foundation-devices / foundation-ur-py Goto Github PK

Hi!

Thanks for implementing this in Python!

Have you considered publishing the package on pypi.org?

Occasionally UR decoding fails to decode a fragment when it shouldn't. It fails to decode the fragment because the checksum included in the body does not match the calculated checksum. I believe the issue is that the calculated checksum is the wrong byte size. Here is an example fragment that reproduces the issue. Other UR libs appear to decode this fragment successfully.

fragment = "UR:CRYPTO-PSBT/20-29/LPBBCSCACFCMCPCYBBRSPTSKHDSSDTSBTKDECHRHPKHKVWMDMNKSGDAOAEAEAEAEAECHPTBBEODKPLETLBLDAMJOPMBNPEPLWMFZLTZTHGOEQZTELTLGAYCHAEAEAEAEAECHPTBBFXSAWKIDLTENBSKPJLFEFLMNCLPKEMTPWPMHMHIOLTKECSAMAEAEAEAEAECHPTBBINTKMWZMYNKNKEZCBGKTHDFEZOPAYNPRMEFTHPKGLTAEAEAEAEADADDNEMMDAAAEAEAEAEAECPAECXHDGWFDVTSPHDOLBNKIGETEECLKOSOXLPJSSNFXSGCLAHESJSVTURDYJZCWSRKNDTADAHTKGHCLAOFSTDLNUYSAASIESFDNRKHSMNJZTONLPSLDWFTDMNINOXEHHNKOTODRWPCHRORHDACLAEETWDVL"

ur_decoder = URDecoder()
status = ur_decoder.receive_part(fragment)

Bytewords.decode throws the exception 'Invalid Bytewords.' because the calculated crc32 checksum does not match the checksum on the body of the fragment. I believe the issue is that the calculated checksum is not returning from crc32_bytes as a 32 bit integer.

For the above example the checksum included in the fragment is 0038eae3 in hex. But the checksum returned by crc32_bytes is 38eae3. So on line 110 of bytewords.py the check if checksum != body_checksum: is false.

When I hard code the byte length in the crc32n function to be 4 bytes this seems to fix the issue, but I don't know what the "correct" way would be to handle this.

I changed:
return n.to_bytes((bit_length(n) + 7) // 8, 'big')

to:
return n.to_bytes(4, 'big')

and this makes this issue go away as far as I can tell