Terraform Plan Parser

Terraform Plan Parser

Parse terraform plan -no-color plain-text output into structured data for programmatic analysis. Designed for TFC remote backends where -json and -out= are unavailable.

Parser location

The parser is a standalone Python module. Set PLAN_PARSER_DIR to its location:

# Example — adjust to your repo layout
export PLAN_PARSER_DIR="$HOME/your-repo/src/framework/parsers"

Import path:

import sys
sys.path.insert(0, os.environ["PLAN_PARSER_DIR"])
from terraform_plain_text_plan_parser import TerraformPlainTextPlanParser

Quick start

1. Capture plan output

PLAN_DIR=$(mktemp -d)
cd iac/{env}
terraform plan -no-color 2>&1 | tee "$PLAN_DIR/{env}-plan.txt"

Always use -no-color. The parser strips ANSI codes but clean input is faster.

2. Parse

from terraform_plain_text_plan_parser import TerraformPlainTextPlanParser

with open(f"{plan_dir}/dev-plan.txt") as f:
    result = TerraformPlainTextPlanParser(f.read()).parse()

Returns a dict with resource_changes list. Each entry:

{
    "address": "module.alb.aws_lb_listener.this[\"http\"]",
    "change": {
        "actions": ["update", "import"],  # or ["create"], ["delete"], ["import"]
        "before": { ... },   # null for create/import-only
        "after":  { ... },   # null for delete
    }
}

3. Classify changes

4. Detect real mutations

On import, many "changes" are just state population (before=null). Filter to real mutations where both before AND after have values:

for rc in result["resource_changes"]:
    if "update" not in rc["change"]["actions"]:
        continue
    before = rc["change"].get("before") or {}
    after = rc["change"].get("after") or {}
    for k in set(list(before) + list(after)):
        if k in ("tags", "tags_all"):
            continue
        bv, av = before.get(k), after.get(k)
        if bv is not None and av is not None and bv != av:
            print(f'{rc["address"]}: {k}: {bv} → {av}')

Import validation target

The goal for Terraform import workflows:

Plan: N to import, 0 to add, 0 to change, 0 to destroy.

• Destroys must be 0 — any destroy is a blocker.
• Adds: internal resources like terraform_data may be expected. Verify.
• Changes: all should be tag normalization or import state population. No real attribute mutations on load balancers, databases, etc.

Use the mutation filter above to prove "no real changes" vs "cosmetic import diffs".

Issue status template

When posting plan results to GitHub issues, use the template at resources/issue-status-template.md.

Known parser limitations

• Plain-text parser cannot distinguish (known after apply) computed values from actual changes. Use -json output (via TFC API plans/{id}/json-output) when available for precise before/after comparison.
• Nested block diffs (e.g. forward { stickiness { ... } }) are flattened into the parent resource's attribute map. Cross-reference with the raw plan text for nested block analysis.

Related skills

• tfc-api — get-plan.sh fetches plan logs; pipe output to this parser
• terraform-dev — inner-loop plan analysis during development

Source: shalomb/agent-skills — distributed by TomeVault.