Skip to main content

dominaite-python

Server-side Python client for the Dominaite merchant API. One call from your backend opens a hosted checkout session; a two-line script tag renders the payment widget on your page. Card details go straight from your customer's browser into the payment widget - they never touch your server, which keeps your PCI scope minimal (SAQ A).

Python 3.9+, standard library only. No requests, no framework, nothing to vendor.

Install

The package name is dominaite on PyPI (verified free 2026-08-17; matches import dominaite, the same pattern Stripe uses). It is not published yet - until it is, install from a checkout:

pip install /path/to/dominaite-python-sdk
# or, while you are working on the SDK itself:
pip install -e /path/to/dominaite-python-sdk

Credentials

You get two values from the Dominaite dashboard, under Online payments -> Website integration, when you create an API key. The secret is shown once - store both like passwords:

  • dmk_... - your API key id. Identifies you; not secret by itself.
  • dms_... - your API secret. Server-side only: environment variable or a config file outside the web root. Never in a browser, never in git, never in logs.

Every request is signed with the secret (HMAC-SHA256) and timestamped. Keep your server clock on NTP - signatures older than 5 minutes are rejected.

Quickstart against dev

Everything you need to go from nothing to a live session on the dev environment.

1. Set your credentials. Both come from the dashboard's Website-integration tab (dev dashboard, dev key - a prod key will not authenticate against dev):

export DOMINAITE_KEY_ID='dmk_...'          # the key id shown on the tab
export DOMINAITE_SECRET='dms_...'          # the secret shown once at key creation
export DOMINAITE_BASE_URL='https://func-dom-gw-payments-dev-gwc-01.azurewebsites.net/api'

That base URL is the dev payments service. Production is https://api.dominaite.com/payments, which is the SDK's default when you pass no base_url.

base_url has to be https://. Every request carries your key id and a signature, and over plain http anyone on the path can read them and replay the request inside the server's 5 minute window, so the constructor raises ValueError rather than let that happen. The one exception is loopback - localhost, 127.0.0.1 and ::1 may use http://, so a local mock server works.

2. Check your signing before you call anything. This runs offline against the published test vector and authenticates nothing, so it can never fail for credential reasons:

python -m pytest tests/test_signing.py

3. Ping before your first mint. One signed GET that creates nothing, so a failure here is your credentials, your signing or your clock and nothing else:

import os

from dominaite import DominaiteClient

client = DominaiteClient(
    os.environ["DOMINAITE_KEY_ID"],
    os.environ["DOMINAITE_SECRET"],
    base_url=os.environ.get("DOMINAITE_BASE_URL", "https://api.dominaite.com/payments"),
)

print(client.ping())
# {'pong': True, 'merchantId': '...', 'serverTime': '...', 'clockSkewSeconds': 0}

Watch clockSkewSeconds: requests start failing once it passes 300, so a drifting number is your warning to fix NTP before payments break.

4. Mint a session (mint.py):

import os

from dominaite import CheckoutRefusedError, DominaiteClient, TransportError

client = DominaiteClient(
    os.environ["DOMINAITE_KEY_ID"],
    os.environ["DOMINAITE_SECRET"],
    base_url=os.environ.get("DOMINAITE_BASE_URL", "https://api.dominaite.com/payments"),
)

try:
    session = client.create_checkout_session(
        amount=2500,                    # minor units: 2500 = 25.00 EUR
        currency="EUR",
        order_reference="order-1042",   # your own order id, shows up in your dashboard
        customer={
            # Pass everything you already know - prefilled fields are hidden from the
            # payer, so the checkout form stays short.
            "firstName": "Ana",
            "lastName": "Kirova",
            "email": "ana@example.com",
        },
        language="bg",                  # widget UI language
        theme="dark",
    )
except CheckoutRefusedError as refusal:
    # Machine-readable: refusal.error_code - see the exception docstring for the codes.
    raise SystemExit("Payment unavailable: " + refusal.error_code)
except TransportError:
    # Network blip - safe to retry with the same idempotency_key.
    raise SystemExit("Payment temporarily unavailable")

print(session["transactionId"], session["cashierKey"], session["cashierToken"])
python mint.py

A transaction id, cashier key and cashier token on stdout means the whole chain works: your credentials, your clock, your signing, and the dev gateway.

