use vsimd::ascii::AsciiCase;
use vsimd::is_isa_type;
use vsimd::isa::{InstructionSet, AVX2, SSE2};
use vsimd::matches_isa;
use vsimd::tools::{read, write};
use vsimd::{SIMD128, SIMD256};
#[inline(always)]
fn charset(case: AsciiCase) -> &'static [u8; 16] {
match case {
AsciiCase::Lower => vsimd::hex::LOWER_CHARSET,
AsciiCase::Upper => vsimd::hex::UPPER_CHARSET,
}
}
#[inline(always)]
unsafe fn encode_bits(src: *const u8, dst: *mut u8, charset: *const u8) {
let x = src.read();
let hi = read(charset, (x >> 4) as usize);
let lo = read(charset, (x & 0x0f) as usize);
write(dst, 0, hi);
write(dst, 1, lo);
}
#[inline(always)]
unsafe fn encode_short(mut src: *const u8, len: usize, mut dst: *mut u8, charset: *const u8) {
let end = src.add(len);
while src < end {
encode_bits(src, dst, charset);
src = src.add(1);
dst = dst.add(2);
}
}
#[inline(always)]
unsafe fn encode_long(mut src: *const u8, len: usize, mut dst: *mut u8, case: AsciiCase) {
let charset = charset(case).as_ptr();
let end = src.add(len / 8 * 8);
while src < end {
let mut i = 0;
while i < 8 {
encode_bits(src, dst, charset);
src = src.add(1);
dst = dst.add(2);
i += 1;
}
}
encode_short(src, len % 8, dst, charset);
}
#[inline(always)]
pub unsafe fn encode_fallback(src: *const u8, len: usize, dst: *mut u8, case: AsciiCase) {
encode_long(src, len, dst, case);
}
#[inline(always)]
pub unsafe fn encode_simd<S: SIMD256>(s: S, src: *const u8, len: usize, dst: *mut u8, case: AsciiCase) {
if cfg!(any(target_arch = "x86", target_arch = "x86_64")) {
if is_isa_type!(S, SSE2) {
return encode_simd_sse2(SSE2::new(), src, len, dst, case);
}
if matches_isa!(S, AVX2) {
return encode_simd_v256(s, src, len, dst, case);
}
}
encode_simd_v128(s, src, len, dst, case);
}
#[inline(always)]
pub unsafe fn encode_simd_v256<S: SIMD256>(
s: S,
mut src: *const u8,
mut len: usize,
mut dst: *mut u8,
case: AsciiCase,
) {
let lut = match case {
AsciiCase::Lower => vsimd::hex::ENCODE_LOWER_LUT,
AsciiCase::Upper => vsimd::hex::ENCODE_UPPER_LUT,
};
if len == 16 {
let x = s.v128_load_unaligned(src);
let y = vsimd::hex::encode_bytes16(s, x, lut);
s.v256_store_unaligned(dst, y);
return;
}
let end = src.add(len / 32 * 32);
while src < end {
let x = s.v256_load_unaligned(src);
let (y1, y2) = vsimd::hex::encode_bytes32(s, x, lut);
s.v256_store_unaligned(dst, y1);
dst = dst.add(32);
s.v256_store_unaligned(dst, y2);
dst = dst.add(32);
src = src.add(32);
}
len %= 32;
if len == 0 {
return;
}
if len >= 16 {
let x = s.v128_load_unaligned(src);
let y = vsimd::hex::encode_bytes16(s, x, lut);
s.v256_store_unaligned(dst, y);
dst = dst.add(32);
src = src.add(16);
len -= 16;
}
if len > 0 {
let charset = charset(case).as_ptr();
encode_short(src, len, dst, charset);
}
}
#[inline(always)]
pub unsafe fn encode_simd_v128<S: SIMD256>(
s: S,
mut src: *const u8,
mut len: usize,
mut dst: *mut u8,
case: AsciiCase,
) {
let lut = match case {
AsciiCase::Lower => vsimd::hex::ENCODE_LOWER_LUT,
AsciiCase::Upper => vsimd::hex::ENCODE_UPPER_LUT,
};
let end = src.add(len / 16 * 16);
while src < end {
let x = s.v128_load_unaligned(src);
let y = vsimd::hex::encode_bytes16(s, x, lut);
s.v256_store_unaligned(dst, y);
dst = dst.add(32);
src = src.add(16);
}
len %= 16;
if len == 0 {
return;
}
let charset = charset(case).as_ptr();
encode_short(src, len, dst, charset);
}
#[inline(always)]
pub unsafe fn encode_simd_sse2(s: SSE2, mut src: *const u8, mut len: usize, mut dst: *mut u8, case: AsciiCase) {
let offset = match case {
AsciiCase::Lower => vsimd::hex::sse2::LOWER_OFFSET,
AsciiCase::Upper => vsimd::hex::sse2::UPPER_OFFSET,
};
let end = src.add(len / 16 * 16);
while src < end {
let x = s.v128_load_unaligned(src);
src = src.add(16);
let (y1, y2) = vsimd::hex::sse2::encode16(s, x, offset);
s.v128_store_unaligned(dst, y1);
dst = dst.add(16);
s.v128_store_unaligned(dst, y2);
dst = dst.add(16);
}
len %= 16;
if len == 0 {
return;
}
let charset = charset(case).as_ptr();
encode_short(src, len, dst, charset);
}