diff --git a/internal/cgen/base/pixconv-submodule.c b/internal/cgen/base/pixconv-submodule.c index 04358e93..3040bb8f 100644 --- a/internal/cgen/base/pixconv-submodule.c +++ b/internal/cgen/base/pixconv-submodule.c @@ -625,6 +625,44 @@ wuffs_base__composite_nonpremul_premul_u32_axxx(uint32_t dst_nonpremul, return (db << 0) | (dg << 8) | (dr << 16) | (da << 24); } +static inline uint64_t // +wuffs_base__composite_nonpremul_premul_u64_axxx(uint64_t dst_nonpremul, + uint64_t src_premul) { + // Extract components. + uint64_t sa = 0xFFFF & (src_premul >> 48); + uint64_t sr = 0xFFFF & (src_premul >> 32); + uint64_t sg = 0xFFFF & (src_premul >> 16); + uint64_t sb = 0xFFFF & (src_premul >> 0); + uint64_t da = 0xFFFF & (dst_nonpremul >> 48); + uint64_t dr = 0xFFFF & (dst_nonpremul >> 32); + uint64_t dg = 0xFFFF & (dst_nonpremul >> 16); + uint64_t db = 0xFFFF & (dst_nonpremul >> 0); + + // Convert dst from nonpremul to premul. + dr = (dr * da) / 0xFFFF; + dg = (dg * da) / 0xFFFF; + db = (db * da) / 0xFFFF; + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint64_t ia = 0xFFFF - sa; + + // Composite src (premul) over dst (premul). + da = sa + ((da * ia) / 0xFFFF); + dr = sr + ((dr * ia) / 0xFFFF); + dg = sg + ((dg * ia) / 0xFFFF); + db = sb + ((db * ia) / 0xFFFF); + + // Convert dst from premul to nonpremul. + if (da != 0) { + dr = (dr * 0xFFFF) / da; + dg = (dg * 0xFFFF) / da; + db = (db * 0xFFFF) / da; + } + + // Combine components. + return (db << 0) | (dg << 16) | (dr << 32) | (da << 48); +} + static inline uint32_t // wuffs_base__composite_premul_nonpremul_u32_axxx(uint32_t dst_premul, uint32_t src_nonpremul) { @@ -1150,6 +1188,91 @@ wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_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__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__bgra_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 sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 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__rgb(uint8_t* dst_ptr, size_t dst_len, @@ -1609,6 +1732,82 @@ wuffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_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] = s0; + d[1] = s1; + d[2] = s2; + d[3] = 0xFF; + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_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 sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 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 // wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src(uint8_t* dst_ptr, size_t dst_len, @@ -1778,6 +1977,64 @@ wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_nonpremul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgra_nonpremul__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -1844,6 +2101,66 @@ wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_nonpremul__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_nonpremul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -1941,6 +2258,63 @@ wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_nonpremul_4x16le__src_ov return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_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__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__bgra_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__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 // wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -2171,6 +2545,37 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgra_premul__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -2337,6 +2742,36 @@ wuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -3871,6 +4306,85 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( return NULL; } +static wuffs_base__pixel_swizzler__func // +wuffs_base__pixel_swizzler__prepare__bgra_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__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__bgra_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__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr__bgra_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__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__BGRA_NONPREMUL_4X16LE: + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src_over; + } + return NULL; + + case WUFFS_BASE__PIXEL_FORMAT__BGRA_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; + + case WUFFS_BASE__PIXEL_FORMAT__RGBA_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__RGBA_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; + } + return NULL; +} + static wuffs_base__pixel_swizzler__func // wuffs_base__pixel_swizzler__prepare__bgrx(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_format dst_pixfmt, @@ -4127,6 +4641,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p, p, dst_pixfmt, dst_palette, src_palette, blend); break; + case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: + func = wuffs_base__pixel_swizzler__prepare__bgra_premul( + p, dst_pixfmt, dst_palette, src_palette, blend); + break; + case WUFFS_BASE__PIXEL_FORMAT__BGRX: func = wuffs_base__pixel_swizzler__prepare__bgrx( p, dst_pixfmt, dst_palette, src_palette, blend); diff --git a/internal/cgen/data/data.go b/internal/cgen/data/data.go index 7cb0ddaa..5c0796bd 100644 --- a/internal/cgen/data/data.go +++ b/internal/cgen/data/data.go @@ -575,10 +575,11 @@ const BasePixConvSubmoduleC = "" + "FFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert dst from premul to nonpremul.