std/vp8: calculate the predicted values

This commit, part of a series, is inspired by pull request PR #168 by
Damian Kaczmarek (with AI assistance), with some modifications.

Updates PR #168
This commit is contained in:
Nigel Tao committed 2026-06-16 14:00:28 +10:00
1 parent 3a74d542ab
commit bb3c85e2ef
8 files changed
+2276 -11

No files matched your search

+1050 -6
View File
@@ -15532,6 +15532,7 @@ struct wuffs_vp8__decoder__struct {
uint16_t f_dequants[4][3][2];
uint32_t f_prob_skip;
uint8_t f_mb_subblock_modes[16];
uint8_t f_yuv_cache[26][32];
uint32_t f_dst_x;
uint32_t f_dst_y;
wuffs_base__pixel_swizzler f_swizzler;
@@ -80975,6 +80976,138 @@ WUFFS_VP8__CLAMP_NO_MORE_THAN_9[16] WUFFS_BASE__POTENTIALLY_UNUSED = {
8u, 9u, 9u, 9u, 9u, 9u, 9u, 9u,
};
static const uint8_t
WUFFS_VP8__CLAMP[1024] WUFFS_BASE__POTENTIALLY_UNUSED = {
0u, 1u, 2u, 3u, 4u, 5u, 6u, 7u,
8u, 9u, 10u, 11u, 12u, 13u, 14u, 15u,
16u, 17u, 18u, 19u, 20u, 21u, 22u, 23u,
24u, 25u, 26u, 27u, 28u, 29u, 30u, 31u,
32u, 33u, 34u, 35u, 36u, 37u, 38u, 39u,
40u, 41u, 42u, 43u, 44u, 45u, 46u, 47u,
48u, 49u, 50u, 51u, 52u, 53u, 54u, 55u,
56u, 57u, 58u, 59u, 60u, 61u, 62u, 63u,
64u, 65u, 66u, 67u, 68u, 69u, 70u, 71u,
72u, 73u, 74u, 75u, 76u, 77u, 78u, 79u,
80u, 81u, 82u, 83u, 84u, 85u, 86u, 87u,
88u, 89u, 90u, 91u, 92u, 93u, 94u, 95u,
96u, 97u, 98u, 99u, 100u, 101u, 102u, 103u,
104u, 105u, 106u, 107u, 108u, 109u, 110u, 111u,
112u, 113u, 114u, 115u, 116u, 117u, 118u, 119u,
120u, 121u, 122u, 123u, 124u, 125u, 126u, 127u,
128u, 129u, 130u, 131u, 132u, 133u, 134u, 135u,
136u, 137u, 138u, 139u, 140u, 141u, 142u, 143u,
144u, 145u, 146u, 147u, 148u, 149u, 150u, 151u,
152u, 153u, 154u, 155u, 156u, 157u, 158u, 159u,
160u, 161u, 162u, 163u, 164u, 165u, 166u, 167u,
168u, 169u, 170u, 171u, 172u, 173u, 174u, 175u,
176u, 177u, 178u, 179u, 180u, 181u, 182u, 183u,
184u, 185u, 186u, 187u, 188u, 189u, 190u, 191u,
192u, 193u, 194u, 195u, 196u, 197u, 198u, 199u,
200u, 201u, 202u, 203u, 204u, 205u, 206u, 207u,
208u, 209u, 210u, 211u, 212u, 213u, 214u, 215u,
216u, 217u, 218u, 219u, 220u, 221u, 222u, 223u,
224u, 225u, 226u, 227u, 228u, 229u, 230u, 231u,
232u, 233u, 234u, 235u, 236u, 237u, 238u, 239u,
240u, 241u, 242u, 243u, 244u, 245u, 246u, 247u,
248u, 249u, 250u, 251u, 252u, 253u, 254u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
255u, 255u, 255u, 255u, 255u, 255u, 255u, 255u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u,
};
static const uint8_t
WUFFS_VP8__SUBBLOCK_MODE_PROBS[10][10][9] WUFFS_BASE__POTENTIALLY_UNUSED = {
{
@@ -81688,8 +81821,7 @@ wuffs_vp8__decoder__decode_coefficients(
uint32_t a_mbx,
uint32_t a_mby,
uint32_t a_seg,
uint32_t a_luma_mode,
uint32_t a_chroma_mode);
uint32_t a_luma_mode);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
@@ -81805,6 +81937,51 @@ wuffs_vp8__decoder__decode_one_subblock_mode(
uint8_t a_top_mode,
uint8_t a_left_mode);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_y4(
wuffs_vp8__decoder* self,
uint32_t a_b);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_y16(
wuffs_vp8__decoder* self,
uint32_t a_mode);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_uv8(
wuffs_vp8__decoder* self,
uint32_t a_b,
uint32_t a_mode);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__reconstruct(
wuffs_vp8__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mbx,
uint32_t a_mby,
uint32_t a_luma_mode,
uint32_t a_chroma_mode);
WUFFS_BASE__GENERATED_C_CODE
static uint32_t
wuffs_vp8__decoder__substitute_dc_top_left(
const wuffs_vp8__decoder* self,
uint32_t a_mode,
uint32_t a_mbx,
uint32_t a_mby);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__prepare_yuv_cache(
wuffs_vp8__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mbx,
uint32_t a_mby);
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__status
wuffs_vp8__decoder__do_decode_image_config(
@@ -82040,8 +82217,7 @@ wuffs_vp8__decoder__decode_coefficients(
uint32_t a_mbx,
uint32_t a_mby,
uint32_t a_seg,
uint32_t a_luma_mode,
uint32_t a_chroma_mode) {
uint32_t a_luma_mode) {
uint32_t v_bc = 0;
uint32_t v_by = 0;
uint32_t v_bx = 0;
@@ -82679,9 +82855,14 @@ wuffs_vp8__decoder__decode_one_macroblock(
a_mbx,
a_mby,
v_seg,
v_luma_mode,
v_chroma_mode);
v_luma_mode);
}
wuffs_vp8__decoder__reconstruct(self,
a_workbuf,
a_mbx,
a_mby,
v_luma_mode,
v_chroma_mode);
return wuffs_base__make_empty_struct();
}
@@ -82830,6 +83011,869 @@ wuffs_vp8__decoder__decode_one_subblock_mode(
return 9u;
}
// -------- func vp8.decoder.predict_y4
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_y4(
wuffs_vp8__decoder* self,
uint32_t a_b) {
uint32_t v_cachey = 0;
uint32_t v_cachex = 0;
uint32_t v_mode = 0;
uint32_t v_z = 0;
uint8_t v_avg = 0;
uint32_t v_tm0 = 0;
uint32_t v_tm1 = 0;
uint32_t v_tm2 = 0;
uint32_t v_a = 0;
uint32_t v_b = 0;
uint32_t v_c = 0;
uint32_t v_d = 0;
uint32_t v_e = 0;
uint32_t v_f = 0;
uint32_t v_g = 0;
uint32_t v_h = 0;
uint32_t v_i = 0;
uint32_t v_p = 0;
uint32_t v_q = 0;
uint32_t v_r = 0;
uint32_t v_s = 0;
uint8_t v_ab = 0;
uint8_t v_bc = 0;
uint8_t v_cd = 0;
uint8_t v_de = 0;
uint8_t v_ef = 0;
uint8_t v_pa = 0;
uint8_t v_qp = 0;
uint8_t v_rq = 0;
uint8_t v_sr = 0;
uint8_t v_abc = 0;
uint8_t v_bcd = 0;
uint8_t v_cde = 0;
uint8_t v_def = 0;
uint8_t v_efg = 0;
uint8_t v_fgh = 0;
uint8_t v_ghi = 0;
uint8_t v_hii = 0;
uint8_t v_pab = 0;
uint8_t v_qpa = 0;
uint8_t v_rqp = 0;
uint8_t v_srq = 0;
uint8_t v_ssr = 0;
uint8_t v_sss = 0;
v_cachey = ((a_b & 12u) + 1u);
v_cachex = (((a_b & 3u) * 4u) + 8u);
v_mode = ((uint32_t)(self->private_impl.f_mb_subblock_modes[a_b]));
if (v_mode == 0u) {
v_avg = ((uint8_t)(((4u +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)]))) / 8u)));
v_z = 0u;
while (v_z < 16u) {
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 2u))][(v_cachex + (v_z & 3u))] = v_avg;
v_z += 1u;
}
} else if (v_mode == 1u) {
v_tm0 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_z = 0u;
while (v_z < 16u) {
v_tm1 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 2u))][(v_cachex - 1u)]));
v_tm2 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + (v_z & 3u))]));
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 2u))][(v_cachex + (v_z & 3u))] = WUFFS_VP8__CLAMP[(((uint32_t)((v_tm1 + v_tm2) - v_tm0)) & 1023u)];
v_z += 1u;
}
} else if (v_mode == 2u) {
v_a = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_e = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)]));
v_f = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 4u)]));
v_def = ((uint8_t)(((v_d +
(2u * v_e) +
v_f +
2u) / 4u)));
v_cde = ((uint8_t)(((v_c +
(2u * v_d) +
v_e +
2u) / 4u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
v_abc = ((uint8_t)(((v_a +
(2u * v_b) +
v_c +
2u) / 4u)));
v_z = 0u;
while (v_z < 4u) {
self->private_impl.f_yuv_cache[(v_cachey + v_z)][v_cachex] = v_abc;
self->private_impl.f_yuv_cache[(v_cachey + v_z)][(v_cachex + 1u)] = v_bcd;
self->private_impl.f_yuv_cache[(v_cachey + v_z)][(v_cachex + 2u)] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + v_z)][(v_cachex + 3u)] = v_def;
v_z += 1u;
}
} else if (v_mode == 3u) {
v_s = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)]));
v_r = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)]));
v_q = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)]));
v_p = ((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)]));
v_a = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_qpa = ((uint8_t)(((v_q +
(2u * v_p) +
v_a +
2u) / 4u)));
v_rqp = ((uint8_t)(((v_r +
(2u * v_q) +
v_p +
2u) / 4u)));
v_srq = ((uint8_t)(((v_s +
(2u * v_r) +
v_q +
2u) / 4u)));
v_ssr = ((uint8_t)(((v_s +
(2u * v_s) +
v_r +
2u) / 4u)));
v_z = 0u;
while (v_z < 4u) {
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + v_z)] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + v_z)] = v_rqp;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + v_z)] = v_srq;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + v_z)] = v_ssr;
v_z += 1u;
}
} else if (v_mode == 4u) {
v_s = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)]));
v_r = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)]));
v_q = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)]));
v_p = ((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)]));
v_a = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_e = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)]));
v_cde = ((uint8_t)(((v_c +
(2u * v_d) +
v_e +
2u) / 4u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
v_abc = ((uint8_t)(((v_a +
(2u * v_b) +
v_c +
2u) / 4u)));
v_pab = ((uint8_t)(((v_p +
(2u * v_a) +
v_b +
2u) / 4u)));
v_qpa = ((uint8_t)(((v_q +
(2u * v_p) +
v_a +
2u) / 4u)));
v_rqp = ((uint8_t)(((v_r +
(2u * v_q) +
v_p +
2u) / 4u)));
v_srq = ((uint8_t)(((v_s +
(2u * v_r) +
v_q +
2u) / 4u)));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_pab;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_abc;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_bcd;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_pab;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_abc;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_bcd;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_rqp;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_pab;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_abc;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_srq;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_rqp;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_pab;
} else if (v_mode == 5u) {
v_r = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)]));
v_q = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)]));
v_p = ((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)]));
v_a = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_e = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)]));
v_ab = ((uint8_t)(((v_a + v_b + 1u) / 2u)));
v_bc = ((uint8_t)(((v_b + v_c + 1u) / 2u)));
v_cd = ((uint8_t)(((v_c + v_d + 1u) / 2u)));
v_de = ((uint8_t)(((v_d + v_e + 1u) / 2u)));
v_cde = ((uint8_t)(((v_c +
(2u * v_d) +
v_e +
2u) / 4u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
v_abc = ((uint8_t)(((v_a +
(2u * v_b) +
v_c +
2u) / 4u)));
v_pab = ((uint8_t)(((v_p +
(2u * v_a) +
v_b +
2u) / 4u)));
v_qpa = ((uint8_t)(((v_q +
(2u * v_p) +
v_a +
2u) / 4u)));
v_rqp = ((uint8_t)(((v_r +
(2u * v_q) +
v_p +
2u) / 4u)));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_ab;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_bc;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_cd;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_de;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_pab;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_abc;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_bcd;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_ab;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_bc;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_cd;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_rqp;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_pab;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_abc;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_bcd;
