2015-07-01 00:09:02 +00:00
|
|
|
// Copyright 2015 The go-ethereum Authors
|
|
|
|
// This file is part of the go-ethereum library.
|
|
|
|
//
|
|
|
|
// The go-ethereum library is free software: you can redistribute it and/or modify
|
|
|
|
// it under the terms of the GNU Lesser General Public License as published by
|
|
|
|
// the Free Software Foundation, either version 3 of the License, or
|
|
|
|
// (at your option) any later version.
|
|
|
|
//
|
|
|
|
// The go-ethereum library is distributed in the hope that it will be useful,
|
|
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
// GNU Lesser General Public License for more details.
|
|
|
|
//
|
|
|
|
// You should have received a copy of the GNU Lesser General Public License
|
|
|
|
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
|
|
|
|
package rlp
|
|
|
|
|
2015-09-09 01:34:15 +00:00
|
|
|
import (
|
|
|
|
"io"
|
|
|
|
"reflect"
|
|
|
|
)
|
|
|
|
|
|
|
|
// RawValue represents an encoded RLP value and can be used to delay
|
2017-06-12 12:45:17 +00:00
|
|
|
// RLP decoding or to precompute an encoding. Note that the decoder does
|
2015-09-09 01:34:15 +00:00
|
|
|
// not verify whether the content of RawValues is valid RLP.
|
|
|
|
type RawValue []byte
|
|
|
|
|
|
|
|
var rawValueType = reflect.TypeOf(RawValue{})
|
2015-07-01 00:09:02 +00:00
|
|
|
|
2022-11-07 18:19:02 +00:00
|
|
|
// StringSize returns the encoded size of a string.
|
|
|
|
func StringSize(s string) uint64 {
|
|
|
|
switch {
|
|
|
|
case len(s) == 0:
|
|
|
|
return 1
|
|
|
|
case len(s) == 1:
|
|
|
|
if s[0] <= 0x7f {
|
|
|
|
return 1
|
|
|
|
} else {
|
|
|
|
return 2
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
return uint64(headsize(uint64(len(s))) + len(s))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// BytesSize returns the encoded size of a byte slice.
|
|
|
|
func BytesSize(b []byte) uint64 {
|
|
|
|
switch {
|
|
|
|
case len(b) == 0:
|
|
|
|
return 1
|
|
|
|
case len(b) == 1:
|
|
|
|
if b[0] <= 0x7f {
|
|
|
|
return 1
|
|
|
|
} else {
|
|
|
|
return 2
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
return uint64(headsize(uint64(len(b))) + len(b))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-09-10 12:48:19 +00:00
|
|
|
// ListSize returns the encoded size of an RLP list with the given
|
|
|
|
// content size.
|
|
|
|
func ListSize(contentSize uint64) uint64 {
|
|
|
|
return uint64(headsize(contentSize)) + contentSize
|
|
|
|
}
|
|
|
|
|
2022-11-07 18:19:02 +00:00
|
|
|
// IntSize returns the encoded size of the integer x. Note: The return type of this
|
|
|
|
// function is 'int' for backwards-compatibility reasons. The result is always positive.
|
2021-05-27 08:19:13 +00:00
|
|
|
func IntSize(x uint64) int {
|
|
|
|
if x < 0x80 {
|
|
|
|
return 1
|
|
|
|
}
|
|
|
|
return 1 + intsize(x)
|
|
|
|
}
|
|
|
|
|
2015-07-01 00:09:02 +00:00
|
|
|
// Split returns the content of first RLP value and any
|
|
|
|
// bytes after the value as subslices of b.
|
|
|
|
func Split(b []byte) (k Kind, content, rest []byte, err error) {
|
|
|
|
k, ts, cs, err := readKind(b)
|
|
|
|
if err != nil {
|
|
|
|
return 0, nil, b, err
|
|
|
|
}
|
|
|
|
return k, b[ts : ts+cs], b[ts+cs:], nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// SplitString splits b into the content of an RLP string
|
|
|
|
// and any remaining bytes after the string.
|
|
|
|
func SplitString(b []byte) (content, rest []byte, err error) {
|
|
|
|
k, content, rest, err := Split(b)
|
|
|
|
if err != nil {
|
|
|
|
return nil, b, err
|
|
|
|
}
|
|
|
|
if k == List {
|
|
|
|
return nil, b, ErrExpectedString
|
|
|
|
}
|
|
|
|
return content, rest, nil
|
|
|
|
}
|
|
|
|
|
2020-09-14 17:23:01 +00:00
|
|
|
// SplitUint64 decodes an integer at the beginning of b.
