Swizzler support src_pixfmt = RGBA_PREMUL

Binary size, before:
 55392 gen/lib/c/clang-9-dynamic/wuffs-base-pixconv.lo
 57880 gen/lib/c/clang-9-static/wuffs-base-pixconv.o
 96816 gen/lib/c/gcc-dynamic/wuffs-base-pixconv.lo
 96736 gen/lib/c/gcc-static/wuffs-base-pixconv.o

After:
 59472 gen/lib/c/clang-9-dynamic/wuffs-base-pixconv.lo
 62072 gen/lib/c/clang-9-static/wuffs-base-pixconv.o
102936 gen/lib/c/gcc-dynamic/wuffs-base-pixconv.lo
102904 gen/lib/c/gcc-static/wuffs-base-pixconv.o
This commit is contained in:
Nigel Tao committed 2021-03-21 21:49:53 +11:00
1 parent 17ae882a96
commit 15b04e65f3
4 files changed
+639 -13

No files matched your search

+306
View File
@@ -1393,6 +1393,93 @@ wuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len2 = dst_len / 2;
size_t src_len4 = src_len / 4;
size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
// TODO: unroll.
while (n >= 1) {
wuffs_base__poke_u16le__no_bounds_check(
d + (0 * 2),
wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(
wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));
s += 1 * 4;
d += 1 * 2;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len2 = dst_len / 2;
size_t src_len4 = src_len / 4;
size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
// TODO: unroll.
while (n >= 1) {
// Extract 16-bit color components.
uint32_t sa = 0x101 * ((uint32_t)s[3]);
uint32_t sb = 0x101 * ((uint32_t)s[2]);
uint32_t sg = 0x101 * ((uint32_t)s[1]);
uint32_t sr = 0x101 * ((uint32_t)s[0]);
// Convert from 565 color to 16-bit color.
uint32_t old_rgb_565 = wuffs_base__peek_u16le__no_bounds_check(d + (0 * 2));
uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11);
uint32_t dr = (0x8421 * old_r5) >> 4;
uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5);
uint32_t dg = (0x1041 * old_g6) >> 2;
uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0);
uint32_t db = (0x8421 * old_b5) >> 4;
// Calculate the inverse of the src-alpha: how much of the dst to keep.
uint32_t ia = 0xFFFF - sa;
// Composite src (premul) over dst (premul).
dr = sr + ((dr * ia) / 0xFFFF);
dg = sg + ((dg * ia) / 0xFFFF);
db = sb + ((db * ia) / 0xFFFF);
// Convert from 16-bit color to 565 color and combine the components.
uint32_t new_r5 = 0x1F & (dr >> 11);
uint32_t new_g6 = 0x3F & (dg >> 10);
uint32_t new_b5 = 0x1F & (db >> 11);
uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0);
wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)new_rgb_565);
s += 1 * 4;
d += 1 * 2;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,
size_t dst_len,
@@ -1886,6 +1973,82 @@ wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr__rgba_premul__src(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len3 = dst_len / 3;
size_t src_len4 = src_len / 4;
size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint8_t s0 = s[0];
uint8_t s1 = s[1];
uint8_t s2 = s[2];
d[0] = s2;
d[1] = s1;
d[2] = s0;
d[3] = 0xFF;
s += 1 * 4;
d += 1 * 3;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr__rgba_premul__src_over(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len3 = dst_len / 3;
size_t src_len4 = src_len / 4;
size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
// Extract 16-bit color components.
uint32_t sa = 0x101 * ((uint32_t)s[3]);
uint32_t sb = 0x101 * ((uint32_t)s[2]);
uint32_t sg = 0x101 * ((uint32_t)s[1]);
uint32_t sr = 0x101 * ((uint32_t)s[0]);
uint32_t dr = 0x101 * ((uint32_t)d[2]);
uint32_t dg = 0x101 * ((uint32_t)d[1]);
uint32_t db = 0x101 * ((uint32_t)d[0]);
// Calculate the inverse of the src-alpha: how much of the dst to keep.
uint32_t ia = 0xFFFF - sa;
// Composite src (premul) over dst (premul).
dr = sr + ((dr * ia) / 0xFFFF);
dg = sg + ((dg * ia) / 0xFFFF);
db = sb + ((db * ia) / 0xFFFF);
// Convert from 16-bit color to 8-bit color.