} else if (v_mode == 6u) {
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_e = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)]));
v_f = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 4u)]));
v_g = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 5u)]));
v_h = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 6u)]));
v_i = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 7u)]));
v_hii = ((uint8_t)(((v_h +
(2u * v_i) +
v_i +
2u) / 4u)));
v_ghi = ((uint8_t)(((v_g +
(2u * v_h) +
v_i +
2u) / 4u)));
v_fgh = ((uint8_t)(((v_f +
(2u * v_g) +
v_h +
2u) / 4u)));
v_efg = ((uint8_t)(((v_e +
(2u * v_f) +
v_g +
2u) / 4u)));
v_def = ((uint8_t)(((v_d +
(2u * v_e) +
v_f +
2u) / 4u)));
v_cde = ((uint8_t)(((v_c +
(2u * v_d) +
v_e +
2u) / 4u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_bcd;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_cde;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_def;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_def;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_fgh;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_def;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_fgh;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_ghi;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_fgh;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_ghi;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_hii;
} else if (v_mode == 7u) {
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_e = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)]));
v_f = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 4u)]));
v_g = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 5u)]));
v_h = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 6u)]));
v_i = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 7u)]));
v_ef = ((uint8_t)(((v_e + v_f + 1u) / 2u)));
v_de = ((uint8_t)(((v_d + v_e + 1u) / 2u)));
v_cd = ((uint8_t)(((v_c + v_d + 1u) / 2u)));
v_bc = ((uint8_t)(((v_b + v_c + 1u) / 2u)));
v_ghi = ((uint8_t)(((v_g +
(2u * v_h) +
v_i +
2u) / 4u)));
v_fgh = ((uint8_t)(((v_f +
(2u * v_g) +
v_h +
2u) / 4u)));
v_efg = ((uint8_t)(((v_e +
(2u * v_f) +
v_g +
2u) / 4u)));
v_def = ((uint8_t)(((v_d +
(2u * v_e) +
v_f +
2u) / 4u)));
v_cde = ((uint8_t)(((v_c +
(2u * v_d) +
v_e +
2u) / 4u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_bc;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_cd;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_de;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_ef;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_bcd;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_def;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_cd;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_de;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_ef;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_fgh;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_cde;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_def;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_efg;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_ghi;
} else if (v_mode == 8u) {
v_s = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)]));
v_r = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)]));
v_q = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)]));
v_p = ((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)]));
v_a = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_b = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex]));
v_c = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)]));
v_d = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)]));
v_sr = ((uint8_t)(((v_s + v_r + 1u) / 2u)));
v_rq = ((uint8_t)(((v_r + v_q + 1u) / 2u)));
v_qp = ((uint8_t)(((v_q + v_p + 1u) / 2u)));
v_pa = ((uint8_t)(((v_p + v_a + 1u) / 2u)));
v_bcd = ((uint8_t)(((v_b +
(2u * v_c) +
v_d +
2u) / 4u)));
v_abc = ((uint8_t)(((v_a +
(2u * v_b) +
v_c +
2u) / 4u)));
v_pab = ((uint8_t)(((v_p +
(2u * v_a) +
v_b +
2u) / 4u)));
v_qpa = ((uint8_t)(((v_q +
(2u * v_p) +
v_a +
2u) / 4u)));
v_rqp = ((uint8_t)(((v_r +
(2u * v_q) +
v_p +
2u) / 4u)));
v_srq = ((uint8_t)(((v_s +
(2u * v_r) +
v_q +
2u) / 4u)));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_pa;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_pab;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_abc;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_bcd;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_qp;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_pa;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_pab;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_rq;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_rqp;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_qp;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_qpa;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_sr;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_srq;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_rq;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_rqp;
} else {
v_s = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)]));
v_r = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)]));
v_q = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)]));
v_p = ((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)]));
v_sr = ((uint8_t)(((v_s + v_r + 1u) / 2u)));
v_rq = ((uint8_t)(((v_r + v_q + 1u) / 2u)));
v_qp = ((uint8_t)(((v_q + v_p + 1u) / 2u)));
v_rqp = ((uint8_t)(((v_r +
(2u * v_q) +
v_p +
2u) / 4u)));
v_srq = ((uint8_t)(((v_s +
(2u * v_r) +
v_q +
2u) / 4u)));
v_ssr = ((uint8_t)(((v_s +
(2u * v_s) +
v_r +
2u) / 4u)));
v_sss = ((uint8_t)(v_s));
self->private_impl.f_yuv_cache[v_cachey][v_cachex] = v_qp;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 1u)] = v_rqp;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 2u)] = v_rq;
self->private_impl.f_yuv_cache[v_cachey][(v_cachex + 3u)] = v_srq;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][v_cachex] = v_rq;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 1u)] = v_srq;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 2u)] = v_sr;
self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex + 3u)] = v_ssr;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][v_cachex] = v_sr;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 1u)] = v_ssr;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 2u)] = v_sss;
self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex + 3u)] = v_sss;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][v_cachex] = v_sss;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 1u)] = v_sss;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 2u)] = v_sss;
self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex + 3u)] = v_sss;
}
return wuffs_base__make_empty_struct();
}
// -------- func vp8.decoder.predict_y16
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_y16(
wuffs_vp8__decoder* self,
uint32_t a_mode) {
uint32_t v_x = 0;
uint32_t v_y = 0;
uint32_t v_z = 0;
uint8_t v_avg = 0;
uint8_t v_val = 0;
uint32_t v_tm0 = 0;
uint32_t v_tm1 = 0;
uint32_t v_tm2 = 0;
if (a_mode == 0u) {
v_avg = ((uint8_t)(((16u +
((uint32_t)(self->private_impl.f_yuv_cache[0u][8u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][9u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][10u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][11u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][12u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][13u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][14u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][15u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][16u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][17u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][18u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][19u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][20u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][21u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][22u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][23u])) +
((uint32_t)(self->private_impl.f_yuv_cache[1u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[2u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[3u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[4u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[5u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[6u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[7u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[8u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[9u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[10u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[11u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[12u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[13u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[14u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[15u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[16u][7u]))) / 32u)));
v_z = 0u;
while (v_z < 256u) {
self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][(8u + (v_z & 15u))] = v_avg;
v_z += 1u;
}
} else if (a_mode == 1u) {
v_tm0 = ((uint32_t)(self->private_impl.f_yuv_cache[0u][7u]));
v_z = 0u;
while (v_z < 256u) {
v_tm1 = ((uint32_t)(self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][7u]));
v_tm2 = ((uint32_t)(self->private_impl.f_yuv_cache[0u][(8u + (v_z & 15u))]));
self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][(8u + (v_z & 15u))] = WUFFS_VP8__CLAMP[(((uint32_t)((v_tm1 + v_tm2) - v_tm0)) & 1023u)];
v_z += 1u;
}
} else if (a_mode == 2u) {
v_x = 0u;
while (v_x < 16u) {
v_val = self->private_impl.f_yuv_cache[0u][(8u + v_x)];
v_y = 0u;
while (v_y < 16u) {
self->private_impl.f_yuv_cache[(1u + v_y)][(8u + v_x)] = v_val;
v_y += 1u;
}
v_x += 1u;
}
} else if (a_mode == 3u) {
v_y = 0u;
while (v_y < 16u) {
v_val = self->private_impl.f_yuv_cache[(1u + v_y)][7u];
v_x = 0u;
while (v_x < 16u) {
self->private_impl.f_yuv_cache[(1u + v_y)][(8u + v_x)] = v_val;
v_x += 1u;
}
v_y += 1u;
}
} else if (a_mode <= 10u) {
v_avg = ((uint8_t)(((8u +
