ckd8s-cdktf-resolver: can't resolve terraform values that are nested in a helm chart

[brent-at-aam]

Description of the bug:

This is kind of an odd one, but we are using cdk8s to synth dagster-user-deployment helm charts and running into an issue.

This particular chart allows you to pass env vars and configMaps/secrets to be referenced in the env and envFrom properties of ContainerSpec. There is also a toggle to pass this down to child k8s jobs that the dagster deployment creates. If that's enabled, it will take relevant entries from your chart values, convert them to json, and set it as a value for another env var.

 - name: DAGSTER_CONTAINER_CONTEXT
   value: {{ include "dagsterUserDeployments.k8sContainerContext" (list $ $deployment) | fromYaml | toJson | quote }}

If I pass a terraform output ref to an env var, for example an S3 bucket name. The chart synth fails with a message like:

# truncated for brevity
Error: Failed serializing construct at path '$PATH' with name '$NAME':
Error: Unable to find output defined for
{
  "k8s":{
    "env":[{"name":"S3_BUCKET","value":"${TfToken[TOKEN.40]}"},...],
    ...
  }
} (Inspected stacks: $STACK_0, $STACK_1) 

My assumption is that since the value is now just a json string, the ResovlerContext is getting the entire json string and can't match the token properly here

Reproduction Steps:

The dagster chart is not importable as it's getting caught in this bug cdklabs/json2jsii#1195, so we just use the Helm class directly:

from cdk8s import App as K8sApp
from cdk8s import Chart, Helm
from cdk8s_cdktf_resolver import CdktfResolver
from cdktf import App as TfApp
from cdktf import TerraformOutput, TerraformStack
from cdktf_cdktf_provider_aws.provider import AwsProvider
from cdktf_cdktf_provider_aws.s3_bucket import S3Bucket
from constructs import Construct


class ExampleStack(TerraformStack):
    bucket: TerraformOutput

    def __init__(self, scope: Construct, stack_id: str):
        super().__init__(scope, stack_id)
        AwsProvider(self, "aws", region="us-east-1")
        bucket = S3Bucket(self, "bucket", bucket="foo")
        self.bucket = TerraformOutput(self, "bucket_output", value=bucket.bucket)


class ExampleChart(Chart):
    def __init__(self, scope: Construct, chart_id: str, tf: ExampleStrack):
        super().__init__(scope, chart_id)
        Helm(
            self,
            "deployment",
            chart="dagster-user-deployments",
            repo="https://dagster-io.github.io/helm",
            version="1.9.6",
            namespace="dagster",
            release_name="foo-bar",
            values={
                "deployments": [
                    {
                        "name": "foo-bar",
                        "env":  [
                            { "name": "S3_BUCKET", "value": tf.bucket.value }
                        ],
                        "includeConfigInLaunchedRuns": { "enabled": True },
                        "port": 3030
                    }
                ],
                "serviceAccount": {"create": True, "name": "foo-bar" }
            }
        )


tf_app = TfApp()
stack = ExampleStack(app, "tf")
k8s_app = K8sApp(resolvers=[CdktfResolver(app=tf_app)]))
ExampleChart(k8ss_app, "k8s", stack)

k8s_app.synth()

Error Log:

Environment:

• Framework Version: 0.0.138
• OS: ubuntu 22.04

---

This is 🐛 Bug Report