ooredoo raastas via bml card

This commit is contained in:
2026-10-03 01:08:33 +05:00
parent 0d4f0074c7
commit 8795d5f758
16 changed files with 370 additions and 42 deletions
+30 -3
View File
@@ -205,8 +205,10 @@ no `action`; their JavaScript posts back to the **same creq URL**. So the creq U
`destValue=token`, `selectChannel=token`, `authMethod=OOB`, `otpDest=`, `formReqType=SUBMIT`
(keep the hidden `creq` / `otpChannels`).
3. **POST channel** → the **OTP entry** page (`otpValue` input). Submit `otpValue=<BML token TOTP>`,
`formReqType=SUBMIT`. A wrong/expired code re-renders the OTP page with text containing
*"incorrect"* / *"expired"* — regenerate the TOTP and retry once.
`formReqType=SUBMIT`. A wrong/expired code re-renders the OTP page (still with `otpValue`) and
*"The OTP code you entered is incorrect Please try again."* — wait for the next TOTP window,
regenerate and retry once. Rejected twice, the payment stops ("The bank rejected the BML token
code"). The app also never sends a code with under 5 s left in its window.
4. On success the ACS returns a form auto-posting **`cres`** to the Mastercard gateway; the gateway
returns a form auto-posting the result (`order.id`, `result=SUCCESS`, …) to
**`transactions/mpgsNotification/<id>`**. Follow both so the verdict is recorded.
@@ -225,6 +227,23 @@ Poll `next-action` until the recorded verdict surfaces:
| `TRANSACTION_CONFIRMED` | Success |
| `TRANSACTION_FAILED` | Declined |
**A decline after 3-D Secure doesn't arrive this way.** In the Ooredoo capture, the card passed
3-D Secure (`mpgsNotification` got `result=SUCCESS`, `gatewayRecommendation=PROCEED`) and was then
declined for insufficient funds. The browser's `?wait=1` went to `?error=1` instead of the
merchant, and the transaction stayed payable: `state` still `QR_CODE_GENERATED`, `hasError: true`,
`allowRetry: true`, and a new `paymentErrorHistory` entry:
```json
{"date":"…","vendor":"mpgs","code":"INSUFFICIENT_FUNDS",
"reason":"Transaction declined due to insufficient funds",
"customerVisibleDescription":"Insufficient funds. Please use another card or payment method."}
```
The same link was then paid successfully after topping up the card. So while polling,
`BmlMerchantCardPayClient` also reads the transaction (the load PATCH,
`BmlMerchantTxnClient.paymentErrors`) and stops with `customerVisibleDescription` as soon as an
entry newer than the ones there before the attempt shows up.
## 7. Return to the merchant
The `mpgsNotification` response is a page whose script sends the browser to
@@ -240,13 +259,21 @@ GET transaction…/<id>?wait=1
(FahiPay's is `fahipay.mv/api/bml/gateway/callback/?…state=CONFIRMED`.)
Ooredoo's callback isn't a redirect: `my.ooredoo.mv/bml/response_new.php?…state=CONFIRMED` is a
200 page whose `<body onload="document.forms['wtmpay'].submit()">` posts the result
(`order_id`, `bml_transaction_id`, `bml_response=CONFIRMED`, `payment_status=success`, …) on to
`www.ooredoo.mv/ooredoo-prod/PaymentGateway/redirect/bml`, which lands on
`/payment-status?order_id=…&status=1`. `returnToMerchant` submits such auto-posting forms too
(up to 2).
**This hop is required.** It's how at least Dhiraagu learns it was paid: a test reload that
stopped at `TRANSACTION_CONFIRMED` charged the card but never topped up, and opening the
`?wait=1` URL in a browser afterwards delivered it. The signature is generated by BML, so the
hop can be replayed later from the transaction id alone.
`BmlMerchantCardPayClient` does it after every confirmed payment (`returnToMerchant`): a browser
UA GET that follows the redirects, up to 3 tries, success = the chain ends on a 2xx page. The
UA GET that follows the redirects and auto-submitted forms, up to 3 tries, success = the chain
ends on a 2xx page. The
merchant host may be behind Cloudflare: plain `curl` got a 403 on `dhiraagu.com.mv`, okhttp got
through.