((uint32_t)(self->private_impl.f_yuv_cache[1u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[2u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[3u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[4u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[5u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[6u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[7u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[8u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[9u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[10u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[11u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[12u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[13u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[14u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[15u][7u])) +
((uint32_t)(self->private_impl.f_yuv_cache[16u][7u]))) / 16u)));
v_z = 0u;
while (v_z < 256u) {
self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][(8u + (v_z & 15u))] = v_avg;
v_z += 1u;
}
} else if (a_mode == 11u) {
v_avg = ((uint8_t)(((8u +
((uint32_t)(self->private_impl.f_yuv_cache[0u][8u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][9u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][10u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][11u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][12u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][13u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][14u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][15u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][16u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][17u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][18u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][19u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][20u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][21u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][22u])) +
((uint32_t)(self->private_impl.f_yuv_cache[0u][23u]))) / 16u)));
v_z = 0u;
while (v_z < 256u) {
self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][(8u + (v_z & 15u))] = v_avg;
v_z += 1u;
}
} else {
v_z = 0u;
while (v_z < 256u) {
self->private_impl.f_yuv_cache[(1u + (v_z >> 4u))][(8u + (v_z & 15u))] = 128u;
v_z += 1u;
}
}
return wuffs_base__make_empty_struct();
}
// -------- func vp8.decoder.predict_uv8
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__predict_uv8(
wuffs_vp8__decoder* self,
uint32_t a_b,
uint32_t a_mode) {
uint32_t v_cachey = 0;
uint32_t v_cachex = 0;
uint32_t v_x = 0;
uint32_t v_y = 0;
uint32_t v_z = 0;
uint8_t v_avg = 0;
uint8_t v_val = 0;
uint32_t v_tm0 = 0;
uint32_t v_tm1 = 0;
uint32_t v_tm2 = 0;
v_cachey = 18u;
v_cachex = ((a_b * 16u) + 8u);
if (a_mode == 0u) {
v_avg = ((uint8_t)(((8u +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 4u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 5u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 6u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 7u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 4u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 5u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 6u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 7u)][(v_cachex - 1u)]))) / 16u)));
v_z = 0u;
while (v_z < 64u) {
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex + (v_z & 7u))] = v_avg;
v_z += 1u;
}
} else if (a_mode == 1u) {
v_tm0 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex - 1u)]));
v_z = 0u;
while (v_z < 64u) {
v_tm1 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex - 1u)]));
v_tm2 = ((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + (v_z & 7u))]));
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex + (v_z & 7u))] = WUFFS_VP8__CLAMP[(((uint32_t)((v_tm1 + v_tm2) - v_tm0)) & 1023u)];
v_z += 1u;
}
} else if (a_mode == 2u) {
v_x = 0u;
while (v_x < 8u) {
v_val = self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + v_x)];
v_y = 0u;
while (v_y < 8u) {
self->private_impl.f_yuv_cache[(v_cachey + v_y)][(v_cachex + v_x)] = v_val;
v_y += 1u;
}
v_x += 1u;
}
} else if (a_mode == 3u) {
v_y = 0u;
while (v_y < 8u) {
v_val = self->private_impl.f_yuv_cache[(v_cachey + v_y)][(v_cachex - 1u)];
v_x = 0u;
while (v_x < 8u) {
self->private_impl.f_yuv_cache[(v_cachey + v_y)][(v_cachex + v_x)] = v_val;
v_x += 1u;
}
v_y += 1u;
}
} else if (a_mode <= 10u) {
v_avg = ((uint8_t)(((4u +
((uint32_t)(self->private_impl.f_yuv_cache[v_cachey][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 1u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 2u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 3u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 4u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 5u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 6u)][(v_cachex - 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey + 7u)][(v_cachex - 1u)]))) / 8u)));
v_z = 0u;
while (v_z < 64u) {
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex + (v_z & 7u))] = v_avg;
v_z += 1u;
}
} else if (a_mode == 11u) {
v_avg = ((uint8_t)(((4u +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][v_cachex])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 1u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 2u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 3u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 4u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 5u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 6u)])) +
((uint32_t)(self->private_impl.f_yuv_cache[(v_cachey - 1u)][(v_cachex + 7u)]))) / 8u)));
v_z = 0u;
while (v_z < 64u) {
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex + (v_z & 7u))] = v_avg;
v_z += 1u;
}
} else {
v_z = 0u;
while (v_z < 64u) {
self->private_impl.f_yuv_cache[(v_cachey + (v_z >> 3u))][(v_cachex + (v_z & 7u))] = 128u;
v_z += 1u;
}
}
return wuffs_base__make_empty_struct();
}
// -------- func vp8.decoder.reconstruct
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__reconstruct(
wuffs_vp8__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mbx,
uint32_t a_mby,
uint32_t a_luma_mode,
uint32_t a_chroma_mode) {
uint32_t v_mode = 0;
uint32_t v_b = 0;
wuffs_vp8__decoder__prepare_yuv_cache(self, a_workbuf, a_mbx, a_mby);
if (a_luma_mode < 4u) {
v_mode = wuffs_vp8__decoder__substitute_dc_top_left(self, a_luma_mode, a_mbx, a_mby);
wuffs_vp8__decoder__predict_y16(self, v_mode);
} else {
v_b = 0u;
while (v_b < 16u) {
wuffs_vp8__decoder__predict_y4(self, v_b);
v_b += 1u;
}
}
v_mode = wuffs_vp8__decoder__substitute_dc_top_left(self, a_chroma_mode, a_mbx, a_mby);
wuffs_vp8__decoder__predict_uv8(self, 0u, v_mode);
wuffs_vp8__decoder__predict_uv8(self, 1u, v_mode);
return wuffs_base__make_empty_struct();
}
// -------- func vp8.decoder.substitute_dc_top_left
WUFFS_BASE__GENERATED_C_CODE
static uint32_t
wuffs_vp8__decoder__substitute_dc_top_left(
const wuffs_vp8__decoder* self,
uint32_t a_mode,
uint32_t a_mbx,
uint32_t a_mby) {
if (a_mode == 0u) {
if (a_mby == 0u) {
if (a_mbx == 0u) {
return 11u;
} else {
return 12u;
}
} else if (a_mby == 0u) {
return 10u;
}
}
return a_mode;
}
// -------- func vp8.decoder.prepare_yuv_cache
WUFFS_BASE__GENERATED_C_CODE
static wuffs_base__empty_struct
wuffs_vp8__decoder__prepare_yuv_cache(
wuffs_vp8__decoder* self,
wuffs_base__slice_u8 a_workbuf,
uint32_t a_mbx,
uint32_t a_mby) {
uint32_t v_i = 0;
uint32_t v_j = 0;
uint64_t v_offset = 0;
uint64_t v_o = 0;
if (a_mbx <= 0u) {
v_j = 0u;
while (v_j < 17u) {
self->private_impl.f_yuv_cache[v_j][7u] = 129u;
v_j += 1u;
}
while (v_j < 26u) {
self->private_impl.f_yuv_cache[v_j][7u] = 129u;
self->private_impl.f_yuv_cache[v_j][23u] = 129u;
v_j += 1u;
}
} else {
v_j = 0u;
while (v_j < 17u) {
self->private_impl.f_yuv_cache[v_j][7u] = self->private_impl.f_yuv_cache[v_j][23u];
v_j += 1u;
}
while (v_j < 26u) {
self->private_impl.f_yuv_cache[v_j][7u] = self->private_impl.f_yuv_cache[v_j][15u];
self->private_impl.f_yuv_cache[v_j][23u] = self->private_impl.f_yuv_cache[v_j][31u];
v_j += 1u;
}
}
if (a_mby <= 0u) {
v_i = 0u;
while (v_i < 32u) {
self->private_impl.f_yuv_cache[0u][v_i] = 127u;
v_i += 1u;
}
v_i = 0u;
while (v_i < 32u) {
self->private_impl.f_yuv_cache[17u][v_i] = 127u;
v_i += 1u;
}
} else {
v_offset = ((uint64_t)(((((a_mby - 1u) * self->private_impl.f_workbuf_yuv_y_stride) + a_mbx) * 16u)));
v_i = 0u;
while (v_i < 16u) {
v_o = (v_offset + ((uint64_t)(v_i)));
if (v_o < ((uint64_t)(a_workbuf.len))) {
self->private_impl.f_yuv_cache[0u][(8u + v_i)] = a_workbuf.ptr[v_o];
}
v_i += 1u;
}
if ((a_mbx + 1u) < self->private_impl.f_mbw) {
while (v_i < 20u) {
v_o = (v_offset + ((uint64_t)(v_i)));
if (v_o < ((uint64_t)(a_workbuf.len))) {
self->private_impl.f_yuv_cache[0u][(8u + v_i)] = a_workbuf.ptr[v_o];
}
v_i += 1u;
}
} else {
while (v_i < 20u) {
v_o = (v_offset + ((uint64_t)(15u)));
if (v_o < ((uint64_t)(a_workbuf.len))) {
self->private_impl.f_yuv_cache[0u][(8u + v_i)] = a_workbuf.ptr[v_o];
}
v_i += 1u;
}
}
v_offset = (self->private_impl.f_workbuf_yuv_y_end + ((uint64_t)(((((a_mby - 1u) * self->private_impl.f_workbuf_yuv_uv_stride) + a_mbx) * 8u))));
v_i = 0u;
while (v_i < 8u) {
v_o = (v_offset + ((uint64_t)(v_i)));
if (v_o < ((uint64_t)(a_workbuf.len))) {
self->private_impl.f_yuv_cache[17u][(8u + v_i)] = a_workbuf.ptr[v_o];
}
v_i += 1u;
}
v_offset = (self->private_impl.f_workbuf_yuv_u_end + ((uint64_t)(((((a_mby - 1u) * self->private_impl.f_workbuf_yuv_uv_stride) + a_mbx) * 8u))));
v_i = 0u;
while (v_i < 8u) {
v_o = (v_offset + ((uint64_t)(v_i)));
if (v_o < ((uint64_t)(a_workbuf.len))) {
self->private_impl.f_yuv_cache[17u][(24u + v_i)] = a_workbuf.ptr[v_o];
}
v_i += 1u;
}
}
v_j = 4u;
while (v_j < 16u) {
self->private_impl.f_yuv_cache[v_j][24u] = self->private_impl.f_yuv_cache[0u][24u];
self->private_impl.f_yuv_cache[v_j][25u] = self->private_impl.f_yuv_cache[0u][25u];
self->private_impl.f_yuv_cache[v_j][26u] = self->private_impl.f_yuv_cache[0u][26u];
self->private_impl.f_yuv_cache[v_j][27u] = self->private_impl.f_yuv_cache[0u][27u];
v_j += 4u;
}
return wuffs_base__make_empty_struct();
}
// -------- func vp8.decoder.get_quirk
WUFFS_BASE__GENERATED_C_CODE
+74
View File
@@ -111,6 +111,80 @@ pri const CLAMP_NO_MORE_THAN_9 : roarray[16] base.u8[..= 9] = [
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 9, 9, 9, 9, 9, 9,
]
pri const CLAMP : roarray[1024] base.u8 = [
0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F,
0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B, 0x3C, 0x3D, 0x3E, 0x3F,
0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x5D, 0x5E, 0x5F,
0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F,
0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F,
0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF,
0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF,
0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7, 0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
]
// RFC 6386 Section 11.5. Subblock Mode Probability Table.