\n if (da != 0) {\n dr = (dr * 0xFFFF) / da;\n dg = (dg * 0xFFFF) / da;\n db = (db * 0xFFFF) / da;\n }\n\n // Convert from 16-bit color to 8-bit color.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint64_t //\nwuffs_base__composite_nonpremul_nonpremul_u64_axxx(uint64_t dst_nonpremul,\n uint64_t src_nonpremul) {\n // Extract components.\n uint64_t sa = 0xFFFF & (src_nonpremul >> 48);\n uint64_t sr = 0xFFFF & (src_nonpremul >> 32);\n uint64_t sg = 0xFFFF & (src_nonpremul >> 16);\n uint64_t sb = 0xFFFF & (src_nonpremul >> 0);\n uint64_t da = 0xFFFF & (dst_nonpremul >> 48);\n uint64_t dr = 0xFFFF & (dst_nonpremul >> 32);\n uint64_t dg = 0xFFFF & (dst_nonpremul >> 16);\n uint64_t db = 0xFFFF & (dst_nonpremul >> 0);\n\n // Convert ds" + "t from nonpremul to premul.\n dr = (dr * da) / 0xFFFF;\n dg = (dg * da) / 0xFFFF;\n db = (db * da) / 0xFFFF;\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint64_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n da = sa + ((da * ia) / 0xFFFF);\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Convert dst from premul to nonpremul.\n if (da != 0) {\n dr = (dr * 0xFFFF) / da;\n dg = (dg * 0xFFFF) / da;\n db = (db * 0xFFFF) / da;\n }\n\n // Combine components.\n return (db << 0) | (dg << 16) | (dr << 32) | (da << 48);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_nonpremul_premul_u32_axxx(uint32_t dst_nonpremul,\n uint32_t src_premul) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * (0xFF & (src_premul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_premul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_premul >> 8)" + ");\n uint32_t sb = 0x101 * (0xFF & (src_premul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_nonpremul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_nonpremul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_nonpremul >> 8));\n uint32_t db = 0x101 * (0xFF & (dst_nonpremul >> 0));\n\n // Convert dst from nonpremul to premul.\n dr = (dr * da) / 0xFFFF;\n dg = (dg * da) / 0xFFFF;\n db = (db * da) / 0xFFFF;\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 da = sa + ((da * ia) / 0xFFFF);\n dr = sr + ((dr * ia) / 0xFFFF);\n dg = sg + ((dg * ia) / 0xFFFF);\n db = sb + ((db * ia) / 0xFFFF);\n\n // Convert dst from premul to nonpremul.\n if (da != 0) {\n dr = (dr * 0xFFFF) / da;\n dg = (dg * 0xFFFF) / da;\n db = (db * 0xFFFF) / da;\n }\n\n // Convert from 16-bit color to 8-bit color.\n da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n" + - "\nstatic inline uint32_t //\nwuffs_base__composite_premul_nonpremul_u32_axxx(uint32_t dst_premul,\n uint32_t src_nonpremul) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * (0xFF & (src_nonpremul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_nonpremul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_nonpremul >> 8));\n uint32_t sb = 0x101 * (0xFF & (src_nonpremul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_premul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_premul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_premul >> 8));\n uint32_t db = 0x101 * (0xFF & (dst_premul >> 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 da = sa + ((da * ia) / 0xFFFF);\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 da >>= 8;\n dr >>= 8;" + - "\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint64_t //\nwuffs_base__composite_premul_nonpremul_u64_axxx(uint64_t dst_premul,\n uint64_t src_nonpremul) {\n // Extract components.\n uint64_t sa = 0xFFFF & (src_nonpremul >> 48);\n uint64_t sr = 0xFFFF & (src_nonpremul >> 32);\n uint64_t sg = 0xFFFF & (src_nonpremul >> 16);\n uint64_t sb = 0xFFFF & (src_nonpremul >> 0);\n uint64_t da = 0xFFFF & (dst_premul >> 48);\n uint64_t dr = 0xFFFF & (dst_premul >> 32);\n uint64_t dg = 0xFFFF & (dst_premul >> 16);\n uint64_t db = 0xFFFF & (dst_premul >> 0);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint64_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul).\n da = sa + ((da * ia) / 0xFFFF);\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Combine components.\n return " + - "(db << 0) | (dg << 16) | (dr << 32) | (da << 48);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul,\n uint32_t src_premul) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * (0xFF & (src_premul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_premul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_premul >> 8));\n uint32_t sb = 0x101 * (0xFF & (src_premul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_premul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_premul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_premul >> 8));\n uint32_t db = 0x101 * (0xFF & (dst_premul >> 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 da = sa + ((da * ia) / 0xFFFF);\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 da >>= 8;\n dr" + - " >>= 8;\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\n" + + "\nstatic inline uint64_t //\nwuffs_base__composite_nonpremul_premul_u64_axxx(uint64_t dst_nonpremul,\n uint64_t src_premul) {\n // Extract components.