shoc-frontend-new/scripts/test-terraform-import-plan-check.py

#!/usr/bin/env python3
"""Deterministic unit tests for the frontend Terraform plan checker."""

from __future__ import annotations

import copy
import json
import re
import subprocess
import sys
import tempfile
import unittest
from pathlib import Path
from typing import Any

from terraform_import_plan_resources import (
    CONTROLLED_UPDATE_ADDRESSES,
    ENVIRONMENT_CONFIG,
    REQUIRED_IMPORT_IDS,
    REQUIRED_RESOURCES,
)

SCRIPT = Path(__file__).with_name("check-terraform-import-plan.py")
REPOSITORY = SCRIPT.parent.parent
BUCKET_POLICY = "module.environment_owned.aws_s3_bucket_policy.site"
BUCKET = "module.environment_owned.aws_s3_bucket.site"
DEPLOY_POLICY = "module.environment_owned.aws_iam_role_policy.github_deploy"
ROLE = "module.environment_owned.aws_iam_role.github_deploy"
DISTRIBUTION = "module.environment_owned.aws_cloudfront_distribution.site"
TAG_ADDRESSES = CONTROLLED_UPDATE_ADDRESSES - {BUCKET_POLICY, DEPLOY_POLICY}


def import_id(environment: str, address: str) -> str:
    expected = REQUIRED_IMPORT_IDS[environment][address]
    assert expected is not None, f"{environment} must pin an import ID for {address}"
    return expected


def distribution_id(environment: str) -> str:
    configured = ENVIRONMENT_CONFIG[environment]["distribution_id"]
    assert isinstance(configured, str), f"{environment} must pin a distribution ID"
    return configured


def pre_adoption_bucket_policy(environment: str) -> dict[str, Any]:
    config = ENVIRONMENT_CONFIG[environment]
    bucket_arn = f"arn:aws:s3:::{config['bucket_name']}"
    source = (
        "arn:aws:cloudfront::396287094661:distribution/"
        f"{distribution_id(environment)}"
    )
    return {
        "Version": "2012-10-17",
        "Statement": [
            {
                "Effect": "Allow",
                "Principal": {
                    "AWS": config["bucket_auto_delete_helper_role_arn"]
                },
                "Action": [
                    "s3:DeleteObject*",
                    "s3:GetBucket*",
                    "s3:List*",
                    "s3:PutBucketPolicy",
                ],
                "Resource": [bucket_arn, f"{bucket_arn}/*"],
            },
            {
                "Effect": "Allow",
                "Principal": {"Service": "cloudfront.amazonaws.com"},
                "Action": "s3:GetObject",
                "Resource": f"{bucket_arn}/*",
                "Condition": {"StringEquals": {"AWS:SourceArn": source}},
            },
            {
                "Effect": "Deny",
                "Principal": {"AWS": "*"},
                "Action": "s3:*",
                "Resource": [bucket_arn, f"{bucket_arn}/*"],
                "Condition": {"Bool": {"aws:SecureTransport": "false"}},
            },
        ],
    }


def bucket_policy(environment: str) -> dict[str, Any]:
    bucket = ENVIRONMENT_CONFIG[environment]["bucket_name"]
    bucket_arn = f"arn:aws:s3:::{bucket}"
    source = (
        "arn:aws:cloudfront::396287094661:distribution/"
        f"{distribution_id(environment)}"
    )
    return {
        "Version": "2012-10-17",
        "Statement": [
            {
                "Effect": "Allow",
                "Principal": {"Service": "cloudfront.amazonaws.com"},
                "Action": "s3:GetObject",
                "Resource": f"{bucket_arn}/*",
                "Condition": {"StringEquals": {"AWS:SourceArn": source}},
            },
            {
                "Effect": "Deny",
                "Principal": {"AWS": "*"},
                "Action": "s3:*",
                "Resource": [bucket_arn, f"{bucket_arn}/*"],
                "Condition": {"Bool": {"aws:SecureTransport": "false"}},
            },
        ],
    }