If it fails, the error tells you which one:

What you see What is wrong
AuthenticationError + INVALID_API_KEY Wrong or revoked key id, or a prod key against dev.
AuthenticationError + INVALID_SIGNATURE Secret does not match the key id.
AuthenticationError + TIMESTAMP_OUT_OF_RANGE Your machine's clock is more than 5 minutes off.
AuthenticationError + IP_NOT_ALLOWED The key has an IP allowlist that does not include you.
CheckoutRefusedError You authenticated fine; the gateway declined to open a session.
TransportError Wrong base URL, or the service is down. Retry with the same key.

5. Render the widget. Store session["transactionId"] against your order, then hand the two cashier values to the page:

<div id="checkout"></div>
<script src="https://bp-checkout.dominaite.com/v2/launcher"
        data-cashier-key="{{ cashier_key }}"
        data-cashier-token="{{ cashier_token }}"></script>

HTML-escape both when templating (Jinja's autoescape does it for you). They are per-payment session values, not your credentials.

6. Learn the outcome from a webhook. The payer finishing on the widget is not your signal that you got paid - your backend does not see that at all. Point an endpoint at your server, verify the signature, and act on payment.succeeded. See Webhooks below; that is the step that closes the loop, not get_status in a loop.

That's the whole integration: the session call, the script tag, the webhook, and your domain bound to your checkout by Dominaite during onboarding.

Amounts are minor units

amount is always an integer in the currency's minor unit: 2500 is 25.00 EUR. A float or a string raises ValueError before anything is sent. The amount is locked server-side - what you pass here is what gets charged; nothing in the browser can change it.

Retries and double-charges

Every create_checkout_session call carries an idempotency key (auto-generated, or pass your own as idempotency_key). Retrying with the same key never opens a second payment - on a timeout, retry with the same key rather than generating a new one.

If the first attempt did land, the retry comes back as a CheckoutRefusedError with a replay code, not as the original session: there are no cashier fields to hand the embed snippet. Use refusal.transaction_id with get_status() to find out what the first attempt did (see Recovering from a replay refusal).

There is a helper that does exactly that:

session = client.create_checkout_session_with_retry(
    amount=2500,
    currency="EUR",
    order_reference="order-1042",
    max_attempts=3,
)

It retries only TransportError (network failures, 5xx, MERCHANT_API_UNAVAILABLE), reuses the one key across all attempts, and backs off between them. Refusals, authentication failures and rate limits are raised immediately.

Sessions expire

A session is valid for 2 hours. If the payer comes back later, create a new session.

Webhooks

Webhooks are how you find out what happened to a payment. Create an endpoint in the Dominaite dashboard, pick the events you care about, and store the whsec_... secret it shows you - like the API secret, it is shown once.

Dominaite POSTs each event to your URL with an X-Webhook-Signature header:

X-Webhook-Signature: t=1755700000,v1=5305bcf1302fdaba8f8c19a20c899e916fb4d2a7d8d547c62529ff87c4697b72

verify_webhook checks it and hands you the decoded event:

import os

from flask import Flask, request

from dominaite import WebhookVerificationError, verify_webhook

app = Flask(__name__)
SECRET = os.environ["DOMINAITE_WEBHOOK_SECRET"]


@app.post("/webhooks/dominaite")
def dominaite_webhook():
    try:
        event = verify_webhook(
            request.get_data(),                          # RAW body, not request.json
            request.headers.get("X-Webhook-Signature", ""),
            SECRET,
        )
    except WebhookVerificationError:
        return "", 400

    if already_handled(event["id"]):                     # delivery is at-least-once
        return "", 200

    enqueue(event)                                       # do the work outside the request
    return "", 200

Pass the raw body. request.json (or any parse-then-re-serialize round trip) gives you different bytes than the ones that were signed, and verification will fail. This is the single most common webhook integration bug.

Verify before you read anything. Until verify_webhook returns, the body is just bytes a stranger POSTed at you. Do not branch on event["type"] or trust an amount from an unverified payload.

The events

payment.succeeded, payment.failed, payment.requires_capture, payment.cancelled, payment.abandoned, payment.refunded, payment.disputed.

payment.succeeded is the only one that means money in hand. requires_capture is an approved hold, not a payment. pending and processing are never webhooked - if you want to show an in-flight state to a customer, poll the session.

