1
0
mirror of https://github.com/vlang/v.git synced 2023-08-10 21:13:21 +03:00
v/vlib/encoding/base64/base64.v

161 lines
4.6 KiB
V
Raw Normal View History

2019-06-22 21:20:28 +03:00
// Copyright (c) 2019 Alexander Medvednikov. All rights reserved.
// Use of this source code is governed by an MIT license
// that can be found in the LICENSE file.
module base64
const (
Index = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
2019-06-22 21:20:28 +03:00
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
62, 63, 62, 62, 63, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 0, 0, 0,
0, 0, 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 25, 0, 0, 0, 0, 63, 0, 26, 27, 28, 29,
30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46,
47, 48, 49, 50, 51]!!
EndingTable = [0, 2, 1]
EncodingTable = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/'
2019-06-22 21:20:28 +03:00
)
/**
* decode - expects a base64 encoded string. Returns its decoded version.
* @param data - the encoded input string.
* @return the decoded version of the input string data.
* NB: if you need to decode many strings repeatedly, take a look at decode_in_buffer too.
*/
2019-06-26 20:03:35 +03:00
pub fn decode(data string) string {
buffer := malloc( data.len * 3 / 4 )
2019-12-07 15:13:23 +03:00
return tos(buffer, decode_in_buffer(data, buffer) )
}
/**
* decode - expects a string. Returns its base64 encoded version.
* @param data - the input string.
2019-12-07 15:13:23 +03:00
* @return the base64 encoded version of the input string.
* NB: base64 encoding returns a string that is ~ 4/3 larger than the input.
* NB: if you need to encode many strings repeatedly, take a look at encode_in_buffer too.
*/
pub fn encode(data string) string {
buffer := malloc( 4 * ((data.len + 2) / 3) )
2019-12-07 15:13:23 +03:00
return tos(buffer, encode_in_buffer(data, buffer))
}
/**
* decode_in_buffer - expects a string reference, and a buffer in which to store its decoded version.
* @param data - a reference/pointer to the input string that will be decoded.
2019-12-07 15:13:23 +03:00
* @param buffer - a reference/pointer to the buffer that will hold the result.
* The buffer should be large enough (i.e. 3/4 of the data.len, or larger) to hold the decoded data.
* @return the actual size of the decoded data in the buffer.
* NB: this function does NOT allocate new memory, and is suitable for handling very large strings.
*/
2019-12-06 23:02:09 +03:00
pub fn decode_in_buffer(data &string, buffer byteptr) int {
mut padding := 0
if data.ends_with('=') {
if data.ends_with('==') {
padding = 2
} else {
padding = 1
}
2019-06-22 21:20:28 +03:00
}
//input_length is the length of meaningful data
input_length := data.len - padding
output_length := input_length * 3 / 4
mut i := 0
2019-06-22 21:20:28 +03:00
mut j := 0
mut b := &byte(0)
mut d := &byte(0)
unsafe{
d = byteptr(data.str)
b = byteptr(buffer)
}
for i < input_length {
mut char_a := 0
mut char_b := 0
mut char_c := 0
mut char_d := 0
if i < input_length {
char_a = Index[d[i]]
i++
}
if i < input_length {
char_b = Index[d[i]]
i++
}
if i < input_length {
char_c = Index[d[i]]
i++
2019-06-22 21:20:28 +03:00
}
if i < input_length {
char_d = Index[d[i]]
i++
}
decoded_bytes := (char_a << 18) | (char_b << 12) | (char_c << 6) | (char_d << 0)
b[j] = decoded_bytes >> 16
b[j+1] = (decoded_bytes >> 8) & 0xff
b[j+2] = (decoded_bytes >> 0) & 0xff
j += 3
2019-06-22 21:20:28 +03:00
}
return output_length
2019-06-22 21:20:28 +03:00
}
/**
* encode_in_buffer - expects a string reference, and a buffer in which to store its base64 encoded version.
* @param data - a reference/pointer to the input string.
2019-12-07 15:13:23 +03:00
* @param buffer - a reference/pointer to the buffer that will hold the result.
* The buffer should be large enough (i.e. 4/3 of the data.len, or larger) to hold the encoded data.
* @return the actual size of the encoded data in the buffer.
* NB: this function does NOT allocate new memory, and is suitable for handling very large strings.
*/
2019-12-07 15:13:23 +03:00
pub fn encode_in_buffer(data &string, buffer byteptr) int {
2019-06-25 20:37:18 +03:00
input_length := data.len
output_length := 4 * ((input_length + 2) / 3)
mut i := 0
mut j := 0
mut d := &byte(0)
mut b := &byte(0)
mut etable := &byte(0)
unsafe{
d = &byte(data.str)
b = &byte(buffer)
etable = &byte(EncodingTable.str)
}
2019-06-25 20:37:18 +03:00
for i < input_length {
mut octet_a := 0
mut octet_b := 0
mut octet_c := 0
if i < input_length {
octet_a = int(d[i])
2019-06-25 20:37:18 +03:00
i++
}
if i < input_length {
octet_b = int(d[i])
2019-06-25 20:37:18 +03:00
i++
}
if i < input_length {
octet_c = int(d[i])
2019-06-25 20:37:18 +03:00
i++
}
triple := ((int(octet_a) << 0x10) + (int(octet_b) << 0x08) + int(octet_c))
2019-06-25 20:37:18 +03:00
b[j] = etable[ (triple >> 3 * 6) & 63 ] // 63 is 0x3F
b[j+1] = etable[ (triple >> 2 * 6) & 63 ]
b[j+2] = etable[ (triple >> 1 * 6) & 63 ]
b[j+3] = etable[ (triple >> 0 * 6) & 63 ]
2019-06-25 20:37:18 +03:00
j += 4
}
padding_length := EndingTable[input_length % 3]
for i = 0; i < padding_length; i++ {
b[output_length - 1 - i] = `=`
2019-06-25 20:37:18 +03:00
}
return output_length
2019-06-25 20:37:18 +03:00
}