d[0] = (uint8_t)(db >> 8);
d[1] = (uint8_t)(dg >> 8);
d[2] = (uint8_t)(dr >> 8);
s += 1 * 4;
d += 1 * 3;
n -= 1;
}
return len;
}
// --------
static uint64_t //
@@ -2416,6 +2579,65 @@ wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len8 = dst_len / 8;
size_t src_len4 = src_len / 4;
size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint64_t s0 = wuffs_base__color_u32__as__color_u64(
wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(
wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));
wuffs_base__poke_u64le__no_bounds_check(d + (0 * 8), s0);
s += 1 * 4;
d += 1 * 8;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len8 = dst_len / 8;
size_t src_len4 = src_len / 4;
size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));
uint64_t s0 =
wuffs_base__color_u32__as__color_u64(wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));
wuffs_base__poke_u64le__no_bounds_check(
d + (0 * 8), wuffs_base__composite_nonpremul_premul_u64_axxx(d0, s0));
s += 1 * 4;
d += 1 * 8;
n -= 1;
}
return len;
}
// --------
static uint64_t //
@@ -4554,6 +4776,85 @@ wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(
return NULL;
}
static wuffs_base__pixel_swizzler__func //
wuffs_base__pixel_swizzler__prepare__rgba_premul(
wuffs_base__pixel_swizzler* p,
wuffs_base__pixel_format dst_pixfmt,
wuffs_base__slice_u8 dst_palette,
wuffs_base__slice_u8 src_palette,
wuffs_base__pixel_blend blend) {
switch (dst_pixfmt.repr) {
case WUFFS_BASE__PIXEL_FORMAT__BGR_565:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGR:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgr__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgr__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
#if defined(WUFFS_BASE__CPU_ARCH__X86_64)
if (wuffs_base__cpu_arch__have_x86_sse42()) {
return wuffs_base__pixel_swizzler__swap_rgbx_bgrx__sse42;
}
#endif
return wuffs_base__pixel_swizzler__swap_rgbx_bgrx;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_premul__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__copy_4_4;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_premul__bgra_premul__src_over;
}
return NULL;
}
return NULL;
}
// --------
WUFFS_BASE__MAYBE_STATIC wuffs_base__status //
@@ -4660,6 +4961,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p,
func = wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(
p, dst_pixfmt, dst_palette, src_palette, blend);
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
func = wuffs_base__pixel_swizzler__prepare__rgba_premul(
p, dst_pixfmt, dst_palette, src_palette, blend);
break;
}
p->private_impl.func = func;
+23 -13
View File
@@ -605,12 +605,15 @@ const BasePixConvSubmoduleC = "" +
"ize_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len3 = src_len / 3;\n size_t len = (dst_len2 < src_len3) ? dst_len2 : src_len3;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t r5 = s[0] >> 3;\n uint32_t g6 = s[1] >> 2;\n uint32_t b5 = s[2] >> 3;\n uint32_t rgb_565 = (r5 << 11) | (g6 << 5) | (b5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 3;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_l" +
"en4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__swap_u32_argb_abgr(\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))))));\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color compon" +
"ents.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sb = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sr = 0x101 * ((uint32_t)s[0]);\n\n // Convert from 565 color to 16-bit color.\n uint32_t old_rgb_565 = wuffs_base__peek_u16le__no_bounds_check(d + (0 * 2));\n uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11);\n uint32_t dr = (0x8421 * old_r5) >> 4;\n uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5);\n uint32_t dg = (0x1041 * old_g6) >> 2;\n uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0);\n uint32_t db = (0x8421 * old_b5) >> 4;\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert from 16-bit color to 565 color and combine the components.\n uint32_t new_r5 = 0x1F & (dr >> 11);\n uint32_t new_g6 = 0x3F & (dg " +
">> 10);\n uint32_t new_b5 = 0x1F & (db >> 11);\n uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)new_rgb_565);\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 *" +
" 2), (uint16_t)rgb_565);\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y_16be(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len2 = src_len / 2;\n size_t len = (dst_len2 < src_len2) ? dst_len2 : src_len2;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 2;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuf" +
"fs_base__pixel_swizzler__bgr_565__index__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[0] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (1 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((" +