\n uint64_t sa = 0xFFFF & (src_premul >> 48);\n uint64_t sr = 0xFFFF & (src_premul >> 32);\n uint64_t sg = 0xFFFF & (src_premul >> 16);\n uint64_t sb = 0xFFFF & (src_premul >> 0);\n uint64_t da = 0xFFFF & (dst_nonpremul >> 48);\n uint64_t dr = 0xFFFF & (dst_nonpremul >> 32);\n uint64_t dg = 0xFFFF & (dst_nonpremul >> 16);\n uint64_t db = 0xFFFF & (dst_nonpremul >> 0);\n\n // Convert dst from nonpremul to premul.\n dr = (dr * da) / 0xFFFF;\n dg = (dg * da) / 0xFFFF;\n db = (db * da) / 0xFFFF;\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint64_t ia = 0xFFFF - sa;\n\n // Composite src (premul) over dst (premul).\n da = sa + ((da * ia) / 0xFFFF);\n dr = sr + ((dr * ia) / 0xFFFF);\n dg = sg + ((dg * ia) / 0xFFFF);\n db = sb + ((db * ia) / 0xFFFF);\n\n // Convert dst from premul to nonpremul.\n " + + " if (da != 0) {\n dr = (dr * 0xFFFF) / da;\n dg = (dg * 0xFFFF) / da;\n db = (db * 0xFFFF) / da;\n }\n\n // Combine components.\n return (db << 0) | (dg << 16) | (dr << 32) | (da << 48);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_nonpremul_u32_axxx(uint32_t dst_premul,\n uint32_t src_nonpremul) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * (0xFF & (src_nonpremul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_nonpremul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_nonpremul >> 8));\n uint32_t sb = 0x101 * (0xFF & (src_nonpremul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_premul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_premul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_premul >> 8));\n uint32_t db = 0x101 * (0xFF & (dst_premul >> 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 da = sa + ((da * ia) / 0xFFFF);\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 da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\nstatic inline uint64_t //\nwuffs_base__composite_premul_nonpremul_u64_axxx(uint64_t dst_premul,\n uint64_t src_nonpremul) {\n // Extract components.\n uint64_t sa = 0xFFFF & (src_nonpremul >> 48);\n uint64_t sr = 0xFFFF & (src_nonpremul >> 32);\n uint64_t sg = 0xFFFF & (src_nonpremul >> 16);\n uint64_t sb = 0xFFFF & (src_nonpremul >> 0);\n uint64_t da = 0xFFFF & (dst_premul >> 48);\n uint64_t dr = 0xFFFF & (dst_premul >> 32);\n uint64_t dg = 0xFFFF & (dst_premul >> 16);\n uint64_t db = 0xFFFF & (dst_premul >> 0);\n\n // Calculate the inverse of the src-alpha: how much of the dst to keep.\n uint64_t ia = 0xFFFF - sa;\n\n // Composite src (nonpremul) over dst (premul)" + + ".\n da = sa + ((da * ia) / 0xFFFF);\n dr = ((sr * sa) + (dr * ia)) / 0xFFFF;\n dg = ((sg * sa) + (dg * ia)) / 0xFFFF;\n db = ((sb * sa) + (db * ia)) / 0xFFFF;\n\n // Combine components.\n return (db << 0) | (dg << 16) | (dr << 32) | (da << 48);\n}\n\nstatic inline uint32_t //\nwuffs_base__composite_premul_premul_u32_axxx(uint32_t dst_premul,\n uint32_t src_premul) {\n // Extract 16-bit color components.\n uint32_t sa = 0x101 * (0xFF & (src_premul >> 24));\n uint32_t sr = 0x101 * (0xFF & (src_premul >> 16));\n uint32_t sg = 0x101 * (0xFF & (src_premul >> 8));\n uint32_t sb = 0x101 * (0xFF & (src_premul >> 0));\n uint32_t da = 0x101 * (0xFF & (dst_premul >> 24));\n uint32_t dr = 0x101 * (0xFF & (dst_premul >> 16));\n uint32_t dg = 0x101 * (0xFF & (dst_premul >> 8));\n uint32_t db = 0x101 * (0xFF & (dst_premul >> 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 da =" + + " sa + ((da * ia) / 0xFFFF);\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 da >>= 8;\n dr >>= 8;\n dg >>= 8;\n db >>= 8;\n\n // Combine components.\n return (db << 0) | (dg << 8) | (dr << 16) | (da << 24);\n}\n\n" + "" + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__squash_align4_bgr_565_8888(uint8_t* dst_ptr,\n size_t dst_len,\n const uint8_t* src_ptr,\n size_t src_len,\n bool nonpremul) {\n size_t len = (dst_len < src_len ? dst_len : src_len) / 4;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n size_t n = len;\n while (n--) {\n uint32_t argb = wuffs_base__peek_u32le__no_bounds_check(s);\n if (nonpremul) {\n argb =\n wuffs_base__color_u32_argb_nonpremul__as__color_u32_argb_premul(argb);\n }\n uint32_t b5 = 0x1F & (argb >> (8 - 5));\n uint32_t g6 = 0x3F & (argb >> (16 - 6));\n uint32_t r5 = 0x1F & (argb >> (24 - 5));\n uint32_t alpha = argb & 0xFF000000;\n wuffs_base__poke_u32le__no_bounds_check(\n d, alpha | (r5 << 11) | (g6 << 5) | (b5 << 0));\n s += 4;\n " + " d += 4;\n }\n return len;\n}\n\n" + @@ -598,44 +599,57 @@ const BasePixConvSubmoduleC = "" + " uint32_t sg = 0x101 * ((uint32_t)s[1]);\n uint32_t sb = 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) | (n" + "ew_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__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_len2 = dst_len / 2;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len2 < src_len8) ? dst_len2 : 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 // Extract 16-bit color components.