From 045b665f1317ac3c285a3af36935abfbbe3a53b3 Mon Sep 17 00:00:00 2001
From: Andrew Zenk <andrew.zenk@particle.io>
Date: Thu, 6 Jun 2024 12:45:59 -0500
Subject: [PATCH] Add support for redaction and regex patterns

---
 modules/s3-logfile/main.tf      |  6 ++++--
 modules/s3-logfile/variables.tf | 18 ++++++++++++++++++
 2 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/modules/s3-logfile/main.tf b/modules/s3-logfile/main.tf
index 717ec46..9601612 100644
--- a/modules/s3-logfile/main.tf
+++ b/modules/s3-logfile/main.tf
@@ -54,7 +54,7 @@ module "s3_processor" {
   }
 
 
-  environment_variables = {
+  environment_variables = merge({
     PARSER_TYPE         = var.parser_type
     FORCE_GUNZIP        = true
     ENVIRONMENT         = var.environment
@@ -66,7 +66,9 @@ module "s3_processor" {
     SAMPLE_RATE_RULES   = jsonencode(var.sample_rate_rules)
     FILTER_FIELDS       = join(",", var.filter_fields)
     RENAME_FIELDS       = join(",", [for k, v in var.rename_fields : "${k}=${v}"])
-  }
+    REDACT_PATTERN      = var.redact_pattern
+    REGEX_PATTERN       = var.regex_pattern
+  }, var.extra_env_vars)
 
   attach_policy = true
   policy        = aws_iam_policy.lambda.arn
diff --git a/modules/s3-logfile/variables.tf b/modules/s3-logfile/variables.tf
index d36fa47..635a92e 100644
--- a/modules/s3-logfile/variables.tf
+++ b/modules/s3-logfile/variables.tf
@@ -93,6 +93,24 @@ variable "rename_fields" {
   default     = {}
 }
 
+variable "redact_pattern" {
+  type        = string
+  description = "A regex pattern to redact sensitive information from the request field. Default is to redact nothing. All groups in the pattern will be replaced with a corresponding number of 'x' characters."
+  default     = ""
+}
+
+variable "regex_pattern" {
+  description = "with `PARSER_TYPE=regex`, will define the regular expression to use for parsing each line in the log file. Honeycomb columns are generated by defining named capture groups. For example, `(?P<name>[a-z]+)` would create a column called 'name' if successfully parsed."
+  type        = string
+  default     = ""
+}
+
+variable "extra_env_vars" {
+  type        = map(string)
+  description = "Extra environment variables to set in the Lambda function."
+  default     = {}
+}
+
 variable "s3_bucket_arn" {
   type        = string
   description = "The full ARN of the bucket storing load balancer access logs."