"size_t)s[1] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (2 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[2] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (3 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 2;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[0] * 4)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index_bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_l" +
"en2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(\n wuffs_base__peek_u16le__no_bounds_check(d + (0 * 2)));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n " +
" return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)s0);\n }\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\n" +
">> 10);\n uint32_t new_b5 = 0x1F & (db >> 11);\n uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)new_rgb_565);\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgba_premul__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n wuffs_base__poke_u16le__no_bounds_check(" +
"\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sb = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sr = 0x101 * ((uint32_t)s[0]);\n\n // Convert from 565 color to 16-bit color.\n uint32_t old_rgb" +
"_565 = wuffs_base__peek_u16le__no_bounds_check(d + (0 * 2));\n uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11);\n uint32_t dr = (0x8421 * old_r5) >> 4;\n uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5);\n uint32_t dg = (0x1041 * old_g6) >> 2;\n uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0);\n uint32_t db = (0x8421 * old_b5) >> 4;\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (premul) over dst (premul).\n dr = sr + ((dr * ia) / 0xFFFF);\n dg = sg + ((dg * ia) / 0xFFFF);\n db = sb + ((db * ia) / 0xFFFF);\n\n // Convert from 16-bit color to 565 color and combine the components.\n uint32_t new_r5 = 0x1F & (dr >> 11);\n uint32_t new_g6 = 0x3F & (dg >> 10);\n uint32_t new_b5 = 0x1F & (db >> 11);\n uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)new_rgb_565);\n\n s += 1 * 4;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nst" +
"atic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__y_16be(uint8_t* dst_ptr,\n size_t dst_len,\n " +
" uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len2 = dst_len / 2;\n size_t src_len2 = src_len / 2;\n size_t len = (dst_len2 < src_len2) ? dst_len2 : src_len2;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t y5 = s[0] >> 3;\n uint32_t y6 = s[0] >> 2;\n uint32_t rgb_565 = (y5 << 11) | (y6 << 5) | (y5 << 0);\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)rgb_565);\n\n s += 1 * 2;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n " +
" const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n const size_t loop_unroll_count = 4;\n\n while (n >= loop_unroll_count) {\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[0] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (1 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[1] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (2 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[2] * 4)));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (3 * 2), wuffs_base__peek" +
"_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[3] * 4)));\n\n s += loop_unroll_count * 1;\n d += loop_unroll_count * 2;\n n -= loop_unroll_count;\n }\n\n while (n >= 1) {\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2), wuffs_base__peek_u16le__no_bounds_check(\n dst_palette_ptr + ((size_t)s[0] * 4)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index_bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__color_u16_rgb_565__as__color_u32_argb_premul(\n w" +
"uffs_base__peek_u16le__no_bounds_check(d + (0 * 2)));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4));\n wuffs_base__poke_u16le__no_bounds_check(\n d + (0 * 2),\n wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\n wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0)));\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__index_binary_alpha__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len2 = dst_len / 2;\n size_t len = (dst_len2 < src_len) ? dst_len2 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_ch" +
"eck(dst_palette_ptr +\n ((size_t)s[0] * 4));\n if (s0) {\n wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)s0);\n }\n\n s += 1 * 1;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\n" +
"" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 =\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwu" +
"ffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len3 < src_len8) ? dst_len3 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 =\n wuffs_base__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__peek_u64le__no_bounds_check(s + (0 * 8)));\n wuffs_base__poke_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 8;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len " +
@@ -621,7 +624,9 @@ const BasePixConvSubmoduleC = "" +
