I am using big.NewInt(int64(e)).Bytes() to convert an int32 to a byte array. Is there a more elegant way to do this?
I expect AQAB to be the base64 encoded value of e
http://play.golang.org/p/M46X7OpZpu
const e = 65537
func base64Encode(b []byte) string {
return strings.TrimRight(base64.StdEncoding.EncodeToString(b), "=")
}
func main() {
fmt.Printf("exp %d\n", e)
b := make([]byte, 4)
binary.BigEndian.PutUint32(b, e)
fmt.Printf("b: BigEndian.PutUint32 %x (Bad) %s\n", b, base64Encode(b))
b2 := make([]byte, 4)
binary.BigEndian.PutUint32(b2, e)
for i := range b2 {
if b2[i] != 0 {
b2 = b2[i:]
break
}
}
fmt.Printf("b2: BigEndian.PutUint32 %x (Good) %s\n", b2, base64Encode(b2))
b4 := big.NewInt(int64(e)).Bytes()
fmt.Printf("b4: big.NewInt(int64(e)).Bytes() %x (Good) %s\n", b4, base64Encode(b4))
}
Output:
exp 65537
b: BigEndian.PutUint32 00010001 (Bad) AAEAAQ
b2: BigEndian.PutUint32 010001 (Good) AQAB
b4: big.NewInt(int64(e)).Bytes() 010001 (Good) AQAB
exp 1
b: BigEndian.PutUint32 00000001 (Bad) AAAAAQ
b2: BigEndian.PutUint32 01 (Good) AQ
b4: big.NewInt(int64(e)).Bytes() 01 (Good) AQ
exp 1000000
b: BigEndian.PutUint32 000f4240 (Bad) AA9CQA
b2: BigEndian.PutUint32 0f4240 (Good) D0JA
b4: big.NewInt(int64(e)).Bytes() 0f4240 (Good) D0JA
Edit:
I've benchmarked b2 and b4:
b2 1000000000 68.1 ns/op 8 B/op 1 allocs/op
b4 200000000 248 ns/op 90 B/op 3 allocs/op
I'll use b2 for now...
An int value can be converted into bytes by using the method int. to_bytes().
The byte type in Golang is an alias for the unsigned integer 8 type ( uint8 ). The byte type is only used to semantically distinguish between an unsigned integer 8 and a byte. The range of a byte is 0 to 255 (same as uint8 ).
For this kind of task I think your first options should always be using encoding/binary
and, if that is insufficient, bitwise math. However, in some cases the overhead of copying data is too large or these safe solutions are too slow:
While I would not call it elegant you can use Go's unsafe
and reflect
* packages to do this very quickly. Just remember, this does not copy the data; rather, it just gives you another "view" of it. And being well- unsafe means that you need to be very careful- (hello units tests and code review) and keep in mind you are breaking Go's memory safety. However, when execution speed is the dominating concern and your team agrees unsafe
is warranted, unsafe
can seldom be beat.
const BYTES_IN_INT32 = 4
func UnsafeCaseInt32ToBytes(val int32) []byte {
hdr := reflect.SliceHeader{Data: uintptr(unsafe.Pointer(&val)), Len: BYTES_IN_INT32, Cap: BYTES_IN_INT32}
return *(*[]byte)(unsafe.Pointer(&hdr))
}
func UnsafeCastInt32sToBytes(ints []int32) []byte {
length := len(ints) * BYTES_IN_INT32
hdr := reflect.SliceHeader{Data: uintptr(unsafe.Pointer(&ints[0])), Len: length, Cap: length}
return *(*[]byte)(unsafe.Pointer(&hdr))
}
*Note: You may want to use SizeOf
rather than a constant. I like the constant better.
Update: here are some benchmark results:
BenchmarkB2 20000000 88.7 ns/op
BenchmarkB4 5000000 309 ns/op
BenchmarkUnsafe 1000000000 2.25 ns/op
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With