//
// Note that the specification (the RFC) differs from the implementation
+1 -2
View File
@@ -13,8 +13,7 @@ pri func decoder.decode_coefficients!(
mbx: base.u32[..= 0x3FF],
mby: base.u32[..= 0x3FF],
seg: base.u32[..= 3],
luma_mode: base.u32[..= 4],
chroma_mode: base.u32[..= 3]) {
luma_mode: base.u32[..= 4]) {
var bc : base.u32
var by : base.u32
var bx : base.u32
+5 -1
View File
@@ -82,8 +82,12 @@ pri func decoder.decode_one_macroblock!(workbuf: slice base.u8, mbx: base.u32[..
} else {
this.decode_coefficients!(
workbuf: args.workbuf, mbx: args.mbx, mby: args.mby,
seg: seg, luma_mode: luma_mode, chroma_mode: chroma_mode)
seg: seg, luma_mode: luma_mode)
}
this.reconstruct!(
workbuf: args.workbuf, mbx: args.mbx, mby: args.mby,
luma_mode: luma_mode, chroma_mode: chroma_mode)
}
// RFC 6386 Section 11.2. Luma Modes.
+675
View File
@@ -0,0 +1,675 @@
// Copyright 2026 The Wuffs Authors.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
//
// SPDX-License-Identifier: Apache-2.0 OR MIT
// --------
// Predicts a 4×4 luma block, the Xs below, based on samples top and left.
//
// a b c d e f g h i
// p X X X X
// q X X X X
// r X X X X
// s X X X X
pri func decoder.predict_y4!(b: base.u32[..= 15]) {
var cachey : base.u32[..= 13]
var cachex : base.u32[..= 20]
var mode : base.u32[..= 9]
var z : base.u32
var avg : base.u8
var tm0 : base.u32
var tm1 : base.u32
var tm2 : base.u32
var a : base.u32
var b : base.u32
var c : base.u32
var d : base.u32
var e : base.u32
var f : base.u32
var g : base.u32
var h : base.u32
var i : base.u32
var p : base.u32
var q : base.u32
var r : base.u32
var s : base.u32
var ab : base.u8
var bc : base.u8
var cd : base.u8
var de : base.u8
var ef : base.u8
var pa : base.u8
var qp : base.u8
var rq : base.u8
var sr : base.u8
var abc : base.u8
var bcd : base.u8
var cde : base.u8
var def : base.u8
var efg : base.u8
var fgh : base.u8
var ghi : base.u8
var hii : base.u8
var pab : base.u8
var qpa : base.u8
var rqp : base.u8
var srq : base.u8
var ssr : base.u8
var sss : base.u8
cachey = ((args.b & 0xC) * 1) + 1
cachex = ((args.b & 0x3) * 4) + 8
mode = this.mb_subblock_modes[args.b] as base.u32
if mode == 0 { // DC.
avg = ((4 +
(this.yuv_cache[cachey - 1][cachex + 0] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 1] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 2] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 3] as base.u32) +
(this.yuv_cache[cachey + 0][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 1][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 2][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 3][cachex - 1] as base.u32)) / 8) as base.u8
z = 0
while z < 16 {
this.yuv_cache[cachey + (z >> 2)][cachex + (z & 3)] = avg
z += 1
}
} else if mode == 1 { // TM.
tm0 = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
z = 0
while z < 16,
inv cachey >= 1,
inv cachex >= 8,
{
tm1 = this.yuv_cache[cachey + (z >> 2)][cachex - 1] as base.u32
tm2 = this.yuv_cache[cachey - 1][cachex + (z & 3)] as base.u32
this.yuv_cache[cachey + (z >> 2)][cachex + (z & 3)] =
CLAMP[((tm1 + tm2) ~mod- tm0) & 1023]
z += 1
}
} else if mode == 2 { // VE.
a = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
e = this.yuv_cache[cachey - 1][cachex + 3] as base.u32
f = this.yuv_cache[cachey - 1][cachex + 4] as base.u32
def = (((d + (2 * e) + f + 2) / 4) & 0xFF) as base.u8
cde = (((c + (2 * d) + e + 2) / 4) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
abc = (((a + (2 * b) + c + 2) / 4) & 0xFF) as base.u8
z = 0
while z < 4 {
this.yuv_cache[cachey + z][cachex + 0] = abc
this.yuv_cache[cachey + z][cachex + 1] = bcd
this.yuv_cache[cachey + z][cachex + 2] = cde
this.yuv_cache[cachey + z][cachex + 3] = def
z += 1
}
} else if mode == 3 { // HE.
s = this.yuv_cache[cachey + 3][cachex - 1] as base.u32
r = this.yuv_cache[cachey + 2][cachex - 1] as base.u32
q = this.yuv_cache[cachey + 1][cachex - 1] as base.u32
p = this.yuv_cache[cachey + 0][cachex - 1] as base.u32
a = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
qpa = (((q + (2 * p) + a + 2) / 4) & 0xFF) as base.u8
rqp = (((r + (2 * q) + p + 2) / 4) & 0xFF) as base.u8
srq = (((s + (2 * r) + q + 2) / 4) & 0xFF) as base.u8
ssr = (((s + (2 * s) + r + 2) / 4) & 0xFF) as base.u8
z = 0
while z < 4 {
this.yuv_cache[cachey + 0][cachex + z] = qpa
this.yuv_cache[cachey + 1][cachex + z] = rqp
this.yuv_cache[cachey + 2][cachex + z] = srq
this.yuv_cache[cachey + 3][cachex + z] = ssr
z += 1
}
} else if mode == 4 { // RD.
s = this.yuv_cache[cachey + 3][cachex - 1] as base.u32
r = this.yuv_cache[cachey + 2][cachex - 1] as base.u32
q = this.yuv_cache[cachey + 1][cachex - 1] as base.u32
p = this.yuv_cache[cachey + 0][cachex - 1] as base.u32
a = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
e = this.yuv_cache[cachey - 1][cachex + 3] as base.u32
cde = (((c + (2 * d) + e + 2) / 4) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
abc = (((a + (2 * b) + c + 2) / 4) & 0xFF) as base.u8
pab = (((p + (2 * a) + b + 2) / 4) & 0xFF) as base.u8
qpa = (((q + (2 * p) + a + 2) / 4) & 0xFF) as base.u8
rqp = (((r + (2 * q) + p + 2) / 4) & 0xFF) as base.u8
srq = (((s + (2 * r) + q + 2) / 4) & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = pab
this.yuv_cache[cachey + 0][cachex + 1] = abc
this.yuv_cache[cachey + 0][cachex + 2] = bcd
this.yuv_cache[cachey + 0][cachex + 3] = cde
this.yuv_cache[cachey + 1][cachex + 0] = qpa
this.yuv_cache[cachey + 1][cachex + 1] = pab
this.yuv_cache[cachey + 1][cachex + 2] = abc
this.yuv_cache[cachey + 1][cachex + 3] = bcd
this.yuv_cache[cachey + 2][cachex + 0] = rqp
this.yuv_cache[cachey + 2][cachex + 1] = qpa
this.yuv_cache[cachey + 2][cachex + 2] = pab
this.yuv_cache[cachey + 2][cachex + 3] = abc
this.yuv_cache[cachey + 3][cachex + 0] = srq
this.yuv_cache[cachey + 3][cachex + 1] = rqp
this.yuv_cache[cachey + 3][cachex + 2] = qpa
this.yuv_cache[cachey + 3][cachex + 3] = pab
} else if mode == 5 { // VR.
