block-accounting/backend/internal/pkg/bip39/bip39.go

package bip39

import (
	"crypto/rand"
	"crypto/sha256"
	"crypto/sha512"
	"encoding/binary"
	"errors"
	"fmt"
	"math/big"
	"strings"

	"golang.org/x/crypto/pbkdf2"
)

var (
	// Some bitwise operands for working with big.Ints.
	last11BitsMask  = big.NewInt(2047)
	shift11BitsMask = big.NewInt(2048)
	bigOne          = big.NewInt(1)
	bigTwo          = big.NewInt(2)

	// wordLengthChecksumMasksMapping is used to isolate the checksum bits from
	//the entropy+checksum byte array.
	wordLengthChecksumMasksMapping = map[int]*big.Int{
		12: big.NewInt(15),
		15: big.NewInt(31),
		18: big.NewInt(63),
		21: big.NewInt(127),
		24: big.NewInt(255),
	}

	// wordLengthChecksumShiftMapping is used to lookup the number of operand
	// for shifting bits to handle checksums.
	wordLengthChecksumShiftMapping = map[int]*big.Int{
		12: big.NewInt(16),
		15: big.NewInt(8),
		18: big.NewInt(4),
		21: big.NewInt(2),
	}

	// wordList is the set of words to use.
	wordList []string

	// wordMap is a reverse lookup map for wordList.
	wordMap map[string]int
)

var (
	// ErrInvalidMnemonic is returned when trying to use a malformed mnemonic.
	ErrInvalidMnemonic = errors.New("Invalid mnenomic")

	// ErrEntropyLengthInvalid is returned when trying to use an entropy set with
	// an invalid size.
	ErrEntropyLengthInvalid = errors.New("Entropy length must be [128, 256] and a multiple of 32")

	// ErrValidatedSeedLengthMismatch is returned when a validated seed is not the
	// same size as the given seed. This should never happen is present only as a
	// sanity assertion.
	ErrValidatedSeedLengthMismatch = errors.New("Seed length does not match validated seed length")

	// ErrChecksumIncorrect is returned when entropy has the incorrect checksum.
	ErrChecksumIncorrect = errors.New("Checksum incorrect")
)

func init() {
	SetWordList(English)
}

// SetWordList sets the list of words to use for mnemonics. Currently the list
// that is set is used package-wide.
func SetWordList(list []string) {
	wordList = list
	wordMap = map[string]int{}

	for i, v := range wordList {
		wordMap[v] = i
	}
}

// GetWordList gets the list of words to use for mnemonics.
func GetWordList() []string {
	return wordList
}

// GetWordIndex gets word index in wordMap.
func GetWordIndex(word string) (int, bool) {
	idx, ok := wordMap[word]
	return idx, ok
}

// NewEntropy will create random entropy bytes
// so long as the requested size bitSize is an appropriate size.
//
// bitSize has to be a multiple 32 and be within the inclusive range of {128, 256}.
func NewEntropy(bitSize int) ([]byte, error) {
	if err := validateEntropyBitSize(bitSize); err != nil {
		return nil, err
	}

	entropy := make([]byte, bitSize/8)
	_, _ = rand.Read(entropy) // err is always nil

	return entropy, nil
}

// EntropyFromMnemonic takes a mnemonic generated by this library,
// and returns the input entropy used to generate the given mnemonic.
// An error is returned if the given mnemonic is invalid.
func EntropyFromMnemonic(mnemonic string) ([]byte, error) {
	mnemonicSlice, isValid := splitMnemonicWords(mnemonic)
	if !isValid {
		return nil, ErrInvalidMnemonic
	}

	// Decode the words into a big.Int.
	var (
		wordBytes [2]byte
		b         = big.NewInt(0)
	)

	for _, v := range mnemonicSlice {
		index, found := wordMap[v]
		if !found {
			return nil, fmt.Errorf("word `%v` not found in reverse map", v)
		}

		binary.BigEndian.PutUint16(wordBytes[:], uint16(index))
		b.Mul(b, shift11BitsMask)
		b.Or(b, big.NewInt(0).SetBytes(wordBytes[:]))
	}

	// Build and add the checksum to the big.Int.
	checksum := big.NewInt(0)
	checksumMask := wordLengthChecksumMasksMapping[len(mnemonicSlice)]
	checksum = checksum.And(b, checksumMask)

	b.Div(b, big.NewInt(0).Add(checksumMask, bigOne))

	// The entropy is the underlying bytes of the big.Int. Any upper bytes of
	// all 0's are not returned so we pad the beginning of the slice with empty
	// bytes if necessary.
	entropy := b.Bytes()
	entropy = padByteSlice(entropy, len(mnemonicSlice)/3*4)

	// Generate the checksum and compare with the one we got from the mneomnic.
	entropyChecksumBytes := computeChecksum(entropy)
	entropyChecksum := big.NewInt(int64(entropyChecksumBytes[0]))

	if l := len(mnemonicSlice); l != 24 {
		checksumShift := wordLengthChecksumShiftMapping[l]
		entropyChecksum.Div(entropyChecksum, checksumShift)
	}

	if checksum.Cmp(entropyChecksum) != 0 {
		return nil, ErrChecksumIncorrect
	}

	return entropy, nil
}

// NewMnemonic will return a string consisting of the mnemonic words for
// the given entropy.
// If the provide entropy is invalid, an error will be returned.
func NewMnemonic(entropy []byte) (string, error) {
	// Compute some lengths for convenience.
	entropyBitLength := len(entropy) * 8
	checksumBitLength := entropyBitLength / 32
	sentenceLength := (entropyBitLength + checksumBitLength) / 11

	// Validate that the requested size is supported.
	err := validateEntropyBitSize(entropyBitLength)
	if err != nil {
		return "", err
	}

	// Add checksum to entropy.
	entropy = addChecksum(entropy)

	// Break entropy up into sentenceLength chunks of 11 bits.
	// For each word AND mask the rightmost 11 bits and find the word at that index.
	// Then bitshift entropy 11 bits right and repeat.
	// Add to the last empty slot so we can work with LSBs instead of MSB.

