Consider Using wallycore for BIP38

[sigkill]

trafficstars

wallycore might be faster for BIP38 brute force attempts. I reviewed the code for btcrpass.py, and notice that it is quite similar to the functions that run in the "brute_force_bip38" projets code. The difference is that project is compiled in CPP, and has no viable tokenlist feature, so even with all the speed, you're stuck due to space constraints.

wallycore - https://wally.readthedocs.io/en/release_0.7.8/crypto/ is probably faster than the present implementation - I looked at btcrpass.py for some time and found that the code is drastically similar to the source code in this other project:

This is used in CPP form in this project: https://github.com/agurodriguez/brute_force_bip38

Specifically: ` def _return_verified_password_or_false_cpu(self, passwords): # BIP38 Encrypted Private Keys l_scrypt = pylibscrypt.scrypt

    passwords = map(lambda p: normalize("NFC", p).encode("utf_8", "ignore"), passwords)
    for count, password in enumerate(passwords, 1):
        if not self.ec_multiplied:
            scrypthash = l_scrypt(password, self.salt, 1 << 14, 8, 8, 64)
            if bip38decrypt_non_ec(scrypthash, self.enc_privkey, self.has_compression_flag, self.has_lotsequence_flag, network_prefix = self.network.prefix_address):
                return password.decode("utf_8", "replace"), count
        else:
            prefactor = l_scrypt(password, self.salt, 1 << 14, 8, 8, 32)
            passpoint = prefactor_to_passpoint(prefactor, self.has_lotsequence_flag, self.enc_privkey)
            encseedb = l_scrypt(passpoint, self.enc_privkey[0:12], 1024, 1, 1, 64)

            if bip38decrypt_ec(prefactor, encseedb, self.enc_privkey, self.has_compression_flag, self.has_lotsequence_flag, network_prefix = self.network.prefix_address):
                return password.decode("utf_8", "replace"), count

    return False, count`

If it were possible to import wallycore.wally_scrypt instead of pylibscrypt.scrypt we should likely see an improvement in passwords per second.

The code that is used in the other project waits for the WALLY_OK output:

Here: https://github.com/agurodriguez/brute_force_bip38/blob/master/src/bip38.hpp and here: https://github.com/agurodriguez/brute_force_bip38/blob/master/src/main.cpp

It is so much faster than btcrecover, but limited again due to the passwordlist and no tokenlist.

[sigkill]

wallycore appears to not support python3.9 - am trying with 3.6 now.