Samuel Hawksby-Robinson c8f9dad554
Feature/key compression (#1990)
## What has changed?

I've introduced to the public binding functionality that will compress and decompress public keys of a variety of encoding and key types. This functionality supports all major byte encoding formats and the following EC public key types:

- `secp256k1` pks
- `bls12-381 g1` pks
- `bls12-381 g2` pks

## Why make the change?

We want shorter public (chat) keys and we want to be future proof and encoding agnostic. See the issue here https://github.com/status-im/status-go/issues/1937

---

* Added basic signature for compresspk and uncompresspk

* Added basic encoding information

* make vendor

* formatted imports for the linter

* Reformatted imports hoping linter likes it

* This linter is capricious

* Added check that the secp256k1 key is valid

* Added test for valid key

* Added multiformat/go-varint dep

* Added public key type handling

* Added key decompression with key type handling

* Added handling for '0x' type indentifying

* Added more robust testing

* Less lint for the linting gods

* make vendor for bls12_381

* Added bls12-381 compression tests

* Added decompress key expected results

* Refactor of typed and untyped keys in tests

* Lint god appeasment

* Refactor of sample public keys

* Implemented bls12-381 decompression

* gofmt

* Renamed decode/encode funcs to be more descriptive

* Added binary bindings for key de/compression

* Refactor of func parameters

gomobile is a bit tempermental using raw bytes as a parameter, so I've decided to use string only inputs and outputs