Each delivery is a flat JSON object - there is no success wrapper, so do not branch on one:

{
  "id": "7f9c24e5-1d1f-4c0a-9b6c-2f3a4d5e6f70",
  "type": "payment.succeeded",
  "createdAt": "2026-08-20T14:00:00Z",
  "data": {
    "transactionId": "0f1e2d3c-4b5a-6978-8796-a5b4c3d2e1f0",
    "status": "succeeded",
    "previousStatus": "pending",
    "kind": "sale",
    "amount": 8440,
    "grossAmount": 8701,
    "surchargeAmount": 261,
    "currency": "EUR",
    "originalTransactionId": null,
    "idempotencyKey": "order-123"
  }
}

Amounts are minor units. On payment.* events amount is what you get paid and grossAmount is what moved on the card; on payment.refunded amount is what went back to the customer.

Delivery, retries, and staying enabled

  • At-least-once. The same event can arrive twice. Dedupe on event["id"] and make your handler idempotent.
  • Answer fast. Return 2xx as soon as the signature checks out and queue the real work. Doing it inline is how endpoints end up timing out and getting retried.
  • Retries follow your endpoint's configured count (default 3, max 10), spaced 1m, 5m, 30m, 2h, 12h.
  • Circuit breaker. An endpoint whose first attempt and every retry fail, over and over, is disabled automatically. Any later successful delivery re-enables it. A disable you did yourself in the dashboard is never undone for you.

Reconcile anyway

Webhooks complement your reconciliation sweep, they do not replace it. There are real loss windows: a chain parked on an endpoint the breaker disabled, an event that never got published. The only thing that closes them is a periodic sweep of your own open orders against get_status. Keep the sweep. It is what catches the payment nobody told you about.

Tolerance and clocks

Deliveries more than 300 seconds away from your clock are rejected as replays. If you see TIMESTAMP_OUT_OF_RANGE on genuine traffic, your server clock has drifted - fix NTP rather than widening tolerance_seconds.

Pass now to keep your own tests deterministic, and sign_webhook to forge a delivery in them:

from dominaite import sign_webhook, verify_webhook

body = '{"id":"evt-1","type":"payment.succeeded","createdAt":"2026-08-20T14:00:00Z","data":{}}'
header = "t=1755700000,v1=" + sign_webhook("whsec_test", "1755700000", body)

event = verify_webhook(body, header, "whsec_test", now=1755700000)

Status polling (fallback)

Polling is the fallback and the reconciliation tool, not the primary path - use webhooks to learn that a payment completed, and use this to sweep your own open orders and to answer "what is this order doing right now".

status = client.get_status(session["transactionId"])
# {"transactionId": ..., "orderReference": "order-1042", "status": "succeeded",
#  "amount": 2500, "currency": "EUR", ...}

status is one of: pending, processing, succeeded, failed, refunded, partially_refunded, cancelled, disputed, requires_capture, abandoned. While the session is still payable the response also carries expiresAt; after that instant a pending session can only become abandoned. An unknown transaction id raises ApiError with http_status == 404.

Those values are also exported as the PaymentStatus enum (and PAYMENT_STATUSES), so you can match on a named member instead of a bare string literal. It subclasses str, so status["status"] == PaymentStatus.SUCCEEDED works directly against what get_status returns.

succeeded is the only value that means the payment is complete. Keep polling on pending, processing and requires_capture - none of them is terminal.

requires_capture is not "unpaid": the payer has already paid and the funds are held awaiting capture. Never treat it as an abandoned order.

Treat any status you do not recognise as still-open as well: a value the API adds later should make you keep polling, never silently close an order that is still live.

Poll after the payer returns to you, or on your order timeout - not in a tight loop; the endpoint is rate limited per key.

Recovering from a replay refusal

When your idempotency key collides with an earlier attempt, the refusal names the transaction it collided with, so you can reconcile instead of minting a second payment:

try:
    session = client.create_checkout_session(...)
except CheckoutRefusedError as refusal:
    if refusal.transaction_id:
        status = client.get_status(refusal.transaction_id)
        # Now you know what the earlier attempt actually did.

refusal.transaction_id is None when the API did not name one (a concurrent-race DUPLICATE_REQUEST knows the key is taken but not yet by which row), so check it before use. The full refusal payload is on refusal.result.