\n uint32_t sa = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 6));\n uint32_t sr = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 4));\n uint32_t sg = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 2));\n uint32_t sb = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 0));\n\n // Convert f" + "rom 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 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgb(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_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_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__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 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 // Compo" + - "site 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_l" + - "en2 < 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 wu" + - "ffs_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 r" + - "eturn 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 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" + + "565);\n\n s += 1 * 8;\n d += 1 * 2;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__bgra_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), wuffs_base__color_u32_argb_premul__as__color_u16_rgb_565(\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__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_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 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\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\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr_565__rgb(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n s" + + "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" + "" + "// --------\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 " + "/ 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 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 (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" + "\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__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_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 // Extract 16-bit color components.\n uint32_t sa = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 6));\n uint32_t sr = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 4));\n uint32_t sg = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(s + 2));\n uint32_t sb = ((uint32_t)wuffs_base__peek_u16le__no_bounds_check(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 d" + - "st 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 * 8;\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" + + "st 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 * 8;\n d += 1 * 3;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgr__bgra_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 = ds" + + "t_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] = s0;\n d[1] = s1;\n d[2] = s2;\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__bgra_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 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\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(" + - "\n wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)));\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__as__color_u32(\n wuffs_base__composite_nonpremul_nonpremul_u64_axxx(d0, 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_nonpremul__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_len4 = dst_len / 4;\n size_t len = (dst_len4 < src_len) ? dst_len4 : 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__peek_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_p" + - "alette_ptr +\n ((size_t)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 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__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_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__swap_u32_argb_abgr(\n 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__c" + - "omposite_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\n" + + "\n wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)));\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__as__color_u32(\n wuffs_base__composite_nonpremul_nonpremul_u64_axxx(d0, 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_nonpremul__bgra_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_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 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_premul__as__" + + "color_u32_argb_nonpremul(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_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_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), wuffs_base__composite_nonpremul_premul_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__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_len4 = dst_len / 4;\n size_t len = (dst_len4 < src_len) ? dst_len4 : 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__peek_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)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 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul__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 sr" + + "c_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__swap_u32_argb_abgr(\n 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__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_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 s0 = wuffs_base__swap_u32_argb_abgr(\n 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_premul__as__color_u32_argb_nonpremul(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__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_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__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke" + + "_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_nonpremul_premul_u32_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + "" + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__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_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] = s0;\n d[1] = s0;\n d[2] = s1;\n d[3] = s1;\n d[4] = s2;\n d[5] = s2;\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__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_len8 = dst_le" + "n / 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),\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__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_len8 = dst_len / 8;\n size_t src_len8 = src_len / 8;\n size_t len = (dst_len8 < src_len8) ? dst_len8 : src_len8;\n uint8_t* d = dst_ptr;\n const uint8_t* s = src_ptr;\n\n siz" + - "e_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__peek_u64le__no_bounds_check(s + (0 * 8));\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 * 8;\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 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_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" + + "e_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__peek_u64le__no_bounds_check(s + (0 * 8));\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 * 8;\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(\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__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__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" + "" + "// --------\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" + "unds_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), wuffs_base__composite_premul_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_premul__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 // TODO: unroll.\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__color_u32__as__color_u64(\n wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)));\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__as__color_u32(\n wuffs_base__composite_premul_nonpremul_u64_axxx(d0, 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__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_len4 = dst_len / 4;\n size_t len = (dst_len4 < src_len) ? dst_len4 : 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__peek_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx" + - "(d0, s0));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__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_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__swap_u32_argb_abgr(\n 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__rgba_nonpremul__src_over(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palett" + - "e_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_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_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_premul__rgba_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__swap_u32_argb_abgr(\n wuffs_base__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n 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__rgba_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 // TODO: unroll.\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__color_u32__as__" + - "color_u64(\n wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)));\n uint64_t s0 = wuffs_base__swap_u64_argb_abgr(\n 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__as__color_u32(\n wuffs_base__composite_premul_nonpremul_u64_axxx(d0, s0)));\n\n s += 1 * 8;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n" + + "),\n wuffs_base__color_u64__as__color_u32(\n wuffs_base__composite_premul_nonpremul_u64_axxx(d0, 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_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_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_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), wuffs_base__composite_premul_premul_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__p" + + "ixel_swizzler__bgra_premul__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_len4 = dst_len / 4;\n size_t len = (dst_len4 < src_len) ? dst_len4 : 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__peek_u32le__no_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(dst_palette_ptr +\n ((size_t)s[0] * 4));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_nonpremul_u32_axxx(d0, s0));\n\n s += 1 * 1;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul__src(\n uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_p" + + "alette_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__swap_u32_argb_abgr(\n 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__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_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_bounds_check(d + (0 * 4));\n uint32_t s0 = wuffs_base__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_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_premul__rgba_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__swap_u32_argb_abgr(\n wuffs_base__color_u64_argb_nonpremul__as__color_u32_argb_premul(\n 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__rgba_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 // TODO: unroll.\n\n while (n >= 1) {\n uint64_t d0 = wuffs_base__color_u32__as__color_u64(\n wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)));\n uint64_t s0 = wuffs_base__swap_u64_argb_abgr(\n wuffs_base__peek_u64le__no_bounds_check(s + (0 * 8)));\n wuffs_base__poke_u32le__no_bo" + + "unds_check(\n d + (0 * 4),\n wuffs_base__color_u64__as__color_u32(\n wuffs_base__composite_premul_nonpremul_u64_axxx(d0, 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__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_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__swap_u32_argb_abgr(\n wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4), wuffs_base__composite_premul_premul_u32_axxx(d0, s0));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n" + + "\n return len;\n}\n\n" + "" + "// --------\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgrw__bgr(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_len3 = src_len / 3;\n size_t len = (dst_len4 < src_len3) ? dst_len4 : 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 wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF000000 | wuffs_base__peek_u24le__no_bounds_check(s + (0 * 3)));\n\n s += 1 * 3;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgrw__bgrx(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palett" + "e_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 wuffs_base__poke_u32le__no_bounds_check(\n d + (0 * 4),\n 0xFF000000 | wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)));\n\n s += 1 * 4;\n d += 1 * 4;\n n -= 1;\n }\n\n return len;\n}\n\n#if defined(WUFFS_BASE__CPU_ARCH__X86_64)\nWUFFS_BASE__MAYBE_ATTRIBUTE_TARGET(\"sse4.2\")\nstatic uint64_t //\nwuffs_base__pixel_swizzler__bgrw__rgb__sse42(uint8_t* dst_ptr,\n size_t dst_len,\n uint8_t* dst_palette_ptr,\n size_t dst_palette_len,\n " + @@ -696,21 +710,23 @@ const BasePixConvSubmoduleC = "" + "ra_nonpremul_4x16le(\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__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul_4x16le__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__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__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 wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__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__copy_8_8;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_nonpremul_4x16le__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__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over;\n }\n return NULL;\n\n case WUF" + "FS_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__rgba_nonpremul__bgra_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__rgba_nonpremul__bgra_nonpremul_4x16le__src_over;\n }\n break;\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__rgba_nonpremul_4x16le__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul_4x16le__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 br" + - "eak;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgrx(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 return wuffs_base__pixel_swizzler__bgr_565__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return wuffs_base__pixel_swizzler__xxx__xxxx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__pixel_swizzler__bgrw__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n return wuffs_base__pixel_swizzler__bgrw_4x16le__bgrx;\n\n case WUFFS_BASE__PIXEL_FORM" + - "AT__BGRX:\n return wuffs_base__pixel_swizzler__copy_4_4;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\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__rgbx;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__rgb(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 return wuffs_base__pixel_swizzler__bgr_565__rgb;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return wuffs_base__pixel_swizzler__sw" + - "ap_rgb_bgr;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__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_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_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_swiz" + - "zler__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_F" + - "ORMAT__RGBX:\n // TODO.