Updated terraform configuration to initialize empty input and ouput d…

…irectories (datacommonsorg#4843)

New variables:

```
gcs_data_bucket_output_folder = "output"
gcs_data_bucket_input_folder = "input"
```

Renamed `dc_gcs_data_bucket_path_override` to `gcs_data_bucket_name`
Renamed `dc_gcs_data_bucket_location` to `gcs_data_bucket_location`

deploy/terraform-custom-datacommons/README.md

@@ -167,7 +167,7 @@ Once the deployment is complete, terraform should output something like:
 cloud_run_service_name = "<namespace>-datacommons-web-service"
 cloud_run_service_url = "https://<namespace>-datacommons-web-service-abc123-uc.a.run.app"
 dc_api_key = <sensitive>
-dc_gcs_data_bucket_path = "<namespace>-datacommons-data-<project-id>"
+gcs_data_bucket_name = "<namespace>-datacommons-data-<project-id>"
 maps_api_key = <sensitive>
 mysql_instance_connection_name = "<project-id>:us-central1:<namespace>-datacommons-mysql-instance"
 mysql_instance_public_ip = "<mysql_ip>"
@@ -178,7 +178,7 @@ redis_instance_port = 6379
 
 ### 6. Load custom data
 
-Upload custom sample data to the GCS bucket specified by the terraform output `dc_gcs_data_bucket_path` (`gs://<your-namespace>-datacommons-data-<your-project-id>`).
+Upload custom sample data to the GCS bucket specified by the terraform output `gcs_data_bucket_name` (`gs://<your-namespace>-datacommons-data-<your-project-id>`).
 From the `website` repository's root directory, run:
 
 ```

deploy/terraform-custom-datacommons/modules/locals.tf

@@ -16,7 +16,7 @@
 
 locals {
   # Data Commons Data Bucket
-  dc_gcs_data_bucket_path = var.dc_gcs_data_bucket_path_override != "" ? var.dc_gcs_data_bucket_path_override : "${var.namespace}-datacommons-data-${var.project_id}"
+  gcs_data_bucket_name = var.gcs_data_bucket_name != "" ? var.gcs_data_bucket_name : "${var.namespace}-datacommons-data-${var.project_id}"
   # VPC Connector CIDR block
   vpc_connector_cidr = cidrsubnet(var.vpc_base_cidr_block, 4, 0)  # Generates the first /28 subnet from the /24 block
 
@@ -56,7 +56,7 @@ locals {
     },
     {
       name  = "OUTPUT_DIR"
-      value = "gs://${local.dc_gcs_data_bucket_path}/output"
+      value = "gs://${local.gcs_data_bucket_name}/${var.gcs_data_bucket_output_folder}"
     },
     {
       name  = "FORCE_RESTART"

deploy/terraform-custom-datacommons/modules/main.tf

@@ -107,12 +107,26 @@ resource "google_sql_user" "mysql_user" {
 }
 
 # Data commons storage bucket
-resource "google_storage_bucket" "dc_gcs_data_bucket" {
-  name     = local.dc_gcs_data_bucket_path
-  location = var.dc_gcs_data_bucket_location
+resource "google_storage_bucket" "gcs_data_bucket" {
+  name     = local.gcs_data_bucket_name
+  location = var.gcs_data_bucket_location
   uniform_bucket_level_access = true
 }
 
+# Input 'folder' for the data loading job.
+resource "google_storage_bucket_object" "gcs_data_bucket_input_folder" {
+  name          = "${var.gcs_data_bucket_input_folder}/"
+  content       = "Input folder"
+  bucket        = "${google_storage_bucket.gcs_data_bucket.name}"
+}
+
+# Output 'folder' for the data loading job.
+resource "google_storage_bucket_object" "gcs_data_bucket_output_folder" {
+  name          = "${var.gcs_data_bucket_output_folder}/"
+  content       = "Output folder"
+  bucket        = "${google_storage_bucket.gcs_data_bucket.name}"
+}
+
 # Generate a random suffix to append to api keys.
 # A deleted API key fully expires 30 days after deletion, and in the 30-day
 # window the ID remains taken. This suffix allows terraform to give API
@@ -360,7 +374,7 @@ resource "google_cloud_run_v2_job" "dc_data_job" {
 
         env {
           name  = "INPUT_DIR"
-          value = "gs://${local.dc_gcs_data_bucket_path}/input"
+          value = "gs://${local.gcs_data_bucket_name}/${var.gcs_data_bucket_input_folder}"
         }
       }
       execution_environment = "EXECUTION_ENVIRONMENT_GEN2"

deploy/terraform-custom-datacommons/modules/outputs.tf

@@ -44,8 +44,8 @@ output "mysql_user_password" {
   sensitive   = true
 }
 
-output "dc_gcs_data_bucket_path" {
-  value = local.dc_gcs_data_bucket_path
+output "gcs_data_bucket_name" {
+  value = local.gcs_data_bucket_name
 }
 
 output "cloud_run_service_name" {

deploy/terraform-custom-datacommons/modules/variables.tf

@@ -67,13 +67,25 @@ variable "google_analytics_tag_id" {
 # Data Commons Cloud Storage bucket variables
 
 # If not set, the default is <namespace>-datacommons-data-<project_id>
-variable "dc_gcs_data_bucket_path_override" {
-  description = "Custom GCS data bucket path."
+variable "gcs_data_bucket_name" {
+  description = "Custom GCS data bucket name."
   type        = string
   default     = ""
 }
 
-variable "dc_gcs_data_bucket_location" {
+variable "gcs_data_bucket_input_folder" {
+  description = "Input data folder in the GCS data bucket"
+  type        = string
+  default     = "input"
+}
+
+variable "gcs_data_bucket_output_folder" {
+  description = "Output data folder in the GCS data bucket"
+  type        = string
+  default     = "output"
+}
+
+variable "gcs_data_bucket_location" {
   description = "Data Commons GCS data bucket location"
   type        = string
   default     = "US"