processors/token_transfer: stop coercing a bytes to_muxed_id to 32 bytes

[tamirms]

Fixes #5984.

The bug

parseV4MapDataForTokenEvents copied a to_muxed_id of type ScvBytes into a fixed 32-byte buffer:

hashBytes := make([]byte, 32)
copy(hashBytes, val)

copy moves min(32, len(val)) bytes, so a value shorter than 32 was right-padded with zeroes and a longer one truncated to its first 32. Either way the emitted event reported a muxed id the contract never sent, silently — and a consumer cannot distinguish a genuine 32-byte value from a 4-byte one zero-extended into the same shape.

Where it came from

The same three lines appear in NewMuxedInfoFromMemo (muxed_info.go), where the source is *m.Hash — an xdr.Hash, i.e. [32]byte. There the fixed size is exact and the copy is just a safe array-to-slice conversion.

Here the source is an xdr.ScBytes, which is []byte and unbounded, so the identical code silently coerces. The 32 was inherited from the classic path, not from the data model: MuxedInfo_Hash.Hash is a []byte, so nothing downstream requires a fixed width.

The fix

Allocate len(val) instead of 32, keeping the defensive copy so nothing aliases the event's XDR buffer.

Per CAP-0067, only classic-derived events map to_muxed_id to bytes, and there it is a 32-byte MEMO_HASH/MEMO_RETURN. A custom SEP-41 contract may put any byte string in that field, so the parser now reports what was emitted and leaves any length policy to the consumer.

The issue floated ignoring such events instead. That trades silent corruption for silent loss: callers discard an event whose parse fails rather than surfacing an error (token_transfer_processor.go), so rejecting would drop the transfer's amount and both addresses over a field carrying no value.

Tests

Three cases added to TestValidContractEventsV4 — shorter than 32, longer than 32, and empty. Each was confirmed to fail against the unfixed parser:

--- FAIL: TestValidContractEventsV4/..._Hash_Memo_Shorter_Than_32_Bytes
--- FAIL: TestValidContractEventsV4/..._Hash_Memo_Longer_Than_32_Bytes
--- FAIL: TestValidContractEventsV4/..._Empty_Hash_Memo
        Should be empty, but was [0 0 0 0 ... 0]

The existing hash-memo case uses exactly 32 bytes, so neither branch was previously covered.

processors/token_transfer passes; gofmt clean, go vet clean under the flags govet.sh uses.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Preserves contract-emitted to_muxed_id byte lengths instead of silently padding or truncating them.

Changes:

• Copies ScvBytes using its original length.
• Adds short, long, and empty-byte regression tests.
• Documents the fix in the changelog.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
processors/token_transfer/contract_events.go	Preserves emitted byte length.
processors/token_transfer/contract_events_test.go	Adds boundary regression coverage.
CHANGELOG.md	Records the bug fix.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.