CCM-13765: Store Files for Virus Scanning

infrastructure/terraform/components/dl/README.md

@@ -40,6 +40,7 @@ No requirements.
 
 | Name | Source | Version |
 |------|--------|---------|
+| <a name="module_file_scanner"></a> [file\_scanner](#module\_file\_scanner) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.29/terraform-lambda.zip | n/a |
 | <a name="module_kms"></a> [kms](#module\_kms) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-kms.zip | n/a |
 | <a name="module_lambda_apim_key_generation"></a> [lambda\_apim\_key\_generation](#module\_lambda\_apim\_key\_generation) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.29/terraform-lambda.zip | n/a |
 | <a name="module_lambda_lambda_apim_refresh_token"></a> [lambda\_lambda\_apim\_refresh\_token](#module\_lambda\_lambda\_apim\_refresh\_token) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.29/terraform-lambda.zip | n/a |
@@ -57,6 +58,7 @@ No requirements.
 | <a name="module_sqs_mesh_download"></a> [sqs\_mesh\_download](#module\_sqs\_mesh\_download) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
 | <a name="module_sqs_pdm_poll"></a> [sqs\_pdm\_poll](#module\_sqs\_pdm\_poll) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
 | <a name="module_sqs_pdm_uploader"></a> [sqs\_pdm\_uploader](#module\_sqs\_pdm\_uploader) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
+| <a name="module_sqs_scanner"></a> [sqs\_scanner](#module\_sqs\_scanner) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
 | <a name="module_sqs_ttl"></a> [sqs\_ttl](#module\_sqs\_ttl) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
 | <a name="module_sqs_ttl_handle_expiry_errors"></a> [sqs\_ttl\_handle\_expiry\_errors](#module\_sqs\_ttl\_handle\_expiry\_errors) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip | n/a |
 | <a name="module_ttl_create"></a> [ttl\_create](#module\_ttl\_create) | https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.29/terraform-lambda.zip | n/a |

infrastructure/terraform/components/dl/cloudwatch_event_rule_item_dequeued.tf

@@ -0,0 +1,19 @@
+resource "aws_cloudwatch_event_rule" "item_dequeued" {
+  name           = "${local.csi}-item-dequeued"
+  description    = "Queue item dequeued event rule"
+  event_bus_name = aws_cloudwatch_event_bus.main.name
+
+  event_pattern = jsonencode({
+    "detail" : {
+      "type" : [
+        "uk.nhs.notify.digital.letters.queue.item.dequeued.v1"
+      ]
+    }
+  })
+}
+
+resource "aws_cloudwatch_event_target" "item_dequeued_scanner" {
+  rule           = aws_cloudwatch_event_rule.item_dequeued.name
+  arn            = module.sqs_scanner.sqs_queue_arn
+  event_bus_name = aws_cloudwatch_event_bus.main.name
+}

infrastructure/terraform/components/dl/lambda_event_source_mapping_file_scanner.tf

@@ -0,0 +1,7 @@
+resource "aws_lambda_event_source_mapping" "file_scanner" {
+  event_source_arn                   = module.sqs_scanner.sqs_queue_arn
+  function_name                      = module.file_scanner.function_name
+  batch_size                         = 10
+  maximum_batching_window_in_seconds = 5
+  function_response_types            = ["ReportBatchItemFailures"]
+}

infrastructure/terraform/components/dl/locals.tf

@@ -5,6 +5,7 @@ locals {
   apim_api_key_ssm_parameter_name      = "/${var.component}/${var.environment}/apim/api_key"
   apim_private_key_ssm_parameter_name  = "/${var.component}/${var.environment}/apim/private_key"
   apim_keystore_s3_bucket              = "nhs-${var.aws_account_id}-${var.region}-${var.environment}-${var.component}-static-assets"
+  unscanned_files_bucket               = "nhs-${var.aws_account_id}-${var.region}-main-acct-digi-unscanned-files"
   ssm_mesh_prefix                      = "/${var.component}/${var.environment}/mesh"
   mock_mesh_endpoint                   = "s3://${module.s3bucket_non_pii_data.bucket}/mock-mesh"
   root_domain_name                     = "${var.environment}.${local.acct.route53_zone_names["digital-letters"]}"

