Why does the decode function of the segwit bech32 encoder/decoder take a hrp (human readable part) as input?

Looking at the code referenced in bip-0173, specifically for example here: https://github.com/sipa/bech32/blob/master/ref/javascript/segwit_addr.js

  function decode (hrp, addr) {
      var dec = bech32.decode(addr);
      if (dec === null || dec.hrp !== hrp || dec.data.length < 1 || dec.data[0] > 16) {
        return null;
      }
      ...

I understand that this is an error checking statement, but why check a (user?) inputted hrp vs the decoded hrp from bech32.decode?

Also I'm assuming that the "dec.data[0] > 16" check is to make sure that the byte at index zero doesn't exceed a value of 16, which would be invalid hex. Is this correct?

kawthuldrok

Posted 2019-01-07T21:21:48.000

Reputation: 83

Answers

The HRP is part of the bech32 encoded string, and in the bech32 decoding API, it is returned along with the payload by the decoder, after checking the checksum.

We still want to compare it with the expected HRP in BIP173, which encodes the chain the software is operating on.

Otherwise you could have a testnet node that accepts mainnet BIP173 addresses or the other way around.

Pieter Wuille

Posted 2019-01-07T21:21:48.000

Reputation: 54 032

I'm assuming that the "dec.data[0] > 16" check is to make sure that the byte at index zero doesn't exceed a value of 16, which would be invalid hex. Is this correct?

No, it's because only versions 0 - 16 are specified. Other versions might behave entirely differently. (As an aside, if it were a comparison for the sake of clamping to the range of a single hex character the test would be >15 not >16).

G. Maxwell

Posted 2019-01-07T21:21:48.000

Reputation: 6 039

Ah, I see now. Also, dec.data[0] is a byte's worth of data (two hex chars), which can hold 256 different states. – kawthuldrok – 2019-01-09T17:37:39.567

From BIP173 > Specification > Segwit address format > Decoding:

Software interpreting a segwit address:

MUST verify that the human-readable part is "bc" for mainnet and "tb" for testnet.

MUST verify that the first decoded data value (the witness version) is between 0 and 16, inclusive

kawthuldrok

Posted 2019-01-07T21:21:48.000

Reputation: 83