-
Notifications
You must be signed in to change notification settings - Fork 15
gateway_params
Merchants often choose to use token providers such as gateways to handle credit card payments. This increases security for end users, merchants and allows for multiple payment models for when the card is not present, as token providers often allow re-use of those cards.
The following diagram demonstrates a sequence of client and server-side interactions leading to the generation and storage of tokens on first time use.
In step 2 of the diagram, the merchant's backend responds to a user request with a checkout page containing token provider parameters.
These parameters contain information in addition to what's specified in basic-card
:
-
merchantId
, required. Serves to identify the merchant at the token provider. By design, this string is opaque to parties other than the token provider. -
publicKey
, optional. A public key that, when present, is used to encrypt card data sent to the token provider. -
tokenProviderURL
, required. This the identifier of the token provider the merchant has a relationship with. This identifier is used to help establish matching payment apps (that interact with that token provider). -
environment
, required. This flag enables the merchant to signal to the payment app (and tokenization provider) that a transaction is either a production ("real" transaction) or a test. The default value of this parameter is "production". -
oneTimeUse
, required. When true, this boolean indicates that the merchant only accepts one time use tokens. When false, the it means that the merchant requests that tokenization provider store the card data for subsequent transactions. The default value of this parameter is "false".
An example token provider tokenized payment method specification which would be given to the payment request API:
enum TokenProviderEnvironment {
"test",
"production"
};
dictionary TokenProviderParameters {
required DOMString merchantId;
required DOMString publicKey;
required DOMString tokenProviderURL;
optional TokenProviderEnvironment environment;
optional Boolean oneTimeUse;
};
dictionary ProviderTokenizedCardRequest {
sequence<DOMString> supportedNetworks; // Defined in Basic Card
sequence<CardType> supportedTypes; // Defined in Basic Card
required TokenProviderParameters tokenProviderParameters;
};
This will materialize in Javascript as follows:
var supportedInstruments = [
{
supportedMethods: ['provider-tokenized-card']
data: {
supportedNetworks: ['amex', 'discover','mastercard','visa'],
supportedTypes: ['credit', 'debit'],
tokenProviderParameters: {
merchantId: 'ZWMQtaTEyVDb8]FKdNaVmVj2yy;rNG',
publicKey: 'live_DmF7xVDZKAUpkY]sxjidtRg@6dVPLt',
tokenProviderURL: 'https://my.payment-gateway.com/webpayment/tokenize',
environment: 'test',
oneTimeUse: false
}
}
},
...,
];
var payment = new PaymentRequest(
supportedInstruments,
details,
options
);
Following steps 3, 4 and 5, assuming the user chooses a provider tokenized payment method, the Payment Handler
will contact the token provider with the parameters provided in the payment method declaration configuring the token provider.
The Payment Handler
will send a request to tokenProviderURL
, passing the parameters merchantId
and environment
, but will also send a Base64-encoded encrypted hash, using the publicKey
provided, of the basic-card
information that would have been sent to the merchant as per the BasicCardResponse
specification.
Optionally, the merchant could provide a oneTimeUse
parameter, which signals to the payment app that this payment method is not to be saved for subsequent use.
It would look something like:
dictionary TokenizationRequest {
required DOMString merchantId;
optional TokenProviderEnvironment environment;
required DOMString cardData;
optional Boolean oneTimeUse;
};
Where cardData is a base-64 encoded string of encrypted data of the following data structure:
dictionary TokenizedCard {
DOMString cardholderName;
required DOMString cardNumber;
DOMString expiryMonth;
DOMString expiryYear;
DOMString cardSecurityCode;
PaymentAddress? billingAddress;
};
This section describes steps 7, and 8.
The token provider responds with two parameters:
-
instrumentToken
: The implementation of the generation of thisinstrumentToken
is left to the token providers -
oneTimeUse
: this gives the opportunity for the tokenization provider to possibly override the preference submitted in the request and describes to thePayment Handler
how to handle the token
dictionary TokenProviderResponse {
DOMString instrumentToken;
Boolean oneTimeUse;
};
Upon receiving this instrumentToken
, the Payment Handler
should generate and store a token-provider-merchant-identifier
for the combination of the:
- card
- token provider
- merchant checkout page origin
This token-provider-merchant-identifier
will be persisted for subsequent use.
This section describes steps 9, 10, 11, 12, 13 and 14 in the diagram, where 11-14 are outside of the scope of the specs described.
After the instrumentToken
is obtained, if oneTimeUse
is false
, the token-provider-merchant-identifier
is generated and stored.
If oneTimeUse
is true, the Payment Handler
will not store this id.
Afterwards, the payment request API returns data similar to what would be returned by basic-card
.
dictionary ProviderTokenizedCardResponse {
DOMString cardholderName;
DOMString tokenizedCardId;
optional DOMString issuerIdentificationNumber;
optional DOMString instrumentToken;
required DOMString tokenProviderURL;
PaymentAddress? billingAddress;
};
In the diagram above, the merchant will then use the instrumentToken
to obtain the token generated by the token provider via server to server interaction.
The implementation of this interaction will be up to the merchant and the token provider. The token provider could also send the actual token to the client if desired, although it would entail trusting the client's computer not to be compromised in the case of long lived tokens.
A response code will be sent to the merchant checkout page indicating success or failure, as per the Payment Request API.
When the Payment Handler
finds the same card, token provider and merchant origin combination, it no longer needs to obtain an instrumentToken
from the token provider.
In which case, the interactions will manifest as follows:
In case either the token provider or the merchant encounter exceptions, they should be handled as one entity.
i.e. the error handling should be no different than the current implementation in Payment Request when there was only one network entity to be reached.
This will minimize UX issues
One of the use-cases of this is with Payment Gateways such as Braintree or Stripe.
Those gateways currently send a nonce
as instrument token to the client. The token that will be long-lived is only exchanged server-to-server.
Another possible implementation is exactly the same as the Nonce Passing, except for providing the actual token to the client, which is then passed to the merchant.
This behavior is entirely up to the merchant and the token provider. A token-provider-merchant-identifier
is still generated and the token is never stored on disk by the Payment Handler
.
This saves one roundtrip by the merchant to the token provider, at the cost of security, but gaining in convenience.
There is a use-case where token providers and payment processors will be different entities. Note that the spec is unchanged. It allows for this use-case and the following diagram illustrates how this could be implemented.
The merchant, when communicating with the Token Provider, may have multiple acquirers registered.
The merchant makes a decision on which acquirer to use depending on the issuerIdentificationNumber
. This is to increase acceptance rates, to diminish costs or to increase redundancy.
The following diagram illustrates this use-case:
The following depicts an example network topology between a merchant, payment gateways and acquirers: