What should we include in our new version of Infracost JSON format?

[alikhajeh1]

We're thinking of updating our JSON format. We'd ❤️ to hear your feedback about the changes we have in mind (infracost breakdown --path=path/to/code --format json). These changes should make it simpler to write policies with tools like Open Policy Agent (OPA) or HashiCorp's Sentinel.

1. we'll bump the version from 0.2 to 0.3.

2. resources will have name, type, and address fields. Previously the name field contained the address. This change follows the convention in the Terraform plan JSON. For example:

   • { ... "name": "aws_instance.web_app" ...} will now be stored as: { ... "type": "aws_instance", "name": "web_app", "address": "aws_instance.web_app" ... }
   • { ... "name": "module.my_api.module.metric_alarm_target_4xx.aws_cloudwatch_metric_alarm.this[0]" ...} will now be stored as { ... "type": "aws_cloudwatch_metric_alarm", "name": "this[0]", "address": "module.my_api.module.metric_alarm_target_4xx.aws_cloudwatch_metric_alarm.this[0]" ... }

3. Fields containing prices, costs, and quantities are now numbers instead of strings.

4. When possible, cost component configuration details will be expanded as a key value list. { "name": "Instance usage (Linux/UNIX, on-demand, m5.4xlarge)" ...} will now be stored as { "name": "Instance usage", "parameters": { "platform": "Linux/UNIX", "purchase_option": "on-demand", "instance_type": "m5.4xlarge" } ... }

5. A JSON schema for the JSON output is now published at https://github.com/infracost/infracost/blob/master/schema/infracost.schema.json.

{
  "version": "0.3",
  "currency": "USD",
  "timeGenerated": "2021-05-31T18:02:49.412164+01:00",
  "projects": [
    {
      "name": "org/repo/subpath (stage)",
      "metadata": {
        "path": "path/to/code",
        "terraformWorkspace": "stage"
      },
      "pastBreakdown": {
        "resources": [
          {
            "address": "aws_instance.web_app",
            "type": "aws_instance",
            "name": "web_app",
            "metadata": {},
            "hourlyCost": 1.017315068493150679,
            "monthlyCost": 742.64,
            "costComponents": [
              {
                "name": "Instance usage",
                "parameters": { "platform": "Linux/UNIX", "purchase_option": "on-demand", "instance_type": "m5.4xlarge" },
                "unit": "hours",
                "hourlyQuantity": 1,
                "monthlyQuantity": 730,
                "price": 0.768,
                "hourlyCost": 0.768,
                "monthlyCost": 560.64
              }
            ]
          }
        ],
        "totalMonthlyCost": 730,
        "totalHourlyCost": 1
      },
      "breakdown": {
        "resources": [...],
        "totalMonthlyCost": 1460,
        "totalHourlyCost": 1
      },
      "diff": {
        "resources": [...],
        "totalMonthlyCost": 730,
        "totalHourlyCost": 1
      }
    }
  ],
  "totalHourlyCost": 1.017315068493150679,
  "totalMonthlyCost": 2920.64,
  "pastTotalHourlyCost": 0,
  "pastTotalMonthlyCost": 0,
  "diffTotalHourlyCost": 1.017315068493150679,
  "diffTotalMonthlyCost": 2920.64,
  "summary": {
    "totalDetectedResources": 2,
    "totalSupportedResources": 2,
    "totalUnsupportedResources": 0,
    "totalUsageBasedResources": 2,
    "totalNoPriceResources": 0,
    "unsupportedResourceCounts": {...},
    "noPriceResourceCounts": {...}
  }
}

Please let us know any ideas/improvements you want to see in our JSON format. What headaches has our old format caused that you want to see rectified?