`Psbt::sign` panics on `non_witness_utxo` with out-of-range vout (upstream fix pending in bitcoin 0.32.x)

[noahjoeris]

Describe the bug

bitcoin::Psbt::spend_utxo in the bitcoin 0.32.x line indexes non_witness_utxo.output[vout] without a bounds check, so Psbt::sign (and sighash_ecdsa) panic on a PSBT input whose non_witness_utxo has fewer outputs than previous_output.vout (e.g. a wrong or truncated previous transaction). This was fixed upstream on master in rust-bitcoin/rust-bitcoin#5500 (returns MissingSpendUtxo instead), but the fix is not in the 0.32.x branches as of 0.32.8.

Wallet::sign_with_signers is not affected, since it validates non_witness_utxo against the outpoint before signing. However, since #505 the recommended flow in docs and examples is to call bitcoin::Psbt::sign directly, which does reach the panic when given a malformed PSBT.

This issue tracks picking up the upstream fix (backport to 0.32.x or bump to a release that includes it).

This issue was found by AI.

To Reproduce

// psbt built by the wallet with a single wpkh input, then corrupted:
let mut psbt = builder.finish().unwrap();
psbt.inputs[0].witness_utxo = None;
psbt.inputs[0].non_witness_utxo = Some(Transaction {
    version: transaction::Version::TWO,
    lock_time: absolute::LockTime::ZERO,
    input: vec![],
    output: vec![], // no outputs
});
psbt.sign(&xpriv, wallet.secp_ctx()); // panics: index out of bounds at bitcoin-0.32.8/src/psbt/mod.rs:626

Expected behavior

Signing a PSBT with an inconsistent non_witness_utxo should return an error rather than panic.

Build environment

• BDK tag/commit: 6fc68462 (master), bitcoin 0.32.8

Which backend(s) are relevant (if any)?

• None / not backend-related (e.g. bdk_chain, bdk_core)

Is this blocking production use?

• No

[Arowolokehinde]

i would like to look into this @noahjoeris