|
|
|
|
// It also returns the remaining data after the integer in 'rest'.
|
|
|
|
func SplitUint64(b []byte) (x uint64, rest []byte, err error) {
|
|
|
|
content, rest, err := SplitString(b)
|
|
|
|
if err != nil {
|
|
|
|
return 0, b, err
|
|
|
|
}
|
|
|
|
switch {
|
|
|
|
case len(content) == 0:
|
|
|
|
return 0, rest, nil
|
|
|
|
case len(content) == 1:
|
|
|
|
if content[0] == 0 {
|
|
|
|
return 0, b, ErrCanonInt
|
|
|
|
}
|
|
|
|
return uint64(content[0]), rest, nil
|
|
|
|
case len(content) > 8:
|
|
|
|
return 0, b, errUintOverflow
|
|
|
|
default:
|
|
|
|
x, err = readSize(content, byte(len(content)))
|
|
|
|
if err != nil {
|
|
|
|
return 0, b, ErrCanonInt
|
|
|
|
}
|
|
|
|
return x, rest, nil
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-07-01 00:09:02 +00:00
|
|
|
// SplitList splits b into the content of a list and any remaining
|
|
|
|
// bytes after the list.
|
|
|
|
func SplitList(b []byte) (content, rest []byte, err error) {
|
|
|
|
k, content, rest, err := Split(b)
|
|
|
|
if err != nil {
|
|
|
|
return nil, b, err
|
|
|
|
}
|
|
|
|
if k != List {
|
|
|
|
return nil, b, ErrExpectedList
|
|
|
|
}
|
|
|
|
return content, rest, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// CountValues counts the number of encoded values in b.
|
|
|
|
func CountValues(b []byte) (int, error) {
|
|
|
|
i := 0
|
|
|
|
for ; len(b) > 0; i++ {
|
|
|
|
_, tagsize, size, err := readKind(b)
|
|
|
|
if err != nil {
|
|
|
|
return 0, err
|
|
|
|
}
|
|
|
|
b = b[tagsize+size:]
|
|
|
|
}
|
|
|
|
return i, nil
|
|
|
|
}
|
|
|
|
|
|
|
|
func readKind(buf []byte) (k Kind, tagsize, contentsize uint64, err error) {
|
|
|
|
if len(buf) == 0 {
|
|
|
|
return 0, 0, 0, io.ErrUnexpectedEOF
|
|
|
|
}
|
|
|
|
b := buf[0]
|
|
|
|
switch {
|
|
|
|
case b < 0x80:
|
|
|
|
k = Byte
|
|
|
|
tagsize = 0
|
|
|
|
contentsize = 1
|
|
|
|
case b < 0xB8:
|
|
|
|
k = String
|
|
|
|
tagsize = 1
|
|
|
|
contentsize = uint64(b - 0x80)
|
|
|
|
// Reject strings that should've been single bytes.
|
2017-12-11 21:47:10 +00:00
|
|
|
if contentsize == 1 && len(buf) > 1 && buf[1] < 128 {
|
2015-07-01 00:09:02 +00:00
|
|
|
return 0, 0, 0, ErrCanonSize
|
|
|
|
}
|
|
|
|
case b < 0xC0:
|
|
|
|
k = String
|
|
|
|
tagsize = uint64(b-0xB7) + 1
|
|
|
|
contentsize, err = readSize(buf[1:], b-0xB7)
|
|
|
|
case b < 0xF8:
|
|
|
|
k = List
|
|
|
|
tagsize = 1
|
|
|
|
contentsize = uint64(b - 0xC0)
|
|
|
|
default:
|
|
|
|
k = List
|
|
|
|
tagsize = uint64(b-0xF7) + 1
|
|
|
|
contentsize, err = readSize(buf[1:], b-0xF7)
|
|
|
|
}
|
|
|
|
if err != nil {
|
|
|
|
return 0, 0, 0, err
|
|
|
|
}
|
|
|
|
// Reject values larger than the input slice.