r = this.yuv_cache[cachey + 2][cachex - 1] as base.u32
q = this.yuv_cache[cachey + 1][cachex - 1] as base.u32
p = this.yuv_cache[cachey + 0][cachex - 1] as base.u32
a = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
e = this.yuv_cache[cachey - 1][cachex + 3] as base.u32
ab = (((a + b + 1) / 2) & 0xFF) as base.u8
bc = (((b + c + 1) / 2) & 0xFF) as base.u8
cd = (((c + d + 1) / 2) & 0xFF) as base.u8
de = (((d + e + 1) / 2) & 0xFF) as base.u8
cde = (((c + (2 * d) + e + 2) / 4) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
abc = (((a + (2 * b) + c + 2) / 4) & 0xFF) as base.u8
pab = (((p + (2 * a) + b + 2) / 4) & 0xFF) as base.u8
qpa = (((q + (2 * p) + a + 2) / 4) & 0xFF) as base.u8
rqp = (((r + (2 * q) + p + 2) / 4) & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = ab
this.yuv_cache[cachey + 0][cachex + 1] = bc
this.yuv_cache[cachey + 0][cachex + 2] = cd
this.yuv_cache[cachey + 0][cachex + 3] = de
this.yuv_cache[cachey + 1][cachex + 0] = pab
this.yuv_cache[cachey + 1][cachex + 1] = abc
this.yuv_cache[cachey + 1][cachex + 2] = bcd
this.yuv_cache[cachey + 1][cachex + 3] = cde
this.yuv_cache[cachey + 2][cachex + 0] = qpa
this.yuv_cache[cachey + 2][cachex + 1] = ab
this.yuv_cache[cachey + 2][cachex + 2] = bc
this.yuv_cache[cachey + 2][cachex + 3] = cd
this.yuv_cache[cachey + 3][cachex + 0] = rqp
this.yuv_cache[cachey + 3][cachex + 1] = pab
this.yuv_cache[cachey + 3][cachex + 2] = abc
this.yuv_cache[cachey + 3][cachex + 3] = bcd
} else if mode == 6 { // LD.
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
e = this.yuv_cache[cachey - 1][cachex + 3] as base.u32
f = this.yuv_cache[cachey - 1][cachex + 4] as base.u32
g = this.yuv_cache[cachey - 1][cachex + 5] as base.u32
h = this.yuv_cache[cachey - 1][cachex + 6] as base.u32
i = this.yuv_cache[cachey - 1][cachex + 7] as base.u32
hii = (((h + (2 * i) + i + 2) / 4) & 0xFF) as base.u8
ghi = (((g + (2 * h) + i + 2) / 4) & 0xFF) as base.u8
fgh = (((f + (2 * g) + h + 2) / 4) & 0xFF) as base.u8
efg = (((e + (2 * f) + g + 2) / 4) & 0xFF) as base.u8
def = (((d + (2 * e) + f + 2) / 4) & 0xFF) as base.u8
cde = (((c + (2 * d) + e + 2) / 4) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = bcd
this.yuv_cache[cachey + 0][cachex + 1] = cde
this.yuv_cache[cachey + 0][cachex + 2] = def
this.yuv_cache[cachey + 0][cachex + 3] = efg
this.yuv_cache[cachey + 1][cachex + 0] = cde
this.yuv_cache[cachey + 1][cachex + 1] = def
this.yuv_cache[cachey + 1][cachex + 2] = efg
this.yuv_cache[cachey + 1][cachex + 3] = fgh
this.yuv_cache[cachey + 2][cachex + 0] = def
this.yuv_cache[cachey + 2][cachex + 1] = efg
this.yuv_cache[cachey + 2][cachex + 2] = fgh
this.yuv_cache[cachey + 2][cachex + 3] = ghi
this.yuv_cache[cachey + 3][cachex + 0] = efg
this.yuv_cache[cachey + 3][cachex + 1] = fgh
this.yuv_cache[cachey + 3][cachex + 2] = ghi
this.yuv_cache[cachey + 3][cachex + 3] = hii
} else if mode == 7 { // VL.
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
e = this.yuv_cache[cachey - 1][cachex + 3] as base.u32
f = this.yuv_cache[cachey - 1][cachex + 4] as base.u32
g = this.yuv_cache[cachey - 1][cachex + 5] as base.u32
h = this.yuv_cache[cachey - 1][cachex + 6] as base.u32
i = this.yuv_cache[cachey - 1][cachex + 7] as base.u32
ef = (((e + f + 1) / 2) & 0xFF) as base.u8
de = (((d + e + 1) / 2) & 0xFF) as base.u8
cd = (((c + d + 1) / 2) & 0xFF) as base.u8
bc = (((b + c + 1) / 2) & 0xFF) as base.u8
ghi = (((g + (2 * h) + i + 2) / 4) & 0xFF) as base.u8
fgh = (((f + (2 * g) + h + 2) / 4) & 0xFF) as base.u8
efg = (((e + (2 * f) + g + 2) / 4) & 0xFF) as base.u8
def = (((d + (2 * e) + f + 2) / 4) & 0xFF) as base.u8
cde = (((c + (2 * d) + e + 2) / 4) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = bc
this.yuv_cache[cachey + 0][cachex + 1] = cd
this.yuv_cache[cachey + 0][cachex + 2] = de
this.yuv_cache[cachey + 0][cachex + 3] = ef
this.yuv_cache[cachey + 1][cachex + 0] = bcd
this.yuv_cache[cachey + 1][cachex + 1] = cde
this.yuv_cache[cachey + 1][cachex + 2] = def
this.yuv_cache[cachey + 1][cachex + 3] = efg
this.yuv_cache[cachey + 2][cachex + 0] = cd
this.yuv_cache[cachey + 2][cachex + 1] = de
this.yuv_cache[cachey + 2][cachex + 2] = ef
this.yuv_cache[cachey + 2][cachex + 3] = fgh
this.yuv_cache[cachey + 3][cachex + 0] = cde
this.yuv_cache[cachey + 3][cachex + 1] = def
this.yuv_cache[cachey + 3][cachex + 2] = efg
this.yuv_cache[cachey + 3][cachex + 3] = ghi
} else if mode == 8 { // HD.
s = this.yuv_cache[cachey + 3][cachex - 1] as base.u32
r = this.yuv_cache[cachey + 2][cachex - 1] as base.u32
q = this.yuv_cache[cachey + 1][cachex - 1] as base.u32
p = this.yuv_cache[cachey + 0][cachex - 1] as base.u32
a = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
b = this.yuv_cache[cachey - 1][cachex + 0] as base.u32
c = this.yuv_cache[cachey - 1][cachex + 1] as base.u32
d = this.yuv_cache[cachey - 1][cachex + 2] as base.u32
sr = (((s + r + 1) / 2) & 0xFF) as base.u8
rq = (((r + q + 1) / 2) & 0xFF) as base.u8
qp = (((q + p + 1) / 2) & 0xFF) as base.u8
pa = (((p + a + 1) / 2) & 0xFF) as base.u8
bcd = (((b + (2 * c) + d + 2) / 4) & 0xFF) as base.u8
abc = (((a + (2 * b) + c + 2) / 4) & 0xFF) as base.u8
pab = (((p + (2 * a) + b + 2) / 4) & 0xFF) as base.u8
qpa = (((q + (2 * p) + a + 2) / 4) & 0xFF) as base.u8
rqp = (((r + (2 * q) + p + 2) / 4) & 0xFF) as base.u8
srq = (((s + (2 * r) + q + 2) / 4) & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = pa
this.yuv_cache[cachey + 0][cachex + 1] = pab
this.yuv_cache[cachey + 0][cachex + 2] = abc
this.yuv_cache[cachey + 0][cachex + 3] = bcd
this.yuv_cache[cachey + 1][cachex + 0] = qp
this.yuv_cache[cachey + 1][cachex + 1] = qpa
this.yuv_cache[cachey + 1][cachex + 2] = pa
this.yuv_cache[cachey + 1][cachex + 3] = pab
this.yuv_cache[cachey + 2][cachex + 0] = rq
this.yuv_cache[cachey + 2][cachex + 1] = rqp
this.yuv_cache[cachey + 2][cachex + 2] = qp
this.yuv_cache[cachey + 2][cachex + 3] = qpa
this.yuv_cache[cachey + 3][cachex + 0] = sr
this.yuv_cache[cachey + 3][cachex + 1] = srq
this.yuv_cache[cachey + 3][cachex + 2] = rq
this.yuv_cache[cachey + 3][cachex + 3] = rqp
} else { // HU.
s = this.yuv_cache[cachey + 3][cachex - 1] as base.u32
r = this.yuv_cache[cachey + 2][cachex - 1] as base.u32
q = this.yuv_cache[cachey + 1][cachex - 1] as base.u32
p = this.yuv_cache[cachey + 0][cachex - 1] as base.u32
sr = (((s + r + 1) / 2) & 0xFF) as base.u8
rq = (((r + q + 1) / 2) & 0xFF) as base.u8
qp = (((q + p + 1) / 2) & 0xFF) as base.u8
rqp = (((r + (2 * q) + p + 2) / 4) & 0xFF) as base.u8
srq = (((s + (2 * r) + q + 2) / 4) & 0xFF) as base.u8
ssr = (((s + (2 * s) + r + 2) / 4) & 0xFF) as base.u8
sss = (s & 0xFF) as base.u8
this.yuv_cache[cachey + 0][cachex + 0] = qp
this.yuv_cache[cachey + 0][cachex + 1] = rqp
this.yuv_cache[cachey + 0][cachex + 2] = rq
this.yuv_cache[cachey + 0][cachex + 3] = srq
this.yuv_cache[cachey + 1][cachex + 0] = rq
this.yuv_cache[cachey + 1][cachex + 1] = srq
this.yuv_cache[cachey + 1][cachex + 2] = sr
this.yuv_cache[cachey + 1][cachex + 3] = ssr
this.yuv_cache[cachey + 2][cachex + 0] = sr
this.yuv_cache[cachey + 2][cachex + 1] = ssr
this.yuv_cache[cachey + 2][cachex + 2] = sss
this.yuv_cache[cachey + 2][cachex + 3] = sss
this.yuv_cache[cachey + 3][cachex + 0] = sss
this.yuv_cache[cachey + 3][cachex + 1] = sss
this.yuv_cache[cachey + 3][cachex + 2] = sss
this.yuv_cache[cachey + 3][cachex + 3] = sss
}
}
pri func decoder.predict_y16!(mode: base.u32[..= 12]) {
var x : base.u32
var y : base.u32
var z : base.u32
var avg : base.u8
var val : base.u8
var tm0 : base.u32
var tm1 : base.u32
var tm2 : base.u32
if args.mode == 0 { // DC.