" 0x101 * ((uint32_t)s[3]);\n uint32_t sr = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sb = 0x101 * ((uint32_t)s[0]);\n uint32_t dr = 0x101 * ((uint32_t)d[2]);\n uint32_t dg = 0x101 * ((uint32_t)d[1]);\n uint32_t db = 0x101 * ((uint32_t)d[0]);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (premul) over dst (premul).\n dr = sr + ((dr * ia) / 0xFFFF);\n dg = sg + ((dg * ia) / 0xFFFF);\n db = sb + ((db * ia) / 0xFFFF);\n\n // Convert from 16-bit color to 8-bit color.\n d[0] = (uint8_t)(db >> 8);\n d[1] = (uint8_t)(dg >> 8);\n d[2] = (uint8_t)(dr >> 8);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n " +
" size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__swap_u32_argb_abgr(\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));\n wuffs_base__poke_u24le__no_bounds_check(d + (0 * 3), s0);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / " +
"3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sb = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sr = 0x101 * ((uint32_t)s[0]);\n uint32_t dr = 0x101 * ((uint32_t)d[2]);\n uint32_t dg = 0x101 * ((uint32_t)d[1]);\n uint32_t db = 0x101 * ((uint32_t)d[0]);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert from 16-bit color to 8-bit color.\n d[0] = (uint8_t)(db >> 8);\n d[1] = (uint8_t)(dg >> 8);\n d[2] = (uint8_t)(dr >> 8);\n\n s += 1 * 4;\n d += 1 * 3;\n " +
" n -= 1;\n }\n\n return len;\n}\n\n" +
" n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__rgba_premul__src(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n while (n >= 1) {\n uint8_t s0 = s[0];\n uint8_t s1 = s[1];\n uint8_t s2 = s[2];\n d[0] = s2;\n d[1] = s1;\n d[2] = s0;\n d[3] = 0xFF;\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__rgba_premul__src_over(uint8_t* dst_ptr,\n " +
" size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len3 = dst_len / 3;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n while (n >= 1) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * ((uint32_t)s[3]);\n uint32_t sb = 0x101 * ((uint32_t)s[2]);\n uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sr = 0x101 * ((uint32_t)s[0]);\n uint32_t dr = 0x101 * ((uint32_t)d[2]);\n uint32_t dg = 0x101 * ((uint32_t)d[1]);\n uint32_t db = 0x101 * ((uint32_t)d[0]);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint32_t ia = 0xFFFF - sa;\n\n " +
" // Composite src (premul) over dst (premul).\n dr = sr + ((dr * ia) / 0xFFFF);\n dg = sg + ((dg * ia) / 0xFFFF);\n db = sb + ((db * ia) / 0xFFFF);\n\n // Convert from 16-bit color to 8-bit color.\n d[0] = (uint8_t)(db >> 8);\n d[1] = (uint8_t)(dg >> 8);\n d[2] = (uint8_t)(dr >> 8);\n\n s += 1 * 4;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\n" +
"" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4),\n wuffs_base__composite_nonpremul_nonpremul_u32_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8" +
"_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len4 < src_len8) ? dst_len4 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__color_u64__as__color_u32(\n wuffs_base__peek_u64le__no_bounds_check(s + (0 * 8))));\n\n s += 1 * 8;\n d += 1 * 4;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len4 < src_len8) ? dst_len4 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__color_u32__as__color_u64(" +
@@ -638,7 +643,9 @@ const BasePixConvSubmoduleC = "" +
"4))));\n wuffs_base__poke_u64le__no_bounds_check(d + (0 * 8), s0);\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len8 = dst_len / 8;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));\n uint64_t s0 = wuffs_base__color_u32__as__color_u64(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke_u64le__no_bounds_check(\n d + (0 * 8), wuffs_base__composite_nonpremul_premul_u64_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler" +
"__bgra_nonpremul_4x16le__index_bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n if (dst_palette_len != 1024) {\n return 0;\n }\n size_t dst_len8 = dst_len / 8;\n size_t len = (dst_len8 < src_len) ? dst_len8 : src_len;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));\n uint64_t s0 = wuffs_base__color_u32__as__color_u64(\n wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4)));\n wuffs_base__poke_u64le__no_bounds_check(\n d + (0 * 8),\n wuffs_base__composite_nonpremul_nonpremul_u64_axxx(d0, s0));\n\n s += 1 * 1;\n d += 1 * 8;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src(\n uint8_t* dst_ptr,\n " +
" size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len8 = dst_len / 8;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint8_t s0 = s[0];\n uint8_t s1 = s[1];\n uint8_t s2 = s[2];\n uint8_t s3 = s[3];\n d[0] = s2;\n d[1] = s2;\n d[2] = s1;\n d[3] = s1;\n d[4] = s0;\n d[5] = s0;\n d[6] = s3;\n d[7] = s3;\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len8 = dst_len / 8;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;\n uint8_t* d = dst_ptr;\n cons" +