Errors

Exception Means Retry?
AuthenticationError Bad credentials, bad signature, clock skew, IP not allowlisted No - fix config
CheckoutRefusedError The gateway refused to open the session (error_code) Depends on the code
ApiError Unexpected response, or a 4xx like an unknown transaction id (http_status, error_code) No
RateLimitError HTTP 429; you are sending faster than the key is allowed (retry_after_seconds) Yes, after you wait
TransportError Network failure or 5xx; you don't know if it landed Yes, same idempotency key
WebhookVerificationError An incoming webhook is not authentic or not fresh (error_code) No - respond 400

All of them inherit from DominaiteError if you only care that the call failed.

Note the two different failure shapes on the create endpoint. A business refusal is HTTP 200 with success: false and raises CheckoutRefusedError; input validation is HTTP 400 and raises ApiError with the code on error_code (currently IDEMPOTENCY_KEY_REQUIRED, exported as VALIDATION_ERROR_CODES). Branch on the exception type, never on the HTTP status.

WebhookVerificationError.error_code is one of MALFORMED_SIGNATURE (wrong header, or a proxy rewrote it), INVALID_SIGNATURE (wrong secret, modified body, or you passed a re-serialized body instead of the raw one), TIMESTAMP_OUT_OF_RANGE (replay, or your clock drifted), and INVALID_PAYLOAD (signed, but not a JSON object).

Rate limits

60 requests per minute per API key, and 120 per minute per IP address. Both are sliding windows, and the IP limit is shared by every key sending from that address, so a busy host can trip it while each key is well inside its own budget.

Going over gets you a RateLimitError. The SDK does not retry it for you, and neither does create_checkout_session_with_retry - answering "you are sending too much" with more traffic is how a short spike turns into a sustained lockout. Wait, then send again:

from dominaite import RateLimitError

try:
    session = client.create_checkout_session(...)
except RateLimitError as limit:
    time.sleep(limit.retry_after_seconds or 60)

retry_after_seconds is what the API asked you to wait, and honouring it is the shortest wait that will work. It is None when the API did not give a number of seconds - back off on your own schedule then.

The usual cause is polling get_status in a tight loop. Poll after the payer returns to you, or on your order timeout, and let webhooks do the rest.

Field lengths

order_reference and idempotency_key are capped at 100 characters each. Characters, not bytes: a 100-character Cyrillic or Greek reference is 200 UTF-8 bytes and the platform takes it.

Running the tests

python -m venv .venv && .venv/bin/pip install pytest
.venv/bin/python -m pytest

tests/test_signing.py reproduces the signing test vector published on the dashboard's Website-integration tab. If it ever fails, the SDK cannot authenticate - fix the signing, never the expected value.

tests/test_webhooks.py pins the cross-SDK webhook vector: the same secret, timestamp and body bytes every Dominaite SDK verifies against. A failure there means the SDK is rejecting genuine deliveries or accepting forged ones. Same rule - fix the code, not the vector.

Release files for dominaite 0.2.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for dominaite 0.2.0
File Size Uploaded
dominaite-0.2.0.tar.gz 38.7 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for dominaite 0.2.0
File Interpreter ABI Platform
dominaite-0.2.0-py3-none-any.whl Python 3 none any Details

Total release size: 65.3 kB

Release files / dominaite-0.2.0.tar.gz

Download URL dominaite-0.2.0.tar.gz
Size 38.7 kB
Tags Source
SHA-256 checksum
How to use checksums
6aa38db4d1bdc4a729a8edfc0ed3be8df68fe421db9a4cb758dcba021ebeffdc
BLAKE2b-256 checksum
How to use checksums
92b422dbce0ef7eab14b109d1416b166e574f58a7d83d11f81cac45f1b8e7d31
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 23, 2026.

Transparency log

Release files / dominaite-0.2.0-py3-none-any.whl

Download URL dominaite-0.2.0-py3-none-any.whl
Size 26.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
55b149fcea82c05daae387e9a618653ef735463cb9425a57add58ec4eff4b30a
BLAKE2b-256 checksum
How to use checksums
6b14b65bb2a5a938e2661405dffb559aa152dcccb87bf7cc24a20c87ed401fb5
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 23, 2026.

Transparency log

Release history Release notifications | RSS feed

0.3.0

2 release files

This release

0.2.0 This release

2 release files

0.1.2

2 release files

0.1.1

2 release files

0.1.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page