Terraform Variable for Packer Manifest post-processor output

1_packer.pseudo-template.json

{
  "builders": [
    {
      "type": "digitalocean",
      ...
    }
  ],
  "post-processors":[
    {
      "type": "manifest",
      "output": "packer.auto.tfvars.json"
    }
  ]
}

2_packer.auto.tfvars.json

{
  "builds": [
    {
      "name": "digitalocean",
      "builder_type": "digitalocean",
      "build_time": 1589911059,
      "files": null,
      "artifact_id": "ams3:63876948",
      "packer_run_uuid": "41577b9e-4c85-ddef-a692-deaf6145783b",
      "custom_data": null
    },
    {
      "name": "digitalocean",
      "builder_type": "digitalocean",
      "build_time": 1589912902,
      "files": null,
      "artifact_id": "nyc1,nyc3,sgp1,lon1,fra1,tor1,sfo2,blr1,ams3:63877376",
      "packer_run_uuid": "94e2e639-4e9b-e806-5248-7a8d7dc7eca4",
      "custom_data": null
    }
  ],
  "last_run_uuid": "94e2e639-4e9b-e806-5248-7a8d7dc7eca4"
}

3_packer-vars.tf

// Define Terraform input variables for packer variables to allow their usage
variable "builds" {
  type = list(
    object(
      {
        name=string,
        builder_type=string,
        build_time=number,
        files=list(object({name=string, size=number})),
        artifact_id=string,
        packer_run_uuid=string
      }
    )
  )
  description = "List of images, as generated by Packer's 'Manifest' post-processor."
}
variable "last_run_uuid" {
  type = string
}

4_main.tf

// Accessing the artifact_id's image ID in a digitalocean_droplet resource
resource "digitalocean_droplet" "my-droplet" {
  image = split(":", var.builds[length(var.builds)-1].artifact_id)[1]
  ...
}

Note that this does not read the custom_data field, as we do not use it.

Additonally, it makes sense to specify the post-processor's output file as <some-name>.auto.tfvars.json in the directory you want to execute terraform in, as that automatically loads the variables for you (no need for -var-file )