dymokomi/luce-js
Source of release 0.6.1
0.6.1 / src / engine / base64.lucb
#==============================================================================================
#
# base64 - Base64 and hex encoding and decoding of byte buffers
#
# DESCRIPTION:
# Port of quickjs.c "Uint8Array base64/hex (tc39 proposal-arraybuffer-base64)"
# helpers (lines 58966-59311): the alphabets and decoding tables, b64_encode,
# b64_skip_ws, from_base64 (the FromBase64 abstract operation), u8a_hex_encode and
# u8a_hex_decode. The C pointer and length pairs are spans here, so every access
# is bounds-checked; these functions need no runtime and are unit-tested in
# tests_typed_arrays.
#
#==============================================================================================
# mark: Alphabets and tables ====================================================================
# The alphabet option of the base64 methods.
let b64_alphabet_base64: i32 = 0
let b64_alphabet_base64url: i32 = 1
# The lastChunkHandling option of the base64 decoders.
let b64_last_loose: i32 = 0
let b64_last_strict: i32 = 1
let b64_last_stop_before_partial: i32 = 2
let b64_enc: const u8[] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"
let b64url_enc: const u8[] = b"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_"
# A decoding table maps a byte to its 6-bit value, to k_ws for whitespace (skipped) or to
# k_er for a byte that is not in the alphabet.
let k_ws: u8 = 64
let k_er: u8 = 65
let b64_dec: u8[256] = [
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_ws, k_ws, k_er, k_ws, k_ws, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_ws, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, 62, k_er, k_er, k_er, 63,
52, 53, 54, 55, 56, 57, 58, 59, 60, 61, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, 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, k_er, k_er, k_er, k_er, k_er,
k_er, 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, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
]
let b64url_dec: u8[256] = [
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_ws, k_ws, k_er, k_ws, k_ws, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_ws, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, 62, k_er, k_er,
52, 53, 54, 55, 56, 57, 58, 59, 60, 61, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, 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, k_er, k_er, k_er, k_er, 63,
k_er, 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, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er, k_er,
]
# mark: Base64 ==================================================================================
## Encode `src` into `dst` (at least 4 * ((src.length + 2) / 3) bytes) with the alphabet
## `alpha`, padding with '='; answer the number of bytes written.
func b64_encode(src: const u8[], dst: u8[], alpha: const u8[]) -> usize:
let len = src.length
var i: usize = 0
var j: usize = 0
while i + 3 <= len:
let v = 65536 * (u32)src[i] + 256 * (u32)src[i + 1] + (u32)src[i + 2]
dst[j + 0] = alpha[(usize)((v >> 18) & 63)]
dst[j + 1] = alpha[(usize)((v >> 12) & 63)]
dst[j + 2] = alpha[(usize)((v >> 6) & 63)]
dst[j + 3] = alpha[(usize)(v & 63)]
i += 3
j += 4
let rem = len - i
if rem == 1:
let v = 65536 * (u32)src[i]
dst[j] = alpha[(usize)((v >> 18) & 63)]
dst[j + 1] = alpha[(usize)((v >> 12) & 63)]
dst[j + 2] = '='
dst[j + 3] = '='
j += 4
elif rem == 2:
let v = 65536 * (u32)src[i] + 256 * (u32)src[i + 1]
dst[j] = alpha[(usize)((v >> 18) & 63)]
dst[j + 1] = alpha[(usize)((v >> 12) & 63)]
dst[j + 2] = alpha[(usize)((v >> 6) & 63)]
dst[j + 3] = '='
j += 4
return j
## The first index at or after `index` of `src` that is not whitespace.
func b64_skip_ws(src: const u8[], index0: usize, dec_table: const u8[]) -> usize:
var index = index0
while index < src.length and dec_table[(usize)src[index]] == k_ws:
