vsimd/
hex.rs

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
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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
use crate::alsw::{self, AlswLut};
use crate::isa::{AVX2, NEON, SSSE3, WASM128};
use crate::mask::{u8x16_highbit_any, u8x32_highbit_any};
use crate::pod::POD;
use crate::vector::{V128, V256, V64};
use crate::{Scalable, SIMD128, SIMD256};

pub const UPPER_CHARSET: &[u8; 16] = b"0123456789ABCDEF";
pub const LOWER_CHARSET: &[u8; 16] = b"0123456789abcdef";

const fn parse_hex(x: u8) -> u8 {
    match x {
        b'0'..=b'9' => x - b'0',
        b'a'..=b'f' => x - b'a' + 10,
        b'A'..=b'F' => x - b'A' + 10,
        _ => 0xff,
    }
}

#[inline(always)]
#[must_use]
pub const fn unhex(x: u8) -> u8 {
    const UNHEX_TABLE: &[u8; 256] = &{
        let mut arr = [0; 256];
        let mut i = 0;
        while i < 256 {
            arr[i] = parse_hex(i as u8);
            i += 1;
        }
        arr
    };
    UNHEX_TABLE[x as usize]
}

#[inline(always)]
pub fn check_xn<S, V>(s: S, x: V) -> bool
where
    S: Scalable<V>,
    V: POD,
{
    let x1 = s.u8xn_sub(x, s.u8xn_splat(0x30 + 0x80));
    let x2 = s.u8xn_sub(s.and(x, s.u8xn_splat(0xdf)), s.u8xn_splat(0x41 + 0x80));
    let m1 = s.i8xn_lt(x1, s.i8xn_splat(-118));
    let m2 = s.i8xn_lt(x2, s.i8xn_splat(-122));
    s.mask8xn_all(s.or(m1, m2))
}

pub const ENCODE_UPPER_LUT: V256 = V256::double_bytes(*UPPER_CHARSET);
pub const ENCODE_LOWER_LUT: V256 = V256::double_bytes(*LOWER_CHARSET);

#[inline(always)]
pub fn encode_bytes16<S: SIMD256>(s: S, x: V128, lut: V256) -> V256 {
    let x = s.u16x16_from_u8x16(x);
    let hi = s.u16x16_shl::<8>(x);
    let lo = s.u16x16_shr::<4>(x);
    let values = s.v256_and(s.v256_or(hi, lo), s.u8x32_splat(0x0f));
    s.u8x16x2_swizzle(lut, values)
}

#[inline(always)]
pub fn encode_bytes32<S: SIMD256>(s: S, x: V256, lut: V256) -> (V256, V256) {
    let m = s.u8x32_splat(0x0f);
    let hi = s.v256_and(s.u16x16_shr::<4>(x), m);
    let lo = s.v256_and(x, m);

    let ac = s.u8x16x2_zip_lo(hi, lo);
    let bd = s.u8x16x2_zip_hi(hi, lo);

    let ab = s.v128x2_zip_lo(ac, bd);
    let cd = s.v128x2_zip_hi(ac, bd);

    let y1 = s.u8x16x2_swizzle(lut, ab);
    let y2 = s.u8x16x2_swizzle(lut, cd);

    (y1, y2)
}

struct HexAlsw;

impl HexAlsw {
    const fn decode(c: u8) -> u8 {
        parse_hex(c)
    }

    const fn check_hash(i: u8) -> u8 {
        match i {
            0 => 1,
            1..=6 => 1,
            7..=9 => 6,
            0xA..=0xF => 8,
            _ => unreachable!(),
        }
    }

    const fn decode_hash(i: u8) -> u8 {
        Self::check_hash(i)
    }
}

impl_alsw!(HexAlsw);

const HEX_ALSW_CHECK: AlswLut<V128> = HexAlsw::check_lut();
const HEX_ALSW_DECODE: AlswLut<V128> = HexAlsw::decode_lut();

const HEX_ALSW_CHECK_X2: AlswLut<V256> = HexAlsw::check_lut().x2();
const HEX_ALSW_DECODE_X2: AlswLut<V256> = HexAlsw::decode_lut().x2();

const DECODE_UZP1: V256 = V256::double_bytes([
    0x00, 0x02, 0x04, 0x06, 0x08, 0x0a, 0x0c, 0x0e, //
    0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, //
]);

const DECODE_UZP2: V256 = V256::double_bytes([
    0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, //
    0x00, 0x02, 0x04, 0x06, 0x08, 0x0a, 0x0c, 0x0e, //
]);