def pre_adoption_deploy_policy(environment: str) -> dict[str, Any]:
    config = ENVIRONMENT_CONFIG[environment]
    bucket_arn = f"arn:aws:s3:::{config['bucket_name']}"
    distribution_arn = (
        "arn:aws:cloudfront::396287094661:distribution/"
        f"{distribution_id(environment)}"
    )
    statements: list[dict[str, Any]] = []
    if environment == "dev":
        statements.append(
            {
                "Sid": "AssumeCdkBootstrapRoles",
                "Effect": "Allow",
                "Action": "sts:AssumeRole",
                "Resource": "arn:aws:iam::396287094661:role/cdk-hnb659fds-*",
            }
        )
    statements.extend(
        [
            {
                "Sid": "DescribeStack",
                "Effect": "Allow",
                "Action": "cloudformation:DescribeStacks",
                "Resource": (
                    "arn:aws:cloudformation:us-east-1:396287094661:stack/"
                    f"{config['cloudformation_stack_name']}/*"
                ),
            },
            {
                "Effect": "Allow",
                "Action": [
                    "s3:Abort*",
                    "s3:DeleteObject*",
                    "s3:GetBucket*",
                    "s3:GetObject*",
                    "s3:List*",
                    "s3:PutObject",
                    "s3:PutObjectLegalHold",
                    "s3:PutObjectRetention",
                    "s3:PutObjectTagging",
                    "s3:PutObjectVersionTagging",
                ],
                "Resource": [bucket_arn, f"{bucket_arn}/*"],
            },
            {
                "Sid": "InvalidateDistribution",
                "Effect": "Allow",
                "Action": [
                    "cloudfront:CreateInvalidation",
                    "cloudfront:GetInvalidation",
                ],
                "Resource": distribution_arn,
            },
        ]
    )
    return {"Version": "2012-10-17", "Statement": statements}


def deploy_policy(environment: str) -> dict[str, Any]:
    bucket = ENVIRONMENT_CONFIG[environment]["bucket_name"]
    bucket_arn = f"arn:aws:s3:::{bucket}"
    distribution_arn = (
        "arn:aws:cloudfront::396287094661:distribution/"
        f"{distribution_id(environment)}"
    )
    return {
        "Version": "2012-10-17",
        "Statement": [
            {
                "Sid": "ReadDeploymentBucket",
                "Effect": "Allow",
                "Action": [
                    "s3:GetBucketLocation",
                    "s3:GetBucketVersioning",
                    "s3:ListBucket",
                    "s3:ListBucketVersions",
                ],
                "Resource": bucket_arn,
            },
            {
                "Sid": "PublishAndRollbackSiteObjects",
                "Effect": "Allow",
                "Action": [
                    "s3:DeleteObject",
                    "s3:DeleteObjectVersion",
                    "s3:GetObject",
                    "s3:GetObjectVersion",
                    "s3:PutObject",
                ],
                "Resource": f"{bucket_arn}/*",
            },
            {
                "Sid": "InvalidateDistribution",
                "Effect": "Allow",
                "Action": [
                    "cloudfront:CreateInvalidation",
                    "cloudfront:GetInvalidation",
                ],
                "Resource": distribution_arn,
            },
        ],
    }


def tag_change(environment: str, address: str) -> dict[str, Any]:
    manager = {
        "HcpTerraformWorkspace": ENVIRONMENT_CONFIG[environment]["workspace_name"]
    }
    before_tags = {
        "Environment": environment,
        "ManagedBy": "cdk",
        "Project": "shoc-frontend",
    }
    after_tags = {
        "Environment": environment,
        "ManagedBy": "terraform",
        "Ownership": "terraform",
        "Project": "shoc-frontend",
    }
    if address == ROLE:
        before_tags.update(manager)
        after_tags.update(manager)
    if address == BUCKET:
        before_tags["aws-cdk:auto-delete-objects"] = "true"
    before: dict[str, Any] = {
        "tags": before_tags,
        "tags_all": before_tags,
    }
    after: dict[str, Any] = {
        "tags": after_tags,
        "tags_all": after_tags,
    }
    if address == DISTRIBUTION:
        before["id"] = distribution_id(environment)
        after["id"] = distribution_id(environment)
    return {"actions": ["update"], "before": before, "after": after}


def policy_change(environment: str, address: str) -> dict[str, Any]:
    before_policy = (
        pre_adoption_bucket_policy(environment)
        if address == BUCKET_POLICY
        else pre_adoption_deploy_policy(environment)
    )
    after_policy = (
        bucket_policy(environment)
        if address == BUCKET_POLICY
        else deploy_policy(environment)
    )
    return {
        "actions": ["update"],
        "before": {"policy": json.dumps(before_policy)},
        "after": {"policy": json.dumps(after_policy)},
    }