"t uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));\n uint64_t s0 =\n wuffs_base__color_u32__as__color_u64(wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));\n wuffs_base__poke_u64le__no_bounds_check(\n d + (0 * 8),\n wuffs_base__composite_nonpremul_nonpremul_u64_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\n" +
"t uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));\n uint64_t s0 =\n wuffs_base__color_u32__as__color_u64(wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));\n wuffs_base__poke_u64le__no_bounds_check(\n d + (0 * 8),\n wuffs_base__composite_nonpremul_nonpremul_u64_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len8 = dst_len / 8;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint64_t s0 = wuffs_base__color_u32__as__color_u64(\n wuffs" +
"_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(\n wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));\n wuffs_base__poke_u64le__no_bounds_check(d + (0 * 8), s0);\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len8 = dst_len / 8;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n >= 1) {\n uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));\n uint64_t s0 =\n wuffs_base__color_u32__as__color_u64(wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));\n wuffs_base__poke_u" +
"64le__no_bounds_check(\n d + (0 * 8), wuffs_base__composite_nonpremul_premul_u64_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 8;\n n -= 1;\n }\n return len;\n}\n\n" +
"" +
"// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4),\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 =" +
" dst_len / 4;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len4 < src_len8) ? dst_len4 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint64_t s0 = wuffs_base__peek_u64le__no_bounds_check(s + (0 * 8));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4),\n wuffs_base__color_u64_argb_nonpremul__as__color_u32_argb_premul(s0));\n\n s += 1 * 8;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n const uint8_t* src_ptr,\n size_t src_len) {\n size_t dst_len4 = dst_len / 4;\n size_t src_len4 = src_len / 4;\n size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n size_t n = len;\n\n // TODO: unroll.\n\n while (n >= 1) {\n uint32_t d0 = wuffs_base__peek_u32le__no_bo" +
@@ -718,15 +725,18 @@ const BasePixConvSubmoduleC = "" +
"BGRA_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n#if defined(WUFFS_BASE__CPU_ARCH__X86_64)\n if (wuffs_base__cpu_arch__have_x86_sse42()) {\n return wuffs_base__pixel_swizzler__bgrw__rgb__sse42;\n }\n#endif\n return wuffs_base__pixel_swizzler__bgrw__rgb;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n return wuffs_base__pixel_swizzler__bgrw_4x16le__rgb;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n return wuffs_base__pixel_swizzler__copy_3_3;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n return wuffs_base__pixel_swizzler__bgrw__bgr;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__rgba_nonpremul(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blen" +
"d blend) {\n switch (dst_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n#if defined(WUFFS_BASE__CPU_ARCH__X86_64)\n if (wuffs_base__cpu_arch__have_x86_sse42()) {\n return wuffs_base__pixel_swizzler__swap_rgbx_bgrx__sse42;\n }\n#endif\n " +
"return wuffs_base__pixel_swizzler__swap_rgbx_bgrx;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n case WUFFS_BASE__PIXEL" +
"_FORMAT__BGRX:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__copy_4_4;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n // TODO.\n break;\n }\n return NULL;\n}\n\n" +
"_FORMAT__BGRX:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__copy_4_4;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_BINARY:\n case WUFFS_BASE__PIXEL_FORMAT__RGBX:\n // TODO.\n break;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__rgb" +
"a_premul(\n wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n switch (dst_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__BGR_565:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgr__rgba_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__rgba_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuff" +
"s_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n#if defined(WUFFS_BASE__CPU_ARCH__X86_64)\n if (wuffs_base__cpu_arch__have_x86_sse42()) {\n return wuffs_base__pixel_swizzler__swap_rgbx_bgrx__sse42;\n }\n#endif\n return wuffs_base__pixel_swizzler__swap_rgbx_bgrx;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wu" +
"ffs_base__pixel_swizzler__bgra_premul__rgba_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src_over;\n }\n return NULL;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__copy_4_4;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_premul__src_over;\n }\n return NULL;\n }\n return NULL;\n}\n\n" +
"" +