infrastructure/terraform/components/dl/module_lambda_file_scanner.tf

@@ -0,0 +1,131 @@
+module "file_scanner" {
+  source = "https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.29/terraform-lambda.zip"
+
+  function_name = "file-scanner"
+  description   = "A function for extracting PDFs from DocumentReference and storing in UnscannedFiles bucket for virus scanning"
+
+  aws_account_id = var.aws_account_id
+  component      = local.component
+  environment    = var.environment
+  project        = var.project
+  region         = var.region
+  group          = var.group
+
+  log_retention_in_days = var.log_retention_in_days
+  kms_key_arn           = module.kms.key_arn
+
+  iam_policy_document = {
+    body = data.aws_iam_policy_document.file_scanner_lambda.json
+  }
+
+  function_s3_bucket      = local.acct.s3_buckets["lambda_function_artefacts"]["id"]
+  function_code_base_path = local.aws_lambda_functions_dir_path
+  function_code_dir       = "file-scanner-lambda/dist"
+  function_include_common = true
+  handler_function_name   = "handler"
+  runtime                 = "nodejs22.x"
+  memory                  = 512
+  timeout                 = 60
+  log_level               = var.log_level
+
+  force_lambda_code_deploy = var.force_lambda_code_deploy
+  enable_lambda_insights   = false
+
+  log_destination_arn       = local.log_destination_arn
+  log_subscription_role_arn = local.acct.log_subscription_role_arn
+
+  lambda_env_vars = {
+    "DOCUMENT_REFERENCE_BUCKET"    = module.s3bucket_pii_data.bucket
+    "UNSCANNED_FILES_BUCKET"       = local.unscanned_files_bucket
+    "UNSCANNED_FILES_PATH_PREFIX"  = var.environment
+    "EVENT_PUBLISHER_EVENT_BUS_ARN" = aws_cloudwatch_event_bus.main.arn
+    "EVENT_PUBLISHER_DLQ_URL"      = module.sqs_event_publisher_errors.sqs_queue_url
+  }
+}
+
+data "aws_iam_policy_document" "file_scanner_lambda" {
+  statement {
+    sid    = "ReadDocumentReferenceBucket"
+    effect = "Allow"
+
+    actions = [
+      "s3:GetObject",
+      "s3:ListBucket",
+    ]
+
+    resources = [
+      module.s3bucket_pii_data.arn,
+      "${module.s3bucket_pii_data.arn}/*",
+    ]
+  }
+
+  statement {
+    sid    = "WriteUnscannedFilesBucket"
+    effect = "Allow"
+
+    actions = [
+      "s3:PutObject",
+    ]
+
+    resources = [
+      "arn:aws:s3:::${local.unscanned_files_bucket}/*",
+    ]
+  }
+
+  statement {
+    sid    = "PutEvents"
+    effect = "Allow"
+
+    actions = [
+      "events:PutEvents",
+    ]
+
+    resources = [
+      aws_cloudwatch_event_bus.main.arn,
+    ]
+  }
+
+  statement {
+    sid    = "SQSPermissionsDLQs"
+    effect = "Allow"
+
+    actions = [
+      "sqs:SendMessage",
+      "sqs:SendMessageBatch",
+    ]
+
+    resources = [
+      module.sqs_event_publisher_errors.sqs_queue_arn,
+    ]
+  }
+
+  statement {
+    sid    = "SQSPermissionsScannerQueue"
+    effect = "Allow"
+
+    actions = [
+      "sqs:ReceiveMessage",
+      "sqs:DeleteMessage",
+      "sqs:GetQueueAttributes",
+      "sqs:GetQueueUrl",
+    ]
+
+    resources = [
+      module.sqs_scanner.sqs_queue_arn,
+    ]
+  }
+
+  statement {
+    sid    = "KMSPermissions"
+    effect = "Allow"
+
+    actions = [
+      "kms:Decrypt",
+      "kms:GenerateDataKey",
+    ]
+
+    resources = [
+      module.kms.key_arn,
+    ]
+  }
+}

infrastructure/terraform/components/dl/module_sqs_scanner.tf

@@ -0,0 +1,35 @@
+module "sqs_scanner" {
+  source = "https://github.com/NHSDigital/nhs-notify-shared-modules/releases/download/v2.0.24/terraform-sqs.zip"
+
+  aws_account_id             = var.aws_account_id
+  component                  = local.component
+  environment                = var.environment
+  project                    = var.project
+  region                     = var.region
+  name                       = "scanner"
+  sqs_kms_key_arn            = module.kms.key_arn
+  visibility_timeout_seconds = 60
+  delay_seconds              = 0
+  create_dlq                 = true
+  sqs_policy_overload        = data.aws_iam_policy_document.sqs_scanner.json
+}
+
+data "aws_iam_policy_document" "sqs_scanner" {
+  statement {
+    sid    = "AllowEventBridgeToSendMessage"
+    effect = "Allow"
+
+    principals {
+      type        = "Service"
+      identifiers = ["events.amazonaws.com"]
+    }
+
+    actions = [
+      "sqs:SendMessage"
+    ]
+
+    resources = [
+      "arn:aws:sqs:${var.region}:${var.aws_account_id}:${local.csi}-scanner-queue"
+    ]
+  }
+}

lambdas/file-scanner-lambda/jest.config.ts

@@ -0,0 +1,8 @@
+import { baseJestConfig } from '../../jest.config.base';
+
+const config = {
+  ...baseJestConfig,
+  displayName: 'file-scanner-lambda',
+};
+
+export default config;

lambdas/file-scanner-lambda/package.json

@@ -0,0 +1,26 @@
+{
+  "dependencies": {
+    "@aws-sdk/client-s3": "^3.908.0",
+    "aws-lambda": "^1.0.7",
+    "digital-letters-events": "^0.0.1",
+    "utils": "^0.0.1"
+  },
+  "devDependencies": {
+    "@tsconfig/node22": "^22.0.2",
+    "@types/aws-lambda": "^8.10.155",
+    "@types/jest": "^29.5.14",
+    "jest": "^29.7.0",
+    "jest-mock-extended": "^3.0.7",
+    "typescript": "^5.9.3"
+  },
+  "name": "nhs-notify-digital-letters-file-scanner-lambda",
+  "private": true,
+  "scripts": {
+    "lambda-build": "rm -rf dist && npx esbuild --bundle --minify --sourcemap --target=es2020 --platform=node --loader:.node=file --entry-names=[name] --outdir=dist src/index.ts",
+    "lint": "eslint .",
+    "lint:fix": "eslint . --fix",
+    "test:unit": "jest",
+    "typecheck": "tsc --noEmit"
+  },
+  "version": "0.0.1"
+}