def make_plan(
    environment: str,
    *,
    mode: str = "import",
    controlled_updates: set[str] | None = None,
) -> dict[str, Any]:
    resources: list[dict[str, Any]] = []
    updates = controlled_updates or set()
    for address, resource_type in REQUIRED_RESOURCES[environment].items():
        if mode == "import":
            change: dict[str, Any] = {
                "actions": ["no-op"],
                "importing": {"id": import_id(environment, address)},
            }
        elif mode == "post-import":
            change = {"actions": ["no-op"]}
        elif address in updates:
            change = (
                tag_change(environment, address)
                if address in TAG_ADDRESSES
                else policy_change(environment, address)
            )
        else:
            change = {"actions": ["no-op"]}
            if address == DISTRIBUTION:
                change["after"] = {"id": distribution_id(environment)}
        resources.append(
            {
                "address": address,
                "mode": "managed",
                "type": resource_type,
                "change": change,
            }
        )
    return {"resource_changes": resources}


def resource(plan: dict[str, Any], address: str) -> dict[str, Any]:
    return next(
        item for item in plan["resource_changes"] if item["address"] == address
    )


def run_checker(
    plan: dict[str, Any],
    environment: str,
    *allowed_updates: str,
    post_import: bool = False,
) -> subprocess.CompletedProcess[str]:
    with tempfile.TemporaryDirectory() as directory:
        path = Path(directory) / "plan.json"
        path.write_text(json.dumps(plan), encoding="utf-8")
        command = [
            sys.executable,
            str(SCRIPT),
            str(path),
            "--environment",
            environment,
        ]
        if post_import:
            command.append("--post-import-no-op")
        for address in allowed_updates:
            command.extend(["--allow-update-address", address])
        return subprocess.run(
            command,
            check=False,
            capture_output=True,
            text=True,
        )


class ImportPlanCheckerTests(unittest.TestCase):
    def assert_passes(
        self,
        plan: dict[str, Any],
        environment: str,
        *allowed_updates: str,
        post_import: bool = False,
    ) -> None:
        result = run_checker(
            plan,
            environment,
            *allowed_updates,
            post_import=post_import,
        )
        self.assertEqual(0, result.returncode, result.stdout + result.stderr)

    def assert_fails(
        self,
        plan: dict[str, Any],
        environment: str,
        *allowed_updates: str,
        post_import: bool = False,
    ) -> None:
        result = run_checker(
            plan,
            environment,
            *allowed_updates,
            post_import=post_import,
        )
        self.assertNotEqual(0, result.returncode, result.stdout + result.stderr)

    def test_cloudfront_function_source_matches_exact_nine_line_join(self) -> None:
        source = (
            REPOSITORY
            / "terraform/live/modules/environment-owned/main.tf"
        ).read_text(encoding="utf-8")
        expected = """  spa_rewrite_code = join("\\n", [
    "function handler(event) {",
    "  var request = event.request;",
    "  var uri = request.uri;",
    "  // No file extension after the last slash -> a client-side route.",
    "  if (uri.lastIndexOf('.') <= uri.lastIndexOf('/')) {",
    "    request.uri = '/index.html';",
    "  }",
    "  return request;",
    "}",
  ])"""
        self.assertIn(expected, source)

    def test_only_dev_has_a_live_root(self) -> None:
        live_roots = sorted(
            path.name
            for path in (REPOSITORY / "terraform/live").iterdir()
            if path.is_dir() and path.name != "modules"
        )
        self.assertEqual(["dev"], live_roots)

    def test_dev_root_pins_import_phase_in_code(self) -> None:
        source = (REPOSITORY / "terraform/live/dev/main.tf").read_text(encoding="utf-8")
        self.assertRegex(source, r"\n\s+adoption_complete\s+= false\n")
        self.assertRegex(source, r"adoption_complete\s+= local\.adoption_complete")
        self.assertNotIn('variable "adoption_complete"', source)
        for root_file in ("main.tf", "imports.tf", "outputs.tf", "providers.tf", "versions.tf"):
            self.assertNotIn(
                "variable ",
                (REPOSITORY / f"terraform/live/dev/{root_file}").read_text(encoding="utf-8"),
                root_file,
            )