"// --------\n\nWUFFS_BASE__MAYBE_STATIC wuffs_base__status //\nwuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p,\n wuffs_base__pixel_format dst_pixfmt,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__pixel_format src_pixfmt,\n wuffs_base__slice_u8 src_palette,\n wuffs_base__pixel_blend blend) {\n if (!p) {\n return wuffs_base__make_status(wuffs_base__error__bad_receiver);\n }\n p->private_impl.func = NULL;\n p->private_impl.transparent_black_func = NULL;\n p->private_impl.dst_pixfmt_bytes_per_pixel = 0;\n p->private_impl.src_pixfmt_bytes_per_pixel = 0;\n\n wuffs_base__pixel_swizzler__func func = NULL;\n wuffs_base__pixel_swizzler__transparent_black_func transparent_black_func =\n NULL;\n\n uint32_t dst_pixfmt_bits_per_pixel =\n wuffs_base__pixel_format__bits_per_pixel(&dst_pixfmt);\n if ((dst_pixfmt_bits_per_pixel == " +
"0) ||\n ((dst_pixfmt_bits_per_pixel & 7) != 0)) {\n return wuffs_base__make_status(\n wuffs_base__error__unsupported_pixel_swizzler_option);\n }\n\n uint32_t src_pixfmt_bits_per_pixel =\n wuffs_base__pixel_format__bits_per_pixel(&src_pixfmt);\n if ((src_pixfmt_bits_per_pixel == 0) ||\n ((src_pixfmt_bits_per_pixel & 7) != 0)) {\n return wuffs_base__make_status(\n wuffs_base__error__unsupported_pixel_swizzler_option);\n }\n\n // TODO: support many more formats.\n\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n transparent_black_func =\n wuffs_base__pixel_swizzler__transparent_black_src;\n break;\n\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n transparent_black_func =\n wuffs_base__pixel_swizzler__transparent_black_src_over;\n break;\n }\n\n switch (src_pixfmt.repr) {\n case WUFFS_BASE__PIXEL_FORMAT__Y:\n func = wuffs_base__pixel_swizzler__prepare__y(p, dst_pixfmt, dst_palette,\n src_palette" +
", blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__Y_16BE:\n func = wuffs_base__pixel_swizzler__prepare__y_16be(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__indexed__bgra_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__INDEXED__BGRA_BINARY:\n func = wuffs_base__pixel_swizzler__prepare__indexed__bgra_binary(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n func = wuffs_base__pixel_swizzler__prepare__bgr(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n" +
" func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n func = wuffs_base__pixel_swizzler__prepare__bgra_premul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n func = wuffs_base__pixel_swizzler__prepare__bgrx(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n func = wuffs_base__pixel_swizzler__prepare__rgb(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n }\n\n p->private_impl.func = func;\n p->private_impl.transparent_black_func = transparent_black_func;\n p->private_impl.dst_pixfmt_bytes_per_pixel = dst_pixfmt_bits_p" +
"er_pixel / 8;\n p->private_impl.src_pixfmt_bytes_per_pixel = src_pixfmt_bits_per_pixel / 8;\n return wuffs_base__make_status(\n func ? NULL : wuffs_base__error__unsupported_pixel_swizzler_option);\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__limited_swizzle_u32_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n uint32_t up_to_num_pixels,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t src_len = wuffs_base__u64__min(\n ((uint64_t)up_to_num_pixels) *\n ((uint64_t)p->private_impl.src_pixfmt_bytes_per_pixel),\n ((uint64_t)(io2_r - iop_r)));\n uint64_t n =\n (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, iop_r, (size_t)src_len);\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n re" +
"turn 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t src_len = ((uint64_t)(io2_r - iop_r));\n uint64_t n =\n (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, iop_r, (size_t)src_len);\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_slice(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (p && p->private_impl.func) {\n return (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n " +
" dst_palette.len, src.ptr, src.len);\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_transparent_black(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n uint64_t num_pixels) {\n if (p && p->private_impl.transparent_black_func) {\n return (*p->private_impl.transparent_black_func)(\n dst.ptr, dst.len, dst_palette.ptr, dst_palette.len, num_pixels,\n p->private_impl.dst_pixfmt_bytes_per_pixel);\n }\n return 0;\n}\n" +
" func = wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n func = wuffs_base__pixel_swizzler__prepare__bgra_premul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n func = wuffs_base__pixel_swizzler__prepare__bgrx(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n func = wuffs_base__pixel_swizzler__prepare__rgb(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:\n func = wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(\n p, dst_pixfmt, dst_palette, src_palette, blend);\n break;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:\n func = wuffs_base__pixel_swizzler__prepare__rgba_premul(\n p, dst_pixfmt, dst_palette, src_palette, blen" +
