contrib: message-capture-parser.py crashes instead of labeling non-printable message types "UNREADABLE" #35883

Is there an existing issue for this?

• I have searched the existing issues

Current behaviour

message-capture-parser.py crashes when a message type is valid UTF-8 but contains non-printable characters, instead of marking it "UNREADABLE" like it's supposed to.

The code does raise UnicodeDecodeError with no arguments. That doesn't work — UnicodeDecodeError needs 5 args to construct, so this actually raises a TypeError, which the except block below doesn't catch. Script dies mid-run.

Since output is only written after the whole file is processed, one bad message type loses the entire parse.

Repro: python3 -c " import struct msgtype = b'\x01bad' + b'\x00'*8 open('msgs_recv.dat','wb').write(struct.pack('<q', 1234567890) + msgtype + struct.pack('<i', 0))" python3 contrib/message-capture/message-capture-parser.py msgs_recv.dat

TypeError: function takes exactly 5 arguments (0 given)

Expected behaviour

Should output "msgtype": "UNREADABLE" and keep going, same as it already does for plain invalid UTF-8.

Fix: try: msgtype_tmp = msgtype.decode() except UnicodeDecodeError: msgtype_tmp = None msg_dict["msgtype"] = msgtype_tmp if msgtype_tmp and msgtype_tmp.isprintable() else "UNREADABLE"

Steps to reproduce

1. cd contrib/message-capture

2. Create a fake capture file with a non-printable message type:

   python3 -c " import struct msgtype = b'\x01bad' + b'\x00'*8 open('msgs_recv.dat','wb').write(struct.pack('<q', 1234567890) + msgtype + struct.pack('<i', 0))"

3. Run the parser on it:

   python3 message-capture-parser.py msgs_recv.dat

4. Script crashes with a TypeError instead of printing UNREADABLE and continuing.

Relevant log output

No response

How did you obtain Bitcoin Core

Compiled from source

What version of Bitcoin Core are you using?

master@5569887

Operating system and version

Windows 11 Pro 10.0.26200

Machine specifications

No response

This seems like it was discovered with AI as opposed to running this on an actual capture and getting an error. Since this is a contrib script, I'm not sure if that is relevant.

Yess, it was hand-built, not from a live capture. I'll build and test it against a real one to confirm.

Compiled bitcoind from source on my machine, started it as an actual running node with -capturemessages -v2transport=1, and connected a real Python p2p client to it over TCP connection with the full protocol handshake.

sent a real wire message with type byte 0x7F - the node received it, didn't recognize it, and wrote it straight to msgs_recv.dat on disk, same as it would for any peer.

Then ran the actual, unmodified message-capture-parser.py (pre-fix, bug still in it) against that real file. It crashed:

TypeError: function takes exactly 5 arguments (0 given)

exit code 1, dies right at the raise UnicodeDecodeError line.

Thx, closing for now. There is no need to open an issue and a pull request. It should be fine to have the discussion in one thread only.