avg = ((16 +
(this.yuv_cache[0x00][0x08] as base.u32) +
(this.yuv_cache[0x00][0x09] as base.u32) +
(this.yuv_cache[0x00][0x0A] as base.u32) +
(this.yuv_cache[0x00][0x0B] as base.u32) +
(this.yuv_cache[0x00][0x0C] as base.u32) +
(this.yuv_cache[0x00][0x0D] as base.u32) +
(this.yuv_cache[0x00][0x0E] as base.u32) +
(this.yuv_cache[0x00][0x0F] as base.u32) +
(this.yuv_cache[0x00][0x10] as base.u32) +
(this.yuv_cache[0x00][0x11] as base.u32) +
(this.yuv_cache[0x00][0x12] as base.u32) +
(this.yuv_cache[0x00][0x13] as base.u32) +
(this.yuv_cache[0x00][0x14] as base.u32) +
(this.yuv_cache[0x00][0x15] as base.u32) +
(this.yuv_cache[0x00][0x16] as base.u32) +
(this.yuv_cache[0x00][0x17] as base.u32) +
(this.yuv_cache[0x01][0x07] as base.u32) +
(this.yuv_cache[0x02][0x07] as base.u32) +
(this.yuv_cache[0x03][0x07] as base.u32) +
(this.yuv_cache[0x04][0x07] as base.u32) +
(this.yuv_cache[0x05][0x07] as base.u32) +
(this.yuv_cache[0x06][0x07] as base.u32) +
(this.yuv_cache[0x07][0x07] as base.u32) +
(this.yuv_cache[0x08][0x07] as base.u32) +
(this.yuv_cache[0x09][0x07] as base.u32) +
(this.yuv_cache[0x0A][0x07] as base.u32) +
(this.yuv_cache[0x0B][0x07] as base.u32) +
(this.yuv_cache[0x0C][0x07] as base.u32) +
(this.yuv_cache[0x0D][0x07] as base.u32) +
(this.yuv_cache[0x0E][0x07] as base.u32) +
(this.yuv_cache[0x0F][0x07] as base.u32) +
(this.yuv_cache[0x10][0x07] as base.u32)) / 32) as base.u8
z = 0
while z < 256 {
this.yuv_cache[0x01 + (z >> 4)][0x08 + (z & 15)] = avg
z += 1
}
} else if args.mode == 1 { // TM.
tm0 = this.yuv_cache[0x00][0x07] as base.u32
z = 0
while z < 256 {
tm1 = this.yuv_cache[0x01 + (z >> 4)][0x07] as base.u32
tm2 = this.yuv_cache[0x00][0x08 + (z & 15)] as base.u32
this.yuv_cache[0x01 + (z >> 4)][0x08 + (z & 15)] =
CLAMP[((tm1 + tm2) ~mod- tm0) & 1023]
z += 1
}
} else if args.mode == 2 { // VE.
x = 0
while x < 16 {
val = this.yuv_cache[0x00][0x08 + x]
y = 0
while y < 16,
inv x < 16,
{
this.yuv_cache[0x01 + y][0x08 + x] = val
y += 1
}
x += 1
}
} else if args.mode == 3 { // HE.
y = 0
while y < 16 {
val = this.yuv_cache[0x01 + y][0x07]
x = 0
while x < 16,
inv y < 16,
{
this.yuv_cache[0x01 + y][0x08 + x] = val
x += 1
}
y += 1
}
} else if args.mode <= 10 { // DCTop.
avg = ((8 +
(this.yuv_cache[0x01][0x07] as base.u32) +
(this.yuv_cache[0x02][0x07] as base.u32) +
(this.yuv_cache[0x03][0x07] as base.u32) +
(this.yuv_cache[0x04][0x07] as base.u32) +
(this.yuv_cache[0x05][0x07] as base.u32) +
(this.yuv_cache[0x06][0x07] as base.u32) +
(this.yuv_cache[0x07][0x07] as base.u32) +
(this.yuv_cache[0x08][0x07] as base.u32) +
(this.yuv_cache[0x09][0x07] as base.u32) +
(this.yuv_cache[0x0A][0x07] as base.u32) +
(this.yuv_cache[0x0B][0x07] as base.u32) +
(this.yuv_cache[0x0C][0x07] as base.u32) +
(this.yuv_cache[0x0D][0x07] as base.u32) +
(this.yuv_cache[0x0E][0x07] as base.u32) +
(this.yuv_cache[0x0F][0x07] as base.u32) +
(this.yuv_cache[0x10][0x07] as base.u32)) / 16) as base.u8
z = 0
while z < 256 {
this.yuv_cache[0x01 + (z >> 4)][0x08 + (z & 15)] = avg
z += 1
}
} else if args.mode == 11 { // DCLeft.
avg = ((8 +
(this.yuv_cache[0x00][0x08] as base.u32) +
(this.yuv_cache[0x00][0x09] as base.u32) +
(this.yuv_cache[0x00][0x0A] as base.u32) +
(this.yuv_cache[0x00][0x0B] as base.u32) +
(this.yuv_cache[0x00][0x0C] as base.u32) +
(this.yuv_cache[0x00][0x0D] as base.u32) +
(this.yuv_cache[0x00][0x0E] as base.u32) +
(this.yuv_cache[0x00][0x0F] as base.u32) +
(this.yuv_cache[0x00][0x10] as base.u32) +
(this.yuv_cache[0x00][0x11] as base.u32) +
(this.yuv_cache[0x00][0x12] as base.u32) +
(this.yuv_cache[0x00][0x13] as base.u32) +
(this.yuv_cache[0x00][0x14] as base.u32) +
(this.yuv_cache[0x00][0x15] as base.u32) +
(this.yuv_cache[0x00][0x16] as base.u32) +
(this.yuv_cache[0x00][0x17] as base.u32)) / 16) as base.u8
z = 0
while z < 256 {
this.yuv_cache[0x01 + (z >> 4)][0x08 + (z & 15)] = avg
z += 1
}
} else { // DCTopLeft.
z = 0
while z < 256 {
this.yuv_cache[0x01 + (z >> 4)][0x08 + (z & 15)] = 0x80
z += 1
}
}
}
pri func decoder.predict_uv8!(b: base.u32[..= 1], mode: base.u32[..= 12]) {
var cachey : base.u32[..= 18]
var cachex : base.u32[..= 24]
var x : base.u32
var y : base.u32
var z : base.u32
var avg : base.u8
var val : base.u8
var tm0 : base.u32
var tm1 : base.u32
var tm2 : base.u32
cachey = 0x12
cachex = (args.b * 16) + 8
if args.mode == 0 { // DC.
avg = ((8 +
(this.yuv_cache[cachey - 1][cachex + 0] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 1] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 2] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 3] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 4] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 5] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 6] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 7] as base.u32) +
(this.yuv_cache[cachey + 0][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 1][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 2][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 3][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 4][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 5][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 6][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 7][cachex - 1] as base.u32)) / 16) as base.u8
z = 0
while z < 64 {
this.yuv_cache[cachey + (z >> 3)][cachex + (z & 7)] = avg
z += 1
}
} else if args.mode == 1 { // TM.
tm0 = this.yuv_cache[cachey - 1][cachex - 1] as base.u32
z = 0
while z < 64,
inv cachey == 0x12,
inv cachex >= 8,
{
tm1 = this.yuv_cache[cachey + (z >> 3)][cachex - 1] as base.u32
tm2 = this.yuv_cache[cachey - 1][cachex + (z & 7)] as base.u32
this.yuv_cache[cachey + (z >> 3)][cachex + (z & 7)] =
CLAMP[((tm1 + tm2) ~mod- tm0) & 1023]
z += 1
}
} else if args.mode == 2 { // VE.
x = 0
while x < 8,
inv cachey == 0x12,
inv cachex >= 8,
{
val = this.yuv_cache[cachey - 1][cachex + x]
y = 0
while y < 8,
inv cachey == 0x12,
inv cachex >= 8,
inv x < 8,
{
this.yuv_cache[cachey + y][cachex + x] = val
y += 1
}
x += 1
}
} else if args.mode == 3 { // HE.
y = 0
while y < 8,
inv cachey == 0x12,
inv cachex >= 8,
{
val = this.yuv_cache[cachey + y][cachex - 1]
x = 0
while x < 8,
inv cachey == 0x12,
inv cachex >= 8,
inv y < 8,
{
this.yuv_cache[cachey + y][cachex + x] = val
x += 1
}
y += 1
}
} else if args.mode <= 10 { // DCTop.
avg = ((4 +
(this.yuv_cache[cachey + 0][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 1][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 2][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 3][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 4][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 5][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 6][cachex - 1] as base.u32) +
(this.yuv_cache[cachey + 7][cachex - 1] as base.u32)) / 8) as base.u8
z = 0
while z < 64 {
this.yuv_cache[cachey + (z >> 3)][cachex + (z & 7)] = avg
z += 1
}
} else if args.mode == 11 { // DCLeft.
avg = ((4 +
(this.yuv_cache[cachey - 1][cachex + 0] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 1] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 2] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 3] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 4] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 5] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 6] as base.u32) +
(this.yuv_cache[cachey - 1][cachex + 7] as base.u32)) / 8) as base.u8
z = 0
while z < 64 {
this.yuv_cache[cachey + (z >> 3)][cachex + (z & 7)] = avg
z += 1
}
} else { // DCTopLeft.
z = 0
while z < 64 {
this.yuv_cache[cachey + (z >> 3)][cachex + (z & 7)] = 0x80
z += 1
}
}
}
+176
View File
@@ -0,0 +1,176 @@
// Copyright 2026 The Wuffs Authors.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.