\n break;\n }\n return NULL;\n}\n\n" + + "eak;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgra_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__bgra_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr_565__bgra_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__bgra_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgr__bgra_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 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__BGRA_NONPREMUL_4X16LE:\n switch (blend) {\n case WUFFS_BASE__PIXEL_BLEND__SRC:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src;\n case WUFFS_BASE__PIXEL_BLEND__SRC_OVER:\n return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_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 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 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__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__RGBA_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 wuffs_base__pixel_swizzler__bgra_premul__rgba_premul__src_over;\n }\n return NULL;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__bgrx(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 return wuffs_base__pixel_swizzler__bgr_565__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return wuffs_base__pixel_swizzler__xxx__xxxx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_BINARY:\n return wuffs_base__pixel_swizzler__bgrw__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE:\n return wuffs_base__pixel_swizzler__bgrw_4x16le__bgrx;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRX:\n return wuffs_base__pixel_swizzler__copy_4_4;\n\n case WUFFS_BASE__PIXEL_FORMAT__RGB:\n // TODO.\n break;\n\n case WUFFS_B" + + "ASE__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__rgbx;\n }\n return NULL;\n}\n\nstatic wuffs_base__pixel_swizzler__func //\nwuffs_base__pixel_swizzler__prepare__rgb(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 return wuffs_base__pixel_swizzler__bgr_565__rgb;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGR:\n return wuffs_base__pixel_swizzler__swap_rgb_bgr;\n\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL:\n case WUFFS_BASE__PIXEL_FORMAT__" + + "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" + "" + "// --------\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__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_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_s" + - "wizzler_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 cons" + - "t 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\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" + "" const BaseUTF8SubmoduleC = "" + diff --git a/release/c/wuffs-unsupported-snapshot.c b/release/c/wuffs-unsupported-snapshot.c index 1b17f927..b7afdb77 100644 --- a/release/c/wuffs-unsupported-snapshot.c +++ b/release/c/wuffs-unsupported-snapshot.c @@ -15369,6 +15369,44 @@ wuffs_base__composite_nonpremul_premul_u32_axxx(uint32_t dst_nonpremul, return (db << 0) | (dg << 8) | (dr << 16) | (da << 24); } +static inline uint64_t // +wuffs_base__composite_nonpremul_premul_u64_axxx(uint64_t dst_nonpremul, + uint64_t src_premul) { + // Extract components. + uint64_t sa = 0xFFFF & (src_premul >> 48); + uint64_t sr = 0xFFFF & (src_premul >> 32); + uint64_t sg = 0xFFFF & (src_premul >> 16); + uint64_t sb = 0xFFFF & (src_premul >> 0); + uint64_t da = 0xFFFF & (dst_nonpremul >> 48); + uint64_t dr = 0xFFFF & (dst_nonpremul >> 32); + uint64_t dg = 0xFFFF & (dst_nonpremul >> 16); + uint64_t db = 0xFFFF & (dst_nonpremul >> 0); + + // Convert dst from nonpremul to premul. + dr = (dr * da) / 0xFFFF; + dg = (dg * da) / 0xFFFF; + db = (db * da) / 0xFFFF; + + // Calculate the inverse of the src-alpha: how much of the dst to keep. + uint64_t ia = 0xFFFF - sa; + + // Composite src (premul) over dst (premul). + da = sa + ((da * ia) / 0xFFFF); + dr = sr + ((dr * ia) / 0xFFFF); + dg = sg + ((dg * ia) / 0xFFFF); + db = sb + ((db * ia) / 0xFFFF); + + // Convert dst from premul to nonpremul. + if (da != 0) { + dr = (dr * 0xFFFF) / da; + dg = (dg * 0xFFFF) / da; + db = (db * 0xFFFF) / da; + } + + // Combine components. + return (db << 0) | (dg << 16) | (dr << 32) | (da << 48); +} + static inline uint32_t // wuffs_base__composite_premul_nonpremul_u32_axxx(uint32_t dst_premul, uint32_t src_nonpremul) { @@ -15894,6 +15932,91 @@ wuffs_base__pixel_swizzler__bgr_565__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr_565__bgra_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__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__bgra_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 sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 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__rgb(uint8_t* dst_ptr, size_t