Add "len" methods to encode and decode
To get length of encoded/decoded string
This commit is contained in:
@@ -3,10 +3,15 @@ package encoding
|
||||
import (
|
||||
"bytes"
|
||||
"fmt"
|
||||
"log"
|
||||
)
|
||||
|
||||
type GSM7Coder struct{}
|
||||
|
||||
// Make sure buffer can fit EncodesInto bytes
|
||||
// Otherwise Encode will allocate memory as it sees fit
|
||||
// Which is fine but not optimal
|
||||
// Preallocate the buffer with the size of EncodesInto bytes
|
||||
func (c *GSM7Coder) Encode(s string, buf *bytes.Buffer) error {
|
||||
// utf8 := *(*[]byte)(unsafe.Pointer(&s))
|
||||
utf8 := []byte(s)
|
||||
@@ -15,10 +20,6 @@ func (c *GSM7Coder) Encode(s string, buf *bytes.Buffer) error {
|
||||
bitshift byte = 0
|
||||
leap, shift bool
|
||||
)
|
||||
tbw := len(utf8) * 7 / 8
|
||||
if buf.Available() < tbw {
|
||||
buf.Grow(tbw)
|
||||
}
|
||||
|
||||
for index, septet := range utf8 {
|
||||
if septet > 0b01111111 {
|
||||
@@ -62,17 +63,18 @@ func (c *GSM7Coder) Encode(s string, buf *bytes.Buffer) error {
|
||||
offset = 1
|
||||
}
|
||||
}
|
||||
log.Println(buf.Cap(), buf.Len())
|
||||
return nil
|
||||
}
|
||||
|
||||
func (c *GSM7Coder) Decode(buf *bytes.Buffer) (string, error) {
|
||||
gsm7 := buf.Bytes()
|
||||
var (
|
||||
offset int = (len(gsm7) / 8) + 1
|
||||
bitshift byte = 0
|
||||
offset int = (len(gsm7) / 8) + 1
|
||||
bitshift byte = 0
|
||||
leap, shift bool
|
||||
)
|
||||
outLength := len(gsm7)*8/7
|
||||
outLength := DecodesInto(buf)
|
||||
lengthDiff := outLength - len(gsm7)
|
||||
gsm7 = append(gsm7, make([]byte, lengthDiff)...)
|
||||
|
||||
@@ -130,7 +132,7 @@ func (c *GSM7Coder) Decode(buf *bytes.Buffer) (string, error) {
|
||||
}
|
||||
// log.Printf("Result: %+v", gsm7)
|
||||
// for _, v := range gsm7 {
|
||||
// log.Printf("%08b", v)
|
||||
// log.Printf("%08b", v)
|
||||
// }
|
||||
return string(gsm7), nil
|
||||
}
|
||||
@@ -138,6 +140,20 @@ func (c *GSM7Coder) Decode(buf *bytes.Buffer) (string, error) {
|
||||
// Allocation free
|
||||
// Which means data MUST have space for value
|
||||
func InsertAt(data *[]byte, index int, value byte) {
|
||||
copy((*data)[index+1:], (*data)[index:])
|
||||
(*data)[index] = value
|
||||
copy((*data)[index+1:], (*data)[index:])
|
||||
(*data)[index] = value
|
||||
}
|
||||
|
||||
func EncodesInto(s *string) int {
|
||||
slen := len(*s)
|
||||
enclen := slen * 7 / 8
|
||||
if slen%8 != 0 {
|
||||
enclen++
|
||||
}
|
||||
return enclen
|
||||
}
|
||||
func DecodesInto(buf *bytes.Buffer) int {
|
||||
blen := buf.Len()
|
||||
declen := blen * 8 / 7
|
||||
return declen
|
||||
}
|
||||
|
@@ -219,7 +219,62 @@ func TestDeletesLastValue(t *testing.T) {
|
||||
}
|
||||
}
|
||||
|
||||
// benchmark
|
||||
// region misc tests
|
||||
func TestGSM7EncodesIntoSmallString(t *testing.T) {
|
||||
input := "Sunshine"
|
||||
expected := 7
|
||||
actual := EncodesInto(&input)
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
func TestGSM7EncodesIntoLargerNot8nString(t *testing.T) {
|
||||
input := "Golden rays play, Chasing night away."
|
||||
expected := 33
|
||||
actual := EncodesInto(&input)
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
func TestGSM7EncodesIntoLarger8nString(t *testing.T) {
|
||||
input := "Ducks are fucking great, they quacks, O quackers, what the fuck."
|
||||
expected := 56
|
||||
actual := EncodesInto(&input)
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
func TestGSM7DecodesIntoSmallString(t *testing.T) {
|
||||
input := []byte{0b11010011, 0b10111010, 0b01111011, 0b10001110, 0b01001110, 0b10111011, 0b11001011}
|
||||
expected := 8
|
||||
actual := DecodesInto(bytes.NewBuffer(input))
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
func TestGSM7DecodesIntoLargerNot8nString(t *testing.T) {
|
||||
input := []byte{0b11000111, 0b00110111, 0b10011011, 0b01011100, 0b01110110, 0b10000011, 0b11100100, 0b11100001, 0b11111100, 0b00011100, 0b00000100, 0b01100111, 0b10000111, 0b11110011, 0b00101100, 0b11010000, 0b00010000, 0b00011101, 0b10011110, 0b10100111, 0b11011101, 0b01100111, 0b10010000, 0b00111011, 0b01111101, 0b01000110, 0b11010011, 0b01000001, 0b11100001, 0b01111011, 0b00111000, 0b11101111, 0b00000010}
|
||||
expected := 37
|
||||
actual := DecodesInto(bytes.NewBuffer(input))
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
func TestGSM7DecodesIntoLarger8nString(t *testing.T) {
|
||||
input := []byte{0b11000100, 0b11111010, 0b01111000, 0b00111101, 0b00000111, 0b10000101, 0b11100101, 0b01100101, 0b10010000, 0b10111001, 0b00111110, 0b01011110, 0b10100111, 0b11011101, 0b01100111, 0b11010000, 0b01011001, 0b01011110, 0b00001110, 0b11010011, 0b01011001, 0b00100000, 0b00111010, 0b10111010, 0b10011100, 0b00000111, 0b11000101, 0b11101011, 0b11100001, 0b11110001, 0b01111010, 0b11001110, 0b00000010, 0b00111101, 0b01000001, 0b11110001, 0b01111010, 0b01111000, 0b10111100, 0b00101110, 0b11001011, 0b11100111, 0b00101100, 0b11010000, 0b00011101, 0b00011101, 0b10100110, 0b10000011, 0b11101000, 0b11101000, 0b00110010, 0b11001000, 0b01011100, 0b00011111, 0b10101111, 0b01011101}
|
||||
expected := 64
|
||||
actual := DecodesInto(bytes.NewBuffer(input))
|
||||
if actual != expected {
|
||||
t.Errorf("Expected %d, but got %d", expected, actual)
|
||||
}
|
||||
}
|
||||
|
||||
// region benchmark
|
||||
func BenchmarkGSM7EncodeSimpleASCIIString(b *testing.B) {
|
||||
coder := &GSM7Coder{}
|
||||
var buf bytes.Buffer
|
||||
|
Reference in New Issue
Block a user