//
// SPDX-License-Identifier: Apache-2.0 OR MIT
pri func decoder.reconstruct!(
workbuf: slice base.u8,
mbx: base.u32[..= 0x3FF],
mby: base.u32[..= 0x3FF],
luma_mode: base.u32[..= 4],
chroma_mode: base.u32[..= 3]) {
var mode : base.u32[..= 12]
var b : base.u32
this.prepare_yuv_cache!(workbuf: args.workbuf, mbx: args.mbx, mby: args.mby)
if args.luma_mode < 4 {
mode = this.substitute_dc_top_left(mode: args.luma_mode, mbx: args.mbx, mby: args.mby)
this.predict_y16!(mode: mode)
} else {
b = 0
while b < 16 {
this.predict_y4!(b: b)
b += 1
}
}
mode = this.substitute_dc_top_left(mode: args.chroma_mode, mbx: args.mbx, mby: args.mby)
this.predict_uv8!(b: 0, mode: mode)
this.predict_uv8!(b: 1, mode: mode)
}
pri func decoder.substitute_dc_top_left(
mode: base.u32[..= 3],
mbx: base.u32[..= 0x3FF],
mby: base.u32[..= 0x3FF]) base.u32[..= 12] {
if args.mode == 0 { // DC.
if args.mby == 0 {
if args.mbx == 0 {
return 11 // DCLeft.
} else {
return 12 // DCTopLeft.
}
} else if args.mby == 0 {
return 10 // DCTop.
}
}
return args.mode
}
pri func decoder.prepare_yuv_cache!(
workbuf: roslice base.u8,
mbx: base.u32[..= 0x3FF],
mby: base.u32[..= 0x3FF]) {
var i : base.u32
var j : base.u32
var offset : base.u64[..= 0x17FE_1FF8]
var o : base.u64
if args.mbx <= 0 {
j = 0x00
while j < 0x11 {
this.yuv_cache[j][0x07] = 0x81
j += 1
}
while j < 0x1A {
this.yuv_cache[j][0x07] = 0x81
this.yuv_cache[j][0x17] = 0x81
j += 1
}
} else {
j = 0x00
while j < 0x11 {
this.yuv_cache[j][0x07] = this.yuv_cache[j][0x17]
j += 1
}
while j < 0x1A {
this.yuv_cache[j][0x07] = this.yuv_cache[j][0x0F]
this.yuv_cache[j][0x17] = this.yuv_cache[j][0x1F]
j += 1
}
}
if args.mby <= 0 {
i = 0x00
while i < 0x20 {
this.yuv_cache[0x00][i] = 0x7F
i += 1
}
i = 0x00
while i < 0x20 {
this.yuv_cache[0x11][i] = 0x7F
i += 1
}
} else {
offset = 0 +
(((((args.mby - 1) * this.workbuf_yuv_y_stride) + args.mbx) * 16) as base.u64)
i = 0x00
while i < 0x10,
inv args.mby > 0,
{
o = offset + (i as base.u64)
if o < args.workbuf.length() {
this.yuv_cache[0x00][0x08 + i] = args.workbuf[o]
}
i += 1
}
if (args.mbx + 1) < this.mbw {
while i < 0x14,
inv args.mby > 0,
{
o = offset + (i as base.u64)
if o < args.workbuf.length() {
this.yuv_cache[0x00][0x08 + i] = args.workbuf[o]
}
i += 1
}
} else {
while i < 0x14,
inv args.mby > 0,
{
o = offset + (0x0F as base.u64)
if o < args.workbuf.length() {
this.yuv_cache[0x00][0x08 + i] = args.workbuf[o]
}
i += 1
}
}
offset = this.workbuf_yuv_y_end +
(((((args.mby - 1) * this.workbuf_yuv_uv_stride) + args.mbx) * 8) as base.u64)
i = 0x00
while i < 0x08,
inv args.mby > 0,
{
o = offset + (i as base.u64)
if o < args.workbuf.length() {
this.yuv_cache[0x11][0x08 + i] = args.workbuf[o]
}
i += 1
}
offset = this.workbuf_yuv_u_end +
(((((args.mby - 1) * this.workbuf_yuv_uv_stride) + args.mbx) * 8) as base.u64)
i = 0x00
while i < 0x08,
inv args.mby > 0,
{
o = offset + (i as base.u64)
if o < args.workbuf.length() {
this.yuv_cache[0x11][0x18 + i] = args.workbuf[o]
}
i += 1
}
}
j = 0x04
while j < 0x10 {
this.yuv_cache[j][0x18] = this.yuv_cache[0][0x18]
this.yuv_cache[j][0x19] = this.yuv_cache[0][0x19]
this.yuv_cache[j][0x1A] = this.yuv_cache[0][0x1A]
this.yuv_cache[j][0x1B] = this.yuv_cache[0][0x1B]
j += 4
}
}
+2
View File
@@ -100,6 +100,8 @@ pub struct decoder? implements base.image_decoder(
mb_subblock_modes : array[16] base.u8[..= 9],
yuv_cache : array[26] array[32] base.u8,
dst_x : base.u32,
dst_y : base.u32,
+293 -2
View File
@@ -65,7 +65,296 @@ the first "./a.out" with "./a.out -bench". Combine these changes with the
// ---------------- VP8 Tests
// No VP8 tests.
void //
initialize_decoder_test_state(wuffs_vp8__decoder* dec) {
// Initialize the top and left luma (Y) pixels to the digits of pi.
dec->private_impl.f_yuv_cache[0x00][0x07] = 0x31;
dec->private_impl.f_yuv_cache[0x00][0x08] = 0x41;
dec->private_impl.f_yuv_cache[0x00][0x09] = 0x59;
dec->private_impl.f_yuv_cache[0x00][0x0A] = 0x26;
dec->private_impl.f_yuv_cache[0x00][0x0B] = 0x53;
dec->private_impl.f_yuv_cache[0x00][0x0C] = 0x58;
dec->private_impl.f_yuv_cache[0x00][0x0D] = 0x97;
dec->private_impl.f_yuv_cache[0x00][0x0E] = 0x93;
dec->private_impl.f_yuv_cache[0x00][0x0F] = 0x23;
dec->private_impl.f_yuv_cache[0x00][0x10] = 0x84;
dec->private_impl.f_yuv_cache[0x00][0x11] = 0x62;
dec->private_impl.f_yuv_cache[0x00][0x12] = 0x64;
dec->private_impl.f_yuv_cache[0x00][0x13] = 0x33;
dec->private_impl.f_yuv_cache[0x00][0x14] = 0x83;
dec->private_impl.f_yuv_cache[0x00][0x15] = 0x27;
dec->private_impl.f_yuv_cache[0x00][0x16] = 0x95;
dec->private_impl.f_yuv_cache[0x00][0x17] = 0x02;
dec->private_impl.f_yuv_cache[0x01][0x07] = 0x88;
dec->private_impl.f_yuv_cache[0x02][0x07] = 0x41;
dec->private_impl.f_yuv_cache[0x03][0x07] = 0x97;
dec->private_impl.f_yuv_cache[0x04][0x07] = 0x16;
dec->private_impl.f_yuv_cache[0x05][0x07] = 0x93;
dec->private_impl.f_yuv_cache[0x06][0x07] = 0x99;
dec->private_impl.f_yuv_cache[0x07][0x07] = 0x37;
dec->private_impl.f_yuv_cache[0x08][0x07] = 0x51;
dec->private_impl.f_yuv_cache[0x09][0x07] = 0x05;
dec->private_impl.f_yuv_cache[0x0A][0x07] = 0x82;
dec->private_impl.f_yuv_cache[0x0B][0x07] = 0x09;
dec->private_impl.f_yuv_cache[0x0C][0x07] = 0x74;
dec->private_impl.f_yuv_cache[0x0D][0x07] = 0x94;
dec->private_impl.f_yuv_cache[0x0E][0x07] = 0x45;
dec->private_impl.f_yuv_cache[0x0F][0x07] = 0x92;
dec->private_impl.f_yuv_cache[0x10][0x07] = 0x30;
// Initialize the top and left chroma-blue (U) pixels to the digits of e.
