromwell-*.jar file.  Example: https://mycicdserver.com/build/cromwell-XX-SNAP.jar.
      If this is specifed, CromwellVersion is ignored.
    Default: ""
  DBSubnetIDs:
    Type: List<AWS::EC2::Subnet::Id>
    Description: Minimum 2 subnets for the Cromwell Database. Private subnets are recommended.
  DBUsername:
    Type: String
    Description: >-
      The master username for the Aurora MySQL RDS cluster that will be used as Cromwell's database
    Default: cromwell
    MaxLength: 16
    MinLength: 1
    AllowedPattern: "^[a-zA-Z]{1}[a-zA-Z0-9]{0,15}$"
    ConstraintDescription: >-
      DBUsername must have 1 to 16 alphanumeric characters. First character must be a letter
  DBPassword:
    Type: String
    NoEcho: true
    Description: >-
      The master password for the Aurora MySQL RDS cluster that will be used as Cromwell's database
    MinLength: 8
    AllowedPattern: '^[ -~][^"@/]{7,}$'
    ConstraintDescription: >-
      DBPassword must have at least 8 printable ASCII characters. Can't contain any of the following: / (slash), "(double quote) and @ (at sign).

  ArtifactBucketName:
    Type: String
    Default: aws-genomics-workflows
    Description: >-
      S3 Bucket where distrubution artifacts and additions scripts are stored
  ArtifactBucketPrefix:
    Type: String
    Default: artifacts  # dist: pin_version
    Description: >-
      Prefix in ArtifactBucketName where distribution artifacts and additions scripts are stored
  S3DataBucketARNs:
    Description: Open datasets on AWS S3 for workflow inputs
    Type: CommaDelimitedList
    Default: "arn:aws:s3:::gatk-test-data/*,arn:aws:s3:::broad-references/*"

  TemplateRootUrl:
    Type: String
    Description: >-
      Root URL for where nested templates are stored
    Default: https://aws-genomics-workflows.s3.amazonaws.com/templates
    ConstraintDescription: >-
      Must be a valid S3 URL
    AllowedPattern: "https://[a-z0-9-./]{0,}s3([a-z0-9-.]+)*\\.amazonaws\\.com/[a-z0-9-./]{3,}"


Conditions:
  NoNamespace: !Equals [ !Ref Namespace, "" ]

Resources:
  GenomicsWorkflowStack:
    Type: AWS::CloudFormation::Stack
    Properties:
      TemplateURL:  !Sub ${TemplateRootUrl}/gwfcore/gwfcore-root.template.yaml
      Parameters:
        Namespace:
          Fn::If:
            - NoNamespace
            - !Sub ${AWS::StackName}
            - !Ref Namespace
        VpcId: !Ref VpcId
        SubnetIds: !Join [",", !Ref SubnetIds ]
        S3BucketName: !Ref S3DataBucketName
        ExistingBucket: !Ref ExistingDataBucket
        DefaultCEMinvCpus: !Ref DefaultCEMinvCpus
        DefaultCEMaxvCpus: !Ref DefaultCEMaxvCpus
        PriorityCEMinvCpus: !Ref PriorityCEMinvCpus
        PriorityCEMaxvCpus: !Ref PriorityCEMaxvCpus
        ArtifactBucketName: !Ref ArtifactBucketName
        ArtifactBucketPrefix: !Ref ArtifactBucketPrefix
        TemplateRootUrl: !Ref TemplateRootUrl
      Tags: !FindInMap ["TagMap", "default", "tags"]

  CromwellStack:
    Type: AWS::CloudFormation::Stack
    DependsOn: GenomicsWorkflowStack
    Properties:
      TemplateURL:  !Sub ${TemplateRootUrl}/cromwell/cromwell-resources.template.yaml
      Parameters:
        Namespace: 
          Fn::If:
            - NoNamespace
            - !Sub ${AWS::StackName}-cromwell
            - !Sub ${Namespace}-cromwell
        GWFCoreNamespace:
          Fn::If:
            - NoNamespace
            - !Sub ${AWS::StackName}
            - !Ref Namespace
        VpcId: !Ref VpcId
        ServerSubnetID:  !Ref ServerSubnetID
        CromwellVersion: !Ref CromwellVersion
        CromwellVersionSpecified: !Ref CromwellVersionSpecified
        CromwellJarUrl: !Ref CromwellJarUrl
        S3DataBucketARNs: !Join [",", !Ref S3DataBucketARNs]
        DBUsername: !Ref DBUsername
        DBPassword: !Ref DBPassword
        DBSubnetIDs: !Join [",", !Ref DBSubnetIDs]
      Tags: !FindInMap ["TagMap", "default", "tags"]


Outputs:
  CromwellVersion:
    Value: !GetAtt CromwellStack.Outputs.CromwellVersion
    Export:
      Name: !Sub "${Namespace}-CromwellVersion"
    
  EC2Instance:
    Value: !GetAtt CromwellStack.Outputs.EC2Instance
    Export:
      Name: !Sub "${Namespace}-EC2Instance"
    Description: >-
      The EC2 Instance ID of your Cromwell Server
  
  PublicIp:
    Value: !GetAtt CromwellStack.Outputs.PublicIp
    Export:
      Name: !Sub "${Namespace}-PublicIp"
    Description: >-
      Cromwell server public IP address

  HostName:
    Value: !GetAtt CromwellStack.Outputs.HostName
    Export:
      Name: !Sub "${Namespace}-HostName"
    Description: >-
      Cromwell server public DNS name

  CloudWatchLogStream:
    Value: !GetAtt CromwellStack.Outputs.CloudWatchLogStream
    Export:
      Name: !Sub "${Namespace}-CloudWatchLogStream"
    Description: >-
      Cromwell server log stream
  
  VpcId:
    Description: >-
      The VPC created for your Nextflow stack.
    Value: !Ref VpcId
  S3DataBucket:
    Value: !GetAtt 'GenomicsWorkflowStack.Outputs.S3BucketName'
    Export:
      Name: !Sub "${Namespace}-DataBucket"
    Description: >-
      S3 bucket for storing genomics workflow input and output data
  BatchDefaultQueue:
    Value: !GetAtt 'GenomicsWorkflowStack.Outputs.DefaultJobQueueArn'
    Export:
      Name: !Sub "${Namespace}-DefaultJobQueue"
    Description: >-
      The default AWS Batch job queue for workflow jobs, based on EC2 SPOT instances
  BatchPriorityQueue:
    Value: !GetAtt 'GenomicsWorkflowStack.Outputs.PriorityJobQueueArn'
    Export:
      Name: !Sub "${Namespace}-PriorityJobQueue"
    Description: >-
      AWS Batch job queue for high priority workflow jobs, based on EC2 On-Demand
      instances
...