* gofmt

* Added function documentation

* Moved multiformat de/compression into api/multiformat ns

* Moved multiformat de/compression into api/multiformat ns

* Changed compress to serialize on API
2020-06-23 11:47:17 +01:00

246 lines
4.2 KiB
Go

package bls12381
import (
"errors"
"math/big"
)
type fp2Temp struct {
t [4]*fe
}
type fp2 struct {
fp2Temp
}
func newFp2Temp() fp2Temp {
t := [4]*fe{}
for i := 0; i < len(t); i++ {
t[i] = &fe{}
}
return fp2Temp{t}
}
func newFp2() *fp2 {
t := newFp2Temp()
return &fp2{t}
}
func (e *fp2) fromBytes(in []byte) (*fe2, error) {
if len(in) != 96 {
return nil, errors.New("input string should be larger than 96 bytes")
}
c1, err := fromBytes(in[:48])
if err != nil {
return nil, err
}
c0, err := fromBytes(in[48:])
if err != nil {
return nil, err
}
return &fe2{*c0, *c1}, nil
}
func (e *fp2) toBytes(a *fe2) []byte {
out := make([]byte, 96)
copy(out[:48], toBytes(&a[1]))
copy(out[48:], toBytes(&a[0]))
return out
}
func (e *fp2) new() *fe2 {
return new(fe2).zero()
}
func (e *fp2) zero() *fe2 {
return new(fe2).zero()
}
func (e *fp2) one() *fe2 {
return new(fe2).one()
}
func (e *fp2) fromMont(c, a *fe2) {
fromMont(&c[0], &a[0])
fromMont(&c[1], &a[1])
}
func (e *fp2) add(c, a, b *fe2) {
add(&c[0], &a[0], &b[0])
add(&c[1], &a[1], &b[1])
}
func (e *fp2) addAssign(a, b *fe2) {
addAssign(&a[0], &b[0])
addAssign(&a[1], &b[1])
}
func (e *fp2) ladd(c, a, b *fe2) {
ladd(&c[0], &a[0], &b[0])
ladd(&c[1], &a[1], &b[1])
}
func (e *fp2) double(c, a *fe2) {
double(&c[0], &a[0])
double(&c[1], &a[1])
}
func (e *fp2) doubleAssign(a *fe2) {
doubleAssign(&a[0])
doubleAssign(&a[1])
}
func (e *fp2) ldouble(c, a *fe2) {
ldouble(&c[0], &a[0])
ldouble(&c[1], &a[1])
}
func (e *fp2) sub(c, a, b *fe2) {
sub(&c[0], &a[0], &b[0])
sub(&c[1], &a[1], &b[1])
}
func (e *fp2) subAssign(c, a *fe2) {
subAssign(&c[0], &a[0])
subAssign(&c[1], &a[1])
}
func (e *fp2) neg(c, a *fe2) {
neg(&c[0], &a[0])
neg(&c[1], &a[1])
}
func (e *fp2) conjugate(c, a *fe2) {
c[0].set(&a[0])
neg(&c[1], &a[1])
}
func (e *fp2) mul(c, a, b *fe2) {
t := e.t
mul(t[1], &a[0], &b[0])
mul(t[2], &a[1], &b[1])
add(t[0], &a[0], &a[1])
add(t[3], &b[0], &b[1])
sub(&c[0], t[1], t[2])
addAssign(t[1], t[2])
mul(t[0], t[0], t[3])
sub(&c[1], t[0], t[1])
}
func (e *fp2) mulAssign(a, b *fe2) {
t := e.t
mul(t[1], &a[0], &b[0])
mul(t[2], &a[1], &b[1])
add(t[0], &a[0], &a[1])
add(t[3], &b[0], &b[1])
sub(&a[0], t[1], t[2])
addAssign(t[1], t[2])
mul(t[0], t[0], t[3])
sub(&a[1], t[0], t[1])
}
func (e *fp2) square(c, a *fe2) {
t := e.t
ladd(t[0], &a[0], &a[1])
sub(t[1], &a[0], &a[1])
ldouble(t[2], &a[0])
mul(&c[0], t[0], t[1])
mul(&c[1], t[2], &a[1])
}
func (e *fp2) squareAssign(a *fe2) {
t := e.t
ladd(t[0], &a[0], &a[1])
sub(t[1], &a[0], &a[1])
ldouble(t[2], &a[0])
mul(&a[0], t[0], t[1])
mul(&a[1], t[2], &a[1])
}
func (e *fp2) mulByNonResidue(c, a *fe2) {
t := e.t
sub(t[0], &a[0], &a[1])
add(&c[1], &a[0], &a[1])
c[0].set(t[0])
}
func (e *fp2) mulByB(c, a *fe2) {
t := e.t
double(t[0], &a[0])
double(t[1], &a[1])
doubleAssign(t[0])
doubleAssign(t[1])
sub(&c[0], t[0], t[1])
add(&c[1], t[0], t[1])
}
func (e *fp2) inverse(c, a *fe2) {
t := e.t
square(t[0], &a[0])
square(t[1], &a[1])
addAssign(t[0], t[1])
inverse(t[0], t[0])
mul(&c[0], &a[0], t[0])
mul(t[0], t[0], &a[1])
neg(&c[1], t[0])
}
func (e *fp2) mulByFq(c, a *fe2, b *fe) {
mul(&c[0], &a[0], b)
mul(&c[1], &a[1], b)
}
func (e *fp2) exp(c, a *fe2, s *big.Int) {
z := e.one()
for i := s.BitLen() - 1; i >= 0; i-- {
e.square(z, z)
if s.Bit(i) == 1 {
e.mul(z, z, a)
}
}
c.set(z)
}
func (e *fp2) frobeniousMap(c, a *fe2, power uint) {
c[0].set(&a[0])
if power%2 == 1 {
neg(&c[1], &a[1])
return
}
c[1].set(&a[1])
}
func (e *fp2) frobeniousMapAssign(a *fe2, power uint) {
if power%2 == 1 {
neg(&a[1], &a[1])
return
}
}
func (e *fp2) sqrt(c, a *fe2) bool {
u, x0, a1, alpha := &fe2{}, &fe2{}, &fe2{}, &fe2{}
u.set(a)
e.exp(a1, a, pMinus3Over4)
e.square(alpha, a1)
e.mul(alpha, alpha, a)
e.mul(x0, a1, a)
if alpha.equal(negativeOne2) {
neg(&c[0], &x0[1])
c[1].set(&x0[0])
return true
}
e.add(alpha, alpha, e.one())
e.exp(alpha, alpha, pMinus1Over2)
e.mul(c, alpha, x0)
e.square(alpha, c)
return alpha.equal(u)
}
func (e *fp2) isQuadraticNonResidue(a *fe2) bool {
c0, c1 := new(fe), new(fe)
square(c0, &a[0])
square(c1, &a[1])
add(c1, c1, c0)
return isQuadraticNonResidue(c1)
}