dec->private_impl.f_yuv_cache[0x11][0x07] = 0x27;
dec->private_impl.f_yuv_cache[0x11][0x08] = 0x18;
dec->private_impl.f_yuv_cache[0x11][0x09] = 0x28;
dec->private_impl.f_yuv_cache[0x11][0x0A] = 0x18;
dec->private_impl.f_yuv_cache[0x11][0x0B] = 0x28;
dec->private_impl.f_yuv_cache[0x11][0x0C] = 0x45;
dec->private_impl.f_yuv_cache[0x11][0x0D] = 0x90;
dec->private_impl.f_yuv_cache[0x11][0x0E] = 0x45;
dec->private_impl.f_yuv_cache[0x11][0x0F] = 0x23;
dec->private_impl.f_yuv_cache[0x12][0x07] = 0x53;
dec->private_impl.f_yuv_cache[0x13][0x07] = 0x60;
dec->private_impl.f_yuv_cache[0x14][0x07] = 0x28;
dec->private_impl.f_yuv_cache[0x15][0x07] = 0x74;
dec->private_impl.f_yuv_cache[0x16][0x07] = 0x71;
dec->private_impl.f_yuv_cache[0x17][0x07] = 0x35;
dec->private_impl.f_yuv_cache[0x18][0x07] = 0x26;
dec->private_impl.f_yuv_cache[0x19][0x07] = 0x62;
}
const char* //
test_wuffs_vp8_decode_predict_uv8() {
CHECK_FOCUS(__func__);
const uint64_t test_cases[13][8] = {
{0x4444444444444444, 0x4444444444444444, 0x4444444444444444,
0x4444444444444444, 0x4444444444444444, 0x4444444444444444,
0x4444444444444444, 0x4444444444444444},
{0x4454445471BC714F, 0x516151617EC97E5C, 0x1929192946914624,
0x6575657592DD9270, 0x627262728FDA8F6D, 0x26362636539E5331,
0x17271727448F4422, 0x5363536380CB805E},
{0x1828182845904523, 0x1828182845904523, 0x1828182845904523,
0x1828182845904523, 0x1828182845904523, 0x1828182845904523,
0x1828182845904523, 0x1828182845904523},
{0x5353535353535353, 0x6060606060606060, 0x2828282828282828,
0x7474747474747474, 0x7171717171717171, 0x3535353535353535,
0x2626262626262626, 0x6262626262626262},
{0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0},
{0x5050505050505050, 0x5050505050505050, 0x5050505050505050,
0x5050505050505050, 0x5050505050505050, 0x5050505050505050,
0x5050505050505050, 0x5050505050505050},
{0x3838383838383838, 0x3838383838383838, 0x3838383838383838,
0x3838383838383838, 0x3838383838383838, 0x3838383838383838,
0x3838383838383838, 0x3838383838383838},
{0x8080808080808080, 0x8080808080808080, 0x8080808080808080,
0x8080808080808080, 0x8080808080808080, 0x8080808080808080,
0x8080808080808080, 0x8080808080808080},
};
wuffs_vp8__decoder dec = {0};
for (int mode = 0; mode < 13; mode++) {
uint64_t w0 = test_cases[mode][0];
uint64_t w1 = test_cases[mode][1];
uint64_t w2 = test_cases[mode][2];
uint64_t w3 = test_cases[mode][3];
uint64_t w4 = test_cases[mode][4];
uint64_t w5 = test_cases[mode][5];
uint64_t w6 = test_cases[mode][6];
uint64_t w7 = test_cases[mode][7];
if (w0 == 0) {
continue;
}
initialize_decoder_test_state(&dec);
wuffs_vp8__decoder__predict_uv8(&dec, 0, mode);
uint64_t h0 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x12][8]);
uint64_t h1 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x13][8]);
uint64_t h2 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x14][8]);
uint64_t h3 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x15][8]);
uint64_t h4 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x16][8]);
uint64_t h5 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x17][8]);
uint64_t h6 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x18][8]);
uint64_t h7 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x19][8]);
if ((h0 != w0) || (h1 != w1) || (h2 != w2) || (h3 != w3) || //
(h4 != w4) || (h5 != w5) || (h6 != w6) || (h7 != w7)) {
RETURN_FAIL(
"mode=%d:" //
"\nhave %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
"\nwant %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64, //
mode, h0, h1, h2, h3, h4, h5, h6, h7, w0, w1, w2, w3, w4, w5, w6, w7);
}
}
return NULL;
}
const char* //
test_wuffs_vp8_decode_predict_y16() {
CHECK_FOCUS(__func__);
const uint64_t test_cases[13][9] = {
{0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A,
0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A,
0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A, 0x5A5A5A5A5A5A5A5A},
{0x98B07DAAAFEEEA7A, 0x5169366368A7A333, 0xA7BF8CB9BEFDF989,
0x263E0B383D7C7808, 0xA3BB88B5BAF9F585, 0xA9C18EBBC0FFFB8B,
0x475F2C595E9D9929, 0x6179467378B7B343, 0x152D00272C6B6700},
{0x4159265358979323, 0x4159265358979323, 0x4159265358979323,
0x4159265358979323, 0x4159265358979323, 0x4159265358979323,
0x4159265358979323, 0x4159265358979323, 0x4159265358979323},
{0x8888888888888888, 0x4141414141414141, 0x9797979797979797,
0x1616161616161616, 0x9393939393939393, 0x9999999999999999,
0x3737373737373737, 0x5151515151515151, 0x0505050505050505},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0, 0, 0},
{0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D,
0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D,
0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D, 0x5D5D5D5D5D5D5D5D},
{0x5757575757575757, 0x5757575757575757, 0x5757575757575757,
0x5757575757575757, 0x5757575757575757, 0x5757575757575757,
0x5757575757575757, 0x5757575757575757, 0x5757575757575757},
{0x8080808080808080, 0x8080808080808080, 0x8080808080808080,
0x8080808080808080, 0x8080808080808080, 0x8080808080808080,
0x8080808080808080, 0x8080808080808080, 0x8080808080808080},
};
wuffs_vp8__decoder dec = {0};
for (int mode = 0; mode < 13; mode++) {
uint64_t w0 = test_cases[mode][0];
uint64_t w1 = test_cases[mode][1];
uint64_t w2 = test_cases[mode][2];
uint64_t w3 = test_cases[mode][3];
uint64_t w4 = test_cases[mode][4];
uint64_t w5 = test_cases[mode][5];
uint64_t w6 = test_cases[mode][6];
uint64_t w7 = test_cases[mode][7];
uint64_t w8 = test_cases[mode][8];
if (w0 == 0) {
continue;
}
initialize_decoder_test_state(&dec);
wuffs_vp8__decoder__predict_y16(&dec, mode);
uint64_t h0 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x01][8]);
uint64_t h1 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x02][8]);
uint64_t h2 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x03][8]);
uint64_t h3 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x04][8]);
uint64_t h4 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x05][8]);
uint64_t h5 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x06][8]);
uint64_t h6 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x07][8]);
uint64_t h7 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x08][8]);
uint64_t h8 = wuffs_base__peek_u64be__no_bounds_check(
&dec.private_impl.f_yuv_cache[0x09][8]);
if ((h0 != w0) || (h1 != w1) || (h2 != w2) || (h3 != w3) || //
(h4 != w4) || (h5 != w5) || (h6 != w6) || (h7 != w7) || (h8 != w8)) {
RETURN_FAIL(
"mode=%d:" //
"\nhave %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64 //
"\nwant %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64 " %016" PRIX64 " %016" PRIX64 " %016" PRIX64 //
" %016" PRIX64, //
mode, h0, h1, h2, h3, h4, h5, h6, h7, h8, //
w0, w1, w2, w3, w4, w5, w6, w7, w8);
}
}
return NULL;
}
const char* //
test_wuffs_vp8_decode_predict_y4() {
CHECK_FOCUS(__func__);
const uint32_t test_cases[10][4] = {
{0x51515151, 0x51515151, 0x51515151, 0x51515151},
{0x98B07DAA, 0x51693663, 0xA7BF8CB9, 0x263E0B38},
{0x43463E49, 0x43463E49, 0x43463E49, 0x43463E49},
{0x61616161, 0x68686868, 0x61616161, 0x36363636},
{0x4B43463E, 0x614B4346, 0x68614B43, 0x6168614B},
{0x394D403D, 0x4B43463E, 0x61394D40, 0x684B4346},
{0x463E4967, 0x3E496786, 0x49678678, 0x6786783F},
{0x4D403D56, 0x463E4967, 0x403D5686, 0x3E496778},
{0x5D4B4346, 0x65615D4B, 0x6C686561, 0x57616C68},
{0x65686C61, 0x6C615736, 0x57361616, 0x16161616},
};
wuffs_vp8__decoder dec = {0};
for (int mode = 0; mode < 10; mode++) {
uint32_t w0 = test_cases[mode][0];
uint32_t w1 = test_cases[mode][1];
uint32_t w2 = test_cases[mode][2];
uint32_t w3 = test_cases[mode][3];
initialize_decoder_test_state(&dec);
dec.private_impl.f_mb_subblock_modes[0] = mode;
wuffs_vp8__decoder__predict_y4(&dec, 0);
uint32_t h0 = wuffs_base__peek_u32be__no_bounds_check(
&dec.private_impl.f_yuv_cache[1][8]);
uint32_t h1 = wuffs_base__peek_u32be__no_bounds_check(
&dec.private_impl.f_yuv_cache[2][8]);
uint32_t h2 = wuffs_base__peek_u32be__no_bounds_check(
&dec.private_impl.f_yuv_cache[3][8]);
uint32_t h3 = wuffs_base__peek_u32be__no_bounds_check(
&dec.private_impl.f_yuv_cache[4][8]);
if ((h0 != w0) || (h1 != w1) || (h2 != w2) || (h3 != w3)) {
RETURN_FAIL(
"mode=%d:" //
"\nhave %08" PRIX32 " %08" PRIX32 " %08" PRIX32 " %08" PRIX32 //
"\nwant %08" PRIX32 " %08" PRIX32 " %08" PRIX32 " %08" PRIX32, //
mode, h0, h1, h2, h3, w0, w1, w2, w3);
}
}
return NULL;
}
// ---------------- Mimic Tests
@@ -91,7 +380,9 @@ the first "./a.out" with "./a.out -bench". Combine these changes with the
proc g_tests[] = {
// No VP8 tests.
test_wuffs_vp8_decode_predict_uv8,
test_wuffs_vp8_decode_predict_y16,
test_wuffs_vp8_decode_predict_y4,
#ifdef WUFFS_MIMIC