address: reject v2-v16 segwit addresses encoded with bech32 (BIP-350)

[Lrifton92]

@

Problem

decodeSegWitAddress is written as a full BIP-350 segwit decoder: it accepts witness versions 0 through 16 (rejecting version > 16). BIP-350 requires that witness version 0 uses bech32, and versions 1 through 16 use bech32m. The reference decoder in the BIP rejects any non-zero witness version that is not bech32m:

if (data[0] == 0 and spec != Encoding.BECH32) or (data[0] != 0 and spec != Encoding.BECH32M):
    return (None, None)

The current code only special-cased versions 0 and 1:

// For witness version 1, the bech32m encoding must be used.
if version == 1 && bech32version != bech32.VersionM {

So a witness program with version 2-16 encoded using the legacy bech32 checksum decoded successfully, in violation of the spec and contrary to the BIP-350 reference decoder.

Reproduction

These are official INVALID_ADDRESS vectors from BIP-350. Both decode successfully on current master and are rejected after this change:

• tb1z0xlxvlhemja6c4dqv22uapctqupfhlxm9h8z3k2e72q4k9hcz7vqglt7rf — witness v2 with bech32 (should require bech32m)
• BC1S0XLXVLHEMJA6C4DQV22UAPCTQUPFHLXM9H8Z3K2E72Q4K9HCZ7VQ54WELL — witness v16 with bech32

Fix

Generalize the version == 1 check to version >= 1, matching the BIP-350 reference decode(). Added a regression test (address/bip350_diff_test.go) covering the v2 and v16 invalid vectors; it fails before this change and passes after.

This aligns the internal reference decoder with BIP-350 for the full v0-16 range the function already claims to handle. Versions 2-16 are not yet surfaced through DecodeAddress today, so this is primarily a spec-conformance / forward-compatibility fix rather than a behavior change for callers of the public API.

Verification

• go build ./... — clean
• go vet ./... — clean
• go test ./... — ok for the address, address/base58, address/bech32 packages
• gofmt -l on touched files — clean
  @

[Lrifton92]

Friendly ping — this is ready for review whenever a maintainer has bandwidth. The change enforces BIP-350 strictly (reject v2-v16 segwit addresses encoded with legacy bech32). Thanks!

[Lrifton92]

Friendly follow-up — this is still ready for review whenever a maintainer has bandwidth. It strictly enforces BIP-350 (rejecting v2–v16 segwit addresses encoded with bech32 instead of bech32m). Small, self-contained change, no conflicts. Thanks!

[Roasbeef]

Ci run approved!

[Roasbeef]

LGTM 🫛