	// Entropy as an int so we can bitmask without worrying about bytes slices.
	entropyInt := new(big.Int).SetBytes(entropy)

	// Slice to hold words in.
	words := make([]string, sentenceLength)

	// Throw away big.Int for AND masking.
	word := big.NewInt(0)

	for i := sentenceLength - 1; i >= 0; i-- {
		// Get 11 right most bits and bitshift 11 to the right for next time.
		word.And(entropyInt, last11BitsMask)
		entropyInt.Div(entropyInt, shift11BitsMask)

		// Get the bytes representing the 11 bits as a 2 byte slice.
		wordBytes := padByteSlice(word.Bytes(), 2)

		// Convert bytes to an index and add that word to the list.
		words[i] = wordList[binary.BigEndian.Uint16(wordBytes)]
	}

	return strings.Join(words, " "), nil
}

// MnemonicToByteArray takes a mnemonic string and turns it into a byte array
// suitable for creating another mnemonic.
// An error is returned if the mnemonic is invalid.
func MnemonicToByteArray(mnemonic string, raw ...bool) ([]byte, error) {
	var (
		mnemonicSlice   = strings.Split(mnemonic, " ")
		entropyBitSize  = len(mnemonicSlice) * 11
		checksumBitSize = entropyBitSize % 32
		fullByteSize    = (entropyBitSize-checksumBitSize)/8 + 1
	)

	// Turn into raw entropy.
	rawEntropyBytes, err := EntropyFromMnemonic(mnemonic)
	if err != nil {
		return nil, err
	}

	// If we want the raw entropy then we're done.
	if len(raw) > 0 && raw[0] {
		return rawEntropyBytes, nil
	}

	// Otherwise add the checksum before returning
	return padByteSlice(addChecksum(rawEntropyBytes), fullByteSize), nil
}

// NewSeedWithErrorChecking creates a hashed seed output given the mnemonic string and a password.
// An error is returned if the mnemonic is not convertible to a byte array.
func NewSeedWithErrorChecking(mnemonic string, password string) ([]byte, error) {
	_, err := MnemonicToByteArray(mnemonic)
	if err != nil {
		return nil, err
	}

	return NewSeed(mnemonic, password), nil
}

// NewSeed creates a hashed seed output given a provided string and password.
// No checking is performed to validate that the string provided is a valid mnemonic.
func NewSeed(mnemonic string, password string) []byte {
	return pbkdf2.Key([]byte(mnemonic), []byte("mnemonic"+password), 2048, 64, sha512.New)
}

// IsMnemonicValid attempts to verify that the provided mnemonic is valid.
// Validity is determined by both the number of words being appropriate,
// and that all the words in the mnemonic are present in the word list.
func IsMnemonicValid(mnemonic string) bool {
	_, err := EntropyFromMnemonic(mnemonic)
	return err == nil
}

// Appends to data the first (len(data) / 32)bits of the result of sha256(data)
// Currently only supports data up to 32 bytes.
func addChecksum(data []byte) []byte {
	// Get first byte of sha256
	hash := computeChecksum(data)
	firstChecksumByte := hash[0]

	// len() is in bytes so we divide by 4
	checksumBitLength := uint(len(data) / 4)

	// For each bit of check sum we want we shift the data one the left
	// and then set the (new) right most bit equal to checksum bit at that index
	// staring from the left
	dataBigInt := new(big.Int).SetBytes(data)

	for i := uint(0); i < checksumBitLength; i++ {
		// Bitshift 1 left
		dataBigInt.Mul(dataBigInt, bigTwo)

		// Set rightmost bit if leftmost checksum bit is set
		if firstChecksumByte&(1<<(7-i)) > 0 {
			dataBigInt.Or(dataBigInt, bigOne)
		}
	}

	return dataBigInt.Bytes()
}

func computeChecksum(data []byte) []byte {
	hasher := sha256.New()
	_, _ = hasher.Write(data) // This error is guaranteed to be nil

	return hasher.Sum(nil)
}

// validateEntropyBitSize ensures that entropy is the correct size for being a
// mnemonic.
func validateEntropyBitSize(bitSize int) error {
	if (bitSize%32) != 0 || bitSize < 128 || bitSize > 256 {
		return ErrEntropyLengthInvalid
	}

	return nil
}

// padByteSlice returns a byte slice of the given size with contents of the
// given slice left padded and any empty spaces filled with 0's.
func padByteSlice(slice []byte, length int) []byte {
	offset := length - len(slice)
	if offset <= 0 {
		return slice
	}

	newSlice := make([]byte, length)
	copy(newSlice[offset:], slice)

	return newSlice
}

// compareByteSlices returns true of the byte slices have equal contents and
// returns false otherwise.
func compareByteSlices(a, b []byte) bool {
	if len(a) != len(b) {
		return false
	}

	for i := range a {
		if a[i] != b[i] {
			return false
		}
	}

	return true
}

func splitMnemonicWords(mnemonic string) ([]string, bool) {
	// Create a list of all the words in the mnemonic sentence
	words := strings.Fields(mnemonic)

	// Get num of words
	numOfWords := len(words)

	// The number of words should be 12, 15, 18, 21 or 24
	if numOfWords%3 != 0 || numOfWords < 12 || numOfWords > 24 {
		return nil, false
	}

	return words, true
}