dst_len, @@ -16353,6 +16476,82 @@ wuffs_base__pixel_swizzler__bgr__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_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] = s0; + d[1] = s1; + d[2] = s2; + d[3] = 0xFF; + + s += 1 * 4; + d += 1 * 3; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgr__bgra_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 sr = 0x101 * ((uint32_t)s[2]); + uint32_t sg = 0x101 * ((uint32_t)s[1]); + uint32_t sb = 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 // wuffs_base__pixel_swizzler__bgr__rgba_nonpremul__src(uint8_t* dst_ptr, size_t dst_len, @@ -16522,6 +16721,64 @@ wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_nonpremul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgra_nonpremul__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -16588,6 +16845,66 @@ wuffs_base__pixel_swizzler__bgra_nonpremul__rgba_nonpremul__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), + wuffs_base__color_u32_argb_premul__as__color_u32_argb_nonpremul(s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_nonpremul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -16685,6 +17002,63 @@ wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_nonpremul_4x16le__src_ov return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_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__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__bgra_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__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 // wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -16915,6 +17289,37 @@ wuffs_base__pixel_swizzler__bgra_premul__bgra_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__bgra_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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + // TODO: unroll. + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4)); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + static uint64_t // wuffs_base__pixel_swizzler__bgra_premul__index_bgra_nonpremul__src_over( uint8_t* dst_ptr, @@ -17081,6 +17486,36 @@ wuffs_base__pixel_swizzler__bgra_premul__rgba_nonpremul_4x16le__src_over( return len; } +static uint64_t // +wuffs_base__pixel_swizzler__bgra_premul__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_len4 = dst_len / 4; + size_t src_len4 = src_len / 4; + size_t len = (dst_len4 < src_len4) ? dst_len4 : src_len4; + uint8_t* d = dst_ptr; + const uint8_t* s = src_ptr; + size_t n = len; + + while (n >= 1) { + uint32_t d0 = wuffs_base__peek_u32le__no_bounds_check(d + (0 * 4)); + uint32_t s0 = wuffs_base__swap_u32_argb_abgr( + wuffs_base__peek_u32le__no_bounds_check(s + (0 * 4))); + wuffs_base__poke_u32le__no_bounds_check( + d + (0 * 4), wuffs_base__composite_premul_premul_u32_axxx(d0, s0)); + + s += 1 * 4; + d += 1 * 4; + n -= 1; + } + + return len; +} + // -------- static uint64_t // @@ -18615,6 +19050,85 @@ wuffs_base__pixel_swizzler__prepare__bgra_nonpremul_4x16le( return NULL; } +static wuffs_base__pixel_swizzler__func // +wuffs_base__pixel_swizzler__prepare__bgra_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__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr_565__bgra_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__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgr__bgra_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__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__BGRA_NONPREMUL_4X16LE: + switch (blend) { + case WUFFS_BASE__PIXEL_BLEND__SRC: + return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src; + case WUFFS_BASE__PIXEL_BLEND__SRC_OVER: + return wuffs_base__pixel_swizzler__bgra_nonpremul_4x16le__bgra_premul__src_over; + } + return NULL; + + case WUFFS_BASE__PIXEL_FORMAT__BGRA_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; + + case WUFFS_BASE__PIXEL_FORMAT__RGBA_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__RGBA_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; + } + return NULL; +} + static wuffs_base__pixel_swizzler__func // wuffs_base__pixel_swizzler__prepare__bgrx(wuffs_base__pixel_swizzler* p, wuffs_base__pixel_format dst_pixfmt, @@ -18871,6 +19385,11 @@ wuffs_base__pixel_swizzler__prepare(wuffs_base__pixel_swizzler* p, p, dst_pixfmt, dst_palette, src_palette, blend); break; + case WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL: + func = wuffs_base__pixel_swizzler__prepare__bgra_premul( + p, dst_pixfmt, dst_palette, src_palette, blend); + break; + case WUFFS_BASE__PIXEL_FORMAT__BGRX: func = wuffs_base__pixel_swizzler__prepare__bgrx( p, dst_pixfmt, dst_palette, src_palette, blend); diff --git a/test/c/std/wbmp.c b/test/c/std/wbmp.c index 0ce8fd58..8c87f237 100644 --- a/test/c/std/wbmp.c +++ b/test/c/std/wbmp.c @@ -254,6 +254,10 @@ test_wuffs_pixel_swizzler_swizzle() { .color = 0x55443300, .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRA_NONPREMUL_4X16LE, }, + { + .color = 0x55443300, + .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRA_PREMUL, + }, { .color = 0xFF443300, .pixfmt_repr = WUFFS_BASE__PIXEL_FORMAT__BGRX,