Reading a DataFrame succeeds even if it is incomplete

[neivv]

If DataFrame::read_dataframe is passed too few bytes, it will still succeed and return a dataframe with an incomplete payload.

Example:

    use websocket::DataFrame;
    let mut data = vec![0x8au8, 0x08, 0x19, 0xac, 0xab, 0x8a, 0x52, 0x4e, 0x05, 0x00];
    println!("Incomplete header:\n  {:?}", DataFrame::read_dataframe(&mut &data[..1], false));
    // This doesn't make sense
    println!("Incomplete payload:\n  {:?}", DataFrame::read_dataframe(&mut &data[..6], false));
    println!("Complete dataframe:\n  {:?}", DataFrame::read_dataframe(&mut &data[..], false));
    data.push(0xff);
    println!("Additional data:\n  {:?}", DataFrame::read_dataframe(&mut &data[..], false));

outputs

Incomplete header:
  Err(NoDataAvailable)
Incomplete payload:
  Ok(DataFrame { finished: true, reserved: [false, false, false], opcode: Pong, data: [25, 172, 171, 138] })
Complete dataframe:
  Ok(DataFrame { finished: true, reserved: [false, false, false], opcode: Pong, data: [25, 172, 171, 138, 82, 78, 5, 0] })
Additional data:
  Ok(DataFrame { finished: true, reserved: [false, false, false], opcode: Pong, data: [25, 172, 171, 138, 82, 78, 5, 0] })

[illegalprime]

Thank's for the issue, the problem seems to be here. This only happens when you give it a non-blocking reader such as a &mut [u8], when reading from a TcpStream or similar it should block until it gets the amount of data in the header length.

With the "Additional Data" test you should see that you have 0xff leftover in your buffer. I think the only issue here is the "Incomplete Payload" scenario and it can be fixed by changing read_to_end to read_exact.

[illegalprime]

I have exams right now so I can't get to it, but a PR is always welcome 😉