Is that really a Dogecoin address? Checking D... vs 0x... before anyone loses coins

I nearly lost 500 DOGE to a dropdown. A casino cashier listed Dogecoin twice, once as the native coin and once as a BEP-20 token on BNB Smart Chain, and the BEP-20 option came first. I copied the address it gave me, pasted it into my wallet and only then saw the 0x at the start. I stopped before sending. Plenty of people don't.

I spent eight years testing payment flows, and this one's a classic edge case: two things with the same name and different rails. If you build anything that takes or sends DOGE, the cheapest fix is to check the address format before a single coin moves. Here's how, with nothing but the Python standard library.

Two families of "DOGE" addresses

Starts with	What it is	Format
D	Native Dogecoin, pay-to-public-key-hash	Base58Check, version byte 0x1E, 34 characters
9 or A	Native Dogecoin, pay-to-script-hash	Base58Check, version byte 0x16
0x	An EVM address (BNB Smart Chain, Ethereum…), where a token borrows the DOGE ticker	0x + 40 hex characters

A native Dogecoin wallet won't send to a 0x address, and coins sent the other way land on a chain the receiving service may never look at. The first character tells you which world you're in, but the first character alone isn't validation: typos and truncated pastes start with D too.

Base58Check in a few lines

A native address is 25 bytes encoded in Base58: one version byte, a 20-byte hash, and a 4-byte checksum, which is the first four bytes of a double SHA-256 of the first 21. Decode, split, recompute, compare.

import hashlib
import re

ALPHABET = "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz"
DOGE_VERSIONS = {0x1E: "P2PKH", 0x16: "P2SH"}   # mainnet: D... and 9.../A...
EVM_RE = re.compile(r"^0x[0-9a-fA-F]{40}$")


def b58decode(s: str) -> bytes:
    n = 0
    for ch in s:
        i = ALPHABET.find(ch)
        if i < 0:
            raise ValueError(f"not base58: {ch!r}")
        n = n * 58 + i
    raw = n.to_bytes((n.bit_length() + 7) // 8, "big")
    pad = len(s) - len(s.lstrip("1"))
    return b"\x00" * pad + raw


def classify(address: str) -> str:
    address = address.strip()
    if EVM_RE.match(address):
        return "evm"   # BEP-20 / ERC-20 style: not a Dogecoin address
    try:
        data = b58decode(address)
    except ValueError:
        return "invalid"
    if len(data) != 25:
        return "invalid"
    payload, checksum = data[:-4], data[-4:]
    if hashlib.sha256(hashlib.sha256(payload).digest()).digest()[:4] != checksum:
        return "invalid"
    kind = DOGE_VERSIONS.get(payload[0])
    return f"doge-{kind.lower()}" if kind else "invalid"

classify returns one of four answers: doge-p2pkh, doge-p2sh, evm or invalid. A Bitcoin legacy address passes the checksum but has version byte 0x00, so it comes back invalid, which is what you'd want in a DOGE field.

What to test

The edge cases I'd put in any test suite for this field:

• a real D… address, and the same address with one character changed (checksum must fail)
• a 9… or A… script address
• 0x followed by 40 hex characters (must be flagged, not rejected silently)
• a Bitcoin address, a Litecoin address, an empty string, a string with spaces around it
• the characters Base58 leaves out on purpose: 0, O, I, l

What to do with the answer

Validation is half the job. The other half is what the person at the screen sees:

1. evm in a DOGE field: say so in words. "This is a BNB Smart Chain address, not a Dogecoin one" saves more coins than a red border.
2. Two networks for one ticker: default to the native chain, label the other one with its chain name, not just "DOGE".
3. No memo field for native DOGE. If your form shows one for Dogecoin, people will assume something's wrong. They'll be right.
4. Show the first and last characters back after paste, so people compare what they meant with what they copied.

None of this needs a node or an API call. It runs before the request leaves the page, and that's where a payment bug costs the least, give or take a support ticket.

I test crypto casino cashiers the way I used to test checkouts. The write-ups are at dogecasino.best, and the raw numbers are on GitHub.