"d);\n break;\n }\n\n p->private_impl.func = func;\n p->private_impl.transparent_black_func = transparent_black_func;\n p->private_impl.dst_pixfmt_bytes_per_pixel = dst_pixfmt_bits_per_pixel / 8;\n p->private_impl.src_pixfmt_bytes_per_pixel = src_pixfmt_bits_per_pixel / 8;\n return wuffs_base__make_status(\n func ? NULL : wuffs_base__error__unsupported_pixel_swizzler_option);\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__limited_swizzle_u32_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n uint32_t up_to_num_pixels,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t src_len = wuffs_base__u64__min(\n ((uint64_t)up_to_num_pixels) *\n ((uint64_t)p->private_impl.src_pixfmt_bytes_per_pixel),\n ((uint64_t)(io2_r - iop_r)));\n uint64_t n =\n (*p->private_impl.func)(dst.ptr, dst." +
"len, dst_palette.ptr,\n dst_palette.len, iop_r, (size_t)src_len);\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_reader(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n const uint8_t** ptr_iop_r,\n const uint8_t* io2_r) {\n if (p && p->private_impl.func) {\n const uint8_t* iop_r = *ptr_iop_r;\n uint64_t src_len = ((uint64_t)(io2_r - iop_r));\n uint64_t n =\n (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, iop_r, (size_t)src_len);\n *ptr_iop_r += n * p->private_impl.src_pixfmt_bytes_per_pixel;\n return n;\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_from_slice(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_" +
"base__slice_u8 dst_palette,\n wuffs_base__slice_u8 src) {\n if (p && p->private_impl.func) {\n return (*p->private_impl.func)(dst.ptr, dst.len, dst_palette.ptr,\n dst_palette.len, src.ptr, src.len);\n }\n return 0;\n}\n\nWUFFS_BASE__MAYBE_STATIC uint64_t //\nwuffs_base__pixel_swizzler__swizzle_interleaved_transparent_black(\n const wuffs_base__pixel_swizzler* p,\n wuffs_base__slice_u8 dst,\n wuffs_base__slice_u8 dst_palette,\n uint64_t num_pixels) {\n if (p && p->private_impl.transparent_black_func) {\n return (*p->private_impl.transparent_black_func)(\n dst.ptr, dst.len, dst_palette.ptr, dst_palette.len, num_pixels,\n p->private_impl.dst_pixfmt_bytes_per_pixel);\n }\n return 0;\n}\n" +
""
const BaseUTF8SubmoduleC = "" +
+306
View File
@@ -16137,6 +16137,93 @@ wuffs_base__pixel_swizzler__bgr_565__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len2 = dst_len / 2;
size_t src_len4 = src_len / 4;
size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
// TODO: unroll.
while (n >= 1) {
wuffs_base__poke_u16le__no_bounds_check(
d + (0 * 2),
wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(
wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));
s += 1 * 4;
d += 1 * 2;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len2 = dst_len / 2;
size_t src_len4 = src_len / 4;
size_t len = (dst_len2 < src_len4) ? dst_len2 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
// TODO: unroll.
while (n >= 1) {
// Extract 16-bit color components.
uint32_t sa = 0x101 * ((uint32_t)s[3]);
uint32_t sb = 0x101 * ((uint32_t)s[2]);
uint32_t sg = 0x101 * ((uint32_t)s[1]);
uint32_t sr = 0x101 * ((uint32_t)s[0]);
// Convert from 565 color to 16-bit color.
uint32_t old_rgb_565 = wuffs_base__peek_u16le__no_bounds_check(d + (0 * 2));
uint32_t old_r5 = 0x1F & (old_rgb_565 >> 11);
uint32_t dr = (0x8421 * old_r5) >> 4;
uint32_t old_g6 = 0x3F & (old_rgb_565 >> 5);
uint32_t dg = (0x1041 * old_g6) >> 2;
uint32_t old_b5 = 0x1F & (old_rgb_565 >> 0);
uint32_t db = (0x8421 * old_b5) >> 4;
// Calculate the inverse of the src-alpha: how much of the dst to keep.
uint32_t ia = 0xFFFF - sa;
// Composite src (premul) over dst (premul).
dr = sr + ((dr * ia) / 0xFFFF);
dg = sg + ((dg * ia) / 0xFFFF);
db = sb + ((db * ia) / 0xFFFF);
// Convert from 16-bit color to 565 color and combine the components.
uint32_t new_r5 = 0x1F & (dr >> 11);
uint32_t new_g6 = 0x3F & (dg >> 10);
uint32_t new_b5 = 0x1F & (db >> 11);
uint32_t new_rgb_565 = (new_r5 << 11) | (new_g6 << 5) | (new_b5 << 0);
wuffs_base__poke_u16le__no_bounds_check(d + (0 * 2), (uint16_t)new_rgb_565);
s += 1 * 4;
d += 1 * 2;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr_565__y(uint8_t* dst_ptr,
size_t dst_len,
@@ -16630,6 +16717,82 @@ wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr__rgba_premul__src(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len3 = dst_len / 3;
size_t src_len4 = src_len / 4;
size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint8_t s0 = s[0];
uint8_t s1 = s[1];
uint8_t s2 = s[2];
d[0] = s2;
d[1] = s1;
d[2] = s0;
d[3] = 0xFF;
s += 1 * 4;
d += 1 * 3;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgr__rgba_premul__src_over(uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len3 = dst_len / 3;
size_t src_len4 = src_len / 4;
size_t len = (dst_len3 < src_len4) ? dst_len3 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
// Extract 16-bit color components.
