pipeline.py

# pipeline.py
import json
import datetime

STATUS_WHITELIST = ["pending", "confirmed", "shipped", "cancelled"]

stats = {
    "processed": 0,
    "failed": 0,
    "skipped": 0,
}


def enrich(d):
    d["total_with_tax"] = round(d["amount"] * 1.21, 2)
    d["processed_at"] = datetime.datetime.utcnow().isoformat()
    return d


def process_data(input_path, output_path):
    with open(input_path, "r") as f:
        records = json.load(f)

    results = []
    failed_records = []

    for record in records:
        if "order_id" not in record or "amount" not in record or "status" not in record:
            stats["failed"] += 1
            failed_records.append(record)
            continue

        if record["status"] not in STATUS_WHITELIST:
            stats["skipped"] += 1
            continue

        res = enrich(record)
        results.append(res)
        stats["processed"] += 1

    with open(output_path, "w") as f:
        json.dump(results, f, indent=2)

    return stats