#[inline(always)]
fn merge_bits<S: Scalable<V>, V: POD>(s: S, x: V) -> V {
    // x:  {0000hhhh|0000llll} xn

    let x1 = s.u16xn_shl::<4>(x);
    // x1: {hhhh0000|llll0000} xn

    let x2 = s.u16xn_shr::<12>(x1);
    // x2: {0000llll|00000000} xn

    s.or(x1, x2)
    //     {hhhhllll|????????} xn
}

#[inline(always)]
fn decode16<S: SIMD128>(s: S, x: V128) -> (V128, V128) {
    let (c1, c2) = alsw::decode_ascii_xn(s, x, HEX_ALSW_CHECK, HEX_ALSW_DECODE);
    (merge_bits(s, c2), c1)
}

#[inline(always)]
fn decode32<S: SIMD256>(s: S, x: V256) -> (V256, V256) {
    let (c1, c2) = alsw::decode_ascii_xn(s, x, HEX_ALSW_CHECK_X2, HEX_ALSW_DECODE_X2);
    (merge_bits(s, c2), c1)
}

#[allow(clippy::result_unit_err)]
#[inline(always)]
pub fn decode_ascii16<S: SIMD128>(s: S, x: V128) -> Result<V64, ()> {
    let (y, is_invalid) = decode16(s, x);

    let ans = if matches_isa!(S, SSSE3 | WASM128) {
        const UZP1: V128 = DECODE_UZP1.to_v128x2().0;
        s.u8x16_swizzle(y, UZP1).to_v64x2().0
    } else if matches_isa!(S, NEON) {
        let (a, b) = y.to_v64x2();
        s.u8x8_unzip_even(a, b)
    } else {
        unreachable!()
    };

    if u8x16_highbit_any(s, is_invalid) {
        Err(())
    } else {
        Ok(ans)
    }
}

#[allow(clippy::result_unit_err)]
#[inline(always)]
pub fn decode_ascii32<S: SIMD256>(s: S, x: V256) -> Result<V128, ()> {
    let (y, is_invalid) = decode32(s, x);

    let ans = if matches_isa!(S, SSSE3 | WASM128) {
        let (a, b) = s.u8x16x2_swizzle(y, DECODE_UZP1).to_v128x2();
        s.u64x2_zip_lo(a, b)
    } else if matches_isa!(S, NEON) {
        let (a, b) = y.to_v128x2();
        s.u8x16_unzip_even(a, b)
    } else {
        unreachable!()
    };

    if u8x32_highbit_any(s, is_invalid) {
        Err(())
    } else {
        Ok(ans)
    }
}

#[allow(clippy::result_unit_err)]
#[inline(always)]
pub fn decode_ascii32x2<S: SIMD256>(s: S, x: (V256, V256)) -> Result<V256, ()> {
    let (y1, is_invalid1) = decode32(s, x.0);
    let (y2, is_invalid2) = decode32(s, x.1);
    let is_invalid = s.v256_or(is_invalid1, is_invalid2);

    let ans = if matches_isa!(S, AVX2) {
        let ab = s.u8x16x2_swizzle(y1, DECODE_UZP1);
        let cd = s.u8x16x2_swizzle(y2, DECODE_UZP2);
        let acbd = s.v256_or(ab, cd);
        s.u64x4_permute::<0b_1101_1000>(acbd) // 0213
    } else if matches_isa!(S, SSSE3 | WASM128) {
        let ab = s.u8x16x2_swizzle(y1, DECODE_UZP1);
        let cd = s.u8x16x2_swizzle(y2, DECODE_UZP1);
        s.u64x4_unzip_even(ab, cd)
    } else if matches_isa!(S, NEON) {
        s.u8x32_unzip_even(y1, y2)
    } else {
        unreachable!()
    };

    if u8x32_highbit_any(s, is_invalid) {
        Err(())
    } else {
        Ok(ans)
    }
}

pub mod sse2 {
    use crate::isa::SSE2;
    use crate::vector::{V128, V64};
    use crate::SIMD128;