    def test_managed_modules_use_direct_pinned_inputs(self) -> None:
        expected = {
            "dev": (
                "local.hosted_zone_id",
                "local.certificate_arn",
                "local.github_oidc_arn",
                "local.cache_policy_id",
            ),
        }
        for environment, values in expected.items():
            source = (
                REPOSITORY / f"terraform/live/{environment}/main.tf"
            ).read_text(encoding="utf-8")
            for name, value in zip(
                (
                    "hosted_zone_id",
                    "certificate_arn",
                    "github_oidc_provider_arn",
                    "cache_policy_id",
                ),
                values,
                strict=True,
            ):
                self.assertIn(f"{name}", source)
                self.assertRegex(source, rf"{name}\s+= {re.escape(value)}")
            self.assertNotRegex(
                source,
                r"(hosted_zone_id|certificate_arn|github_oidc_provider_arn|cache_policy_id)\s+= module\.inventory",
            )

    def test_exact_import_plan_passes_for_every_environment(self) -> None:
        for environment in REQUIRED_RESOURCES:
            with self.subTest(environment=environment):
                self.assert_passes(make_plan(environment), environment)

    def test_import_missing_extra_wrong_type_and_cross_environment_fail(self) -> None:
        for mutation in ("missing", "extra", "wrong-type", "cross-environment"):
            plan = make_plan("dev")
            if mutation == "missing":
                plan["resource_changes"].pop()
            elif mutation == "extra":
                plan["resource_changes"].append(
                    {
                        "address": "module.inventory.aws_route53_zone.site",
                        "mode": "managed",
                        "type": "aws_route53_zone",
                        "change": {
                            "actions": ["no-op"],
                            "importing": {"id": "Z00000000000000000000"},
                        },
                    }
                )
            elif mutation == "wrong-type":
                plan["resource_changes"][0]["type"] = "aws_s3_object"
            else:
                resource(plan, DISTRIBUTION)["change"]["importing"]["id"] = (
                    REQUIRED_IMPORT_IDS["staging"][DISTRIBUTION]
                )
            with self.subTest(mutation=mutation):
                self.assert_fails(plan, "dev")

    def test_import_rejects_mutation_and_invalid_metadata(self) -> None:
        for actions in (["create"], ["update"], ["delete"], ["delete", "create"]):
            plan = make_plan("dev")
            plan["resource_changes"][0]["change"]["actions"] = actions
            with self.subTest(actions=actions):
                self.assert_fails(plan, "dev")
        plan = make_plan("dev")
        plan["resource_changes"][0]["change"]["importing"] = {"id": ""}
        self.assert_fails(plan, "dev")

    def test_post_import_no_op_passes(self) -> None:
        self.assert_passes(
            make_plan("staging", mode="post-import"),
            "staging",
            post_import=True,
        )

    def test_post_import_rejects_import_metadata_and_update(self) -> None:
        plan = make_plan("dev", mode="post-import")
        plan["resource_changes"][0]["change"]["importing"] = {"id": "unexpected"}
        self.assert_fails(plan, "dev", post_import=True)
        plan = make_plan("dev", mode="post-import")
        plan["resource_changes"][0]["change"]["actions"] = ["update"]
        self.assert_fails(plan, "dev", post_import=True)

    def test_every_allowed_controlled_diff_passes(self) -> None:
        for environment in REQUIRED_RESOURCES:
            for address in CONTROLLED_UPDATE_ADDRESSES:
                with self.subTest(environment=environment, address=address):
                    self.assert_passes(
                        make_plan(
                            environment,
                            mode="controlled",
                            controlled_updates={address},
                        ),
                        environment,
                        address,
                    )

    def test_full_exact_controlled_allowlist_passes(self) -> None:
        addresses = tuple(sorted(CONTROLLED_UPDATE_ADDRESSES))
        self.assert_passes(
            make_plan(
                "dev",
                mode="controlled",
                controlled_updates=set(addresses),
            ),
            "dev",
            *addresses,
        )

    def test_tag_update_rejects_extra_attribute_and_wrong_value(self) -> None:
        plan = make_plan("dev", mode="controlled", controlled_updates={ROLE})
        resource(plan, ROLE)["change"]["after"]["assume_role_policy"] = "{}"
        self.assert_fails(plan, "dev", ROLE)
        plan = make_plan("dev", mode="controlled", controlled_updates={ROLE})
        resource(plan, ROLE)["change"]["after"]["tags"]["ManagedBy"] = "attacker"
        self.assert_fails(plan, "dev", ROLE)

