a71aecec52
* Split wallet functionality into multiple traits In this commit we refactor LockedWallet into multiple traits that provide functionality related to a subset of total wallet functionality. This has the benefit of making it clear which methods are helper methods that are only intended to be used in a very specific setting, and which methods are part of the internal wallet API that other parts of the wallet can use. * Rework TransactionOutput and TransactionOutPoint to case classes * Add extension methods for flattening lists of assertions * Segregate confirmed and unconfirmed balance methods * Add test for FutureUtil.sequentially * Add trace logging of balance fetching * Process outgoing TXOs Move TX processing into separate trait, add internal API method Unify DB representation of TXOs Prior to this commit we stored TXO information across diferent tables, with joins and tuples needed a bunch of places to keep track of everything we needed. In this commit we unify the tables, leaving us with only one table for TXOs. |
||
---|---|---|
.. | ||
src/main/scala/org/bitcoins/wallet | ||
README.md |
wallet
This is meant to be a stand alone project that can be used as a cold storage wallet and hot wallet.
Features
- utxo storage
- key storage
- key generation
- coin selection
- transaction building
- fee calculation
Design choices
- Private key material is just stored once, as the mnemonic code used to initialize the wallet
- Addresses we hand out to users are stored with their BIP44/BIP49/BIP84 paths and script types, so that everything we need for spending the money sent to an address is derivable.
- The wallet is a "dumb" wallet that acts mostly as a database of UTXOs, transactions and
addresses, with associated operations on these.
The wallet module does very little verification of incoming data about transactions,
UTXOs and reorgs. We're aiming to write small, self contained modules, that can be
composed together into more fully fledged systems. That means the
chain
andnode
modules does the actual verification of data we receive, andwallet
just blindly acts on this. This results in a design where you can swap outnode
for a Bitcoin Core full node, use it with hardware wallets, or something else entirely. However, that also means that users ofwallet
that doesn't want to use the other modules we provide have to make sure that the data they are feeding the wallet is correct.
Database structure
We store information in the following tables:
- TXOs - Contains both the information needed to spent it as well as information related to wallet state (confirmations, spent/unspent etc)
- Addresses - must reference the account it belongs to
- Accounts
Mnemonic encryption
The mnemonic seed to the Bitcoin-S wallet is written to disk, encrypted. The file name is
$HOME/.bitcoin-s/$NETWORK/encrypted_bitcoin-s_seed.json
. We store it in a JSON object
that looks like this:
{
"iv": "initializationVector",
"cipherText": "encryptedCipherText",
"salt": "saltUsedInEncryption"
}
The parts that's relevant to this part of the wallet is WalletStorage.scala
(where we handle
the actual reading from and writing to disk), EncryptedMnemonic.scala
(where we convert an
encrypted mnemonic to a cleartext mnemonic) and AesCrypt.scala
(where do the actual
encryption/decryption).
We use AES encryption for this, block cipher mode and PKCS5 padding. The wallet password is fed
into the PBKDF2 key stretching function, using SHA512 as the HMAC function. This happens in
PBKDF2.scala
.