Terraform variable pointing to current file name - terraform

Question

Is there some special variable available in Terraform configuration files which would point to current file name?
I'd like to use it for description fields in various resources, so that someone seeing these resources in the systems would know where is the master definition for them.
e.g.
in myinfra.tf
resource "aws_iam_policy" "my_policy" {
name = "something-important"
description = "Managed by Terraform at ${HERE_I_WOULD_LIKE_TO_USE_THE_VARIABLE}"
policy = <<EOF
[...]
EOF
}
And I would hope the description becomes:
description = "Managed by Terraform at myinfra.tf"
I tried ${path.module} but that only gives "filesystem path of the module where the expression is placed", so pragmatically speaking - everything but the file name I want.

Answer 1

Here's what I can share. Use the data external resource to call an external script that would get the directory/file name and then return it back as a string or any other type that your resources require. Obviously it's not exactly what you wanted as you'll get the dir/file name indirectly but hopefully it helps for others or even yourself for use-cases.
We use that only for azurerm and for very complex integrations that are not yet supported with the current provider versions. I have have not tested it specifically for AWS but since it's a core Terraform resource provider, I'm guessing it might work across the board.
data "external" "cwd" {
program = ["./script.sh"]
query = {
cwd = "${path.cwd}"
}
}
resource "aws_iam_policy" "my_policy" {
name = "something-important"
description = "Managed by Terraform at ${data.external.dir_script.result.filename}"
policy = <<EOF
[...]
EOF
This is how my script looks like:
#!/bin/sh
#echo '{"cwd":"for_testing"}' | ./dir_name.sh | xargs
PIPED=`cat`
errPrint "INFO: Got PIPED data:\n$PIPED"
DIR=`jq -r .cwd <<< $PIPED`
cd $DIR
filename=`ls | grep \.tf$ | xargs`
errPrint "INFO: Returning this as STDOUT:${filename}"
echo "{\"name\":\"$filename\"}"
You need to be that the data from the script needs to return a valid JSON object.
The program must then produce a valid JSON object on stdout, which will be used to populate the result attribute exported to the rest of the Terraform configuration. This JSON object must again have all of its values as strings. On successful completion it must exit with status zero.
Unfortunately, like the others mentioned, there's no other way to get the current file name being 'applied'.

Answer 2

I think you might benefit from using something like yor from Bridge Crew.
From the project's README:
Yor is an open-source tool that helps add informative and consistent tags across infrastructure-as-code frameworks such as Terraform, CloudFormation, and Serverless.
Yor is built to run as a GitHub Action automatically adding consistent tagging logics to your IaC. Yor can also run as a pre-commit hook and a standalone CLI.
So basically, it updates your resources tags with things like:
tags = {
env = var.env
yor_trace = "912066a1-31a3-4a08-911b-0b06d9eac64e"
git_repo = "example"
git_org = "bridgecrewio"
git_file = "applyTag.md"
git_commit = "COMMITHASH"
git_modifiers = "bana/gandalf"
git_last_modified_at = "2021-01-08 00:00:00"
git_last_modified_by = "bana#bridgecrew.io"
}
Maybe that would be good enough to provide what you're trying to do?
As far as my testimony, I have not used yor since my tagging uses a different approach. Instead of having "raw" tags, we use a label module that builds the tags for us and then merges in local tags.
Just sharing this info FYI in case it helps.