    def test_tag_update_requires_complete_adopted_tag_sets(self) -> None:
        plan = make_plan("dev", mode="controlled", controlled_updates={BUCKET})
        del resource(plan, BUCKET)["change"]["after"]["tags"]["Ownership"]
        self.assert_fails(plan, "dev", BUCKET)

    def test_role_trust_change_is_rejected(self) -> None:
        plan = make_plan("dev", mode="controlled", controlled_updates={ROLE})
        role = resource(plan, ROLE)["change"]
        role["before"]["assume_role_policy"] = '{"Statement":[]}'
        role["after"]["assume_role_policy"] = '{"Statement":[{"Effect":"Allow"}]}'
        self.assert_fails(plan, "dev", ROLE)

    def test_bucket_policy_rejects_malicious_principal_and_extra_statement(self) -> None:
        for mutation in ("principal", "extra"):
            plan = make_plan(
                "dev",
                mode="controlled",
                controlled_updates={BUCKET_POLICY},
            )
            policy = copy.deepcopy(bucket_policy("dev"))
            if mutation == "principal":
                policy["Statement"][0]["Principal"] = {"AWS": "*"}
            else:
                policy["Statement"].append(
                    {
                        "Effect": "Allow",
                        "Principal": {"AWS": "*"},
                        "Action": "s3:*",
                        "Resource": "*",
                    }
                )
            resource(plan, BUCKET_POLICY)["change"]["after"]["policy"] = json.dumps(
                policy
            )
            with self.subTest(mutation=mutation):
                self.assert_fails(plan, "dev", BUCKET_POLICY)

    def test_deploy_policy_rejects_resource_action_and_extra_statement(self) -> None:
        for mutation in ("resource", "action", "extra"):
            plan = make_plan(
                "staging",
                mode="controlled",
                controlled_updates={DEPLOY_POLICY},
            )
            policy = copy.deepcopy(deploy_policy("staging"))
            if mutation == "resource":
                policy["Statement"][0]["Resource"] = "*"
            elif mutation == "action":
                policy["Statement"][0]["Action"].append("iam:PassRole")
            else:
                policy["Statement"].append(
                    {
                        "Sid": "Extra",
                        "Effect": "Allow",
                        "Action": "s3:*",
                        "Resource": "*",
                    }
                )
            resource(plan, DEPLOY_POLICY)["change"]["after"]["policy"] = json.dumps(
                policy
            )
            with self.subTest(mutation=mutation):
                self.assert_fails(plan, "staging", DEPLOY_POLICY)

    def test_policy_updates_require_exact_pre_adoption_state(self) -> None:
        for environment in REQUIRED_RESOURCES:
            for address in (BUCKET_POLICY, DEPLOY_POLICY):
                plan = make_plan(
                    environment,
                    mode="controlled",
                    controlled_updates={address},
                )
                change = resource(plan, address)["change"]
                before = json.loads(change["before"]["policy"])
                before["Statement"].append(
                    {
                        "Sid": "UnexpectedDrift",
                        "Effect": "Deny",
                        "Action": "*",
                        "Resource": "*",
                    }
                )
                change["before"]["policy"] = json.dumps(before)
                with self.subTest(environment=environment, address=address):
                    self.assert_fails(plan, environment, address)

    def test_controlled_update_rejects_unknown_and_replace_paths(self) -> None:
        for field, value in (
            ("after_unknown", {"tags": {"ManagedBy": True}}),
            ("replace_paths", [["tags"]]),
        ):
            plan = make_plan(
                "dev",
                mode="controlled",
                controlled_updates={ROLE},
            )
            resource(plan, ROLE)["change"][field] = value
            with self.subTest(field=field):
                self.assert_fails(plan, "dev", ROLE)

    def test_nonallowlisted_update_and_unused_allowlist_fail(self) -> None:
        plan = make_plan("dev", mode="controlled", controlled_updates={ROLE})
        self.assert_fails(plan, "dev", BUCKET_POLICY)
        plan = make_plan("dev", mode="controlled", controlled_updates=set())
        self.assert_fails(plan, "dev", ROLE)


if __name__ == "__main__":
    unittest.main()