etld 1.1.0

Official Python SDK for the ETL-D API - Stateless Data Middleware for AI Agents.

ETL-D SDK: The Agentic Data & Finance Middleware

Stop wasting LLM context windows and tokens on basic data parsing and complex financial logic. ETL-D is a stateless, deterministic middleware designed specifically for AI Agents, No-Code automation (n8n, Make), and high-performance integrations. We handle chaotic global data and B2B financial pipelines using algorithmic parsing with an LLM-Routing Fallback to guarantee 100% accuracy and perfect JSON schemas.

🚀 The Arsenal (Available Endpoints) ETL-D goes far beyond basic string manipulation. Your AI agents gain instant access to:

• B2B Parsing: Transform raw Norma 43 (N43) bank statements and complex nested XMLs into flat, queryable JSONs.
• Financial Operations: Generate SEPA Direct Debit (PAIN.008) XMLs, fetch historical Forex rates, and autonomously map messy transactions to GAAP/PGC accounting codes.
• Data Engineering Magic: Generate E-commerce Cartesian product variants, normalize physical measurement units, and intelligently split chaotic contact strings.
• Core Enrichment: E.164 phone validation, ISO 8601 relative date parsing (timezone-aware), and global address structuring with LLM-fallback for complex regions (e.g., UAE, Japan).

🤖 The Agentic Workflow (Autonomous Billing) ETL-D V1.0+ introduces features strictly designed for autonomous systems that handle their own billing lifecycles.

1. Zero-Touch Agent Provisioning

If your AI agent is booting up for the first time and lacks an API key, it can provision one programmatically. No emails, no logins.

import etld
from etld.client import EtlDClient

client = EtlDClient(api_key="BOOTSTRAP") # Key not needed for provisioning

# Provision a new session
provision = client.create_provisioning_session()
print(f"Pay here: {provision['checkout_url']}")

# Poll for the key
status = client.poll_provisioning_status(provision['poll_id'])
if status['status'] == 'ready':
    print(f"Your API Key: {status['api_key']}")

2. Handling 402 Payment Required

When an agent exhausts its credits, the API returns a HTTP 402. The SDK intercepts this and raises a PaymentRequiredError containing a checkout_url. The agent can autonomously navigate to the URL, pay for a top-up, and resume execution.

from etld.exceptions import PaymentRequiredError

try:
    client.enrich_date("next Tuesday")
except PaymentRequiredError as e:
    print(f"Credits exhausted. Top up at: {e.checkout_url}")
    # Agent can now wait, or notify human, or even use a headless browser to pay!

⚡ Async Batch Processing Process 10,000+ records without blocking your server. Queue the batch, and the SDK automatically polls the background worker until the job is done.

task_id = client.enrich_batch("date", ["next Tuesday", "27/10/2023"])
results = client.wait_for_batch(task_id)
print(results)

Installation

pip install etld