    #[inline(always)]
    #[must_use]
    pub fn decode_nibbles(s: SSE2, x: V128) -> (V128, V128) {
        // http://0x80.pl/notesen/2022-01-17-validating-hex-parse.html
        // Algorithm 3

        let t1 = s.u8x16_add(x, s.u8x16_splat(0xff - b'9'));
        let t2 = s.u8x16_sub_sat(t1, s.u8x16_splat(6));
        let t3 = s.u8x16_sub(t2, s.u8x16_splat(0xf0));
        let t4 = s.v128_and(x, s.u8x16_splat(0xdf));
        let t5 = s.u8x16_sub(t4, s.u8x16_splat(0x41));
        let t6 = s.u8x16_add_sat(t5, s.u8x16_splat(10));
        let t7 = s.u8x16_min(t3, t6);
        let t8 = s.u8x16_add_sat(t7, s.u8x16_splat(127 - 15));
        (t7, t8)
    }

    #[inline(always)]
    #[must_use]
    pub fn merge_bits(s: SSE2, x: V128) -> V64 {
        let lo = s.u16x8_shr::<8>(x);
        let hi = s.u16x8_shl::<4>(x);
        let t1 = s.v128_or(lo, hi);
        let t2 = s.v128_and(t1, s.u16x8_splat(0x00ff));
        let t3 = s.i16x8_packus(t2, s.v128_create_zero());
        t3.to_v64x2().0
    }

    pub const LOWER_OFFSET: V128 = V128::from_bytes([0x27; 16]);
    pub const UPPER_OFFSET: V128 = V128::from_bytes([0x07; 16]);

    #[inline(always)]
    #[must_use]
    pub fn encode16(s: SSE2, x: V128, offset: V128) -> (V128, V128) {
        let m = s.u8x16_splat(0x0f);
        let hi = s.v128_and(s.u16x8_shr::<4>(x), m);
        let lo = s.v128_and(x, m);

        let c1 = s.u8x16_splat(0x30);
        let h1 = s.u8x16_add(hi, c1);
        let l1 = s.u8x16_add(lo, c1);

        let c2 = s.u8x16_splat(0x39);
        let h2 = s.v128_and(s.i8x16_lt(c2, h1), offset);
        let l2 = s.v128_and(s.i8x16_lt(c2, l1), offset);

        let h3 = s.u8x16_add(h1, h2);
        let l3 = s.u8x16_add(l1, l2);

        let y1 = s.u8x16_zip_lo(h3, l3);
        let y2 = s.u8x16_zip_hi(h3, l3);

        (y1, y2)
    }
}

#[cfg(test)]
mod algorithm {
    use super::*;

    #[test]
    #[ignore]
    fn check() {
        fn is_hex_v1(c: u8) -> bool {
            matches!(c, b'0'..=b'9' | b'a'..=b'f' | b'A'..=b'F')
        }

        fn is_hex_v2(c: u8) -> bool {
            let x1 = c.wrapping_sub(0x30);
            let x2 = (c & 0xdf).wrapping_sub(0x41);
            x1 < 10 || x2 < 6
        }

        fn is_hex_v3(c: u8) -> bool {
            let x1 = c.wrapping_sub(0x30 + 0x80);
            let x2 = (c & 0xdf).wrapping_sub(0x41 + 0x80);
            ((x1 as i8) < -118) || ((x2 as i8) < -122)
        }

        for c in 0..=255_u8 {
            let (v1, v2, v3) = (is_hex_v1(c), is_hex_v2(c), is_hex_v3(c));
            assert_eq!(v1, v2);
            assert_eq!(v1, v3);
        }
    }

    #[test]
    #[ignore]
    fn hex_alsw() {
        HexAlsw::test_check();
        HexAlsw::test_decode();
    }
}