uint32_t sa = 0x101 * ((uint32_t)s[3]);
uint32_t sb = 0x101 * ((uint32_t)s[2]);
uint32_t sg = 0x101 * ((uint32_t)s[1]);
uint32_t sr = 0x101 * ((uint32_t)s[0]);
uint32_t dr = 0x101 * ((uint32_t)d[2]);
uint32_t dg = 0x101 * ((uint32_t)d[1]);
uint32_t db = 0x101 * ((uint32_t)d[0]);
// Calculate the inverse of the src-alpha: how much of the dst to keep.
uint32_t ia = 0xFFFF - sa;
// Composite src (premul) over dst (premul).
dr = sr + ((dr * ia) / 0xFFFF);
dg = sg + ((dg * ia) / 0xFFFF);
db = sb + ((db * ia) / 0xFFFF);
// Convert from 16-bit color to 8-bit color.
d[0] = (uint8_t)(db >> 8);
d[1] = (uint8_t)(dg >> 8);
d[2] = (uint8_t)(dr >> 8);
s += 1 * 4;
d += 1 * 3;
n -= 1;
}
return len;
}
// --------
static uint64_t //
@@ -17160,6 +17323,65 @@ wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_nonpremul__src_over(
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len8 = dst_len / 8;
size_t src_len4 = src_len / 4;
size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint64_t s0 = wuffs_base__color_u32__as__color_u64(
wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(
wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)))));
wuffs_base__poke_u64le__no_bounds_check(d + (0 * 8), s0);
s += 1 * 4;
d += 1 * 8;
n -= 1;
}
return len;
}
static uint64_t //
wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over(
uint8_t* dst_ptr,
size_t dst_len,
uint8_t* dst_palette_ptr,
size_t dst_palette_len,
const uint8_t* src_ptr,
size_t src_len) {
size_t dst_len8 = dst_len / 8;
size_t src_len4 = src_len / 4;
size_t len = (dst_len8 < src_len4) ? dst_len8 : src_len4;
uint8_t* d = dst_ptr;
const uint8_t* s = src_ptr;
size_t n = len;
while (n >= 1) {
uint64_t d0 = wuffs_base__peek_u64le__no_bounds_check(d + (0 * 8));
uint64_t s0 =
wuffs_base__color_u32__as__color_u64(wuffs_base__swap_u32_argb_abgr(
wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))));
wuffs_base__poke_u64le__no_bounds_check(
d + (0 * 8), wuffs_base__composite_nonpremul_premul_u64_axxx(d0, s0));
s += 1 * 4;
d += 1 * 8;
n -= 1;
}
return len;
}
// --------
static uint64_t //
@@ -19298,6 +19520,85 @@ wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(
return NULL;
}
static wuffs_base__pixel_swizzler__func //
wuffs_base__pixel_swizzler__prepare__rgba_premul(
wuffs_base__pixel_swizzler* p,
wuffs_base__pixel_format dst_pixfmt,
wuffs_base__slice_u8 dst_palette,
wuffs_base__slice_u8 src_palette,
wuffs_base__pixel_blend blend) {
switch (dst_pixfmt.repr) {
case WUFFS_BASE__PIXEL_FORMAT__BGR_565:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgr_565__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGR:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgr__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgr__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
#if defined(WUFFS_BASE__CPU_ARCH__X86_64)
if (wuffs_base__cpu_arch__have_x86_sse42()) {
return wuffs_base__pixel_swizzler__swap_rgbx_bgrx__sse42;
}
#endif
return wuffs_base__pixel_swizzler__swap_rgbx_bgrx;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_premul__rgba_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_premul__src_over;
}
return NULL;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
switch (blend) {
case WUFFS_BASE__PIXEL_BLEND__SRC:
return wuffs_base__pixel_swizzler__copy_4_4;
case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:
return wuffs_base__pixel_swizzler__bgra_premul__bgra_premul__src_over;
}
return NULL;
}
return NULL;
}
// --------
WUFFS_BASE__MAYBE_STATIC wuffs_base__status //
@@ -19404,6 +19705,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p,
func = wuffs_base__pixel_swizzler__prepare__rgba_nonpremul(
p, dst_pixfmt, dst_palette, src_palette, blend);
break;
case WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL:
func = wuffs_base__pixel_swizzler__prepare__rgba_premul(
p, dst_pixfmt, dst_palette, src_palette, blend);
break;
}
p->private_impl.func = func;
+4
View File
@@ -270,6 +270,10 @@ test_wuffs_pixel_swizzler_swizzle() {
.color = 0x55443300,
.pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__RGBA_NONPREMUL,
},
{
.color = 0x55443300,
.pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__RGBA_PREMUL,
},
};
const struct {