index += 1
return index
# Implements the FromBase64 abstract operation.
# src: the input string (must be ASCII/latin1)
# dst: output buffer (its length is the maximum output)
# dec_table: b64_dec or b64url_dec (selects valid characters)
# last_chunk: b64_last_loose, b64_last_strict, or b64_last_stop_before_partial
# *p_read: set to number of input characters consumed
# *p_err: set to true on error, false on success
# Returns: number of bytes written to dst
func from_base64(src: const u8[], dst: u8[], dec_table: const u8[], last_chunk: i32,
p_read: usize*, p_err: bool*) -> usize:
let src_len = src.length
let max_len = dst.length
var read: usize = 0
var written: usize = 0
var acc: u32 = 0
var seen: i32 = 0
var index: usize = 0
*p_err = false
if max_len == 0:
*p_read = 0
return 0
while true:
if seen == 0:
# Fast path: decode complete groups of 4 valid characters. Breaks out on
# whitespace, padding, invalid chars, or capacity.
while index + 4 <= src_len and written + 3 <= max_len:
let v0 = (u32)dec_table[(usize)src[index]]
let v1 = (u32)dec_table[(usize)src[index + 1]]
let v2 = (u32)dec_table[(usize)src[index + 2]]
let v3 = (u32)dec_table[(usize)src[index + 3]]
if (v0 | v1 | v2 | v3) >= 64:
break
let v = (v0 << 18) | (v1 << 12) | (v2 << 6) | v3
dst[written] = (u8)(v >> 16)
dst[written + 1] = (u8)(v >> 8)
dst[written + 2] = (u8)v
written += 3
index += 4
read = index
if written >= max_len:
*p_read = read
return written
# Slow path: handle whitespace, padding, partial groups, capacity.
index = b64_skip_ws(src, index, dec_table)
if index == src_len:
if seen > 0:
if last_chunk == b64_last_stop_before_partial:
*p_read = read
return written
if last_chunk == b64_last_strict:
*p_err = true
return 0
# loose
if seen == 1:
*p_err = true
return 0
break
*p_read = src_len
return written
let ch = src[index]
index += 1
if ch == '=':
if seen < 2:
*p_err = true
return 0
index = b64_skip_ws(src, index, dec_table)
if seen == 2:
if index == src_len:
if last_chunk == b64_last_stop_before_partial:
*p_read = read
return written
*p_err = true
return 0
if src[index] == '=':
index += 1
index = b64_skip_ws(src, index, dec_table)
else:
*p_err = true
return 0
# After padding, only whitespace is allowed
if index != src_len:
*p_err = true
return 0
if last_chunk == b64_last_strict:
let mask: u32 = 0xF if seen == 2 else 0x3
if (acc & mask) != 0:
*p_err = true
return 0
break
let v = (u32)dec_table[(usize)ch]
if v >= 64:
*p_err = true
return 0
# Check remaining capacity before committing to this group
let remaining = max_len - written
if (remaining == 1 and seen == 2) or (remaining == 2 and seen == 3):
*p_read = read
return written
acc = (acc << 6) | v
seen += 1
if seen == 4:
dst[written] = (u8)(acc >> 16)
dst[written + 1] = (u8)(acc >> 8)
dst[written + 2] = (u8)acc
written += 3
acc = 0
seen = 0
read = index
if written >= max_len:
*p_read = read
return written
if seen == 2:
dst[written] = (u8)(acc >> 4)
written += 1
elif seen == 3:
dst[written] = (u8)(acc >> 10)
dst[written + 1] = (u8)(acc >> 2)
written += 2
*p_read = src_len
return written
# mark: Hex =====================================================================================
let u8a_hex_digits: const u8[] = b"0123456789abcdef"
## Encode `src` into `dst` (at least 2 * src.length bytes) as lowercase hex; answer the
## number of bytes written.
func u8a_hex_encode(src: const u8[], dst: u8[]) -> usize:
for i in 0..<src.length:
dst[i * 2] = u8a_hex_digits[(usize)(src[i] >> 4)]
dst[i * 2 + 1] = u8a_hex_digits[(usize)(src[i] & 0xF)]
return src.length * 2
# Decode hex string to bytes.
# Returns bytes written. Sets *p_read to chars consumed, *p_err on error.
func u8a_hex_decode(src: const u8[], dst: u8[], p_read: usize*, p_err: bool*) -> usize:
let src_len = src.length
let max_len = dst.length
var written: usize = 0
var i: usize = 0
*p_err = false
if (src_len & 1) != 0:
*p_err = true
return 0
while i < src_len and written < max_len:
let hi = cutils.from_hex((i32)src[i])
let lo = cutils.from_hex((i32)src[i + 1])
if hi < 0 or lo < 0:
*p_err = true
return 0
dst[written] = (u8)((hi << 4) | lo)
written += 1
i += 2
*p_read = i
return written