|
|
|
|
if contentsize > uint64(len(buf))-tagsize {
|
|
|
|
return 0, 0, 0, ErrValueTooLarge
|
|
|
|
}
|
|
|
|
return k, tagsize, contentsize, err
|
|
|
|
}
|
|
|
|
|
|
|
|
func readSize(b []byte, slen byte) (uint64, error) {
|
|
|
|
if int(slen) > len(b) {
|
|
|
|
return 0, io.ErrUnexpectedEOF
|
|
|
|
}
|
|
|
|
var s uint64
|
|
|
|
switch slen {
|
|
|
|
case 1:
|
|
|
|
s = uint64(b[0])
|
|
|
|
case 2:
|
|
|
|
s = uint64(b[0])<<8 | uint64(b[1])
|
|
|
|
case 3:
|
|
|
|
s = uint64(b[0])<<16 | uint64(b[1])<<8 | uint64(b[2])
|
|
|
|
case 4:
|
|
|
|
s = uint64(b[0])<<24 | uint64(b[1])<<16 | uint64(b[2])<<8 | uint64(b[3])
|
|
|
|
case 5:
|
|
|
|
s = uint64(b[0])<<32 | uint64(b[1])<<24 | uint64(b[2])<<16 | uint64(b[3])<<8 | uint64(b[4])
|
|
|
|
case 6:
|
|
|
|
s = uint64(b[0])<<40 | uint64(b[1])<<32 | uint64(b[2])<<24 | uint64(b[3])<<16 | uint64(b[4])<<8 | uint64(b[5])
|
|
|
|
case 7:
|
|
|
|
s = uint64(b[0])<<48 | uint64(b[1])<<40 | uint64(b[2])<<32 | uint64(b[3])<<24 | uint64(b[4])<<16 | uint64(b[5])<<8 | uint64(b[6])
|
|
|
|
case 8:
|
|
|
|
s = uint64(b[0])<<56 | uint64(b[1])<<48 | uint64(b[2])<<40 | uint64(b[3])<<32 | uint64(b[4])<<24 | uint64(b[5])<<16 | uint64(b[6])<<8 | uint64(b[7])
|
|
|
|
}
|
|
|
|
// Reject sizes < 56 (shouldn't have separate size) and sizes with
|
|
|
|
// leading zero bytes.
|
|
|
|
if s < 56 || b[0] == 0 {
|
|
|
|
return 0, ErrCanonSize
|
|
|
|
}
|
|
|
|
return s, nil
|
|
|
|
}
|
2020-11-04 18:29:24 +00:00
|
|
|
|
|
|
|
// AppendUint64 appends the RLP encoding of i to b, and returns the resulting slice.
|
|
|
|
func AppendUint64(b []byte, i uint64) []byte {
|
|
|
|
if i == 0 {
|
|
|
|
return append(b, 0x80)
|
|
|
|
} else if i < 128 {
|
|
|
|
return append(b, byte(i))
|
|
|
|
}
|
|
|
|
switch {
|
|
|
|
case i < (1 << 8):
|
|
|
|
return append(b, 0x81, byte(i))
|
|
|
|
case i < (1 << 16):
|
|
|
|
return append(b, 0x82,
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
case i < (1 << 24):
|
|
|
|
return append(b, 0x83,
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
case i < (1 << 32):
|
|
|
|
return append(b, 0x84,
|
|
|
|
byte(i>>24),
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
case i < (1 << 40):
|
|
|
|
return append(b, 0x85,
|
|
|
|
byte(i>>32),
|
|
|
|
byte(i>>24),
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
|
|
|
|
case i < (1 << 48):
|
|
|
|
return append(b, 0x86,
|
|
|
|
byte(i>>40),
|
|
|
|
byte(i>>32),
|
|
|
|
byte(i>>24),
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
case i < (1 << 56):
|
|
|
|
return append(b, 0x87,
|
|
|
|
byte(i>>48),
|
|
|
|
byte(i>>40),
|
|
|
|
byte(i>>32),
|
|
|
|
byte(i>>24),
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
|
|
|
|
default:
|
|
|
|
return append(b, 0x88,
|
|
|
|
byte(i>>56),
|
|
|
|
byte(i>>48),
|
|
|
|
byte(i>>40),
|
|
|
|
byte(i>>32),
|
|
|
|
byte(i>>24),
|
|
|
|
byte(i>>16),
|
|
|
|
byte(i>>8),
|
|
|
|
byte(i),
|
|
|
|
)
|
|
|
|
}
|
|
|
|
}
|