Service-to-Service Transform with CSV Response

Use this pattern when a JSON payload must be sent to an HTTP service that returns CSV — for example, a pricing or reporting service used by a legacy system.

Input: json → Output: csv

Steps used: validate-json, http-transform, select, count

Pipeline

id: pricing-sync

input:
  format: json

steps:
  - validate-json:
      schema: ./schemas/pricing-request.schema.json

  - http-transform:
      url: https://api.example.com/v1/pricing/export
      method: POST
      timeout: 30
      headers:
        Authorization: Bearer ${PRICING_TOKEN}
        X-Source-System: pipectl
      expect-format: csv

  - select:
      fields: [sku, region, price, currency]

  - count:
      message: Rows returned from pricing service

output:
  format: csv

Sample input

{
  "account_id": "acct-123",
  "effective_date": "2026-03-22",
  "items": [
    {"sku": "SKU-001", "region": "AU"},
    {"sku": "SKU-002", "region": "NZ"}
  ]
}

What happens

1. validate-json — validates the request payload before sending it anywhere
2. http-transform — POSTs the JSON to the pricing service; the response is CSV and becomes the new payload
3. select — keeps only the four columns needed downstream, dropping any extra fields the service returns
4. count — prints how many rows were returned

Notes

• expect-format: csv tells http-transform to parse the response as CSV. The response Content-Type must be text/csv.
• This example requires a reachable HTTP service at the configured url.
• Environment variable substitution in headers (e.g. ${PRICING_TOKEN}) is not automatic — use your shell or a secrets manager to inject the value before running.

Run it

PRICING_TOKEN=mytoken pipectl run pricing-sync.yaml < pricing-request.json